WPS6826


Policy Research Working Paper                     6826




            Cash Transfers and Child Labor
                               Jacobus de Hoop
                                Furio C. Rosati




The World Bank
Development Economics Vice Presidency
Partnerships, Capacity Building Unit
March 2014
Policy Research Working Paper 6826


  Abstract
  Cash transfer programs are widely used in settings                                child labor and hours worked and cushion the effect of
  where child labor is prevalent. Although many of                                  economic shocks that may lead households to use child
  these programs are explicitly implemented to improve                              labor as a coping strategy. Boys experience particularly
  children’s welfare, in theory their impact on child labor                         strong decreases in economic activities, girls in household
  is undetermined. This paper systematically reviews the                            chores. The findings underline the usefulness of cash
  empirical evidence on the impact of cash transfers,                               transfers as a relatively safe policy instrument to improve
  conditional and unconditional, on child labor. The                                child welfare, but also point to knowledge gaps, for
  authors find no evidence that cash transfer interventions                         instance regarding the interplay between cash transfers
  increase child labor in practice. On the contrary, there                          and other interventions, that should be addressed in
  is broad evidence that conditional and unconditional                              future evaluations to provide detailed policy advice.
  cash transfers lower both children’s participation in




  This paper is a product of the Partnerships, Capacity Building Unit, Development Economics Vice Presidency. It is part
  of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy
  discussions around the world. Policy Research Working Papers are also posted on the Web at http://econ.worldbank.org.
  The authors may be contacted at jdehoop@ucw-project.org.




         The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development
         issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the
         names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those
         of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and
         its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent.


                                                       Produced by the Research Support Team
                                       Cash Transfers and Child Labor
                                         Jacobus de Hoop* & Furio C. Rosati°i




Keywords: Cash Transfers, Child Labor, Impact Evaluation, Review
JEL codes: I28, I38, O20

         i
           *Understanding Children’s Work. ° Understanding Children’s Work, University of Rome Tor Vergata, IZA Funding for
this paper was provided by the United States Department of Labor. This paper does not necessarily reflect the views or policies of the
United States Department of Labor, nor does the mention of trade names, commercial products, or organizations imply endorsement
by the United States Government. The paper benefited greatly from the valuable comments and suggestions of Marco Manacorda, the
editor, and the anonymous referees. The findings, interpretations, and conclusions expressed in this paper are entirely those of the
authors. They do not necessarily represent the views of Understanding Children’s Work or its partner organizations, the International
Labor Organization, UNICEF, and the World Bank. Corresponding author: Jacobus de Hoop, email address: jdehoop@ucw-
project.org; phone: +39-06-43412008; street address: International Labor Office for Italy and San Marino, Via Panisperna 28, 00184
Rome, Italy.

                                                                  1
        The International Labor Office (2013a) estimates that over 120 million 5- to 14-year-old

children were involved in labor in 2012, accounting for approximately 10% of the children in this

age group. Child labor constitutes a violation of children’s basic rights. It may have detrimental

short- and long-run effects on children’s lives, potentially lowering their school participation and

learning in school and possibly affecting their mental and physical health. Beyond its direct effects

on the children concerned, child labor may result in negative externalities. The economic literature

assumes that parents determine the child labor supply by weighing its harmful effects against its

potential benefits (primarily contributions to current household income). This literature shows that

child labor supply is likely to be inefficiently high even if parents are altruistic and externalities are

absent, thus providing a rationale for policy intervention (see, among others, Cigno and Rosati

2005; Edmonds 2007; and Udry 2006).

        This paper aims to add to our understanding of the role of policy interventions by

systematically reviewing the evidence on the impact of cash transfers, unconditional and
                                                             1
conditional, on child labor in developing countries.             Unconditional cash transfers provide

households with an income transfer, for instance to reduce poverty, address household vulnerability,

and encourage household investment in the human capital of their children. Conditional cash

transfer programs also provide an income transfer, but on the condition that the members of the

households receiving the transfer adhere to specific behavioral requirements. The behavioral

conditions are typically in the area of health (e.g., health checkups and attendance at health-related

seminars) and human capital accumulation (e.g., regular school attendance of children in the

household). 2

        Cash transfers are a particularly relevant category of interventions. They are widely used in

settings where child labor is prevalent and are often explicitly implemented to improve children’s

welfare. Nonetheless, in theory, their impact on child labor is undetermined. In fact, even increases

in school participation do not necessarily translate to reductions in child labor because school
participation and child labor are not mutually exclusive activities. The extensive evidence on the

beneficial effects of cash transfers on school participation (see, e.g., Baird et al. 2013; Fiszbein and

Schady 2009; Rawlings and Rubio 2005; Saavedra and Garcia 2012) thus provides little guidance

regarding their effects on child labor.

        A key issue that must be emphasized from the outset is that child labor is not a concept that

easily translates into statistical indicators. Child labor can affect children in different ways, and its

consequences for children’s welfare cannot be captured by a single indicator. Damage to child

health, for instance, depends on the participation in - and the length of exposure to - hazardous

activities and occupations. Similarly, the effects on the accumulation of human capital depend on

participation in work, work schedule and hours worked, sector of employment, and occupation. 3

Thus, detailed information is necessary to fully understand how policy interventions affect

children’s welfare. 4 Such detailed information, however, is seldom available. Instead, most studies

focus on participation in work, with a few studies discussing the impact on working hours. To

complicate matters further, there is substantial variation in the way “work” is defined across

different studies. Some studies focus on specific activities (such as work in agriculture), whereas

others use a more general definition (such as work in economic activities or household chores). 5

        To ensure consistency and incorporate evidence from the largest possible number of studies,

we mainly discuss the impact of cash transfer programs on children’s participation in economic

activities. To the extent possible, we complement this discussion with a separate analysis of the

impact of cash transfer programs on hours worked. Despite its limitations, the literature’s focus on

participation in work provides important insights. First, impact on participation serves as a useful

first-order approximation to examine whether cash transfers affect the working behavior of

children. Second, keeping children out of work is a sufficient condition to prevent detrimental

effects on outcomes such as health and human capital accumulation. Third, participation in work is

typically the key outcome in policy discussions and is an important outcome from a legal

perspective (most countries in the world have adopted child labor laws that set a minimum age
                                                   3
below which children are not allowed to work). Moreover, by focusing on participation in work, we

are able to highlight some of the limitations of the research conducted to date and the main

questions that remain to be answered in future evaluations. 6

       The remainder of this review is organized as follows. Section 1 provides the necessary

background. It heuristically describes why the effects of cash transfers on child labor are

theoretically undetermined, and it introduces the procedure that we used to identify the relevant

studies for this review. Section 2 discusses the impact of two subsets of unconditional cash

transfers: programs designed to support poor households’ investments in children’s human capital

and old age pension schemes. Section 3 discusses the average impact of conditional cash transfer

schemes on the intensive and extensive margin of child labor, the impact of conditional cash

transfers on child labor compared to their impact on school participation, heterogeneity by poverty,

age, and gender, spillover effects, long-run effects, determinants of program effects, protection from

shocks, and variations on the basic conditional cash transfer scheme. Section 4 discusses and

concludes.


<<A>> 1. Background


<<B>> 1.1. Theoretical Framework


       The effects of cash transfers on child labor cannot be determined a priori, as we briefly

discuss below. There are two main channels through which cash transfers may affect child labor: (i)

by modifying the propensity to attend school and (ii) by changing the returns to child labor.

Because a detailed theoretical discussion of these issues is beyond the scope of this paper, we

present a heuristic explanation in what follows. This discussion draws on the theoretical framework

presented in De Hoop and Rosati (2014). 7

       Consider a unitary household in which parents maximize utility over current consumption

(which depends partly on children’s work), children’s education, and leisure. For simplicity, and


                                                  4
because the cash transfer programs we consider are targeted at the poor, assume that households are

credit constrained. As a result, they may under-invest in children’s education and in productive

assets. Assume also that the adult labor supply is fixed, fertility is exogenous and fixed at one child,

and school participation is dichotomous (i.e., the child either does not attend school or spends a

fixed amount of time in school). The basic characteristics of an altruistic overlapping generation

model that are essential for our analysis are captured by this very simple set of assumptions.

       Within this framework, the effects of cash transfers, both conditional and unconditional,

depend on whether the transfer program affects school participation. 8 If the school participation of

the child is not affected (e.g., in case of children who would be in school also in absence of the cash

transfer program) and the household consumes the transfer (i.e., the transfer results only in an

income effect), then child labor should decrease. 9 However, if the household invests (part of) the

transfer in productive assets, the returns to child work may increase, thereby counterbalancing the

income effect and possibly resulting in increased child labor. 10

        Figure 1 helps explain what happens when school participation is affected by the transfer.

Household consumption is plotted on the vertical axis, and child leisure is plotted on the horizontal

axis. The budget constraint for a household that does not receive a transfer and does not send its

child to school is given by the solid line ABC. The downward slope of the budget constraint reflects

the decrease in consumption as the child works less (i.e., consumes more leisure). Household

consumption is positive even if the child does not work and spends all available time on leisure

(point B) because the household also relies on income from other household members.

       If the child begins to attend school following a transfer, the time available to the child for

leisure and work is reduced (here, from C to F). Moreover, the household incurs the cost of

education (e.g., school fees). If the transfer does not fully compensate the monetary cost of

attending school, the budget constraint shifts downwards (for instance, to the dashed line GHF).

Here, the distance DA represents the monetary costs of attending school, and DG represents the

transfer. The change in child labor is undetermined because both feasible consumption and leisure
                                                   5
are reduced, and the final outcome depends on the relative change in the (utility) value of the two. If

the transfer exceeds the monetary cost of attending school, the budget constraint shifts upwards (for

instance, to the dashed line IJF). Child labor should unambiguously decrease if the household does

not invest the transfer in productive assets. However, in both cases, the effect on child labor also

depends on whether the household decides to invest (part) of the transfer in productive assets, which

may increase the returns to child labor.


<<B>> 1.2. Identifying Relevant Studies


       To identify potentially relevant studies that evaluate the impact of cash transfer

interventions on children’s work outcomes, we began with a literature search covering Google

Scholar’s electronic bibliographical database, the World Bank Development Impact Evaluation

Initiative (DIME) database, the Poverty Action Lab, the Social Science Research Network (SSRN),

Network of Networks for Impact Evaluation (NONIE), and the International Initiative for Impact

evaluation (3IE). 11 To determine which papers to include in our discussion, we split the identified

papers into two subgroups: peer-reviewed papers and non-reviewed papers. We included all of the

peer-reviewed papers in our discussion. We included non-reviewed studies if they applied a

plausible strategy to address endogenous program placement and self-selection into the program.

We did not limit our overview to randomized controlled trials (RCTs) and considered other

methodologies, such as regression discontinuity designs, natural experiments, and propensity score

matching studies. 12

       Table 1 provides an overview of the studies included in our primary analysis: seven studies

of unconditional cash transfer schemes and 23 studies of conditional cash transfer schemes. The

majority of studies, 23 in total, focus on cash transfer programs implemented in Latin America and

the Caribbean (column 1). Five studies focus on Mexico’s flagship conditional cash transfer

scheme, called Oportunidades. 13 Over half of the studies appeared in a peer-reviewed journal

(column 4). Fourteen studies are based on an RCT (column 5). The age range covered differs
                                                  6
substantially across studies, an issue that should be considered when interpreting our results

(column 6). In most studies, the main outcome variable on which we focus includes economic

activities for pay or for the household, although in some cases, the impact estimate refers to a

narrower (economic activities for pay) or a broader outcome variable (economic activities or

household chores) (column 7). 14


<<A>> 2. Unconditional Cash Transfers


<<B>> 2.1. Unconditional Cash Transfer Programs Targeted at Poor Households with

Children


       We discuss three unconditional cash transfer programs for which rigorous evidence is

available: Ecuador’s Bono de Desarrollo Humano, Malawi’s Social Cash Transfer Scheme, and

South Africa’s Child Support Grant (the results are summarized in Figure 2). 15 The evaluations of

these programs suggest that unconditional cash transfers that aim to encourage investment in human

capital tend to lower participation in economic activities. IV estimates using randomly assigned

eligibility status as an instrument indicate that Ecuador’s Bono de Desarrollo Humano had a

particularly strong effect. It lowered children’s participation in economic activities by 17 percentage

points for 6- to 17-year-old children (Schady and Araujo 2006) and by 25 percentage points for 11-

to 16-year-old children (Edmonds and Schady 2012), and it resulted in substantial reductions in

work for pay. It appears that South Africa’s Child Grant did not affect the time allocation of 10-

year-olds. However, propensity score dose-response estimates indicate that it did affect adolescents

aged 15 to 17; the probability that they worked outside the home was 21% if they started receiving

the grant at the age of 14 and 13% if they started receiving the grant at the infant or pre-school age

(DSD, SASSA, and UNICEF 2012). 16

       Covarrubias, Davis, and Winters (2012) find that Malawi’s Social Cash Transfer Scheme

increased household investment in productive agricultural assets. Perhaps as a result, there is


                                                  7
evidence of reductions in child labor, especially in domestic work, outside the household; however,

“the time freed seems to be replaced with greater involvement in within-household tasks”.

Participation in household chores, for instance, increased significantly by 8 percentage points

according to the authors’ propensity score matching estimates. Although participation in work in the

family farm did not change significantly, hours worked in the family farm or family business

increased. 17 Miller and Tsoka (2012) provide difference-in-difference estimates using the same data,

confirming that participation in household chores increased significantly, whereas participation in

work for pay decreased.


<<B>> 2.2. Old Age Pensions and the Role of Credit Constraints


       In our theoretical discussion, we assumed that households are credit constrained. If they are

not credit constrained, then household investment in education and productive assets (and,

concomitantly, child labor) should not change upon the receipt of a fully anticipated unconditional

cash transfer. Two studies examine whether credit constraints matter for child labor by studying the

impact of old-age pensions. Pensions are highly institutionalized and represent an anticipated and

relatively certain future income stream for the household. Economic theory suggests that in the

absence of credit constraints, households will follow the optimal smooth path of consumption and

investment. Therefore, we should not observe any discontinuity in the behavior of households, and

the fraction of working children in particular, just above and just below the pension age.

       Edmonds (2006) uses data from South Africa’s old-age pension scheme to test this

proposition. South Africa’s pensions are means tested and, as a result, primarily cover the

comparatively deprived black population of South Africa. The benefits provided by the pension

scheme are large: in 1999, they represented approximately 125% of median per capita income of

South Africa’s black population. School participation of 13- to 17-year-old children in the

household increases substantially when an eligible elderly person reaches the pension age, an effect

that is especially relevant for male pensioners. Children’s participation in economic activities does
                                                  8
not decline significantly when an elderly person in the household becomes eligible for the old-age

pensions. However, there is evidence of a significant decline in daily hours worked. 18 Boys

experience larger reductions in time spent on economic activities, whereas girls appear to

experience larger reductions in time spent on household chores. 19

        De Carvalho Filho (2012) confirms these findings for Brazil’s social pension scheme. The

author exploits a 1991 social security reform that increased the minimum benefit provided to

pension beneficiaries in rural areas and reduced the minimum eligibility age, comparing households

that became eligible to receive old-age benefits as a result of the reform to households that were

nearly eligible after the reform.20 Reduced form estimates indicate that the reform significantly

increased girls’ school participation and reduced their participation in economic activities for pay.

IV estimates that disentangle this effect (using eligibility status as an instrument) show that

increasing pension benefits by 100 Reais (approximately US$50) increases girls’ school enrollment

by nearly 10 percentage points and reduces the probability that girls work for pay by 3.6 percentage

points. It appears that effects of the program differ by the gender of the pension beneficiary, as only

girls living in a household where a female received the pension experienced a significant reduction

in economic activities for pay.

        In summary, the evidence from studies of pension schemes points to the relevance for child

labor of transfer schemes that do not have human capital accumulation as one of their objectives.

Moreover, it lends support to the hypothesis that unconditional cash transfers affect child labor at

least in part because they mitigate the effect of credit constraints.


<<A>> 3. Conditional Cash Transfers


<<B>> 3.1. Impact on Participation in Work and Hours Worked


        Conditional cash transfers have been evaluated extensively, allowing us to discuss their

impact in detail. Figure 3 synthesizes the evidence on the average impact of the conditional cash


                                                    9
transfer programs on participation in child labor. For some programs, we only have disaggregated

estimates by gender and/or age. Because we want to begin with a comparison of the average

program effects before discussing heterogeneous effects, we impute the average program effect for

those programs by taking the unweighted mean of the impact estimates given for different age and

gender groups (these imputed estimates are marked with a ǂ in the table). 21 For other programs, we

have estimates of average program effects from more than one study, in which case we show all of

the estimates. 22 To the greatest extent possible, we separately show impact estimates for urban and

rural areas. 23 For brevity, we do not provide a discussion of the identification strategy employed in

each individual paper, but Table 1 provides a basic description.

        The results indicate that although there is considerable variation across programs,

conditional cash transfer programs tend to reduce child labor. Impact estimates range from no

statistically significant change in eight of the 16 studies to a reduction in child labor of 10

percentage points for Cambodia’s CESSP scholarship program (Ferreira, Filmer, and Schady 2010).

We do not observe statistically significant increases in child labor for any of the programs, an

important finding given the theoretically ambiguous effect of conditional cash transfers on child

labor discussed above. The results suggest that the effects of any household investments in

productive assets and activities that draw children into work are offset by a stronger income and

substitution effect that keeps children in school and out of work. 24

        Figure 4 displays the impact of cash transfer programs on weekly hours worked. 25 Because

each of the included studies sets hours worked equal to zero for children who do not work, these

results effectively represent the combined effect of the included conditional cash transfer programs

on the intensive and extensive margin of child labor. 26 Although fewer studies focus on hours

worked than on participation in work, the results are qualitatively similar: conditional cash transfer

programs tend to reduce hours worked, and none of the studies finds evidence of a significant

increase in hours worked. In general, reductions in hours worked are modest (the average reduction

is approximately one hour and a half a week).
                                                   10
<<B>> 3.2. Comparing Impacts on School Participation and Child Labor


        The impact of cash transfer programs on child labor is correlated with their impact on

school participation. Of the eight studies that find a significant reduction in child labor, six also find

a significant increase in school participation (enrollment or attendance, depending on the outcome

examined in the study), and of the eight studies that find no significant reduction in child labor, only

three find a significant increase in school participation. However, the correlation between program

impact on child labor and school participation is not perfect. When we regress the child labor

impact estimates on the school participation impact estimates (results not displayed), we find

evidence of substantial and statistically significant co-movement. Each percentage point increase in

school participation is associated with a reduction in child labor of 0.31 percentage points,

suggesting that child labor is a key part of households’ human capital investment decisions.

However, the coefficient on school participation impacts is significantly different from -1,

indicating that changes in school participation are not fully mirrored in changes in child work.

        In fact, some of the differences in program impacts on school participation and participation

in work are remarkable. As noted above, there are cases in which conditional cash transfers had a

significant effect on education but no effect on child labor (e.g., Path in Jamaica). Perhaps more

unexpectedly, there are cases in which conditional cash transfers had a significant negative effect on

child labor but no effect on school participation (e.g., female school stipends in Pakistan). It is,

therefore, evident that complex adjustments in household behavior occur when cash transfers are

received (changes in working hours and/or in leisure time, changes in the type of activities

performed) and that the impact of a cash transfer on child labor is not necessarily the reciprocal of

its impact on education.


<<B>>3.3. Heterogeneous Effects




                                                   11
<<C>> 3.3.1. Heterogeneity by Income


        Above, we presented the results for the average impact of conditional cash transfer

programs. We now examine possible differences associated with some characteristics of the

beneficiaries: income (in this subsection) and age and gender (in the next). 27 A priori, there are

reasons to expect that the impact of conditional cash transfers will differ depending on the level of

household income. 28 Most importantly, because poor households are more likely to be affected by

credit constraints, transfers targeted to these households are more likely to affect investment in

human capital and to reduce inefficiently high levels of child labor.

        Two studies of the PRAF conditional cash transfer scheme in Honduras find that although

the program did not significantly reduce child labor overall (Figure 3), the effects of the program

differed depending on household income and were statistically significant for the poorer households

(Figure 5). Galiani and McEwan (2013) divide their sample of municipalities into height-for-age

quintiles (arguing that height for age provides an indication of poverty in the municipality) and find

that the program reduced children’s participation in economic activities by eight percentage points

(significant at the 5% level) in the poorest quintile. In the richest quintile, no statistically significant

change in child labor could be observed. Glewwe and Olinto (2004) interact their treatment

indicator with the log of per capita expenditure and find a qualitatively similar result (not displayed

in Figure 5): as the per capita expenditure decreases by one percentage point, the impact of PRAF-II

on children’s participation in economic activities increases significantly by 0.45 percentage points.

        Sparrow (2007) divides his sample into four per capita consumption quantiles to examine

the heterogeneous effects of Indonesia’s Jaringan Pengaman Social. He too finds that program

impact on child labor increases with poverty. In the lowest consumption quantile, children’s

participation in economic activities decreased by 4 percentage points, whereas in the combined

upper two quantiles, it decreased by only 3.3 percentage points (both are significant at the 1%

level). Only Dammert (2009) does not identify a differential impact of cash transfers on child labor

                                                    12
by income level. She investigates the effects of Nicaragua’s Red de Protección Social by interacting

a treatment dummy with marginality quintiles (based on a locality level marginality index). 29 The

impact estimates by marginality quintile are volatile, and the estimates for the four richest quintiles

are not significantly different from the impact coefficient for the poorest quintile.

        In conclusion, the impact of conditional cash transfers on child labor generally appears to be

larger for the poor. This finding is in accordance with the findings of Fiszbein and Schady (2009),

who observe that “numerous studies have shown larger [conditional cash transfer] program effects

among households that are poorer at baseline” on school participation. It is also in accordance with

Edmonds and Schady (2011), who find that reductions in child labor as a result of Ecuador’s Bono

de Desarrollo Humano unconditional cash transfer scheme are concentrated in the poorest

households. These results lend further support to the hypothesis that the mitigation of credit

constraints, which are more likely to be binding for poor households, is a key channel through

which cash transfers reduce child labor.


<<C>> 3.3.2. Heterogeneity by Age and Gender


        It is not obvious a priori whether we should expect stronger changes in work participation

for older or younger children. On the one hand, children’s participation in work, and hence the

margin for improvement in child labor outcomes, increases with age. On the other hand, as children

become older, their returns to work are likely to increase, making work a more attractive alternative

to compliance with a conditional cash transfer schooling requirement. To examine this issue, we use

data from all studies that examine the effect of a conditional cash transfer scheme on child labor for

two or more age groups. Using a regression, we test whether the impact on the younger and older

age group differs (Column (1) of Table 2). 30 We find no evidence of a significant correlation,

suggesting that the effect of conditional cash transfer schemes on child labor is not clearly

heterogeneous by age.



                                                   13
              In Column (2) of Table 2, we test whether the effect of conditional cash transfers is

heterogeneous by gender. To do so, we use the results from all of the studies that show impact

results separately for boys and girls. We find that conditional cash transfer schemes result in a 3.3

percentage point stronger reduction in child labor for males than females. In fact, a substantial

number of studies finds a significant impact on child labor among boys and no significant impact on

child labor among girls. Borraz and González (2009), who examine the impact of the PANES

conditional cash transfer program in Montevideo, Uruguay, are the only authors to find a significant

decrease in child labor for girls, but no significant decrease for boys. 31

              To better understand the differential impact of conditional cash transfer interventions on the

work of boys and girls, Figure 6 provides results from three studies that disaggregate the overall

impact of conditional cash transfers across different work activities by gender. Boys primarily

experience reductions in economic activities for pay. Oportunidades, for example, resulted in a

significant reduction in economic activities for pay conducted by boys, but not for girls (Skoufias

and Parker 2001). Similarly, the CESSP scholarship program in Cambodia (Ferreira, Filmer, and

Schady 2009) and the PRAF program in Honduras (Galiani and McEwan 2013) appear to have had

a stronger impact on work for pay and work outside the home, respectively, for boys than for
         32
girls.        Girls, in contrast, appear to experience larger reductions than boys in household chores

(Oportunidades in Mexico), 33 economic activities without pay (the CESSP program in Cambodia),

and work at home (PRAF in Honduras). 34

              Similar results are obtained by Del Carpio and Macours (2010), who focus on Atención a

Crisis, a one-year randomized pilot building on the Red de Protección Social cash transfer scheme

in Nicaragua. The authors test for differences in reductions in hours worked by boys and girls in

different activities during the week before the interview. 35 Fixed effects estimates indicate that the

reduction in participation in economic activities of boys as a result of the basic conditional cash

transfer program exceeded that of girls by more than one hour a week. 36 For household chores


                                                       14
(cooking, cleaning, washing, and caring for younger siblings), there was no significant difference

between boys and girls. 37

        It appears that a focus on economic activities does not reveal the full impact of conditional

cash transfers on girls. Girls are more likely to participate in household chores than in economic

activities; consequently, the impact of cash transfers on work performed by girls is likely to be

underestimated if we focus only on participation in economic activities. More elaborate survey

modules on children’s time use are necessary to fully understand the impact of cash transfers,

particularly on girls’ activities.


<<B>> 3.4. Spillover Effects and General Equilibrium Effects


        Conditional cash transfer programs may affect children who are not direct beneficiaries. At

the household level, for example, income effects may alter the probability that the siblings of the

beneficiary work. Conditional cash transfer programs may also result in spillover effects in the local

labor market. For example, pulling a substantial number of children out of work might affect

conditions in the local labor market and increase the marginal returns to child labor. 38

        Two evaluations suggest that such spillover effects are not highly relevant. Ferreira, Filmer,

and Schady (2009) and Galiani and McEwan (2013) compare the siblings of eligible children in

their treatment group to the siblings of eligible children in their control group for the CESSP

program in Cambodia and PRAF in Honduras, respectively. Neither study finds compelling

evidence that the siblings of eligible children in the treatment group altered their participation in

work in comparison with the siblings of eligible children in the control group. 39 However, a

different result is found by Barrera-Osorio et al. (2008, working paper version), who examine the

spillover effects of Colombia’s Subsidios Condicionados a la Asistencia Escolar conditional cash

transfer scheme on child labor. Within the sample of households that registered two children for the

program, there was evidence that children, especially girls, spent more hours in work if they were

not assigned to the program while their sibling was assigned to the program.
                                                  15
        Buddelmeyer and Skoufias (2004) investigate whether village-level spillover effects can be

observed for the Oportunidades conditional cash transfer scheme in Mexico. They exploit the fact

that within randomly selected intervention villages, only poor households were eligible to

participate in the program. Non-poor households in intervention villages were not eligible and can

thus be compared to non-poor households in control villages to identify spillover effects. The

authors find no significant village-level spillover effects on children’s participation in economic

activities.


<<B>> 3.5. Long-run Effects


        The impact evaluations discussed thus far examine the impact of conditional cash transfers

at one particular point in time (mainly shortly after the program began). However, the impact of a

program may vary significantly over its lifetime (King and Behrman 2009). For instance, if program

operation improves as providers become more experienced, the impact of the program may be

amplified over time. If the impact of the program depends on the duration of exposure to the

program, impact estimates based on data collected relatively soon after participants enter the

program may differ substantially from estimates based on data collected at a later stage.

        Behrman, Parker, and Todd (2011) examine the impact of Oportunidades in the longer run

and discuss whether these estimates differ from the short-run results. Propensity score estimates

suggest that the probability that boys who were 14 to 16 years old in 2003 (5.5 years after the

program was first implemented) worked was 14 percentage points lower in Oportunidades

communities than in communities that had never received benefits of the cash transfer scheme. 40

There is no evidence that work participation changed for girls in this age group (who were less

likely to work in the first place). The strong reduction in work by boys in the long run compared to

the modest impact in the short run (Skoufias and Parker (2001) found that boys aged 14-15 reduced

work participation by approximately 4 percentage points) suggests that the beneficial impact of the

Mexican conditional cash transfer program is compounded over time. It is possible that reduced
                                                 16
probabilities of dropping out of school in individual grades (an issue alluded to in Schultz, 2004)

begin to add up. The latter interpretation appears to be confirmed by Behrman, Parker, and Todd

(2011), who also register strong improvements in school participation for girls up to 18 and boys up

to 21 years old.


<<B>> 3.6. Differences in Program Impact


       We have observed that the estimates of the impact of conditional cash transfers on child

labor vary substantially. Based on the available information and without developing a new in-depth

quantitative and qualitative analysis for each program, it is impossible to fully identify the reasons

behind this variation. However, we can obtain some basic insight into the relationship between

program characteristics and program impact on child labor. In this section, we examine the role of

schooling conditionality and the size of the transfer.


<<C>> 3.6.1. Effect of the Conditionality


       A key question is, of course, whether the impact of conditional cash transfers on child labor

exceeds that of unconditional cash transfers. This question is not easy to answer by comparing the

effects of the unconditional and conditional cash transfers presented thus far. The decision to attach

a condition might be endogenous depending on the expected impact of the program in the target

population. Moreover, the studies included in this review do not always discuss the exact conditions

of the programs, how these conditions are communicated to beneficiaries, and to what extent

conditions are enforced. Hence, it may be the case that programs that are nominally conditional are

unconditional in practice.

       However, a few recent studies allow us to shed some light on the effects of schooling

conditions. Schady and Araujo (2006) and Edmonds and Schady (2012) exploit a glitch in the

rollout of the Bono de Desarrollo Humano program, which resulted in some beneficiary households

incorrectly believing that the cash transfers were provided conditional on school attendance.

                                                  17
Although the conclusions of the two papers are not entirely uniform, it appears that the effect of the

program on child labor was similar in the households that believed that the program was conditional

on school participation and in the households that did not.

       At the time of writing, preliminary evidence was also available from an experiment in which

households in rural Morocco were randomly selected to receive conditional and unconditional cash

transfers (Benhassine et al. 2012). Although we caution that these results are not conclusive, they

too suggest that changes in the time spent working in the household business are similar (and

insignificant) for conditional and unconditional cash transfers. 41 The similar impact of conditional

and unconditional cash transfers on child work is in contrast with more abundant evidence

suggesting that conditional cash transfers have a stronger impact on school participation than do

unconditional cash transfers (for a review, see Baird et al. 2013). It is evident that more information

is necessary to understand the effect of the schooling condition on child labor.


<<C>> 3.6.2. Amounts Transferred


       There is some experimental evidence on the impact of the size of the transfer on school

participation (e.g., Baird, McIntosh, and Özler 2011). No such evidence is available for child labor.

To examine the association between amounts transferred and changes in child labor, we regressed

the average impact of conditional cash transfer programs on the amount of money transferred as a

percentage of average household income (results not displayed). 42 We found no evidence that larger

transfer amounts are associated with a larger reduction in child labor. The lack of a negative

relationship between amounts transferred and changes in child labor is also apparent from the

individual studies. 43 The CESSP scholarship program in Cambodia, for instance, resulted in the

second strongest decrease in child labor of all evaluated conditional cash transfer programs even

though it provided only very modest transfers (equal to 2% to 3% of the total expenditures of the

average recipient household). Uruguay’s PANES, in contrast, provided income transfers equal to

approximately 50% of average self-reported pre-program household income and yet does not appear
                                                  18
to have lowered child labor. This finding is in line with the evidence synthesized by Kremer and

Holla (2009), suggesting that relatively small costs (for example, the costs of school uniforms) are

sufficient to keep children out of school and that relatively small subsidies can generate “sizeable

movements” in the take-up of health and education interventions. Together with the finding that

cash transfers have a stronger impact on the poor, this finding lends further support to the

hypothesis that the mitigation of credit constraints is a key channel through which cash transfers

affect child labor. Possibly, once sufficient cash has been transferred and credit constraints are no

longer binding, transferring additional cash to the household has no effect on child labor.


<<B>> 3.7. Protection from Shocks


       There is substantial evidence that households in developing countries use child labor to cope

with income shocks. Beegle, Dehejia, and Gatti (2006) use data from a household panel survey in

Tanzania to show that households increase child labor in response to transitory income shocks.

Similarly, Duryea, Lam, and Levison (2007) find that unemployment shocks experienced by male

household heads in Brazil significantly increase the probability that a child will enter the labor

force. Guarcello, Mealli, and Rosati (2010) show that in Guatemala, exposure to negative shocks

strongly influences household decisions and pushes children into work. An important question is

whether social protection programs, such as conditional cash transfers, can serve as a safety net

preventing income shocks from causing children to drop out of school and enter work.

       De Janvry et al. (2006) test whether the conditional cash transfers provided by the

Oportunidades program in Mexico protect children from household level shocks, including illness

of the head of household, loss of employment by the head of household, and natural disasters such

as drought and harvest failure. The authors find evidence of state dependence in schooling: children

may leave school as the result of a shock (particularly illness of the household head and locality-

level natural disasters) and there is a substantial risk that they will not return to school at a later

stage. Oportunidades strongly reduced the risk that pupils would leave school as a result of such
                                                  19
shocks. For instance, the risk that a child will drop out of school as a result of illness of the

household head or a locality-level disaster (1.7 and 3.2 percentage points in the absence of

Oportunidades, respectively) was virtually reduced to zero by the conditional cash transfer scheme.

No such protective effect could be observed for child labor, however. The same two shocks, illness

of the household head and locality-level natural disasters, increase children’s participation in

economic activities by 2.2 and 4.7 percentage points, respectively, in the absence of the program.

This increase was not significantly different in Oportunitades villages, suggesting that the protective

effect of the conditional transfer program was “not sufficient to reduce the use of child work as a

crucial element of [household] risk-coping strategies”.

        Following a similar approach, Fitzsimons and Mesnard (Forthcoming) examine whether

Colombia’s Familias en Acción conditional cash transfer scheme protected children from the

detrimental effects of the permanent departure of their father from the household due to death or

divorce. The authors show that the departure of the father is a quasi-random event accompanied by

a substantial reduction in household income. Children who experience the departure of their father

are approximately five percentage points less likely to attend school and approximately three

percentage points more likely to work. In contrast with De Janvry et al. (2006), Fitzsimons and

Mesnard (Forthcoming) find that conditional cash transfers strongly mitigate both the detrimental

effect of the departure of the father on school enrollment and the detrimental effect on participation

in work. 44


<<B>> 3.8. Variations on the Basic Conditional Cash Transfer Scheme


        Some studies examine whether changes in the basic setup of the conditional cash transfer

program affect their impact. Del Carpio and Loayza (2012) rely on a randomized experiment in

Nicaragua to compare the impact of simple conditional cash transfers to that of conditional cash

transfers combined with a “grant for productive investments” to start a new income-generating non-

agricultural activity. The grant was provided to households conditional on the development of a
                                                  20
business plan and was accompanied by technical assistance and training in basic business skills.

The two variants of the conditional cash transfer scheme do not appear to have had the same impact

on child participation in economic activities or household chores: the conditional cash transfers in

combination with the productive investment grant reduced child labor by 0.94 hours a week,

whereas the basic conditional cash transfer reduced child labor by 1.76 hours a week (the difference

between the point estimates is statistically significant at the 5% level).

        Barrera-Osorio et al. (2011) rely on a randomized experiment in Colombia to compare the

impact of simple conditional cash transfers to that of two variations on the traditional conditional

cash transfer scheme. The first variation provided a regular cash transfer that was equal to two-

thirds of the basic conditional cash transfer. The remaining accumulated one-third was transferred

shortly before the start of the following school year, thus potentially helping households cope with

savings constraints that keep students from proceeding to the next grade. The second variation also

lowered the regular cash transfer payment, but instead of transferring the accumulated funds prior to

the new school year it provided a large bonus for graduating from secondary school. This bonus

was provided earlier to students who enrolled for tertiary education (upon graduation) than students

who did not. The effect of the first variation on child work appears to be similar to that of the

simple cash transfer scheme. The second variation, however, resulted in a more pronounced

reduction in child work, in particular for graduating pupils (who were also markedly more likely to

enroll for tertiary education).

        Glewe and Olinto (2004) and Galiani and McEwan (2013) compare the impact of receiving

Honduras’ PRAF-II conditional cash transfers in isolation (discussed in more detail above) to the

impact of receiving the conditional cash transfer in combination with direct investments in the

communities’ health and education facilities. As shown in Figure 3, neither of these studies found

an overall significant reduction in child labor as a result of the pure conditional cash transfer

scheme (although both found effects for the poorest households). Glewe and Olinto (2004) also find

no overall effect of the cash transfers combined with supply-side interventions. Galiani and
                                                   21
McEwan (2013), however, find that the conditional cash transfers in combination with investments

in health and education facilities did result in a statistically significant decrease in economic

activities outside the household and household chores. 45

       Finally, Yap, Sedlacek, and Orazem (2002) investigate the impact of Brazil’s Programa de

Erradicacao de Trabalho Infantil (PETI). This program was targeted at poor households in rural

areas of the country where the prevalence rates of child labor are high. Similar to Brazil’s Bolsa

Escola program, PETI provided a cash transfer conditional on school participation. However, PETI

was explicitly designed to reduce child labor. For this purpose, it required households to enroll their

child in an after-school education program. The content of the after-school education program

differed by community and could include academic and physical education components. After-

school education essentially doubled the length of the school day for participating children. The

authors find that PETI reduced child labor by 5 to 25 percentage points in different regions. 46 Taken

at face value, these results suggest that the program resulted in strong reductions in child labor

compared to, for example, Bolsa Escola, which provides pure conditional cash transfers and

reduced participation in economic activities by 8.7 percentage points in rural areas (Ferro, Kassouf,

and Levison 2010).

       Taken together, the studies discussed in this section suggest that the impact of conditional

cash transfers depends partly on their integration with other interventions. Interventions that aim to

improve income-generating activities may reduce the impact of conditional cash transfers on child

labor, possibly by generating increased demand for children’s time within the household. In the case

of Nicaragua, for example, a plausible explanation for the weaker program effect when the basic

conditional cash transfer is combined with a grant for productive investments could be that children

are employed in the newly developed household business.

       Combining conditional cash transfers with supply-side interventions such as the provision of

health and education facilities and after-school education may increase impacts on child work. This

finding seems intuitive because these supply-side interventions reduce the incentives or the time
                                                  22
available to the child for work. However, further research is needed to better understand how cash

transfer and supply-side interventions interact. Is their combined effect simply equal to the sum of

the effect of the individual interventions? Or is there a synergy, such that the effect of the cash

transfers and supply-side interventions is mutually reinforcing? This is a key topic for future

research, not only for child labor but also for outcomes such as school participation.


<<A>> 4. Conclusion


        Cash transfer schemes are not often designed and implemented with the aim of reducing

child labor, although social protection is recognized as one of the main instruments to address child

labor (International Labor Office 2013b; Understanding Children’s Work 2010). However, as this

review shows, cash transfers have a strong potential to address child labor. We have not identified a

single program that increased child labor (although indications of negative spillover effects warrant

further research). On the contrary, there is broad evidence that cash transfers, conditional and

unconditional, lower both the extensive and intensive margin of child labor. Moreover, cash

transfers appear to cushion the effect of economic shocks that may lead households to use child

labor as a coping strategy.

        The effects of the cash transfer programs on child labor are heterogeneous. There are

differences by the child’s gender. Boys tend to experience a larger reduction in participation in

economic activities, whereas girls experience relatively larger reductions in involvement in

household chores. Moreover, reductions in child labor are particularly pronounced when

beneficiaries are poor, signaling that the mitigation of credit constraints, which force households to

use child labor as a consumption smoothing mechanism, are a key channel through which cash

transfers lower child labor. This interpretation is confirmed by studies of old-age pension schemes

that allow for a more direct test of the role of credit constraints.

        Preliminary evidence suggests that the impact of conditional cash transfers depends partly

on their integration with other interventions. Combining conditional cash transfers with supply-side
                                                    23
interventions such as the provision of health and education facilities and/or after-school education

may increase impacts on child labor. In contrast, interventions that aim to improve income-

generating activities may reduce the impact of conditional cash transfers on child labor, possibly by

generating increased demand for children’s time within the household. These results indicate that to

reduce child labor, careful attention should be paid to the integration of conditional cash transfers

with other interventions.

       Our findings are in accordance with Edmonds (2007) and Fiszbein and Schady (2009) who

also find that cash transfer programs are generally a promising tool to reduce child labor, drawing

on a subset of the studies discussed in this paper. 47 In conclusion, the use of cash transfers as an

anti-poverty strategy seems to be effective to reduce child labor. The same does not necessarily hold

for all anti-poverty and income-generating interventions. The encouragement of entrepreneurship

through microfinance programs, for instance, can increase child labor (e.g., Augsburg et al. 2012;

Nelson 2011). This finding further underlines that cash transfers are a useful policy instrument to

improve child welfare and suggests that they are unlikely to have detrimental effects on child labor,

even when they are not implemented or designed to address it.

       However, important knowledge gaps must be addressed to provide more detailed policy

advice. The main gaps are because, as mentioned, cash transfers are seldom implemented with a

reduction in child labor as one of their main objectives and are therefore typically not assessed in

depth against this outcome. As a result, we know relatively little about the program characteristics

that determine cash transfer programs’ effects on child labor, and we do not clearly understand why

some of these programs have no effect on child labor. The role of design elements that have been

tested appears to be limited. There is little evidence that schooling conditions affect program impact

on child labor. The latter finding is surprising in light of recent research indicating that schooling

conditions matter for school participation, and this finding warrants further research. The size of the

transfer relative to household income also appears to have little influence on reductions in child

labor. Some conditional cash transfer projects that transfer substantial sums of money have no
                                                  24
effect on child labor, whereas other programs that provide only a small subsidy result in strong

changes.

       Beyond examining the program characteristics that determine the impact of cash transfers

on child labor, there are more questions open for future research. A key issue, which we outlined in

the introduction, is the measurement of child labor. Most impact evaluations focus on economic

activities. This approach potentially results in the underreporting of program impact on activities

performed by girls because they are more likely to be involved in household chores. Additionally,

as a result of the focus on participation in economic activities (or in one of its subcomponents), we

have little evidence on the extent to which the interventions prevent and reduce (i) the worst forms

of child labor, including hazardous work, and (ii) long working hours that keep children from

learning in school.

       More systematic evidence on extensions of basic cash transfer schemes and the interplay

between cash transfer schemes and other social protection and supply-side interventions would also

be important. In practice, cash transfer interventions are rarely implemented in isolation, and

interaction effects may well determine their impact. This review provides preliminary evidence

suggesting that the impact of cash transfer schemes and education interventions may be mutually

reinforcing, whereas combining cash transfer schemes with services that aim to foster income-

generating activities may have a detrimental effect on child labor. However, much work remains to

be done to understand the extent to which synergy effects drive the combined effect of cash transfer

and supply-side interventions.

       Finally, we know very little about the relative cost effectiveness of cash transfers in

reducing child labor and how their cost effectiveness compares to other interventions. This issue has

remained largely unexplored in most impact evaluations focusing on child labor. Few of the studies

included in our review are explicit on the cost of implementing a cash transfer program (other than

the transfer amounts). Evaluations of the impact of other categories of interventions on child labor


                                                 25
are equally unlikely to discuss the cost of implementing the program per beneficiary. More explicit

cost effectiveness analyses will be crucial in guiding governments in the elimination of child labor.




                                                 26
          Notes

           1
             Previous work discussing evidence on the relationship between cash transfers and child labor, building on the earlier
papers included also in this review, includes Edmonds (2007) and Fiszbein and Schady (2009). Edmonds (2007) also discusses the
effect of a range of other interventions on child labor. Fiszbein and Schady (2009) also examine the effects of conditional cash
transfers on various of other outcomes (including adult labor).
           2
             Over the past decade, developing country governments have begun to adopt conditional cash transfers as social protection
instruments at a rapid pace. See Fiszbein and Schady (2009) for a comprehensive review of the recent proliferation of conditional
cash transfer schemes in developing countries.
           3
             The three principal international conventions on child labor recognize this complexity and set the legal boundaries that
define children’s work that is targeted for elimination. The ILO Convention No. 138 determines the minimum age below which
children should not work and the minimum ages for light and hazardous work. The ILO Convention No. 182 specifies the worst
forms of child labor that are prohibited for all children under the age of 18. The United Nations Convention on the Rights of the
Child aims to protect children from economic exploitation and from performing any work that is likely to be hazardous or to interfere
with the child's education or to be harmful to the child's health or physical, mental, spiritual, moral, or social development. The
international legal standards contain a number of flexibility clauses left to the discretion of the competent national authority in
consultation (where relevant) with worker and employer organizations (e.g., minimum ages, scope of application). There is no single
legal definition of child labor across countries, and concomitantly, there is no single standard statistical measure of child labor
consistent with national legislation across countries. (Text in this footnote is adapted from standard Understanding Children's Work
description.)
           4
              Moreover, we must keep in mind that schooling, work, and other activities, including leisure and sleep, are jointly
determined.
           5
             There is also variation in the reference period. Some studies consider work in the seven days prior to the household
survey, whereas some studies examine work in the past 12 months. Finally, some studies present results for a few separate categories
of activities instead of focusing on a comprehensive indicator for participation in work.
           6
             Initially, most evaluations of the impact of cash transfer interventions on human capital accumulation focused on self-
reported school enrollment or attendance. A new generation of studies focuses on more elaborate outcomes, such as performance on
standardized tests. A similar development has not taken place for child labor, and we hope that this review can serve as a starting
point.
           7
             For a formal model, see for instance Ravallion and Wodon (2000).
           8
             For a theoretical discussion of the effects of cash transfers on school participation see, inter alia, Fiszbein and Schady
(2009).
           9
             If leisure is a normal good.
           10
              For a discussion of investment of cash transfers in productive assets, see Fiszbein and Schady (2009).
           11
               The full literature search was conducted in early 2011 by Understanding Children’s Work (UCW) to build a
comprehensive database of child labor impact evaluations. The database is updated regularly and can be found at http://www.ucw-
project.org/impact-evaluation/inventory-impact-evaluations.aspx. We considered works identified to the end of 2012. We considered
only papers written in English and did not consider dissertation chapters that did not appear as separate (working) papers. Of course,
relevant impact evaluations continue to appear after this date. Authors of relevant papers not included in this database are invited to
share their papers.
           12
              We exclude simulation studies, such as Bourguignon, Ferreira, and Leite (2003), studies that capture the effect of
multiple programs at once, such as Cardoso and Portela Souza (2003), and studies of in-kind transfers, such as Ravallion and Wodon
(2000) and Kazianga, de Walque, and Alderman (2009).
           13
              We do not discuss Rubio-Codina (2010) and Schultz (2004) in the main text because these studies rely on the same data
and find results comparable to those of Skoufias and Parker (2001).
           14
              A table with the exact definition of the outcome variable for each individual study is available on request.
           15
              Figure 2 does not include the dose-response estimates for South Africa’s Child Support Grant (DSD, SASSA, and
UNICEF 2012) discussed in the text, which differ in nature from the “binary” treatment results discussed in the remainder of this
paper.
           16
              This result should be interpreted with some care because the authors do not show whether differences between
individuals treated from an early age and those treated from the age of 14 are significant.
           17
              The program did not affect children’s participation in care for other children in the household, care for adults in the
household, or work outside the household for income (other than domestic work). The estimates are not always robust for different
specifications.
           18
              That is, -0.6 hours for children living in a household with a male pensioner and -0.5 hours for children living in a
household with a female pensioner.
           19
              Edmonds (2006) discusses several alternative hypotheses that could explain the observed changes in child labor and
school participation in response to the realization of anticipated income but concludes that credit constraints are the most plausible
explanation.
           20
              To correct for age-specific trends not related to the reform, the author compares difference-in-differences estimates for
rural households to difference-in-differences estimates for urban households (which were not affected by the reform) in a triple-
difference framework.
           21
              To calculate the standard errors associated with these estimates, we assume that the covariance between the individual
estimates is zero. Note that the overall impact of the program may not be statistically significant even if some of the underlying
estimates for age and gender subgroups are statistically significant.
                                                                 27
           22
              If we have an estimate of average program effects from one study and estimates of disaggregated program effects from
another study, we show only the estimate of the average program effect.
           23
              Although it seems reasonable to take the unweighted means of impact estimates across age and gender groups, the same
procedure cannot be applied across urban and rural areas that may differ drastically in terms of population size. When we have
estimates for both rural and urban areas for the same program this is indicated after the program name.
           24
              We examined whether the average the impact estimates of randomized controlled trials are different from the impact
estimates of quasi-experimental studies and find that this is not the case. This finding is interesting in light of the ongoing debates
regarding the validity of randomized evaluations vis-à-vis quasi-experimental approaches (Deaton 2010; Duflo and Kremer 2005;
and Ravallion 2009).
           25
              To ensure comparability, we multiply the hours worked by seven for studies that consider daily working hours.
           26
              To correct for the censoring of the outcome variable, Attanasio et al. (2010), Dammert (2009), Del Carpio and Loayza
(2012), and Ferreira, Filmer, and Schady (2009) use Tobit regressions. Skoufias and Parker (2001) use a Heckit procedure.
           27
              Bando, Lopez-Calva, and Patrinos (2005) examine whether the impact of Mexico’s Oportunidades was stronger for
indigenous households.
           28
              See Fiszbein and Schady (2009) for a further discussion.
           29
              In her paper, Dammert shows how impact estimates for the richest four quintiles differ from the impact estimate for the
poorest quintile. To ensure that the results are comparable with those of the other studies, we imputed the impact of the program for
each of the five quintiles. We followed a procedure similar to that discussed in footnote 21 to calculate standard errors for these
estimates. Dammert (2009) also examines heterogeneity along household per capita expenditure quintiles. These results are not
qualitatively different from the marginality index results; therefore, we decided not to present them in Figure 5.
           30
              For each study, we focus on two age groups, within the 7-14 age range, if possible.
           31
              The effects of South Africa’s (unconditional) Child Grant also appear to be stronger for adolescent girls than for boys
(DSD, SASSA, and UNICEF 2012).
           32
              For Galiani and McEwan (2013) this result refers to the two poorest quintiles.
           33
              Making purchases for the family, making clothes for family members, taking a family member to school, work, the
health center, or the hospital, cleaning the house, washing and ironing clothes, cooking, fetching water or firewood, disposing of
trash, and caring for small children, elderly family members, or sick individuals.
           34
              Moreover, Dubois and Rubio-Codina (2012) find that Oportunidades lowered teenage girls’ participation in care for
younger siblings.
           35
              Not displayed in Figure 6 because the study focuses on the intensive instead of extensive margin of child labor.
           36
              For this set of estimates, only the differential impact of the intervention on boys versus girls is given. The impact on boys
and girls themselves is not available.
           37
              These estimates do not correct for truncation of the outcome variable, but other estimates provided in the paper suggest
that most results are robust to corrections for censoring.
           38
              A similar argument underlies the well-known theoretical work of Basu and Van (1998), who argue that pulling all
children out of work through a ban may jolt the labor market to another equilibrium in which adult wages are higher and children do
not work.
           39
              Galiani and McEwan (2013) find some evidence of reductions in child work in the poorest quintile.
           40
              We do not consider the older age groups also discussed in the paper.
           41
               Because the results are preliminary, we have not included them in our discussion of the impact of unconditional and
conditional cash transfers on the intensive margin of child labor.
           42
              Amounts partly based on figures provided in Fiszbein and Schady (2009).
           43
              We have no estimate for the amount transferred as a percentage of household income for the Female School Stipends
program in Pakistan.
           44
              Several other studies do not directly test whether conditional cash transfers protect children from participation in work
when the household is hit by an economic shock. However, they do suggest that conditional cash transfers can reduce child labor
during economic downturns. Maluccio (2005), for instance, shows that Red de Protección Social reduced participation in economic
activities among children living in Nicaragua’s coffee-growing regions during a sharp downturn in coffee prices in 2001 and 2002.
Sparrow (2007) finds similar results in his study of Indonesia’s Jaringan Pengaman Social emergency conditional cash transfer
program. However, these results contrast with those of Amarante, Ferrando, and Vigorito (2011), who find no effect of Uruguay’s
anti-crisis conditional cash transfer program on child labor.
           45
              The point estimates for conditional cash transfers in combination with investments in health and education facilities
exceed those of conditional cash transfers only, but the estimated coefficients are not significantly different from each other.
           46
              Moreover, there is some evidence of spillover effects at the locality level. The likelihood of working decreased for
children from non-program households in treatment localities. However, for these same children the likelihood of working long hours
(10 or more per week) increased.
           47
              Fiszbein and Schady (2009) also caution that potential detrimental spillover effects within households deserve further
scrutiny.




                                                                   28
References

Alam, Andaleeb, Javier E. Baez, and Ximena V. Del Carpio. 2011. “Does Cash for School

       Influence Young Women’s Behavior in the Longer Term? Evidence from Pakistan.” World

       Bank Policy Research Working Paper 5669.

Amarante, Verónica, Mery Ferrando, and Andrea Vigorito. 2011. “School Attendance, Child Labor,

       and Cash Transfers. An Impact Evaluation of PANES.” PIERI Working Paper 2011-22.

Attanasio, Orazio, Emla Fitzsimons, Ana Gomez, Martha Isabel Gutiérrez, Costas Meghir, and

       Alice Mesnard. 2010. “Children’s Schooling and Work in the Presence of a Conditional

       Cash Transfer Program in Rural Colombia.” Economic Development and Cultural Change

       58 (2): 181–210.

Augsburg, Britta, Ralph de Haas, Heike Harmgart, and Costas Meghir. 2012. “Microfinance,

       Poverty, and Education.” NBER Working Paper 18538.

Baird, Sarah, Craig McIntosh, and Berk Özler. 2011. “Cash or Condition? Evidence from a Cash

       Transfer Experiment.” Quarterly Journal of Economics 126 (4): 1709–53.

Baird, Sarah, Francisco H. G. Ferreira, Berk Özler, and Michael Woolcock. 2013. “Relative

       Effectiveness of Conditional and Unconditional Cash Transfers for Schooling Outcomes in

       Developing Countries: A Systematic Review.” Campbell Systematic Reviews 2013:8.

Barrera-Osorio, Felipe, Marianne Bertrand, Leigh L. Linden, and Francisco Perez-Calle. 2008.

       “Conditional Cash Transfers in Education: Design Features, Peer and Sibling Effects.

       Evidence from a Randomized Experiment in Colombia.” World Bank Policy Research

       Working Paper 4580.

Barrera-Osorio, Felipe, Marianne Bertrand, Leigh L. Linden, and Francisco Perez-Calle. 2011.

       "Improving the Design of Conditional Transfer Programs: Evidence from a Randomized

       Education Experiment in Colombia." American Economic Journal: Applied Economics 3

       (2): 167-95.


                                               29
Bando, Rosangela, Luis F. Lopez-Calva, and Harry Anthony Patrinos. 2005. “Child Labor, School

       Attendance, and Indigenous Households: Evidence from Mexico.” World Bank Policy

       Research Working Paper 3487.

Basu, Kaushik, and Pham Hoang Van. 1998. “The Economics of Child Labor.” American Economic

       Review 88 (3): 412–27.

Beegle, Kathleen, Rajeev H. Dehejia, and Roberta Gatti. 2006. “Child Labour and Agricultural

       Shocks.” Journal of Development Economics 81 (1): 80–96.

Behrman, Jere R., Jorge Gallardo-García, Susan W. Parker, Petra E. Todd, and Viviana Vélez-

       Grajales. 2012. “Are Conditional Cash Transfers Effective in Urban Areas? Evidence from

       Mexico.” Education Economics 20 (3): 233–59.

Behrman, Jere R., Susan W. Parker, and Petra E. Todd. 2011. “Do Conditional Cash Transfers for

       Schooling Generate Lasting Benefits? A Five-Year Followup of

       PROGRESA/Oportunidades.” Journal of Human Resources 46 (1): 93–122.

Benhassine, Najy, Florencia Devoto, Esther Duflo, Pascaline Dupas, and Victor Pouliquen. 2012.

       “Unpacking the Effects of Conditional Cash Transfer Programs: Experimental evidence

       from Morocco.” Unpublished Manuscript.

Borraz, Fernando, and Nicolás González. 2009. “Impact of the Uruguayan Conditional Cash

       Transfer Program.” Cuadernos de Economía 46 (Novembre): 243–71.

Bourguignon, François, Francisco. H. G. Ferreira, and Phillippe. G. Leite. 2003. “Conditional Cash

       Transfers, Schooling, and Child Labor: Micro Simulating Brazil’s Bolsa Escola Program.”

       World Bank Economic Review 17 (2), 229–54.

Buddelmeyer, Hielke, and Emmanuel Skoufias. 2004. “An Evaluation of the Performance of

       Regression Discontinuity Design on PROGRESA.” IZA Discussion Paper 827 and World

       Bank Policy Research Working Paper 3386.

Cardoso, Eliana, and André Portela Souza. 2003. “The Impact of Cash Transfers on Child Labor

       and School Attendance in Brazil.” Working Paper.
                                                30
Cigno, Alessandro, and Furio. C. Rosati. 2005. The Economics of Child Labour. Oxford University

       Press.

Covarrubias, Katia, Benjamin Davis, and Paul Winters. 2012. “From Protection to Production:

       Productive Impacts of the Malawi Social Cash Transfer Scheme.” Journal of Development

       Effectiveness 4 (1): 50–77.

Dammert, Ana C. 2009. “Heterogeneous Impacts of Conditional Cash Transfers: Evidence from

       Nicaragua.” Economic Development and Cultural Change 58 (1): 53–83.

Dubois, Pierre, and Marta Rubio-Codina. 2012. “Child Care Provision: Semiparametric Evidence

       from a Randomized Experiment in Rural Mexico.” Annales d'Economie et de Statistique

       105/106: 155–84.

Deaton, Angus. 2010. “Instruments, Randomization, and Learning about Development.” Journal of

       Economic Literature 48 (2): 424–55.

De Carvalho Filho, Irineu Evangelista. 2012. “Household Income as Determinant of Child Labor

       and School Enrollment in Brazil: Evidence from a Social Security Reform.” Economic

       Development and Cultural Change 60 (2): 339–435.

De Hoop, Jacobus, and Furio C. Rosati. 2014. “Does Promoting School Attendance Reduce Child

       Labour? Evidence from Burkina Faso’s BRIGHT Project.” Economics of Education Review

       39: 78-96.

De Janvry, Alain, Frederico Finan, Elisabeth Sadoulet, and Renos Vakis. 2006. “Can Conditional

       Cash Transfer Programs Serve as Safety Nets in Keeping Children at School and from

       Working when Exposed to Shocks?” Journal of Development Economics 79 (2): 349–73.

Del Carpio, Ximena V., and Norman V. Loayza. 2012. “The Impact of Wealth on the Amount and

       Quality of Child Labour.” World Bank Policy Research Working Paper 5959.

Del Carpio, Ximena V., and Karen Macours. 2010. “Leveling the Intra-household Playing Field:

       Compensation and Specialization in Child Labour Allocation.” In Child Labour and the

       Transition between School and Work (Research in Labour Economics, Volume 31), ed.
                                               31
       Randall K.Q. Akee, Eric V. Edmonds, and Konstantinos Tatsiramos, 259–96. Bingley:

       Emerald Group Publishing Limited.

DSD, SASSA, and UNICEF. 2012. The South African Child Support Grant Impact Assessment:

       Evidence from a survey of children, adolescents and their households. Pretoria: UNICEF

       South Africa.

Duflo, Esther, and Michael Kremer. 2005. “Use of Randomization in the Evaluation of

       Development Effectiveness." In Evaluating Development Effectiveness, ed. George K.

       Pitman, Osvaldo N. Feinstein, Gregory K. Ingram, 205–39. New Brunswick: Transaction

       Publishers.

Duryea, Suzanne, David Lam, and Deborah Levison. 2007. “Effects of Economic Shocks on

       Children’s Employment and Schooling.” Journal of Development Economics 84 (1): 188–

       214.

Edmonds, Eric V. 2006. “Child Labour and Schooling Responses to Anticipated Income in South

       Africa.” Journal of Development Economics 81 (2): 386–414.

Edmonds, Eric V. 2007. “Child Labour.” In Handbook of Development Economics, Volume 4, ed.

       T. P. Schultz, J. Strauss, 3607–709. Amsterdam: Elsevier Science.

Edmonds, Eric V., and Norbert Schady. 2012. “Poverty Alleviation and Child Labor.” American

       Economic Journal: Economic Policy 4 (4): 100–24.

Ferreira, Francisco H. G., Deon Filmer, and Norbert Schady. 2009. “Own and Sibling Effects of

       Conditional Cash Transfer Programs.” World Bank Policy Research Working Paper 5001.

Ferro, Andrea R., Ana Lucia Kassouf, and Deborah Levison. 2010. “The impact of conditional cash

       transfer programs on household work decisions in Brazil.” In Child Labour and the

       Transition between School and Work (Research in Labour Economics, Volume 31), ed.

       Randall K.Q. Akee, Eric V. Edmonds, and Konstantinos Tatsiramos, 193–218. Bingley:

       Emerald Group Publishing Limited,.


                                               32
Fiszbein, Ariel, and Norbert Schady. 2009. Conditional Cash Transfers: Reducing Present and

       Future Poverty. The World Bank, Washington D.C.

Fitzsimons, Emla, and Alice Mesnard. Forthcoming. “Can Conditional Cash Transfers Compensate

       for a Father’s Absence?” World Bank Economic Review.

Galiani, Sebastian, and Patrick J. McEwan. 2013. “The Heterogeneous Impact of Conditional Cash

       Transfers.” Journal of Public Economics 103: 85–96.

Gee, Kevin A. 2010. “Reducing Child Labour through Conditional Cash Transfers: Evidence from

       Nicaragua’s Red de Proteccion Social.” Development Policy Review 28 (6): 711–32.

Glewwe, Paul, and Pedro Olinto. 2004. “Evaluation of the Impact of Conditional Cash Transfers on

       Schooling: An Experimental Analysis of Honduras’ PRAF Program.” Working Paper.

Guarcello, Lorenzo, Fabrizia Mealli, and Furio C. Rosati. 2010. “Household Vulnerability and

       Child Labor: The Effect of Shocks, Credit Rationing, and Insurance.” Journal of Population

       Economics 23 (1): 169–98.

International Labour Office. 2013a. Marking Progress Against Child Labour: Global Estimates and

       Trends 2000-2012. Geneva.

International Labour Office. 2013b. World Report on Child Labour: Economic Vulnerability, Social

       Protection and the Fight against Child Labour. Geneva.

Kazianga, Harounan, Damien de Walque, and Harold Alderman. 2009. “Educational and Health

       Impacts of Two School Feeding Schemes. Evidence from a Randomized Trial in Burkina

       Faso.” World Bank Policy Research Working Paper 4976.

King, Elizabeth M., and Jere R. Behrman. 2009. “Timing and Duration of Exposure in Evaluations

       of Social Programs.” World Bank Research Observer 24 (1): 55–82.

Kremer, Michael, and Alaka Holla. 2009. “Pricing and Access: Lessons from Randomized

       Evaluations in Education and Health.” In What Works in Development: Thinking Big and

       Thinking Small, ed. Jessica Cohen, William Easterly, 91–129. Washington, DC: The

       Brookings Institution.
                                               33
Levy, Dan, and Jim Ohls. 2007. “Evaluation of Jamaica’s PATH Program.” Mathematica Policy

       Research Report.

Maluccio, John A. 2005. “Coping with the “Coffee Crisis” in Central America: The Role of the

       Nicaraguan Red de Protección Social.” FCND Discussion Paper 188.

Maluccio, John A., and Rafael Flores. 2005. “Impact Evaluation of a Conditional Cash Transfer

       Program: The Nicaraguan Red de Protección Social” International Food Policy Research

       Institute Research Report 141.

Miller, Candace, and Maxton Tsoka. 2012. “Cash Transfers and Children’s Education and Labour

       among Malawi’s Poor.” Development Policy Review 30 (4), 499–522.

Nelson, Leah K. 2011. “From Loans to Labour: Access to Credit, Entrepreneurship, and Child

       Labor.” Working Paper.

Ravallion, Martin 2009. “Evaluation in the Practice of Development.” World Bank Research

       Observer 24 (1): 29–53.

Ravallion, Martin, and Quentin Wodon. 2000. “Does Child Labour Displace Schooling? Evidence

       on Behavioural Responses to an Enrollment Subsidy.” Economic Journal 110 (March):

       C158–C175.

Rawlings, Laura B., and Gloria M. Rubio. 2005. “Evaluating the Impact of Conditional Cash

       Transfer Programs.” World Bank Research Observer 20 (1): 29–55.

Rubio-Codina, Marta. 2010. “Intra-household Time Allocation in Rural Mexico: Evidence from a

       Randomized Experiment.” In Child Labour and the Transition between School and Work

       (Research in Labour Economics, Volume 31), ed. Randall K.Q. Akee, Eric V. Edmonds, and

       Konstantinos Tatsiramos, 219–57. Bingley: Emerald Group Publishing Limited.

Saavedra, Juan Esteban, and Sandra Garcia. 2012. “Impact of Conditional Cash Transfer Programs

       on Educational Outcomes in Developing Countries: A Meta Analysis.” Rand Working Paper

       WR-921-1.


                                               34
Schady, Norbert R., and Maria Caridad Araujo. 2006. “Cash Transfers, Conditions, School

       Enrollment, and Child Work: Evidence from a Randomized Cash Transfer Experiment in

       Ecuador.” World Bank Policy Research Working Paper 3930.

Schultz, T. Paul. 2004. “School Subsidies for the Poor: Evaluating the Mexican Progresa Poverty

       Program.” Journal of Development Economics 74 (1): 199–250.

Skoufias, Emmanuel, and Susan W. Parker. 2001. “Conditional Cash Transfers and their Impact on

       Child Work and Schooling: Evidence from the PROGRESA Program in Mexico.” Economía

       2 (1): 45–96.

Sparrow, Robert. 2007. “Protecting Education for the Poor in Times of Crisis: An Evaluation of a

       Scholarship Programme in Indonesia.” Oxford Bulletin of Economics and Statistics 69 (1):

       99–122.

Udry, Christopher. 2006. “Child Labor.” In Understanding Poverty, ed. Abhijit Vinayak Banerjee,

       Roland Bénabou, and Dilip Mookherjee, 243–58. Oxford: Oxford University Press.

Understanding Children’s Work. 2010. Joining Forces against Child Labour: Inter-agency Report

       for The Hague Global Child Labour Conference of 2010. Geneva.

Yap, Yoon-Tien, Guilherme Sedlacek, and Peter F. Orazem. 2002. “Limiting Child Labor through

       Bahavior-Based Income Transers: An Experimental Evaluation of the PETI Program in

       Rural Brazil.” Working Paper.




                                                35
Tables and Figures

                                     Figure 1. Changes in the budget constraint when the transfer affects school participation.




Source: Authors

Notes: Household consumption is plotted on the vertical axis and child leisure is plotted on the horizontal axis. The solid line ABC represents the budget constraint for a
household that does not receive a transfer and does not send its child to school. The dotted and dashed lines represent the budget constraints for households that do send their
child to school and (i) receive no cash transfer (DEF), (ii) receive a cash transfer smaller than the cost of education (GHF), and (iii) receive a cash transfer that exceeds the cost of
education (IJF). The segments AD, DG, and DI respectively represent the monitary cost of education, a transfer smaller than the cost of education and a transfer exceeding the
cost of education. The segment FC represents the fixed amount of time a child spends in school.
                                                                                          36
                              Figure 2. Unconditional cash transfers tend to reduce child labor, although results are not uniform.

                             0.150
                                                                                                                    0.110***
                             0.100                                      0.077**                           0.080*

                             0.050                                                  0.021
                             0.000

                            -0.050

                            -0.100                                                            -0.077***
                                                             -0.099*                                                                       -0.100***
                            -0.150                                                                                             -0.120***

                            -0.200   -0.172***

                            -0.250
                                                 -0.245***
                            -0.300
                                     Economic Economic Economic Household Work in        Paid Household Household Work for Work for
                                      activities activities activities chores  family domestic chores,   chores, pay, boys pay, girls
                                     for pay or for pay or for pay, IV        farm or    work    boys     girls
                                         for        for                        family   outside
                                     household, household,                    business household
                                         IV         IV
                                     4. Schady     5. Edmonds and         2. Covarrubias, Davis, and             3. Miller and Tsoka (2012)
                                         and        Schady (2011)              Winters (2012)
                                      Araujo
                                       (2006)
                                       Bono de Desarrollo Humano,                           Social Cash Transfer Scheme, Malawi
                                                Ecuador

Source: Authors

Note: Change in the probability of working as a result of the unconditional cash transfer programs displayed on the horizontal axis. *** p<0.01, ** p<0.05, * p<0.1.




                                                                                       37
                                                     -0.12
                                                                         -0.10
                                                                                        -0.08
                                                                                                         -0.06
                                                                                                                          -0.04
                                                                                                                                                         -0.02
                                                                                                                                                                                        0.00
                                                                                                                                                                                                0.02

              10. CESSP Scholarship Program,
              Cambodia (Ferreira et al. 2009) ǂ




                                                             -0.104***
       8. Bolsa Escola, rural Brazil (Ferro et al.
                         2010)




                                                                            -0.087***
              22. Oportunidades, urban Mexico
                   (Behrman et al. 2012) ǂ

       25. Red de Proteccion Social, Nicaragua                                          -0.067**
             (Maluccio and Flores 2005)
                                                                                                -0.056**



           16. Jaringan Pengaman Social, rural
                Indonesia (Sparrow 2004)
                                                                                                       -0.051***




          28. Female School Stipends, Pakistan
                   (Alam et al. 2011)
                                                                                                               -0.049**




      8. Bolsa Escola, urban Brazil (Ferro et al.
                        2010)
                                                                                                                                  -0.025***




          16. Jaringan Pengaman Social, urban




38
                Indonesia (Sparrow 2004)
                                                                                                                                          -0.014**




          15. PRAF-II, Honduras (Glewwe and
                     Olinto 2004)
                                                                                                            -0.049




           14. PRAF-II, Honduras (Galiani and
                    McEwan 2013)

       11. Familias en Accion, urban Colombia
              (Attanasio et al. 2010) ǂ
                                                                                                                                  -0.024 -0.023




     20. Oportunidades, rural Mexico (Skoufias
                and Parker 2001) ǂ
                                                                                                                                                -0.020




        11. Familias en Accion, rural Colombia
                                                                                                                                                                                                       Figure 3. Conditional cash transfer programs tend to reduce the prevalence of child labor.




               (Attanasio et al. 2010) ǂ

        13. Subsidios Cond. a la Asistencia
      Escolar, Colombia (Barrera-Osorio et al.,…

       17. Path, Jamaica (Levy and Ohls 2007)
                                                                                                                                                                 -0.004 -0.003 -0.001




         29. PANES, Uruguay (Amarante et al.
                                                                                                                                                                                               0.014




                       2011)
Source: Authors

Note: Change in the probability of working as a result of the conditional cash transfer programs displayed on the horizontal axis. ǂ indicates that the estimate is a weighted average
of multiple age and gender groups. To minimize text on the horizontal axis, we only display the first author of the study if the study has more than 2 authors. *** p<0.01, **
p<0.05, * p<0.1.




                                                                                         39
                                 Figure 4. Conditional cash transfer programs tend to reduce weekly hours worked by children.

                               0.50
                                                                                                                                         0.07
                               0.00

                               -0.50

                               -1.00
                                                                                                                      -0.91
                               -1.50                                                              -1.19***
                                                                                 -1.54*
                               -2.00                         -1.76***

                               -2.50

                               -3.00

                               -3.50

                               -4.00       -3.65***
                                          27. Red de      24. Atencion a 11. Familias en         10. CESSP              20.        11. Familias en
                                          Proteccion    Crisis, Nicaragua Accion, urban         Scholarship       Oportunidades,    Accion, rural
                                       Social, Nicaragua (Del Carpio and   Colombia               Program,         rural Mexico       Colombia
                                          (Gee 2010)      Loayza 2012) (Attanasio et al.         Cambodia         (Skoufias and    (Attanasio et al.
                                                                             2010) ǂ           (Ferreira et al.   Parker 2001) ǂ       2010) ǂ
                                                                                                   2009) ǂ

Source: Authors

Note: Change in hours worked as a result of the conditional cash transfer programs displayed on the horizontal axis. ǂ indicates that the estimate is a weighted average of multiple
age and gender groups. To minimize text on the horizontal axis, we only display the first author of the study if the study has more than 2 authors. *** p<0.01, ** p<0.05, * p<0.1.




                                                                                          40
              Figure 5. The impact of conditional cash transfers on child labor tends to be stronger in poorer households and communities.

                                0.020
                                                                                                                       0.001                                                                                                                                                                                0.003
                                0.000
                               -0.020                                                       -0.011                                                -0.009
                               -0.040                                                                                                                                              -0.033***
                               -0.060                                                                                                                                        -0.039***
                                                               -0.050**
                                                                                                                                                                       -0.040***
                                                                                                                                                                                                                                                                                                                                      -0.063
                               -0.080                                                                                                                                                      -0.072
                                        -0.079***                                                                                                                                                                                                                               -0.080
                               -0.100
                               -0.120
                               -0.140                                                                                                                                                                                                                                                                                                                             -0.127


                                                                                              Third poorest quintile




                                                                                                                                                                                                                                                                                                             Third poorest quintile
                                                                  Second poorest quintile




                                                                                                                                                                                                                                                                                  Second poorest quintile
                                            Poorest quintile




                                                                                                                                                    Richest quintile

                                                                                                                                                                          Poorest 1-25 percentile

                                                                                                                                                                                                         Poorest 25-50 percentile
                                                                                                                        Fourth poorest quintile




                                                                                                                                                                                                                                    Richest 50 percentiles

                                                                                                                                                                                                                                                             Poorest quintile




                                                                                                                                                                                                                                                                                                                                                                    Richest quintile
                                                                                                                                                                                                                                                                                                                                        Fourth poorest quintile
                                           14. PRAF, Honduras (Galiani                                                                                                     16. Jaringan                                                                      26. Red de Proteccion Social,
                                                and McEwan 2013)                                                                                                         Pengaman Social,                                                                     Nicaragua (Dammert 2009)
                                                                                                                                                                        Indonesia (Sparrow
                                                                                                                                                                              2004)

Source: Authors

Note: Change in the probability of working as a result of the conditional cash transfer programs displayed on the horizontal axis by poverty quintile or percentile. *** p<0.01, **
p<0.05, * p<0.1.

                                                                                                                                                                                                    41
                              Figure 6. Boys experience stronger reductions in economic activities, girls in household activities.

                               0.100
                                                                                                                                                                                                                                                                                                                                          0.074**
                                                                                                                                                                                                                                                                                                 0.043
                               0.050
                                                                               0.000                                                     0.000
                               0.000
                                       -0.006*                                                                       -0.006
                              -0.050                                                                                                                         -0.020*                                                                                                                                                                                                                                                        -0.023
                                                                                                                                                                  -0.040***                                                                                                                                                                                                                                                       -0.034*

                              -0.100                                                                                                                                                                                                                     -0.088***                                                                                                                                                                                                                        -0.092***
                                                                                                                                                                                                                  -0.120***                                                                                                                                                          -0.111***
                              -0.150


                                                                                Economic activities for pay, girls




                                                                                                                                                                                                                                                            Economic activities for pay, girls
                                           Economic activities for pay, boys




                                                                                                                                                               Household chores, boys




                                                                                                                                                                                                                     Economic activities for pay, boys




                                                                                                                                                                                                                                                                                                                                            Economic activities without pay, girls




                                                                                                                                                                                                                                                                                                                                                                                                                                                                 Household chores, boys
                                                                                                                                                                                                                                                                                                  Economic activities without pay, boys
                                                                                                                                          Farm work, girls




                                                                                                                                                                                                                                                                                                                                                                                       Economic activities for pay or for

                                                                                                                                                                                                                                                                                                                                                                                                                            Economic activities for pay or for
                                                                                                                       Farm work, boys




                                                                                                                                                                                        Household chores, girls




                                                                                                                                                                                                                                                                                                                                                                                                                                                                                             Household chores, girls
                                                                                                                                                                                                                                                                                                                                                                                              household, boys

                                                                                                                                                                                                                                                                                                                                                                                                                                   household, girls
                                          20. Oportunidades, Mexico (Skoufias                                                                                                                                        10. CESSP Scholarship                                                                                                                                               14. PRAF, Honduras
                                                   and Parker 2001)                                                                                                                                                   Program, Cambodia                                                                                                                                                  (poorest and second
                                                                                                                                                                                                                      (Ferreira et al. 2009)                                                                                                                                           poorest quintile) (Galiani
                                                                                                                                                                                                                                                                                                                                                                                          and McEwan 2013)

Source: Authors

Note: Change in the probability of involvement in economic and household activities as a result of the conditional cash transfer programs displayed on the horizontal axis by
gender. To minimize text on the horizontal axis, we only display the first author of the study if the study has more than 2 authors. *** p<0.01, ** p<0.05, * p<0.1.

                                                                                                                                                                                                                    42
                                                                                                   Table 1. Programs and studies discussed in main text
                                                                                                             Published in peer-                                              Main age-
ID Country                  Program                             Studies                                      reviewed journal      Methodology                               range            Activities included in main outcome variable
    (1)                     (2)                                 (3)                                          (4)                   (5)                                       (6)              (7)
Unconditional Cash Transfers
1. Brazil                   Old age pensions                    de Carvalho Filho (2012)                      Yes                       Triple difference                    10-14            Economic activities for pay
2. Malawi                   Social Cash Transfer Scheme         Covarrubias, Davis, and Winters (2012)        Yes                       PSM (see footnote)                   0-17             Multiple underlying activities
3. Malawi                   Social Cash Transfer Scheme         Miller and Tsoka (2012)                       Yes                       Dif in dif (see footnote)            6-18             Economic activities for pay
4. Nicaragua                Bono de Desarollo Humano            Schady and Araujo (2006)                      No                        Household level RCT                  6-17             Economic activities for pay or for household
5. Nicaragua                Bono de Desarollo Humano            Edmonds and Schady (2012)                     Yes                       Household level RCT                  11-16            Economic activities for pay or for household
6. South Africa             Child Support Grant                 DSD, SASSA, and UNICEF (2012)                 No                        PSM Dose response analysis           10 and 15-17     Economic activities for pay or for household
7. South Africa             Old age pensions                    Edmonds (2006)                                Yes                       RDD                                  13-17            Economic activities for pay or for household

Conditional Cash Transfers
8. Brazil                       Bolsa Escola                    Ferro, Kassouf, and Levison (2010)              In edited volume           PSM                                  6-15          Economic activities for pay or for household
9. Brazil                       PETI                            Yap, Sedlacek, and Orazem (2002)                No                         Regression (see footnote)            7-14          No definition of work given
10. Cambodia                    CESSP Scholarship Program Ferreira, Filmer, and Schady (2009)                   No                         RDD                                  7-18          Economic activities for pay
11. Colombia                    Familias en Accion              Attanasio et al. (2010)                         Yes                        Probit and regression (see footnote) 10-17         Economic activities for pay
12. Colombia                    Familias en Accion              Fitzsimons and Mesnard (Forthcoming)            Yes                        Regression (see footnote)            10-17         Economic activities for pay or for household or chores
13. Colombia                    Subsidios Condicionados a la Barrera-Osorio et al (2011)                        Yes                        Child level RCT                      grades 6-11 Work last week as a primary activity
                                Asistencia Escolar
14. Honduras                    PRAF-II                         Galiani and McEwan (2013)                       Yes                        Cluster RCT                          6-12          Economic activities for pay or for household
15. Honduras                    PRAF-II                         Glewwe and Olinto (2004)                        No                         Cluster RCT                          6-12          Economic activities for pay or for household
16. Indonesia                   Jaringan Pengaman Social        Sparrow (2007)                                  Yes                        Natural experiment IV                10-18         Economic activities for pay or for household
17. Jamaica                     Path                            Levy and Ohls (2007)                            No                         RDD                                  6-17          Economic activities for pay or for household or chores
18. Mexico (rural)              Oportunidades                   Buddelmeyer and Skoufias (2004)                 No                         Cluster RCT                          6-12          Economic activities for pay or for household
19. Mexico (rural)              Oportunidades                   De Janvry et al. (2006)                         Yes                        Cluster RCT                          8-18          Economic activities for pay or for household
20. Mexico (rural)              Oportunidades                   Skoufias and Parker (2001)                      Yes                        Cluster RCT                          8-17          Economic activities for pay or for household
21. Mexico (rural, long-run) Oportunidades                      Behrman, Parker, and Todd (2011)                Yes                        Cluster RCT and PSM                  15-16         Economic activities for pay or for household
22. Mexico (urban)              Oportunidades                   Behrman et al. (2012)                           Yes                        PSM                                  12-14         Economic activities for pay
23. Nicaragua                   Atención a crisis               Del Carpio and Macours (2010)                   In edited volume           Cluster and household level RCT 6-15               Hours worked last week for pay or for household
24. Nicaragua                   Atención a crisis               Del Carpio and Loayza (2012)                    No                         Cluster and household level RCT 7-14               Hours worked last week for pay or for household or in chores
25. Nicaragua                   Red de Protección Social        Maluccio and Flores (2005)                      No                         Cluster RCT                          7-13          Economic activities for pay or for household
26. Nicaragua                   Red de Protección Social        Dammert (2009)                                  Yes                        Cluster RCT                          7-13          Economic activities for pay or for household
27. Nicaragua                   Red de Protección Social        Gee (2010)                                      Yes                        Cluster RCT                          7-13          Hours worked last week (no further definition given)
28. Pakistan                    Female School Stipends          Alam, Baez, and Del Carpio (2011)               No                         RDD                                  12-19         Economic activities for pay or for household
29. Uruguay                     PANES                           Amarante, Ferrando, and Vigorito (2011) No                                 RDD                                  6-17          Working (no further definition given)
30. Uruguay                     PANES                           Borraz and Gonzalez (2009)                      Yes                        PSM                                  6-15          Economic activities for pay or for household or chores
Source: Authors
Note: PSM stands for propensity score matching, RCT stands for randomized controlled trial, and RDD stands for regression discontinuity design. The identification strategy of Covarrubias, Davis, and Winters (2012) and Miller and Tsoka (2012) is
enhanced by a randomization procedure in which 8 groups of villages (containing 23 villages in total) were randomly divided into an equally sized treatment group and control group (4 groups of villages each). The age range for the study by Covarrubias,
Davis, and Winters (2012) is not explicitly mentioned and we deduce the reported age range from the descriptive statistics.Yap, Sedlacek, and Orazem (2002) compare children in nine treatment municipalities to children in nine control municipalities that
were either scheduled for incorporation in the program or had applied to be incorporated. The identification strategy of Attanasio et al. (2010) and Fitzsimons and Mesnard (Forthcoming) is enhanced by the fact that control households are drawn from
purposefully chosen similar non-program municipalities.

                                                                                                                            43
  Table 2. Heterogeneity of conditional cash transfer child labour impacts
                                                       (1)           (2)
Dummy for older age group                            0.003
                                                    (0.015)

Dummy for boys                                                            -0.033
                                                                         (0.017)*
Number of observations                                        24            32
Source: Authors
Note: Column 1 shows the results of regressing changes in child labor for 2 age
groups per study (within the 7-14 age range, if possible) on a dummy taking the value
1 for the older age group (and a constant). Column 2 shows the results of regressing
changes in child labor for boys and girls on a dummy taking the value 1 for boys (and
a constant). *** p<0.01, ** p<0.05, * p<0.1.




                                         44