THE DETERMINANTS OF CHILD LABOR: THEORY AND EVIDENCE

The Determinants of Child Labor:
Theory and Evidence
Drusilla K. Brown
Tufts University
Alan V. Deardorff and Robert M. Stern
University of Michigan
September, 2002
Recent RSIE Discussion Papers are available on the World Wide Web at:
2
THE DETERMINANTS OF CHILD LABOR:
THEORY AND EVIDENCE
By
Drusilla K. Brown
Tufts University
Alan V. Deardorff
University of Michigan
and
Robert M. Stern
University of Michigan
September 2002
I. Introduction
The specter of small children toiling long hours under dehumanizing conditions has
precipitated an intense debate concerning child labor over the past decade and a half. As
during the midst of the 19th century industrial revolution, policymakers and the public
have attempted to come to grips with the causes and consequences of child labor.
Coordinating a policy response has revealed the complexity and moral ambiguity of the
phenomenon of working children.
Although child labor has been the norm throughout history, the fact of children
working and the difficult conditions under which children work occasionally become
more evident. In the midst of the 19th century, child labor became more visible because
children were drawn into an industrial setting. Currently, child labor has become more
visible because of the increase in the number of children producing goods for export.
Our purpose here is not to provide a definitive diagnosis of the causes and
consequences of child labor, but rather to review the existing theoretical, empirical, and
historical literature as to why and when children work. As will become clear, technology
and other demand-side factors interact subtly with household dynamics, culture, and
market and political failures to determine the labor force participation rate and
educational attainment of children.
In section II, we review the theoretical literature and some incidental empirical
evidence concerning household decision- making and its implications for work and school
choices for children. Unitary models are analyzed first, followed by models with
multiple agency. We consider household decision- making, itself, along with market
characteristics that constrain the choices that families make concerning their children. In
particular, considerable recent theoretical and empirical attention has focused on the role
that market failures play in child labor. Failures that emerge in financial, spot labor, and
human capital markets can all give rise to more child labor than is economically efficient
or in the interest of families. We will also find that political failure can play a significant
role in determining the allocation of children’s time. We then turn, in section III, to a
discussion of the empirical evidence on these and related issues based on survey research
of household decision-making.
The demand side of child labor is then broached in section IV. Here we consider the
role that technological change, or lack thereof, can play in creating employment
opportunities for children. Central to this literature is the extent to which mothers, young
women, or machines can duplicate and, therefore, replace the special labor qualities
provided by child workers. Evidence of current work by children is compared to the
historical record on the role of technological change and child labor during the 19th
century industrial revolution.
A more thorough treatment of the rise and fall of child labor during the 19th century is
presented in section V. There remains considerable disagreement among historians
concerning the nature of child labor during the western industrial revolution. Historians
4
continue to debate issues concerning the role that child workers played in the industrial
setting, the impact that the industrial revolution had on total and sectoral employment of
children, and the role of legal restrictions on hours worked, the age at which children
could enter the labor force, compulsory education, and the public provision of
educational services. Nevertheless, there are many distinct parallels between child labor
today and the experience with child labor in the west two centuries ago. We conclude
section V with an attempt to draw some lessons from the historical experience.
We then turn to a series of specific issues thought to be important in affecting both
the supply and demand for children. Section VI addresses the conflicting effects that
trade openness has on child labor. Section VII provides a critical review of the empirical
evidence on the impact of compulsory education laws in the United States. Finally,
section VIII addresses the all- important question concerning the value of an education
and the determinants of education quality. Conclusions follow in Section IX.
II. Theories of the Supply of Child Labor
We begin with a discussion of the supply-side determinants of child labor. First, we
will consider the theory of household decision-making in a perfectly competitive context.
(Empirical implementation of this model will be discussed in the following section.) We
then turn to consider several market imperfections that impact households along with
empirical evidence supporting the relevance of these market failures for the determining
child labor.
A Basic Model of Household Decision-Making
A generic Becker (1981) type household decision model such as the one articulated
by Rosenzweig and Evanson (1977), Pörtner (2001c), or Cignati and Rosato (2000) and
summarized by Schultz (1997) assumes that the household acts to maximize utility,
which is a function of the number of children, the schooling per child, the leisure time per
child, the leisure of the parents, and a composite consumption good. These goods are
produced using a composite commodity purchased in the market place and the time of
household members. The time inputs to produce the composite consumption good can be
supplied by the mother or by the children. Household income can be earned by selling
goods produced in a household enterprise or by working as a wage laborer. Inputs to the
production of the household enterprise good include physical assets owned by the family
and by parent and child labor. Markets for labor, goods, and capital are taken to be
perfectly competitive, at least initially.
The husband allocates time between market work and leisure; the mother allocates
time among market work, child rearing, and home production; and children allocate time
among market work, education, leisure, and home production.
5
Uncompensated cross-elasticities in this model concerning children are the following:
· An increase in the father’s wage raises the implicit price of his leisure and will
lead to substitution toward the child’s education if the child’s education and
the father’s leisure are substitutes. An increase in the father’s wage will also
raise household income. If a quality-child is a normal good, then education
will rise.
· An increase in the mother’s wage increases the opportunity cost of each birth,
thereby lowering the optimal family size. To the extent that child quality is a
substitute for child quantity, the fall in the optimal family size will raise
investment in education. However, to the extent that the mother’s work in the
home is a substitute for child work in the home, child leisure and education
may decline when the mother’s wage rises. Finally, the rise in the mother’s
wage will raise the demand for all normal goods. Quality children may be
among these, in which case educational attainment will rise.
· An increase in the child’s wage works through several channels to alter the
amount of education. First, an increase in the child’s wage raises the
opportunity cost of time spent in school. Second, an increase in the child’s
wage raises the return to each birth. To the extent that the subsequently larger
family size leads families to trade off quality for quantity of children,
educational attainment will decline further.
· The impact of an increase in the child’s wage also depends on whether leisure
and education are complements or substitutes. If leisure and education are
complements, then the rise in the cost of leisure will induce a decline in the
demand for education. However, if they are substitutes, a rise in the wage will
raise the demand for education. In order to determine the net effect of the
child’s wage, we have to weigh the income and substitution effects. If the
contribution of the child’s work to household income is small, then the
substitution effect will dominate and the child will increase work and reduce
education.
· An increase in land holdings or other family assets should increase income,
thereby increasing educational attainment.
The Quality-Quantity Trade-Off
Most, theoretical analys is hypothesizes a tradeoff between the quantity and quality of
children, as reviewed by Schultz (1997). However, Rosenzweig and Evanson (1977)
allow the quantity-quality tradeoff to emerge as a by-product of the impact of the
mother’s wage on the number of children. In this case, the increase in the mother’s wage
raises the opportunity cost of the labor- intensive enterprise of raising children. The fall
in the number of children in the family frees resources available to increase child quality.
For example, the services that children provide to their parents may be defined as the
product of the number of children and their average quality. In that case, quality and
quantity are inherently substitutes. However, Cigno and Rosati (2000) note that this
depends on the presumption that the net cost of a child is negative.
6
Variation in Child Quality across Siblings
Investment in child quality typically varies across children. There are several theories
as to why this would be the case, as reviewed by Ejernæs and Pörtner (2002), associated
with birth order. They identify three different possible explanations, associated with
household budget constraints, biology, and returns to scale in household production.
Household Budget Constraints. Even if parents would like to equalize educational
expenditures across children, they may not do so if they lack access to capital markets or
if they do not realize the value of borrowing against future income. In this event, the
level of spending on first and last born will be higher than the family average for two
reasons. First, as noted by Cigno and Rosati (2000), families that are liquidity
constrained cannot spend the return on their investment in their children until they have
entered the labor force. Once the oldest children in the family begin working, the
household budget constraint is relaxed, permitting more investment in the human capital
of younger siblings.
Second, time spent in a smaller family is longer for first- and last-born than for
middle children (Birdsall, 1991). Note, however, that when a family is liquidityconstrained,
the youngest children in the family receive their bonus in the form of greater
educational attainment since their family will be smallest during their school-age years.
By comparison, the oldest children will receive their bonus in the form of greater
maternal attention as infants. In addition, the last children to be born in the family will
enter when the parents are at the peak of their earning power, thus further biasing human
capital formation in the youngest children in the family (Parish and Willis, 1993).
Birdsall’s insights also suggest some interesting interactions between mother’s work
and child’s work. She reasons that mothers who work are not at a corner solution in the
time allocation across children, whereas mothers who do not work allocate all of their
available time to nurturing. Birdsall finds that mothers who engage in market work
outside the home spread their maternal resources across their children more evenly than
mothers who engage exclusively in home production.
The impact of a mother’s market work on human capital formation is reversed,
however. A mother who reduces hours of market work as the number of children in the
family rises, in order to increase the maternal time spent with each child, also lowers
family income. The negative impact on household income may create an incentive to
withdraw older children from school and send them to work. In other words, a family
can use income of their older children to reallocate the mother’s time from periods in
which her family is small toward those periods in which her family is large. Thus, once
again, we might expect to observe a first-born to begin work earlier than subsequent
children.
7
Biology may play a secondary role. First-borns and children born to older women
tend to have lower birth weight than middle children. These middle children, then, may
have more potential to acquire human capital. Parents may draw conclusions concerning
a child’s genetic endowments and, therefore, ability to acquire human capital, by
observing such factors as how prone they are to illness, genetic disorders and ability to
grow physically based on birth weight. Gender, of course, may also play a role in the
parent’s eva luation of the earnings potential of an educated child.
Ejrnæ and Pörtner (2002) use a biology-based argument to explain birth-order effects
in human capital investment. They assume that parents make fertility decisions
sequentially. They have one child, then observe the genetic endowment. Based on the
observed outcome of the first child, they make a decision as to whether to have a second,
and so on. The objective function of the parents is to maximize an index of human
capital embodied in their children. The upshot of this process is that, as soon as they
have a child with higher than averaged expected genetic ability, they stop having children
and focus a disproportionate amount of resources on that last child.
Thus, even if genetic abilities arrive randomly, parents stop having children or reduce
the rate at which subsequent children are conceived, once an above average child is born.
As a consequence, the last-born child will receive more human capital than children born
higher in the birth sequence.
Returns to Scale in Household Production. Differences in innate ability are not the
only reason that investment in human capital may vary across children in a single family.
Up to this point, we have assumed constant returns to scale in household production. For
the purposes of this analysis, this implies that children in the household in the same
cohort can be assigned the same set of tasks. However, Chernichovsky (1985) suggests
that there may be returns to scale for some tasks. Consequently, children within an age
cohort may be assigned different tasks. Some may be engaged in household production
and others in acquiring human capital.
Along a similar vein, Levison (1991) argues that parents may be diversifying their
investment in children. Placing all children in school may expose the family to excess
risk from income shocks. As a consequence, some children in the family may be
assigned the task of acquiring skills that have immediate market value, such as that which
can be acquired with on-the-job training.
Children as Insurance
The Ejrnæ and Pörtner model also offers an explanation for the inverse relationship
between family size and education. Large families arise when the random birth of the
above average child occurs only after multiple draws from the birth distribution. Such
families, by virtue of their large size, are constrained in their ability to invest even in the
most innately able of their children. Overall investment is therefore lower than for small
families, and investment in above average children is also reduced.
8
In this model, children are being used as a savings vehicle. Parents are optimally
investing in the number and quality of children to maximize the market value of the
family as a whole. Of course, in some economic environments, there may be savings
vehicles that are better investments than children. In countries that do not have welldeveloped
financial markets, land holdings may offer the most attractive rate of return. If
the return to education is low and the return to land is high, then family wealth is
maximized by having a large number of child- farmers.
Parents are also motivated to have children as a form of insurance in economic
environments in which insurance cannot be purchased at an actuarially fa ir price; a point
made by many authors as reviewed by Pörtner (2001b). Parents may be particularly
motivated to use children as insurance instruments when land tenure rights are uncertain.
DeVany and Sanchez (1977) found that land reform in Mexico, which made it impossible
for land to be bought, sold, leased, or mortgaged, resulted in large family size.
Having children for insurance purposes can have several consequences, as explored
by Pörtner (2001b). Since it is the nature of insurance that income is sacrificed to reduce
uncertainty, the return to the last child born may be negative. Thus, family size is larger
than it would otherwise be. Large family size, of course, translates into fewer resources
for human capital investment and, thus, early entry into the labor force. Furthermore, to
the extent that children are used to stabilize income, child labor will be positively
correlated with the severity and frequency of negative income shocks. By contrast,
higher expected future income will lower the demand for children for insurance purposes.
The use of children as a form of insurance also provides some insight into the role of
parental education in determining child labor, even after controlling for current income.
Educated parents are likely to ha ve higher expected future income and, therefore, be less
likely to incur the expense today of having children to insure against low income in the
future. Smaller family size for a given present income translates into more resources for
human capital forma tion. Thus, educated parents may have fewer, more educated
children because of a reduced need to insure against future poverty.
In teasing out the role of parental education empirically, we will be particularly
interested in those studies that control no t only for income but also for expected future
income and family size. That is, is parental education positively correlated with human
capital investment in their children because the parents have higher future income and a
smaller number of children, or because they have a deeper understanding of the value of
education?
Education, particularly of the mother, has a secondary impact on human capital
formation. Child mortality is lower for educated mothers, thus requiring fewer costly
births to achieve the targeted family size. More resources are therefore left to invest in
surviving children. In this framework, positive income surprises raise fertility. Parents
who receive what they believe to be a temporary windfall are likely to invest some of it in
having more children. These additional children then provide additional income in the
9
future when household income returns to a more typical level. As a consequence, income
surprises that ultimately prove to be permanent can produce some curious empirical
results. Families that have persistently high income that is unexpected are likely to have
a larger family size and less human capital formation than we would normally expect.
Thus, income growth will more slowly reduce family size when it is unexpected.
Child mortality has conflicting affects on family size. On the one hand, if a child is
more likely to survive infancy, the expected rate of return per birth is higher, thus raising
the optimal family size. However, the increase in the probability of surviving childhood
raises expected future household income, thereby lowering the optimal family size.
Human Capital Formation
The decision to educate one’s children has an inter-temporal aspect, as discussed by
many authors, most notably Becker (1974). Baland and Robison (2000) make a
particularly direct connection of human capital formation to child labor when evaluating
the efficiency characteristics of household decisions.
They note that when parents are altruistic toward their children, have the ability to
leave a bequest to their children, and have free access to capital markets, then investment
in their children’s education will be efficient. Parents in this setting optimize by equating
the earnings of the last hour of a child’s labor to the present discounted value of earnings
that would accrue to the family due to the last hour of human capital acquisition in
school. That is, the parents act to maximize the value of the dynasty’s income.
Capital Market Failure
Problems with inefficient child labor arise when families are credit-constrained, as
noted by Laitner (1997), Parsons and Goldin (1989), and Jacoby and Skoufias (1997),
and as analyzed by Baland and Robinson. For example, if parents expect family income
to be rising over time, then they may find it optimal to borrow against the future so as to
smooth consumption across time. That is, it is optimal for savings to be negative when
children are young. However, if parents do not have access to credit markets, then they
have to rely on internal assets. In the child-labor scenario, parents borrow from the future
by putting their children to work rather than investing in human capital that will make
their children more productive in the future. Such a strategy, while optimal for the family
in this constrained situation, is not efficient. The present discounted value of another
hour of schooling is greater than the return to another hour of work.
There is an abundance of indirect empirical evidence, discussed below, concerning
the role of credit constraints and educational attainment. However, Dehejia and Gatti
(2002) test the hypothesis directly. They estimate a basic model of child labor
determination for a panel of 172 countries for the years 1950-60, 1970, 1980, and 1995.
The credit-constraint variable is proxied by the share in GDP of private credit issued by
10
deposit- money banks. They find that a one standard deviation increase in the share of
credit in GDP is associated with a 10 percent standard deviation decrease in child labor.
They conclude that families with access to credit are considerably less likely to put
children to work during a period of economic volatility than parents without access to
credit.
Similarly, Jacoby and Skoufias (1997) study the effects of incomplete financial
markets on child labor through their analysis of time allocation of children ages 5 to 18
included in the Village Level Studies Survey, 1975-1978, of ten villages in semi-arid
India. This work is particularly interesting because it attempts to disentangle creditmarket
failure from insurance-market failure. This decomposition is accomplished by
comparing how families cope with anticipated seasonal variations in income with
unanticipated shocks due to variations in rainfall. The use of child labor to smooth
seasonal income variations reflects incomplete capital markets, while the use of child
labor to smooth variations in rainfall reflects the unavailability of insurance.
Jacoby and Skoufias found that parents make significant use of child labor to selfinsure.
Small farms were found to be particularly poorly insured. They do not have
access to seasonal borrowing. In contrast, large farms appear to have access to insurance
but not seasonal credit. However, it is unclear how the consequent irregular school
attendance affects human capital accumulation. Over the course of three years, a typical
child in a household with limited access to credit acquired 98% of the human capital of a
child completely insured against household specific risk.
Nontransferability of Household Assets
The analysis of Baland and Robinson (2000) suggests that as long as asset markets
are functioning and there are transfers between parents and children, parents will make
efficient working decisions for their children. This is the case if the parents are altruistic
toward their children and intend to leave them a bequest, or if children are altruistic
toward their parents and intend to support them during retirement. However, Bommier
and Dubois (2002) argue that when children have a disutility of work, then even if
parents are altruistic toward their children and have access to credit, the amount of child
labor will still be inefficiently high.
Bommier and Dubois consider the particular case in which parents choose the amount
of child labor and that labor reduces the future productivity of their children. These
children, when grown, work and provide a transfer to their parents. Children set the size
of the transfer to their parents by equating at the margin their own utility of consumption
to the pleasure they get from their parent’s consumption.
If a parent makes the child work, that work lowers the child’s future income and,
therefore, the transfer to the parent. Thus, the child can punish the parent for forcing
them to work as a child by reducing the size of the transfer.
11
They key here is that the transfer declines because child labor makes those children,
once grown, less productive than they would have been had they gone to school. That is,
the parents are punished for inefficient formation of human capital in their children.
However, the unhappiness that the children feel while working does not affect their future
income and so does not affect the size of the transfer they pay to their parents. Thus,
parents pay no penalty for imposing this distress on their children, resulting in inefficient
child labor even though asset markets are functioning properly.
Failure in the Markets for Land or Labor
The model of household decision-making outlined above presumes that households
have access to perfectly functioning markets for land and labor. However, as a practical
matter, there may be several types of market failure that will alter the optimizing
decisions combining children and other household assets. Skoufias (1995) emphasizes
the importance of the difficulties that families may have in employing labor or in leasing
land. There may be, for example, high monitoring costs associated with the use of nonfamily
labor. As a consequence, families may have difficulty adjusting toward the
household’s desired cultivated area, given their reluctance to employ labor from the spot
market.
The implications of labor- market failure for child labor are significant. In a model
with perfect labor and land markets, the investment in children should be positively
associated with land holdings through the income effect. However, in the presence of
labor and land-market failure, a family with large land holdings may use the children to
work the land rather than invest in human capital. Indeed, Skoufias (1995) finds in
empirical analysis of six villages in a semi-arid region of India for the period 1975-1984,
that the larger the number of adult males and children in the household, the smaller the
amount of land leased out and the greater the amount of land leased in by the family.
This is the case even after controlling for other household assets, other work
opportunities, and education level of the household head.
Market imperfections in the capital and labor markets can further interact, with
adverse consequences for children. To the extent that land is used as a savings vehicle,
land holdings may be dispersed over a large number of families rather than concentrated
in the hands of a small number of large efficient farms. However, optimal use of the land
may require child labor inputs if there are also significant monitoring and moral hazard
issues with hired labor.
Labor market failure can also contribute to child labor when it is accompanied by
adult unemployment, as analyzed by Basu (2000). Basu considers the impact of an adult
minimum wage. If the statute specifies a wage that is above the equilibrium level, then
adult unemployment may emerge. Parents may bridge the gap in earnings by putting
their children to work. The analysis by Basu is part of a general observation concerning
the interrelationship between income inequality and child labor. Ranjan (2001)
concludes that in an economy where child labor is inefficient – that is, the return to
12
education outweighs the forgone earnings of a child, but poor households with an
uneducated head do not have access to credit markets – then greater income inequality is
associated with more child labor. In contrast, Rogers and Swinnerton (2001) emphasize
the opposite. For economies that cannot support the entire population without child
labor, increased inequality reduces child labor. In this case, if all families have an equal
share of household income then all families will require child labor to survive. However,
if income is unevenly distributed, then families in the upper half of the distribution may
be viable without putting their children to work. In this case, the number of working
children will decline.
Bargaining Failure
Several theories propose the possibility that bargaining failure is a contributing factor
in child labor. Becker (1993) and Baland and Robinson (2000) make a compelling case
that non-altruistic parents fail to invest in an efficient level of human capital in their
children because the child cannot pre-commit to repay the loan made by the parent to the
child while in school.
Genicot (1998) suggests that even when parents are altruistic toward their children,
bargaining with the parent’s employer may give rise to child labor. He argues that for
households at a very low level of income in which the intake of nutrients is suboptimal,
an employer may seek to increase a worker’s productivity by paying a higher wage. The
expectation is that the higher wage would be spent on food, thereby making the worker
more productive. However, if the worker has a family, some of the increased wage may
be spent increasing the consumption of family members other than the worker. Thus, the
sought after productivity-enhancing benefits of the higher wage will not be realized.
In order to internalize the leakage, the employer may seek to employ all members of
the family, including the spouse and children. The reward to the family of supplying the
child for work is not only the child’s wage but also the increment to the parent’s wage
because he/she is no longer sharing productivity-enhancing food with other members of
the family.
Non-Altruistic Parents
Up to this point, we have, for the most part, discussed family decision-making under
the assumption that parents are at least partly altruistic toward their children. However, it
is interesting to examine also the possibility that parents are willing to have children only
if they receive an adequate return on their investment. We consider how the two theories
differ in terms of their implications for the treatment of children. Although it may seem,
at first, that there should be strong testable contrasts between the two models, this is not,
in fact, the case. This is akin to Becker’s (1974) Rotten Kid Theorem, except that in the
case of child labor we are considering the Rotten Parent Theorem.
13
Consider, for example, the analytical framework of Cigno and Rosati (2000). In their
model of non-altruistic parents, families abide by some set of rules requiring each child to
pay an amount T to their parents when an adult. T itself is a function of human capital
formation activity and childhood consumption. Parents have an incentive to maximize
the value of their offspring once they grow to adulthood because it also maximizes the
value of their income in old age.
Concerns for the time-consistency of the arrangement presumed by Cigno and Rosati
lead to the possibility that children will not make the transfer T to their parents as
expected. That is, Cigno and Rosati have not addressed the commitment problem raised
by Baland and Robinson (2000) and Becker (1993).
Lopez-Calva and Miyamoto (1999) attempt to address the child’s commitment
problem by constructing a social norm that produces children who find it optimal to
fulfill their filial obligations to their parents. They consider an infinite-horizon overlapping
generations model. Each generation has to choose how much to educate its
children, how much to save, and how much to transfer to its parents when they retire. If
one generation chooses to educate its children, those children will be more productive in
the future. However, the parent will get a payoff from the investment in its children only
if the children decide to make a transfer back to the parent at retirement.
Whether or not the transfer in retirement is received depends on whether the family is
participating in the following cultural norm: If I pay a transfer to my parents when they
retire, my children will pay a transfer to me when I retire. Each generation chooses to
participate in the cultural norm only if its own utility under the norm is higher than
without it. Thus, an efficiency-enhancing social structure exists that requires each
participant to contribute before receiving a benefit, eliminating the commitment problem
identified by Becker and Baland and Robinson.
Although the Lopez-Calva-Miyamoto model is essentially a model of retirement, it
gives parents a material interest in the human capital formation of their children. That is,
they now have an incentive to maximize the earning power of their children even if the
parents are not altruistic. Under what conditions will such a cultural norm emerge as an
equilibrium? A high rate of return on education and a low rate of return on physical
capital formation will raise the rate of return on the cultural norm relative to savings and
the holding of physical capital as a vehicle for providing for retirement.
There is considerable statistical evidence of non-altruistic parents, such as Burra
(1995), Gupta (2000), and Parsons and Goldin (1989). Ray (2000) finds mixed evidence
that the altruistic feeling that parents have for their children increases with household
income.
14
The Mother’s Stature in the Household
The unified model above presumes that parents jointly solve a family maximization
problem. However, alternatively, we may assume that the parents maximize a weighted
family welfare function, where the weights depend on the bargaining power of the mother
and father. It is well established that households in which the mother has more
bargaining power than is typical spend more on children’s clothes and food and less on
tobacco and alcohol. (See, for example, Kanbur and Haddad, 1994).
However, Basu and Ray (2001) reason that the relationship between maternal stature
in the home and the incidence of child labor is not monotonic. In fact, they find that a
balance of power between parents is more likely to reduce child work than a family in
which all decision- making is concentrated in the hands of a single parent. The reasoning
is straightforward. Consider a household in which both parents have disutility for child
labor but enjoy the goods that the added income can bring. If either parent completely
controls the purse strings, then the additional income earned by children will be devoted
entirely to the goods that they desire. Hence, at the margin, the powerful parent will
weigh their own disutility for child work against the value they themselves place on the
goods purchased by that labor.
By contrast, consider decision making in a balanced household in which each parent
controls half of the household income. In this case, each parent weighs their disutility
from putting their child to work against the goods that can be purchased with only half of
the child’s income. Clearly, both parents will chose a smaller amount of child labor
when they only control half of the income, because the benefit is now smaller than before
but the psychic cost of putting their child to work is just as great.
The psychic cost of child labor has a public- good quality in the household. That is,
each parent suffers the cost of putting their child to work, whether or not they enjoy the
benefit of the proceeds. By contrast, spending the proceeds of the child’s labor ha s a
private- good quality. When one parent controls all household spending, the powerless
parent has no ability to inject his/her willingness to pay to reduce the level of child labor.
Analyzing household data from the Nepal Living Standards Survey (June 1995), Basu
and Ray find that child labor is highest when the father is dominant in the household and
lowest when there is a balance of educational attainment in the household. Further, while
it is preferable to have an educated mother in lieu of an educated father, children fair best
in a balanced household.
Children’s Stature in the Household
The willingness of children to work, aside from their parents’ requirement that they
do so, may also play some role in determining the level of child labor. As with mothers,
an increase in the share of household income earned by children may enhance their role
in decision making in the family. Moehling (1995), in her empirical analysis of early 20th
15
century urban America, finds that working children received a larger share of household
resources than nonworking children.
Indeed, some of the most challenging theory concerning child labor attempts to
simultaneously determine the amount of child labor and the amount of bargaining power
that the child has in the household (Moehling, 1995 and Bourguignon and Chiappori,
1994). These models are complex due to the fact that the amount of bargaining power
that the child has is determined by the fraction of household income earned, but the
fraction of household income earned is in turn an outcome of the bargain over how much
the child works.
If we resolve this debate in favor of determining bargaining power as a function of
potential earnings power, as suggested by Basu (1999), we can provide an explanation for
the startling rise in the stature of the child throughout the 20th century. For it remains
something of a mystery as to why current day parents continue to invest in the formation
of human capital in their children well past the age of twenty years old.
For the sake of argument, assume that bargaining power in the household is a
function of potential, as opposed to actual, earnings. In such a case, one of the impacts of
an upward-trending worker productivity profile is to raise the stature of younger members
of the household. This is the case because younger members of the household ultimately
will be more productive and thus earn more than their older counterparts when these
younger members finally do enter the labor force. Thus, one of the effects of the
technology revolution was to ramp up the worker productivity profile and tilt bargaining
power toward younger members of the household.
Basu (1999) notes that such bargaining games have potential to reach deeply into the
sociological construction of families. For example, the game described above may not
have a unique equilibrium. As a consequence, child labor may be deeply imbedded in a
complex interaction between bargain and outcome that may not be readily amenable to
policy intervention.
Tuttle (1999) adds a second dimension to the issue of child stature within the home.
She argues that one of the effects of technological change in the textile mills was to
create employment opportunities for children distinct from their parents. That is, some
children were employed in the textile factories not in a subcontracting relationship with
their parents, but directly by the plant manager. As a consequence of these new
employment opportunities, parents found themselves in competition with mill managers
for the labor services of their children. The consequent increased bargaining power of
children raised their stature in the home.
Intrahousehold Externalities
Parental bargaining models raise the important question of whether members of a
family are willing to share their personal capital with other members of the family. Basu,
16
Narayan, and Ravllion (2001) theorize that family members will share their personal
capital, such as literacy, with other family members, provided that it does not alter their
relative bargaining power. So, for example, a literate husband may be unwilling to share
the benefits of his literacy with his illiterate wife, if sharing erodes his pre-eminent
decision- making authority in the household.
However, there are cases in which sharing the external effects of literacy may
improve the prospects of the literate member of the household. For example, a literate
mother who is able to help her children with their schoolwork may raise the return to
education. This may accrue benefits to her in the form of greater transfers from her
children in her old age.
Multiple Equilibria
Models that are characterized by some discrete switching regime are commonly
characterized by multiple equilibria. For example, Basu and Van (1998) model a family
in which altruistic parents withdraw their children from the labor force once adult wages
have reached some critical level. As a consequence, the supply of labor is increasing in
the wage below this critical level. Then, once the critical level is reached, parents begin
withdrawing their children from the labor force. Consequently, the supply of labor
begins to bend back. Once child labor has been reduced to zero, the supply of labor
resumes its upward slope.
As a consequence of this configuration, the demand for labor may intersect the supply
of labor more than once. There are then two stable equilibria, a low-wage equilibrium
characterized by child labor and a high- wage equilibrium in which children are all
attending school. Developing countries may be stuck in this low-wage child labor trap.
The low-wage child labor trap is characteristic of a number of dynamic models, including
Basu (1999). Such models provide a basis for compulsory education policies designed to
eliminate the low-wage equilibrium and are discussed below.
Economic Crisis
Economic volatility can affect household decision-making through a number of
channels. On the one hand, a decline in economic activity that reduces current
employment opportunities relative to the future may lower the opportunity cost of an
education relative to its future payoff. Thus, families may decide to increase educational
attainment. However, for families that are credit-constrained or lack access to
employment insurance, the impact may be the opposite. Children are withdrawn from
school and put to work in order to span the economic downturn.
There is considerable evidence that families in developing countries adjust labormarket
activity of the children in response to shocks. Jacoby and Skoufias (1997) find
that parents in rural India withdraw their children from school during an unanticipated
17
decline in crop income. Duryea (1998) finds that paternal unemployment during the
school year reduces the probability of grade advancement for boys and girls.
Behrman, Duryea, and Szekely (1999), find that for 18 Latin American and
Caribbean countries, macroeconomic instability, as measured by volatility of
international terms of trade and GDP, has played a dominant role in the slowdown in
educational attainment since the early 1980s. Similarly, Flug, Spilimbergo, and
Wachtenheim (1998), analyzing cross-country panel data, find a significant negative
correlation between schooling and macroeconomic activity.
Skoufias and Parker (2002) study the impact of economic shock variables on time use
by Mexican 12-17 year olds using the National Mexican Urban Employment Survey.
Analysis is conducted on families during the economic crisis of 1995 and the recovery
period of 1998-1999. Shocks are measured by such variables as the male and female
unemployment rate. They find that, on impact, Mexican families largely turn to older
adult males and females to augment household income, though there is some measurable
effect on the schooling of children. However, shocks have a significant effect on whether
children continue in school in the next school year. The effect is most notable for female
children, suggesting that these girls are replacing the mother’s work in home production.
Finally, “safety net” programs had a significant effect on the effect of macroeconomic
shocks on investment in human capital.
In comparison, Cameron (2002) analyzes the effect of the economic crisis in
Indonesia during the late 1990s on education, labor force participation, and health. Based
on data from four villages in the 100 Villages Survey, school attendance dropped slightly
at the onset of the crisis but is now higher than pre-crisis levels. Fewer children are also
working, though the ones that do are working longer hours. Of children ages 15-19 in
1998, nearly half worked 35 hours or more per week, and 14% worked 55 hours or more.
However, Cameron’s analysis is not entirely supported by Manning (2000). While it
is true that the there is only a small change in aggregate labor- force-participation rates
and enrollment rates in 1998, as compared to 1997, Manning documents a dramatic
increase in the number of street children in Indonesia. Children have become a common
sight, selling food, drinks, and newspapers at most intersections, particularly in Jakarta.
The Department of Social Affairs estimates that children working in this capacity have
risen from 10,000-15,000 before the crisis to around 50,000 in 1999.
Lim (2000) found similar results for the Philippines. Enrollment rates for primary
school fell from 99.2 percent in 1997-98 to 98.1 percent in 1998-99. This change for
elementary students is quite small. However, the enrollment rate for secondary students
fell by 7.2 percent and the enrollment rate for high school students dropped from 76
percent to 70 percent. Labor- force-participation rates also rose for children ages 10-14,
from 9.6 percent to 10.6 percent. For males, the rate rose from 11.7 percent to 13.4
percent.
18
Some additional evidence can be found from the way that families respond to other
types of unanticipated adverse events. Pitt and Rosenzweig (1990) study the effects of an
increase in infant morbidity on the time allocation of families. Based on analysis of the
1980 National Socioeconomic Survey of Indonesia, a high rate of child morbidity
increases the time of teenage daughters spent in home production and reduces their
formal labor-force participation and educational attainment.
III. Household Survey Research
Theories of the time allocation of children discussed in the previous section have
been subject to extensive empirical testing. In a seminal contribution to the empirical
literature on family structure, Rosenzweig and Evanson (1977) simultaneously estimated
the relationship between fertility, sex-specific school enrollment, and child labor-force
participation rates for 189 rural districts of India in 1961.1
Rosenzweig and Evanson’s work subsequently spawned an industry of survey
research on the demographic and economic variables that determine the supply of
children to the labor force. Although the results vary somewhat from country to country,
some underlying themes emerge consistently across all studies. This research provides us
with the most consistent evidence available on the structure of the household and the
determinants of the supply of child labor.
The existing body of survey research on child labor is summarized and discussed in
the Appendix. Here we will confine ourselves to a report on the results obtained by
Rosenzweig and Evanson and then discuss some general lessons to be drawn from this
body of literature.
The fertility equations generally support the hypothesis that the returns to children are
higher in larger families. A ten percent increase in the child wage rate is associated with
a six percent rise in family size. Large land holdings also increase family size. However,
family size declines as the opportunity cost of children rises. A rise in the wages paid to
the mother by ten percent would decrease family size by almost eight percent.
Furthermore, children appear to be normal goods. A ten percent rise in the wages paid to
adult males raises family size by three percent. The authors attribute this to a pure
income effect on the number of children, because fathers do not participate in household
production. Father’s education appears to have no significant impact on educational
attainment of children, but mothers with education above the primary level have smaller
families.
Turning now to the education equations, variables that raise the productivity of child
labor, such as land holdings, land productivity, and the child wage, all lower educational
attainment. However, variables that raise the opportunity cost of children also raise
1 Numerous econometric issues arise in Rosenzweig and Evanson (1977) and are discussed in Schultz
(1997).
19
educational attainment. Most notably, in regions where the mother’s wage is high, she
substitutes away from mothering toward market work. However, the children she does
have are more highly educated than average. This is particularly the case for her
daughters.
Finally, there are the labor force participation equations. As predicted, all variables
that raise the return to child labor and increase family size also increase the incidence of
child labor. However, a rise in the mother’s wage that lowers family size also lowers
child labor. Fur ther, mothers who have more than a primary level of education also are
more likely to educate their children.
Survey Research and Child Labor: What Have We Learned?
1. Education Status of Parents. Parental education plays a persistent and significant role
in lowering the incidence of child labor, above and beyond the impact on family income.
The results presented on this are quite robust, as reviewed by Strauss and Thomas (1995).
In some cases, such as Cote d’Ivoire, the parent’s level of education overwhelms all other
family characteristics.
Several theoretical contributions on the determinants of child labor emphasize the
importance of educating a single generation of parents and the long-term implications for
decision- making in future generations. The theoretical mechanism draws attention to the
impact that an education has on the parent’s human capital and income. That is,
educated parents earn enough income to afford to educate their own children.
However, the empirical evidence very strongly suggests that a parent’s education
affects future generations above and beyond the impact on household wealth. There are
several possible explanations. For example, educated parents have a greater appreciation
for the value of an education, whereas uneducated parents may simply want to believe
that the human-capital decisions made by their own parents were correct. In any event,
cost-benefit analysis of programs that concentrate on educational attainment must look
beyond the impact that an education has on a future parent’s income stream and
incorporate the implications for human-capital formation by subsequent generations.
It has also been argued that the human capital of an education provides the family
with an asset that can be used in the event of adverse economic events, or that it increases
a family's access to formal capital markets. That is, the credit constraint facing a family
is relaxed if there is a literate member of the household.
One might also argue that education is a family tradition or reflects family values.
Becker (1991) offers us an opportunity to put some analytical structure on such a case for
family tradition. It may be the case that families that have a tradition of educating their
children are precisely those who have found some mechanism for solving the intertemporal
bargaining problem between parents and children. Recall from the analysis of
Baland and Robinson (2000), that if parents are not altruistic toward their children or if
20
the optimal bequest is negative, then parents will under- invest in the human capital of
their children. Children cannot credibly pre-commit to repaying education loans that the
parents undertake on their behalf. However, there are, in fact, inter-temporal bargains
that can be struck that resolve the impasse even when parents are not altruistic. It is
arguably the case that families that have a tradition of educating their young, even though
poor, are precisely those who have managed to find a solution to the inter-temporal
bargaining problem.
If bargaining issues are highly relevant to understanding the education of children,
then we have also found a solution to why parents with extensive land holding do not
educate their children. In this case, parents can compensate their children for caring for
them when old, that is, not by caring for their children when young but by manipulating
the bequest of land.
2. School Quality. Several studies point to the importance of school quality as an
important determinant of schooling and work. However, school quality is virtually never
measured directly. It is quite possibly the case that, when a family is poised to move
children out of the workforce into school and fails to do so, the culprit is poor schools.
Poor school quality is found to be weakly important in rural Ghana (Lavy, 1996) and very
important for Africa generally (Bonnet, 1993). It should be noted, though, that even if
poor school quality lowers the value of formal education, there is an abundance of
empirical evidence across Latin America, Africa and Asia that the return to education is
still quite high and more than offsets the foregone income of children in school.
3. The Poverty Hypothesis. The role of household income in determining child-labor
decisions needs further study. Clearly, there is a very strong cross-country negative
correlation between child labor and per capita GDP. However, the role of family income
is not so predominant in explaining variations within a community. We do observe, for
some but not all countries, that household expenditures play a central role in child labor
decisions.
At first blush, this evidence suggests that there are some external effects across
families that make it difficult to put children in school even as income rises or, equally,
difficult to put children to work when income is critically low. In particular, none of the
studies does a very good job of measuring school quality. The role of cost of schooling,
when this is measured, suggests that it may be acting as a proxy for quality. In this case,
parents who have the financial ability to forgo the income from their children may still
not choose schooling if the quality of schools is very poor. However, if we consider the
apparent contradiction between the cross-country negative correlation between household
income and child labor, together with the limited significance of income in the survey
research studies, this seeming paradox is not so difficult to understand.
Theory tells us to expect a correlation between income and child labor under a couple
of different circumstances. On the one hand, if a quality child is a normal good, then
there should be a straightforward negative correlation between income and child labor.
On the other hand, consider the income-child labor connection if credit-constrained
21
parents are using child labor to transfer income from the future into the present. In this
case, the desire to reallocate income backward through time will occur only if current
income is lower than expected future income. Thus, child labor responds not to the level
of income today, but rather to the level of income today relative to future income.
As argued by Baland and Robinson, child labor is a device for transferring resources
from the future into the present. Children who work do not invest in human capital that
would make them more productive in the future. A family will choose to make this intertemporal
shift in household resources when current income is low relative to future
income. Thus, it is not the absolute level of family income that matters for the child labor
decision but, rather, the current level relative to future income. There may be families
that are quite poor and do not have any reason to expect any change in the future. Such
families have no reason to attempt to smooth consumption by putting their children to
work.
None of the studies include income relative to expected future income, explicitly. In
some cases, the researcher attempts to measure permanent income rather than current
income. Such a variable should be significant only if a quality child is a normal good,
since a permanent income measure washes away all information about current income
relative to future income.
In contrast, nearly all of the studies include measures of household human and
physical capital. We may reasonably take these assets as a proxy for the future potential
of a household. In such a case, we would expect that a family with low current income,
but a lot of household assets, would attempt to shift income from the future into the
present through child labor. However, no such pattern emerges. Income does indeed
tend to be negatively correlated with child labor. However, household assets are
negatively correlated with child labor, as well. Thus, we cannot interpret income and
household assets today as a measure of income today relative to income tomorrow.
Rather, it appears to be the case that household assets serve to relax the borrowing
constraint for otherwise credit-constrained households.
How then can we interpret the dominant role of income in cross-country
comparisons? There are a large number of variables that are correlated with GDP that
may, in fact, be driving the correlation between income and child labor. For example,
high- income countries also have well developed capital markets. Thus, families in highincome
countries may more readily smooth consumption over time without resorting to
child labor.
High- income countries also tend to have adopted more sophisticated technology. As
a consequence, the return to education will be higher. In a child labor force participation
equation that includes some proxy for the return to education and measures of access to
capital markets, income may have little additional explanatory power.
4. Children as Complementary Inputs to Household Production. Household assets play
an important role in the child- labor decision. One might expect that the greater the
22
wealth a family has, the lower the probability of child labor. However, there are a
number of assets that require a complementary input of labor, and families may expect to
get that labor from their children. Tapping the human capital of mothers in the family
also requires an increase in child- labor inputs in home production. Thus, a strategy of
increasing access to capital markets may not always lower child labor, at least in the short
run.
Nevertheless, the significant role of household assets lends some evidence to the
possibility that incomplete credit markets give rise to inefficiently high levels of child
labor. For example, the presence of older children in the home considerably reduces the
probability of child labor. Note that there is a measurable impact above and beyond the
contribution that the older siblings themselves make to family income. This is
particularly the case for older brothers, who embody the greatest human capital. In
addition, a parent’s education reduces child labor for reasons other than the impact of
education on the parent’s productivity. It is possible that a parent’s education is viewed
as a marketable asset, or it may be a reflection of the informational externalities
associated with the value of formal education.
What is not clear is why family assets matter. On the one hand, households with
assets can more readily weather adverse events. That is, these assets provide the
household with the ability to manage uncertainty and, as a consequence, child labor is not
required for this purpose. However, families with assets may also have more access to
capital markets or can, themselves, fund a child’s education without a formal loan. That
is, household assets help families transfer household income intertemporally.
In either case, expanding access to formal capital markets to families who otherwise
lack collateral may lead to a reduction in labor force participation rates for children.
However, it is also the case that placing constraints on household decision- making, such
as mandatory schooling, may at least inhibit the family from turning to internal assets that
can be accessed only if children work more. Providing working mothers with firm- level
childcare may also help reduce the reliance on older daughters to care for their younger
siblings.
5. Age. It is clear that older children are more likely to work than younger children. As
children grow older and acquire skills, the opportunity cost of schooling rises. This is
particularly the case for adolescent boys, who are increasingly able to perform physically
demanding tasks as they approach maturity. Thus, it appears that it will be more
challenging and costly from a policy point of view to induce older male children to
remain in school.
6. Siblings. The role of siblings in the household does not appear to be a major
deterrent to schooling once we control for other household characteristics. The only
exception is that there is evidence in some cases that mid-aged children are caring for
younger siblings.
23
When evidence that older children are caring for younger children is combined with
the fact that the presence of an older sibling in the house generally raises the probability
of schooling, it is possible to make a case that parents are diversifying their humancapital
investments in their child assets. The oldest children acquire human capital in the
form of on-the-job training and the youngest children receive formal education.
However, this interpretation of the evidence does not accord well with the other
significant result: the presence of siblings in the same age range tends to raise the
probability of school and lower the probability of work.
Rather it seems more natural, first, to view children in the middle age range, 10-14, as
complements, sharing housework and schooling opportunities. Second, when we observe
older children making schooling possible for their younger siblings, this is likely
evidence that older siblings help relax the liquidity constraint in the presence of capitalmarket
failure. Third, when we observe mid-aged children caring for younger siblings, it
is to help the family make optimal use of the mother’s human capital in the form of
marketable skills. Thus, policies that focus on lowering fertility may not be particularly
effective in reducing child labor.
To the extent that parents diversify their children’s assets, this appears to occur along
gender lines. In Latin America, parents are more likely to engage in the formal schooling
of their daughters, whereas in Africa parents are more likely to school their sons.
7. Connections to History. Several results from the survey literature dovetail
interestingly with analysis of the British industrial revolution that we will discuss in
section V. For example, family subcontracting in the textile mills and mines was
common in the first half of the 19th century. As discussed below, there are a couple of
possible explanations for this phenomenon. For example, employers may choose this
subcontracting relationship in order to internalize the productivity benefits of food
consumption. Or families may seek to internalize the benefits of hard working adults on
the productivity of their assistants.
In any event, such incentive problems appear to continue to give rise to child labor
today. For example, landholding families prefer to employ family members rather than
hire labor on the spot market. Similar subcontracting relationships exist between parents
and children in a host of production settings.
In addition, family size increased during the early part of the British industrial
revolution when the return to child labor and the return to adult labor were declining. But
family size began to decline in the second half of the 19th century when adult wages were
raising the opportunity cost of having children.
8. Technology. Finally, we do not see strong evidence that child labor is driven by the
needs of industry. Children are far more likely to be working in a rural setting rather than
in an urban setting where factories are located. In addition, labor force participation rates
rise with a child’s age, strongly suggesting that the productivity of a child increases, the
larger and stronger the child is. If child workers were valued for their small stature and
24
tiny fingers, we should have observed the opposite. To the extent that child labor is a
demand-side phenomenon, it appears to occur primarily within the household. Families
with a household enterprise or a large tract of land tend to want to put their children to
work. That is, the household’s physical assets are most efficiently employed when the
child’s time is used as a complementary input.
Nevertheless, it remains the case that there is some elasticity in the supply of labor by
children. Many studies find that an increase in employment opportunities, as represented
by a higher wage, will draw children into employment. Thus, if there are demand-side
factors drawing children into employment, parents will oblige.
IV. Demand Side Factors in Child Labor
The demand side of the market for child labor has two distinct dimensions. We most
commonly think of the demand for child workers arising as a consequence of specific
features that children have. It has been argued that the small stature of a child’s body or a
child’s hands make them particularly effective at performing certain tasks, e.g., Marx
(1867).
However, technological advances can have effects on the demand for child labor
counter to those identified by Marx. Levy (1985), for example, notes that during the
1970s, the availability of credit for Egyptian farmers lowered the cost of technologyintensive
inputs. The opportunity to mechanize in sectors such as fruits and vegetables
reduced production of more labor- intensive production such as cotton. The demand for
child labor, therefore, declined with mechanization. Mechanization has a particularly
strong impact on the work of young children who are normally assigned such menial
tasks as pumping water.
Indeed, the demand for child labor can be understood as part and parcel of the
demand for unskilled relative to skilled labor. Skill-biased technological change will
lower the demand for unskilled labor including that provided by children.
Concomitantly, the rise in the demand for skilled labor will raise the return to education,
providing an additional channel through which technological parameters determine the
fraction of time that a child spends working.
Turning now to Levy’s empirical results, first, Levy finds, that a rise in the return to
children increases family size. For the purposes of the current discussion, a 10 percent
increase in cotton’s acreage share in a region increased fertility by 1.5 percent. Between
1969 and 1979, cotton’s acreage share declined by 28%, causing a 4.2 percent decline in
the rural fertility rate. However, output and value added per acre had a significantly
positive impact on family size. Similarly, a child’s wage also raised family size, but the
effect was statistically insignificant.
25
The female wage, as expected, lowered fertility, since the opportunity cost of rearing
children rises with the mother’s wage. By contrast, a higher male wage raised fertility;
thus, even the number of children appears to be a luxury good.
Equations estimating school enrollment mirror results for child labor. Child labor
appears to discourage schooling in this context. Interestingly, the only exception is the
cotton acreage variable. Cotton picking occurs during the summer months when school
is not in session. and therefore it is not in competition for the child’s time spent in school.
A similar phenomenon appears to have emerged during the British industrial
revolution, as will be discussed in detail below. When textile mills were powered by
water and machinery was wooden, children were highly valued for their small size. In
the case of waterpower, the machinery was low to the ground in order to be close to the
waterpower source. In the case of wooden framed looms, legs on the machines were
short in order to minimize mechanical failure due to vibrations. Children were also used
in the mining sector to traverse narrow passages, perform menial tasks, and separate and
wash the mined ore.
However, technological change throughout the 19th century reduced the usefulness of
child workers. The switch to coal as an energy source, the introduction of metal
machinery, mechanical devices for dressing the ore, and larger and more complex mines
eliminated or greatly reduced the demand for child labor.
Admassie (2002) makes a similar argument concerning the cause of child labor in
Ethiopia. There is a fairly strong correlation between the incidence of child labor and
agriculture’s share of GDP. Although there are several possible explanations for this,
Admassie argues that when the production system is “backwards and labor intensive,”
there is a greater demand for child workers.
Corroborating evidence is provided by Swaminathan (1998) in a study of the city of
Bhavnagar in the state of Gujarat (India). Based on a census-type survey of working
children in 1995, rising household income was correlated with increased child labor in
such occupations as diamond cutting, ship-breaking, cleaning plastic cement bags, and
plaiting plastic ropes. In the case of plastic ropes, Swaminathan describes a production
process by which children are used as human shuttles, weaving the ropes held by their
parents.
It is debatable whether children, in fact, do possess special characteristics particularly
valued in an industrial setting. Several studies, e.g., Boyden et al. (1998), have
considered the issue carefully and have found little evidence that children are faster or
more accurate than adults. They note, for example, that children do not make the most
detailed and expensive carpets. This task is assigned to adults.
Similarly, the U.S. Department of Labor (1997) notes that children stitching soccer
balls in Sialkot, Pakistan are paid less per ball stitched than adults. Adults receive 20 to
30 Pakistani rupees per ball while children are paid 20 to 22 rupees per ball. The
26
difference is accounted for by quality. Most high-quality balls are stitched by adults in
stitching centers. Children are not assigned this task because they are not strong enough
to make the required stitches.
In order to determine the extent to which children have specific and highly valued
skills, Diamond and Fayed (1998) estimated the Hicksian elasticities of complementarity
between adult and child labor using Egyptian household survey data for the year 1990-91.
They found that adult female labor is a substitute for adult male and child labor, but child
and adult male labor are complements. Indeed, the elasticity of substitution between
children and adult females is estimated to be -11.4419, quite a high figure. Nevertheless,
it remains the case that adult male and male child labor are complementary. 2 However,
this result does not necessarily imply that there are skills that boys have that cannot be
replicated by adult males. Rather, it is more likely that adult males have special
characteristics that cannot be replicated by children.
The results obtained by Diamond and Fayed provide an interesting perspective on
some historical evidence that will be discussed below. Several historians note that the
fraction of employees who are children varied dramatically across facilities. For
example, while child workers in England were found to be particularly valuable in 19th
century textile factories, plant managers in Ghent, northern Massachusetts and New
Hampshire viewed young women, instead, as a source of ‘nimble fingers’.
Further, Diamond and Fayed found that adult males and female labor are
complements with capital, but child labor is a substitute. This last piece of evidence is
most relevant to our current discussion. The authors interpret the evidence to mean that
children perform tasks requiring little or no skill. As a consequence, labor-saving devices
most readily displace child workers.
Indeed, Nardinelli (1980) argues that the demand for child labor in the English cotton
mills in the early part of the 19th century was precisely of this sort. Even though
technological innovations made the machines easier to run, that task was not wholly
taken over by children. Rather, child labor was complementary to adult labor. Children
were assigned the task of picking up waste cotton, running errands, and assisting older
workers. Better-organized plants required fewer of these secondary workers, thus
depressing the demand for children.
Further, Nardinelli makes the case that the decline in child labor through the middle
of the 19th century in England was largely driven by technological change. In 1830, mills
began to adopt self-acting spinning mules. The self-acting mules broke threads less often
than the hand-operated mules. Piecing the broken threads together had been a task
performed by children, but it was no longer necessary due to the technological
improvements of the self-acting mules.
2 Results obtained by Diamond and Fayed (1998) are broadly similar to those obtained in other studies as
surveyed by Hamermesh (1993).
27
The fall in the demand for children around the turn of the 18th century in the cotton
mills was also a consequence of technological change. Prior to the invention of steamdriven
plants, mills derived their power from water and thus were often located far from
densely populated areas. In order to make up for the labor shortage in rural areas, mills
employed children. By contrast, steam-powered mills could locate closer to urban
centers where adult labor was plentiful.
Galbi (1996) further argues that the fall in the demand for child labor around the
middle of the 19th century arose from the fact that by 1835 an entire generation of
children had grown up in the factories and thus was well suited to factory work. Prior to
this time, factories could only employ adults without that experience, who may not have
had of the mental discipline necessary to work in a factory. The evidence on this issue is
discussed further below.
The above discussion leads us to two important conclusions. First, in many cases,
young women are perfectly, or perhaps even more, capable of performing the tasks
assigned to children. That is, child labor and the labor of young women are very close
substitutes in most cases. Second, in tasks where the child’s small stature is the valued
characteristic, child labor is a close substitute for capital. It is evident from the history of
the industrial revolution as well as from a review of current child-labor practices that
child labor is rendered almost completely irrelevant by modern technology.
Mechanization eliminates the need for children to perform menial tasks such as carrying
water, and precision equipment substitutes for tiny bodies and fingers.
However, some of the most profoundly disturbing and depraved dimensions of child
labor do, in fact, involve the special physical features of childhood. Consider the demand
that arises from pedophilia or those with a peculiar taste for degrading children.
Similarly, two and three year old children, used as camel jockeys in the United Arab
Emirates, are prized for the precise timbre of their terrified shrieks in the ears of the
racing camels. Children are also valued runners in drug trafficking because they are
more difficult to detect by law enforcement officials and, once caught, the punishment is
less severe than for an adult runner. Thus, child runners are less costly for drug dealers
than adults.
Further, some employers believe, rightly or wrongly, that child workers are more
compliant, honest and easily disciplined in the work place. For example, it has been
noted by some historians of the British industrial revolution that corporal punishment,
while routinely used against child workers, was rarely used as a strategy for controlling
adults (Nardinelli, 1982). However, some researchers have argued, as discussed below,
that employers preferred to hire entire families so that parents could be enlisted in
disciplining their own children.
The level of economic activity also exercises a strong impact on the demand for child
workers. Duryea and Arends-Kuenning (2001) found significant wage effects for 14-16
year old urban Brazilian children, based on a household survey for the period 1977-1998
conducted by the IBGE. The opportunity cost of schooling is measured by the state-time
28
variations in the wages of low-skilled males. In equations controlling for adult family
income, child labor was markedly pro-cyclical. A rise in the unskilled wage increased
the number of children employed. Similar wage effects have been found by Jacoby and
Skoufias (1997) for Peru, by Levison, Moe, and Knaul (1999) for Mexico, and by Binder
(1999) for Mexico.
However, the demand for child labor arises, perhaps most commonly, in the
household. This is particularly the case when child labor is complementary with other
household assets such as land, with the capital associated with a household enterprise, or
with young children. As discussed above, certain types of failure in the markets for labor
and land tenancy can lead parents to turn to the internal market for labor. Aunts, uncles,
and the elderly, as well as children in the household who have a personal or financial
stake in the enterprise, are considered to be more reliable, honest, and disciplined than
labor purchased on the spot market. The empirical evidence on this point is discussed in
Sections II and III.
V. Economic History of the Decline in Child Labor
In attempting to understand the phenomenon of child labor today, it is very useful to
examine the evidence concerning the decline in child labor in some of the major
developed countries over the past century and a half. Several historians argue that the
rise and fall of child employment during the industrial revolution reveal a number of
relevant lessons for dealing with issues of child labor today. The rather complex dance
involving technological change, family dynamics, and social construction during the last
two centuries has been explored carefully from a number of analytical and
methodological points of view. Below we attempt to give at least a flavor of this analysis
together with a discussion of lessons for contemporary policy formation.
The Rise and Fall of Child Labor During the British Industrial Revolution
In order to set the stage for the discussion of child labor during the industrial
revolution, it is first worth documenting the extent and conditions of child labor prior to
the British industrial revolution. Tuttle (1999) argues that children were already
commonly working prior to the industrial revolution. She offers anecdotal evidence that,
during the transition of child- labor contracts at the end of the 18th century in the rural
sector from in-kind payments to money wages, the conditions and hours of work became
more onerous. For example, since children increasingly came to live at home and
commute to work, they may have had to add five or ten miles of walking per day on top
of the tasks performed at their place of employment.
In some cases, the working hours were brutally long. Apprentice seamstresses might
work from 4 a.m. until 12:00 a.m. particularly during spring and fall. Further, while it is
the case that children came to work in factories, by the middle of the 19th century the
range of sectors in which children worked was largely the same as it had been at the
29
beginning of the century. Over half of children who worked were to be found in the
agricultural sector throughout the century.
The primary change in child employment pertained to the location of children who
worked in textiles. When domestic jennies were replaced by large water frames, many
children and their parents lost their livelihood resulting in falling wages. Textile workers
responded by seeking employment in the factories, attempting another trade, or entering
the unskilled labor market. In some cases, female children left home to become domestic
servants. Work for girls as domestic servants was generally regarded as more attractive
than factory work. The girls received payment in-kind and acquired skills for use as an
adult. However, around the beginning of the 20th century, young women began to take
over this occupation from young girls.
Some engagements for boys, by comparison, were quite hazardous. For example,
work as a chimney sweep was dangerous and brutal. Young children were compelled to
enter the chimneys often through threat of violence. In this case, the small stature of
children aged 4-10 was highly valued owing to the narrowness of chimneys. The demand
for the labor of small boys declined when chimneys were widened and mechanical
devices were developed to sweep narrow chimneys. The practice was completely
eliminated after the Law of 1875 that required chimney sweeps to be licensed and
conform to all laws regulating age of work.
Factory work began for children with the introduction of water power. These plants
were, of necessity, located near waterfalls in rural regions of England where there was a
shortage of adult workers. Pauper children were commonly brought from the
workhouses of London to operate the machinery. Working conditions in some facilities
were truly appalling. Children in some factories were reportedly worked to complete
exhaustion, living and working in filthy, poorly ventilated, and dangerously hot
conditions.
However, over the 19th century, the pauper apprentices disappeared from the
factories and were replaced by wage labor. There remains considerable disagreement as
to why this transition occurred. Some have argued that the cost of maintaining
apprentices became prohibitively expensive. Others have argued that the Factory Acts
played a cent ral role. It is also possible that the first children used in the factories, upon
turning 18 years of age, continued to work in the factories but were no longer
apprentices. Rather, they were employed as day laborers. Technological change played a
role as well. Textile factories using Watt's Steam Engine, introduced in 1769, were
located in urban areas and employed free labor.
Child Workers in the Early Stages of the British Industrial Revolution
While Tuttle (1999) and others argue that children were already working well before
the onset of the industrial revolution, two changes began to emerge toward the end of the
18th century. First, the sectors in which children worked underwent a transition, with
30
child workers entering textiles, coal, copper, and tin in droves. Second, children's work
became central to the operation of the plant in some sectors. Rather than working
alongside their parents, children now went off to work on their own.
Was this transformation due to a change in technology that raised the marginal value
product of child workers? Or was it due to parental greed, poverty, or profit-maximizing
behavior on the part of firms? One can tease out some of these different influences using
simple demand-supply reasoning. If the number of children rises and the wage paid to
children also rises, then we have a demand/technology driven change. However, if
employment rises and wages fall then we have a supply-side phenomenon.
The evidence from the earliest phases of the industrial revolution does not settle this
question as neatly as we might hope. For, while we know that the sectoral composition
of child employment changed, we do not know whether the number of children working
and the number of hours worked rose or fell.
Horrell and Humphries (1995) analyze data on 1781 English working-class
households for the period from1787 to 1872. Household budget data are available for
several occupations and geographical locations, providing information on household
composition, earnings, occupations, and family expenditures. They found limited
evidence that industrialization lowered the age at which children entered the labor force.
During the 1820s and 1830s, the labor force participation rate for children aged 5 to 9 in
mining, factory work, and high-wage agricultural work rose. Perhaps most importantly
for purposes of this discussion, the participation rates for 10 to 14 year olds rose for all
occupations between 1781-1816 and 1817-1839. After 1840, participation rates then
declined for nearly all occupations.
One might argue that children in industrial families were entering the labor force
because of rising child wages, particularly relative to adults. But this is not the case after
1840. Horrell and Humphries suggest that it is more likely that putting children to work
was the norm, but that families in the industrial sector had more opportunities to put their
children to work than families in other sectors. It is also the case that many of the
children that worked in factories were initially pauper apprentices with little say in how
much they worked or what compensation they received. These children were essentially
enslaved. As a consequence, it is difficult to say what wage they would have
commanded in a competitive market.
Although technological change appears to be a prime candidate for explaining the
pattern of child employment in the first half of the 19th century, parental selfishness is
occasionally offered as a competing explanation. In fact, throughout the discourse on the
evils of child labor in the 19th century, references to lazy and slothful parents are
common.
Horrell and Humphries produce several pieces of evidence that dispute the lazy parent
hypothesis. First, in the early stages of the industrial revolution, children in industrial
households contributed less to family income than children in non- industrial households
31
outside of agriculture. Second, children contributed less than a third of household
income. The father’s wages were the mainstay of the household, contributing 60-95
percent of the household total. Furthermore, the child’s contribution declined after 1840.
Declines were most rapid in low-wage agriculture, mining, outworking, and trades
families. The decline in the contribution made by children in ind ustrial families occurred
more slowly. The only sector in which children increased their contribution to household
income during the middle of the 19th century was in high- wage agriculture. At the very
least, these numbers dispute the view promulgated by some historians that child workers
were the primary source of income for many industrial households.
In addition, Horrell and Humphries consider the impact of the father’s wage on child
labor. Interestingly, even though wages were rising and child labor- force-participation
rates falling in the second half of the 19th century, one cannot make the case for causality.
The problem is that child labor-force-participation rates rose in those families that also
were experiencing a rise in the father’s wage. The wage increase for adult males was
most pronounced in industrial families during the middle of the century. However, in
spite of their rising wages, these fathers were increasingly sending their children to work.
This piece of evidence strongly supports the contention that children worked when
opportunities existed, but that families were often constrained to supply less child labor
than they considered optimal.
Horrell and Humphries offer a second argument that counters the allegations of
parental selfishness as a source of the shifting sectoral employment of children, or at least
that characterizes parental selfishness in a less venal form. This analysis draws from the
intra- household bargaining literature. The technological change that created a market for
the labor of children independent of their parents provided children a meaningful
alternative to working under the terms offered by their parents. That is, parents found
themselves in competition with factory owners for the labor of their children. The
intensification of competition increased the bargaining power of children in the
household.
According to this line of reasoning, children were attracted to the employment
opportunities offered to them by the factory jobs. Thus, children willingly offered
themselves for employment, and the sectoral shift arose from a desire to seek
employment where compensation was paid in cash rather than in-kind, as would be the
case in a household enterprise. Of course it was the case that, for most children, parents
managed to control most of the wages paid. Nevertheless, children were still better off
because the rise in their contribution to household income also raised their bargaining
power within the household. Thus, the only sense in which parental greed enters in the
story is that children would not have sought outside employment if the terms offered to
them by their parents had better reflected the contribution that the child was making to
household production. However, it would not have been the case, according to this line
of reasoning, that parents were forcing their unwilling children into factory jobs.
The issue of autonomous agency is relevant to the debate about child labor today. For
example, Iverson (2002) has argued that some child workers that migrate from rural
32
Karnataka (India) to urban centers do so because of a weak bargaining position in their
parents’ home. Iverson finds that children who migrate against their parents’ will do so
primarily as a consequence of household discord. The costs to these children of
migrating in terms of the lost opportunity to accumulate human capital are considerable.
Technological Change
Though there are supply-side considerations in the changing role of children, Tuttle
(1999) argues that the overwhelming force at work was skill-biased technological change.
As we approach the last quarter of the 18th century, children worked alongside their
parents. The children were not paid directly. Rather parents were paid a premium,
reflecting the greater productivity of the parent-child work team. The question then is:
"what is it about the industrial revolution that changed the role of children?"
Bolin-Hort (1989) makes the case that children were originally drawn into the
industrial setting by their parent s in a sub-contracting arrangement. For example, textile
factory managers found that they could best manage the plant by hiring a single mulespinner.
The mule-spinner would then hire and manage his own piecers and scavengers.
However, Tuttle challenges this view because, while children did enter the factories as
piecers, they had many other tasks as well. Rather, Tuttle (1999) makes the case for
technological change on two grounds. The first concerns the nature of the industrial
work setting. Among other characteristics, children were more willing to submit to the
discipline of the factory floor.
Tuttle also makes the case that the inventions of the early industrial revolution
embodied unskilled- labor biased technological change. For example, Arkwright's roller
spinning replaced human operated machines with human attended machines. The water
frame replaced an adult spinner with a machine minder. Cartwright's automation of the
handloom reduced the need for effort and strength. In addition, the machines themselves
were more easily tended by workers with small stature. The water frame was close to the
ground, and the power- looms were situated close together on the factory floor. Further,
technological change provided for a division of labor, with many small tasks none of
which required more than a couple of minutes of instruction to learn.
In support of the division of labor hypothesis, Tuttle (1999) reports estimates of a
return to education outside of industry on the order of 9 to 42.5 percent. However, within
industry what mattered were not critical thinking skills but coordination, strength, and the
willingness to do repetitive tasks.
The rise in the relative productivity of child workers was reflected in their wages.
Between 1830 and 1860 child-worker wages rose relative to adults. For example, Tuttle
provides evidence that, between 1833 and 1848, the wage of child piecers doubled
relative to adult spinners, rising from 0.21 to 0.42. Thus, according to the technology
hypothesis, in the early stages of the industrial revolution, technological change had an
33
unskilled-labor bias. Firms responded by hiring more unskilled labor (children) and less
skilled labor (adults). Tuttle (1999) presents evidence that the relative wages of unskilled
labor rose during this period, and we will discuss evidence below that the real wages of
skilled workers declined, as well.
However, by the middle of the 19th century the bias began to shift toward skilled
labor, thus lowering the relative demand for children, moving them into peripheral tasks
and raising the return to education and experience. The shift in the bias toward skill is
reflected in a rise in the real return to adult labor after the middle of the century, a process
that persisted through to the end of the 20th century.
As Cunningham (2000) argues, while child labor may have been central during the
first two stages of the industrial revolution, working children became increasingly
marginalized by the beginning of the 20th century. By 1911, children in England were
most commonly employed as messengers. By the end of the 20th century, children were
allowed to keep whatever wages they earned at after-school jobs and received
considerable income supplements from their parents
Technological innovation, however, is to some degree endogenous. It has been
argued that some of the machine designs in the textile sector were specifically tailored to
the small stature of children. This line of reasoning suggests that the driving force behind
the rise in child employment in this sector is a consequence of a rise in supply. Inventors
responded by constructing equipment that suited small bodies. Nevertheless, Tuttle
argues that the fundamental factor was still technology.
The earliest weaving machines were low to the ground so as to be close to the water
that provided the energy source. Second generation machines were set so close to the
ground that even child workers were unable to operate them comfortably or safely.
Deformities in the leg bones of children were common and attributed to the fact that
children spent each day squatting before the looms. Thus, Tuttle asserts that the wooden
framed looms were set low to the ground to minimize the vibrations that would have
disrupted operations had the legs been tall enough to accommodate adult operators.
Children in the Mining Sector
As just argued, the greater demand for children in the textile sector appears to have
been related to technological change. However, the demand for children in the mines
appears to be largely driven by the energy and raw material demands of the growing
industrial sectors. Between 1842 and 1856, coal mining in Great Britain greatly
expanded, with the number of mines and workers quadrupling during this 15-year period.
Unlike in the textile sectors, children in the mines typically worked alongside their
parents or other family members. Humphries (1981) argues that mining provided for a
division of labor between hewers and colliers, on the one hand, whose tasks required skill
and strength and those, on the other hand, whose tasks could be performed by small
34
children. The productivity of the child assistants was enhanced by the efforts of the adult
workers. Thus, adults had an incentive to internalize the external benefits of their own
hard work by hiring their own children as assistants.
In most mines, each family worked in an isolated area, permitting them to identify the
product of their own labor. By contrast, in those aspects of mining in which there were
large economies of scale, such as when ore was loosened by dynamite, subcontracting
was less common. Humphries also argued that families had an incentive to bring their
children to the mines even if they were not productive. All hewers put their coal in a
single pile for transport to the surface. Once at the surface, the colliers decided what
fraction went to each hewer. The larger the hewer’s team of workers the larger the share
of the pile received. Thus, even if children had contributed nothing, the colliers still
remunerated their team with a greater share.
The intermediate demand for coal and ore increased the demand for labor in the
mining sector, and there were some aspects of technology that had implications for child
employment. Advances in the ventilation system beginning in the 1760s made it possible
to increase the depth of a mine. A system of trap doors moved the air between the
bottom of the shaft and the surface. Tiny children spent their days crouched in corners
throughout the mine, opening and shutting the doors as needed. In addition, the
introduction of rails and wheeled coal carts reduced the strength needed to transport the
coal. Thus, even though hewing was still performed by skilled and strong adults,
children took on the task of putters, moving the coal to the surface.
Interestingly, beginning in 1853, Shetland ponies were introduced into the mines,
displacing child putters. As a consequence, child labor in the mines began to decline
during the second half of the 19th century. In fact, in those mines with tall coal seams,
such as in Cumberland, horses were always used to move the coal. Children were never
used for conveyance.
The history of coal mining in Great Britain also demonstrates that technological
change can alter the demand for children in some very dramatic ways. Although early
technological advancement raised the demand for children in textiles, the failure of
technological change raised the demand for children in mining.
Child workers were valued for their ability to move the coal and ore through low
passages. In mines with little potential, the mine-owners did not want to go to the capital
expense of building roomy passageways that could accommodate adult putters. Thus,
children were substituting for capital, not adult labor. By contrast, mines that appeared to
be richly endowed were more intensely capitalized with tall passageways that could be
traversed by adults. In this connection, Tuttle establishes a statistically significant
negative relationship between seam thickness and the child-to-adult labor ratio in English
mines in the mid 19th century.
It is entirely possible that many mines that would have been unprofitable without
child labor were brought into service because of the availability of children. Those mines
35
with thin veins of ore or coal would not produce a sufficient quantity to justify the
investment in tall passages but, in their absence, could not be mined by adults. Child
labor made these marginal mines economically viable.3
In order to tease out the relative influence of demand and supply on the employment
of children in the mines, Tuttle looked at evidence on the movement of child wages. As
with the textile sector, if wages are rising along with employment then demand factors
are paramount. In addition, it is interesting to determine whether the demand for children
is arising largely as a consequence of a demand for coal and ore generally, or whether
technological reorganization of the mining industry particularly impacted child workers.
If the impact is intermediate-demand driven, then wages for adults and children in mining
should rise in tandem. However, if changes in the organization of the mines particularly
raised the demand for children, then child wages should rise relative to adults.
Evidence from the middle of the 19th century suggests that there was little change in
the relative wages of child and adult workers in the mines. However, it should be noted
that this evidence is not definitive, since children were typically paid by their parents and
so may not have been paid their marginal value product.
Child labor in the mines and factories of Great Britain peaked around 1835.
However, the path from that point was not directly downward. After 1870, children
began to leave employment, and by the 20th century children were almost completely out
of the mines. Nearly all workers were 16 years old or older. In addition, there were no
longer any women or children working at the surface dressing the ores.
Why Did Child Labor Decline?
Several explanations for the decline in child labor have been offered by economic
historians. These include: (1) the rise of the bread winner-homemaker household; (2) the
rise in wages of adult males that accompanied industrialization; (3) government
intervention; (4) a rise in the return to education that accompanied industrialization; and
(5) technological change that reduced the demand for unskilled labor.
While the bread winner-homemaker family emerged as a norm for upper-income
families at the beginning of the 19th century, economic forces must underlie its adoption
throughout all income strata. In this connection, Cunningham (2000) argues that
however we explain the decline in child labor, we must account for the rise in stature of
3 A similar phenomenon occurred in the decline in child labor in the fruit and
vegetable canning industry in the United States, as discussed below. Child labor declined
most quickly in those canneries tha t adopted new technology, whereas child labor
persisted in the rural canneries that lacked an incentive to invest in labor-saving
technology.
36
children throughout the twentieth century. Between 1830 and 2000, the cultural
perception of children was transformed from one in which children were seen as small
adults with much the same clothing, privileges, and responsibilities as large adults, to a
perception of children as treasures on which all discretionary income is spent.
Change in income arguably lies behind this sea change in the perception of children.
While real wages grew slowly during the first half of the 19th century, wages rose steadily
throughout the second half of the century. Tuttle’s (1999) review of the evidence
suggests that real wages grew on the order of one percent per year for five decades,
perhaps precipitating the change in family structure identified by sociologists. Similarly,
Goldin (1981) found that the probability that a child was working in Philadelphia in 1880
was negatively correlated with the father’s income.
Brezis (2001) carefully documents the correlation between adult wages, fertility and
child labor throughout the 19th century. In an analysis of the English cotton industry,
between 1800 and 1840, capital per worker in the industrial sector declined in every
decade from 396 in 1800 to 335 in 1830. Real wages in the cotton industry fell during
the same period from an index of 98 down to 45. By comparison, in the second half of
the century, capital per worker rose in every single decade as did real wages. Family
fertility decisions responded to the wage pattern with a lag. Fertility remained constant
for the first four decades of the century and then began to rise, peaking in 1870 and then
declining.
This u-shaped pattern of real wages observed in the 19th century may be due to the
following dynamic path. At the beginning of the century, the population growth rate was
higher than savings per unit of capital. Thus, capital per worker was declining, giving
rise to a fall in real wages and a rise in fertility and child labor. Growing wealth in the
hands of capital owners raised the savings rate so that capital accumulation accelerated.
The dynamics are thus later reversed, with capital per worker and real wages rising, and
fertility and child labor declining.
In contrast, education legislation is unlikely to have been an important factor in the
middle of the 19th century. Mandatory education was not legislated by the British
Parliament until 1876, and education was not provided free to the poor until the Free
Education Act of 1891. Children had begun leaving the factories and attending school
two decades prior to the enactment of these laws. Evidence of rising school enrollment
began in 1833. By the end of the 19th century, most English children were attending or
had attended school.
Parents’ perceptions of the return to education were in part driven by changes in the
demand for labor. However, the school curriculum changed as well between 1844 and
1900. Some time was spent teaching occupations and marketable skills to children,
rather than simply focusing on reading, writing and arithmetic.
Rahikainen (2001) offers a similar analysis of child labor legislation in 19th century
Finland. Child labor became common in Finland during the industrialization period from
37
1840 to 1870, but then began to decline after the mid-1870s. By 1889, just before the
first Labour Protection Law came into force, children constituted only 3.4 percent of the
Finish labor force, falling from a century high of 25 percent. Furthermore, while the
legislation applied to children working in factories, these children made up only about
one percent of all working children. Most working children worked in the rural regions
of Finland without legal protections.
Nardinelli (1980) nevertheless argues that there is weak evidence of the effectiveness
of the English Factory Act of 1835, which constrained the hours of work per week for a
child to no more than 48. While it is true that child labor was already on the decline in
the cotton mills for the two decades before the Act was passed, the decline in the
employment of children occurred more slowly in the silk indus try that was not subject to
the Act, although child employment did decline in the 1840s and 1850s.
Change in technology over the century also caused the demand for skills possessed by
children to wax and wane. Prior to the industrial revolution, children aided their parents
who performed tasks that required skill and strength. The beginnings of mechanization
still required skill and strength, but the assistance of children enhanced the efficiencies
achieved with the new machines. By contrast, the earliest phase of industrialization at the
end of the 18th century created a demand for child workers, independent of their parents.
This is the case since the first industrial machines replaced the skilled craftsmen and
artisans of pre-industrial England. Ho wever, as we move through to the end of the 19th
century, technological change reduces the demand for all workers, particularly children.
A similar process emerged in the extractive industries. By the end of the 19th century,
the essential task of preparing the material brought to the surface was completely
automated. Consequently, the work of women and children was no longer required.
Cunningham (2000) also notes the absence of child labor in the new industries
emerging at the end of the 19th and early 20th centuries. For example, children played no
role in the electrical or chemical sectors. By 1911, one-quarter of the working boys in
Britain under the age of fifteen were employed as messengers. More recently, the
process of marginalizing child workers has culminated over time in the reduced use of
children as newspaper carriers.
The impact of new technology that replaced child workers was also emphasized by
Brown, Christiansen and Philips (1992) in their analysis of the decline in child labor in
the U.S. fruit and vegetable canning industry between 1880 and 1920. During this period
the proportion of workers in the U.S. fruit and vegetable industry fell from 18 to 3
percent. Family income rose over this period along with changes in technology and the
emergence of laws regulating child labor and education. However, there was
considerable cross-state variation, which is suggestive of the fundamental forces that
reduced child labor. Brown et al. were particularly concerned with separating out the
effects of technological change from the passage of compulsory schooling and child labor
laws.
38
Fruit and vegetable processing and canning has two fundamental steps. The
preparation of fruit and vegetables for processing is a somewhat haphazard and seasonal
enterprise in comparison to the cooking and canning process. As a consequence, cooking
was more amenable to continuous processing and was, therefore, more readily adapted to
new production technologies. Workers in the cook rooms were hired from craft labor
markets and standards for consistent worker performance were high. By comparison,
workers in the preparation rooms were hired from the spot market and paid a piece rate.
Considerable variation across workers was thus tolerated by employers.
Mothers brought their children to the preparation rooms in part to work, but also in
part due to a lack of other childcare. The availability of child labor provided an
important supplement to the adult workforce during peak harvesting season. Parents
were also enlisted to discipline child workers.
Two types of canneries co-existed in the United States throughout the early part of the
20th century. Rural canneries processed one or two locally grown products and,
therefore, were idle throughout most of the year, providing little incentive to invest in the
most modern equipment and technology. By comparison, urban canneries attempted to
remain in operation throughout the year by processing a wide range of products. These
canneries were the first to adopt newly available technology and strove to develop an
efficient and continuous production process. Child labor was seen as increasingly
incompatible with developing technology that typically involved huge vats of boiling or
pressurized liquid.
However, mothers continued to bring their children to these increasingly dangerous
plants in order to supplement family income and due to a lack of childcare. Indeed, plant
inspectors monitoring the presence of children encountered more resistance from mothers
than from urban employers. In some cases, urban canners lobbied in favor of child-labor
laws in order to prohibit children in the workplace that they saw increasingly as a source
of annoyance, reducing the productivity of their mothers.
Notably, however, the passage of laws prohibiting children in the canneries included
exceptions for rural canneries. This evidence suggests that the change in technology
adopted in urban centers created an incentive to remove children from the workplace,
thus generating support for child- labor laws. However, the laws did not apply to those
canneries in which child labor was still productive.
It is also notable that urban employers, who supported laws regulating the presence of
children, vigorously opposed restrictions on hours of work by their adult female
employees. They were motivated to keep workers on the job to fully utilize their capital,
particularly during seasonal peaks. Thus, laws regulating ages and hours of work were
tailored to suit the technology rather than the interests of the workers.
Empirical evidence is available to support these conclusions. Child labor declined far
more quickly in urban than in rural sectors. During the decline in the urban sectors,
families migrated to the rural sector to find employment for their children. This fact
39
argues strongly for a demand-side story of the reduction in child labor, rather than forces
coming from the supply side. From a statistical point of view, the rise in capital per
worker had a much stronger role than the rise in family income in explaining the decline
in child employment in the canning industry.
Cogan (1982) also attributes the decline in black teenage employment in the United
States between 1950 and 1970 to technological advancement. Between 1950 and 1970,
the labor force participation rate for black males ages 16-19 declined from 50 to 33
percent, and then declined further to 25 percent by 1978. Several possible explanations
for the decline have been put forward, but none account for more than a small fraction of
the decline. These include expanded coverage of the Federal Minimum Wage Law,
population concentration in urban centers where employment opportunities have been
declining, growth in the size of the black teenage cohort, and improved educational
opportunities.
Rather, Cogan argues, based on analysis of census data, that mechanization in
farming lowered the demand for low-skilled agricultural employment. In 1950, 45
percent of black teenagers were employed in agriculture, but by 1970 agricultural
employment accounted for an insignificant portion of black teenage employment.
Employment trends for black teenage males were mirrored for agriculture employment
generally. As a consequence of mechanization, employment in the farming sector
declined by 56 percent between 1950 and 1970. By contrast, in the urban north, there
was a veritable explosion in the black teenage population, tripling in this period, with no
attendant decline in the labor force participation rate for this group.
The fall in agricultural employment during this period is associated with a dramatic
rise in urban employment opportunities. The associated rise in the cost of labor may have
provided an incentive to adopt labor saving technologies. For, during this 20- year period,
capital per worker in U.S. agriculture tripled. Further, technological change was most
pronounced in cotton production, with the widespread adoption of the mechanical cottonpicker.
In 1950, 99 percent of cotton was picked by hand, but by 1970 virtually all cotton
was picked mechanically.
There is one other interesting conclusion to draw from this evidence. Technological
change rather than employment opportunities in the urban sector appears to have been the
primary causal factor. For if a rise in the demand for labor outside of agriculture were
providing an incentive to adopt labor saving technology, labor would have flowed from
north to south where the largest gains in productivity were emerging.
One might wonder, though, why rural southern black teenagers were not absorbed
into other occupations. Cogan argues that the rise in the federal minimum wage in 1956
swept away jobs at the bottom end of the wage distribution. It is notable that black
teenage employment also collapsed in the southern wood mills, which were covered by
federal wage regulations. However, employment expanded in the exempt service sectors.
40
A second feature of the maturing industrial revolution that contributed to the decline
in child labor has been emphasized by Galbi (1996) in England and by Rahikainen (2001)
in Finland. Work in factories required new emotional skills, as enumerated by Galbi:
regular attendance, consistent work effort, respect for tools and machinery used but not
owned, tolerance for close supervision, and the ability to work in close quarters with
other workers.
There is some evidence that factory managers preferred to hire children because they
could more readily adapt to the work environment of a factory. In fact, some managers
believed that if a person did not begin work in a factory as a small child, they could not
acquire the necessary state of mind as an adult. According to this logic, then, child labor
began to decline when the first generation of child workers matured into adult factory
workers.
Of the British factory managers surveyed by the Factory Queries of 1833, 84 percent
of respondents strongly preferred employees who had been working in the factory since
“infancy.” In fact, very few managers employed workers that had begun work in the
factory as an adult. The human capital formed from experience working in a factory was
well rewarded. Male migrant children ages 11-15 earned an 18 percent wage premium
for two years of additional experience working in a factory and females earned a 26
percent premium.
One additional episode in U.S. history is instructive in understanding the striking
roles that the demand for child labor and the endogeneity of legal institutions can play in
child employment. At the onset of World War II, children once again returned to the
labor force. Aruga (1988) notes that, between 1940 and 1944, school enrollment for 15
to 18 year olds declined by 24 percent and the number of employed children aged 14 to
17 rose by two million, a 200 percent increase. Obviously the labor shortage precipitated
by the war effort was a driving force.
Not surprisingly, the change in economic conditions precipitated a change in law.
Forty-four state legislatures considered amending their child labor laws in 1943.
California and Massachusetts granted their governors broad powers to permit children to
work under any circumstances they specified. In Delaware, 14 year olds were permitted
to work as early as 5 a.m. and as late as midnight. Public campaigns by the federal
government and the media to get children back to school were largely ineffective. In the
period immediately after the end of WWII, children continued to work. In October 1946,
there were nearly three times as many children aged 14 to 17 in the labor force as
compared to1940.
The Supply Side of Child Labor
Although there is much evidence concerning the role of technological change on the
demand for children, we must also consider the evidence on the supply side. One of the
most penetrating statistical analyses of theories of human capital acquisition applied to
41
late 19th century historical data is that undertaken by Parsons and Goldin (1989). They
tested three alternative theories concerning household decision-making.
Model I assumed that parents make child labor/education decisions to maximize
family wealth. Child labor arises when the rate of return on an education falls below the
market rate of interest. If a child is working, families should accumulate assets equal to
the value of a child’s work for transfer to the child in the future. In Model II, the families
in Model I are credit constrained. Child labor arises as a mechanism for transferring
household assets from the future into the present. Children will work only if household
assets are zero. In Model III, parents are selfish but cannot control the income of their
children once they become adults. In this case, child labor arises due to a desire to have
their children earn income during a period in which parents control their children and
their income. Child labor and substantial household assets will co-exist.
The theories are tested on a large-scale micro data set of 6,800 industrial families in
the United States for the period 1889-1890. The results are striking.
· Model I was rejected outright. Parents appeared to move to locations where
there were opportunities to put their children to work even if it entailed a more
than offsetting decline in their own wages.
· Parents did not allow children to retain their own wages or accumulated assets
equal to the value of their child’s labor for transfer as a bequest.
· Model II was also rejected. Parents of working children accumulated
substantial assets for transfer into the future but not paid as a bequest to their
children.
· The evidence was consistent with Model III. Parents put their children to
work even though there were assets available to school them, and the
subsequent assets acquired were not transferred to the child in the form of a
bequest.
· The gains from child labor were almost entirely illusory. In order to secure
employment opportunities for their children, parents moved to locations in
which adult wages were correspondingly depressed.
Thus, Parsons and Goldin provide compelling evidence that parents were, at least in
part, motivated by their own selfish interests and sought out employment opportunities
for their children.
Opposition of parents to schooling children is also discussed by Riney-Kehrberg
(2001) in an analysis of public documents and first hand accounts regarding late 19th and
early 20th century children in the United States and New Zealand. She evaluates the
impact of compulsory schooling laws and technological change and argues that the
intervention of reformers failed in part because they simply condemned the practice of
requiring children to work without understanding the underlying economics. Rural
children were particularly difficult to help because they worked under the close
supervision of their parents. Enforcing child labor and compulsory schooling laws would
have entailed overturning the authority of the parents in their home, actions that public
42
officials were loath to undertake. In cases where rural children actually came to class,
they promptly fell asleep, probably as a result of the long hours of farm work already
undertaken. Not only were parents resistant to educating their children because they were
needed for farm work, parents also feared that education would open up new
opportunities other than farming. This, in fact, proved to be the case.
What Can the Historical Experience Teach us About Child Labor Today?
The central question for our current purposes is what can we learn from the historical
experience that can be applied to child labor today. Currently, children around the world
work in a range of occupations, many similar to pre- industrial Britain such as agriculture,
mining, brick making, domestic service, and in the shops of artisans and craftsmen.
Children also tend to work in partially modernized labor- intensive enterprises.
Cunningham’s (2000) answer to the foregoing question is “not very much,” at least based
on our current understanding of historical events. We, however, will argue that the
historical experience provides an informative perspective on the present, especially if we
include a subtle interpretation of the survey research literature. But let us first consider
Cunningham’s conclusions.
First, were children fundamental to the development of industry during the 19th
century? While it is undoubtedly the case that child workers constituted a large fraction
of the industrial workforce, there are some key exceptions. In northern Massachusetts, in
New Hampshire, and in the Voortmans mills in Ghent (Belgium), young women were
thought to have the special features of nimble fingers, discipline, and malleability to work
in a factory setting. Thus, while it may be the case that the particular skill-bias of
technological change in the first half of the 19th century raised the employment of
children, young women may have possessed the requisite skills and diminutive stature, as
well.
In fact, children as a fraction of the labor force varied greatly across plants and
regions. Cunningham (2000) cites the following figures: In the Voortmans mills, 3.7%
of the labor force was under 15 in 1842, and 10 percent in 1859; in Alsace, one-third or
more of the labor force was under 15 during the 1820’s; in a sample of 43 Manchester
mills, 32.4 percent of the labor force was under 16 in 1833; and for 29 mills in Glasgow,
48.3 percent of workers were under 16.
Second, the evidence on the impact of state regulations is ambiguous. We can
observe several different configurations: (1) child labor in spite of regulations prohibiting
it; (2) child labor declining well before regulations were enacted and (3) child labor
declining precipitously in the wake of legal restrictions. As discussed above, child labor
was already on the decline before the Factory Act of 1935. However, later in the century,
laws appear to have had some impact. Following the 1872 Mines Regulation Act in
Britain, children under 15 as a proportion of the total labor force declined from 10.5
percent in 1871 to 6 percent in 1881.
43
Third, the evidence on technological change is mixed. On the one hand, some
technologies were specifically designed for use by children. However, on the other hand,
there is considerable variation in child employment across plants using the same
technology. One notable example is the use of technology in the cotton industry. In the
American south, children and whole families were employed, whereas in the American
north, employers relied on the employment of young women.
To some degree, these cross-region variations appear to be the result of supply-side
considerations rather than demand-side considerations. In both the cotton industry in the
American south and the U.S. canning industry, families insisted on having their children
in the workplace. It has been argued that employers in many cases saw these children as
a nuisance and thus supported laws banning children in the workplace.
Finally, deep cultural forces certainly played a role in the decline in child labor.
Child labor was nonexistent in Japan throughout the end of the 19th and beginning of the
20th centuries in spite of the absence of legal restrictions. Further, the decline in child
labor in the West was accompanied by the emergence of the romantic conceptualization
of childhood. This phenomenon may be in part a consequence of the decline in infant
mortality, as argued by Kabeer (2000). The rise in income may also have played a role in
the change in the perception of children if children are a luxury good.
Contrasting views of the role of children in the industrial revolution are offered by
Tuttle (1999) and Nardinelli (1991). In Tuttle's view, child labor was prevalent
throughout the British industrial revolution but confined to a small number of industries.
Children provided the core of the industrial work force in mining and textiles throughout
the first half of the 19th century. Children were drawn into these industries because they
were attractive employees from the point of view of industrialists. Children were ideally
suited to work the machines, and their nature was more compatible with the new
industrial regime. Furthermore, technological innovation brought changes that increased
the demand for children as primary workers.
Nardinelli argues that children were key to the early phases of the industrial
revolution, but were largely phased out by the 1830s. Technological change reduced the
demand for children by the 1830s. Children that remained served largely to assist adult
workers and did not displace adult employment.
However, it is arguably the case that the poorly capitalized mines of the 19th century
and the rural canneries of the 20th century provide the most useful insight into child labor
today. Inadequate capital formation that gives rise to the triplet effects of low adult
wages, tasks that can be performed by children, and a low return to education lie at the
heart of child labor in most developing countries. The extent of capital- market failure
that accompanies macroeconomic instability, and inadequate legal institutions that lead to
inefficiently low physical capital formation and technological innovation in developing
countries, are as important in determining child labor as is the inability of families to
borrow to cover education expenses.
44
There remains considerable debate both among historians and students of child labor
today as to the precise role that special features of children played in early
industrialization. At one extreme, one can argue that children were and still are
fundamental. There are tasks that only small children can perform unless replaced by
high precision machinery or other sophisticated tools. Certainly children working as
chimney sweeps or putters in the marginally productive mines were performing tasks that
could not also be performed by adults or even youths.
However, it is also argued that the nimble fingers, keen eyesight and small stature
were fundamental to the operation of the weaving machines of the early 19th century and
the knotting of fine carpets today. There is considerable anecdotal evidence, summarized
by Tuttle (1999, p. 249), that many employers in developing countries prefer to employ
children. Children are valued for the ability and willingness to perform minute tasks for
long periods of time in cramped, poorly lit facilities. If such is the case, then child labor
arises largely from the demand side due to the needs of technology.
A more moderate position is that early industrialization created the opportunity for
employment of children under such circumstances. However, these tasks could be
performed by adults as well. Thus, the existence of child labor must, at least in part, be a
supply-side phenomenon arising as a consequence of the family’s need for the income of
children or similar considerations.
An ILO (1996) document supports the view that adults, or at least female adult labor,
are closely substitutable for child labor. The ILO report cites as evidence a study of
2,000 weavers in the hand-knotted carpet industry, indicating that children were not more
likely to make the finest knots. Further, the finest carpets with the greatest density of
small knots are woven by adults, not children.
Corroborating evidence is provided by Swaminathan (1998) in a study of the city of
Bhavnagar in the state of Gujarat (India). Based on a census-type survey of working
children in 1995, rising household income was correlated with increased child labor in
such occupations as diamond-cutting, ship-breaking, cleaning plastic cement bags, and
plaiting plastic ropes. Children worked at simple repetitive tasks that did not require skill
or training. In addition, the work involved drudgery and exposed the children to
workplace hazards.
As argued by Horrell and Humphries in the context of the British industrial
revolution, Swaminathan maintains that economic growth that accompanied liberalization
created new work opportunities for children that were exploited by their parents.
However, the work did not involve tasks that could not also have been accomplished by
adults or machines.
While the ILO and other evidence is not conclusive, it certainly is suggestive of the
fact that children are not essential to any phase of industrialization from a technological
point of view. It also remains the fact that in many industries in which children work,
45
there is available technology employed in industrialized countries that would completely
eliminate tasks that can be performed only by children.
The policy implications of the distinction between demand and supply side effects are
quite sweeping. When supply-side considerations are paramount, policies that prohibit
child labor in industry will likely only divert children into alternative occupations.
However, if children are employed due to the needs of technology, then laws that prohibit
children from certain dangerous or grueling occupations will provide firms with an
incentive to adopt technological innovations that mimic the special qualities of child
workers, provided such laws are enforced.
We also see a second fascinating parallel between the historical experience and
evidence gleaned from survey research. In household production and farming prior to the
industrial revolution, during the early stages of industrialization of textiles, and
throughout the industrialization of the extractive industries, children worked alongside
their parents. Several historians go to great lengths to document family subcontracting
arrangements that took place in the mines and, to a lesser degree, in the textile mills.
Parents brought their children to the mines as soon as they were able to make even the
smallest contribution, working alongside fathers who were extracting coal or mothers
who were dressing the ore. Children also worked as piecers alongside their mothers in
the textile factories.
A couple of different arguments are put forward to explain the preference for
employing one’s own family members, including the existence of external effects from
the production process. However, whatever the reason, we observe a similar
phenomenon with child labor today. Parents who can employ their children in a family
business will choose to do so even at the expense of the formation of human capital.
Skoufias (1995) has suggested that agency problems in the employment of spot labor in
the farming sector are to blame. Gennicott (1998) makes a theoretical argument that
firms hire families in order to internalize the benefits of food consumption within the
family.
Whatever the reason, the proclivity of parents to employ their own children in
partially modernized businesses has important policy implications. Policies that promote
the development of micro-enterprises may, at least in the short run, give rise to more
child labor rather than less even if they raise family income.
VI. Globalization
Although globalization has raised the awareness of child labor, there is also a
perception that globalization has resulted in an increase in the amount of child labor. The
commonly cited mechanism is that trade between an unskilled-labor abundant/developing
country and a skilled labor abundant/developed country will raise the relative rate of
return to unskilled labor in the developing country. This change lowers the return to
education and raises the opportunity cost of an education, thereby stimulating child labor.
46
There are several theoretical models of the interaction between world markets and child
labor. We briefly review these below.
Trade and the Return to Labor
Maskus (1997) models an economy that produces an export and an import-competing
good with some sector-specific factors. Though adult labor is mobile between sectors,
child labor is employed only in an informal sector that supplies inputs to the export
sector. If we take globalization to be an expanded opportunity to engage in international
trade, then a larger export sector will raise the demand for child labor inputs.
Presumably, the rise in the demand for child labor will be accompanied by a rise in the
child’s wage.
We might conclude, at first, that the rise in the child’s wage will raise the return to
current work relative to the return to education, thereby increasing child labor. In fact, to
the extent that developing countries specialize in goods that are unskilled-labor intensive,
the return to education may indeed fall.
However, Basu and Van (1998) and Basu (2002) model the child-labor decision
assuming altruistic parents. Once the wages of adults reach some critical level beyond
which the family’s survival is reasonably assured, parents will withdraw their children
from the labor force. Thus, in this context, any positive income effects that accompany
trade openness will help families approach or even exceed the critical adult-wage level at
which child labor begins to decline.
We must also consider the possibility that child labor is, in fact, used intensively in
the production of the import-competing good or in the nontraded sectors. In this case,
increased trade openness will lower the demand for child labor. If we can make the
added assumption that trade is also raising income, then both the substitution effects and
income effects in such cases would operate in favor of reduced child labor. However, it
should be kept in mind that the returns to education may still decline for a country that is
specializing in the goods produced by unskilled labor.
Trade and Credit Constraints
Ranjan (2001) considers the inter-relationship between trade and the returns to
education within the context of credit constraints. He models an economy that has a
comparative advantage in unskilled- labor- intensive goods. His particular focus concerns
the dynamics of the transition out of child labor. Unskilled credit-constrained parents are
able to make a transition out of the child-labor equilibrium only after the wage of
unskilled adults rises above the subsistence level. Once unskilled wages rise sufficiently
so that one generation of unskilled parents is able to educate one generation of children,
the dynasty makes the transition to a high wage/educated labor path. Thus, all
subsequent generations of children are educated, as well.
47
Openness will tend to lower the return to skilled labor and to raise the return to
unskilled labor. If the return to education is strongly positive but families with unskilled
labor are credit-constrained, then trade openness will relax the credit constraint. More
children will become educated even though trade lowers the return to an education.
The impacts of trade openness and credit constraints are further explored by Jafarey
and Lahiri (2002). They consider a two-period model in which an export good and an
import-competing good are produced using fixed inputs of skilled and unskilled labor.
Families are rich or poor depending on whether the household head is skilled or
unskilled. Children can become skilled by receiving training rather than working in the
first period. Parents decide whether to educate their children by maximizing utility over
two periods, the utility deriving from both household consumption and the pleasure of
having educated children.
In this context, educating a child implies loss of income in the current period, gain in
consumption in the second period, and the utility of seeing one’s child educated. Thus, at
the margin, the pecuniary returns to education are negative, which must be balanced by
the psychic value of having an educated child. As a consequence, poor parents choose
less education than rich parents. This is the case since poor households have lower
consumption and, therefore, a higher marginal utility of income.
A reduction in the interest rate raises the present discounted value of an education.
Thus, to the extent that openness implies access to credit at a rate below that prevailing
on the domestic market, child labor is reduced and educational attainment is increased.
Opening to trade also raises the price of unskilled- labor intensive goods, thereby
raising the wage of unskilled workers and also the wages of child labor. On the one
hand, both factors reduce the return to education. On the other hand, the income of
poorer families rises, thereby increasing their demand for educated children. Thus, the
impact of goods trade on child labor is ambiguous.
Basu and Chau (2002) alter the above analysis to provide for the possibility that labor
markets in developing countries are not fully integrated into world markets. Thus, an
increase in the supply of unskilled labor will result in a decline in the return to unskilled
labor. As a consequence, families that attempt to use child labor as a form of borrowing
also raise the supply of unskilled labor. The subsequent decline in family income makes
child labor yet that much more likely.
The Basu-Chau analysis takes place in an agrarian setting in which each peasant
household is matched with one landlord. During the off-season, families have to live on
savings from the previous period. The family can augment consumption if the landlord
offers the family a loan during lean periods and is repaid by the peasant household during
the harvest season. The peasant family can separate from a particular landlord only after
the loan is repaid.
48
In a particularly bad year, the size of the loan ma y be so large that it can be repaid
only by supplying the labor of the household’s children as well as the parents. Since
parents are altruistic toward their children by assumption, they will undertake such a
large loan only if the benefits of consumption smoothing outweigh the disutility of
requiring their children to work.
Despite the benefits of consumption smoothing made possible by the credit offered
by the landlord, a family may actually be worse off as a consequence of entering into the
linked labor-credit contract. The increase in the supply of labor in an agrarian
community causes the marginal product of labor during the harvest season to decline,
reducing the going wage and lowering future consumption.
One might expect that families can anticipate the decline in harvest wages and,
therefore, would refrain from entering into a linked contract. However, Basu and Chau
argue that there is a prisoner’s dilemma among peasant households. No single family
itself has an impact on the wage. Thus, each family has an incentive to take advantage of
the consumption smoothing made possible by a linked contract.
Opening to trade in this model has some surprising implications for debt bondage, the
incidence of child labor, and the welfare of peasant households. We presume that
expanded trade opportunities raise the return to unskilled labor, including agrarian labor.
On impact, the increase in the wage during the harvest season creates a greater
incentive to smooth consumption across the lean and harvest seasons. Child labor may
be required to pay off the larger loan. However, the rise in the wage will also give rise to
more savings during the harvest season, to be used to sustain the family during the offseason.
In the steady state, the increase in the wage raises consumption in both the
harvest and off-season, eliminating the demand for more consumption smoothing. Thus,
while there may be some increase in child labor accompanying the initial opening to
trade, the steady-state impact is neutral with regard to child labor. Furthermore, peasant
families are better off as a result of the increase in wages.
To the extent that openness also entails greater access to credit markets, peasant
families may be able to borrow at less onerous interest rates. The smaller interest
payments reduce the pressure on peasant families to offer their children for work in order
to service the household debt.
Finally, there is one interesting twist in the Basu-Chau analysis. As noted above,
peasant families have a coordination problem. They may be better off agreeing
collectively to refuse linked labor-credit contracts if the subsequent decline in wages
makes them worse off. To the extent that participating in the international arena also
implies buying into a set of international labor standards that prohibits forced labor
contracts, peasant families may find help in coordinating their labor supply decisions.
There is one import caveat to the analysis above. Most of the models discussed
conclude that improved access to credit markets that might emerge with globalization
49
will relax the credit constraints faced by families or erode the monopolistic control of
local money-lenders that gives rise to exorbitant interest payments.
However, the L?pez-Calva-Miyamoto (1999) analysis relies on a slightly different
mechanism. Recall that these authors use a culturally constructed retirement scheme to
create a financial interest in the schooling of children on the part of the parents. Parents
invest in the education of their children because it makes them more productive in the
future. That enhanced productivity will accrue to the parents in the form of larger
transfers when they retire.
Now, the impact of improved access to capital ma rkets on this retirement
arrangement depends on the form that this access takes. If, on the one hand, globalization
lowers the rate of return that the working generation can earn on their savings, then
savings as a vehicle for providing for retirement becomes less attractive. Such families
find a greater relative rate of return on inter- generational contracts. Thus, they will renew
their interpersonal connections and take a greater interest in the education of their
children.
However, on the other hand, if integration of world asset markets provides the
working generation with greater access to high yielding investments, then the rate of
return on savings will rise above the rate of return on investing in the human capital of
their children. As a consequence, the social norm will begin to break down and the
commitment problem that children have in bargaining with their parents over an
education re-emerges.
Trade and the Political Economy of Child Labor
In the above discussion of household decision-making, child labor choices are deeply
affected by institutional characteristics such as the quality of schools, the availability of
credit, and laws regulating child employment and compulsory school attendance. One
might argue that such institutions, while not exogenous, may be set endogenously with an
eye toward maximizing GNP. However, Shelburne (2001) considers the political
economy context in which such laws are set. He argues that the social and economic
institutions that proscribe child labor reflect the economic interests of those in power.
An interesting example of this phenomenon is provided by Lea (1975). The U.S.
Federal Child Labor Act of 1916 gained particular support from northern employers who
saw a competitive advantage in supporting federal child labor legislation that would
disadvantage southern producers. The advantage arose from the fact that southern
employers tended to employ whole families whereas northern producers employed
mostly young women.
Similarly, Walker (1970) argues that the AFL-CIO’s primary motivation for
supporting child labor legislation was to reduce the competition between children and
adults for jobs. The interests of the children played only a secondary role.
50
In addition, when considering the motivation for the English factory acts of the
1830s, it has been noted by many authors (e.g., Rose, 1998) that constraining the length
of the workweek for women and children favored steam mills over water mills. Long
hours in water mills during some parts of the year were necessary to make up for lost
time when water levels were low. More recently, the Sanders Amendment to the U.S.
Tariff Act of 1930 was used to punish Sococitrico Cutrale Ltd., Brazil’s largest juice
exporter, for using forced child labor to pick oranges. While the amendment was
seemingly virtuous, Basu (1999) argues that the causal factor driving the charge against
Cutrale was not concern for Brazilian children but rather workforce reductions in their
Florida plants.
Turning now to Shelburne’s analysis in a closed economy context, the returns to other
factors of production in the economy may be positively impacted by an increase in the
supply of unskilled labor. The real return to skilled labor is certainly increasing in the
supply of unskilled labor. Physical capital may benefit as well. However, capital’s
overall interest may be in having access to a larger pool of skilled workers, which is
made possible when children attend school rather than work.
Consider now how opening to trade affects the political economy of social decisionmaking.
The returns to capital and skilled and unskilled labor are now set on world
markets and are independent of the local supply of child labor. The economic incentives
to preserve institutions that predispose families to put their children to work are thus not
as great as in the closed economy context.
In fact, the incentive now lies in the opposite direction. Countries that are unskilledlabor
abundant enjoy an improvement in the terms of trade when the supply of unskilled
labor worldwide declines. Participating in international arrangements that deter child
labor has the benefit of reducing the supply of unskilled labor world-wide, thereby raising
the wages of unskilled labor and improving the terms of trade of deve loping countries.
Empirical Evidence on Openness and Child Labor
Cigno, Rosati and Guarcello (2002) analyze cross-country evidence from the World
Bank’s Development Indicators for 1980, 1990, 1995, and 1998. They regress trade
openness (exports plus imports as a fraction of GNP) on the incidence of child labor, real
income per capita, health policy (public health expenditure share of GDP), and skill
composition (fractions of labor force over 25 that completed primary and secondary
education). Child labor is measured by the 10-14 year-old labor- force-participation rate
and the primary school nonattendance rate.
If skill composition is not controlled for, trade raises the 10-14 year olds’ labor- forceparticipation
rate but has no effect on the nonattendance rate. If skill composition is
controlled for, then openness has no significant effect on child labor or nonattendance.
51
However, if trade openness is measured by the Sachs-Warner index, then trade openness
reduces child labor.
Shelburne (2001) undertakes similar empirical analysis. He regresses child labor (the
ILO’s labor- force-participation rate for 10-14 year olds) on per capita GDP, a measure of
trade openness (imports and exports as a fraction of GNP) and economic size (GNP) for
1996 for 113 countries. Child labor is negatively correlated with income, size and
openness.
Edmonds and Pavcnik (2002) review the available empirical evidence for Vietnam,
focusing on the 1993 liberalization of the rice market. Rice constitutes 44 percent of food
expenditures, accounts for 70 percent of all farmland in Vietnam and is the most common
arena in which children work. An increase in the price of rice, therefore, raises the
opportunity cost of a child’s time spent in school. It also raises the opportunity cost of
adult labor. Two impacts are possible. First, parents may engage in less home work, thus
requiring children to replace them in the home. Second, parents may decide to have
smaller families in order to reduce the time spent in child rearing. Finally, there is an
income effect. The rise in household income may raise the demand for quality children
and relax any credit constraints the family may face.
Given the ambiguous nature of the theoretical discussion, household survey data are
analyzed to determine the impact of two rounds of liberalization in the rice market on
child labor. Between 1993 and 1998, the average price of rice rose 29 percent compared
to the consumer price index in Vietnam. Edmonds and Pavcnik argue that the increase
was in part due to liberalization of border controls, most notably a relaxation in the rice
export quota.
The empirical analysis estimates the probability of work as a function of gender and
age distribution in the family, variations in the rice price, availability of schooling, labormarket
conditions, land and resource endowments, and integration into the Vietnamese
economy. Edmonds and Pavcnik find that a 30 percent increase in the price of rice is
associated with a 10 percentage point reduction in the probability of a child working.
Further, the impact is strongly pronounced for rural children who are most likely to be
engaged in rice production.
However, for urban children, a rise in the price of rice corresponds to a rise in the
probability of child labor. For these children, a 30 percent increase in the price of rice
corresponds to a 5 percentage point increase in the probability of child labor.
Furthermore, the beneficial effect of the increase in the price of rice accrues exclusively
to children in land-holding families. A 30 percent increase in the price of rice is
correlated with a 19 percentage point increase in child labor in households with
negligible land holdings, whereas for land- holding families, child labor declines by 9.5
percentage points.
To some extent, the rise in the incidence of child labor in landless families, when rice
prices rise, appears to be a consequence of the coincident increase in the wages paid to
52
children. Therefore, for poor families the income effect associated with the rise in the
adult wages on child labor is offset by the opportunity cost of education associated with a
rise in the child wages. By contrast, in land-holding families, the rise in the price of rice
also raises the price of land. This either relaxes the credit constraint or it increases
resources in families in which quality children are a luxury good enough to reduce child
labor.
However, another factor may be in play that is suggested by the analysis of Skoufias
(1995). He finds that land holdings raise the demand for child labor and increase
imperfections in the market for spot labor. It is thus possible that the benefit of
liberalization arises in Vietnam because it improves the functioning of the spot market.
Consequently, land holders are less likely to turn to internal markets to work their land.
This interpretation of the evidence is consistent with another piece of the analysis. In
landholding families, the rise in the price of rice reduces child labor in the home because
adults are engaged in more home labor. This suggests that the household was able to
make better use of the spot market to work the land, freeing adult family members to
engage in home work and freeing children to attend school. By comparison, in landless
families, the increase in the price of rice lowers the household production of all family
members and increases their market work. Thus, it appears that whatever happened in
Vietnam raised the employment of landless families in the production of rice at the
expense of home production. Landholding families then substituted toward home
production and education of their children.
VII. Compulsory Schooling and Child Labor Laws
Several justifications for these laws have been put forward. One possibility is that
parents are not sufficiently altruistic toward their children. In this case, society may have
a broad interest in preventing parents from extracting services from their children.
However, Dessy and Pallage ( 2001) note that child- labor laws have a role to play
even if parents are altruistic toward their children. They argue that inefficient child labor
emerges as a consequence of coordination problems between households and firms.
Firms are reluctant to locate capital in markets that lack a supply of skilled labor. Parents
are reluctant to educate their children if there are no jobs for skilled labor. In such a
setting, laws that prohibit child labor and impose compulsory schooling can provide the
needed signal to firms that the requisite supply of skilled workers will be forthcoming.
Firms respond to the anticipated supply of skilled labor by installing the capital that in
turn makes acquisition of the requisite human capital optimal ex post for the family.
A case for compulsory labor laws can also be made in an economy characterized by
multiple equilibria, as analyzed by Dessy (2000). In this model, multiple equilibria arise
for reasons similar to those in Basu and Van (1999). Altruistic parents use a switching
rule based on the level of the adult wage to determine whether children should work or go
to school. Basu and Van (1999) suggest that a ban on child labor can help move the
53
economy out of the low wage-child labor equilibrium. Dessy (2000) argues in favor of
compulsory school laws. Parents are compelled to send their children to school rather
than to work. Due to the compelled increase in education, parents respond by lowering
fertility. That is, they trade off quantity for the quality of children. The law thus
accelerates the transition to a high-education/low- fertility equilibrium.
The Evidence from Compulsory Education Laws in the United States
The first compulsory education law in the United States was passed by Massachusetts
in 1852. All states had such laws by 1918.
It is useful to consider some of the more careful statistical analysis of the impact of
laws regulating entrance to the labor market and compulsory schooling. Angrist and
Krueger (1991) develop a “natural experiment” type statistical technique for evaluating
the impact of compulsory schooling laws on school attendance. The 1960-1980 U.S.
censuses collected information on the “quarter of birth” and “school attendance as of
April 1.” Angrist and Krueger argue that if compulsory school laws are effective,
teenagers who are 16 years old as of April 1 and live in a state that requires students to
remain in school only until they are 16 are less likely to be attending school at the time of
the census than 16 year old teenagers who live in states with a mandatory school age of
17 or 18. They find a statistically significant effect of compulsory schooling laws for
1960 and 1970, thus supporting the hypothesis that laws affect schooling behavior.
Acemoglu and Angrist (1999) perform similar analysis on the same data looking for
the impact of child labor laws on school attendance. They find, for example, that boys
born in states that required 9 years of school before entering the work force spent 0.26
more years in school than boys born in states requiring 6 or fewer years of schooling.
The Angrist-Krueger technique was then applied to earlier periods in U.S. history.
Margo and Finegan (1996) have analyzed the schooling choices of 14- year olds reported
in the1900 federal census. In this study, 14- year olds are broken into two groups: (1)
those teenagers who are already 14 at the beginning of the 1900 school year; and (2)
those who are not yet 14 at the beginning of the school year. Margo and Finegan
hypothesize that if mandatory school laws are effective, the younger 14- year olds living
in a state with a compulsory schooling law should be more likely to be in school than
older 14-year olds. However, no such difference should exist for 14-year olds in states
without compulsory schooling laws. Margo and Finegan find that compulsory school
laws have a positive and statistically significant impact on the decision to obtain some
schooling for younger 14- year olds. However, the laws had no discernible effect on the
probability of fulltime school attendance.
They then considered the impact of compulsory school laws combined with laws that
regulate the minimum age of work. The addition of child-labor restrictions is likely to
have an additional effect on school attendance, because child-labor laws were more
aggressively enforced than mandatory education laws at that time. In this case, the
54
combination of laws has a statistically significant impact on school choice. Young 14-
year olds were 18 to 21 percent more likely to obtain some schooling if their access to the
labor market was legally restricted. However, the laws did not significantly increase the
probability of being in school fulltime.
The statistical evidence presented above has been criticized, most notably by
Moehling (1999). She argues that the laws mandating school attendance are, themselves,
endogenous and tend to follow the decline in child labor rather than precipitate it. That
is, cross-state differences in technology, immigration, and real wages are driving both the
change in educational attainment and the laws regulating school attendance. Thus,
despite the fact that compulsory education laws, child- labor laws, and school attendance
are correlated, it is not a causal relationship.
Moehling draws on the fact that most laws around the turn of the 20th century applied
to 13-year old, but not 14-year old children. Therefore, when work-school patterns for
13- and 14-year olds are similar, it is unlikely that legal restrictions are affecting
household behavior. However, we can detect a role for legal restrictions if 13- and 14-
year olds make different work choices particularly in those states with compulsory
education laws.
Moehling looks at occupation rates – the proportions of youth that identify some form
of employment as their main use of time, as opposed to school. Then, in order to control
for differences in the economic conditions across states that might be driving both the
legislative process and schooling choice, she first looks at the difference in occupation
rates for 13- and 14- year olds in each state prior to the introduction of compulsory
schooling laws. This gives a baseline against which to compare the difference in
occupation rates for 13- and 14-year olds after some states passed compulsory education
laws. Moehling also includes a number of other economic and demographic variables
that have been shown to play a significant role in child labor decisions, as discussed
above.
Moehling finds that the probability that a 14- year old boy would be working fell
substantially between 1890 and 1900 in states with newly enacted compulsory education
laws. However, she observes a statistically similar decline in labor force participation in
states without such laws, thus suggesting that the laws themselves were not the causal
factor for boys.
Similarly, the labor force participation rates for 13- and 14- year old girls in states that
did pass compulsory education laws also fell between 1880 and 1990. By contrast, 13
and 14 year-old girls in states that did not pass compulsory education laws had increased
labor force participation during the decade. Thus, for girls, there is a negative
correlation across states between the passage of laws and labor-force-participation
rates.
The above evidence, thus, suggests that compulsory education laws might be
affecting work-school choices made by (or for) girls. However, Moehling argues that
55
such an inference is not correct. Her reasoning follows from the fact that there is no
differential effect on girls covered and not covered by the law within a single state. That
is, the employment choices by 13- year old girls covered by compulsory education laws
are mirrored by 14-year old girls in the same state but not covered by the law. From this,
Moehling infers that the failure of some states to pass laws requiring 13-year old girls to
attend school, and the increase in the employment of 13-year old girls in these same
states, are being simultaneously driven by other economic factors. These other factors
are similarly driving behavior by 14-year old girls not regulated by legislation.
Moehling then goes on to consider Margo and Finegan’s hypothesis that schooling is
affected by the combination of child labor and compulsory education laws. Once again,
the laws do not seem to be driving behavior. The only case in which 13-year
olds behave differently than 14-year olds occurs for boys in states with no legislative
change. In states with no laws regulating either compulsory education or minimum age
of employment, the labor-force-participation rate for 14-year old boys rose between 1890
and 1900, whereas the labor- force-participation rate for 13-year old boys declined during
the same period. Thus, the results are running precisely counter to the expectation that
laws affect behavior!
In response to the rising labor-force-participatio n rates for girls in the last decade of
the 19th century, there was a burst of legislative activity shortly after 1900. In 1900, 24
states had laws regulating minimum age of employment. By 1910, 43 states had such
laws. Perhaps more importantly, by 1909, 34 states had enacted legislation providing for
inspectors assigned to enforce child- labor laws.
Moehling then applies her statistical technique to the 1900 and 1910 censuses. In this
case, the estimated effect of legal restrictions on school attendance, at least, appears to be
positive but statistically insignificant for some groups. However, the impact is small
relative to the time-series change.
What can we conclude from this evidence? First, the more carefully executed the
statistical analysis, the weaker is the evident effect of legal restrictions on child schooling
and labor decisions. Second, it appears that for carefully crafted laws, such as those
enacted in the last quarter of the 20th century in England, there is some impact of
legislation on behavior at the margin. However, when the age limits specified by the
laws are substantially at odds with optimizing decisions by households, they have little
effect. For example, the laws written in the United States around 1900 tended to specify
14 years as the cut-off between schooling and work. However, Moehling’s evidence
clearly suggests that 14 years of age was not viewed as a significant work-school
boundary for many U.S. households at that time. Similarly, recently enacted laws
regulating work in Brazil have had no effect on household decisions. Thus, mandatory
school laws and minimum age of employment are, at best, a complement to other policies
designed to alter the family’s perception of the appropriate age at which children should
begin working.
56
Finally, the results of Margo and Finegan on the one hand, and Moehling, on the
other, are not as inconsistent as they may seem at first. Margo and Finegan focus on the
1900 census because it not only asks whether a child views school as the main occupation
but also whether the child is in school at all. Moehling, by contrast, looks at several
decades of data and, so, is only able consider whether the laws are affecting a child’s
perception as to whether school is the main occupation. Neither study finds an impact of
compulsory schooling laws or child-labor laws on the child’s perception of his/her main
occupation. That is, neither study finds that the legal restrictions increase the probability
of fulltime schooling.
More recent work has focused on the period 1910-1940. Goldin and Katz (1999)
identify this period as the golden age of the high school movement in the United States.
In 1910, 9 percent of American youth earned a high school diploma, but by 1935 this
figure had risen to 40 percent. During the same period, public resources committed to
education as a fraction of community income rose steadily. In 1910, the national average
was 0.9 percent, but by 1925 it had risen to 2.5 percent and by 1970, 4.5 percent of
income was spent on public education.
The returns to education were a fundamental factor in the high school movement
during this period. Based on data from the Iowa State Census of 1915, Goldin and Katz
estimated that each year of high school increased annual earnings by 12 percent. Quality
education was so appreciated by parents that families often moved to districts with
quality schools. Property values began to reflect the value of the educational
infrastructure, making school construction a good investment for property owners.
The intrusion of the local government on private education decisions was justified
primarily on the basis of imperfections in the market for human capital formation.
Goldin and Katz argue that a sort of social contract was formed between the older
members of the community who could fund education. The elderly would then receive
benefits from an educated community during retirement.
The sustainability of the arrangement depended on educated children remaining in the
community in which they were educated and providing services to the elderly. A wide
distribution of income could undermine the intergenerational contract since the wealthy
are not liquidity constrained. Communities characterized by ethnic and language
differences or recent migrations are less likely to sustain such a commitment.
Goldin and Katz tested the social contract hypothesis by considering the role that
private and public variables play in determining the high school enrollment rate.
Between 1910 and 1928, the high school graduation rate was positively correlated with
wealth or income and negatively correlated with the wage in manufacturing. The latter
captures the opportunity cost of schooling. Greater access to public universities also
increased high school enrollment. It can be argued that the availability of a college
education increases the rate of return on a high school education.
57
High school graduation was also positively correlated with the fraction of a state’s
population that is 65 or older. Automobile registration per capita was introduced as a
measure of income distribution and had a very powerful effect on the high school
graduation rate.
One piece of evidence obtained by Goldin and Katz that runs counter to evidence of
current developing countries is that the high school movement was most pronounced in
rich agricultural areas that were dotted with small towns and villages. As we have seen
above, the rate of return to an education in the rural sector tends to be distinctively lower
than in the urban sector in most developing countries. Yet the high school movement in
the United States began in the states with the largest rural sectors. Goldin and Katz’s
theory suggests that the cohesion of rural communities makes it easier to capture the
external benefits of a high school education. Thus, even though that rate of return may be
lower in rural areas, the ability to overcome the failure in financing education is greater.
Goldin and Katz also have noted that the property owners in communities that had
invested in a quality school system were able to capitalize their investment into the value
of their property. Families with children were attracted to communities that had quality
school systems, thereby bidding up the price.
This piece of evidence suggests another form of market failure that may limit the
ability of communities to develop a public school system. Families that make the
original capital investment in facilities and search costs for teachers will only be able to
reap the benefits of their investment during the period in which their children are in
school. Such a period is almost certainly shorter than the life of the capital investment.
Having future returns capitalized into the value of their homes allows the original
families to capture all of the returns from their original investment. However, if property
rights are not well established, as is the case in many developing countries, then families
will under- invest in their property, including under- investing in community capital such
as schools.
Lleras-Muney (2001) analyzes U.S. census data for 1960 to determine the impact of
compulsory education laws between 1914 and 1939. She regresses the number of years
of education for each individual on individual characteristics such as gender and race, a
set of parameters characterizing compulsory education laws pertaining to the individual,
characteristics of the state in which the individual resided when 14 years old, and a set of
cohort, state and region-specific effects. Results suggest that a one- year increase in the
number of years of required education raises the number of days in school by 18. Some
additional results are even more revealing. First, legal changes that tightened existing
restrictions had a greater impact on educational attainment than laws that initiated
compulsory education in a state. Second, the laws had no effects on blacks. It is possible
that this group had much lower returns to education or that the laws may not have been
enforced for this group. Also, the laws may not have applied due to a lack of available
schools and resources.
58
Lleras-Muney then considered whether educational attainment and laws regulating
compulsory education were both driven by common factors or whether the laws
themselves increased years in school. In order to test for endogeneity, changes in the law
were regressed on average education level in the state, state characteristics, regional
dummies, state effects, cohort effects, and region/cohort effects. Higher state wealth, a
higher urbanization rate, and a large immigrant population resulted in more stringent
laws. By comparison a larger fraction of blacks in the state gave rise to weaker laws and
per capita expenditures on education, and manufacturing wages and percent of workforce
in manufacturing had no significant effect on the passage of laws. These results suggest
that activists groups concerned with the external effects of education were the driving
factor behind legal restrictions. Finally, these results suggest that legal restrictions were
most likely to affect children in the bottom half of the income distribution.
Thus, while compulsory education laws passed at the beginning of the 20th century
may have had little impact, subsequent legal restrictions may have been quite effective.
Indeed, Krueger’s (1996) analysis of laws passed in Britain suggests a similar result. In
the United Kingdom, the legal school- leaving age was changed in 1947, rising from 14 to
15 and then again in 1973, rising from 15 to 16. For each cohort, the modal age at which
children left school coincided with the legal requirement and no more than 5 percent of
children left school before the legal age.
VIII. The Value of an Education
Much of the literature on the education-work tradeoff for children in developing
countries emphasizes the problem of poor schools or the irrelevance to future work
opportunities. Boyden and Levison (2000) report on a survey of results concerning the
impact of education on adult wages in developing countries. They conclude that an
educated adult will earn on the order of 11 percent per year of education more than an
uneducated adult in developing countries. However, the debate on this issue is intense.
We turn now to consider the evidence concerning the parameters and policies that
determine the return to education.
School Quality
Studies estimating the impact of school quality on the skills acquired by children have
been plagued by econometric problems, as discussed in Glewwe (2002). We discuss a
number of these studies below, summarizing their main results and pointing out some of
the problems in interpreting these results.
Harbison and Hanushek (1992) examine the determinants of primary school
performance of children in rural areas of northeast Brazil. Performance in reading and
math are measured by standardized tests administered in 1981, 1983 and 1985. School
quality was measured by ten characteristics: building characteristics; writing materials
(e.g., chalk, notebooks, pencils, etc.); availability of textbooks; and graded classrooms.
59
The only variable that did not significantly improve test scores was the availability of
graded classrooms. Children appear to learn as well or better in mixed-age classrooms
than in an age-segregated environment. With regard to the characteristics of the teacher,
neither the teacher-pupil nor teacher experience variables were significant, but teacher
salaries were significant, and teacher training improved student performance in math.
Glewwe and Jacoby (1994) examine the 1988-89 middle school achievement test
scores for Ghana in reading and math. Of all teacher quality variables measured, only
teacher experience mattered to educational attainment. Experienced teachers were more
skilled at inducing students to remain in school. By contrast, repairing leaky classrooms
and provisioning classrooms with chalkboards were much more important to student
success, with libraries providing an additional albeit smaller benefit.
A similar study of primary school for Jamaican students in 1990 found few variables
that affected student performance. However, students did benefit from routine eye
exams, recent teacher training, routine testing of student performance, and most
importantly, the use of textbooks.
Kingdon (1996) evaluates the impact of several school-quality variables on class 8
reading and math performance for 1991. Teacher variables included years of general
education, years of teacher training, marks received on official teacher exams, years of
teaching experience, and salary. School variables included class size, 17 school building
characteristics, and hours per week of instruction. The significant variables were teacher
exam marks, teacher years of education, physical characteristics of the school building,
and hours of instruction.
All of the above studies attempt to control, though imperfectly, for innate ability,
motivation, and parental inputs. Econometric difficulties also arise due to omitted
variables, such as teacher motivation, and sample-selection bias.
In order to address these limitations, more recent research has focused on the impact
of certain policies such as vouchers or private schools using randomized trials. For
example, Jamison, et al. (1981) conducted a randomized trial on a group of first-grade
classrooms in Nicaragua. Forty-eight classrooms received radio mathematics instruction,
twenty received mathematics workbooks, and twenty served as controls. After one year,
students receiving radio mathematics instruction performed one standard deviation higher
than the controls on standardized tests, and students receiving the math text performed
one-third of a standard deviation higher.
Subsequently, Heyneman (1984) studied first and second graders in the Philippines.
Once again, classrooms were randomly assigned to three groups. Group I received math,
science, and Filipino textbooks, one for each pupil. Group II received the same
textbooks, but each pair of students had to share one set of books. Group III received no
additional instructional material. Student s in Groups I and II performed similarly on
standardized tests and about 0.4 standard deviations higher than Group III.
60
Glewwe (2002) reports on a series of recent randomized studies in rural Kenyan
primary schools. Six studies evaluated the impact of a package of textbooks, school
uniforms and construction materials; textbooks only; block grants; flip charts; a package
of teacher incentives; and treatment of intestinal parasites.
Some results are as follows. The package of textbooks, school uniforms, and
construction materials did not improve performance but did reduce the dropout rate. As a
consequence, class size increased by an average of 35 percent. Researchers were unable
to disentangle the educational impact of the program from the downward bias induced
through larger class size. Textbooks alone did not have a significant impact either,
although the better students did appear to benefit. It is possible, however, that the level
of the textbooks chosen was not appropriate. The typical median child in grades 3-5
could not read the books provided. Nor did flip charts mounted on walls or easels affect
student performance. Finally, treatment of intestinal parasites raised attendance but did
not improve performance.
Natural experiments provide a third approach for studying the impact of school
quality on educational performance. Case and Deaton (1999) study the impact of adding
resources to schools in South Africa after 1993. They find that enough resources to
reduce average class size from 40 to 20 increases grade attainment by 1.5 to 2.5 years and
raises reading scores, conditional on years of schooling, by as much as two additional
years of schooling in a larger classroom.
Angrist and Lavy (1999) study the impact of class size in Israel using data from the
early 1990s on 3rd, 4th and 5th graders. They find a strong negative impact of class size on
student performance. A one-standard deviation reduction in class size raises reading
scores by 0.2 to 0.5 standard deviations and math scores by 0.1 to 0.3 standard deviations.
Policies that affect school choice have also been studied. In Colombia, vouchers
were available for poor urban students between 1992 and 1997. The number of vouchers
was limited. So in cases in which demand exceeded supply, recipients were determined
by lottery. Angrist et al. (2001) find that lottery winners completed more grades due to
reduced grade repetition, though the impact is very small. Lottery winners also had
higher test scores in math and reading.
Card and Krueger (1996) argue that a natural experiment can also be found by
studying the differential effects in educatio n of segregation in the American south in the
early part of the 20th century. Levels of education spending varied dramatically across
students without regard for their innate ability or family wealth. There was a particularly
notable difference between North Carolina and South Carolina, with discrimination
against black students being particularly egregious in the latter. North Carolina was
among the most progressive states in terms of black education, whereas South Carolina
was one of the least. However, by the middle of the 20th century, expenditures by race
were broadly similar in the two states.
61
Card and Krueger focus on the role that pupil-teacher ratio and expenditures per pupil
play in determining educational attainment and earnings. They analyze data from the
1960, 1970 and 1980 censuses to identify trends in education and earnings for men born
in the Carolinas between 1900 and 1959.
Blacks born in North Carolina between 1900 and 1909 attended classes with 14 fewer
pupils per teacher than in South Carolina and acquired 0.65 more years of education. The
pattern is reversed for whites. The typical white classroom in South Carolina had four
fewer pupils than white classrooms in North Carolina and acquired 0.67 more years of
education. Earnings reflect the greater educational attainment of black males in North
Carolina, earning six percent more than black males from South Carolina. In both cases,
the gaps narrowed as we move through the century.
Acemoglu and Angrist (1999) argue that a similar natural experiment was performed
by U.S. states that passed compulsory education laws in the early part of the 20th century.
They find, for example, that for males working between 1960 and 1980, men born in a
state requiring a minimum of nine years of schooling acquired 0.26 more years of
schooling than those born in a state that required six or fewer years. The exogenously
imposed additional years of schooling is then used to estimate the return to education.
They find private returns on the order of 6 to 10 percent.
Based on the above discussion, the evidence on school quality and school
performance and how we define school quality is not definitive. Glewwe (2002) argues
that we have learned a great deal about methodology but not much about school quality
from the last 20 years of research.
Return to Cognitive Ability
Based on the above discussion, it seems clear that schools in developing countries are
not very adept at imparting cognitive ability. That raises a second question: What are the
returns to cognition in developing counties?
Bossiere, Knight and Sabot (1985) examine the returns to cognitive ability in Kenya
and Tanzania. They estimate a standard wage equation including work experience and
years of schooling, with the addition of scores on reading, math, and the Raven’s test of
abstract thinking ability. The results are quite pronounced. Measures of cognitive ability
are always statistically significant and years of schooling and the Raven’s test are not.
However, when the cognitive measures are omitted, years of schooling become
significant. The authors conclude that years of schooling raise wages because they
impart skills, not because they provide credentials to innately able individuals.
Glewwe (1996) performs similar analysis on workers in Ghana and also finds that
education imparts cognitive skills that are rewarded in the market place. Glewwe and
Jacoby (1994) also calculate rates of return to specific educational infrastructure in
Ghana. The rate of return from providing books is 6-7 percent and for blackboards 13-24
62
percent. More broadly, the rate of return to an additional year of schooling was estimated
to be 4-6 percent.
The role of formal education in developing skills that are rewarded in the marketplace
has been confirmed for Pakistan by Alderman et al. (1996), for Morocco by Angrist and
Lavy (1997), and for South Africa by Moll (1998). Cognitive skills have also been found
to raise self-employment income in Ghana by Jolliffe (1998). Though, interestingly,
cognitive skills do not raise farm income. The authors suggest that the returns to
numeracy and literacy are low in farming, which in developing countries involves routine
activities unaffected by technological advances.
The results obtained by Jolliffe also caste a new light on the role of land holdings on
child labor/education decisions. We hypothesized above that imperfections in the market
for spot labor in the agricultural sector led families to turn to the internal market. Thus,
large land holdings transla ted into more child labor. However, it is also the case that the
returns to education in farming appear to be low, at least in Ghana. Thus, parents who
expect to leave a legacy of farming to their children do not have an incentive to obtain an
education for them even if they are altruistic toward their children.
Some of the econometric problems with a cross-section of schools within a country
can be ameliorated with evidence from a cross-section of countries. Lee and Barro
(1997) relate test scores, dropout rates, and grade repetition rates to various measures of
school and family inputs for 58 countries. The key family variables are family income,
parents’ education level, and father’s occupation. School resources include pupil-teacher
ratios, expenditure per pupil, teacher salary and education level, length of school year,
and teaching materials. Test scores are from examinations in mathematics, science, and
reading conducted by the International Association for the Evaluation of Education
Achievement and International Assessment of Educational Progress and cover primary
and secondary students. Results show a very strong impact of family inputs on student
achievement, including both family income and educational achievement of parents. In
contrast to other studies, Lee and Barro find that the pupil-teacher ratio lowers
achievement while the average salary of teachers is positive and significant. As with
other studies, expenditure per pupil is not important, nor is the length of the school year.
Lee and Barro find that similar factors determine the dropout and repetition rates.
Duflo (2001) finds that the school building program undertaken by the Indonesian
government in the mid 1970s had a strong impact on educational attainment and wages.
Between 1973 and 1978, the Indonesian government built 61,000 primary schools. Duflo
finds that children aged 2 to 6 received an additional 0.12 -0.19 more years of schooling
for each school constructed per 1000 children. The increase in education translates into a
1.5 to 2.7 percent increase in wages for each additional school built per 1,000 children.
The implicit rate of return to education ranges from 6.8 to 10.6 percent.
The rate of return to schooling can be found from the ratio of additional income to
forgone income. One estimate of the return to education in India (The Probe Team,
1999) finds that a primary education in India raises income by 20-100 percent. A
63
discount rate of 3.7-14.9 percent would be necessary to make education profitable.
However, Basu (1997) finds that the effective interest rate faced by poor borrowers in
India may be as high as 16-200 percent. Low rates of return to education in other
developing countries were found by Saha and Sarkar (1999) for India and by Cohen and
House (1999) for urban Khartoum.
There are a number of possible explanations for the low rate of return to education in
developing countries. It may, for example, be a reflection of the absence of knowledgeusing
capital or technology. However, Fleisher and Wang (2001) hypothesize that the
rate of return may be a consequence of imperfectly competitive characteristics of the
market for skilled labor in developing countries. They base their views on the analysis of
the return to education in China. In particular, the ratio of wages of skilled to unskilled
workers is far smaller than the comparable labor productivity figures for urban workers.
However, as is the case in other studies, the wage-productivity gap is not as pronounced
in the rural sector where the returns to education are fairly small.
Fleischer and Wang argue that the skill-wage compression and low returns to
schooling in China are the result of restrictions on worker mobility and unexploited
economies of scale. The degree of monopsonistic exploitation is startling. They estimate
that the marginal product of aggregate labor in urban markets in their sample is 7.7 times
the average earnings.
However, in more fluid Chinese labor markets, the return to education appears to
approach levels achieved in other countries. Wan, Maruyama, and Kikuchi (2000), based
on a small sample of workers, find that in Harbin City the rate of return for managers
with a college degree was 13.8 percent.
Indeed, any policy that interferes with the labor- market function deters human capital
formation and encourages child labor. Access to good jobs for women will raise the
return to education, as noted by Anker (2000).
Market Work and the Impact on Schooling
Child labor will be deterred to the extent that formal schooling and work are in
competition for a child’s time. Based on a review of the available evidence, Anker
(2000) concludes that work on the order of two-three hours does not interfere with
schooling. Further, it remains the case for many children that the most valuable form of
education is acquired through work in apprenticeships or in the family business. In these
cases, work and education coincide.
IX. Conclusions
We have presented above a description and discussion of the state of scholarly debate
on the supply and demand-side determinants of child labor. Rather than attempt a
64
recapitulation of the above material, we would like to take the final section to draw some
closing conclusions for the policy debate.
First, the evidence clearly suggests that child labor is a consequence of both the
supply of and the demand for child workers. The conventional wisdom that child labor is
fundamentally driven by any single cause such as poverty, greed, or ‘nimble fingers’ is
not supported by the evidence. While it is clear, based on both cross-country and
household survey research, that poverty increases the incidence of child labor, it is also
the case that child labor surges when employment opportunities present themselves.
Though we might be tempted to conclude that child labor is fundamentally a
consequence of supply-side forces, the sudden rise in the labor force participation rate of
children and the decline in high school enrollment in the United States during World War
II clearly demonstrate the powerful role that the demand for child workers can play in the
overall determination of child employment. Corroborating evidence emerges from the
role of technological change in the employment of young black males in the United
States between 1950 and 1970.
Second, it is important to bear in mind that parents are the single largest employer of
children. In many cases, parents employ their children in the household, family
enterprise, family farm, or even on the factory floor in order to keep the family intact.
However, it also appears to be the case that families turn to internal markets because
parents face a host of incentive problems when non-family members are employed.
Asset market failure provides an additional obstacle to optimal human capital acquisition.
Finally, although the theoretical and empirical evidence suggest that there are many
factors that drive child labor, the precipitous decline in the hours that children worked
and the improvement in the conditions under which they worked in the West between the
middle of the 19th and 20th centuries suggests that in the correct economic, policy, and
cultural environment, eliminating child labor is an obtainable social objective.
However, a policy response that targets a single dimension of child labor is unlikely
to be efficient or even effective. Child labor may decline as incomes in developing
countries rise, but there is nothing in theory or evidence that tells us such an outcome is
inevitable. Perhaps more importantly, there are many policy tools available that can help
alter family and firm decision making prior to a rise in income, such as a package of
improved educational opportunities, increased inflow of child-displacing technologies
from industrial countries, more efficient capital and labor markets, and educational
attainment subsidies. Efforts to eliminate child labor must take all aspects of the problem
into account and draw upon these and other mechanisms that have the potential for
reducing child labor without inducing further hardship.
65
APPENDIX
SURVEY RESEARCH
Author: Chiswick (1986)
Hypothesis: American Jews are among the most successful ethnic group in the United
States. The author attempts to determine whether the greater economic achievement is
the consequence of human capital investment in children in addition to formal schooling.
If this is the case, then the presence of children should have a greater negative effect on
female labor supply in Jewish than in non-Jewish households.
Data: White females aged 25 to 64 in 1970 with at least one foreign born parent taken
from the 1970 Census of the Population.
Model: Female labor supply is a function of the real wage, marginal productivity in the
home, and other income sources, education, age, marital status, on-the-job training and
number of children born younger than six and 6 to 18 and speaker of Yiddish.
Regression method: OLS
Results: Even though Jewish women have a higher labor supply than the total population,
the number of children under six in the household has a strong and significant negative
effect on labor supply. Having a child under six lowers the probability of working for
non-Jewish women by 14 percent but lowers the probability for Jewish women by 26
percentage points. Similar results are observed for weeks of work. Jewish women with
children aged 6 to 18 are also less likely to work than non-Jewish women but the impact
is only about one-quarter the size. Interestingly, the labor supply of Jewish women
whose children are 18 or older is the same as that for Jewish women who never worked
and are more likely to be working that non-Jewish women.
Comments: Spencer (1992) undertook similar analysis for Canadian women for 1981.
The Census of Canada asks respondents whether they are Jewish. By comparison,
Chiswick identified Jews by whether the spoke Yiddish. Spencer claims that he finds no
significant difference labor supply Jewish women with small children are and their non-
Jewish counterparts. However, Chiswick (1992) disputes Spencer's interpretation of the
Canadian results.
Author: Emerson and Souza (2002)
Hypothesis: Evidence form sociology and psychology literature find that siblings higher
in the birth order tend to have higher innate ability and, thus, may attract more human
capital investment by parents than younger siblings. However, in the presence of income
constraints, parents may be unable to invest in older children but can invest in younger
children with the aid of income earned by first borns.
66
Data: 1998 Pesquisa Nacional por Amostragem a Domicílio (PNAD).
Model: Each child in the sample is assumed to allocate time between work as a child
laborer and attending school. The analysis focuses on families with three children and a
mother that is 40 or older. These criteria are designed to isolate complete families and
abstract away from problems associated with endogenous fertility.
Estimation method: bivariate probit. The probability of being in school or working is a
function of birth order, gender of the oldest child, the child's age, the child's race, the
father's and the mother's schooling, the father's and the mother's age and status as rural or
urban household.
Results: For the unrestricted sample, first born boys are least likely to be in school and
last born boys are least likely to work. First born girls are less likely to attend school but
no more or less likely to work. The mother's age is negatively correlated with the
probability that a male child will work and a larger family implies a higher probability of
work for both males and females. The various restricted models give similar results.
Discussion: Last born males are less likely to work than their first born siblings. This
result is consistent with the hypothesis that poor families can send a child to school only
if at least some of them work. Parents send the oldest one to work because he or she
commands the highest wage and, therefore, the opportunity cost of an educating older
children is higher than for younger ones. First born children are less likely to be in
school but no more likely to be working. However, since home production is not
considered work in the survey, these girls who are not in school are likely aiding in home
production.
Author: Robles-Vásquez and Abler (2000).
Hypothesis: Some theory suggests that poverty is an important contributor to child labor.
Recent economic volatility in Mexico provides an opportunity to determine the extent of
the role that poverty plays in child labor. In addition, in 1994 the mandatory years of
schooling increased from six to nine accompanied by considerable fluctuations in
education spending and a reduction in farm subsidies over the sample period..
Data: Mexican Encuesta nacional de Ingresos y Gastos de los Hogares. This is a
household survey of socioeconomic characteristics, incomes and expenditures. Data are
taken from surveys conducted in 1984, 1989, 1992 and 1996. Labor force participation is
analyzed for children over 11 years of age.
Dependant Variables: The probability of full- time schooling, the probability that an
employed child engages in market work (as opposed to a family enterprise), the number
of hours worked for a child employed in market work.
67
Independent Variables: age, years of education, sibling congestion, biological offspring
of head of household, children in the household 5 years or younger, males aged 18 and
older, females ages 18 and older, household size, single-parent household, household
head’s years of education, head’s age, head’s gender, employment status of head,
household income (net of the child’s earnings), household income from corn or bean
sales, profits from a family enterprise or financial assets, dummy variables for farm
households and nonagricultural enterprise, ownership of a refrigerator, students per
school in the community and regional dummies.
Results:
· An increase in one year of age lowers the probability of full-time schooling by
0.11 for urban boys and 0.19 for rural girls.
· An increase in the number of years of schooling increases the probability of being
a full-time student by 0.03 for urban boys and 0.08 for rural girls.
· A boy in an urban household reporting profits from a family enterprise or
financial assets has a 0.13 lower probability of full- time schooling than other
urban boys.
· A boy in a rural household reporting such profits is 0.10 less likely to be a fulltime
student than other rural boys.
· Household income has little effect on child labor independent of other factors.
Doubling household income only increases the probability of being a fully-time
students by 0.01 for rural girls and 0.03 for rural boys.
· Household demographic and school characteristics have little impact of the
probability of full- time schooling.
· Girls are less likely to be in full-time school in male- headed households than in
female headed households.
· The income and wage elasticities of supply of labor by a working boy are very
small. For example, a doubling of household income net of a working child’s
earnings decreases hours worked by rural boys by only 1.9 percent.
· An increase in one year of age raises hours worked by 7.6 percent (3.4 hours) for
urban boys and 3.6 percent (1.5 hours) at the sample mean.
Discussion: As in other studies, parents with a family enterprise employ their own
children rather than school them. This result is significant because one of the strategies
that families have used to cope with the stresses of reform is to open household
enterprises. This may have the effect of increasing child labor as parents employ their
own children to run the business.
Author: Edmonds and Turk (2002).
Hypothesis: Rural reforms, removal of subsidies for capital intensive industry and the
new Enterprise Law of 2000 may all increase labor intensive production. The consequent
increase in the demand for unskilled labor may increase child labor. However, the rise in
68
income and the 1991 legislation that mandates and provides for primary school education
may lower child labor.
Data: Vietnam Living Standards Surveys, 1992/93 and 1997/98.
Dependant Variables: Children ages six to fifteen working for pay; working in family
agriculture; working in a household run business.
Results: Girls are more likely to work than boys, particularly for families under great
financial stress. Child labor is higher in rural settings. Ethnic minority children are more
likely to work than non- minority children. Children in families that have recently
migrated from a rural setting to an urban setting are more likely to work. Children who
work in newly established household enterprises are more likely to work than other
children.
Discussion: This paper finds that the rise in living standards have played an important
role in reducing child labor in the 1990s. However, the study does not control for family
characteristics, etc., that may be correlated with household income.
Author: Ray (1999)
Hypothsis: There is a positive association between poverty and child labor and a negative
association between poverty and schooling.
Data: Peru Living Standards Measurement Survey in 1994 and the Pakistan Integrated
Household Survey of 1991.
Dependant Variable: Child labor hours of paid worked, years of schooling
Results:
· A previously non-poor Pakistani family will increase the hours worked by each
child in the household by 500/year if the family drops below the poverty line.
However, the poverty variable is weak and statistically insignificant for Peru.
· In both countries, urban children work less than rural children.
· Parental education status has a significant negative impact on child labor supply,
with the effect being considerably more pronounced in Pakistan.
· Child labor supply is increasing in the child wage in both countries.
· In Peru, a rise in the adult wage lowers the labor supply of boys and girls.
· In Pakistan, a rise in the mother’s wage raises the hours worked by her daughters.
Note that this is market work, not household work.
· When household chores performed by girls are included in the definition of the
supply of labor, the poverty hypothesis fails for Pakistan. Dropping below the
poverty line leads Pakistani girls to reduce their time spent performing household
chores and increase market work. However, the total number of hours does not
rise.
69
· Falling below the poverty line does not reduce years of schooling for Peruvian
girls but does reduce years of schooling for Pakistani children. The schooling of
Pakistani girls is more negatively impacted by poverty than for boys.
· Educated parents are more likely to keep their children in school than uneducated
parents.
· Rising child wages reduce schooling in Pakistan but not in Peru.
· A rise in a Pakistani mother’s wage reduces schooling for her daughter. The
evidence appears to suggest that when a mother’s work opportunities improve,
she pulls her daughter out of school and takes the girl to work alongside her.
Discussion: In the author’s view, Peruvian children are not as adversely affected by
poverty as Pakistani children in part due to the relatively high value placed on education
by Peruvian parents and the ability of Peruvian children to combine work and school.
70
References
Acemoglu, Daron and Joshua Angrist. 1999. “How Large are the Social Returns to
Education? Evidence from Compulsory Schooling Laws.” Working Paper 7444.
Cambridge, MA: National Bureau of Economic Research, December.
Aderinto, Adeyinka A. 2000. “Social Correlates and Coping Measures of Street-Children:
A Cmparative Study of Street and Non-Street Children in South-Western Nigeria.”
Child Abuse and Neglect, vol. 24, no. 9, pp. 1199-1213.
Admassie, A. 2002. “Child Labour and Schooling in the Context of a Subsistence Rural
Economy: Can They Be Compatible?” International Journal of Education
Development, forthcoming.
Ahlburg, Dennis A., Brian P. McCall, and Ragui Assad. 2001. “Schooling in Egypt; The
Impact of Delayed Entry and Working while in School.” Economic Policy Initiative
Consortium, Final Report. University of Minnesota.
Ahmed, Aftikhar. 2000. “Getting Rid of Child Labour.” ILO/IPEC Working Paper.
Geneva: International Labour Office.
Alderman, Harold. 1996. “The Returns to Endogenous Houman Capital in Pakistan’s
Rural Wage Labour Market.” Oxford Bulletin of Economic Statistics, vol. 58, no. 1,
pp. 229-254
Alderman, Harold and Elizabeth M. King. 1998. “Gender Differences in Parental
Investment in Education.” Structural Change and Economic Dynamics, vol. 9, no. 4
(December), pp. 453-468.
Angrist, Joshua and Victor Lavy. 1997. “The Effect of A Change in Language of
Instruction on the Returns to Schooling in Morocco.” Journal of Labor Economics,
vol. 15, no. 1, S48-76.
Angrist, Joshua and Victor Lavy. 1999. “Using Maimonides’ Rule to Estimate the Effect
of Class Size on Scholastic Achievement.” Quarterly Journal of Economics, vol. 114,
no. 2, pp. 533-575.
Anker, Richard. 2000. “Conceptual and Research Frameworks for the Economics of
Child Labour and its Elimination.” IPEC Working Paper. Geneva: International
Labour Office.
Aruga, Natsuki. 1988. “’An’ Finish School’: Child Labor During World War II.” Labor
History, vol. 29, no.4 (Fall), pp. 498-530.
71
Bachman, Sarah L. 2000. “The Political Economy of Child Labor and Its Impacts on
International Business.” Business Economics, vol. 35, no. 3 (July), pp. 30-41.
Baland, Jean-Marie and James A. Robinson. 2000. “Is Child Labor Inefficient?” Journal
of Political Economy, vol. 108, no. 4, pp. 663-679.
Basu, Alaka M. 1993. Family Size and Child Welfare in an Urban Slum: Some
Disadvantages of Being Poor but Modern. In, Cynthia B. Lloyd, Fertility , Family
Size and Structure. NY: Population Council.
Basu, Arnab K. and Nancy H. Chau. 2002. “Exploitation of Child Labor and the
Dynamics of Debt Bondage.” Working Paper 2002-06. Ithaca, NY: Department of
Applied Economics and Management, Cornell University.
Basu, Kaushik. 1997. Analytical Development Economics. Cambridge, MA: The MIT
Press.
Basu, Kaushik. 1999a. “Child Labor: Cause, Consequence, and Cure, with Remarks on
International Labor Standards.” Journal of Economic Literature, vol. 37, no. 1
(September), pp. 1083-1119.
Basu, Kaushik. 1999b. “International Labor Standards and Child Labor.” Challenge, vol.
42, no. 5 (September-October), pp. 80-93.
Basu, Kaushik. 2000. “The Intriguing Relation between Adult Minimum Wage and Child
Labour.” Economic Journal, vol. 110, no. 462 (March), pp. C50-61.
Basu, Kaushik. 2002. “A Note on Multiple General Equilibria with Child Labor.”
Economics Letters, vol. 74, issue 3, February, pp. 301-308.
Basu, Kaushik, Ambar Narayan, and Martin Ravallion. 2001. “Is Knowledge Shared
Within Households?” Labour Economics, vol. 8, no. 6 (December), pp. 649-665.
Basu, Kaushik and Ranjan Ray. 2001. “The Collective Model of the Household and an
Unexpected Implication for Child Labor: Hypothesis and an Empirical Test.”
Working Paper 2813. Washington, D.C.: The World Bank.
Basu, Kaushik and Phang H. Van. 1998. “The Economics of Child Labor.” American
Economic Review, vol. 88, pp. 412-427.
Becker, Gary 1993. Human Capital. Chicago: University of Chicago Press.
Becker, Gary. 1974. “A Theory of Social Interaction.” Journal of Political Economy, vol.
82, pp. 1063-1094.
Becker, Gary. 1981. A Treatise on the Family. Cambridge: Harvard University Press.
72
Behrman, Jere R., Suzanne Duryea, and Miguel Szekely. 1999. “Schooling Investments
and Aggregate Conditions : A Household Survey-Based Approach for Latin America
and the Caribbean.” Inter-American Development Bank, Research Department,
Working Paper no. 407, Washington, D.C.
Behrman, Jere R. and James C. Knowles. 1999. “Household Income and Child Schooling
in Vietnam.” The World Bank Economic Review, vol. 13, no. 2, pp. 211-256.
Berg, Maxine and Pat Hudson. 1992. “Rehabilitating the Industrial Revolution.” The
Economic History Review, New Series, vol. 45, Issue 1, February, pp. 24-50.
Birdsall, Nancy. 1991. “Birth Order Effects and Time Allocation.” In T. Paul Schultz
(ed.), Research in Population Economics. A Research Annual, vol. 7. Greenwich, CN
and London: JAI Press.
Blunch, Niels-Hugo and Dorte Verner. 2000. “Revisiting the Link between Poverty and
Child Labor: The Ghanaian Experience.” Working Paper 2488, October. Washington,
D.C.: The World Bank.
Boch, John. 2002. “Evolutionary Demography and Intra-household Time Allocation:
School Attendance and Child Labor Among the Okavango Delta Peoples of
Botswana.” American Journal of Human Biology, vol. 14, no. 2, pp. 206-221.
Bommier, Antoine and Pierre Dubois. 2002. “Rotten Parents and Child Labor.” Cahier de
Recherche 2002-01, Institut National de la Recherche Agronomique, I’inra de
Toulouse.
Bourguignon, Francois and Pierre-Andre Chiappori. 1994. “The Collective Approach the
Household Behavior.” In Richard Bludell, Ian Preston, and Ian Walker, eds., The
Measurement of Household Welfare. U.K.: Cambridge University Press.
Boyden, Jo, and Deborah Levison. 2000. “Children as Economic and Social Actors in the
Development Process.” Working Paper 2000:1. Stockholm: EGDI.
Boyden, Jo, B. Ling, and W. Myers. 1998. What Works for Working Children. Florence
and Stockholm: UNICEF International Child Development Centre.
Brezis, Elise S. 2001. “Long-Run Growth and Demographic Transition; Social Classes,
Demographic Transition and Economic Growth.” European Economic Review, vol.
45, pp. 707-717.
Brown, Martin, Jens Christiansen, and Peter Philips. 1992. “The Decline of Child Labor
in the U.S. Fruit and Vegetable Canning Industry: Law or Economics?” Business
History Review, vol. 66, Winter, pp. 723-770.
73
Burra, Neera. 1995. Born to Work. Delhi: Oxford University Press.
Cameron, Lisa A. 2002. “The Impact of the Indonesian Financial Crisis on Children:
Data from 100 Villages Survey.” Victoria, Australia: University of Melbourne,
March.
Camps-Cura, Enriqueta. 1998. “Transitions in Women’s and Children’s Work Patterns
and Implications for the Study of Family Income and Household Structure: A Case
Study from the Catalan Textile Sector (1850-1925). The History of the Family, vol. 3,
no. 2, pp. 137-153.
Canagarajah, Sudharshan and Harold Coulombe. 1997. “Child labor and Schooling in
Ghana.” Working Paper 1844, November. Washington, D.C.: The World Bank.
Canagarajah, Sudharshan and Saji Thomas. 1997. “Ghana’s Labor Market (1987-92).”
Working Paper 1752, April. Washington, D.C.: The World Bank.
Card, David and Alan Krueger. 1996. “School Resources and Student Outcomes: An
Overview of the Literature and new Evidence from North and South Carolina.”
Working Paper 5708. Cambridge, MA: National Bureau of Economic Research.
Case, Anne and Angus Deaton. 1999. “School Inputs and Educational Outcomes in South
Africa.” Quarterly Journal of Economics, vol. 114, no. 3, pp. 1047-1084.
Chernichovsky, D. 1985. “Socioeconomic and Demographic Aspects of School
Enrollment and Attendance in Rural Botswana.” Economic Development and
Cultural Change, vol. 13, no. 4, pp. 551-563.
Chiswick, Barry R. 1986. “Labor Supply and Investment in Child Quality: A Study of
Jewish and Non-Jewish Women.” The Review of Economics and Statistics, vol. 68,
issue 4, November, pp. 700-703.
Chiswick, Barry R. 1992. “Labor Supply and Investment in Child Quality: A Study of
Jewish and Non-Jewish Women: A Reply.” The Review of Economics and Statistics,
vol. 74, issue 4, November, pp. 726-727.
Cigno, Alessandro, Rufio C. Rosati, and Lorenzo Guarcello. 2002. “Does Globalization
Increase Child Labor? World Development, forthcoming.
Cigno, Alessandro and Furio C. Rosati. 2000. “Why do Indian Children Work, and is it
Bad for Them?” Discussion Paper no. 115. Bonn: IZA, February.
Coady, David P. and Susan W. Parker. 2002. “A Cost-Effectiveness Analysis of
Demand- and Supply-Side Education Interventions: The Case of PROGRESA in
Mexico.” FCND Discussion Paper No. 127, March. Washington D.C.: International
Food Policy Research Institute.
74
Cogan, John. 1982. “The Decline in Black Teenage Employment: 1950-70.” American
Economic Review, vol. 72, no. 4 (September), pp. 621-638.
Cohen, B and W. House. 1994. Education, Experience and Earnings in the Labor Market
of a Developing Country: The Case of Urban Khartoum. World Development, vol. 22,
pp. 1549-1565.
Connelly, Rachel, Deborah S. DeGraff, and Deborah Levison. 1996. “Women’s
Employment and Child Care in Brazil.” Journal of Development and Cultural
Change, vol. 44, no. 3 (April), pp. 619-656.
Cunningham, Hugh. 2000. “The Decline of Child Labour: Labour Markets and Faminly
Economies in Europe and North America Since 1830.” Economic History Review,
vol. 53, no. 3, pp. 409-428.
Dehejia, Rajeev H. and Roberta Gatti. 2002. “Child Labor: The Role of Income
Variability and Access to Credit across Countries.” Working Paper 9018, June.
Cambridge, MA: National Bureau of Economic Research.
Dessy, Sylvain E. 2001. “A Defense of Compulsive Measures against Child Labor.”
Journal of Development Economics, vol. 62, no. 1 (June), pp. 261-275.
Dessy, Sylvain E. and Stéphane Pallage. 2001. “Child Labor and Failures.” Journal of
Development Economics, vol. 65, pp. 469-476.
Diamond, Charles and Tammy Fayed. 1998. “Evidence on Substitutability of Adult and
Child Labour.” The Journal of Development Studies, vol. 34, no. 3 (February), pp. 62-
70.
Dorman, Peter. 2001. Child Labour in the Developed Economies.” ILO/IPEC Working
Paper. Geneva: International Labour Office.
Drèze, Jean and Geeta Gandhi Kingdon. 2001. “School Participation in Rural India.”
Review of Development Economics, vol. 5, no. 1 (February), pp. 1-24.
Duflo, Esther. 2001. “Schooling and Labor Market Consequences of School Construction
in Indonesia: Evidence from an Unusual Policy Experiment.” American Economic
Review, vol. 91, no. 4 (September), pp. 795-813.
Duryea, Suzanne. 1998. “Children’s Advancement Through School in Brazil: The Role
of Transitory Shock to Household Income.” Inter-American Development Bank,
Office of the Chief Economist, Working Paper no. 376 (July), Washington, D.C.
75
Duryea, Suzanne and Mary Arends-Kuenning. 2001. “School Attendance, Child Labor
and Local Labor Markets in Urban Brazil.” International Development Bank,
November.
Edmonds, Eric and Nina Pavcnik. 2002. “Does Globalization Increase Child Labor?
Evidence from Vietnam. Working Paper 8760, February. Cambridge, MA: National
Bureau of Economic Research.
Edmonds, Eric and Carrie Turk. 2002. “Child Labor in Transition in Vietnam.” Working
Paper 2774, February. Washington, D.C.: The World Bank.
Ejrnæs, Mette and Claus Chr. Pörtner. 2002. “Birth Order and the Intrahousehold
Allocation of Time and Education.” University of Copenhagen, May.
Emerson, Patrick M. and André Portela Souza. 2002. “Birth Order, Child Labor and
School Attendance in Brazil, manuscript, May.
Felt, Jeremy P. 1970. “The Child Labor Provisions of the Fair Labor Standards Act.”
Labor History, vol. 11, no. 4 (Fall), pp. 467-481.
Fishback, Price V. 1998. “Operations of ‘Unfettered’ Labor Markets: Exit and Voice in
American Labor Markets at the Turn of the Century.” Journal of Economic
Literature, vol.36, issue 2, June, pp. 722-765.
Fleisher, Belton and Xiaojun Wang. 2001. “Skill Differentials, Return to Schooling and
Market Segmentation in a Transition Economy: The Case of Mainland China.”
Manuscript, Department of Economics, Ohio State University.
Flug, Karnit, Antonio Spilimbergo, and Erik Washtenheim. 1998. “Investment in
Education: Do Economic Volatility and Credit Constraints Matter?” Journal of
Development Economics, vol. 55, pp. 465-481.
Galbi, Douglas A. 1997. “Child Labor and the Division of Labor in the Early English
Cotton Mills.” Journal of Population Economics, vol. 10, no. 4, pp. 357-375.
Genicot, Garance. 1998. “Malnutrition and Child Labor.” Department of Economics,
University of California, Irvine.
Glewwe, Paul. 2002. “Schools and Skills in Developing Countries: Education Policies
and Socioeconomic Outcomes.” Journal of Economic Literature, vol. 40, June, pp.
436-482.
Glewwe, Paul and Hanan Jacoby. 1994. “Student Achievement and Schooling Choice in
Low-Income Countries: Evidence from Ghana.” Journal of Human Resources, vol.
29, no. 3, pp. 843-864.
76
Goldin, Claudia. 1981. “Family Strategies and the Family Economy in the Late
Nineteenth Century: The Role of Secondary Workers’.” In T. Hershberg, ed.,
Philadelphia: Work, Space, Family and Group Experience in the Nineteenth Century.
Oxford: Oxford University Press, pp. 277-310.
Goldin, Claudia and Lawrence F. Katz. 1999. “Human Capital and Social Capital: The
Rise of Secondary Schooling in America, 1910 to 1940.” Journal of Interdisciplinary
History, vol. 29, no. 4, pp. 683-723.
Goldin, Claudia and Donald O. Parsons. 1981. “Economic Well-Being and Child Labor:
The Interaction of Family and Industry.” Working Paper 707, Cambridge, MA:
National Bureau of Economic Research.
Gunn, Susan E. and Zenaida Ostos. 1992. “Dilemmas in Tackling Child Labour: The
Case of Scavenger Children in the Philippines.” International Labour Review, vol.
131, no. 6, pp. 629-646.
Gupta, Manash R. 2000. “Wage Determination of a Child Worker: A Theoretical
Analysis.” Review of Development Economics, vol. 4, no. 2, June.
Hamermesh, Daniel S. 1993. Labour Demand. Princeton, NJ: Princeton University Press.
Harbison, Ralph and Eric Hanushek. 1992. Educational Performance of the Poor:
Lessons from Rural Northeast Brazil. Washington, D.C.: Oxford University Press for
the World Bank.
Hentschel, Jesko and William F. Waters. 2002. “Rural Poverty in Ecuador: Assessing
Local Realities for the Development of Anti-Poverty Programs.” World Development,
vol. 30, issue 1, January, pp. 33-47.
Heyneman, Stephen P., Dean T. Jamison, and Xenia Montenegro. 1984. “Textbooks in
the Philippines: Evaluation of the Pedagogical Impact of Nationwide Investment.”
Education Evaluation Policy Analysis, vol. 6, no. 2, pp. 139-150.
Holleran, Philip M. 1993. “Child Labor and Exploitation in Turn-of-the-Century Cotton
Mills.” Explorations in Economic History, vol. 30, pp. 485-500.
Horrell, Sara and Jane Humphries. 1995. “The Exploitation of Little Children”: Child
Labor and the Family Economy in the Industrial Revolution.” Explorations in
Economic History, vol. 32, pp. 485-516.
Humphries, Jane. 1981. “Protective Legislation, the Capitalist State and Working-Class
Men; the Case of the 1842 Mines Regulation Act.” Feminist Review, vol. 7, pp. 1-33
77
Ilahi, Nadeem. 2001. “Children’s Work and Schooling: Does Gender Matter? Evidence
from the Peru LSMS Panel Data.” Working Paper 2745, December. Washington,
D.C.: The World Bank.
IFPRI. 2001. “Fourth Report on the PRAF Series. PRAF/IDB Phase II: Analysis of the
Situation before the Beginning of Distribution of Vouchers and Project
Implementation. Washington, D.C.: International Food Policy Research Institute.
Iversen, Vegard. 2002. “Autonomy in Child Labor Migrants.” World Development, vol.
30, no. 5, pp. 817-834.
International Labour Office. 1996. Child Labour: Targeting the Intolerable. Geneva:
International Labour Office Publications.
Jacoby, Hanan G. and Emmanuel Skoufias. 1997. “Risk, Financial Markets, and Human
Capital in a Developing Country.” Review of Economic Studies, vol. 64, no. 3, pp.
311-335.
Jafarey, Saqib and Sajal Lahiri. 2002. “Will Trade Sanctions Reduce Child Labour? The
Role of Credit Markets.” Journal of Development Economics, vol. 68, pp. 137-156.
Jamison, Dean, Barbara Searle, Klaus Galda, and Stephen Heyneman. 1981. ‘Improving
Elementary Mathematics Education in Nicaragua: An Experimental Study of the
Impact of Textbooks and Radio on Achievement.” Journal of Education Psychology,
vol. 73, no. 4, pp. 556-567.
Jensen, Peter and Helena Skyt Nielsen. 1997. “Child Labour or School Attendance?
Evidence from Zambia.” Journal of Population Economics, vol. 10, no. 4, pp. 407-
424.
Jolliffe, Dean. 1998. “Skills, Schooling and Household Income in Ghana.” World Bank
Economic Review, vol. 12, no. 1, pp. 81-104.
Kabeer, Naila. 2000. “Inter-Generational Contracts, Demographic Transitions and the
‘Quantity-Quality’ Tradeoff: Parents, Children and Investing in the Future.” Journal
of International Development, vol. 12, no. 4 (May), pp. 463-482.
Kanbur, R. and Haddad, L. 1994. “Are Better-off Households more Equa l or Less
Equal?” Oxford Economic Papers, vol. 46, pp. 445-458.
Kassouf, Ana Lúcia, Martin McKee, and Elias Massialos. 2001. “Early Entrance to the
JobMarket and Its Effect on Adult Health: Evidence from Brazil.” Health Policy and
Planning, vol. 16, no. 1, pp. 21-28.
78
Kindgon, Geeta. 1996. “The Quality and Efficiency of Private and Public Education: A
Case Study of Urban India.” Oxford Bulletin of Economic Statistics, vol. 58, no. 1,
pp. 57-82.
Kolk, Ans and Rob van Tulder. 2002. “The Effectiveness of Self- Regulation: Corporate
Codes of Conduct and Child Labour.” European Management Journal, vol. 20, no. 3,
pp. 260-271.
Krase, Douglas and Douglas Mahony. 1998. “Illegal Child Labor in the United States:
Prevalence and Characteristics.” Working Paper 6479. Cambridge, MA: National
Bureau of Economic Research.
Krueger, Alan B. 1997. “Observations on International Labor Standards and Trade.” In
M. Bruno and B. Pleskovic (eds.), Annual World Bank Conference on Development
Economics. Washington, D.C.: The World Bank.
Laitner, John. 1997. “Intergenerational and Interhousehold Economic Links.” In Mark R.
Rosenzweig and Oded Stark, eds., Handbook of Population and Family Economics,
vol. 1A. Amsterdam: Elsevier, pp. 189-238
Lea, Arden J. 1975. “Cotton Textiles and the Federal Child Labor Act of 1916.” Labor
History, vol 16., no. 4 (Fall), pp. 484-494.
Lee, Jong-Wha and Robert J. Barro. 1997. “Schooling Quality in a Cross-Section of
Countries, Working Paper 6198. Cambridge: National Bureau of Economics
Research.
Leinback, Thomas R. and Adrian Smith. 1994. “Off-Farm Employment, Land and Life
Cycle: Transmigrant Households in South Sumatra, Indonesia.” Economic
Geography, vol. 70, issue 3, July, pp. 273-296.
Levine, James A., Robert Weisell, and Simon Chevassus. 2002. “Looking at Child
Labor.” Science, vol. 296, no. 5570, May 10, pp. 1025-1026.
Levison, Deboray and Karine S. Moe. 2001. “Youth Education and Work in Mexico.”
World Development, vol. 29, no. 1, pp. 167-188.
Levy, Victor. 1985. “Cropping Pattern, Mechanization, Child Labor and Fertility
Behavior in a Farming Economy: Rural Egypt.” Economic Development and Cultural
Change, vol. 33, no.4, pp. 777-791.
Lim, Joseph Y. 2000. “The East Asian Crisis and Child Labour in the Philippines.”
ILO/IPEC Working Paper. Geneva: International Labour Office.
79
Lleras-Muney, Adriana. 2001. “Were Compulsory Attendance and Child Labor Laws
Effective? An Analysis from 1915 to 1939.” New York, NY: Columbia University,
April.
Lokshin, Michael M., Elena Glinskaya, and Marito Garcia. 2000. “The Effect of Early
Childhood Development Programs on Women’s Labor Force Participation and Older
Children’s Schooling in Kenya.” Working Paper 2376, June. Washington, D.C.: The
World Bank.
L?pez-Calva, Luis Felipe, and Koji Miyamoto. 1999. “Filial Obligations and Child
Labor.” Manuscript, Centro de Estudios Econ?micos, El Colegio de México.
Manning, Chris. 2000. “The Economic Crisis and Child Labour in Indonesia.” ILO/IPEC
Working Paper. Geneva: International Labour Office.
Margo, Robert A. and T. Aldrich Finegan. 1996. “Compulsory Schooling Legislation and
School Attendance in Turn-of-the Century America: A “Natural Experiment’
Approach. Historical Paper 89. Cambridge, MA: National Bureau of Economic
Research.
Maskus, Keith. 1997. “Core Labor Standards: Trade Impacts and Implications for
International Trade Policy.” World Bank International Trade Division, mimeo.
Marx, Karl. 1867. Capital: A Critique of Political Economy.
Mitchell, Daniel J.B. 1980. “The Impact of Child Labor Laws on the Kinds of Jobs Held
by Young School Leavers.” Journal of Human Resources, vol. 15, no. 3, pp. 396-408.
Moehling, Carolyn. 1998. “State Child Labor Laws and the Decline of Child Labor.”
Explorations in Economic History, vol. 36, no. 1 (January), pp. 72-106.
Moehling, Carolyn. 1995. “The Intrahousehold Allocation of Resources and the
Participation of Children in Household Decision-Making: Evidence from Early
Twentieth Century America.” Northwestern University, mimeo.
Moll, Peter. 1998. “Primary Schooling, Cognitive Skills and Wages in South Africa.”
Economica, vol. 65, pp. 263-284.
Moore, Karen. 2000. “Supporting Children in their Working Lives: Obstacles and
Opportunities within the International Policy Environment.” Journal of International
Development, vol. 12, no. 4 (May), pp. 531-548.
Nardinelli, Clark. 1980. “Child Labor and the Factory Acts.” The Journal of Economic
History, vol. 40, issue 4, December, pp. 739-755.
80
Nardinelli, Clark. 1990. Child Labor and the Industrial Revolution. Bloomington: Indiana
University Press.
Nardinelli, Clark. 1982. “Corporal Punishment and Children’s Wages in Nineteenth-
Century Britain.” Explorations in Economic History, vol. 19, no. 3, pp. 283-95
Parish, William L. and Robert J. Willis. 1993. “Daughters, Education, and Family
Budgets: Taiwan Experiences.” Journal of Human Resources, vol. 28, no. 4. pp. 863-
898.
Parsons, Donald O. and Claudia Goldin. 1989. “Parental Altruism and Self-Interest:
Child Labor Among Late Nineteenth-Century American Families.” Economic
Inquiry, vol. 27, no. 4, October, pp. 637-659.
Pastore, José. 2000. “Labor Standards and International Trade: The Case of Child Labor
in Brazil. University of São Paulo, Brazil (June).
Patrinos, Harry Anthony and George Psacharopoulos. 1995. “Educational Performance
and Child Labor in Paraguay.” International Journal of Educational Development,
vol. 15, no. 1, pp 47-60.
Patrinos, Harry Anthony and George Psacharopoulos. 1997. “Family Size, Schooling and
Child Labor in Peru – An Empirical Analysis.” Journal of Population Economics,
vol. 10, no. 4, pp. 387-405.
Pitt, Mark M. and Mark R. Rosenzweig. 1990. “Estimating the Intrahousehold Incidence
of Illness: Child Health and Gender-Inequality in the Allocation of Time.
International Economic Review, vol. 31, no. 4 (November), pp. 969-989.
Plank, David N. 1986. “State Action and the Distribution of School Enrollments in
Brazil, 1970.” Economics of Education Review, vol. 5, no. 4, pp. 403-414.
Pörtner, Claus Chr., 2001a. “Children and Household Behavior in Developing
Countries.” University of Copenhagen, July.
Pörtner, Claus Chr. 2001b. “Children as Insurance.” Journal of Population Economics,
vol. 14, no. 1, pp. 119-136.
Pörtner, Claus Chr. 2001c. “Children’s Time Allocation in the Laguna Province, the
Philippines.” The World Bank, September.
Psacharopoulos, George. 1997. “Child Labor versus Educational Attainment. Some
Evidence from Latin America.” Journal of Population Economics, vol. 10, no. 4, pp.
377-386.
81
Rahikainen, Marjatta. 2001. “Children and ‘the Right to Factory Work’: Child Labour
Legislation in Nineteenth-Century Finland.” Scandinavian Economic History Review,
vol. 49, no. 1, pp. 41-62.
Ranjan, Priya. 2001. “Credit Constraints and the Phenomenon of Child Labor.” Journal
of Development Economics, vol. 64, issue 1, pp. 81-102.
Ranjan, Priya. 1999.” An Economic Analysis of Child Labor.” Economics Letters, vol.
64, no. 1 (July), pp. 99-105,
Ravallion, Martin and Quentin Wodon. 1999. “Does Child Labor Displace Schooling”
Evidence on Behavioral Responses to and Enrollment Subsidy.” Working Paper
2116, May. Washington, D.C.: The World Bank.
Ray, Ranjan. 2000. “Analysis of Child Labor in Peru and Pakistan: A Comparative
Study.” Journal of Population Economics, vol. 13, pp. 3-19.
Ray, Ranjan. 1999. “How Child Labour and Child Schooling Interact with Adult
Labour.” Working Paper 2179, September. Washington, D.C.: The World Bank.
Riney-Kehrberg, Pamela. 2001. “The Limits of Policy: Rural Children and Work in the
United States and New Zealand, 1870-1920.” History of the Family, vol. 6, issue 1,
pp. 51-67.
Robles-Vásquez, Héctor and David G. Abler. 2000. “Education and Labor Force
Participation by Mexican Children during Structural Adjustment: A Microeconomic
Analysis.” Working Paper 00-05.University Park, PN: Population Research Institute,
April.
Rogers, Carol Ann and Kenneth A. Swinnerton. 2000. “Inequality, Productivity and
Child Labor: Theory and Evidence.” Georgetown University, manuscript.
Rose, Jim. 1998. “Child Labor, Family Income, and the Uruguay Round.” Quarterly
Journal of Austrian Economics, vol. 1, no. 4 (Winter), pp. 75-87.
Rosenzweig, M. R. and R.E. Evensson. 1977. “Fertility, Schooling, and the Economic
Contribution of Children in Rural India: An Econometric Analysis.” Econometrica,
vol. 45, no. 5, pp. 1065-1079.
Rosenzweig, Market R. and T. Paul Schultz. 1989. “Schooling, Information and
Nonmarket Productivity: Contraceptive Use and Its Effectiveness.” International
Economic Review, vol. 30, no. 2 (May), pp. 457-477.
Sadoulet, Elisabeth, Federico Finan, and Alain de Janvry. 2001. “How Effective are
Educational Subsidies Programs for the Rural Poor? Progresa in Mexico.” Berkley,
CA: University of California, manuscript.
82
Saha, B. and S. Sarkar. 1999. “Schooling, Informal Experience, and Formal Sector
Earnings: A Study of Indian Workers. Review of Development Economics, vol. 3, pp.
187-199.
Sawada, Uasuyuki and Michael Lokshin. 2001. “Household Schooling Decisions in Rural
Pakistan.” Working Paper 2541. Washington, D.C.: The World Bank.
Schultz, T. Paul. 1997. “Demand for Children in Low Income Countries.” In Mark R.
Rosenzweig and Oded Stark, eds., Handbook of Population and Family Economics,
vol. 1A. Amsterdam: Elsevier, pp. 349-430.
Shelburne, Robert C. 2001. “An Explanation of International Variation in the Prevalence
of Child Labour.” World Economy, vol. 24 (March), pp. 359-378.
Sharif, Mahammed and Ranjan K. Saha. 1993. “The Observed Landholding-Fertility
Relationship – Is It Monotonic?” Journal of Development Studies, vol. 29, no. 2
(January), pp. 319-341.
Skoufias, Emmanuel. 1995. “Household Resources, Transactions Costs, and Adjustment
through Land Tenancy.” Land Economics, vol. 71, no. 1, pp. 42-56.
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.” FCND Discussion Paper No. 123, October. Washington D.C.: International
Food Policy Research Institute.
Skoufias, Emmanuel and Susan W. Parker. 2001. “Labor Market Shocks and their
Impacts on Work and Schooling: Evidence from Urban Mexico.” FCND Discussion
Paper No. 129, March. Washington D.C.: International Food Policy Research
Institute.
Spencer, Byron G. 1992. “Labor Supply and Investment in Child Quality: A Study of
Jewish and Non-Jewish Women: A Comment.” The Review of Economics and
Statistics, vol. 74, issue 4, November, pp. 721-725.
Swaminathan, Madhura. 1998. “Economic Growth and the Persistence of Child Labor:
Evidence from an Indian City.” World Development, vol. 26, no. 8, pp 1513-1528.
The Probe Team. 1999. Public Report on Basic Education in India. New Dehli: Oxford
University Press.
Tuttle, Carolyn. 1999. Hard at Work in Factories and Mines. The Economics of Child
Labor During the British Industrial Revolution. Boulder: Westview Press.
83
U.S. Department of Labor. 1997. By the Sweat and Toil of Children, Volume IV:
Consumer Labels and Child Labor. Washington, D.C.: U.S. Department of Labor,
Bureau of International Labor Affairs.
Walker, Roger W. 1970. “The A.F.L. and Child-Labor Legislation: An Exercise in
Frustration.” Labor History, vol. 11, no. 3 (Summer), pp. 323-340.
Wan, Tianhong, Atshushi Maruyama, and Masao Kikuchi. 2000. “Rural-Urban Migration
and Labor Markets in China: A Case Study in a Northeastern Province. The
Developing Economies, vol. 38, no. 1, pp. 80-104.