Вы находитесь на странице: 1из 23

Child Labor and Schooling in the Philippines*

Chris Sakellariou
School of Humanities and Social Science,
Nanyang Technological University, Singapore
E-mail: acsake@ntu.edu.sg

1. Introduction

In June 1999, the International Labor Organization (ILO) adopted a new convention on child
labor – No. 182, moving away from minimum age standards and an overall aggressive stance
against child labor, apparently toward a gradual approach targeting hazardous work
conditions, long hours, and in general eliminating the worst of child labor. This could be
related to some recent studies advocating a gradual and multi-faceted approach to increase
schooling attendance and make-up for short-term financial losses to the family (such as
Grootaert and Patrinos 1999).

Child labor is prevalent throughout the developing world. ILO estimates indicate that
between 1990 and 1997, the number of children working full-time increased from about 78
million to 120 million. Estimates of the number of economically active children (either full-
time or part-time) varied from 250 million (ILO 1997) to about 400 million (UNICEF 1997).
Less than 5 percent of the world’s “working” children produce goods for the export sector
(Boyden et al 1998); nevertheless, this sector is usually targeted by trade bills or threat of
sanctions. As a result, children are often pushed into work under even more hazardous
conditions.

More importantly, the definition of child labor is plagued by at least one problem. It relates to
the need for distinguishing between the “bad” child labor, which is performed at the expense
of education, and the “less harmful” child work, which includes activities that can be
combined with schooling, such as household activities. An empirical investigation should be
distinguishing between the two, since child labor is not necessarily “bad” and policy
intervention may lead to an inferior outcome. However, when work activity is at the expense
of a child’s human capital accumulation, it is easier to argue in favor of policy intervention.

There has been an increasing interest in the study of child labor especially in the 1990s,
however, the overwhelming majority of empirical studies are for African (for example,
Canagarajah and Coulombe 1997; Jensen and Nielsen 1997; Grootaert 1999) and
South American countries (for example, Levison 1991; Patrinos and Psacharopoulos 1995;
1997, Cartwright and Patrinos 1999). Some new nationwide studies exist for South Asia (Ray
2000; Maitra and Ray 2002; Dreze and Gandhi-Kingdom 2001; Cockburn 2001), but very
few for South-East Asia, with the exception of the ones by Sakellariou & Lall (2000) for the

* Unpublished paper, 2004.


** The author would like to thank the Department of Economics, De La Salle University for making the
data used in this paper available.

http: www.childprotection.org.ph
Featured Paper for June 2004
Philippines, using less than ideal data from the 1991/92 Labor Force and Family
Expenditure1,2.

In this study I use a recent, large, nationwide representative survey (1999 Annual Poverty
Indicator Survey) to analyze child time allocation decisions within households. Attention is
paid to the effects of poverty, fertility and composition of the household, the opportunity cost
of schooling, and an attempt is made to incorporate demand factors into the model. The data
allows for the distinction between children who only school and children who combine
school and work, as well as the distinction between paid and unpaid work; however the latter
distinction is not incorporated into the model due to the small number of children, once the
children combining school and work are separated according to age group and type of work
activity.

Section 2 provides some background information, while section 3 discusses the theoretical
framework and the determinants of child labor. Section 4 discusses the data, along with
descriptive statistics on children’s activity. Section 5 presents the model and issues related to
what model is appropriate in modeling child time allocation decisions, followed by a
discussion of the empirical results in Section 6. Finally, in Section 7 I present the main
findings and policy implications.

2. The Case of the Philippines

The Philippines has free public education to grade 6, but only 88 percent of children in the
under-12 age group and 83 percent of children in the 6-17 age group were enrolled in 1999
(1999 Annual Poverty Incidence Survey), and an even smaller percentage actually attended.
Many of these children eventually drop out. School enrollment has increased moderately
compared to a decade earlier, when 82 percent of children under 12 attended school (1990
Census).

According to UNICEF estimates, about 3.7 million children worked in 1988, and their
number increased to between 5 and 7 million in 1994 (Sancho-Liao 1994). Children work in
both the formal and informal sectors. About 50 percent of plantation workers in the vegetable
industry (such as that of Benguet) are children. Children are also employed in the agricultural
1
The data had two main drawbacks: it did not allow the identification of children combining work and school
and did not cover children under 10 years old.
2
Some narrowly focused studies exist, such as those by Suzan Gunn and Zenaida Ostos, “Dilemmas in
Tackling Child Labor: the Case of Scavenger Children in the Philippines”, International Labor Review 131
(1992):629-646, G. Mergos, “The Economic Contribution of Children in Peasant Agriculture and the Effect of
Education: Evidence from the Philippines”, Pakistan Economic Review 31 (1992):189-201 and Nelia Sancho-
Liao, “Child Labor in the Philippines: Exploitation in the Process of Globalization of the Economy”, Labor,
Capital and Society 27 (1994): 270-281 for the Philippines, and D. DeTray, “Children’s Work Activities in
Malaysia”, Population and Development Review 9 (1983): 437-455.

http: www.childprotection.org.ph 2
Featured Paper for June 2004
sector for wages that are half of the adult wage. The garment industry employs a large
number of girls between the ages of 5 and 14 trough a system of sub-contractors.

The minimum age for general employment in the Philippines is 15 years (Child Protection
Act of 1992 and Republic Act No. 7610), however, children under 15 may be employed
provided that the employer secures a work permit from the Department of Labor and
Employment and the protection, health, safety and morals of the child can be ensured. The
Republic Act No. 7658 amended Act No. 7610 by prohibiting children below 15 years from
employment, except when they work in a family-run business (Reuter Library Report 1993).
The Bureau of Women and Young Workers is in charge of enforcing child labor laws. This
Bureau coordinates and collaborates with governmental agencies and non-governmental
organizations, but has no inspectors. It depends on inspectors from the Labor Standards
Division and the Welfare Division. However, the number of inspectors is very small, making
the task of monitoring of child labor laws particularly difficult. There are other problems,
such as gaps in the laws governing child labor; while the labor code protects children
working in the industry, the majority of children working in agriculture, small industries and
domestic work settings are left unprotected (Sakellariou and Lall 2000). Finally, while the
basic laws are in place, there is a lack of an effective mechanism to constantly monitor the
situation of child labor and act for the protection of children. There is a tendency for action to
depend on a complaint being filed first, rather than independent action by the monitoring
departments (Institute for Labor Studies, Manila 1994).
3. Theoretical Framework and Determinants

Framework

From the human capital theory perspective, the decision of child school attendance versus
child work is affected by the perceived return to investing in a child’s education in
comparison to investing in other assets. High direct and indirect costs of education, in
relation to family income, lower the perceived returns to children’s education. Likewise, low
quality of education will also lower the perceived benefits of children’s education. Poverty,
therefore, may be considered the most important determinant of school attendance versus
work activity. Because of insufficient family income, cost of schooling and other constraints,
the household may be compelled to keep children away from school.

Fertility is a determinant of child labor, since the number of children in the family determines
the supply of potential workers in the family. Of particular importance is the trade-off
between quality and quantity of children, which implies that we cannot consider
schooling/work decisions without taking fertility into account (Becker 1991; Jensen and
Nielsen 1997), irrespectively of whether schooling and fertility are simultaneously3 modeled
or the household size is assumed to be exogenous to the schooling decision. Evidence from
both developed and developing countries points to a trade-off between number of children
and educational attainment (for example, Knodel et al 1990; Knodel and Wongsith 1991).

3
Rosenzweig and Evenson (1977), examined simultaneously decisions which are jointly associated with child
investment, i.e., family size, schooling, and child labor force participation.

http: www.childprotection.org.ph 3
Featured Paper for June 2004
The underlying mechanism for this effect seems to be the dilution of resources within the
family that are available for each child (see, for example, Patrinos and Psacharopoulos 1997).

The interaction between schooling and fertility could, however, lead to higher levels of
education for the individual child. For example, Chernichovsky in his study of rural
Botswana (1985) found that to be the case. That is, the larger the number of children in the
household the higher the levels of education for the individual child. He argued that there
might be diminishing returns to labor in the household, which reduces the indirect costs of
schooling. The result could be specialization and role assignment within the household.
Levison (1991) also found evidence of specialization within the household for rural Brazil as
well as Collings (1983) for indigenous people in rural Peru. Cultural values are also of
importance. In the study by Collins for example, the incidence of children combining work
and schooling for the Aymara people in highland rural Peru is high, as the Aymara value
education and, given the resource constraints, families channel resources to educating at least
a subset of children in the family.

Empirical evidence suggests that the observed magnitude of the effect of household size on
child labor depends on a combination of factors such as family culture, sibling sex, age and
activities, as well as birth order of children. So, the effect of family size must be examined in
relation to the characteristics and activity of siblings. For example, DeGraff et al (1992)
reported that in the Philippines, the relationship between household size and child work
depends on the sex and birth order of the child, and the presence of older siblings decreases
the likelihood of paid work by a child (suggesting a substitution effect), however, no such
effect was found in the case of domestic work (see also, Grootaert and Kampur 1995).

Determinants of child labor and schooling

The determinants of schooling/child work decision can be grouped into 5 sets of


characteristics: (1) Child characteristics; (2) Parent characteristics; (3) Household
characteristics; (4) Cost of schooling; (5) Demand factors. While for (1) to (4) there is a wide
array of variables that can be employed using a large, multipurpose, nationally representative
household survey, information on employment opportunities of children is not included in
such surveys.

Relevant child characteristics are the child’s age and its square, as well as the sex of the
child. While we can be fairly certain that the probability of child labor is increasing with the
child’s age, the sign and magnitude of the effect of the gender of the child depends on
cultural factors and demand conditions.

Both father’s and mother’s characteristics are important determinants. Here we focus on their
education and employment status. It is expected that the probability of child labor
(probability of the child being enrolled in school) is decreasing (increasing) with parents’
level of education. The effect of parents’ employment status (especially the father’s) on child
activity, besides the income effect, is affected by the nature of their employment (such as
being self-employed), among other things. On the other hand the effects of interaction terms
with the sex of the child are ambiguous, as they depend on cultural and other factors.

http: www.childprotection.org.ph 4
Featured Paper for June 2004
Household characteristics relate to: (1) certain characteristics of the head of household such
as age and gender; (2) household characteristics such as purchasing power and ownership of
a business or farm; (3) the size and composition of the household, such as number, gender
and age composition of children in the household. In households headed by women there are
usually less resources available, however, the effect on child labor is ambiguous as families
headed by women tend to be smaller, which increases the amount of resources available per
household member.

Children in poor households are expected to have a higher probability of engaging in market
employment. The link between poverty and child labor has been established in several
studies, such as the comparative study covering four countries (Grootaert and Patrinos 1999)
and by Blunch & Verner (2000), among others. However, in some other empirical studies
this link is questioned (Nielsen 1998; Canagarajah and Coulombe 1997; Sasaki and
Temesgen 1999). The effects of the household owning a family business or agricultural land
are more complex, and the sign of the effect may depend on nature of employment activity as
well as the nature of the business owned. One would probably expect that ownership of
family business or farm would decrease the probability of the child working for wages,
because of an income effect, and would increase the probability of the child working in the
family business or farm. The combination of these two effects is ambiguous; however, it
could result in a lower probability of school enrollment.

The rest of the household characteristics relate to the size and composition of the household.
Although the argument that family size reduces the probability of school enrollment and
increases the probability of child labor is intuitively appealing (a result the often finds
support in empirical studies), it is often found that the number of children in the household
has a positive effect on school enrollment. Chernichovsky (1985) argued that such a positive
effect could be attributed to diminishing returns to labor in the household, resulting in lower
indirect costs of schooling, which, in turn, results to role assignment in the household.
Differentiating the number of siblings by sex and age further enriches the measure of the
substitutability of the child’s work for the work of other family members.

Assuming economic incentives affect demand for schooling, price - represented by the
schooling expenses facing the individual child - enters the demand for schooling function.
Household surveys, however, contain information on education expenses (such as fees paid),
which are incurred only when children are enrolled in school. This problem can, to a large
extent, be overcome by averaging education expenses by community (department) - 148 in
the case of the Philippines. This variable would, then, be expected to reflect the cost of
schooling by community. Likewise, averaging the distance to school by community can
capture the opportunity cost of commuting to school. Given the weakness of the cost of
schooling measures, even if the coefficients are statistically significant, the sign is not a
foregone conclusion. For example, a positive (negative) coefficient in an equation
determining the probability of school enrollment (child labor) would probably be capturing
school quality instead of cost of schooling (for example, see Coleman 1988; Cartwright
1999).

http: www.childprotection.org.ph 5
Featured Paper for June 2004
Capturing the effect of demand factors is the most problematic part of the investigation
because of the absence of data on relevant information such as employment opportunities of
children. There is not much that can be done in that respect, besides using imperfect proxies,
such as a cluster average of children’s wages as a proportion of adult wages across many
communities and location measures, hoping that they capture part of the effect of demand
factors.

4. Data and Preliminary Results

Data

Most previous attempts to study the economics of child labor suffer from some data problem,
be it the lack of national-level data, reliability of the data, number of relevant observations in
the data set, inability to identify the children who combine schooling and work from
schooling only, or the inability to distinguish type of employment (paid, unpaid, employed in
a family business etc). In this study I use the 1999 Annual Poverty Indicator Survey
(APIS99)4, a large, nationally representative survey of households and their members, with
information on different indicators related to poverty and the socio-economic profile of
families and other information relating to their living conditions. It contains detailed
information on the educational attainment of the family members 5 years old and over; it also
determines whether a family member aged between 6-24 years are currently attending formal
school and if not, the reason for not currently attending school. The information on economic
characteristics covers employment status of each family member 5 years old and over (for
different categories of employment, including unpaid household employment), class of
worker, number of days worked during the past quarter and salaries and wages from
employment of each working family member.

A working file on children aged 6 to 17 years was obtained, which contains 57,473
observations of which 55,790 are usable. About 83 percent of children are only attending
school, 3.8 percent are combining school and work and 4.3 percent are only working. The
remaining 9.1 percent are neither in school nor working, and about 50 percent are in the 6 to
8 age bracket. Of those who combined school and work 80 percent worked without pay on a
family operated farm or business. Of those who only worked, about 50 percent worked
without pay on a family operated farm or business.

Preliminary results

A large pool of children in the 6-8 age group do not attend school, and more so in rural areas
(Table 1). School attendance has an inverted U-shape irrespectively of location and gender. It
peaks after the age of 8 years and by the age of 12 years it declines sharply (Chart 1).
Regardless of location, more girls attend school than boys. This has been reported in studies
for South American countries (Cartwright 1999; Turbay and Acuńa 1998). This suggests that
a large number of boys drop out after primary school.

4
Jointly undertaken by the National Statistics Office of the Philippines, the World Bank mission office and the
UNDP.

http: www.childprotection.org.ph 6
Featured Paper for June 2004
Table 1: Schooling/Work Activity by Gender and Location, 1999 (percent)

Age/Activity Boys Girls Urban Rural All


6-8: School only 82.8 83.7 87.5 78.1 83.2
School and work 0.7 0.6 0.3 1.1 0.6
Work only 0.1 0.1 0.0 0.2 0.1
Home care only 16.4 15.7 12.2 20.7 16.0
100 100 100 100 100
9-11: School only 92.2 94.4 95.8 90.2 93.3
School and work 3.3 2.5 1.5 4.6 2.9
Work only 0.8 0.2 0.2 0.9 0.5
Home care only 3.6 2.8 2.4 4.2 3.2
100 100 100 100 100
12-14: School only 81.4 88.4 89.8 78.6 84.8
School and work 6.7 4.5 3.1 8.7 5.6
Work only 6.4 1.9 2.0 6.9 4.2
Home care only 5.5 5.2 5.0 5.8 5.3
100 100 100 100 100
15-17: School only 60.1 75.0 73.8 58.0 67.2
School and work 8.1 5.2 4.3 10.1 6.7
Work only 20.4 7.1 8.8 21.4 14.1
Home care only 11.2 12.7 13.1 10.3 11.9
100 100 100 100 100
All: School only 79.9 85.9 87.1 77.3 82.8
School and work 4.5 3.1 2.2 5.8 3.8
Work only 6.4 2.1 2.6 6.5 4.3
Home care only 9.1 8.9 8.0 10.3 9.0
100 100 100 100 100

A simple tabulation of schooling and work activities by household income quintile suggests a
direct or indirect link of these activities to poverty. Over 95 percent of primary school age
children and over 90 percent of secondary school age children in the highest household
income quintile report schooling as the only activity, compared to about 84 percent and 62
percent, respectively, for children in the lowest quintile (Table 2). Ten times more (eight
times more) children of primary (secondary) school age work and are not enrolled in school,
while about four times more (three times more) children of primary (secondary) school age
are exhibiting only home care activity. Similar, though less pronounced, differences are
found for children who combine work and school.

http: www.childprotection.org.ph 7
Featured Paper for June 2004
Table 2: Schooling and Work Activities by Income Quintile, 1999 (percent)

Age/Activity Q1 Q2 Q3 Q4 Q5
Age 6-12
School only 84.2 86.3 89.5 90.0 95.2
School and work 2.3 3.0 2.7 1.7 1.4
Work only 1.0 0.8 0.3 0.3 0.1
Home care only 12.5 10.0 7.5 7.6 3.3
100 100 100 100 100
Age 13-17
School only 61.6 66.8 71.0 76.8 90.6
School and work 8.5 7.9 7.1 5.2 3.1
Work only 16.0 14.0 11.6 8.9 2.2
Home care only 13.9 11.2 10.3 8.9 4.1
100 100 100 100 100

Children’s contribution to household income is substantial. On average, a child worker


provides over 9 percent of household income in urban areas and over 8 percent in rural areas
(Table 4). This is only the direct, monetary contribution and does not include the indirect
contribution from home care activities, which allows mothers to engage in paid work.

One of the aims of this study is to compare the child labor incidence between the beginning
and the end of the 1990s. A comparison of the evidence from the 1999 APIS with evidence
from a decade earlier (Sakellariou and Lall 2000), shows that the overall incidence of child
labor for children of age 10 and over in the Philippines is roughly unchanged, at a little over
12 percent (Table 3), and a little over 8 percent for children of age 6 and over 5. This finding
can be discussed in the context of a hypothesized link of child labor to poverty. Poverty
estimates in the Philippines were revised downwards from 40.6 percent at the beginning of
the 1990s to 32.1 percent in 1994 and updated estimates show further reductions in 1997
(Balisacan 1999). One would expect that reductions in poverty would have been matched by
lower child labor incidence rates, provided that the incidence of child labor is intimately
linked to poverty. Further discussion hinges on the sign and size of the coefficients of the
poverty variables, presented in Section 6.

5
This figure cannot be compared to the corresponding figure in 1991, as the data used in the 1991 study did not
cover children less than 10 years.

http: www.childprotection.org.ph 8
Featured Paper for June 2004
Table 3: Incidence of Child Labor: 1991 vs 1999 (percent)

1991* 1999
Age
10 3.3 3.8
11 4.9 5.1
12 6.7 7.5
13 9.0 8.8
14 13.5 13.6
15 16.9 17.5
16 21.1 20.1
17 29.0 25.8
All 12.3 12.2
Region
NCR: Manila 1.7 2.4
Ilocos 8.6 11.7
Cagayan 19.5 21.9
Central Luzon 9.2 6.4
S. Luzon 7.3 7.8
Bicol 20.8 14.6
W. Visayas 16.1 12.0
C. Visayas 16.5 12.1
E. Visayas 21.5 13.6
W. Mindanao 10.7 12.8
N. Mindanao 19.9 27.8
S. Mindanao 15.1 13.1
C. Mindanao 19.4 16.5
Cordillera 9.5 14.5

* Source: Sakellariou & Lall, 2000. Note: The data used in Sakellariou & Lall, 2000 did
not include children less than 10 years.

Table 4: Children’s’ Earnings: Contribution to Family Income (percent)

Share of Family Income


Age Urban Rural
10 4 4
11 4 4
12 6 5
13 7 6
14 7 8

http: www.childprotection.org.ph 9
Featured Paper for June 2004
15 9 9
16 10 9
17 11 10
All 9.4 8.4

5. The model and Econometric Methodology

The approach to modeling the supply of child labor depends on the assumption used about
the nature of the decision making process within the household. Most previous studies of
child labor estimate a single binary model, although several studies employing multivariate
analysis exist. I focus on the choice between the two most utilized alternatives in multivariate
analysis of child labor, that of the simultaneous decision making hypothesis within the
household, versus the hierarchical (sequential) one. If decisions are hypothesized to be made
simultaneously, a multinomial choice model, such as a multinomial logit is a natural choice.
A hierarchical decision making process, on the other hand, would call for a sequential choice
model.

There are several benefits and drawbacks of the two alternative formulations. The most
frequently cited drawback of the multinomial logit model is that the model requires the
assumption of the independence of irrelevant alternatives (IIA); that is, removing any of the
alternatives in the model does not alter the relative probabilities of choosing the remaining
alternatives6. In general, the consequences of the requirement of IIA are more serious when
the choices, as viewed by the decision maker/s, include close substitutes (such as work for
wages vs. work for a family business)7. Another, less serious drawback of the multinomial
logit model is the requirement that the set of explanatory variables is the same for all
alternatives. This, again, would lead to complications, mainly, if the alternatives include
more than one work options (i.e., work for wages vs. work for a family business) and less so
for choosing between school and work.

The fundamental drawback of the sequential decision making process (and, therefore, the
sequential probit model) is the nature of the hypothesis itself. It is unlikely that household
decisions (whether made jointly or by a benevolent “dictator”) are made sequentially; in fact,
it would be difficult to argue that such decisions are not made simultaneously. Furthermore,
if one was to adopt the sequential decision making hypothesis, the question of the order of
the sequence would still be, largely, arbitrary. I, therefore, adopt the hypothesis that
households make decisions on the supply of child labor in a simultaneous manner, and adopt
a multinomial logit model8.

The probability of a child having activity j (with choices: j=1 (school only); j=2 (school and
work); j=3 (work only); j=4 (no activity/home care)), is given by:
6
Cockburn (2001), in his study of rural Ethiopia, tested the hypothesis and was inclined to reject it, although the
tests were not conclusive.
7
An alternative model would be the multinomial probit model, which is not subject to the IIA assumption;
however, computational difficulties with this model allow only a small number of alternatives (see for example,
Grootaert and Patrinos, 1999).
8
The results of the two approaches are often qualitatively similar (see for example, Grootaert and Patrinos,
1999).

http: www.childprotection.org.ph 10
Featured Paper for June 2004
a βX α βX
Pj = e j + j / Σke k + k k ; j, k = 1, 2, 3, 4
Choice j=4 is made the excluded choice and the predicted probabilities, therefore,
indicate the impact of the explanatory variable on the probability that a child attends school
only, combines school and work or works only, relative to the same variable’s impact on the
probability that the child exhibits no work or school activity.

6. Findings and Policy Implications

Findings

It is considered important to derive and discuss the determinants of child activity within two
separate age groups (very young versus older children), as one would expect interaction of
age with variables entered in the regression. Table 5 presents the results of the multinomial
logit models for all children of age 6 to 12 and 13 to 17 years. The individual coefficient
results are significant for about 45 to 65 percent of the coefficients. In the case of young
children the model under-predicts the probability of combining school and work and only
working, due to the small number of children in these categories. The discussion of results is
organized by determinant of child activity, rather than by stage in the decision making
process, so that it is easier to observe the differential effect of each determinant on the 3
stages of the decision making process.

Child Characteristics
Both age and gender are positively correlated with the decision to enroll in school and the
effect of age (and gender for older children) is substantial. Girls of secondary school age
have a 10 percentage point higher probability of being only in school. Being older by one
year increases the probability of work without schooling by 13 and 5.5 percentage points
respectively for primary and secondary age children; however, these estimates on the effect
of age conceal any gender differences (see below for gender differences).

Parent Characteristics
Fathers’ education has a strong effect on both schooling and work decisions, increasing the
probability of the child only schooling and decreasing the probability of combining school
and work as well as the probability of work without schooling. The effect of higher education
of mothers is of similar nature but less pronounced.

http: www.childprotection.org.ph 11
Featured Paper for June 2004
Table 5: Multinomial Logit Results: Probability Derivatives at the Mean (%) – All

Children 6 - 12 years Children 13 - 17 years


Variable School School Work School School Work
only and work only only and work only
AGE 13.9*** 0.8*** -0.2*** 5.8*** 5.5*** 3.9***
AGE SQ. -0.7*** -0.03*** 0.01** -0.4*** -0.2*** -0.04***
FEMALE CHILD 0.1 -0.3 -0.2*** 9.7* -5.0*** -7.4***
YEARS ED. FATHER 0.4*** -0.01*** -0.02*** 1.7*** -0.2*** -0.8***
YEARS ED. FATHER*FEM -0.01 0.01 0.01* 0.03*** 0.1*** 0.1***
YEARS ED. MOTHER 0.1*** -0.01** -0.002 0.3* -0.1 -0.1**
YEARS ED. MOTHER*FEM -0.01 -0.01 0.003 0.01 0.04 0.03
FATHER EMPL FOR WAGES 0.04 0.05 0.01** -0.1 0.1 0.4
FATHER EMPL FOR WAGES*FEM 0.1 0.1 -0.02 -0.8*** 1.0*** 2.0***
FATHER SELF-EMPLOYD -1.0 0.7*** 0.1*** -1.2*** 2.6*** 2.0***
AGE OF HEAD -0.3*** 0.03 0.01 0.1 -0.02 -0.1
AGE OF HEAD SQ. 0.01*** 0.000 -0.000 0.001 0.000 0.000
FEMALE HEAD -0.5 0.3** 0.1*** -1.5 0.8* 1.2**
OWNS FARM -0.3 0.1 -0.03** 1.2*** 1.6*** -0.2***
LOW INCOME (Lowest 10%) -2.8*** 0.1** 0.1 -9.2*** 0.7** 2.8
SIZE OF HOUSE 0.01*** 0.01*** 0.000 0.02 0.001 -0.03***
# OF CHILDREN 0-5 YEARS -0.1*** -0.01* 0.003 0.1 -0.1 0.04
# OF CHILDREN 0-5 YEARS*FEM 0.004 0.01 0.000 -0.2 0.1 -0.1*
# OF BOYS 6-12 YEARS -0.1*** 0.000 -0.01* 0.02 0.01 -0.03
# OF GIRLS 6-12 YEARS -0.05 0.004 0.001 -0.03 0.04 -0.03
# OF BOYS 13-17 YEARS 0.03 0.01 -0.003 -0.1 -0.004 0.04
# OF GIRLS 13-17 YEARS -0.02 - 0.000 -0.002 0.1 -0.1 0.1
SCHOOL FEES (/100 pesos) 0.2*** -1.0*** -0.000 0.3 -0.4*** -0.05***
LONG DISTANCE TO SCHOOL -0.2*** 0.04 0.01 -0.1** 0.1*** 0.1***
CHILD/ADULT WAGE (%) -0.04*** 0.005 -0.001* 0.1 -0.1*** -0.01
URBAN 0.7 -0.5*** -0.04** 3.5 -1.4*** -2.7***
ILOCOS 1.4 -0.5* 0.03 -1.0* -1.5 4.6***
CAGAYAN 0.3 0.1 0.5* -4.9 2.3** 4.7***
LUZON 1.0 -0.7*** 0.1 0.1 -2.8** 2.9***
BICOL -0.4 -0.4** 0.3 -2.5 -1.6 4.4***
VISAYAS 0.7 -0.4 0.2 2.9*** -2.4 2.1***
MINDANAO -2.0** 0.3 0.4 -1.5* -0.6 4.1***
CAR 1.7*** -0.2 0.1 7.0*** -0.6** -1.0**
ARMM -5.8*** -0.8*** 0.4 8.6*** -3.8*** -1.5
CARANGA -1.6 0.04 0.5 -4.3 -0.5 5.7***
Predicted Probability (%) 95 1 0.1 83 4 5
Actual Frequency (%) 88 2 0.6 77 6 8
Log likelihood -11,282 -15,134
Prob > Chi-sq 0.000 0.000
Pseudo R-squared 0.246 0.164
N 34,496 21,269
Note: Comparison activity is: no activity/home care. (***) indicates significance at the 99% level. (**)
indicates significance at the 95% confidence interval. (*) indicates significance at the 90% confidence interval.

http: www.childprotection.org.ph 12
Featured Paper for June 2004
Fathers’ wage employment results in a higher probability that older (mostly male) children
are working, but the effect is not statistically significant. The effect of fathers’ self-
employment, on the other hand, decreases the probability of school enrollment for children of
both age groups and increases the probability of (mostly older children) combining school
and work as well as working only (presumably in a family business). Mother’s employment
status is not included in the specification, as mother’s employment tends to affect child labor
(and vice-versa).

Household Characteristics
Children in households headed by women are less likely to only attend school and more
likely to work, suggesting that the dominant effect is due to the lack of resources in such
households. The effect of ownership of agricultural land obtained suggests that the income
effect dominates the substitution effect; children are more likely to attend school, although
this effect is evident only in the case of older children.

As noted earlier, the link between poverty and child labor is stressed in many earlier studies;
however, in a minority of studies this link is questioned. Our results provide at least moderate
evidence in support of this link: While children from poor households (lowest decile) 9 are
much less likely to enroll in school (by 3 and 10 percentage points, respectively, for younger
and older children), and are not more likely to work, the latter effect is rather small. That is,
low family income does not seem to be strongly associated with work activity. The estimated
effect of poverty (belonging in the lowest decide of the income distribution) on child labor
may have been affected by the inclusion of other variables, such as “size of house”.
Furthermore, given the comparison activity in the estimation of the model, any change in the
probability of child labor is in comparison to another “bad”, namely “no activity/home care”.
Nevertheless, the weak association between poverty and child labor may suggest that the
finding that the child labor incidence in the Philippines has not decreased between 1991 and
1999 despite decreases in the incidence of poverty during the same period may not be that
surprising; the process from poverty reduction to declining child labor incidence may be very
slow.

The results on the presence and age composition of children in the household do not uncover
any discernable effects on children’s activity. Although information on the number of school
age children in the household not attending school was available, it was not entered in the
specification because of endogeneity concerns.

9
In some earlier studies family income included the income from children’s work, raising endogeneity concerns
and, therefore, the possibility that the effect of poverty on child labor is underestimated. In our case, however,
the family income measure is a binary variable (belongs in the lowest decile of the income distribution); it
measures the relative position of the family on the income scale compared to other families, at the national
level.

http: www.childprotection.org.ph 13
Featured Paper for June 2004
Cost of Schooling
The results on the effect of the cost of schooling proxy highlight the imperfection of this
proxy as a measure of the cost of schooling. While significant coefficients are obtained, the
signs of some of the coefficients are counter-intuitive; almost certainly, what is captured here
is some other effect, such as school quality, rather than the cost of schooling 10. On the other
hand, the other proxy of the opportunity cost of schooling - distance to school - is well
behaved; the probability of school enrollment decreases and the probability of working
increases, the longer the distance to school.

Demand for Child Labor


The attempt of incorporating demand for child labor into the model (using a cluster average
of children’s wages across many communities as a proportion of adult wages) produced some
significant coefficients, however, the results are not clear-cut; while, as expected, the
probability of school enrollment for young children decreases with a higher child to adult
wage, the rest of the results are not clear-cut. It is not clear, therefore, whether this proxy
captures the effect of demand factors.

Location
The effect of urbanity on the decision to work is significant; children in urban areas have a
higher probability of only schooling and are less likely to work compared to children in rural
areas. The regions with higher probability of school enrollment (without working), compared
to the NCR-Manila, are the Cordillera Autonomous Region, and Visayas, while in almost all
regions children are more likely to engage in work activity, compared to the capital region of
Manila.

Urban versus Rural Areas

In deriving urban/rural differences as well as gender differences (see below), the sample used
consists of children 10 to 17 years old, as the number of children under 10 years is quite
small.

Comparing the results between urban and rural areas (Tables A2 in the appendix), reveals
many similarities and some differences. One notable difference is the effect of children’s age;
while in urban areas the probability of a child attending school increases with age, in rural
areas the probability of only attending school decreases by about 5 percentage points for each
additional year of age and the probability of combining school and work or only working
increases by about 4 percentage points. Other differences relate to the intensity of the effect
of parents employment status and lack of resources in the household. In rural areas the effects
of parents’ education are stronger, compared to urban areas, and the same goes for the effects
of father’s self-employment. Finally, the effects of belonging to households headed by

10
The cost of schooling (school fees only) variable may be correlated with the urban-rural dichotomy variables
(i.e., rural/urban residence and ownership of farm).

http: www.childprotection.org.ph 14
Featured Paper for June 2004
females are more evident in rural areas (lower probability of schooling and higher probability
of working).

Gender Effects
The effect of age for children 10-17 years old is clearly different between genders (Table
A3). While for boys one additional year of age decreases the probability of schooling only
and increases the probability of working by about 4 percentage points, the probability of girls
attending school increases by about 1.5 percentage points for each additional year of age.

The favorable effect of parents’ education, while qualitatively similar between genders, is
much more evident in the case of boys compared to girls. Similarly for the unfavorable effect
of having a self-employed father which is much more evident for boys. Finally, the
prevailing income effect of owning a family farm increases the probability of attending
school only, and more so for girls.

7. Main findings - Policy Implications

A number of results in this study can be discussed in a policy context. The first relates to the
link between poverty, schooling and child labor. It is manifested in the effects of belonging
to a poor household (lowest decile of income distribution) and belonging to a household
headed by a female. Both characteristics reflect, directly or indirectly, a lack of resources in
the household. It was found that children in poor households are less likely to attend school
(but not necessarily more likely to work); such households can be the target of policy
intervention

Another troubling finding is the detrimental effect of ownership of a small family business on
school enrollment and child labor in the Philippines. Having a self-employed father increases
the probability of working without schooling, especially for older children.

The effect of the opportunity cost of schooling can be discussed in relation to the role of the
quality of education and the need for an increase in education expenditure in the Philippines,
funded by the state with generous help from international organizations. As noted earlier, the
effect of the cost of schooling proxy (cluster of school fees across 148 departments),
probably reflects the effect of quality of education across communities in the Philippines. If
so, it would suggest that families take quality of education provided in their community into
account when making decisions about their children’s time allocation. Parents will send
children to school when the perceived private return to education exceeds the return to any
alternative allocation of children’s time. Improving the quality of schooling provided will go
a long way in increasing school attendance. Another possibility worth exploring is public
funding of the private education sector in the Philippines (along the lines of a voucher
system).

Recent results based on theoretical models (such as those by Basu and Van 1998); Basu
2000; Dessy 2000; Dessy and Knowles 2001; Ranjan 1999; Jaferey and Lahiri 2002, among

http: www.childprotection.org.ph 15
Featured Paper for June 2004
others), provide useful clues as to the applicability of popular arguments in favor of child
labor bans, compulsory education laws and related legislation. These results seem to suggest
that banning child labor in poor countries is not expected to produce the intended results,
especially a ban that can only be enforced in the formal sector, which can reduce the well-
being of children by forcing them to work in the informal sector under worse work
conditions. Making education compulsory is a preferred alternative; even under free and
universal access to education, without compulsory education laws an under-development trap
with high incidence of child labor and high fertility rates is at work. Finally, low parental
incomes in combination with lack of access to credit markets can generate child labor.

The above-mentioned recommendations, in fact any policy designed to reduce the problem of
child labor and increase school enrollments, will be expensive and cannot be implemented
with the existing budget allocations. Funding from international organizations is essential;
however, on the domestic front, education budgets need to be increased, even considering the
scarcity of funds that the government of a developing economy usually faces.

Concluding, the shift in emphasis in the new International Labor Organization (ILO)
convention on child labor (No. 182, 1999), away from its previous offensive for immediate
eradication and minimum age standards, towards a softer approach seems well conceived, as
long as international codes regulating and protecting children’s rights are upheld. This is
because, first, this approach is in agreement with recent evidence on the consequences of
child labor bans and second, because it will now be easier accepted and acted upon in
countries with high incidence of child labor.

http: www.childprotection.org.ph 16
Featured Paper for June 2004
References

Balisacan, A. M. (1999). “Poverty Profile in the Philippines: An Update and Re-examination


of Evidence in the Wake of the Asian Crisis,” Working Paper (School of Economics, UP
Diliman, Quezon City).

Basu, Kaushik and Pham Hoang Van (1998). “The Economics of Child Labor,” American
Economic Review 88: 412-427.

Basu, Kaushik (2000). “The Intriguing Relation Between Adult Minimum Wage and Child
Labor,” Economic Journal 110, C50-C61.

Becker, Gary (1997). A Treatise on the Family, Enlarged Edition (Harvard University Press,
Cambridge MA, 1991); also see Jensen and Nielsen .

Blunch, Niels-Hugo and Verner Dorte (2000). “Revisiting the Link Between Poverty and
Child Labor: The Ghanaian Experience,” Centre for Labor and Social Research, Department
of Economics, The Aarhus School of Business (Aarhus, Denmark).

Boyden, Jo Ling, Birgitta and William Myers (1998). What Works for Working Children?,
(Stockholm and Florence: Unicef).

Canagarajah, S. and H. Coulombe (1997). “Child Labor and Schooling in Ghana”, Policy
Research Working Paper 1844, The World Bank.

Chernichovsky, D. (1985). “Socioeconomic and Demographic Aspects of School Enrollment


and Attendance in Rural Botswana”, Economic Development and Cultural Change 32: 319-
332.

Cockburn, J. (2001). “Child Labor versus Education: Poverty Constraints or Income


Opportunities,” forthcoming, Development and Change.

Coleman, J. (1988). “Social Capital in the Creation of Human Capital,” American Journal of
Sociology 94: S95-S120.

Cartwright, Kimberly (1999). “Child Labor in Colombia”, in C. Grootaert and H. Patrinos,


eds, The Policy Analysis of Child Labor: A Comparative Study (New York, St. Martin’s
Press, 1999): 63-102.

______________ and Harry A. Patrinos (1999). “Child Labor in Urban Bolivia”, in C.


Grootaert and H. Patrinos, eds, The Policy Analysis of Child Labor: A Comparative Study
(New York, St. Martin’s Press, 1999): 103-130.

DeGraff, D. Bilsborrow, R. and A. Herrin (1992). “The Implications of High Fertility for
Children’s Time Use in the Philippines”, in C.B. Lloyd, ed, Fertility, Family Size and

http: www.childprotection.org.ph 17
Featured Paper for June 2004
Structure – Consequences for Families and Children, Proceedings of a Population Council
Seminar (New York, June).

DeTray, D. (1983). “Children’s Work Activities in Malaysia”, Population and Development


Review 9: 437-455.

Dessy, Sylvain (2000). “A Defense of Compulsive Measures Against Child Labor,” Journal
of Development Economics 62: 261-275.

____________ and John Knowles (2001). “Why is Child Labor Illegal?,” mimeographed (U.
S. National Bureau of Economic Research).

Dreze, J. and G. Gandhi-Kingdom (2001). “School Participation in Rural India,” Review of


Development Economics 5: 1-24.

Grootaert, Christaan and Ravi Kambur (1995). “Child Labor: An Economic Perspective”,
International Labor Review 134: 187-203.

Grootaert, Christaan and H. A. Patrinos (1999). “Policies to Reduce Child Labor,” in C.


Grootaert and H. Patrinos, eds, The Policy Analysis of Child Labor: A Comparative Study
(New York, St. Martin’s Press, 1999): 151-162.

Gunn, Suzan and Zenaida Ostos (1992). “Dilemmas in Tackling Child Labor: the Case of
Scavenger Children in the Philippines”, International Labor Review 131: 629-646.

Institute for Labor Studies. Comprehensive Study on Child Labor in the Philippines,
Monograph Series No. 1, Manila, Philippines, 1994.

International Labor Organization, IPEC at a Glance, ILO, Geneva, 1997.

Jafarey, Saqib and Sajal Lahiri (2002). “Will Trade Sanctions Reduce Child Labor? The Role
of Credit Markets,” Journal of Development Economics 68: 137-156.

Jensen, Peter and Helena Nielsen (1997). “Child Labor or School Attendance? Evidence
from Zambia”, Journal of Population Economics 10 (1997):407-424.

Knodel, J. Havanon, N. and W. Sittitrai (1990). “Family Size and Education of children in
the Contest of Rapid Fertility Decline”, Population and Fertility Review 16: 31-62.

_____________ and M. Wongsith (1991). “Family Size and Children’s Education in


Thailand: Evidence from a National Sample, Demography 28: 119-131.

Levison, D. (1991). “Children’s Labor Force Activity and Schooling in Brazil” (Ph.D.
dissertation, University of Michigan).

http: www.childprotection.org.ph 18
Featured Paper for June 2004
Maddala, G. S. (1983). Limited-Dependentand Qualitative Variables in Econometrics
(Cambridge University Press, New York).

Maitra, P. and R. Ray (2002). “The Joint Estimation of Child Participation in Schooling and
Employment: Comparative Evidence from Three Countries,” Oxford Development Studies
30: 41-62.

Mergos, G. (1992). “The Economic Contribution of Children in Peasant Agriculture and the
Effect of Education: Evidence from the Philippines”, Pakistan Economic Review 31: 189-
201.

Nielsen, Helena (1998). “Child Labor and School Attendance: Two Joint Decisions,” Centre
for Labor Research – WP 98-15 (Aarhus, Denmark).

Patrinos, Harry A. and George Psacharopoulos (1995). “Educational Performance and Child
Labor in Paraguay”, International Journal of Educational Development 15: 47-60.

______________ and George Psacharopoulos (1997). “Family Size, Schooling and Child
Labor in Peru – An Empirical Analysis,” Journal of Population Economics 10: 387-405.

Ranjan, Priya (1999). “An Economic Analysis of Child Labor,” Economics Letters 64: 99-
105.

Ray, R. (2000). “Analysis of Child Labor in Peru and Pakistan.” Journal of Population
Economics 13: 3-19

Reuter Library Report, Philippine Congress Passes Labor Law (November 1993).

Rosenzweig, M. and R. Evenson (1977). “Fertility, Schooling and the Economic


Contribution of Children in Rural India: An Economic Analysis”, Economertica 45:1065-79.

Sakellariou, Chris and Ashish Lall (2000). “Child Labor in the Philippines,” Asian Economic
Journal 14 (2000): 233-253.

Sancho-Liao, Nelia (1994). “Child Labor in the Philippines: Exploitation in the Process of
Globalization of the Economy”, Labor, Capital and Society 27: 270-281.

Sasaki, Masaru and Temesgen, Tilahun (1999). “Children in Different Activities: Child
Labor and Schooling in Peru,” Draft, World Bank (Washigton D. C).

Turbay, C. and E. Acuńa (1998). “Child Labor and Basic Education in Colombia,” in M.C.
Salazar and W. Alarcon Glasinovich, eds, Better Schools, Less Child Work: Child Work and
Education in Brazil, Colombia, Ecuador, Guatemala and Peru (Florence, Italy, UNICEF
International Child Development Centre).

UNICEF, The State of the World’s Children, New York, 1997.

http: www.childprotection.org.ph 19
Featured Paper for June 2004
Appendix

Table A1: Definitions of Variables and Mnemonics


Mnemonic Definition
Child characteristics
AGE Age of child
AGE SQ. Age of child squared
FEMALE CHILD Female child
Parent characteristics
YEARS ED. FATHER Years of education of father
YEARS ED. FATHER*FEM Years of education of father x female child
YEARS ED. MOTHER Years of education of mother
YEARS ED. MOTHER*FEM Years of education of mother x female child
FATHER EMPL. FOR WAGES Father employed (excluding self-employment)
FATHER EMPL. FOR WAGES*FEM Father employed for wages x female child
FATHER SELF-EMPLOYED Father self-employed
Household characteristics
AGE OF HEAD Age of household head
AGE OF HEAD SQ. Age of household head squared
FEMALE HEAD Female head of household
OWNS FARM Household owns agricultural land
LOW INCOME (Lowest 10%) Household in the lowest income quintile
SIZE OF HOUSE Size of house in square meters
# OF CHILDREN 0-5 YEARS Number of other children 0-5 years
# OF CHILDREN 0-5 YEARS*FEM Number of other children 0-5 years x female child
# OF BOYS 6-12 YEARS Number of other boys 6-12 years
# OF GIRLS 6-12 YEARS Number of other girls 6-12 years
# OF BOYS 13-17 YEARS Number of other boys 13-17 years
# OF GIRLS 13-17 YEARS Number of other girls 13-17 years
Cost of schooling proxies
SCHOOL FEES (/100 pesos) Cluster average of school fees across 148 domains
LONGDISTANCE TO SCHOOL Travel time to school more than 30 minutes
Demand proxy
CHILD/ADULTWAGE (%) Cluster average of children’s wage as a proportion of
adult wage across 148 domains
Location
URBAN Urban household (excluded: rural)
ILOCOS Ilocos region
CAGAYAN Canayan region
LUZON Luzon region
BICOL Bicol region
VISAYAS Visayas region
MINDANAO Mindanao region
CAR Cordillera (autonomous) region
ARMM ARMM region
CARANGA Caranga region
(excluded: NCR – Manila)

http: www.childprotection.org.ph 20
Featured Paper for June 2004
Table A2: Multinomial Logit Results: Probability Derivatives at the Mean (%) – Urban/Rural

Children 10 - 17 years Children 10 - 17 years


Urban Areas Rural Areas
Variable School School Work School School Work
only and work only only and work only
AGE 3.0** 1.1*** 0.6*** -4.6** 3.7*** 3.8***
AGE SQ. -0.1*** -0.03*** 0.01** 0.01*** -0.1*** -0.06***
FEMALE CHILD 4.2** -1.3 -1.3*** 8.8** -6.4*** -5.5***
YEARS ED. FATHER 0.9*** -0.1*** -0.2*** 1.1*** -0.05*** -0.6***
YEARS ED. FATHER*FEM -0.1 0.02 0.04 0.3*** 0.1*** -0.05**
YEARS ED. MOTHER 0.1*** -0.01 -0.04* 0.2 -0.1 -0.07
YEARS ED. MOTHER*FEM 0.04 0.01 0.01 -0.04 -0.01 0.03
FATHER EMPL. FOR WAGES -0.1 0.3 0.00 1.7** -0.8 0.8*
FATHER EMPL. FOR WAGES*FEM -0.1** 0.6*** 0.5*** -0.02 0.6 0.5*
FATHER SELF-EMPLOYED -0.6*** 1.7*** 0.5*** -1.4*** 3.1*** 1.5***
AGE OF HEAD 0.01 0.01 -0.04 -0.6** 0.2 0.1
AGE OF HEAD SQ. 0.00 -0.00 0.00 0.01*** -0.00 -0.01
FEMALE HEAD -0.6 0.4 0.3* -2.5 1.7* 1.2**
OWNS FARM 0.3*** 1.3*** -0.1** 0.7*** 1.2*** -0.3**
LOW INCOME (Lowest 10%) -4.1*** -0.6*** 0.6 -6.6*** 1.8 2.2
SIZE OF HOUSE 0.00 0.00 -0.01*** 0.04*** 0.01*** -0.03**
# OF CHILDREN 0-5 YEARS -0.01 -0.01 0.01 -0.02 -0.1* 0.06
# OF CHILDREN 0-5 YEARS*FEM -0.06 0.04 -0.03* -0.2 0.2 -0.04
# OF BOYS 6-12 YEARS -0.01 0.03 -0.00 0.06 -0.07 -0.06*
# OF GIRLS 6-12 YEARS -0.01 0.03 0.02* -0.03 0.05 -0.07*
# OF BOYS 13-17 YEARS 0.01 -0.00 0.00 -0.02 0.06 0.03
# OF GIRLS 13-17 YEARS 0.01 -0.1*** 0.02 -0.08 0.06 0.03
SCHOOL FEES (/ 100 pesos) 0.1 -0.1*** -0.01*** 0.9 -0.8*** -0.00
LONG DISTANCE TO SCHOOL -0.02 0.06*** 0.02** -0.3 0.3*** 0.1***
CHILD/ADULT WAGE (%) 0.02 -0.02** -0.01* 0.1* -0.1*** -0.01**
ILOCOS 1.6 -0.9 0.4* - - -
CAGAYAN -0.9 1.9*** 0.2 -7.3* 6.9** 0.1
LUZON 1.5* -1.3** 0.5*** -0.3* -2.5*** -0.9***
BICOL -0.9 0.1 0.9** 0.4* -2.5*** -0.4*
VISAYAS 1.9** -1.1* 0.3** 2.3 -1.4 -1.2*
MINDANAO -0.9 0.3 0.8** -3.2 2.3 0.04
CAR 3.6*** -0.01 -0.3* 3.0* 1.0* -2.3*
ARMM 3.7*** -1.3** -0.3 8.7 -6.8*** -2.4***
CARANGA -2.6 1.2 1.3** -2.2 -0.3 0.2
Predicted Probability (%) 93 2 1 84 7 4
Actual Frequency (%) 86 3 4 75 8 10
Log likelihood -9,110 -10,719
Prob > Chi-sq 0.000 0.000
Pseudo R-squared 0.187 0.174
N 20,258 15,634
Note: Comparison activity is: no activity/home care. (***) indicates significance at the 99% level. (**) indicates
significance at the 95% confidence interval. (*) indicates significance at the 90% confidence interval.

http: www.childprotection.org.ph 21
Featured Paper for June 2004
Table A3: Multinomial Logit Results: Probability Derivatives at the Mean (%) – by Gender

Children 10 - 17 years Children 10 - 17 years


Boys Girls
Variable School School Work School School Work
only and work only only and work only
AGE -3.9* 2.5*** 3.7*** 1.6*** 1.1*** -0.1
AGE SQ. -0.02*** -0.06*** -0.06*** -0.1*** -0.03*** 0.02
YEARS ED. FATHER 1.2*** -0.1*** -0.5*** 0.8*** -0.04*** -0.2***
YEARS ED. MOTHER 0.2*** -0.04 -0.08 0.1*** -0.03 -0.03
FATHER EMPL. FOR WAGES 0.2 -0.2 0.7** 0.6*** 0.8*** 0.2***
FATHER SELF-EMPLOYED -1.9*** 2.5*** 1.7*** -0.3*** 1.9*** 0.2***
AGE OF HEAD 0.1 -0.06 -0.05 -0.4** 0.1 0.05
AGE OF HEAD SQ. 0.00 0.00 0.00 0.01*** -0.00 -0.00
FEMALE HEAD -1.3 0.7 0.7 -1.1 0.9** 0.6***
OWNS FARM 0.2*** 1.6*** -0.3* 0.9*** 0.6*** -0.1**
LOW INCOME (Lowest 10%) -6.4*** 0.9** 1.7 -4.3*** 0.1*** 1.0
SIZE OF HOUSE 0.02 -0.00 -0.02*** 0.02*** 0.01*** -0.01***
# OF CHILDREN 0-5 YEARS -0.05 -0.05* 0.03 -0.06 0.02 0.00
# OF BOYS 6-12 YEARS 0.09 -0.03 -0.04 -0.05 0.03 0.00
# OF GIRLS 6-12 YEARS 0.04 0.03 -0.03 -0.06 0.02 0.01
# OF BOYS 13-17 YEARS 0.00 0.01 0.03 0.04 -0.03 -0.01
# OF GIRLS 13-17 YEARS 0.07 -0.1** 0.01 -0.9 0.04 0.04
SCHOOL FEES (/ 100 pesos) 0.3 -0.3*** -0.00*** 0.3 -0.3*** -0.00
LONG DISTANCE TO SCHOOL -0.1 0.1*** 0.07*** -0.1 0.1*** 0.03**
CHILD/ADULT WAGE (%) 0.06** -0.07*** -0.02*** 0.04 -0.03*** 0.00
URBAN 3.1** -2.1*** -2.1*** 1.4* -0.6 -0.3
ILOCOS -0.3 -1.8 3.8*** 2.6* -1.4 0.3
CAGAYAN -5.5 2.1* 4.8*** 1.1* 0.2 0.2
LUZON 2.1 -3.4*** 1.7** 1.5 -1.8* 0.9***
BICOL -1.5 -1.7 3.7*** 1.3 -1.6* 0.4
VISAYAS 1.9 -2.5 1.6** 3.2*** -1.7 0.4**
MINDANAO -2.1 -0.6 3.2*** 0.4 -0.4 1.0***
CAR 5.4*** -1.5 -0.6* 3.5*** -0.5* -0.1**
ARMM 7.3*** -3.8** -0.04** 4.4 -2.6*** -0.7
CARANGA -5.4 -0.1 4.7** 0.5 -0.5 1.2**
Predicted Probability (%) 87 4 3 92 2 1
Actual Frequency (%) 77 7 9 86 4 3
Log likelihood -11,713 -8,143
Prob > Chi-sq 0.000 0.000
Pseudo R-squared 0.199 0.160
N 18,621 17,271
Note: Comparison activity is: no activity/home care. (***) indicates significance at the 99% level. (**) indicates
significance at the 95% confidence interval. (*) indicates significance at the 90% confidence interval.

http: www.childprotection.org.ph 22
Featured Paper for June 2004
Chart 1: Children Enrolled in School and Not Working

100

In school only (%)


80
Boys
60 Girls
40 Urban
Rural
20

0
6 7 8 9 10 11 12 13 14 15 16 17
Age

Chart 2: Children Working (and Not Enrolled in School)

35

30
Work only (%)

25
Boys
20 Girls
15 Urban
Rural
10

0
6 7 8 9 10 11 12 13 14 15 16 17
Age

n.b.
Any republication by the web users, in whatever form, without the written consent by the
author, is prohibited. The text may be used only for internal information needs.
E –mail: acsake@ntu.edu.sg

Prof. Chris Sakellariou


School of Humanities and Social Science,
Nanyang Technological University, Singapore

http: www.childprotection.org.ph 23
Featured Paper for June 2004

Вам также может понравиться