You are on page 1of 19

European Journal of Economics, Finance and Administrative Sciences

ISSN 1450-2275 Issue 27 (2010)


© EuroJournals, Inc. 2010
http://www.eurojournals.com

A Comparative Analysis of Rural and Urban


Child Labor in Pakistan

Rana Ejaz Ali Khan


Department of Economics, the Islamia University of Bahawalpur, Bahawalpur Pakistan
Tel: +92 0345 8724744
E-mail: ranaejazalikhan@iub.edu.pk

Tasnim Khan
Department of Economics, the Islamia University of Bahawalpur, Bahawalpur Pakistan
Tel: +92 0346 8806011
E-mail: tasneem_iub@hotmail.com

Rashid Sattar
Department of Economics, the Islamia University of Bahawalpur, Bahawalpur Pakistan
Tel: +92 0334 7071203
E-mail: rashid.sattar@iub.edu.pk

Abstract
The paper presents a comparative analysis of determinants of child labor in urban and rural
areas. A simple theoretical model (Probit) of household’s decision about child’s time
allocation is used. From the econometric data sets of urban and rural areas of two districts
of Pakistan, evidence is provided suggesting that urban and rural children have different
determinants of child labor reflecting their different socio-economic background, e.g.
education of head of household (as a continuous as well as binary variable) impacts the
child schooling positively but the effect of continuous variable is much stronger for urban
households, employment status of head of household impacts the child schooling positively
in urban areas but negatively in rural areas. Mother’s employment is complement to child
schooling in urban areas but substitute in rural areas. It negatively impacts the child labor
in urban but positively in rural areas. Poverty affects the child schooling three times more
strongly in rural households than urban ones. Gender discrimination is schooling
participation is higher in rural households but in labor force participation of children, it is
higher in urban households. Part-time labor force participation of children increases by
incremental change in age in urban households but it decreases in rural households. A few
number of determinants are same for both groups of children. For policy implication
different sets of policies are required for urban and rural areas to enhance child welfare.

Keywords: Child Labor, Child Schooling, Poverty, Urban, Rural, Pakistan.

1. Introduction
The socio-economic structure of rural-urban areas differs for whole of the third world. High poverty
and stagnation are the main characteristics of rural Pakistan. Rural population survives under extremely
poor conditions deprived of health and education facilities. Rural labor market is characterized by
under-employment, seasonality of job opportunities, lack of alternative jobs, lack of skill, strong
Electroniccopy
Electronic copy available
available at:
at:https://ssrn.com/abstract=2567431
http://ssrn.com/abstract=2567431
140 European Journal of Economics, Finance and Administrative Sciences - Issue 27 (2010)

occupational bondage and absence of agro-based industry. The rural areas of Pakistan have gained
disproportionately from the development process of the country. The socio-economic change has been
experienced mostly by urban residents, where as rural population, representing the majority lag much
behind in social and economic aspects. Arif (2000) opined that in 1990s poverty has adversely affected
the poor families in Pakistan specifically of rural ones, which results into decline of child welfare.
Child labor varies across urban and rural areas within the country. Access to primary school
specifically of girls is comparatively difficult in rural areas. The child labor that is basically assumed
an informal sector phenomenon also differs for both areas1. Though the urban informal sector
households have the same characteristics of those of rural ones, reflecting the artificial division of
economy that researchers and policy-makers have created. One justification of this division, in
practical terms may be that urban informal sector is generally more accessible to researchers and thud
yields more from complete data. Moreover, when studying the education of children and legal
intervention for child labor it seems that greater coverage and efficiency can initially be achieved in the
urban setting. But it would be overly ambitious to make recommendations about implementing
programs on rural areas which are feasible in urban and peri-urban areas. However, it may be hoped
that lessons learned in urban context may be useful in the formulation of future comprehensive plans
which will take place in rural areas. That is why some researchers2 have analyzed child labor for urban
and rural areas separately.
Bhalotra and Attifield (1998) have shown no evidence of differential treatment of higher birth-
order or elderly children in intra-household allocation of resources in rural Pakistan. Bhalotra and
Heady (2003) have analyzed the official survey data from rural Pakistan and Ghana to search the effect
of ownership of land a specific characteristic of rural community, on schooling and work decision of
children. They found an interesting result that children in land-rich households are often more likely to
be in work than the children of land-poor households. It needs reexamine the matter by primary data
and a comparative analysis of rural and urban areas. Ersado (2005) has made a comparative analysis of
rural and urban children activities for Nepal, Peru and Zimbabwe. The study estimated that poverty is
the main cause of child labor in rural areas but there is lack of support for poverty hypothesis in urban
areas. The improving access to credit has greater potential for alleviating child labor and enhancing
school attendance in rural areas but availability of child-care options promises greater impact in urban
areas. However, the efforts to bolster adult educational level and wage will help curb the prevalence
and intensity of child labor and likelihood child school for both urban and rural areas3.
We are going to make a comparative analysis of urban and rural areas to see the differential
impact of explanatory variables on children activities. Since the incidence of child labor is significantly
high4 and school enrolment of children is as low as 66 percent in Pakistan, it offers a good case study
for such an investigation. We expect significant variations in the determinants of child labor in rural
and urban areas, which if true, may warrant different strategies from policy makers aimed at addressing
this issue.

1
Though some researchers have analyzed the child labor collectively for urban and rural areas, for instance, Burki and
Shahnaz (2003) for Pakistan; Maitra and Ray (2002) for Pakistan.
2
See for instance, Sharif 1994 for rural India; Nielsen 1998 for rural areas of Zabmia; Sawada and Lokshin 2000 for child
schooling in rural Pakistan; Hazarika 2001 for girls schooling in rural Pakistan; Cigno, et. al. 2001 for rural India; Nielsen
and Dubey 2002 for rural India; Bhalotra and Heady 2003 for rural Pakistan; Ali and Khan 2004 for urban Pakistan; Hou
2009 for rural and urban separately for Pakistan.
3
The study has utilized the official data of Nepal, Peru and Zimbabwe but we will discuss Pakistan with primary data.
Majority of the studies on child labor has used official data, for instance Burki and Shahnaz (2001) for Pakistan; Maitra
and Ray for rural Pakistan (2002) and Hou (2009) for Pakistan.
4
Labor force participation of children is 8.27 percent in overall areas, 3.2 percent in urban and 10.26 percent in rural areas
(FBS 1996). See CUTS (2003) for such type of rural urban disparity in magnitude of child labor in South Asian countries.

Electroniccopy
Electronic copy available
available at:
at:https://ssrn.com/abstract=2567431
http://ssrn.com/abstract=2567431
141 European Journal of Economics, Finance and Administrative Sciences - Issue 27 (2010)

2. Definitions, Data Collection and Estimation Model


To make a comparative analysis of children activities in urban and rural areas, we apply the model
separately for urban and rural data.

2.1. Definitions
A child is defined as a person who is 5-15 years old and the child labor is defined as “the participation
of school-age children (5-15 years) in the labor force, i.e. work for wage or in household enterprises to
earn a living for themselves or to support household income”. Urban areas are defined as the areas
within the boundaries of municipal committee and the rural areas are those outside the boundaries of
municipal and town committees. Though the locality of the town committee differs from urban and
rural locality in socio-economic conditions and some surveys termed it as semi-urban areas but we
have analyzed two broader communities, i.e. urban and rural.

2.2. Data Collection


In this paper the empirical analysis of rural urban differences of child labor in Pakistan, is based on the
data from two districts of Pakistan. The survey covered more than four thousand households from
urban and rural areas. The distinguishing feature of this survey is that it covers 5-15 years old children,
while previous studies (see for instance, Ray 2000; Burki and Shahnaz 2001; Burki and Shahnaz 2003)
have covered 10-14 years old children using labor force surveys.

2.3. Estimation Model


We examined the determinants of children activities for urban and rural areas separately by using a
series of sequential probit model. In the first regression, the urban households are assumed to use a
sequential decision process, keeping the schooling of their children as a priority for the welfare of their
children. The sequential choices making the welfare of the child are assumed as: (i) schooling, (ii)
schooling and work, (iii) work only, and (iv) neither school nor work. This leads to the following four
choices, and choice probabilities, to be estimated for each child:
P1= Probability to go to school and not to work
P2= Probability to go to school and to work
P3= Probability not to go to school but to work
P4= Probability neither to go to school nor to work
In the sequential probit model, the probabilities for the four choices are determined as follows,
P1 = f (b1X)
P2 = [1 - f (b1X)] f (b2X)
P3 = [1-f (b1X)] [1-f (b2X)] f (b3X)
P4 = [1-f (b1X)] [1-f (b2X)] [1-f (b3X)] f (b4X)
Where f represents the standard normal distribution function, and b1, b2, b3 and b4 are vectors of
the model parameters. To detangle the determinants of children’s activities, four groups of explanatory
variables (X1-Xn), i.e. child characteristics, head of household characteristics, parent characteristics,
and household characteristics have been taken. For the second regression, the model is the same where
a rural child’s activity (P1, P2, P3, or P4) is a function of same explanatory variables. The definitions of
explanatory variables are presented in table-1.

Table 1: Definitions of Dependent and Explanatory Variables Used in the Probit Model

VARIABLES DEFINITIONS
Dependent Variables
P1 [Child goes to school only] • 1 if child goes to school and not to work, 0 otherwise
P2 [Child goes to school as well as to work] • 1 if child goes to school and to work, 0 otherwise
P3 [Child goes to Work only] • 1 if child does not go to school but to work, 0 otherwise

Electronic copy available at: https://ssrn.com/abstract=2567431


142 European Journal of Economics, Finance and Administrative Sciences - Issue 27 (2010)
P4 [Child neither goes to school nor work] • 1 if child neither go to school nor to work, 0 otherwise
Independent Variables
1. Child Characteristics
BORD (Birth-order of child) • Birth-order of child in his/her brothers and sisters
CGEN (Gender of child) • 1 if child is male, 0 otherwise
CAGE (Child’s age) • Child’s age in completed years (for 5-15 years age group)
CAGESQ (Child’s age squared) • Child’s age squared (for 5-15 years age group)
CEDU (Child’s education) • Child’s education in completed years
2. Head of household Characteristics
HGEN (Gender of the head of household) • 1 if the head of household is male, 0 otherwise
HEDU (Head of the household’s education) • Head of the household’s completed years of education
HLIT (Head of the household’s literacy status) • 1 if the head of the household is literate, 0 otherwise
HEMP (Head of household’s employment) • 1 If head of household is employed, 0 otherwise
HY (Head of household’s income) • Head of household’s income per month (in 000 Rupees)5
3. Parent characteristics
FEDU (Father’s education) • Father’s education in completed years of education
FLIT (Father’s literacy status) • 1 if father is literate, 0 otherwise
FEMP (Father’ employment) • 1 if father is employed, 0 otherwise
FY (Father’s income) • Father’s income per month (in 000 Rupees)
MEDU (Mother’s education) • Mother’s completed years of education
MLIT (Mother’s literacy status) • 1 if mother is literate, 0 otherwise
MEMP (Mother’s employment) • 1 if mother is employed, 0 otherwise
MY (Mother’s income) • Mother’s income per month (in 000 Rupees)
4. Household Characteristics
ASST (Household’s ownership of assets) • 1 if the household owns of assets, 0 otherwise
HHPCY (Household’s per capita Income) • Household’s per capita income (in 00 Rupees) per month
HPOVTY (Household’s poverty status) • 1 if household is below poverty line, otherwise 0
HHSIZ (Household/family size) • Number of household/family members
HHNUC (Nucleus household/family) • 1 if household is nucleus, otherwise 0
CHILD04 • Number of children up to 4 years of age in the household
CHILD515 • Number of children (5-15 years) in the household
CHILD16 • Number siblings (16 years or above) in the household

3. Estimated Results and Discussion


The objective of the study is comparative analysis of determinants of children activities in rural and
urban areas by econometric model but the qualitative results have also been compiled. If composition
of activities of children vary according to locality of the household, it is possible that they have been
differently affected by explanatory variables. The activities of children in rural and urban areas are
shown in table-2.

Table 2: Activities of Children (5-15 years) in Rural and Urban Areas (Percentage)

Activities Urban Rural Overall


School-going Children 55.15 36.95 47.04
Children Combining School and Work 5.13 8.35 6.24
Child Laborers 9.94 15.51 12.72
No-School, No-Work/Home-care Children 29.48 55.87 34.49

The rural-urban differences in school participation are evident as a significantly higher ratio of
urban children is going to school. The urban children have advantage in their welfare. On the other

5
The explanatory variables like the education of head of household, employment status of head of household and income
level of head of household are likely to be endogenous and thus may result in biased estimates. We will apply the
sensitivity test for the robustness of the results. Same test will be applied for employment of father and mother, and
household income or poverty status.

Electronic copy available at: https://ssrn.com/abstract=2567431


143 European Journal of Economics, Finance and Administrative Sciences - Issue 27 (2010)

hand, in all the remaining activities, the ratio of rural children is higher, where children welfare is
lower.
The summary statistics for urban and rural areas are shown in tables-3 and 5 respectively. In the
tables mean and standard deviations are shown. The figures in the parenthesis are the standard
deviations. A rural-urban difference is also noted here: average per capita per month income of urban
households (who are producing child labor) is Rs.503 as opposed to Rs.479 of rural households.
Similarly, among the child laborers the average household size is 7 and 7.9 respectively for urban and
rural areas. The parents’ education also makes a difference for urban and rural sample. For example 49
percent of urban school-going children (as against 40 percent of rural areas) had literate fathers and
13.21 percent of urban child laborers (as against 3.3 of rural areas) had literate fathers.

Table 3: Summary Statistics (Mean and Standard Deviation) for Urban Areas

Children Going to Children Going to Children do not going Children neither going
Variable
School Only School as well as Work to School but Work to School nor Work
1. Child Characteristics
BORD 2.2467 (1.2170) 2.2105 (1.2281) 2.2750 (1.2807) 2.3966 (1.9546)
CGEN 0.5454 (0.4995) 0.5303 (0.5009) 0.4750 (0.5054) 0.4814 (0.4831)
CAGE 9.1883 (2.7695) 9.8340 (3.0014) 10.6000 (2.6584) 9.7037 (3.5389)
CAGESQ 92.0454 (52.8219) 105.6801 (639.2866) 119.2500 (53.0504) 106.22 (71.3428)
CEDU 3.8441 (2.6075) 4.2307 (2.3376) 1.8373 (1.9506) 0.5106 (1.9795)
2. Head of Household Characteristics
HGEN 0.9632 (0.0921) 0.9821 (0.1028) 0.9989 (0.9269) 0.9747 (0.9921)
HEDU 10.0000 (5.3394) 6.9433 (6.0597) 0.9500 (1.6478) 1.8245 (2.1889)
HLIT 0.4768 (0.4362) 0.2160 (0.5009) 0.1138 (0.1874) 0.0740 (0.1924)
HEMP 0.9155 (0.2789) 0.8299 (0.3764) 0.5000 (0.5052) 0.6666 (0.4803)
HY 5203.57 (7567.90) 2626.51 (6573.80) 1295.00 (593.53) 1577.77 (1061.32)
3. Parent Characteristics
FEDU 8.4740 (5.3857) 6.9271 (6.0769) 0.9911 (1.2328) 1.5121 (2.3629)
FLIT 0.4935 (0.4253) 0.2137 (0.4872) 0.1321 (0.1321) 0.0873 (0.1824)
FEMP 0.8729 (0.3198) 0.8714 (0.3938) 0.5321 (0.5291) 0.6829 (0.5463)
FY 6799.44 (8046.53) 2263.31 (5729.43) 1397.93 (733.35) 1447.69 (1153.59)
MEDU 4.35065 (6.2387) 2.1538 (6.5828) 0.9750 (1.1068) 0.6666 (2.1483)
MLIT 0.4619 (0.4672) 0.1308 (0.4296) 0.0250 (0.1581) 0.0741 (0.2668)
MEMP 0.0885 (0.2583) 0.1978 (0.3867) 0.2600 (0.5063) 0.5925 (0.5007)
My 1040.58 (4610.37) 926.57 (3989.60) 670.00 (698.05) 662.96 (619.64)
4. Household Characteristics
ASST 0.8220 (0.2689) 0.7137 (0.3900) 0.6000 (0.4961) 0.5555 (0.5063)
HHPCY 2001.47 (2719.80) 1394.06 (2285.52) 303.23 (180.91) 379.73 (170.70)
HPOVTY 0.3441 (0.4766) 0.8461 (0.3679) 0.9500 (0.2207) 0.9629 (0.1924)
HHSIZ 6.5194 (1.7089) 6.8016 (1.7957) 7.0250 (1.8043) 6.9259 (2.3521)
HHNUC 0.37013 (0.4844) 0.3117 (0.4813) 0.2500 (0.4935) 0.3333 (0.5547)
NCHILD 3.6233 (1.4099) 3.8461 (1.5462) 4.3000 (1.6361) 4.2592 (2.0864)
CHILD04 0.3441 (0.5759) 0.3684 (0.5754) 0.4000 (0.4961) 0.4074 (0.6360)
CHILD515 3.2727 (1.3149) 3.4210 (1.3528) 3.6750 (1.3471) 3.6296 (1.6904)
CHILD16 0.5779 (0.9887) 0.7611 (1.1875) (1.1000) (1.4815) 0.7037 (1.3247)
No of Observations 3912 407 705 2070
Total Number of Observations 7094

The sequential probit results for urban and rural areas are shown in table-4 and 6 respectively.
The tables represent the marginal probabilities, parameter estimates and t-statistics. The bold figures
are probability derivatives, normal figures are parameter estimates and figures in the parenthesis are t-
statistics. The results for the four outcomes (school, combining school with work, full-time work and
home-care) are discussed here.

Electronic copy available at: https://ssrn.com/abstract=2567431


144 European Journal of Economics, Finance and Administrative Sciences - Issue 27 (2010)
Table 4: Sequential Probit Results for Urban Areas

First Stage Probability that Second Stage Probability Third Stage Probability that Third Stage Probability that
Variable the Child Goes to School that the Child Goes to the Child does not go to the Child neither go to
Only School as well as Work School but Work School nor Work
-1.9140 -0.5248 -1.8993 1.9447
Constant
-7.018 (-1.5307) -7.889 (-2.7117) -9.6724 (-1.7065) 8.8971 (2.2766)
1. Child Characteristics
-0.0047 -0.0084 -0.0115 0.0186
BORD -0.1844 (-1.3302)* -0.0200 (-1.2863)** -0.5023 (-2.2656)** 0.1224 (1.4903)*
0.0753 0.0098 0.0428 -0.0722
CGEN 0.1279 (1.7103)** 0.0632 (2.0167)** 0.2220 (1.6251)* -0.1756 (-1.2869)*
0.1742 0.0852 0.0709 -0.0822
CAGE 0.8739 (2.1273)** 0.3144 (1.3768)* 0.4913 (1.9148)** -0.9059 (-1.2991)*
-0.0076 -0.0001 -0.0026 0.0049
CAGESQ -0.0503 (-2.5559)** 0.0266 (-1.2966)* -0.0294 (-1.6934)** 0.4578 (1.4620)*
0.0857 0.0111 -0.0258 -0.0795
CEDU 1.3113 (1.8849)** 1.3361 (4.2011)** 1.0809 (-1.5641)* -2.6741 (-3.0463)**
2. Head of Household Characteristics
-0.0010 -0.2096 0.0010 -0.0859 -0.3691
HGEN -0.5225 (-1.3910)* -0.2639 (-1.6821)** 0.0316 (0.0721) (-0.0210)
0.1491 0.0002 -0.2830 -0.2299
HEDU 0.6074 (1.3095)** 0.3286 (0.0100) 0.9852 (-1.7205)** -4.7104 (-2.2618)**
0.2104 -0.2142 -0.2129 -0.3104
HLIT -0.7710 (1.8557)** -3.1542 (-1.3837)* 0.9862 (-2.9821)** -1.0620 (-1.3866)*
0.0081 -2.6921 -0.1703 -0.2576
HEMP 0.5738 (1.6094)** -1.3140 (-0.5567) -1.0275 (-1.5086)* -1.2183 (-1.8622)**
0.0331 -0.0037 -0.0063 0.0295
HY 0.7319 (1.2348)* -0.1439 (-0.9395) -0.8931 (-1.1484)* 0.3221 (0.9341)
3. Parent Characteristics
0.2135 -0.0043 -0.0727 -0.1061
FEDU 6.4799 (1.4434)* -0.2834 (-0.1581) -7.1645 (-1.7131)** -4.6047 (-1.6348)*
0.2916 -0.3277 -0.1966 -2.4144
FLIT 1.2317 (2.2603)** -0.9718 (-0.5143) -0.8957 (-3.1426)** -1.3621 (-0.3445)
0.3216 -2.4661 2.3629 0.1321
FEMP 0.2319 (0.0913) -1.2574 (-0.9624) 1.6921 (1.0291) 0.6136 (0.1536)
0.1237 0.0239 -0.2196 -0.6737
FY 0.1192 (0.9725) 0.2082 (0.0853) -1.3361 (-1.9542)** -0.4319 (-1.0593)
0.2366 -0.0046 -0.1347 -0.1398
MEDU 0.1580 (1.6535)** -0.0443 (-1.8397)** -0.1969 (-1.2886)* -0.1079 (-2.0691)**
0.2917 0.3919 -0.4380 0.3592
MLIT 0.6721 (2.1031)** 1.0762 (1.3722)* 1.1495 (-1.7627)** 1.3761 (2.1310)**
0.1674 0.5064 -0.0381 -0.1844
MEMP 0.7702 (1.2991)* 1.4463 (0.2018) -1.1043 (-1.3180)* -0.2246 (-1.3533)*
0.0003 -0.0035 -0.0062 -0.0001
My 0.1120 (1.5902)* 0.5122 (-1.2858)* -0.6116 (-1.3937)* 00.2367 (-1.4330)*
4. Household Characteristics
0.0805 -0.0163 -0.0395 -0.0398
ASST 1.0158 (1.8925)** -0.3137 (-1.7514)** -0.6482 (-1.6460)** -0.3165 (-1.7057)**
0.0169 -0.1769 -0.1143 -0.0014
HHPCY 0.6945 (6.7555)** -0.2555 (-1.9755)** -0.7083 (-1.8429)** -1.1364 (-1.3786)*
-0.1814 -1.4725 0.2768 0.3544
HPOVTY 1.1315 (-3.1874)** 1.3316 (-0.5165) 1.4107 (2.3741)** 1.8315 (1.8018)**
-0.0194 0.0068 0.0481 0.1107
HHSIZ -0.1254 (-1.7426)** 0.2936 (1.2929)* 0.1807 (2.5195)** 0.4532 (2.0796)**
0.1723 -2.4157 70.1914 1.7781
HHNUC 0.3861 (0.0222) -2.9397 (-3.0709)** 0.8131 (0.4125) 0.6841 (1.1052)
-0.0287 0.0323 0.0914 0.0187
NCHILD -0.3687 (-1.4062)* 0.4833 (2.2820)** 0.3513 (1.5507)* 0.3688 (1.3168)*
0.1550 0.0186 0.2039 0.1060
CHILD04 0.6313 (1.2753)* 0.3957 (1.2891)* 0.2380 (1.8594)** 0.8077 (3.1312)**
-0.2258 0.0006 0.1323 0.0526
CHILD515 -0.3394 (-1.6576)** 0.7183 (2.0551)** 0.4140 (1.2877)* 0.6548 (2.9017)**
0.1067 1.6020 -139.65 -0.1780
CHILD16 1.6983 (2.4536)** 0.9736 (1.0182) 0.5871 (-0.3344) 0.9853 (-2.0922)**
Log of Likelihood Function -3376.9 -376.88 -660.1 -1597.1
Number of Observation 3912 407 705 2070
R-Squared 0.6833 0.6057 0.7090 0.6282
Percent Correct Prediction 0.8704 0.8419 0.8825 0.8987
* Indicates significant at 10 percent level and ** indicates significant at 5 percent level

Electronic copy available at: https://ssrn.com/abstract=2567431


145 European Journal of Economics, Finance and Administrative Sciences - Issue 27 (2010)
Table 5: Summary Statistics (Mean and Standard Deviation) for Rural Areas

Children Going to Children Going to Children not going to Children neither going
Variable
School Only School as well as Work School but Work to School nor Work
1. Child Characteristics
BORD 2.2516 (1.2145) 1.6923 (0.9281) 2.2682 (1.2653) 2.4074 (1.3939)
CGEN 0.6483 (0.49927) 0.5769 (0.5038) 0.4878 (0.5060) 0.4814 (0.5091)
CAGE 9.1871 (2.7605) 12.6153 (2.4670) 10.6829 (2.6780) 9.5555 (3.6514)
CAGESQ 91.97 (52.6576) 165.00 (56.8914) 121.1219 (53.7369) 104.1481 (72.9012)
CEDU 3.1377 (2.6032) 7.2307 (2.3376) 0.9024 (2.0347) 0.6296 (1.8218)
2. Head of Household Characteristics
HGEN 0.9461 (0.7921) 0.9276 (0.0437) 0.9844 (0.0732) 0.9784 (0.9138)
HEDU 5.9354 (5.3823) 2.1923 (4.0597) 0.9268 (1.6338) 1.0370 (2.3118)
HLIT 0.3531 (0.4807) 0.1667 (0.38069) 0.0237 (0.026) 0.0527 (0.2262)
HEMP 0.8161 (0.2780) 0.8322 (0.8811) 0.5122 (0.5060) 0.7474 (0.4465)
HY 4162.25 (7560.80) 2403.84 (1275.76) 982.92 (591.1439) 1533.33 (1080.95)
3. Parent Characteristics
FEDU 6.9996 (5.4276) 3.9872 (4.5439) 0.8274 (1.7610) 1.0021 (2.1150)
FLIT 0.4014 (0.5001) 0.1904 (0.3162) 0.0334 (0.019) 0.0689 (0.3161)
FEMP 0.9100 (0.27421) 0.9226 0.8946 0.5122 (0.5060) 0.7407 (0.4465)
FY 6285.61 (7930.46) 2864.42 (1431.4) 1282.92 (0591.14) 581.62 (986.26)
MEDU 3.2903 (6.2636) 1.1153 (3.4447) 0.6707 (1.0932) 0.1666 (2.1483)
MLIT 0.0589 (0.4405) 0.0117 (0.2041) 0.0027 (0.0000) 0.0064 (0.1622)
MEMP 0.0890 (0.2576) 0.0946 (0.3258) 0.1522 (0.5060) 0.6666 (0.4803)
My 915.8064 (4605.72) 663.87 (1025.67) 534.53 (693.17) 421.96 (605.31)
4. Household Characteristics
ASST 0.7661 (0.2780) 0.6692 (0.4296) 0.5853 (0.4987) 0.4814 (0.5091)
HHPCY 1389.52 (1213.04) 527.90 (213.98) 279.49 (180.23) 311.09 (119.27)
HPOVTY 0.6588 (0.4769) 0.8333 (0.3806) 0.8947 (0.3153) 0.8684 (0.3425)
HHSIZ 6.5290 (1.7067) 7.6923 (1.1922) 7.9739 (1.7857) 6.8518 (2.2988)
HHNUC 0.3677 (0.4837) 0.03846 (0.1961) 0.2439 (0.4889) 0.3333 (0.5547)
NCHILD 3.8258 (1.4056) 4.0384 (1.3410) 4.2926 (1.6162) 4.0370 (1.9311)
CHILD04 0.6000 (0.7898) 0.2083 (0.4148) 0.3157 (0.6710) 1.1315 (0.8437)
CHILD515 3.1822 (1.1390) 3.5000 (1.5036) 2.2631 (0.9334) 1.7368 (1.2010)
CHILD16 0.7294 (0.9806) 1.0416 (1.1220) 0.8421 (0.6882) 0.7105 (0.8976)
No. of Observations 2915 713 1223 3039
Total No. of Observation 7890

Table 6: Sequential Probit Results for Rural Areas

Second Stage: Fourth Stage:


First Stage: Probability Third Stage: Probability
Probability that the Probability that the
Variable that the Child Goes to that the Child does not
Child Goes to School as Child neither go to
School Only go to School but Work
well as Work School nor Work
-0.2020 -0.2356 -1.3815 2.2931
constant
12.1513 (-0.6281) -19.8056 (-1.1716) -51.0742 (-1.0541) 42.8905 (0.5258)
1. Child Characteristics
-0.5345 0.0335 -0.1236 0.0249
BORD
-0.2068 (-1.2924)* 0.4256 (1.0539) -1.9625 (-1.4410)* 0.9882 (1.6842)**
0.1081 0.0310 0.0043 -0.1140
CGEN
0.8842 (1.8903)** 1.3051 (2.3501)** 1.7971 (2.06141)** -0.0253 (-1.5900)*
0.1233 -0.0417 0.0988 -0.0975
CAGE
1.3209 (2.5239)** -1.5904 (-1.3782)* 0.6777 (1.3527)* -4.3026 (-1.3531)*
-0.0098 0.0000 -0.0041 0.0054 0.2039
CAGESQ
-0.0938 (-3.2080)** 0.0554 (0.9147) -0.2505 (-1.0482) (1.5341)*
0.0420 0.0426 -0.0290 -0.1193
CEDU
0.3178 (2.2820)** 1.4594 (3.5327)** -0.2625 (-1.5569)* -1.6181 (-2.764)**
2. Head of Household Characteristics
-0.0310 -0.0081 0.4616 0.1429
HGEN
-1.5161 (-1.3628)* -0.9267 (-1.8237)** 0.9218 (0.9778) 0.2143 (0.7324)
0.0736 -0.2563 -0.0910 -0.1564
HEDU
7.9150 (1.8230)** -0.9371 (-0.4562) -11.2680 (-1.6360)* 7.2362 (-1.3614)*

Electronic copy available at: https://ssrn.com/abstract=2567431


146 European Journal of Economics, Finance and Administrative Sciences - Issue 27 (2010)
0.2646 -0.3559 -0.2417 -0.3988
HLIT
1.7646 (1.4596)* -4.5559 (-1.4765)* -2.4670 (-2.3524)** -10.3988 (-1.8729)**
-0.0276 -10.3429 -0.1227 0.1714
HEMP
-0.6433 (-1.7612)** -1.6974 (-0.5197) -11.6049 (-1.2856)* 12.9852 (2.0091)**
0.0130 -0.0823 -0.0771 0.0003
HY
0.2295 (2.1941)** -0.2900 (-1.3623)* -0.8021 (-1.2901)* 0.3269 (1.0603)
3. Parent Characteristics
0.0953 0.2555 -0.0829 0.3606
FEDU
7.7829 (2.2569)** 0.6456 (0.4548) -10.0509 (-1.6164)* 8.4095 (0.4736)
0.3828 -0.5559 -0.2943 0.4164
FLIT
1.5821 (1.2954)* 0.8821 (-1.4765)* -0.2388 (-2.2378)** 0.2736 (2.2109)**
0.0076 0.0029 -0.0671 0.1637
FEMP
0.3194 (0.3843) 0.7562 (0.9217) -0.6752 (-0.4166) 1.8231 (0.3496)
-0.0130 -0.0740 -0.0372 0.0261
FY
-0.2416 (-0.1940) 0.3529 (-1.6412)* -1.2789 (-1.4217)* 0.9468 (0.2437)
0.1197 0.0567 -0.0250 -0.0325
MEDU
1.1519 (2.1148)** 0.7521 (0.0087) -6.9334 (-1.9526)** -0.0325 (-1.3139)*
0.4311 4.8555 -0.3046 -15.6660
MLIT
5.8344 (2.5339)** 0.4855 (1.0156) -4.7830 (-3.2611)** -1.5666 (-0.8386)
-0.0061 -0.0800 0.1172 0.1528
MEMP
0.6103 (-1.2664)* -0.3631 (-2.4365)** 5.7558 (1.7513)** 1.8498 (1.3441)*
0.0324 -0.0269 0.0169 -0.0001
My
0.5869 (2.1372)** -0.3463 (-1.2863)* 0.2799 (0.1524) -0.4036 (-0.9246)
4. Household Characteristics
0.0921 0.0833 0.0133 0.1156
ASST
1.3055 (2.2695)** -3.3463 (2.1528)** 11.4481 (1.3663)* 1.6716 (1.3812)**
0.0010 -0.0096 -0.0428 -0.0018
HHPCY
0.1738 (2.6271)** -0.2077 (-1.8572)** -2.0296 (-1.9323)* -1.2341 (-1.4483)*
-0.3649 1.1526 0.1881 0.3908
HPOVTY
-0.1049 (-2.1068)** 0.2774 (0.4028) 1.4707 (1.2875)* 10.7908 (2.1410)**
0.0294 -0.0053 0.0981 0.1632
HHSIZ
0.3891 (1.3007)* 1.3582 (-2.6224)** 6.6860 (2.1559)** 5.5380 (1.6845)**
-0.2549 -0.0323 -0.0198 0.2441
HHNUC
-0.4392 (-2.2952)** -0.5105 (-1.3128)* -2.6742 (-2.1359)** 1.6256 (1.1543)
-0.0105 0.0063 -0.0619 0.2596
NCHILD
-0.1689 (-2.4428)** -2.5337 (2.5172)** -20.9528 (-3.9661)** 11.6743 (1.5112)*
-0.1434 0.1741 -32.2610 0.1676
CHILD04
-1.16583 (-1.3520)* 0.2369 (2 .0870)** -3.2226 (-0.4484) 0.5675 (2.8153)**
-0.0025 0.0246 0.0422 0.4486
CHILD515
-0.5025 (-3.6316)** 0.3587 (2.0163)** 1.2381 (1.9716)** 5.4486 (2.4298)**
0.3484 1.5337 -0.1070 -0.1905
CHILD16
0.3484 (1.5517)* 1.6036 (0.9076) -10.8942 (-1.2980)** -6.4905 (-2.4665)**
Log of Likelihood Function -2480.11 -586.21 -1020 2844.6
Number of Observation 2915 713 1223 3039
R-Squared 0.6645 0.6597 0.7270 0.6227
Percent Correct Prediction 0.8764 0.9475 0.9083 0.9243
* Indicates significant at 10 percent level and ** indicates significant at 5 percent level.

3.1. School Participation


The first stage results show the probability of school participation of children. Here we are going to
compare the impact of explanatory variables on child schooling for urban and rural areas.

3.1.1. Child Characteristics


We have found that birth-order negatively affects the schooling decision of children in urban and rural
areas but it is stronger in rural areas. It contradicts the results by Bhalotra and Attifield (1998 for rural
Pakistan), where no differential treatment of children by their birth-order exists in Pakistan. The
negative birth-order (in his/her brothers and sisters) shows that the younger brothers and sisters have
lower probability to go to school. This may be due to resource constraint effect but other explanation
may be the delayed enrolment of children in schools in urban and rural areas. It means that the first
school enrolment is more delayed in rural as compared to urban areas. Alternatively resource constraint
effect and delayed first enrolment of children are severe in rural areas. The estimated results of child’s

Electronic copy available at: https://ssrn.com/abstract=2567431


147 European Journal of Economics, Finance and Administrative Sciences - Issue 27 (2010)

age implied that the probability of going to school increases at a decreasing rate by increase in child’s
age for both urban and rural areas. But the probability for rural child is lower than urban one. It
revealed the fact that in rural areas the first school enrolment of children is more delayed than in urban
areas.
The probit estimation of present study has shown that boys are more likely to go to school as
compared to girls in urban as well as rural areas. But gender disparity is higher in rural areas as
compared to urban areas though Bhalotra and Attifield (1998) have found little evidence of gender
differences among Pakistani rural children in intra-household resource allocation. There may be several
possible explanations for the distinct gender gap in urban as well as rural areas but higher gender
disparity in rural areas needs consideration. The rural areas have specific characteristics and
socioeconomic structure so the nature of impact of a child’s region of residence on her/his schooling is
highly sensitive to child’s gender. The lack of female schools in rural areas also explains this result
(see Hazarika 2001). As a policy implication improving access to primary schools will reduce the
gender gap in rural areas. Moreover, the religious and socio-cultural values that keep women at home
are comparatively more conservative in rural communities. Even if a girls’ school is available in the
village the girls often are kept at home due to seclusion ethic. From a longer term perspective, policies
have to aggressively erode the pillars that support seclusion norms (see also Sharif 2000). The other
factors may be the higher opportunity cost of daughter’s education in rural areas, intra-households
discrimination in households, relatively higher level of poverty which has a gender aspect, strict
cultural and social customs, low rewards attached to girls’ education by parents and low female teacher
quality and availability in rural areas.
We have found that the current year of the education of a child has positive impact on the
probability to continue schooling in urban as well as rural areas. It is further found that urban children
are more likely than their rural counterparts to continue their schooling, that is, they are less likely to
drop out from school than the rural children. The results regarding first enrollment in school, gender
disparity in schooling and school dropout corroborate the national figures about education in urban and
rural areas of the country.

3.1.2. Head of Household Characteristics


The characteristics of head of the household are critical in determining the child’s schooling. It is
generally perceived that head of the household’s education plays a positive role in his decision to send
the child to school. It indicates important complementary relationship between the education of the
head of the household and child’s schooling. To capture the effect of head of household’s education on
child schooling, we have included two types of variables regarding head of household’s education, i.e.
continuous variable representing the number of years of education of head of household, and binary
variable representing whether the head of household is literate or illiterate6 (same type of variables
have been used to capture the effect of parent’s education on children activities in coming pages). We
have estimated a complementary relationship between the education of head of household (as a
continuous variable as well as binary variable) and child’s schooling in urban and rural areas. The
positive impact of education of head of household on child’s schooling in urban areas is 7 percentage
points stronger than in rural areas. It needs some explanation, why the education of head of households
has smaller impact on child schooling in rural areas as compared to urban ones. One may argue that
rural educated adults have low education returns so they perceive the education of their children low-
rewarding. Even if sometimes the educated heads of households would perceive the education of their
children positively, but non-availability of schools and lower quality of education in rural areas as
compared to urban areas results into lower school participation of children in rural areas.
Conceptually, employment status of head of household affects schooling positively, i.e. the
children from employed head of the household are more likely to go to school. We have found the

6
Literacy status of the head of household, father and mother is defined as the completion of at least five years of formal
schooling by the individual.

Electronic copy available at: https://ssrn.com/abstract=2567431


148 European Journal of Economics, Finance and Administrative Sciences - Issue 27 (2010)

same kind of results for urban areas but for rural areas there is surprising result, that is, children from
employed head of household are 2.7 percent less likely to go to school. The possible explanation for
this difference may be that in agrarian community, the land-owners and well-to-do households send
their children nearby good private schools. On the other hand majority of the labor class in villages is
employed by these land-owner households and income of this labor class is so low that it can not send
its children to school. Similarly, informally self-employed households cannot afford schooling but they
have a better opportunity for their children to make them involved in work. In addition there is high
rate of disguised unemployment in rural communities. From the policy point of view, for the rural
areas it is needed to enhance not only the employment but also the productivity and wages/incomes.
The income level of head of household affects the schooling of children positively both in urban as
well as rural areas but the impact is slightly stronger in urban areas7.

3.1.3. Parent Characteristics


The parent characteristics like the education of parents, employment status and income level of parents
are important for child schooling. Specifically, the differential effect of mother’s and father’s education
on the acquisition of schooling by their offspring is important to analyze. Why should parents’
education matter in determining child’s schooling attainment? Parental schooling may be a proxy for a
host of unobservable determinants, such as parental preferences for education and parents’ assistance
to children in school work.
The probit estimation has shown that father’ education as a continuous variable had a positive
impact on the child’s schooling in both urban and rural areas though it is more effective in urban areas
as compared to rural ones. It shows that behavior of the father’s education is similar to that of head of
household. It is further found that father’s education as a binary variable, i.e. literacy status has shown
an impressive positive effect on child’s schooling and it is more evident in rural areas. The impact of
mother’s education on child schooling is also comparable for urban and rural households that is an
urban child from literate mother is 29 percent more likely to go to school while a rural child from
literate mother is 43 percent more likely to go to school.
The mother’s employment results into 16 percent more probability for the child to go to school
in urban areas while it results into 6 percent less probability for the child to go to school in rural areas.
The explanation may be that in the urban areas, the average years of education of women are higher,
they are comparatively highly paid and they contribute more to household income, which results into
more probability of child schooling. It is further supported by the positive impact of mother’s income
on child schooling. On the other hand for the rural areas, the employed women are under-paid and are
usually engaged in informal employment or household enterprises. Their contribution in household and
participation in decision-making is non-recognized. They have lower educational level and are doing
casual and unskilled labor which decrease their perception about child schooling.

3.1.4. Household Characteristics


The household characteristics play an important role in children’s activities. The ownership of assets
by the household is generally a measure of a household’s wealth. We have found that ownership of
assets positively affects children’s schooling in both urban and rural households. The ownership of
assets makes the household stable against incomer shocks through credit procurement or sale of the
assets. It is further found that household per-capita income impacts the schooling of the children
positively in both areas. The lower effect of household per-capita income on child schooling in rural as
compared to urban areas may be explained by the summary statistics. The household per-capita income
is much lower in rural as compared to urban areas so incremental addition to household per-capita

7
Some studies have used the employment status and educational level of head of household (or father and mother) as a
proxy for financial status (or income level) of the individual due to the problem of endogeneity of explanatory variables
(see for instance Burki and Shahnaz 2003). To overcome the problem we have used a sensitivity test, i.e. by including or
excluding the HHY (head of household’s income) the econometric estimates remained unchanged. So we have included
the head of household’s income as an explanatory variable in the analysis.

Electronic copy available at: https://ssrn.com/abstract=2567431


149 European Journal of Economics, Finance and Administrative Sciences - Issue 27 (2010)

income results into less increase in schooling probability in rural as compared to urban areas. From the
policy point of view, rural households need more financial support to send their children to school.
For both urban and rural areas, it is evident from the probit results that poverty affects the child
schooling negatively, that is evident by the probit estimation, but the effect is three times stronger in
rural areas. The stronger poverty effect reflects the stark income disparity in rural areas. Severe poverty
is prevalent in rural areas as 65 percent of the households of school-going children are living below
poverty line in rural areas as compared to 34 percent of that category in urban areas.
To analyze the effect of household size on children activities, we have included two types of
variables related to household size, i.e. continuous variable---number of household members and
binary variable---whether the family is nuclear or not. It is found that the family size (as continuous
variable) impacts the schooling of children negatively in urban areas and positively in rural areas. The
negative impact on child schooling by increase in household members explains the resource
competition effect within the household. As household per-capita income is lower in rural areas and
poverty is widely spread, the resource competition effect becomes severe and child welfare slides down
on priority list in the form of diminishing schooling. On the other hand for rural areas, the larger family
size results into surplus of supply of labor and increase in the probability of schooling of children. It is
further found that children from nuclear families in urban areas are 17 percent more likely to go to
school but 25 percent less likely to go to school in rural areas.
We have found that the number of pre-school-age children in the household has shown no
significant effect in urban areas, while in rural areas, the presence of such children decreases the
schooling probability for school-age children. The number of school-age children in the household has
shown a negative impact on the schooling decision in urban as well as rural areas. But the effect is
much stronger in urban households. If we take into consideration, the resource competition effect, the
negative impact of presence of school-age children in the rural household should be more severe as
these households have comparatively lower household per-capita income, and they are poorer than
urban counterparts, but the probit estimation has shown the reverse results. The explanation may be
that there exists combined family system in rural areas of Pakistan and there is surplus of labor supply
in these households. That is why an increase in number of children in rural households decreases the
school participation of children lesser in rural areas as compared to urban ones. On the other hand
though the urban households are better in financial terms, the education cost is also higher which may
decrease school participation of children by increase in number of children.
The presence of prime-age children in the household boosts the propensity for child schooling
both in urban and rural households but the impact is three times stronger in rural households. The
explanation may be that the presence of prime-age children contributes to household income which
ultimately enhances school participation of school-age children. Sawada and Lokshin (2000:19) had
supported the notion for rural Pakistan, that is, once a child is picked up as a “winner” of educational
investment within the family, his or her education is supported by the elder siblings’ resource
contribution. Resultantly the resource contribution in the household by prime-age children is higher in
rural households.

3.2. Part-Time Work


The second stage estimation eliminates from the sample the children who only go to school. The
probability to be determined for the remaining sample is that of combining school and work or part-
time labor force participation of children.

3.2.1. Child Characteristics


We have found that in urban areas, the birth-order of the child negatively effects the decision of part-
time work, that is younger child is less likely to do part-time work in urban areas. While for rural areas
the birth-order has shown no significant results. For the gender of the child, the results have shown that
boys are 0.98 percent more likely to do part-time work as compared to girls in urban areas but 3.1
percent more likely than those in rural areas. It means the gender aspect regarding part-time work is

Electronic copy available at: https://ssrn.com/abstract=2567431


150 European Journal of Economics, Finance and Administrative Sciences - Issue 27 (2010)

severe in rural households. It explained that cultural norms are weaker along with existence of paid-
employment opportunities in urban areas, so gender disparity in part-time work is low in urban areas. It
corroborates the results of summary statistics where in urban areas 53 percent of part-time child
laborers are boys but in rural areas they are 64 percent.
The probability for the child to do part-time work increases (at decreasing rate) by incremental
change in age in urban areas but in rural areas it decreases in a linear way (see also Maitra and Ray
2002 for such type of results for rural Pakistan). The higher probability for the urban child to do part-
time work in higher age-groups may be explained by the fact that at high age, the cost of schooling
increases, so to support the educational expenditures of schooling, the children have to do part-time
work. Moreover, by the increase in age, children are physically able to do some light work, so the
parents make them combine school and work to decrease the opportunity cost of schooling. But why
such probability decrease in rural areas and increase in urban ones? The explanation may be that there
are less schooling opportunities for children in rural areas, so there is less probability to combine
school with work. Secondly, there are more wage employment opportunities in urban areas, so urban
children avail more chances to combine school with work. Thirdly, there are higher wage rates for
children in urban areas which make the urban children to combine school and work. Fourthly, there is
less need for unpaid domestic work or seasonal work in urban households due to industrial structure of
the cities. On the other hand, in rural areas, due to seasonal demand for labor and need for unpaid work
in agrarian households, the children have less opportunity to combine school with work. Fifthly, the
cost of living and the cost of schooling is higher in urban centers as compared to rural ones, so the
parents have to decide for their children to combine school with work.
The current years of education of child increase the probability to do part-time work in urban as
well as rural areas. It explained that children have to work to support the educational expenditures in
both urban and rural areas.

3.2.2. Head of Household and Parent Characteristics


It is found that the children from female-headed households are more likely to do part-time work in
both urban and rural areas. The educational level of the head of household as a dummy variable, i.e.
whether the head of household is illiterate or literate has also shown the same kind of results for both
areas, i.e. children from illiterate head of household are more likely to do part-time work but the impact
is stronger in rural areas. Similarly, the mother’s income has shown a negative impact on the decision
for the child to do part-time work for both urban and rural areas, but the effect is stronger for rural
areas.

3.2.3. Household Characteristics


The ownership of assets by the household has shown negative impact on urban children’ part-time
work and positive impact on rural ones. The explanation may be that presence of assets in urban
households enhances the financial status of the household and decreases the fluctuation in household
income (assets stand as a proxy for wealth). The asset-owner urban households do not make their
children to do part-time work but they send their children to school (see first stage results). On the
other hand, the presence of assets in rural households makes their children to do part-time work. The
contradiction between the behavior of urban and rural households may be explained as poverty is more
prevalent in rural households (ADB 2002:2; See also Ray 2000:8 for India), income disparity is high
and the majority of the households have assets of minor value like the carts, oxen etc., so they are
prone to engage their children in part-time work.
The poverty status of the household has also shown diverging impact on the decision of part-
time work in urban and rural households. The urban children from poor households are less likely to do
part-time work, but the rural children from such households are more likely to do part-time work. The
explanation may be that majority of the urban child labor, either full-time or part-time is involved in
waged-employment like work in workshops, small enterprises, factories, etc. while majority of the
rural child labor, either full-time or part-time is involved in the family enterprises or family

Electronic copy available at: https://ssrn.com/abstract=2567431


151 European Journal of Economics, Finance and Administrative Sciences - Issue 27 (2010)

employment and seasonal employment. The type of employment in urban areas requires full-time
utilization so the urban children are less likely to do part-time work, when the household falls below
poverty line. On the other hand, the type of employment in rural areas requires comparatively less
time, so the children have better opportunity to do part-time work when the household falls into
poverty.
The family size (as a continuous variable) has shown diverging effect on part-time work in
urban and rural areas. In the urban households the incremental change in number of family members
increases the part-time probability but in rural households it decreases the activity. The positive impact
in urban households explains the income dilution effect of household which compels the parents to
engage their children in part-time work. The negative impact in rural households may be explained by
the fact that combined family system is prevalent in rural areas, which slides down the demand for
labor within the household. So children are less likely to combine school and work in larger
households.

3.3. Child Labor


The third stage of the estimation looks only at the children who are neither in school, nor doing part-
time work. It determines the probability that they will participate in labor force as full-time worker, i.e.
work for wages or in household enterprises.

3.3.1. Child Characteristics


It is found that child labor is negatively related with birth-order of the child in both urban and rural
households, i.e. elder the child, the more likely it is for him/her to do work full-time. The elder children
of the rural households are more likely than their urban counterparts to do full-time labor.
The probit estimation of third stage has shown that boys are more likely to do work than girls in
both urban and rural households but in urban household, such disparity is higher. It corroborates with
first stage results, where boys are more likely to go to school as compared to girls in both type of
households, but disparity is high in rural areas. The phenomena in both stage results have same
explanations, as given in section 3.1.1.
As concerns the age of the child, the results have shown that an incremental change in age of
child would result into increase in probability of child labor in urban areas by 7 percent and in rural
areas by 9.8 percent. It may be concluded that rural children enter labor force earlier than urban ones.
Alternatively the rural children are dropped out of school at an earlier age as compared to their urban
counterparts. It is supported by the summary statistics where the average age of the child laborers in
urban areas is 10.6 years while in rural areas it is 9.6 years. Educational level of the child has shown a
negative impact on full-time work. One additional year of education of child would decrease the
probability for full-time work by 2.5 percent in urban and 2.9 percent in rural areas. As the current
educational level of the child significantly reduces his/her labor force participation ceteris paribus, so a
trade-off exists between child labor and schooling for both urban and rural children.

3.3.2. Characteristics of Head of Household


It is estimated that on average an additional year of education of head of household would result into
decreasing the probability of full-time work by 28 percent in urban and 9 percent in rural households.
The impact in urban areas is 3 times stronger than in rural ones. Similarly the children from illiterate
head of households are 21 and 24 percent more likely to work in urban and rural households
respectively. The results corroborate the findings of the first stage results, i.e. head of household’s
education has stronger positive effect on child’s schooling in urban as compared to rural areas.
The employment status and income level of the head of household negatively affects the child
labor both in urban and rural households. The head of household’s employment effect is stronger in
urban areas. It corroborates the head of household’s employment effect on school participation of
urban children only (see section 3.1.2). A possible explanation may be the prevalence of under-
employment and disguised unemployment in rural households; lower productivity of labor and lower

Electronic copy available at: https://ssrn.com/abstract=2567431


152 European Journal of Economics, Finance and Administrative Sciences - Issue 27 (2010)

wages in rural areas and ultimately lower income leading to higher poverty in rural as compared to
urban households. It is further corroborated by the summary statistics that the incomes of the heads of
the households of child labor producing households are Rs.1295 and Rs.982.9 respectively for urban
and rural areas. As concerns the income of head of household, it is found that incremental change in
income of the head of household decreases the child labor in both urban and rural households but the
effect is much stronger in urban households. The result is supported by the first stage results where
income level of head of household positively and strongly affects the school participation in urban as
compared to rural households.

3.3.3. Parent Characteristics


The mother’s education and father’s education has shown a negative effect on child labor in urban as
well as rural households. But father’s education (continuous as well as binary variable) is more
effective in reducing child labor in rural households while mother’s education (continuous as well as
binary variable) is more effective in urban households. Mother’s education in urban households has
also shown stronger effect on child schooling than of father’s. So the general perception that mother’s
education plays more effective role in reducing child labor is negated in rural households. We think,
the difference in bargaining power within the household between father and mother determines the
decision-making. In rural areas, the women are generally depressed by social norms and do not have a
fair say in household decisions. The female literacy rate is comparatively lower in rural areas, which
makes their bargaining power low. Moreover, the employment opportunities for rural educated mothers
are lower which keeps their bargaining power lower and they do not affect the child labor as effectively
as that in urban households. On the other hand, the female adults (mothers) are comparatively more
educated in urban households than their rural counterparts and have more employment and income
opportunities, so they enjoy good status in their households so their education decreases the child labor
more effectively.
In urban households, there is a stronger negative effect of mother’ education on child labor, as
compared to father’ education, it is corroborated by the first stage results for urban households, i.e.
mother’s education shows a stronger positive effect on child’s school as compared to father’s. In rural
households the mother’s education effect is weaker on child labor as compared to father’s, but on
child’s schooling mother’s education effect is stronger than father’s like urban households.
Regarding the mother’s employment, we have found that it has a negative effect on child labor
in urban and positive effect in rural households. The results are corroborated by first stage results
where urban children from employed mothers are more likely to go to school and rural children from
such mothers are less likely to go to school. A possible explanation for the complementarity between
the mother’s employment and child employment in rural households may be that majority of them
comes from lower strata of the society. They work in informal sector, cottage industries and as
domestic workers. They have meager income (in many cases no income) and are forced to engage their
girls with them in work. Due to poverty (lack of schooling facilities as well) their daughters can not go
to school and mothers hesitate to leave them at home in their absence. They prefer to engage their
daughters in work.
It is found that for urban communities, father and mother’s income has negative impact on child
labor, though the mother income effect is very mild. For the rural communities, father’s income has a
negative impact on child labor but mother’s income has shown no significant effect. The father’s
income effect in urban communities is stronger than in rural communities. The explanation may be that
rural adult’s (father’s) income level are much lower than their urban counterparts, as fathers are the
main contributors in household income, so the rural households sink in to poverty so deeply, where
marginal increase in father’s income can do a slight effect on household welfare and ultimately on
children’s withdrawal from labor. The seasonal employment which has the characteristic of income
fluctuation is more prevalent in rural communities, so an increase in father’s income results into lower
effect on child labor. Majority of the rural labor force is employed in agriculture or agro-business,
though this labor force is employed throughout the year, but their mode of payment is usually seasonal.

Electronic copy available at: https://ssrn.com/abstract=2567431


153 European Journal of Economics, Finance and Administrative Sciences - Issue 27 (2010)

Similarly, a significant ratio of the rural labor force is under-utilized, which ultimately keeps their
income so low that a marginal increase in their income can not enhance the welfare of the household.

3.3.4. Characteristics of Household


The ownership of assets has shown contrasting effects on child labor in urban and rural households.
The children from asset-owner households are less likely to work in urban households but more likely
to work in rural households (see also Nielsen 1998 for rural Zambia; Sharif 1994 for rural India). For
the urban areas, the possible explanation may be that the asset-owner households have ability to
manage economic uncertainty, they can absorb the adverse economic shocks and they have the easy
access to capital market. So the child labor is not required for the purpose. For the rural households, the
assets like live stock, small piece of land, tractor etc, absorbs child labor due to required labor
intensity.
The child labor activity is negatively related to the household per-capita income and poverty
status of the household in both urban and rural households. It corroborates with the first stage results.
The urban households are more sensitive about child labor to these changes. The lower impact of
household per-capita income and poverty in rural households reflects the parent’s tendency towards
child labor, that may be due to poor educational facilities in rural areas, poor educational infrastructure
and low qualified teachers in rural areas. All or some of these factors enhance the negative perception
of the education and positive work tendency.
The probability of child labor is positively related to the household size (as a continuous
variable) in urban as well as rural areas. The relationship is stronger in rural areas. As concerns the
composition of household, the presence of school-age children in the household increases the child
labor force participation in both urban and rural areas but the effect is stronger in urban areas. The
explanation may be that though the resource competition exists in both areas, but due to surplus of
supply of labor in the rural households rural children are comparatively less likely to work as compared
to their urban counterparts.

3.4. Home-care Activity


The fourth stage estimations of model sees all the remaining urban and rural children, that is, the
children not going to school, not doing part-time or full-time work but are doing home-care or remain
in state of no-school and no-work.

3.4.1. Child Characteristics


The home-care activity of the children is positively related to the birth-order of the child in urban and
rural households. Higher the birth-order, that is younger the child in brothers and sisters, it is more
likely for the child to do home-care. It explained the fact of delayed first school enrolment. The birth-
order effect is stronger for rural households, so the first enrolment of children is more delayed in rural
households. The result supported the first-stage estimation results.
The gender of the child is an important determinant of home-care activity of children. We have
found that, girls are more likely in urban as well as rural areas to do home-care as compared to boys.
Rural girls are almost two times more likely to do home-care than urban girls. It showed that all the
factors which involve the girls in home-care activities are stronger in rural households as compared to
urban ones. It supported the notion that school probability is higher for urban as compared to rural girls
(see also first stage results). Furthermore, home-care activity is negatively related to the age of the
child in both urban and rural households but the effect is stronger for rural households. The notion
again confirms the fact that school enrolment is more delayed in rural households.
The home-care activity is found negatively related to the number of years of child’s schooling
in urban as well as rural households, but the relation is stronger in rural households. It is worth noting
from the policy point of view that educational level of the child has stronger negative impact on child
labor and home-care activity in rural as compared to urban households. That is, provision of schooling

Electronic copy available at: https://ssrn.com/abstract=2567431


154 European Journal of Economics, Finance and Administrative Sciences - Issue 27 (2010)

facilities will lower down the child labor and home-care activity more sharply in rural households as
compared to urban ones.

3.4.2. Head of Household Characteristics


Head of household’s education decreases the likelihood of home-care in urban and rural communities.
But the education of head of household has a stronger impact on child’s home-care activity in urban
households, as compared to rural households. It corroborates the first and third stage results that,
education of the head of household has stronger positive effect on child’s schooling in urban
households as compared to rural ones, and a stronger negative impact on child labor in urban
households as compared to rural ones. It means that, an increase in adult education as a policy
instrument to enhance school participation and to pull down the child labor may be more effective in
urban as compared to rural areas.
The head of household’s employment has shown divergent effect on home-care activity of
children in urban and rural households. In urban households children from employed head of
household are less likely to do home-care but in rural areas children from such households are more
likely to do home-care. A possible explanation for the positive impact in rural areas may be the lower
wages and income levels, and existence of under-employment, semi-employment and disguised
employment in rural areas resulting into inability of the households to afford education cost. Similarly,
unavailability of child employment opportunities keep the children involved in home-care in rural
areas.

3.4.3. Parent Characteristics


As concerns the parent characteristics the father’s education as a continuous variable has shown
negative impact on home-care activity of children in urban households but it has shown insignificant
results for rural households. The mother’s education (as a continuous variable) has shown positive
impact on home-care activity in both urban and rural households but the impact is stronger in urban
households. It corroborates the first and third stage results for urban and rural households, i.e. mother’s
years of education impact schooling of the children positively and child labor negatively and the
impact is stronger for urban households. This phenomenon may be explained by two aspects (i) female
adult literacy is much lower in rural households (ii) school availability, quality of education and infra-
structure is unsatisfactory in rural areas. The child labor and home-care both are less affected by
mother’s education.

3.4.4. Household Characteristics


The home-care activity is differently affected by ownership of assets in urban and rural areas. In urban
areas the children from asset-owner households are less likely to do home-care, while in rural areas
they are more likely to do home-care. It explains the different behavior of asset-owner households for
urban and rural areas. In urban areas, the parents from such households prefer their children to go to
school and restrict their part-time work, child labor and home-care activity. In rural areas, the asset-
owner households first prefer their children’ schooling then home-care rejecting part-time work and
child labor. Similarly, the home-care activity of the children is negatively related to the household per-
capita income in both urban and rural areas and the effect is stronger for rural households.
The household size affects the home-care activity of children in the same way for urban and
rural households. Furthermore, combined family system in rural households results into surplus of
labor supply in the households, which makes the children stay at home for home-care. The home-care
activity is positively related to the number of the school-age children in urban as well as rural
households. The resource competition effect is evident again. More the number of school-age children
in the household, less are they likely to go to school (firsts stage results), more likely to work (third
stage results), and more likely to remain at home due to surplus of labor in the households.

Electronic copy available at: https://ssrn.com/abstract=2567431


155 European Journal of Economics, Finance and Administrative Sciences - Issue 27 (2010)

4. Conclusion and Policy Implications


The model and estimation we presented above allow us to analyze the rural urban differences of
children activities. The conclusions of the study are summarized as below:
• Gender disparity in school participation and in labor force participation of children (boys are
more likely to go to school as well as to participate in labor force) exists in both urban and rural
areas. The gender discrimination in school participation is higher in rural areas but in labor
force participation it is higher in urban areas. So from the policy point of view regarding
elimination of gender discrimination in child welfare, rural areas need more stress. Availability
of high quality schools and decrease in cost of schooling would result into decrease in gender
disparity in child welfare.
• The first school enrolment of children is delayed in both urban and rural areas but it is more
delayed in rural areas as compared to urban areas. The rural children dropout of schooling and
enter labor force at an earlier age as compared to urban children. The current years of education
enhance the probability to continue schooling in both urban and rural households but
probability is higher for urban areas. It also decreases the probability to work full-time in both
areas so a trade off exists between schooling and child labor.
• The birth-order effect exists in child welfare in both urban and rural areas as younger brothers
and sisters have lower probability to go to school. On the other hand elder children are more
likely to do full-time work in both urban and rural households.
• The probability for the child to do part-time work increases (at a decreasing rate) by increase in
age in urban households but it decreases in rural households in linear relation.
• Head of household’s education (as a continuous as well as binary variable) impacts the child’s
schooling positively but the effect is much stronger for urban households. That is rural
households perceive the child’s schooling non-productive which need policy stress regarding
awareness to parents about child schooling.
• Employment status of head of household has shown positive impact on child schooling in urban
households but negative impact in rural households. It revealed the existence of disguised
employment and low productivity of labor in rural areas. It has also shown diverging effect on
home-care activity of children, i.e. it lowers home-care activity in urban areas but raises in rural
areas.
• The income level of head of household increases the school participation of children and
decreases the child labor in both urban and rural areas but the effect is stronger in urban areas.
The result leads to policy direction of provision of schools in rural areas.
• In urban areas, mother’s employment is complement to child’s schooling but in rural areas, it is
substitute to child’s schooling that is children accompany their mothers in work in rural areas.
• The mother’s education effect (positive) on child schooling is stronger than father’s in both
urban and rural households, but its effect (negative) on child labor is stronger than father’s in
only urban households. The father’s education effect (negative) on child labor is stronger than
mother’s in rural households.
• The ownership of assets by the household has shown contrasting effects on part-time work,
child labor and home-care activity of children in urban and rural households. The urban
children are less likely to do part-time work, full-time work and home-care but rural ones are
more likely to do this activity in asset-owner households.
• Household per-capita income affects the schooling positively and child labor negatively in
urban as well as rural households but such type of effects are stronger in urban households.
• The poverty status of the household negatively affects the child schooling three times strongly
in rural as compare to urban areas. The poverty alleviation specifically in rural areas needs a
part of policy formulation in the long run, while for the short-term the stipends to the school-
going children in the form of food, school bags and even cash is proposed to slide down the
disparity in child welfare in education perspective.

Electronic copy available at: https://ssrn.com/abstract=2567431


156 European Journal of Economics, Finance and Administrative Sciences - Issue 27 (2010)

•Poverty status of the household affects urban children’ part-time work negatively but rural
children’ positively. It reflects the different structure of employment in urban and rural areas.
• Family size of the household (as a continuous variable) affects the part-time work of urban and
rural children differently. An incremental member in the household increases the part-time
work in urban households but it decreases that in rural households which shows surplus of labor
supply in rural households.
The results of the study clearly indicated that it needs different education and child labor
programs for urban and rural areas. In addition to increasing the future productivity of children, the
provision of urban utilities in rural areas specifically of education would likely produce the effect on
school participation of children. Thus well-targeted rural programs may be seen as optimal economic
investment that would affect both the current and future welfare of rural households and children.

References
1] ADB (2002) Poverty in Pakistan: Issues, Causes and International Perspective. Asian
Development Bank (ADB). Islamabad.
2] Ali, K. and R. E. A. Khan (2004) “Simultaneous Decision Making of Child Schooling and
Child Labor in Pakistani Urban Households”, Lahore Journal of Economics, 9 (1):127-148.
3] Arif, G. M. (2000) “Recent Rise in Poverty and Its Implications for Poor Households in
Pakistan”, Pakistan Development Review, 39(4):453-70.
4] Bhalotra, S. and C. Heady (2003) “Child Farm Labor: The Wealth Paradox”, World Bank
Economic Review, 17(2):197-227.
5] Bhalotra, S. and L. F. Attifield (1998) “Intrahousehold Resource Allocation in Rural Pakistan:
A Semiparametric Analysis”, Journal of Applied Econometrics, 13(5):463-80.
6] Burki, A. A. and L. Shahnaz (2001) “The Implications of Household Level Factors for
Children’s Time Use in Pakistan”, Paper presented at International Conference on Child Labor
in South Asia, New Delhi, 15-17 October, 2001.
7] Burki, A. A. and L. Shahnaz (2003) “School Attendance, Child Labor or Home Production?
Gender Bias in Household Choice”, Lahore University of Management Sciences. Lahore.
8] Cigno, A., F. C. Rosati and Z. Tzannatos (2001) “Child Labor, Nutrition and Education in
Rural India: An Economic Analysis of Parental Choice and Policy Options”, Social Protection
Discussion Paper #0131. The World Band, Washington, D.C.
9] CRPRID (2002) Pakistan Human Condition Report 2002. Center for Research on Poverty
Reduction and Income Distribution (CRPRID), Islamabad.
10] CUTS (2003) Child Labor in South Asia. CUTS Center for International Trade, Economics and
Environment. India.
11] Ersado, L. (2005) “Child Labor and Child Schooling in Urban and Rural Areas: Comparative
Evidence from Nepal, Peru and Zimbabwe”, World Development, 33(3):455-480.
12] FBS (1996) Child Labor Survey. Federal Bureau of Statistics (FBS). Islamabad.
13] Hazarika, G. (2001) “The Sensitivity of Primary School Enrolment to the Cost of Post-Primary
Schooling in Rural Pakistan: A Gender Perspective”, Education Economics, 9(3):237-44.
14] Hou, X. (2009) Wealth: Crucial but not Sufficient Evidence from Pakistan on Economic
Growth, Child Labor and Schooling. Policy Research Working Paper No. 4831. The World
Bank. Washington, D.C.
15] Maitra, P. and R. Ray (2002) “The Joint Estimation of Child Participation in Schooling and
Employment: Comparative Evidence from Three Continents”, Oxford Development Studies,
30(1):41-62.
16] Nielsen, H. S. (1998) “Child Labor and School Attendance: Two Joint Decision”, CLS-WP 98-
15. Center for Labor Market and Social Research, Aarhus, Denmark.
17] Nielsen, H. S. and A. Dubey (2000) “Child Labor in Rural India: A Micro-Economic
Perspective”, The Indian Journal of Labor Economics, 45(3):479-496.

Electronic copy available at: https://ssrn.com/abstract=2567431


157 European Journal of Economics, Finance and Administrative Sciences - Issue 27 (2010)

18] Ray, R. (2000) “Child Labor, Child Schooling, and Their Interaction With Adult Labor:
Empirical Evidence for Peru and Pakistan”, World Bank Economic Review, 14(2):347-367.
19] Sawada, Y. and M. Lokshin (2000) “Household Schooling Decisions in Rural Pakistan”, Study
conducted by Japanese Ministry of Education, The Foundation for Advance Studies on
International Development, and Matsushita International Foundation, Japan.
20] Sharif, M. (1994) “Child Participation, Nature of Work and Fertility Demand: A Theoretical
Analysis”, Indian Economic Journal, 40 (4).
21] Sharif, R. (2000) “The Seclusion Ethic and Educational Attainment and Well-being of
Adolescent Girls”, Briefing Notes in Economics, 52:1-8.

Electronic copy available at: https://ssrn.com/abstract=2567431

You might also like