Data And Descriptive Statistics

Published Date: 02 Nov 2017

This paper is aimed at providing an assessment of the validity of the notion that school attendance and child labour are substitutable for each other. The paper specifically aims at examining whether an improvement in the access to school represented by the cost and distance to school are effective policy tools to use to enhance school attendance and combat child labour. The relationship between child labour and school attendance is different for boys and girls, may depend on the duration of work as well as on the type of work that is either within or outside the household.

The data is from a household survey drawn from rural Pakistan and has information on the duration and incidence of child labour on children between 10 and 14 years old as well as information on school attendance. The restriction of the data to children between the ages of 10 and 14 years is in line with the ILO’s definition of child labour in the Pakistani context. The data set also has information on community, household and individual variables. Also included is information on factors of direct and indirect cost (distance) of schooling.

The average weekly hours children spend in different work activities is presented in Table 1. From the data, on average, children spend approximately 16 hours working. While girls spend about 23 hours working a week, boys spend only a total of 9 hours working per week. This large difference can be attributed to the contributions girls make by helping with household chores. In most rural parts of developing countries, men and boys usually do not contribute to household chores. On the other hand, whiles

Table 2 below displays the full descriptive statistics of the independent variables in the data set. Cost of schooling is measured by three variables namely direct cost of primary school education, distance to closest primary school and distance to closest middle school. The average for direct cost of schooling is about Rs. 126 per year. Distances to the closest middle and secondary school are the other measures of cost of education. Their effect on labour supply is however uncertain. The assumption is that a high transportation cost associated with schools that are further from the community may impact on school attendance by reducing school. The logic therefore is that while there is a negative relationship between distance of community from the closest middle and secondary school, the relationship will be positive with child labour. Table 1 shows that averagely, a child must travel 4.5 kilometres to a secondary school and 6 kilometres to a middle school. Closeness of community to school on the other hand will make it child labour easier on children. This therefore implies a positive relationship between child labour and distance from school.

The statistics conditional on child work status suggest a positive relationship between labour supply and direct primary schooling costs (Table XXXX).

Table 1: Average Weekly Hours of Work by 10-14 Year Old Children

Hours of Work

Total

Male

Female

Total Number of Hours

15.75 (24.47)

8.67 (18.98)

23.36 (27.26)

Hours per week worked in household

13.50(22.50)

6.29 (15.14)

21.27 (25.61)

Hours per week worked out of household

20.25(10.18)

2.39 (11.32)

2.11 (8.78)

The statistics in Table 4 suggest there is a positive relation between labour supply and distance to school. On average, a non-working child must travel about 4 km to a middle school while one engaged in extra-household labour must travel about 9 km.

Table 2: Descriptive Statistics of Full Sample

Mean hours worked per week

 

 

 

Variable

N

Mean

Std. Dev.

Child Attributes

 

 

 

Male = 1

2199

0.518

0.500

Age in years

2199

11.851

1.420

Ever Attended School = 1

2197

0.605

0.489

Educational Attributes

 

 

 

Distance to closest secondary school

2199

4.479

7.257

Annual cost of Primary school education

2127

125.612

168.244

Distance to closest middle school

2156

6.028

11.066

Proportion of local primary school with electricity

2154

0.372

0.37

Proportion of local primary school with water

2154

0.542

0.394

Proportion of local primary school with brick wall

2154

0.838

0.273

Household Characteristics

 

 

 

Highest education level of Male Adults

2106

4.227

4.74

Highest education level of Female Adults

2173

0.657

2.204

Total household Landholding

2199

2.892

10.207

Household size

2199

9.865

4.599

Number of male members aged 0-4

2199

0.635

0.858

Number of female members aged 0-4

2199

0.703

1.004

Number male members aged 5-9

2199

0.94

1.01

Number female members aged in 5-9

2199

0.906

0.999

Community and Regional Attributes

 

 

 

Cluster average daily male weekly wage (Rupees)

2199

39.969

11.722

Distance of community from Tehsil capital (km)

2199

21.531

15.808

Access to canal irrigation

2199

0.609

0.488

Visited by agric extension worker in the past 6 months

2199

0.547

0.498

Access to paved road

2186

0.684

0.465

Community variables are also likely to have an effect on household decisions. Some of these variables include access to agricultural extension officers and irrigation canals, proximity to school and village accessibility, can significantly affect agricultural productivity. These factors may affect the demand for child labour. Employment opportunities are also concentrated in more populous areas and as such, the degree of remoteness of a community can have an impact on the demand for child labour.

Figure 1: Frequency Distribution of Children by Gender and Age

The sample is evenly distributed between the 2 gender groups, 52 per cent male and 48 per cent female. The average age for the sample is 11.85 years. The largest age group is the 12 year-olds (26.5 per cent), followed by the 10 year-olds (25.5 per cent the) 14 year-olds (18 per cent), 11 year-olds (15.1 per cent) and the 13 year-olds (15 per cent).

The sample is generally from large household size. The average family size is 9.9. Large family sizes can reduce the pressure on to work on individual children marginally and as a result, children from larger households are more likely to attend school compared to children from smaller households. The level of education tends to be higher among the male adults compared to the female adults. While he males have on average 4.23 years of education, the female adults only have 0.66 years of education.

Table 3 indicates that compared to children who do not work, children who work outside of the households come from families that have lower landholdings and less educated adults. It can be concluded from the data then that in total Pakistan, children who are engaged in labour outside of the household are from poorer families. Tables 2 and 3 indicate that there is no marked difference between children who do not work and those who do only work inside of the household.

Table 3: Descriptive Statistics of Selected Variables by Child Work Status

Variable

Non-Working

Working

Children engaged in work outside of household

Child Attributes

Mean

Std. Dev.

Mean

Std. Dev.

Mean

Std. Dev.

Age

11.581

1.352

12.060

1.438

12.418

1.420

Ever Attended School = 1

0.845

0.362

0.420

0.494

0.270

0.445

Educational Attributes

 

 

 

 

 

 

Annual cost of Primary school education

125.783

154.971

131.850

180.803

152.226

192.691

Distance to closest middle school

4.040

8.457

7.294

12.377

9.284

10.532

Distance to closest secondary school

4.420

7.192

4.614

7.501

6.674

8.537

Proportion of local primary school with electricity

0.371

0.373

0.365

0.365

0.382

0.380

Proportion of local primary school with water

0.541

0.395

0.545

0.396

0.473

0.406

Proportion of local primary school with brick wall

0.841

0.267

0.830

0.280

0.805

0.267

 Household Attributes

 

 

 

 

 

 

Highest education level of Female Adults

0.827

2.503

0.515

1.934

0.193

1.105

Highest education level of Male Adults

4.826

4.987

3.815

4.544

2.219

3.327

Total household Landholding

2.812

10.230

2.865

9.312

0.425

1.231

Household size

10.255

5.276

9.678

4.101

8.723

2.808

Number of male members aged 0-4

0.620

0.872

0.655

0.854

0.546

0.779

Number of female members aged 0-4

0.743

1.125

0.690

0.927

0.603

0.745

Number of male members aged 5-9

0.952

1.066

0.947

0.968

0.993

0.982

Number of female members aged 5-9

0.925

1.086

0.889

0.941

0.723

0.887

Mean hours worked per week

Variable

N

Mean

Std. Dev.

Works within the household

2199

0.531

0.499

Works off farm, hrsout >0

2180

0.056

0.230

child works - hrswk ~=0

2071

0.587

0.492

Total hours worked past week

2199

15.746

24.468

Hours per weeks worked in household

2199

13.500

22.151

Hours per weeks worked outside household

2193

2.252

10.181

Child Attributes

Variable

N

Mean

Std. Dev.

Male = 1

2199

0.518

0.500

Age in years

2199

11.851

1.420

Ever Attended School = 1

2197

0.605

0.489

Household Attributes

Variable

N

Mean

Std. Dev.

Highest education level of Male Adults

2106

4.227

4.740

Highest education level of Female Adults

2173

0.657

2.204

Total household Landholding

2199

2.892

10.207

Household size

2199

9.865

4.599

Number of male members aged 0-4

2199

0.635

0.858

Number of female members aged 0-4

2199

0.703

1.004

Number male members aged 5-9

2199

0.940

1.010

Number female members aged in 5-9

2199

0.906

0.999

Community and Regional Attributes

Variable

N

Mean

Std. Dev.

Cluster average daily male weakly wage (Rupees)

2199

39.969

11.722

Distance of community from Tehsil capital (km)

2199

21.531

15.808

Access to canal irrigation

2199

0.609

0.488

Visited by agric extension worker in the past 6 months

2199

0.547

0.498

Access to paved road

2186

0.684

0.465

Educational Attributes

Variable

Obs

Mean

Std. Dev.

Distance to closest secondary school

2199

4.479

7.257

Annual cost of Primary school education

2127

125.612

168.244

Proportion of local primary school with electricity

2154

0.372

0.370

Proportion of local primary school with water

2154

0.542

0.394

Proportion of local primary school with brick wall

2154

0.838

0.273

Distance to closest middle school

2156

6.028

11.066

Average Weekly Hours of Work by 10-14 Year Old Children

Hours of Work

Total

Male

Female

Total Number of Hours

15.75 (24.47)

8.67 (18.98)

23.36 (27.26)

Hours per week worked in household

13.50(22.50)

6.29 (15.14)

21.27 (25.61)

Hours per week worked out of household

20.25(10.18)

2.39 (11.32)

2.11 (8.78)

4. Model Specification and Variable Choice

I have chosen to use two dependent variables, namely school attendance status of the child (sch) and the working a status of the child (work). The dichotomous nature of these 2 variables allows for the examination of the probability of whether a child goes to school or work in reaction to changes in the cost of schooling. The sample also allows for the testing of the effect of child labour supply on whether or not a child works at home or in the. Both school attendance and work are binary variables. School attendance takes on a value of 1 if the child is in school and 0 if the child is not. Working status also takes on a value of 1 if the child works and 0 if the child does not.

The allocation of a child’s time is likely to be jointly determined by work and school attendance. This means that a child’s participation in the work and school are both endogenous and as such it would be wrong to use one as an endogenous variable of the other on the one hand and ignoring the endogeneity problem may also leads to spurious correlation between the schooling outcome and work status. To solve the issue of endogeneity, one can use instrumental variables (IV) to correct for the effect of endogeneity between the two variables. I will however not attempt to use the IV approach since I cannot seem to come up with a valid instrument that affects child labour without having a direct effect on school attendance. I have therefore decided to use a reduced form approach to analyse the connection between school, work status and the direct cost of schooling and the degree of substitution between child work and schooling. For this reason, I use 2 different models to analyse the effect of the direct cost of schooling on school attendance and another one to examine the effect of the direct cost of schooling on child work status. Since schooling competes with economic activities in the use of a child’s time, they can be thought of as substitutes in the sense that an increase in the price of one leads to a decrease in the consumption of the other and vice versa. Therefore, a decrease in the direct and indirect cost of schooling should lead to a reduction in the child work and an increase in child school attendance. The model I intend to use to analyse this relationship is thus;

Schi =δ1 + β1 dircosti + γ1 Xi + μi 1

Worki =δ2 + β2 dircosti + γ2 Xi + νi 2

where,

Schi = school attendance for ith child

Worki = working status of ith child

dircosti = cost of schooling

Xi = independent variables

δ, β, γ = coefficients to be estimated

μi, νi = error terms depicting individual unobservable effects