Professional Documents
Culture Documents
Chap 5
Chap 5
5.1 Introduction
This chapter provides main findings of this study. This chapter consists of two main sections
first section explain elementary data analysis and second section describe the empirical
estimation of models. First section 5.2 comprised on frequency distributions of working child
characteristics, school going child characteristics and household characteristics. While 5.3
section explore the statistical analysis of primary data, correlation analysis and econometric
analysis of child labor, child schooling and household poverty model. Finally section 5.4
gives some concluding remarks.
In this section an effort has been made to describe the statistics collected through a sample
survey in descriptive way. Let it be clear that sample were collected from the households
members, as they know much of the information about the reasons of household poverty and
other factors affecting child labor and child schooling. Here different factors affecting child
labor, child schooling and household poverty are described in frequency distribution form
with some explanation of it. Here following frequency distributions and their sub sections are
made to present the characteristics of working children, school going children and households
to accurate the collected data precisely.
Working children characteristics includes work starting age, reason of work, type of labor
activity, per month earn income, involvement in paid and unpaid work and control on earn
income. All these are explaining as follows.
According to sample survey 23% of households children started work at the age of 12 year in
Multan city while in case of Shujaabad district 17% of households children started work at
the age of 10 year. This table shows that 20% of overall households’ children started work at
the age of 10 year. So, these estimates pretend that child labor started at pre maturity age.
Table 5.1 the percentage distribution of household by children work starting age
This table describes the reasons behind the problem of child labor in two selected districts.
The survey data reveals that 45% children involved in labor activity to supplement their
family income, 25% children involved in labor activity because they are not interested in
school. While 13% children doing work to manage house.
The following table reveals that 43% of children involved in own father business activity,
25% children engaged in shopkeeper activity and 14% involved in other activities. While
children more involved in own father business activity in case of Shujaabad district than
Multan city. Only 1% household children involve in factory or industry work, it is also
manifest that children have less techniques for the participation in industry or factory work.
This table describes that 29% of working children earned income 3 to 5 thousand rupees per
month. The survey data shows that more children earned 7 to 9 thousand rupees in Shujaabad
district than in Multan city. This is also reveals that wage rates lower in Shujaabad district
than Multan city district.
Table 5.4 the percentage distribution of household by children earned income per month
The survey data reveals that 64% (among which 38% children have no control on earned
income and 70% have some control) household children worked under certain paid
conditions while 30% engage in unpaid work. Here the data also shows that children have
more average control on earned income in case of Multan city as compared to Shujaabad.
Characteristics of school going children includes school enrollment age, type of education
children want to be interested, favorite and not favorite subject, inside home working hours
and school absent days. All these are explaining as follows.
The survey data describes 26% of households enrolled their siblings at the age of 5 year and
23% enrolled at the age of 3 year. In case of Multan city 2 households never enrolled their
children in school while in Shujabad district all households enrolled their children. The
survey data also reveals that in case of Multan city more children enrolled in school at the age
of 5 year than in Shujaabad.
Table 5.6 the percentage distribution of household by children school enrollment age
In survey there are four types describe which usually household children wants. The collected
data shows that majority of household’s children wants to get school education and this trend
is greater in Shujaabad district than Multan city, while only 8% household’s children want to
obtain any technical vocational training.
Table 5.7 the percentage distribution of household by type of education children want
In the survey study it was assumed that all the children of households who are assist their
families in domestic work considered not as child labor because this is an unpaid job. The
survey data shows that about 26% of school going children doing inside home work more
than six hours and this trend greater in Multan city than Shujaabad.
Table 5.8 the percentage distribution of household by inside home work duration of school
going children
Table 5.9 the percentage distribution of household by school going children favorite subject
The survey data shows that 41% of overall household’s school going children is not
interested in learning mathematics and this trend is greater in Multan city than in Shujaabad.
While in Shujaabad district 41% of households’ school going children is not like the
computer subject as compared to other subjects.
Table 5.10 the percentage distribution of household by school going children not favorite
subject
The data of survey shows that 71% of household school going children absent one day in a
week from school and this percentage is greater in Multan city as compared to Shujaabad.
While 20% household’s school going children are absent twice in a week. Overall
distribution shows that children school absent days tendency greater in Multan than Shjaabad.
Table 5.11 the percentage distribution of household by absent days of school going children
The survey data describes that about 34% households in which only one person earn income
and this trend more prevail in Shujaabad district than in Multan city. The collected data also
reveals that only 6% of households in which 5 to 6 or more income earners. Overall collected
data also pretend that number of working persons is greater in Shujaabad than in Multan city.
Table 5.12 the percentage distribution of household by number of working person
The survey data shows that about 53% households worked on daily wages and this trend is
greater in Multan as compared to Shujaabad district. While about 40% of households worked
monthly wage and this percentage more in Shujaabad than in Multan city. While only 7%
household worked on weekly wages.
Table 5.13 the percentage distribution of household by mode of income payment
The survey data reveals that about 35% household’s father and mother have no personal
expenses and this tendency greater in Shujaabad than in Multan city. The data also shows that
26% of father and other family members have no personal expenses; this is mostly happen in
Multan city as compare to Shujaabad district and about 14% household mothers have no
personal expenses.
Table 5.14 the percentage distribution of household members have no personal expenses
The data of following table shows that about 43% households in which only father have the
decision making authority and this tendency greater in district of Multan city than in
Shujaabad. While about 41% households in which both father and mother have the decision
making authority and this more in Shujaabad as compare to Multan city, only 5% households
decision made by other family members rather than father or mother.
The survey data describes that about 38% fathers (head of household) are literate either they
are primary, middle, metric and more and this tendency more in Multan city than Shujaabad
district. While about 47% both parents are literate and this trend higher in district of
Shujaabad as compare to Multan city. The survey data also reveals that only 9% mothers’
head of household are educated. The following table shows that 26% households are
illiterate.
In survey three types of assets are taken to estimate the financial position of households, table
shows that 40% of households own all assets and this trend is greater in Multan city than in
Shujaabad. In the sample survey it is reported that 7% households have no ownership of the
following assets. Mostly households of Shujaabad district have less ownership of assets.
The survey data reveals that 60% of households obtain water from deep well pump and this
percentage greater in Shujaabad district than Multan city. The survey data also reported that
33% of households obtain water from Govt. sources and this percentage greater in Multan
city than Shujaabad district. While only 2% household obtained drinking water from other
sources.
Low income of people allows them to received non-labor income either from their relatives
or from Govt. The survey data reveals that about 28% households received non-labor income
among which 95% received from relatives and 5% from Govt. The following data also
pretend that relatives are more financial support the poor household than Govt.
In this study it is also examined that physical structure of houses is an indicator of household
poverty that’s why this term incorporated in this study. The survey data shows that 13%
households have own mud made houses and this trend greater in district of Multan city than
Shujaabad. The data also reveals that 87% households have own cemented made houses.
It is basically considered that poor people purchase necessary goods daily and they have low
purchasing power. The survey data reveals that 32% households buy grocery daily and this
trend is higher in district of Multan city than Shujaabad. While 40% households buy grocery
monthly. While 29% households purchase necessary goods weekly.
The following table reveals that Shujaabad district more populated than Multan city. More
children in labor activity in Multan city and more children go to school in Shujaabad district.
The female male ratio is greater in Multan city than Shujaabad district with less dependency
ratio. About 99% household have per capita income less than Rs.13365.
Several versions of child labor, child schooling and household poverty based on econometric
analysis are available in literature. However the use of some specific versions depends upon
the objectives of the study as well as the data available on parameters of basic concerns.
Therefore to analyze the models precisely three econometric techniques are used.
E-views software was used to determine the statistical analysis of primary data for each
model. To have the clear idea about the structure of the variables, averages, standard
deviations, skewness and kurtosis are given in following table. Table contains six columns.
First contains the name of variables. The mean of the variables are reported in column 2;
column 3 is for median, while standard deviation of the variables is presented in the column 4
and skewness are presented in column 5 and 6 present kurtosis respectively.
Additional descriptive analysis of child labor model provides the comprehensive profile of
working children and its determinants. The following table indicates that the 40%
household’s children involve in labor activity and average age of working child is 11.78
years. The average passed education of working child is 4.3 its means that children left
school before completing primary standard of education. The average family size of
household in the sample is that about 9.4 percent. On the average about 73 percent
households children want to get school education and average age of households head is 43
years. Average per month income of households is Rs.28721 with a standard deviation
Rs.23510; similarly the other variables disclose the sample statistic and about 21 percent
households living below the poverty line.
The following table indicates that mean of total personal expenses (TPE) of households is
Rs.11296.7 with the standard deviation Rs.10498.72. The 22 percent households are living
below the poverty line. The average per capita income (PCI) of households is Rs.3241.2 with
the standard deviation Rs.3334.16. The average number of working person (NOHWP) in a
household is 2.5. The average of households received non-labor (HRNLI) from relatives or
Govt. is 38 percent. About 32 percent poor households bought necessary goods daily
(HBGD) and 40 percent household children involved in labor activity. The following table
also shows that household engaged in outside work about 8 to 9 hours.
Correlation shows the degree of association between dependent and two or more independent
variables. The following correlation matrix describes degree of interdependence between
pairs of variables. To explore the association between variables we use a criterion which is
explained in chapter 4.
If we analyze the results of the correlation coefficient in terms of child labor, correlation
coefficient describes the extent of the relationship between child labor and independent
variables. Here we start one by one analysis of coefficients: child labor activity weak positive
associated with household poverty (PLS), children labor activity moderate positively related
with age of working children (AGE) which means that as the children grow old labor activity
increases. Children want school education (CSE) weak negatively related with child labor,
showing that more children want to get school education less will be the child labor. Obtained
education (EDU) of working children weak negatively related with child labor. Household
family size (FS) did not influence child labor activity, both variables does not impact each
other. Household head age (HA) weak positive related with child labor. While household
monthly income very weak positive related with children labor activity.
The following table reveals that household monthly income (HMI), dependency ratio (DR)
and female male ratio (FMR) are not correlated with child schooling so, they does not
influence child schooling. While households Head education (HHE), children want school
education (CSE) very weak positive correlated with child schooling.
The correlation coefficient of total personal expenses of household (TPE) and per capita
income (PCI) weak negatively related with household poverty line status which shows that
higher the total personal expenses lower will be households faced poverty, similarly higher
per capita income means that lower the number of household’s lies below the poverty line.
The number of household working person (NOHWP) and children labor activity (CL) weak
positively related with household poverty line status which means that higher the number of
working person in a households more chances to fall below the poverty line status. There are
two reasons behind this phenomena first all working person involve in same family business
so, they have no personal income, second more personal spending of working person left less
income of other family members so, they remain under the shadow of poverty and third all
working person work against low wage rates. Household received non-labor income
(HRNLI) very weak positive correlated with household poverty line status. The coefficient of
household bought grocery daily (HBGD) not correlated with household poverty line status.
Household outside home working hours (HOHWH) weak negatively related with household
poverty line status which means that as the outside home working hours increase household
poverty decrease.
The results in this chapter are discussed in the light of hypothesis formulated in chapter 4.
The following tables are interprets the Logit estimates of child labor, child schooling and
household poverty model. Table contains five columns. First contains the name of variables.
The estimated parameters of the variables are reported in column 2; column 3 is for standard
error of the coefficients, while asymptotic Z-statistic of the parameters of the explanatory
variables is presented in the column 4 and probability values are presented in column 5
respectively. While this study used two tailed test of significance or z-statistics for
determining the acceptance or rejection of null hypothesis in order to check the reliability of
the point estimates.
The logistic estimates of child labor model are given below. Where the head age, head age
square, working children age, working children education in completed years, children want
school education, family size, household poverty line status and log of household monthly
income are taken as independent variables and child labor is used as the dependent variable.
With this arbitrary choice, it is observed that the head age (HA) of households is explaining
child labor positively. The coefficient of head age is highly significant. It means that as the
age of household increases child labor also increases. This can be seen as the age of
household head increase working capacity declining after the certain age and it is difficult for
him to bear all financial affairs of his household that why they allow their children to
participate in labor market. In the sample study mean age of household head is 43 years. The
estimation of head age variable is indirectly reconciled with the result of Togunde and
Richardson (2006).
The coefficient of head age square has negative and significant impact on child labor.
Children of upper age household head more involve in labor activity while children of lower
age household head less involve in labor activity. The same result is found by Khan (2003).
Estimation of the model describes that age of child has positive and highly significant impact
on child labor. As the age of working child increases, its impact becomes highly significant
and the value of coefficient also increases. The statistic describes that 1 year increase in child
age increase 1.0 1 percent chances of child involvement in labor activity. The estimate of
child age variable reconciled with the results of Blunch and Verner (2000), Admassie (2003),
Aldaba et al. (2004), Dayioglu, Togunde and Richardson (2006), Hou (2009), Ahmad (2012),
Webbink et al., Onyemauwa and W.A and T.N (2013).
The variable EDU represent the past education standard of working children in years and is
introduced in the model to see whether child past education influenced child labor or not. The
variable is highly significant having negative association with the child labor. The coefficient
of the variable indicates that 1 percent increase in child education decrease 1.09 percent child
involvement in labor force participation. The estimate child education variable is also
consistent with the results of Amin et al. (2006).
No one can observe the child interest in school education in many society, therefore to see the
impact of children want school education variable included in the model for determine child
labor decision. The coefficient for this variable is negative and insignificant. It describes that
more child want to obtain school education fewer would be children involve in labor force
participation. This result indicates that 1 percent change in children interest decreased child
labor 0.59 percent.
It is observed that the family size (FS) of household is explaining child labor positively. The
coefficient of household family size is significant. The coefficient of the variable has positive
association with the child labor decision. This reveals that children belonging to a large
family increase the probability to participate in labor market. This result indicates that 1
percent increase in family size increase child labor 0.13 percent. The estimate of household
family size variable reconciled with the results of Qureshi et al. (2014), Ahmad (2012),
Togunde and Richardson (2006) and Toor (2005).
The coefficient of log of household monthly income (LHMI) has positive sign and
insignificant. Higher income of household leads more children involved in labor activity. In
developing countries low or uneducated parents less wants to invest in children education so,
their children more involve in labor activity. The results of this variable also satisfy the
wealth paradox (i-e as the household become wealthier start child labor increases). The
estimate of log of household monthly income reconciled with the results of Lima et al. (2015)
and Abdullahi et al. (2016).
The variable of PLS represent the household poverty line status and is included in the model
to see whether household poverty effected child labor or not. The variable is significant and
has positive association with child labor. The coefficient of the variable indicates that 1
percent increase in household poverty increase 1.06 percent chances of children to involve in
labor activity. The estimate of household poverty variable consistent with the results of
Blunch and Verner, Ray (2000), Aldaba et al. (2004), Ersado (2005), Hou, Kim (2009),
Onyemauwa et al. (2013) and Qureshi et al. (2014).
The logistic estimates of child schooling model are given below. Where the female male
ratio, dependency ratio, household head education, log of household monthly income and
children want school education are taken as independent variables and child schooling is used
as the dependent variable.
Table 5.31 Logistic Estimate of Child Schooling Model
The result of constant term shows that the effect of intercept term on child schooling is less
important. The value of McFadden R 2 which assures the quality of our estimates shows that
only 26 percent variations in the model are due to explanatory variables.
Estimation of child schooling model describes that household head education has positive and
significant impact on child schooling. More educated household head is in favor of more
children schooling than less educated household head. It describe that household head
education significantly increases child schooling. The result indicates that 1 percent increases
in head education increases child schooling 0.37 percent. The estimate of household head
education variable reconciled with the results of Berenger and Chouchan (2015), Onyemauwa
et al. (2013), Grigoli and Sbrana (2011), Khan (2003), Shapiro and Tambashe (2001), Ray
(2000) and Jensen and Nielsen (1997).
No one can ignore the child interest in school education in any society, therefore to see the
impact of children want school education variable included in the model for determine child
schooling. The coefficient for this variable is positive and significant. It describes that more
child want to obtain school education greater would be child schooling. This result indicates
that 1 percent change in children interest increased child schooling 2.4 percent.
An important variable in the model is log of household monthly income. Log of household
monthly income has a positive coefficient with insignificant value. The sign of the variable is
positive indicating that if the household income increases they will be prefer more children
schooling. The estimated results of the variable contradict with wealth paradox (i-e as the
household become wealthier they will prefer less child schooling) presented by Lima et al.
(2015) and Abdullahi et al. (2016).
The variable dependency ratio (DR) of household indicates the position of household
economically active and non-active person. The coefficient of this variable is also according
to expectation showing negative impact of dependency ratio on child schooling. This result
describes that fewer presence of economically active person in household decreases child
schooling, however this result is insignificant.
The logistic estimates of household poverty model are given below. Where the household
bought grocery daily, per capita income, number of household working person, household
received non-labor income, total personal expenses, household outside home working hours
and child labor are taken as independent variables and household poverty line status is used
as the dependent variable.
The result of constant term shows that the effect of intercept term on household poverty is
less important. The value of McFadden R 2 which assures the quality of our estimates shows
that only 46 percent variations in the model are due to explanatory variables.
In this model we hypothesized that household living below the poverty line mostly they are
working on daily wages so, they bought grocery or necessary goods daily, the estimated
result is as, as was expected in the hypothesis. The result is insignificant and coefficient has
positive value. The positive sign of the variable indicating that if household become poorer,
they will more buy necessary goods daily. The coefficient of the variable indicates that 0.07
percent increment in household poverty due to 1 percent change in household bought
necessary goods daily.
Per capita income of household is another important variable of household poverty model.
The results of the model state highly significant negative impact on household poverty. The
statistic describes that 1 percent decline in per capita income increase poverty 0.0013 percent.
Another important variable is number of household working person, the coefficient sign of
this variable against the expectation showing positive impact on household poverty. But
positivity of the coefficient may be due to that reasons, that mostly working persons involved
in same family business and their participation not changed the productivity and hence
income level, more personal spending of household working person left fewer income for
other members. The results of the model state highly significant positive impact of household
working person on poverty.
Household received non-labor income is another important determinant of household poverty
model. The coefficient sign of this variable is positive and significant. The sign of this
variable is indicating that an increment in household poverty leads the 1.05 percent
attainment of non-labor income. From this result it can be concluded that household poverty
line status positively related with household received non-labor income and same result is
found as was expected.
Keeping in mind the importance of household members outside home working hours, a
variable HOHWH, average outside home working hours included in the model. The
estimated result is as, as was expected in the hypothesis. The statistic shows that the
increment in average outside home working hours decline household poverty. The result is
significant and coefficient has negative value. From this result it can be concluded that if
average household outside home working hours increased household poverty decreased. The
coefficient of the variable indicates that 0.19 percent decline in household poverty due to 1
percent increase in average outside home working hours.
5.4 Conclusion
In this chapter, there was analysis of the data with the help of descriptive analysis and
econometric analysis. Descriptive analysis provide more deep insight of this study and
pretend how own child and household characteristics. While through econometric tools we
are able to determine how own child and household characteristics influence child labor,
child schooling and household poverty. Then there were estimation of the models of child
labor, child schooling and household poverty and interpretation of the selected variables.