ECOSENTIMENTS JOURNAL, VOLUME 1, ISSUE 1, 2013
EMPIRICS OF CHILD LABOUR: A STUDY OF AN INDIAN STATE
MANAS GHOSE Faculty, Raja N.L. Khan Women’s College, Vidyasagar University, Midnapore (West), India
Abstract Using the 62nd round (2005-2006) unit level National Sample Survey data, the study has found that incidence of child labour is still high in India. The worst form of child labour i.e. distress child labour though not significant is prevalent. Incidence of child labour is lower for the families headed by a female member. Incidence of child labour is higher in rural areas than the urban areas. The paper explores that the child labour is mainly caused by poverty. Maximum female education has a positive impact on child labour. Larger number of children within the families implies greater possibility of child labour. This underlines the need for family planning and poverty eradication to root out the evil of child labour. Key words: child labour, family planning, poverty, education. INTRODUCTION There is enough literature on child labour which argues that the principle cause of child labour is poverty. Availability of good schools, distribution of mid-day meal for children at schools, subsidy to the parents who send their children to schools can reduce child labour to a great extent. Nowadays parents are always altruistic about their children, with some exception, so they 1
ECOSENTIMENTS JOURNAL, VOLUME 1, ISSUE 1, 2013
do not like send their children to work if they can afford not to. This has been called “luxury axiom�. A number of studies support this view (Basu and Van, 1998; Basu, 1999; Ray, 2000; Basu and Tzanntos, 2003; Emerson and Souza, 2003; Edmonds and Pavcnik, 2005; Edmonds, 2005). However, the problem of child labour is a multifaceted hydra. It is simply not possible to comprehend the problem using some simple theoretical constructs or causal empirics. In fact, the problem transcends well beyond the narrow boundary of a single discipline. The NSSO 62nd round unit level data provides us a wide arena of information regarding several aspects of child labour. Some of these aspects are considered here. The age and sex of the child labour, nature of the job they perform, major determinants of child labour and their family characteristics - all are considered in this paper. In order to comprehend the problem it is first necessary to distinguish between child labour and exploitation of child labour. It has been accepted that a certain amount of child labour will persist under the family environment, which is non-exploitative. This is not only inevitable but also desirable1. At the same time, there are other forms of child work such as in hazardous occupations, factories and other organized establishments, begging and prostitution, which are repressive and should not be allowed to continue. The Indian Government has taken some steps to alleviate this monumental problem. In 1989, India invoked a law that made the employment of children under age 14 illegal, except in family-owned factories. However, this law is rarely followed, and does not apply to the employment of family members. Thus, factories often circumvent the law through claims of hiring distant family. In addition, in rural areas, there are few enforcement mechanisms, and punishment for factories violating the mandate is minimal, if not nonexistent.
1
Sociologists argue that such minimal sharing of family chores is a joyful activity and an essential part of the socialization process of the child. However when such jobs hinder the basic development of child and her rights, they turn out to be exploitative. In the standard literature (Basu and Van 1998: Basu, Das and Dutta 2007, Bhalotra and Heady 2003), we deal with this second type of child labour.
2
ECOSENTIMENTS JOURNAL, VOLUME 1, ISSUE 1, 2013
Again, there is a definitional problem involving child labour. Barriers of counting child labour arise in developing country like India and it can be described in two ways: one is due to various conceptual disparities and another is due to methodological bottleneck and/or ignorance. However, in this paper we have used a variant of NSSO (National Sample Survey Organisation) definition of child labour. NSSO identified a set of activities as ‘economic activity’ that includes all productive activities, household or market-oriented, undertaken by a child (6-14 years of age) in a paid or unpaid capacity. However, in this paper we have also included some uneconomic activities such as begging, prostitution etc. and the children who are engaged with such activity within the purview of child labour2. Further NSSO introduce a new concept - potential child labour defined as the children worked as a paid or unpaid labour within or outside the family in the childhood age but did not work due to sickness or other reasons or the children who sought or seeking/available for work (Unemployed). This definition is very important. It identifies those fringe areas where the possibility to enter the labour market remains exceptionally high. Inclusion of such concepts within the purview of child labour helps us to broaden the concept of child labour in a meaningful way. In this paper, we want to give a panoramic view about pattern and causes of child labour in an Indian state with the emphasis on verifying the inverted U hypothesis. As the paper proceeds, a brief description of our data and methodology is given. The section that follows describes the results of basic econometric analysis and provides concluding remarks. DATA SOURCES AND METHODOLOGY For empirical analysis, we have used the National Sample Survey Organization (NSSO) 62nd (2005-2006) round Unit Level data on Employment and Unemployment Situation in India. It is a detailed all India data covering all the major states and union territories of India. However, for this paper, we have selected only one state, Orissa for our study. The choice of the state is purposive. There are several reasons for selecting the state. The state shows wide socioeconomic, cultural and religious variations. However, in the recent decades, the state has made
2
In fact, they may be referred to as “distress child labour”. Obviously, they are the worst form of child labour.
3
ECOSENTIMENTS JOURNAL, VOLUME 1, ISSUE 1, 2013
some progress in agrarian growth and rural development. Hence it will be of interest to study how far such changes are reflected in the use and allocation of child labour. Most of the research on this area held the same view that child labour is mainly caused by economic factors such as income and assets (Goldin, 1979; Horan and Hargis, 1991; Bonnet, 1993; Basu and Van, 1998; Basu, 1999; Ray, 2002; Bhalotra and Heady 2003; Basu and Tzanntos, 2003; Emerson and Souza, 2003; Edmonds and Pavcnik, 2005; Edmonds, 2005; Basu, Das, Dutta, 2007). But in this paper we aim to analyse the influence of family education level, number of children in the family and ratio of consumer to adult worker. Again, our data enables us to compare the situations in the rural areas with the urban areas, male-headed families as against the female-headed families etc. Since not all the sample families use child labour, the dependent variable is truncated at zero for some families. Hence, the use of Ordinary Least Squares (OLS) technique is inappropriate for this analysis. Instead, we propose to use the Tobit regression technique that is suitable for such analysis. RESULTS AND DISCUSSIONS Basic information We now consider some basic socio-economic features of the sample state; they are given in tables 1, 2 and 3. We have worked out the results on 3013 families, out of these 60.77% are rural families and 39.23% are urban families. Classifying another way we have found that 90.74% families are headed by a male member and 9.26% families are headed by a female member. Table 1 Demographic profile of the sampled population Child sex ratio (0-14) years
941
Child sex ratio, below 6 years Child sex ratio (6-14) years Sex ratio of child labour (6-14 years) Sex ratio of child labour (10-14 years) Sex ratio of potential child labour (6-14 years)
951 935 923 1004 1000
Sex ratio of potential child labour (10-14 years) 926 Adult sex ratio 958 nd Source: Estimation based on unit level data of NSS 62 Round, 2005-06, EmploymentUnemployment Survey.
4
ECOSENTIMENTS JOURNAL, VOLUME 1, ISSUE 1, 2013
From table 1 we have seen that adult sex ratio (958) is not bad but child sex ratios are 941 (0-14 years) and 951 (below 6 years) are relatively bad. This is a clear reflection of misuse of modern technology. Sex ratio of child labour for the age group 6-14 years is 935 and it is 1004 for the age group 10-14 years. This reflects the fact that female child comes to labour market at a higher age than the male child does. However, the situation is reversed if we consider the potential child labour where it is 1000 for the age group 6-14 years and 926 for 10-14 years. This is because parents are not bothered about their girl child. They are sick at a lower age when their immunity power is low and they get recovery as age increases. Table 2 Educational attainment by the adults Average years of schooling by adult male
6.6
Average years of schooling by adult female
4.2
Average years of schooling by the family head
4.5
Literacy rate
68
Source: Same as Table 1 Next we consider the educational attainment of the sample families. Literacy rate of the sampled families is bad (68%) and average years of schooling by adult male (6.6) as well as adult female (4.2) both are low. Average years of schooling by adult female is much lower than average years of schooling by adult male that imply prevalence of high gender deprivation. Table 3 Poverty and access to public distribution of the sample Percentage of APL families
12.61
Percentage of BPL families
30.20
Percentage of Antyodaya families
2.12
Percentage of families which have others type ration card
11.78
Percentage of families which have no ration card
43.18
Total
100
Source: Same as Table 1 5
ECOSENTIMENTS JOURNAL, VOLUME 1, ISSUE 1, 2013
Our next important issue is the incidence of poverty. Basu and Van (1998) epitomized the issue in their famous luxury axiom. It is argued that families that can barely sustain their subsistence might opt for child labour. The official estimate of poverty - percentage of Below Poverty Line (BPL) families is high enough (about 30%) while that of the very poor (Antyodaya families) is insignificant (about 2%). However, a sizeable portion of the families (43.18%) has no ration card. It might be a reflection of illegal transnational migration (possibly from the neighboring countries) and poor administration. In that case, the conditions of these families are very precarious. Incidence of child labour Incidence of child labour for different categories is shown in table 4, 5 and 6. By incidence we mean the percentage of children, (6-14) years, working as child labour. Table 4 Sex wise incidence of child labour (percentage of respective total child of 6-14 years) Name of the variables
Male
Female
Incidence of child labour, (6-14) years
31.32
30.94
Incidence of child labour, (10-14) years
18.64
20.04
Incidence of potential child labour, (6-14) years
7.21
7.71
Incidence of potential child labour, (10-14) years
4.47
4.43
Source: Same as Table 1 Table 4 shows the sex wise incidence of child labour. From the table we have seen that incidence of child labour is higher for the male child for the age group 6-14 years and lower for the age groups 10-14 years. This fact again establishes that girl child comes to labour market at a higher age. But the situation is reversed for potential child labour. This fact again establishes that female child is continuously neglected.
6
ECOSENTIMENTS JOURNAL, VOLUME 1, ISSUE 1, 2013
Table 5 Region wise incidence of child labour (percentage of respective total child of 6-14 years) Name of the variables
Rural
Urban
Incidence of child labour (6-14) years
31.99
29.56
Incidence of child labour (10-14) years
19.20
19.52
Incidence of potential child labour (6-14) years
6.74
8.66
Incidence of potential child labour (10-14) years
3.95
5.31
Source: Same as Table 1 An important dimension is the rural-urban divide. This factor did not get proper attention in the mainstream literature of child labour. The incidence of child labour is higher in rural areas due to the abundance of employment opportunities, for instance, helper in the farm work. However, in the urban areas such opportunities are missing. However, incidence of potential child labour is higher in urban areas. That may be due to higher unemployed child in the urban areas. Table 6 Incidence of child labour according to sex of the house head (percentage of respective total child of 6-14 years) Name of the variables
Male headed families
Incidence of child labour (6-14) years
31.49
Female families 26.26
Incidence of child labour (10-14) years
19.43
17.88
Incidence of potential child labour (6-14) 7.47 years Incidence of potential child labour (10-14) 4.31 years Source: Same as Table 1
headed
7.56 6.15
Table 6 helps to compare the incidence of child labour between the households, which are headed by a male member, and the households, which are headed by a female member. From this table we have seen that incidence of child labour, for both the age groups, is relatively higher for the families headed by a male member compared to the families headed by a female member. 7
ECOSENTIMENTS JOURNAL, VOLUME 1, ISSUE 1, 2013
However, incidence of potential child labour is relatively larger for the female-headed families compare to male-headed families. This implies that the conversion of child labour from potentiality to reality is much lower for the female-headed families. Pattern of Child labour Let us now consider the pattern of child labour- what are the type of jobs our children are doing. Tables 7 and 8 show the pattern of child labour for our sample. NSSO divides total child labour into eight classes. These eight classes may be grouped into four broad categories: (i) child labour in household enterprises, (ii) child labour as regular/casual wage labour, (iii) child labour in domestic services, and (iv) destitute child labour. It is clear from table 7 that household enterprises dominate for both the age groups; it is near 36% for both the age groups. Wage labour takes the second best position in our sample; it is 32.64% for the age group 6-14 years and 34.93% for the age group 10-14 years. Unlike the household enterprise and domestic child labour, this category shows a different trend. For this category, incidence increases as age increases, that means children begin to earn at a higher age. The incidence of domestic child labour is also sizable; it is near 28% for the age group 6-14 years but it is 26% for the age group 10-14 years. That means as age increases children divert their work from domestic servant to wage labour or destitute. Again, the last category is worst from welfare point of view. It represents the distress child labour because in this category children’s activities are begging, prostitution, etc representing the worst form of child exploitation. It is disgraceful fact that this category constitutes near 3% for both the age groups. More child come to this activity as age rises.
8
ECOSENTIMENTS JOURNAL, VOLUME 1, ISSUE 1, 2013
Table 7 Pattern of child labour (% of child labour) Categories
6-14 years
10-14 years
Household enterprises (including unpaid family workers)
36.78
35.70
Wage labour
32.64
34.93
Domestic servants
27.96
25.93
Destitutes
2.62
3.44
Source: Same as Table 1
Table 8 Pattern of potential child labour (% of potential child labour) Categories
6-14 years
10-14 years
Worked but did not work due to sickness
29.31
28.86
Worked but did not work due to other reasons
36.78
36.50
Unemployed
33.91
34.64
Did not seek but was available for work
0.00
0.00
Source: Same as Table 1 Table 8 shows the pattern of potential child labour. NSSO divides potential child labour into six classes, these six classes may be grouped into four broad categories to facilitate our analysis: (i) Worked but did not work due to sickness (ii) Worked but did not work due to other reasons (iii) Unemployed (iv) Did not seek but was available for work. From table 8 we have seen that nearly 34% of total potential child labours are unemployed. This implies high supply pressure of child labour in the child labour market that reduces child wage as well as adult wage rate (if we assume that child labour and adult labour are substitute). Percentage of child unemployment increases as age increases that implies as age increases more and more children come to child
9
ECOSENTIMENTS JOURNAL, VOLUME 1, ISSUE 1, 2013
labour market. Nearly 29% potential child labour worked but did not work due to sickness. This implies a poor health condition of the children. Results of Basic Econometric Analysis The above analysis was descriptive. To get a proper picture some further analytical study on the determinants of child labour is necessary. However since the incidence of child labour is probabilistic, there may be a large number of families without any child labour. Consequently, Ordinary Least Squares technique is inapplicable here. Instead, one should use some truncated variable technique for this purpose. In our analysis, we have used the tobit regression technique.
Table 9 Tobit regression results (dependent variable: total child labour) Dependent variable
Total child labour 6-14 Total child labour 10-14 years
years
Independent variables
Coefficient(t statistic)
coefficient(t statistic)
Per capita consumption
-0.29252E-03(-2.2341)**
-0.42438E-03(-2.6285)***
No. of child in the family
0.53866(12.040)***
0.40147(7.5469)***
Average years of education by adult -0.82010E-02(-0.70428)
-0.37554E-02(-0.26614)
male Maximum female education(years)
0.40427E-01(3.6549)***
0.71865E-01(5.3108)***
Ratio of consumer to male worker
0.70100E-01(2.1354 **
0.97689E-01(2.4947)**
Ratio of consumer to female worker 0.19522(5.7663)***
0.20754(5.0758)***
Constant
-2.9583(-18.954)***
-3.6206(-18.703)***
Total child labour
(28.999)***
(23.084)***
Log-Likelihood
-1774.7485
-1384.2035
No. of observations
3013
3013
*Significant at 10% level, **Significant at 5% level, ***Significant at % level Table 9 shows the tobit regression results. In both the cases dependent variable is total child labour; first case is for the age group 6-14 years, second case is for the age group 10-14 years, and in both the cases, we regress our dependent variable on the same set of explanatory 10
ECOSENTIMENTS JOURNAL, VOLUME 1, ISSUE 1, 2013
variables. They are per capita consumption, number of children in the family, average years of education, maximum female education (years), ratio of consumer to male worker, ratio of consumer to female worker and a constant term. Here we use per capita consumption as the proxy of per capita income. From the regression result, we have seen that in both the regressions per capita consumption is highly significant and sign of the coefficient is negative implying the fact that as per capita consumption or income increases extent of child labour decreases. This proves that child labour is caused by poverty. Our second factor is number of children in the family. For our data, the relevant coefficient is highly significant for both the regression. That means child labour increases with the number of children in the family. When number of children increases amount of consumption increases but income remains same so to survive the family adults send their child to work. Our third and fourth factors reflect the education level of the family. From the regression result, we have seen that an average year of education by adult male is insignificant in both the cases. That means schooling of the adult male have no significant impact on child labour. However, maximum female education (years) is highly significant in both the regressions and sign of the coefficient is positive. That implies as years of education by the adult female increases extent of child labour is also increases. This may be because of as female education increases then female worker shifts their work from informal/domestic to formal that increases demand for labour for the informal/domestic work. Our fifth and sixth factors measure the work burden of the familial duties and side by side, these variables measure the family sustainability. The work burden of the family will be less if these ratios are small. On the background of income, a family will be more sustainable if the ratio of consumer to worker is low. From the regression result, we have seen that ratio of consumer to male worker and ratio of consumer to female worker are highly significant for both the regression equations. This reflects the fact that as ratio of consumer to male worker and ratio of consumer to female worker falls family gets away from poverty trap and they have no necessity to send their children to work. 11
ECOSENTIMENTS JOURNAL, VOLUME 1, ISSUE 1, 2013
Table 10 Tobit regression result (dependent variable: total child labour 6-14 years) Dependent variable
Rural child labour
Urban child labour
Independent variables
Coefficient(t statistic)
coefficient(t statistic)
Per capita consumption
-0.63910E-03(-2.3097)**
-0.22497E-03(-1.9963)**
No. of child in the family
0.58862(9.9098)***
0.47460(7.0279)***
Average years of education by adult -0.33343E-01(-2.1321)**
0.28275E-01(1.4716)
male Maximum female education(years)
0.50429E-01(3.4655)***
0.27068E-01(2.1518)**
Ratio of consumer to male worker
0.44145E-01(2.32638)**
0.17828(3.5426)***
Ratio of consumer to female worker 0.20003(4.6099)***
0.18561(3.4671)***
Constant
-2.5599(-12.248)***
-3.4132(-12.515)***
Total child labour
(23.219)***
(17.410)***
Log-Likelihood
-1120.3090
-644.74433
No. of observations
1831
1182
Table 10 shows two regression results one for rural and another for urban. In both the regression, our dependent variable, total child labour aged 6-14 years, is regressed on same set of explanatory variables as earlier. From the regressions result, we have seen that in both the regressions per capita consumption is significant and sign of the coefficient is negative implying the fact that as per capita consumption or income increases extent of child labour decreases. This proves that child labour is caused by poverty. Our second factor is number of children in the family. For our data, the relevant coefficient is highly significant for both the regression. That means child labour increases with the number of children in the family. When number of children increases amount of consumption increases but income remains same so to survive the family adults send their child to work. 12
ECOSENTIMENTS JOURNAL, VOLUME 1, ISSUE 1, 2013
Our third and fourth factors reflect the education level of the family. From the regression result, we have seen that an average year of education by adult male has a significantly negative impact on child labour for the rural areas but it is insignificant for urban areas. However, maximum female education (years) is highly significant in both the regressions and sign of the coefficient is positive. That implies as years of education by the adult female increases extent of child labour is also increases. Our fifth and sixth factors measure the work burden of the familial duties and side by side, these variables measure the family sustainability. The work burden of the family will be less if these ratios are small. On the background of income, a family will be more sustainable if the ratio of consumer to worker is low. From the regression result, we have seen that ratio of consumer to male worker and ratio of consumer to female worker are highly significant for both the regression equations. This reflects the fact that as ratio of consumer to male worker and ratio of consumer to female worker falls family gets away from poverty trap and they have no necessity to send their children to work. CONCLUSION Our analysis shows that the incidence of child labour is still high in India. The worst form of child labour i.e. distress child labour though not significant is prevalent. Incidence of child labour is lower for the families headed by a female member. Incidence of child labour is higher in rural areas than the urban areas. The negligence towards female children continues intact. From the regression result we can conclude that the child labour is mainly caused by poverty. Maximum female education has a positive impact on child labour. Larger number of children within the families implies greater possibility of child labour. This underlines the need for family planning and poverty eradication to root out the evil of child labour. REFERENCES 1. Baland, J-M and James A Robinson (2000): ‘Is Child Labour Inefficient’, Journal of Political Economy, Vol.108, No.4, pp.663– 679. 2. Basu, K (1999): ‘Child Labor: Cause, Consequence, and Cure, with Remarks on International Labor Standards’, Journal of Economic Literature, Vol.XXXVII, pp.1083–1119.
13
ECOSENTIMENTS JOURNAL, VOLUME 1, ISSUE 1, 2013 3. Basu, K (2000): ‘The Intriguing Relation between Adult Minimum Wage and Child Labor’, The Economic Journal, Vol. 110, No.462, pp.C50– C61. 4. Basu, K and P H Van (1998): ‘The Economics of Child Labor’, American Economic Review, Vol. 88, No.3, pp.412– 427. 5. Basu, K, Das Sanghamitra and Bhaskar Dutta (2007): ‘Child Labour and Household Wealth: Theory and Empirical Evidence of an Inverted-U’, DP NO.2736, IZA, Germany. 6. Bhalotra, S and C Heady (2003): ‘Child Farm Labour; The Wealth Paradox’, World Bank Economic Review, Vol. 17, No.2. pp.197-227. 7. Cain, M T (1977): ‘The Economic Activities of Children in a Village in Bangladesh’, Population and Development Review, Vol. 3, No.3, pp.201– 227. 8. Carpio Ximena V Del (2008): ‘Does Child Labor Always Decrease With Income? An Evaluation in the context of a Development Program in Nicaragua’, WP No.4694, World Bank. 9. Cigno, A and F C Rosati (2000): ‘Why Do Indian Children Work and Is it Bad for Them?’, (mimeo), University of Florence. 10. Dreze, J and Geeta Gandhi Kingdon (2001): ‘School Participation in Rural India’, Review of Development Economics, Vol.5, No.1, pp.1 –24. 11. Emerson, P M and A P Souza (2000): ‘Is There a Child Labor Trap? Intergenerational Persistence of Child Labor in Brazil’, Available online at http://people.oregonstate.edu/~emersonp/childlabor/childlabor.pdf, Accessed on February 20, 2013. 12. Fields, G. S. (1995): Trade and labour standards: A review of the issues, Paris: Organisation for Economic Cooperation and Development. 13. Grootaert, C and R Kanbur (1995): ‘Child labor: an economic perspective’, International Labor Review Vol.134, No.2, pp.187– 203. 14. International Labour Organization (1996): ‘Child Labour Surveys: Results of Methodological Experiments in Four Countries, 1992–93’, International Programme on the Elimination of Child Labour, ILO, Geneva. 15. Jensen, P and H S Nielsen (1997): ‘Child labour or school attendance? Evidence from Zambia’, Journal of Population Economics, Vol.10, pp.407–424. 16. Khasnabis, Ratan and Tania Chatterjee (2007): ‘Enrolling and Retaining Slum Children in Formal Schools: A Field Survey in Eastern Slums of Kolkata’, Economic and Political Weekly, Vol.42, No.22, pp.2091-2098.
14
ECOSENTIMENTS JOURNAL, VOLUME 1, ISSUE 1, 2013 17. Maskus, K E and J A Holman (1996): ‘The economics of child labor standards’, Discussion Paper 96/10, Department of Economics, University of Colorado, Boulder, CO. 18. Parsons, D O and C Goldin (1989): ‘Parental altruism and self-interest: child labour among late nineteenth century American families’, Economic Inquiry, Vol. 27, No.4, pp.637–659. 19. Psacharopoulos, G (1997): ‘Child Labor versus Educational Attainment: Some Evidence from Latin America’, Journal of Population Economics, Vol.10, No.4, pp.377–386. 20. Ranjan, P (1999): ‘An economic analysis of child labor’, Economic Letters, Vol.64, Issue 1, pp.99– 105. 21. Ranjan, P (2001): ‘Credit constraints and the phenomenon of child labour’, Journal of Development Economics, Vol.64, Issue 1, pp.81– 102. 22. Ravallion, Martin and Quentin Wodon (2000): ‘Does Child Labour Displace Schooling? Evidence from Behavioural Responses to an Enrollment Subsidy’, The Economic Journal, Vol.110, No.462, pp.C158–C175. 23.
Ray,
R
(1999):
‘Poverty,
Household
Size
and
Child
Welfare
in
India’,
Available
at
http://www.utas.edu.au/__data/assets/pdf_file/0020/208244/1999-01.PDF, Accessed on January 15, 2013. 24. Ray, R (2000): ‘Poverty, Household Size and Child Welfare in India’, Economic and Political Weekly, Vol.35, No.39, pp.3511-3520. 25. Ray, R (2002): ‘The Determinants of Child Labour and Child Schooling in Ghana’, Journal of African Economies, Vol. 11, No.4, pp. 561-590.
15