literature reviewschool of economics, finance and management

1、 Child Activities in South Asia and Sub-Saharan Africa: A Comparative AnalysisPublished in P. Lawrence and C. Thirtle, eds., Africa and Asia in Comparative Development, (London: Macmillan), September, 2001.Sonia BhalotraUniversities of Bristol & Cambridge and Christopher HeadyUniversity of BathC

2、orrespondence to : Dr Sonia Bhalotra, Department of Applied Economics, Cambridge CB3Tel 01223 513045; Email: srb32hermes.cam.ac.uk Acknowledgements: We are grateful to Fiona Coulter and Catherine Porter for excellent research assistance. This work was funded by the Economic and Social Research Counc

3、il (grant number R000237121) and the Department for International Development as part of its Employment and Labour Markets Programme.AbstractThis paper profiles child activities in the rural areas of Ghana and Pakistan using large nationally representative household survey data that use comparable d

4、efinitions in the two countries. The data refer to 7-14 year olds in Ghana and 10-14 year olds in Pakistan. School enrollment rates for Ghanaian boys and girls and Pakistani boys are between 70-80%. However, a Pakistani girl is less than half as likely as the other children in the sample to be at sc

5、hool. Children in Ghana do not work for wages but, in Pakistan, 6% of boys and 12% of girls are in fairly full-time wage labour, to the clear neglect of schooling. Just more than a third of children in both countries are engaged in work on the household farm or enterprise. There is great variation i

6、n their hours of work but they average to approximately half-time levels. In both countries, there is evidence that own-farm work lowers school attainment though self-employment and school are more easily combined by children in Ghana than in Pakistan. A substantial fraction of children in both coun

7、tries are neither in work nor in school, and this fraction is especially large amongst girls. The presence of this group indicates that child labour is not the only barrier to child schooling. We compare the determinants of child labour in the two countries, including household living standards, hou

8、sehold human capital and demographics, and community-level data on schools and infrastructure. The data describe prominent differences in the environment that children grow up in. We then present a summary of the determinants of the variation in child work across households within each country. Inte

9、resting contrasts across country and gender are highlighted.Keywords: child labour, school attendance, poverty, female education, household size, home production, Ghana, PakistanChild Activities in South Asia and Sub-Saharan Africa: A Comparative AnalysisSonia Bhalotra and Christopher Heady1. Introd

10、uctionWhile South Asia has the largest number of working children, Sub-Saharan Africa has the highest incidence of child labour. Child work participation rates are 41% in Africa as compared with 21% in Asia and 17% in Latin America (Ashagrie, 1998). Comparative work is a first step in gaining an ins

11、ight into the universality of the problem of child work. South Asia and Sub-Saharan Africa are clearly very different environments, their common ground being that the average household, at least in rural areas, is poor. We compare the determinants of child labour in the two countries, including hous

12、ehold living standards, household human capital and demographics, and community-level data on schools and infrastructure. The data describe prominent differences in the environment that children grow up in. We then present a summary of the determinants of the variation in child work across household

13、s within each country. Interesting contrasts across country and gender are highlighted.There have been a number of empirical studies of child labour in recent years. However, as Basu (1999) points out, there remains considerable scope for good empirical work in this field. At this stage, patterns ar

14、e only just beginning to emerge among the variety of results in the literature, corresponding to the vast variety of regions, types of child work, and empirical specifications. The main contribution of this paper is that it uses comparable micro-data on two countries where child participation rates

15、in work are currently very high, in an attempt to determine whether any patterns can be discerned amidst the vast (observed and unobserved) heterogeneity in household conditions. Existing beliefs about the causes and consequences of child labour have tended to be shaped by case studies. These typica

16、lly interview working children. An advantage of using large scale representative household surveys, as is done here, is that we have comparable information for children who do not work.The paper is organised as follows. Section 2 profiles child activities in the early 1990s in the rural areas of Gha

17、na and Pakistan, and looks at the correlation of child labour with education. In Section 3, we discuss a range of variables that theory and existing evidence suggest have an impact on child labour. Section 4 compares the central tendencies of these potential explanatory variables across Ghana and Pa

18、kistan so as to describe the similarities and contrasts in the environments in which children live. In Section 5, we compare the means of the variables across the sub-samples of working and non-working children, for each country and each gender. This indicates the likely significance of the variable

19、s. In a natural extension of this analysis, Section 6 pools household-level data on working and non-working children and tobit models of hours of child work provide estimates of the size and significance of the range of variables considered, holding the others constant at their mean levels. The anal

20、ysis of Sections 4-6 is restricted to child labour on the household farm, because this is the type of child work that is both more prevalent and more directly comparable across the two countries. Conclusions and policy implications are presented in Section 7. 2. A Profile of Child Activities - By Ge

21、nder and CountryIn this section we present empirical evidence on school attendance as well as on the prevalence, intensity, and nature of child labour in rural areas. We discuss the extent to which school and work are combined by children, and we highlight the fact that a substantial proportion of c

22、hildren, especially girls, are neither in school nor at work (though they may well be engaged in domestic work). Interesting inter-country and gender differences in child activities emerge.The DataThe data are drawn from the Ghana Living Standards Survey (GLSS) for 1991/2 (wave 3) and the Pakistan I

23、ntegrated Household Survey (PIHS) for 1991. The GLSS contains 4552 households, with an average household size of 4.5 members, giving a total of 20403 individuals. Almost 50 per cent of males and 45 per cent of females are aged under 15. The Pakistan Integrated Household Survey for 1991 contains 4795

24、 households. On account of a much larger mean household size of 7.5 members, we have a sample of 36,109 individuals. About 43.5% of males and 45.5% of females in the sample are aged under 15. The GLSS collects data on employment for persons 7 years or older whereas the cut-off is at the age of 10 in

25、 the PIHS. While there are important differences in the two datasets, their structure and coverage are sufficiently similar to allow some interesting cross-national comparisons. Participation rates of rural children in school and in different kinds of work are presented in Tables 1-2 for Ghana and 4

26、-5 for Pakistan, for 7-17 year olds in three age groupings. Data on 15-17 year olds is of interest in so far as it illustrates how the school and work patterns of under-15s evolve with age. The discussion that follows refers to 10-14 year olds. Participation Rates in SchoolIn rural Ghana, 79% of boy

27、s and 72% of girls aged 10-14 years are “currently in school”. In contrast, school attendance in Pakistan is remarkably larger for boys as compared to girls of all ages. Amongst 10-14 year olds, 72% of boys and only 31% of girls are in school. Except for girls in Pakistan who appear to withdraw into

28、 the household at an early age, school participation is lower in the 7-9 and 15-17 year ranges suggesting late entry and early exit (the latter, especially for girls). Consistent with this, participation in all sorts of work, for boys and girls, tends to increase steadily with age.Participation Rate

29、s in WorkIn Ghana, 49% of boys and 44% of girls undertake work on the household farm, about a further 3% of each gender are engaged in household enterprises, while less than 1% report any employment outside the household. In Pakistan, 22% of boys and 28% of girls work on the household farm, about 2%

30、 of both genders work in a household enterprise, and 6% of boys and 12% of girls work outside the home. The striking difference between the two countries is that children in Pakistan engage in wage work outside the household, which Ghanaian children do not. The data show that this is more or less fu

31、ll-time work The wage economy is better developed in Pakistan, possibly on account of the relative scarcity of land in South Asia as compared with Sub-Saharan Africa. Therefore, for adults and children alike, wage employment presents productive opportunities where the returns to self employment are

32、relatively limited. Child work outside the home in Pakistan is analysed in Bhalotra (1999). and that, for girls, it is predominantly seasonal agricultural work, whereas the large fraction of boys who work outside the home are engaged in non-agricultural work.Data on domestic work, which includes fet

33、ching firewood or water, cooking, cleaning, laundry, shopping and child care, were collected for boys and girls in Ghana and for girls alone in Pakistan. Virtually all children participate in domestic work of some sort. In addition (not shown in the Tables), about 5% of Pakistani girls engage in hom

34、e work for sales, an activity in which boys have no part. It appears unusual for children to be active in more than one type of work though a small fraction of Pakistani children do combine wage work and farm work (see Tables 3 and 6).Hours of WorkHow hard do working children work? See Table 7. Mean

35、 hours increase with age for all groups other than Pakistani girls working for the household. Work intensity for the household is similar among girls and boys in Ghana. Pakistani girls exhibit lower intensity work, while boys in Pakistan work harder than the other sub-groups. The Table shows wide di

36、spersion around mean hours but, overall, household work would not appear to be full time work. Indeed, as we shall see below, many children combine farm work with school attendance. On the other hand, wage work, which is restricted to children in Pakistan, is more or less full-time work. Boys work o

37、n average 45 hours a week while girls work about 30 hours a week. Competition between Work and SchoolingRefer to Tables 3 and 6. Of Ghanaian children who work on the household farm, almost three in four boys and girls are at the same time in school. Combining farm work and school would appear to be

38、less easily done in Pakistan, where almost half the boys but only one in ten farm-working girls manage it The correlation of school attendance (a binary variable for the individual) with work-participation and hours of work was examined for 7-17 year olds, holding constant age, household size, curre

39、nt household expenditure per capita, and all cluster-specific effects. The conditional correlation of work participation with school participation in Ghana is positive but increasing hours of work did appear to reduce the probability of school attendance. In Pakistan, both participation and hours of

40、 child work are negatively correlated with school attendance (results available from the authors on request). Similarly, in Ghana, virtually all boys and almost half of the girls combine working on the household enterprise with going to school while, in Pakistan, this is rare amongst the boys and un

41、known amongst girls. Child wage work is virtually absent in Ghana but in Pakistan, where it occupies about 6% of boys and 12% of girls, it clearly interferes with schooling: less than 1% of children combine wage work and school. Overall, in all its forms, child work in Pakistan is much more evidentl

42、y in competition with school attendance han is the case in Ghana.A remarkable fact is that, in both countries, a substantial proportion of children neither work nor go to school. In Pakistan, this is 35% of girls and 10% of boys and in Ghana, it is 14% of girls and 8% of boys. These fractions are es

43、pecially large among girls. Therefore, if the main concern is with low educational attainment (and the gender gap therein), then policies designed to discourage child labour may be rather less important than policies that directly promote school attendance.3. Potential Determinants of the Variation

44、in Child Work In this Section, we discuss the variables that both theory and existing evidence suggest have an impact on child labour. This underlies the empirical analysis in the remainder of the paper.Bhalotra and Heady (1999, section 4) present a theoretical model of household choice relating to

45、the supply of child labour. The key issue is the allocation of child time between work and school. This time allocation will depend on the relative advantages of work and school, both in the short-term and in the long-term.The main quantifiable short-term advantage of work is the wage that is receiv

46、ed. In the context of working for the household, which section 2 showed to account for most of the child labour in Ghana and Pakistan, the wage will typically not be explicit. Instead, it is necessary to look for variables that determine the marginal product of child work. These include child charac

47、teristics such as age and sex, household characteristics such as land at the disposal of the household as well as household size and composition, and community level characteristics such as local infrastucture and indicators of local labour demand (for example, the region of the country or recent ch

48、anges in local economic well-being). The main quantifiable measures of the short-term attractiveness of schooling is the access to schooling and quality of schooling, both of which are defined at the level of the community.The long-term advantages of work and schooling depend on their ability to rai

49、se the childs future earnings. This is hard to measure in itself. However, there are observable variables that affect the relative weight that households put on the short-term and long-term advantages. In a perfect capital market, the relative weights will just depend on the rate of interest, but th

50、is is an extreme assumption. In reality, it is reasonable to think of households having differential access to credit and thus placing different weights on current and future benefits. As access to credit typically varies with income, this corresponds to the popular belief that children are more lik

51、ely to work the poorer the households they come from. We therefore include household income in the analysis, proxied by the natural logarithm of per capital food expenditure. In addition, household assets such as land and the availabiltiy of local risk-sharing institutions may influence households a

52、ccess to credit, and thus their ability to keep their children out of work.These economic considerations will be modified by household and community attitudes to the importance of education and the appropriate treatment of children. This might depend on the age and sex of the child and whether he or

53、 she is the child of the household head. There is also evidence of birth order effects on the distribution of household investments across children (e.g. Das Gupta (1987), Butcher and Case (1994). Attitudes to education will depend on the education of the childs parents, although it is important to

54、distinguish the effect of parental education on attitudes from its effects on household income. Attitudes will also depend on which ethnic and religious groups the household belongs to.Many of these variables have been shown to play a role in determining childrens labour market participation by stud

55、ies based on similar household surveys, such as Canagarajah and Coulombe (1998), Jensen and Nielsen (1996) and Patrinos and Psacharopoulos (1997)4. Variation in Explanatory Variables Between Ghana and Pakistan4.1. Child Work on the Household FarmHenceforth, the discussion will concentrate on farm wo

56、rk which employs by far the largest proportion of children in both countries and which is comparable across them. The vast majority of working children in developing countries, whether in Brazil or Pakistan or Ghana are engaged in agricultural work. Yet, this work is severely understudied as compare

57、d with the more visible forms of work in Latin America and Asia, which involve children in labour-intensive manufacturing. 4.2. Contrasting Environments for Children in Ghana and PakistanThe discussion here is a summary of the broad magnitudes relating to some interesting contrasts between the rural

58、 regions of the two countries in the environments in which children work. Recall that children in Ghana are 7-14 and in Pakistan are 10-14 years old. Refer to Tables 8 and 9.Approximately half the children in each sample are first-born children and something like three in four are children of the ho

59、usehold head. However, cross country similarities stop at child characteristics. It is not uncommon in Ghana for a household to own more than one plot of land, but this is unusual in Pakistan. In Ghana, households appear to have access to freely available or village owned land, which is not the case in Pakistan. Sharecropping is much more prevalent in Pakistan than in Ghana, involving almost 40% of households as compare