Report on Child Work-Load in Schools

How Much Work Is Too Much?
Impact of Child Work Hours on Schooling – Evidence of Cameroon

By

Arline Kengne Kamga
University of Yaounde II – Soa

Faculty of Economics and Management
Email : arlinekamga@yahoo.fr
Tel: (237) 99 46 91 56

Abstract
The paper examines the linkages between child work hours and school performance of 11 391 children aged 5-17 years using Cameroon based on National Household Survey (ECAM 3, 2007). A bivariate probit model was used to select only those children who work and attend school. The endogeneity of hours of child labor has been corrected using the method of the double least squares (2LS). The non linearity between school performance of children and work hours was taken into account through the inclusion square work hours and cube of work hours of children in the estimation. School performance of children was estimated using a Tobit model with double truncation. The main finding of econometric analysis is that, the first hours of child labor have a negative effect on school performance. The result suggest that a child can work up to somewhere between 13-34 hours a week without suffering a loss in her school performance.
Keywords: Child Work Hours, School performance and Cameroon.

1. Introduction
Child labour is a pervasive problem throughout the world, especially in developing countries. This can be explained by the fact that, from a development perspective, the problem of child labour in Africa is not one of enforcement of international labour standards, but it is one of poverty and household survival. In general, child labour takes place in what can be called “informal child labour” , that is to say children working in small businesses.
The International Labor Office (ILO) defines child labor as any activity other than study or playing, remunerated or unremunerated, that is carried out by a person under the age of 15 (14 in certain countries). According to the International Labour Organization (ILO), for example, 218 million children aged 4-15 were trapped in child labour in 2004 of whom 126 million were in what ILO refers to as ‘hazardous’ work (ILO 2006a). Of these children, 69% were engaged in agriculture, 22% were in services, and 9% were employed in industry. While Asia and the Pacific region harbour the largest population of child labourers, Sub Saharan Africa (SSA) is top in terms of activity rate with 26.4% of the children aged 5-14 engaged in economic activities followed by Asia and the Pacific region (18.8%).
A number of factors are responsible for the high incidence of child labour in developing countries, many attributing this to poverty and poverty related factors. It is argued that households that do not have enough resources to sustain the family have no choice but make their children engage in various activities to make ends meet. In such cases, not having the children work puts the very existence of the family at risk. Limited access to (quality) schooling is also among the factors identified as encouraging child labour. In areas where there is little or no access to (quality) schooling, parents may consider child work as an opportunity to help their children develop future “career”.
In Cameroon, between 1996 and 2001, thanks to the combined effects of economic growth and the redistribution of its fallout, poverty had noticeably fallen; dropping from 53.3% to 40.2%. By contrast, between 2001 and 2007, the economy did not post similar performances; resulting in poverty remaining virtually flat over the period (ECAM 3, 2007).
According to MICS (2006), 31% working children aged 6 to 14 years old in the labour force. Overall, 2% of Cameroonian children aged 5 to 14 years are in paid employment outside the household and 11% are involved in unpaid work for the benefit of a third person. In addition, nearly 83.3% of children performed in household chores. At the same time, the net enrollment rate in primary schools is 80%, namely it is 82% among boys and 77% among girls. This rate increases with the child’s age, level of maternal education and quintile of wealth index. Moreover, the net enrollment ratio in secondary education is 38% including 39% among males and 37% among girls. It is lowest in the North (10%), the Far North (14%) and rural (19%).
In addition, the stability of poverty observed at the national level masks a striking contrast between areas of residence. In urban areas, indeed, the monetary poverty rate that stood at 17.9% in 2001 declined by 5.7 percentage points over the period between 2001-2007; while it rose by 3 percentage points in rural areas during the same period and only stood at 55.0% in 2007 (ECAM 3 , 2007).
Most of the recent concern over child labour, as is evident from the rapidly expanding literature on the subject, stems from the belief that it has a detrimental effect on human capital formation. This is reflected in the close attention that child schooling has receiving in several studies on child labour.
Much of this evidence is on the impact of children’s labour participation rates, rather than hours worked by children, on child schooling. This reflects the fact that data on child labour hours is much more difficult to obtain than that on child labour participation rates. However, from a policy viewpoint, knowledge of the impact of child labour hours on a child’s school school performance is as important as that of child labour force participation rates. This raises the question: Is there an “acceptable” threshold of weekly hours of work beyond which school performance are negatively impacted?
The reminder of the paper is organized as follows. After briefly reviewing related studies in Section 2, Section 3 describes the survey data and gives a data description of tabulations on child labor and school participation trends in Cameroon based on National Household Survey (ECAM 3, 2007), Section 4 outlines the empirical methodology employed to analyze the impact of child labour hours on a child’s school performance in Cameroon. Section 5 discusses the estimation results obtained and where relevant showing evidence of similar findings from other studies while the final section concludes the paper.

2. Review of literature
This section reviews some of the existing theoretical and empirical literature on household schooling and child labor decisions from the developing world (for reviews, see Basu, 1999; Brown et al., 2002; Dar et al., 2002; Edmonds, 2007; Edmonds and Pavcnik, 2005; Glewwe, 2002; Hannum and Buchman, 2004; Udry 2003).
Theoretical explanations have been given on how households decide children’s participation in schooling and labour activities. Rosenzweig and Evenson (1977), Rosenzweig (1978) and DeTray (1983) were among the pioneers to take account the interrelationships between children’s multiple activities, insofar as they recognized the potential trade-offs between children’s labor force participation et their school attendance. They were also among the first to analyze children’s hours of market work in conjunction with their schooling. The studies by Rosenzweig and Evenson (1977) and Rosenzweig (1978) are, focused on the economic value of children to their parents and the implications of the children for fertility. Detray (1983) also writes for the discussion on the economic value of children of that time period but frame his results broadly; he specially mentions, for example, possible effects of children’s household work on their time in school.
The 1990s saw a boom in empirical studies of child work, often analyzed jointly with the school attendance or enrollment of the children (see Edmonds, 2008; Orazem and Gunnarsson, 2003; Bhalotra and Tzannatos, 2003 for recent reviews).
Basu and Van (1998) state that child participation in work activities occurs under the conditions of multiplicity of equilibriums in the labor market. In a one-period model, they show that the type of activities that households choose for their children are part of any of two stable equilibriums : a high-wage equilibrium in which children do not work and a low-wage one in which both adults and children do. Basu and Van derive their conclusion from two axioms: the substitution axiom which postulates that child and adult labor are perfect substitutes and the luxury axiom which states that the household chooses not to send children to work if household income from non-child source is high enough.
Baland and Robinson (2000) also studied households’ decision mechanism of whether to send child to work or to school. In a two-period model taking into account the future returns from investment in children education, they demonstrate that parents will choose a socially efficient level of child labor if they are not subject to borrowing constraints or if they can make a bequest or receive a transfer from their children. But, if parents lack access to the credit market or if there is neither transfer nor bequest possibility, they will engage their children in the labor market at an inefficiently high level.
The empirical literature has mainly been occupied with the schooling decision, merely viewing child labor as the lack of schooling. Chao and Alper (1998) analyze the access to basic education in Ghana for children between 10-14 years of age. Two supply-side factors that reduce participation are identified, namely (1) distance to primary school, and (2) pupil-teacher ratio at the primary level.
More recently, the literature has moved to incorporate the work decision and thus analyzing schooling and child labor jointly. Nielsen (1998) analyzes child labor and schooling in Zambia. A gender gap is established, as boys are found more likely to go to school than girls. However, there does not seem to exist any gender related differences in the working decision. Transport costs in the form of walking distance to school affects schooling adversely. Nielsen does not find a positive relation between poverty and child labor, and thus raises doubts to the claim of poverty being a main determinant of child labor, see, e.g., Grootaert and Kanbur (1995).
Using a micro level data, a number of studies investigated the causes and consequences of child labour, with a particular emphasis on the link between child labor and schooling. These studies suggests that the direct and indirect costs of schooling affect household schooling and child labor decisions (Ravallion and Wodon, 2000; Edmonds, 2005). In the developing world, households face direct costs of schooling, such as tuition, fees, donations, books, supplies, uniform, transportation, private tutoring, and miscellaneous costs. If the cost (direct as well as indirect) of sending children to school is high, then poor households will be forced not to send their children to school or to take their children out of school.
The empirical literature on child labour has focused attention on its causes (i.e. its determinants) rather than its effects. There is relatively little evidence in the published literature on the impact of a child’s labour hours on his/her educational experience, especially on his/her performance at school. The present study provides Cameroon evidence on this issue, which is national and international concern.
This paper adds to the existing literature on child labour in Africa by assessing the evidence on the welfare cost that child labour entails on human capital. Previous investigations include the studies of Canaragarajah and Neilsen, 1999; Akabayashi and Psacharopoulos, 1999; Heady, 2000; Rosati and Rossi (2001) and Djenouassi, 2009. The general consensus that emerges from the results of these studies is that child labour is harmful to human capital accumulation. For example, Akabayashi and Psacharopoulos (1999) observe that “a trade-off between hours of work and study exists…(;) hours of work are negatively correlated to reading and mathematical skills through the reduction of human capital investment activities”. Heady (2000) similarly observe on Ghanaian data that work has a substantial effect on learning achievement in the key areas of reading and mathematics. Rosati and Rossi (2001), using data from Pakistan and Nicaragua, conclude that an increase in the hours worked by children significantly affects their human capital accumulation.
Based on the analysis of a national survey (ECAM 3, 2007); the present study seeks to determine the effect of work on children’s schooling. It examines whether a relationship exists between the hours of children’s work and schooling outcomes by attempts to identify a “threshold”, if one exists, where the effect of work on schooling changes from benign or positive to harmful, in that school performance is negatively affected.
3. Data and Descriptive Analysis
This study investigates how many hours can be undertaken before negative effects on school performance are observed, using the third Cameroon Households Consumption Survey (ECAM 3, 2007) conducted by the National Institute of Statistic (INS). ECAM 3 is a national survey of household consumption. It concerns the household as well as individuals who belong to this household. This survey differs from investigations ECAM I and II in that, it includes sections on economic and domestic activities of household members aged 5 years at least.
The survey questionnaire contains 17 sections covering the main theme of poverty (13 sections) and other related topics. Section 3 which deals with items relating to education (the different household members), section 4 that will focus on the economic activities of children over 5 years and Section 5 which provides information on domestic activities of household members. The database consists of a single file that provides each individual personal information, as well as, those relating to his household.
In calculating the child’s labour hours, the study uses the standard ILO definition of child work, i.e. economic activity, which includes work provided on the labour market and work for household farms and enterprises, even if it is unpaid.
The survey covered a sample of 11 391 children and 17 550 households, which were interviewed on the nature of the economic activities of each child within the household, the consequences and challenges faced by each child while in employment, and the amount of time the child spent on his/her studies and recreational activities as well as on economic activities and household chores.
Descriptive evidence
Table 1: Enrollment rates among Cameroonian children in 2007
Age Urban Rural
Male Female Total Male Female Total
5 82.5 86.2 84.2 50.2 45.6 47.9
6 90.9 91.1 91.0 67.5 58.4 63.0
7 95.4 95.7 95.5 79.3 75.4 77.2
8 96.1 95.3 95.7 84.3 70.8 77.8
9 97.6 96.3 96.9 89.3 82.9 85.9
10 97.9 96.2 97.1 84.8 76.8 80.9
11 97.2 96.9 97.1 88.5 84.1 86.5
12 96.8 95.0 96.0 82.5 79.5 81.1
13 95.5 89.3 92.3 84.7 75.6 80.4
14 88.3 84.6 86.3 85.8 69.5 77.9
15 80.2 77.2 78.7 75.6 60.4 68.5
16 76.5 70.8 73.5 73.3 49.7 61.0
17 64.6 67.8 66.2 70.3 38.7 55.3
Total 89.0 87.6 88.3 77.9 67.2 72.6
Source: from Cameroon Households Consumption Survey (ECAM 3, 2007)
Table 1 show child enrollment participation rates by age, sex and location. In 2007, there are 88.3% of children attending school in urban areas against 72.6% in rural areas. These disparities are exacerbated when one is restricted to children 5-9 years old. In this age, the enrollment rate is 92.8% in urban areas against 69.8% in rural areas. It seems that children residing in rural areas start school late.
For all age groups, current school attendance rate are lowest in rural areas compared to urban areas. This lower enrollment rates in rural areas in Cameroon may reflect a lack of access to good schools. For the age group 5-10 years, disaggregating by age and sex shows that enrollment rate difference by gender grows wider with age in urban areas.
Table 2: Employment rates among Cameroonian children in 2007
Age Urban Rural
Male Female Total Male Female Total
5 1.7 4.6 3.0 11.2 8.6 10.0
6 4.3 3.3 3.8 21.2 21.6 21.4
7 9.4 5.9 7.5 33.9 36.3 35.1
8 11.3 11.7 11.5 45.8 43.8 44.8
9 13.5 15.7 14.6 45.5 56.1 51.0
10 13.9 16.2 15.0 61.8 60.3 61.1
11 14.7 19.0 16.8 58.2 59.9 59.0
12 21.0 25.3 22.9 64.9 66.4 65.6
13 28.0 20.8 24.2 67.0 67.2 67.1
14 27.6 22.3 24.7 66.7 70.1 68.3
15 35.1 24.7 29.8 68.6 69.6 69.1
16 36.2 32.4 34.2 70.5 76.2 73.5
17 41.0 28.5 34.6 71.6 75.7 73.5
Total 19.6 17.3 18.4 50.9 51.8 51.3
Source: from Cameroon Households Consumption Survey (ECAM 3, 2007)
On the other hand, child employment rates go in an opposite direction to enrollment, as shows in Table 2. There is a higher employment rates in rural areas compared to urban areas. This, possibly suggest that dropping out of school is at least partly driven by employment decisions. Labor force participation grows with age. In urban areas, 18.4% of children worked whereas 51.3% did in rural areas. This difference is more pronounced among children 10-14 years-old (20.7% against 64.1%). It is increasingly apparent that the age where most children work is that of 15-17 years-old. By focusing on a comparative analysis by gender in urban and rural areas, it is noted that unlike the urban, the proportion of economically active girls (51.8%) exceeds that of boys Milieu Rural (50.9 %).

Table 3: Children time allocation into employment, schooling, and/or both (%)
By residence By sex
Activity Urban Rural All Boys Girls All

Study Only 74.8 37.0 48.9 50.3 47.5 48.9
Work Only 5.1 15.6 12.3 10.3 14.3 12.3
Work and Study 13.4 35.8 28.7 31 26.3 28.7
Neither 6.7 11.6 10.1 8.4 11.9 10.1
Total 100 100 100 100 100 100
Source: from Cameroon Households Consumption Survey (ECAM 3, 2007)
Work and study are not mutually exclusive categories, and do not exhaust the list of possibilities. Somme children are reported attending school, while at the same time performing some form of paid or unpaid work. Others are reported doing apparently nothing (neither attending school, nor working). We created four mutually exclusive and exhaustive categories: work only, study only, work and study neither work nor study. As Table 3 shows, the majority of children (over 50 %) study only. The second largest category (29 %) attends school and works at the same time. The rest work only (about 12 %) or neither work nor study (10%). This table suggests that in Rural Cameroon, the proportion of children who are both working and going to school at the same time is 35.8 %.
Table 4: Child labour and schooling participation by sector (%)
Work Only Work and Study
Urban Rural Urban Rural
Type of activity
Farmer 12.4 29.4 87.6 70.6
Employee from Trade Sector 23.0 36.9 77.0 63.1
Employee out of Trade Sector 44.7 82.7 55.3 17.3
Industry 62.3 44.0 37.7 56.0
Services 31.4 35.9 68.6 64.1
Total 27.4 30.3 72.6 69.7
Employment status
Independent 55.4 55.9 44.6 44.1
Regularly employed 86.5 77.4 13.5 22.6
Unpaid family worker 11.8 29.3 88.2 70.7
Paid family worker 21.4 19.6 78.2 80.4
Total 27.5 30.3 72.5 69.7
Source: from Cameroon Households Consumption Survey (ECAM 3, 2007)
About 41% of the children in the 5-17 age groups are reported to be engaged in paid or unpaid work in Cameroon. In rural areas, there are 30.3% of working children and only 69.7% combining work and education. In addition, 82.7% of working children are employed only outside of business while 64% of children who combine work and school are employed in the service sector. In urban areas, most working children are employed only in industry (62.3%) and are regularly employed.
4. Estimation Methodology
Conceptual model
The econometric analysis of the data sets is based on a two-part estimation methodology.
(a) The study uses a bivariate probit model to estimate the determinants of the household’s decision to put the child in one of four observables states, namely, the child (i) attends school and does not work, (ii) attends school and works, (iii) neither attends school nor works, and (iv) works and does not attend school. However, analysis of the effect of child labor on school performance means that it focuses only on children who work and attend.
Child labour and school attendance is two activities mutually exclusive of each other because the working time is part of the time taken on education and vice versa. In this context, work opportunities and schooling for children choices are interdependent. Using a bivariate probit model is therefore necessary to test the probability of children working and / or going to school. Both dependent variables are the dichotomous variable “work” that takes the value 1 when the child participates in the labor force and 0 otherwise, and the dichotomous variable “school” that takes the value 1 if the child participates in the system education and 0 otherwise.
In the Bivariate probit, let the latent variable represent the decision of working and represent the decision of schooling. Therefore the general specification for a two-equation model would be:

Where and are error terms with normal distributions, and is coefficient of correlation between the two equations.

In this regard, we can consider that and represents the expected net gain to household i, respectively, to work and children to attend school. However, only choices if the child works and 0 otherwise and, if the child attends school and 0 otherwise are observed.
and are row vectors of exogenous variables which determine respectively, working and schooling propensities and and are associated parameter column vectors.
(b) The exercise, then, moves on from estimating participation/non-participation rates to estimating learning measures with special attention paid to the impact of child labour hours, consistent with the principal objective of this study. The simultaneity between the schooling outcomes and child labour hours is recognized by jointly estimating them as a two-equation simultaneous equation system, using 2SLS method of estimation.
The School performance of the child expressed as a linear function of the child work hours , the square of the work hours , the cube of the work hours and child, family, household head and education system characteristics.
The inclusion of both the child work hours variable , and its square is designed to allow and test for the possibility that the impact of work hours on the school performance , changes direction beyond a certain critical value of child work hours . That possibility exists if, as we generally observe in the estimations, and are each statistically significant and have reverse sign. In that case, the critical value of child work hours, at which its impact on school performance reverses direction, is given by:

Where , are the estimated values.
This approach can also be improved by introducing into the equation for school performance of children, which allows two critical values of work hours and . That possibility exists if, as we generally observe in the estimations, , and are each statistically significant and have reverse sign. In that case, the critical value of child work hours, at which its impact on school performance reverses direction, is given by the solution of the equation:

Where , and are the estimated values.
The child labour hours equation of the child expressed as a linear function of the child, family, household head and education system characteristics.
The simultaneity of the decisions about schooling and child labor was used to summarize these two equations in the following system:

Where, the inverse Mills ratio is obtained using the method Lee (1983) from estimating the bivariate probit model of work and school attendance. Its introduction into the equation of working hours will correct the selection bias because we select only the sample children who combine work and school attendance. Formally, lambda is given by:

Where and represent respectively the density function and cumulative function of the normal distribution. is the probability of a child combining work and school.
In the system, there is an endogeneity of child work hours as an explanatory variable in an equation that estimates its impact on school performance of the child. By “endogeneity”, we mean that child work hours are determined by child’s schooling variables as well as vice versa. There are several reasons for this endogeneity. For example, a child’s labour market status could refer her school performance as much as the other way around. Consequently, the estimates in the regression of the child’s schooling variables on the labour market status are likely to be inaccurate. Heady (2000), in his study on Ghanaian data, recognizes the endogeneity issue but does not tackle it in the estimation.
Unlike Ranjan and Lancaster (2005) who try to correct this endogeneity by using the instrumental variables (IV) method of estimation, we use the two-stage least squares (2LS) method of estimation because, the instrumental variables method faces problems of choice of instruments and their validity.
2LS method of estimation
Stage 1:
Formally, the reduced equations of the system can be given by:

Where

This stage involves:
• Estimate each equation of the reduced form;
• Determine the predicted values , of , respectively.
Stage 2:
The second step is to estimate the structural equations of the form by replacing the endogenous variables by their predicted values. .
is variable between 0.1 and 1.67 so, it is truncated. The estimation method suitable for this variable is the Tobit model with double truncation .

Measure of school performance
Much of the recent literature has used test scores as a measure of school performance in studying how this learning outcome variable is impacted by child work hours. The present study departs from this practice for, principally, two reasons.
First, tests scores are not available for children in the data that have been considered in this study. Second, the possession of reading, language and mathematical skills that the test scores measure, offer only a very limited picture of “school performance”, especially in the context of developing country.
As Orazem and Gunnarsson ((2003) point out the “years of schooling completed” measure is only appropriate for parents and adults. A more appropriate measure for this study is the “schooling for age” (SAGE) variable that measures schooling attainment relative to age. It is given by:

Where is the age of first registration of the child . The fact that a child have attended nursery school or not is tackle in the variable , it will be higher for children who were in a nursery school.
Empirical Model
Variables used in the estimation are presented in this section. Previous studies in sub-Saharan Africa suggest that labour participation is influenced by different child, parent and household characteristics. Consequently, these factors are assumed to be important determinants of work/school participation.
Child characteristics
Work Hours is the number of hours worked per week by a child 5 to 17 years.
Age is a variable that measures the child’s age in years. Most activities on cocoa farms are heavy tasks that are not appropriate for children with inadequately developed muscles. It is therefore more likely that older children will be more involved in market work. Also due to the delay in enrolling children in school, it is more likely that older children will be enrolled in school. We hypothesized AGE CHILD to be positively related to WORK and negatively related to SCHOOL. The model includes a quadratic in child age to determine any nonlinearity in the relationship
Sexe_Child indexes the gender of the child (1 = male, 2 = female). Some authors have emphasized that boys are more likely to be involved in the labor market while girls are more likely to do more housekeeping work (Patrinos and Psacharopoulos, 1995; Psacharopoulos and Arriagada, 1989). A recent study by Canagarajah and Coulombe (1998) in Ghana came out with gender discrimination, with boys being more likely to go to school than girls.
Relationship to the head of the household is a variable that measure the relationship of child with the household. It is indexes (1 = No relationship; 2 = Child; 3 = other relative Head; 4 = Spouse of the head of the household).
Family environment
Household size is the household family size. Generally, large households have more problems to resolve (sickness, etc.), which leave them with insufficient capital to send all the children to school. Also, a large family may have more labor availability. It is hypothesized that household size is positively related to work and negatively to school.
Poor household is a variable that measure the standard of living of the household. It is indexes (1 = if the household is poor and 2 = if the household is non poor).
Charateristics of the Head of household
Sexe_HeadH indexes the gender of the head of the household (1 = male, 2 = female).
Head education indexes the level of education of the head of the household (1 = no formal education; 2 = primary school; 3 = secondary 1; 4 = secondary 2; 5 = Higher education ).
Head Activity is indexes (1= if the head of the household works in the public sector; 2 = if the head of the household works in the private formal sector; 3 = if the head of the household works in the informal agriculture sector; 4 = if the head of the household works in the informal non-agriculture sector; 5 = if the head of the household is unemployed and 6 = if the head of the household is inactive).
Place of residence indexes (1 = rural; 2 = urban).
Region is indexes (1 = Douala ; 2 = Yaounde ; 3 = Adamaoua ; 4 = Center ; 5 = East ; 6 = Far North ; 7 = Littoral ; 8 = North ; 9 = North West ; 10 = West ; 11 = South ; 12 = South West).
Characteristics of the education system
Sub-system is indexes (1 = francophone, 2 = Anglophone).
Type of school is indexes (1 = public; 2 = private secular; 3 = private confessional; 4 = others).
5. Econometrics Results
The estimated coefficients of working hours are negative and statistically significant. In other words, the findings support the hypothesis that working hours have a negative effect on schooling, and this, in the first hour of work. Thus, even a limited number of hours hurting the education of a child. The explanation is simple, whatever the context, work captures the energy of students which puts them in an inferior position compared to those who do not participate in the work force.

However, the positive coefficient estimated for the square for hours of work indicates that the negative marginal effect of work hours on education weakens gradually as working hours increase. Unlike the results obtained by Rajan and Lancaster (2005), the estimation results do not show a U-shaped curve representing the schooling of children according to their hours of work. Because, the negative and significant coefficient estimated for the cube working hours shows that above a certain value of the number of hours worked, the effect changes sign again.

From the results, it appears that the relationship between school performance of the child and the number of hours of work is governed by a function of the form:

where is the school performance of the child and the number of hours of work.
Solving the equation gives the following results : and .

This result shows that the first hours (from one hour to 12 hours) have a negative effect on school performance. Which approximates the results obtained by Rajan and Lancaster (2005), Phoumin and Fukui (2006) and Guarcello et al. (2007), although the specification adopted in this work is of the form .
This negative influence can be explained by the integration of the child and/or their parents work in the use of child’s time, and the fact that this case concerns the youngest children (under 10 years ). From 13 hours per week and up to 34 hours, child labor has a positive effect on school performance in Cameroon. This slice of time is equal to older children who do a good matching between their work and schooling.

We find supportive evidence, for Basu and Van’s (1998) luxury axiom that poverty drives child labour. While poverty reduces the probability of child school performance, it significantly increases the intensity of child labour. Our result shows that poverty appear to be the main culprit of the prevalence of child labour in Cameroon.

The educational level of the head of the household significantly improves child school performance and decrease the likelihood of child labour and intensity of work. Then, children from households headed by a person with at least primary education are more (less) likely to attend school (work). Phoumin and Fukui (2006) also find inverse association between child work participation and head education. This finding reinforces the widely accepted notion that parental education is the most consistent determinant of child education and employment decisions.

Table 6: Determinants of school performance of children aged 5-17 years in Cameroon
Variables B t-stat
Child Characteristics
Work Hours -0.043*** -3.219
Work Hours_squared 0.002*** 3.02
Work Hours_cube -0.00003** -2.443
Age 0.062*** 7.857
Age2/100 -0.151*** -4.518
Sexe_Child male 0.004 0.779
Relationship to head of household
No relationship -0.04 -1.006
Child -0.025 -0.712
Another relative head -0.048 -1.378
Family environment
Household Size -0.018*** -3.838
Poor household -0.026*** -4.582
Characteristics of the Head of Household
Sexe_HeadH female 0.021*** 3.405
Head Education
Primary 0.029*** 3.497
Secondary 1 0.060*** 5.496
Secondary 2 0.090*** 7.288
Higher Education 0.171*** 7.984
Head Activity
Public -0.028 -0.923
Private formal -0.004 -0.126
Informal agriculture -0.064** -2.198
Non-agricultural informal -0.029 -0.989
Unemployed -0.154*** -3.363
Inactive -0.038 -1.011
Place of residence_rural -0.021*** -3.224
Region
Douala 0.333*** 4.205
Yaounde 0.270*** 4.745
Adamaoua 0.149*** 2.677
Center 0.189** 2.513
East 0.213*** 2.666
Littoral 0.219*** 2.778
North 0.068*** 3.421
North West 0.184*** 3.714
West 0.208*** 2.775
South 0.264*** 3.605
South West 0.203*** 2.798
Characteristics of the education system
Sub-system_ francophone -0.046*** -3.162
Type of school
Public 0.028 0.744
Privé Secular 0.070* 1.764
Privé confessionnal 0.065* 1.68
Constant 0.306** 2.417
Source : ECAM 3. Number of observations = 3884 ; Log Likelihood = 1088,926 ; Prob>Chi2 = 0,000 ; Sigma = 0,151 (0,0024). Note : modalities female, spouse of head of household, non-poor, man, out of school, retired, urban, Far North, anglophone and others are remaining for the sex of the child, relationship to head of household, living standards, the sex of head of household, education level of head of household, socio-economic group of head of household, place of residence, region, sub-system and the type of school. *** (**) [*] represent significance levels at 1% (5%) [10%] respectively.

6. Conclusion
This study aimed to analyze the impact of work hours on children’s academic performance in Cameroon. To do so, it was postulated that the working hours of children’s influence on child labor but it is not linear. The children involved in this analysis were those who work and attend together.
Three key events have marked this study:
(a) The first item consisted of the construction of a variable that capture the academic performance of children, taking into accounts the particularities of the educational system in Cameroon. This is the phenomenon of skipping classes, the heterogeneity in age of first entry and passage through the maternal or not of the child.
(b) The second point emphasized methodological issues and the functional form of school performance. A bivariate probit model was used to select only those children who work and attend school. The endogeneity of hours of child labor has been corrected using the method of the double least squares (2LS). The non linearity of the school performance of children was taken into account through the square and cube of work hours of children. School performance of children was estimated using a Tobit model with double truncation.
(c) The main findings of econometric analysis that the first hours of child labor have a negative effect on school performance. Child labor is beginning to have a positive effect on school performance from 13 hours per week, but this effect becomes negative up to the 34th hours of work.
7. References
Adjiwanou V. (2005), « Impact de la pauvreté sur la scolarisation et le travail des enfants de 6-14 ans au Togo », Centre d’Etudes et de Recherches sur le Développement International (CERDI).
Akabayashi, H. and Psacharopoulos, G. (2002), The Trade-off between Child Labour and Human Capital Formation: A Tanzanian case study, Journal of Deve¬lopment Studies, vol.35, n° 5, pp. 120-140.
Amin, A. (1993), Le travail des enfants au Cameroun : analyse d’une situation, IPEC/ILO, Bureau de l’OIT pour l’Afrique centrale, Yaoundé.
Anker, R. (2000), L’économie du travail des enfants : un cadre de mesure, Revue Internationale du Travail, Vol.139, n° 3, pp. 289-317.
Baland, J.M. and Robinson, J. A.(2000), Is Child Labour Inefficient, Journal of Political Economy, Vol. 108, n° 4.
Basu, K. and Van, PH.(1998), The economic of child labour, American Economic Review, Vol. 88, n° 3, pp. 412-427.
Becker, G. S. and Tomes N. (1986), Human Capital and the Rise and Fall of Families, Journal of Labor Economics, vol. 4, n° 3.
Becker, G.S. (1993), Human capital: A theoretical and empirical analysis with difference to education, 3ème edition, The Chicago University press, Chicago.
Bhukuth A. (2004), Le travail des enfants : une revue de littérature économique récente, Ethique et économique.
BIT (2004), Le travail des enfants : un manuel à l’usage des étudiants, Genève.
Bonnet, M. (1993), Child labour in Africa, International Labour Review, vol.132, n° 3, pp. 371-389.
Boyden, J., Ling ,B. and MYERS, W. (1998), What Works for Working Chil¬dren, UNICEF, Centre International pour le Développement de l’Enfant (CIDE) , Florence et Stockholm.
Canagarajah, S. and Coulombe, H. (1997), Child labour and schooling in Ghana, Washington, Policy research working paper n° 1844, Banque mondiale.
Canagarajah, S. and Nielsen, H. S. (1999), Child Labor and Schooling in Africa: A Comparative Study, Social Protection Papers Series, n° 9916. The World Bank.
Cigno, A. and Rosati, F. C. (2001), Child labor handbook. Washington D.C., Banque Mondiale.
Cockburn, J. (2000), Child labour in rural Ethiopia: Income constraints or income opportunities, Working Paper, Oxford, April 9 – 10.
Diallo, Y. (2000), Les déterminants du travail des enfants en Côte d’Ivoire, Centre d’Economie du Développement(CED), Université Montesquieu -Bordeaux IV, France.
Diallo, Y. and Koné, K. (2001), Pauvreté des ménages et phénomène du travail des enfants en Côte d’ivoire, Centre d’Economie du Développement (CED), Université Montesquieu -Bordeaux IV, France.
Dumas C. and Lambert S. (2006), Trajectoires de scolarisation et de travail des enfants, Organisation Internationale du Travail, Genève, Suisse.
Grootaert, C. (1998), Child labor in Côte d’ivoire: incidence and determinants, Washington, Policy research working paper n° 1905, Banque mondiale.
Heady, C. (2000), What is the Effect of Child Labour on Learning Achievement? Evidence from GHANA, Innocenti Working Papers n° 79 (Florence, UNICEF Inno¬centi Research Centre), Octobre.
Jacoby, H., Skoufias, E. (1997), Risk, financial markets and human capital in a developing country, Review of Economic Studies 64, 311-335.
Lachaud, J.-P. (2005), Le travail des enfants et la pauvreté en Afrique : un réexa¬men appliqué au Burkina Faso, Document de travail, CED, Université Montesquieu Bordeaux, France.
Maitra, P., Ray, R. (2000), The joint estimation of child participation in schooling and employment: Comparative evidence from three continents, Journal of Economic Literature.
MICS (2006), Rapport principal de l’enquête par grappe à indicateurs multiples (MICS 3), Institut National de la Statistique du Cameroun.
Nielsen, H. S. (1998), Child labor and school attendance : two joint decisions, Denmark, working paper 98-15, Centre for labor market and social research, octobre.
Noumba I. (2006), Les Déterminants de la déperdition scolaire dans l’enseignement secondaire au Cameroun, African Economic Research Consortium (AERC).
Noumba, I. (2002a), Une évaluation de la performance de l’enseignement secondaire au Cameroun, Rapport final de recherche soumis à AERC, octobre.
OCDE (2003), Combattre le travail des enfants. Un bilan des politiques, Paris.
Rodgers, G. and Standing, G. (1981), Child work, poverty and underdevelopment, Geneva, ILO.

Skills

Posted on

March 8, 2018

Submit a Comment

Your email address will not be published.