UNIVERSITY OF ECONOMICS UNIVERSITY OF ECONOMICS ERASMUS UNVERSITY ROTTERDAM HO CHI MINH CITY INSTITUTE OF SOCIAL STUDIES VIETNAM THE NETHERLANDS VIETNAM – THE NETHERLANDS PROGRAMME FOR M A IN DEVELOPM[.]
UNIVERSITY OF ECONOMICS HO CHI MINH CITY VIETNAM ERASMUS UNVERSITY ROTTERDAM INSTITUTE OF SOCIAL STUDIES THE NETHERLANDS VIETNAM – THE NETHERLANDS PROGRAMME FOR M.A IN DEVELOPMENT ECONOMICS THE DIRECT AND INDIRECT IMPACT OF CHILD LABOR ON EDUCATIONAL ACHIEVEMENT: EVIDENCE FROM VIETNAM BY NGUYEN TAN PHUC MASTER OF ARTS IN DEVELOPMENT ECONOMICS HO CHI MINH CITY, November 2017 UNIVERSITY OF ECONOMICS HO CHI MINH CITY VIETNAM INSTITUTE OF SOCIAL STUDIES THE HAGUE THE NETHERLANDS VIETNAM - NETHERLANDS PROGRAMME FOR M.A IN DEVELOPMENT ECONOMICS THE DIRECT AND INDIRECT IMPACT OF CHILD LABOR ON EDUCATIONAL ACHIEVEMENT: EVIDENCE FROM VIETNAM A thesis submitted in partial fulfilment of the requirements for the degree of MASTER OF ARTS IN DEVELOPMENT ECONOMICS By NGUYEN TAN PHUC Academic Supervisor: Assoc Prof Nguyen Huu Dung Ph.D HO CHI MINH CITY, November 2017 Contents CHAPTER I: INTRODUCTION Problem statement Research objectives and methods Structure of thesis CHAPTER II: LITERATURE REVIEW Theoretical framework Review of empirical studies CHAPTER III: RESEARCH METHODOLOGY 11 Empirical models 11 The data: Young Lives Round – 2013 15 Data description 16 CHAPTER IV: RESEARCH RESULTS 19 Overview of child labor and education in Vietnam 19 Summarize the data 21 The estimation results for Whole sample 24 Further estimation 26 Results of two – stage least squares regression 29 CHAPTER V: CONCLUSION 32 REFERENCE 35 APPENDICES 38 Abstract This study explores the direct and indirect impact of child work on educational achievement of children at the age from 11 to 20 across rural and urban areas in Vietnam, using the data of Young Lives Round complemented in 2013 Given the characteristics of individual, household and schooling which are controlled in estimation, the results indicate that there is negative relationship between hours worked and math scores of children, but the impact in the rural areas is different from that in the urban In urban, exhaustion while working or doing other activities besides learning is responsible for weak performance in schools Meanwhile, school dropouts and delays because of working is the main reason of low educational outcomes of children in rural Further, schooling attributes contribute to the increase in math scores of children, especially those in rural, raising the necessities of improving qualities of education in those regions This study also uses a set of factors including income earned from crops, household shocks, and community – level rice price as instruments of hours worked variable But after Hausman examination, the Ordinal Least Square (OLS) results are preferred due to weak instruments JEL Classification: I21, J13, J22, O15 Key words: child work, educational achievement, Vietnam CHAPTER I: INTRODUCTION Problem statement Child labor is described as the engagement of children in various activities (paid or unpaid) that keep them from their childhood According to Global Child Labor Trends, there are approximately 10.6 percent of children at the age of – 17 in the world participate in workforce Following the first National Child Labor Survey in 2014, this number in Vietnam is about 9.6 percent Most of them living in the countryside, being involve in agricultural activities or their family businesses over 42 hours per week Consequently, around 96.2 percent of them were not going to school Many argue that working of children would bring many disadvantages for the development of them following reasons One disadvantage concerns the increase in the risks of issues which could harm seriously physical and mental health of children when they participate in workforce It is obviously to see that jobs employing children as workforce normally are described as unskilled types of work, together with poor quality in working conditions The occupational injuries potentially appear from operations with dangerous equipment, heavy loads or poison exposure Moreover, throughout history of the world, child labor often relates to illegal activities such as slavery, drug trade, child prostitution and human trafficking These kinds of abuses cause both physically and mentally traumas for the whole life of children involved According to UNICEF, Sub – Saharan Africa has the highest percentage of young workers around the world At the group of – 14 ages, child labor in this country accounts for 28 percent, while countries such as Middle East and North Africa and East Asia and the Pacific are about 10 percent Following International Labor Organization, using the sample of 26 countries, there is a quarter of children who suffer injuries while working In United States, industries using children as employment force accounts for the higher injury rates than the average level every year In term of Vietnam, according to Hanoi school of Public Health report, about 23000 children were injured by sharp objectives, and half of these children got damaged when they were working Further, more than 60 percent of injuries caused by machines in young workers, mainly at the age of older 14 There was investigation in 2001 conducted by the Ho Chi Minh City Department of Labor, Invalids, and Social Affairs, found that child labor problem still remains at least of 24 districts throughout Vietnam, especially in rural areas More than 90 percent of companies that employ children as workforce without legal license, and children have to work under hazardous conditions, particularly in environments of gold mines, timber operations or cargo transport In addition, the survey implemented by Statistical Information and Monitoring Programme which disentangle the state of child labor in four countries Cambodia, Laos, Mongolia, and Vietnam reports that 43 percent of children aged – 14 and 51 percent of children aged 15 – 17 are suffering dangerous conditions at their work In 2015, The Guardian documented that about 3000 children in Vietnam were trafficked to the UK for working and debt paying purpose, with regard to illegal businesses such as brothels or cannabis farms The second drawback of child employment is to prevent children from their childhood that they should have The fact is, children will miss the opportunity to attend school and to acquire academic education They also have no time playing outside with those at the same age In turn these factors lead to the statement that children are limited for developing fully their understandings, awareness and knowledge Education plays a vital role for the long – term growth of people, especially children It is believed that, living in a competitive world, candidates with a high education and knowledge will have more choices in labor market than others The reason for this could be explained that children acquire from basic to advanced level of accumulated understandings of a field through learning From that, they will be able to have a general perspective for any situations, circumstances, and abilities Knowledge also helps them avoid faults and build on achievements from the past Moreover, spending more time in school improves their soft skills as well, including responsibility, time management, disciplines, organizational and social skills These strengths contribute significantly to their success and quality of life in future In term of country level, education is described as a main tool for sustainability, economic development and social welfare One of the prior goals of both developed and developing nations in the world is to improve the total capacity and quality of their human resources from their own country or from others Every year, they invest large amount of money in human capital as well as enhance the awareness of residents about benefits of acquiring knowledge Educational achievement is considered as the instructional goals or learning objectives of education It is described as outcomes that an individual obtains from learning activities in academic institutions, such as in schools, colleges, universities, etc Students reveal educational performance by their ability of knowledge, understanding and skills acquisitions in a particular field (numeracy, literacy, science, art, computing, etc.) through distinct measurements like scores or grades on tests, level of academic degrees, and number of educational certificates Educational achievement is normally employed to measure ability of one person It shows the outcome that an individual obtains from his or her engagement of education, focusing on what he or she could actually do, rather than the level of participation in education Educational achievement also helps to setting standardized assessments for distinguishing capacities of students in schools, contributing to household decisions whether a child should continue education or not, as well as enhancing the motivation of education engagement of children Research objectives and methods This study mainly aims to investigate the direct and indirect effect of child labor on educational achievement which measured by the cognitive achievement such as mathematics performance for children in Vietnam The exploration from this research could solve some following problems At first, educational achievement (or educational performance) is seen as one of important factor which highly influence long – term development as well as enhance the quality of life of an individual, household, society or even a country Therefore, it is necessary and reasonable for exploring the effect of a historical and prevalent problem like employment of children at work, on their educational achievement Second, previous studies primarily concentrate on school attendance or school enrollment, and use this measurement as the indicators of learning achievement However, this approach is unable to estimate truly the harmful which is caused by child labor For example, working could harm potential achievement or gains acquired from education even attaining school or not According to Christopher (2000), it could over – estimate the effect of working at the early age in case children enroll in poor education or school, but they can improve their knowledge through their job Besides, it also would under – estimate the child labor due to the scenario in which student in spite of going to school but having no time learning or completing homework after working Consequently, it is obviously necessary to employ other indicators for educational achievement than simply using school enrollment rates Unlike previous studies, this study employs new measurement of learning achievement, the mathematics performance of children Third, by employing rich set of control variable, including individual characteristics, household characteristics, and schooling characteristics, the results suggest which factor of an individual could potentially affect educational achievement, together with child work For example, it is considered that older children are likely to outperform younger children in term of both working and learning achievement due to their higher physical health As a result, the intensity of work will not affect performance of older children as much in comparison to younger ones Finally, it will show the general picture as well as support the factual evidence about consequences of using children as labor force and its effect (both direct and indirect) on human growth and learning, of one typical developing country like Vietnam These results will help policy makers in taking child labor into consideration, then building appropriate and effective policies, also contribute to academic field of employment of children at work as well as learning achievement for later related studies Following this research objective, the chosen empirical method in this study is based on the model set by Heady (2000), using the Young Lives Round data in Vietnam which covers 839 children at the age from 11 to 20 In which, the total effect is measured when regressing educational achievement on child labor while excluding schooling attributes out of estimated equations, given the control of individual characteristics and household characteristics of children The direct effect, however, is exposed in the same analysis but keep the schooling attributes constant The estimations in this study are Ordinal Least Square (OLS) and Two – stage Least Square with a set of instrumental variables for robustness check In my expectation, weak performance in education is driven by the incidence of child work, and this negative impact is different across urban and rural areas Structure of thesis This study is divided into five main sectors followed: The literature review related to the research problems and methodologies will be shown in section II Section III describes the theory, chosen empirical model, data sample and data requirements used in this study, plus the suggestion of potential problems and solutions Section IV reports the results of regressions as well as tests Finally, section V conclude some remarks, including main findings, policy implications and limitations appeared in this study CHAPTER II: LITERATURE REVIEW This chapter provide the theoretical framework about the decision in whether children take part in labor force or continue their education is made within household as well as the correlation between child work and educational achievement of children After that, some related studies and researches are discussed for the purpose of further understandings about this academic field Theoretical framework Following the definition of ILO, child labor is “work that deprives children of their childhood, their potential and their dignity, and that is harmful to physical and mental development” (ILO 2004, p.16) Heady (2000) described child labor as paid or unpaid activities which provided on the labor market as well as on household farms or companies This approach of description of child work excludes the domestic works in the households (including taking care of ill members or younger children, cleaning, washing, cooking, etc) He employed a group of questions to ask children about their economic activities for measuring the intensity of child work, such as whether a child had worked in the past twelve months, how many weeks and how many hours per week they had worked in the past twelve months Mavrokonstantis (2011) pointed out that child labor should be considered as the economic works, including paid activities outside the household and unpaid activities inside the household The educational achievement of a child is the indicator of the school output of one individual, which is derived from an educational production function, given the student inputs An educational production function is determined as follows: E = f (X1, X2, X3) (1) Where E represents the school output of a student, in other says, the educational achievement of a student X1 includes factors around the school environment, such as the teaching methods and materials, the school infrastructures, the length of time that student use for schooling X2 comprises the environmental influences on education outside the school, like educational backgrounds of parents, or motivation for education of a student X3 represents factors which measure the initial level oriented towards learning of student The intensity of child labor, which is measured by the time allocated on economic works, theoretically affect educational achievement following numerous ways On the one hand, child labor will reduce educational outcome of a child due to the fact that time allocation is scarce resource, thus the increase in hours worked will lead to the decrease in input factors of educational production function, such as hours spent on attending schools or extra classes, and hours spent on accomplishing homework Additionally, working could cause exhaustion, lack of energy and ability to learn academic knowledge On the other hand, child labor is considered to enhance the educational performance of children by allowing them to apply academic knowledge they have learnt at school in real life More than this, working could provide children not only specific experiences related to jobs, but also a number of soft skills such as time management, responsibility, communication, confidence, problem solving, etc In turn, these elements will help to increase the educational outcome of children in their schools Alternatively, child labor could have no effect on educational achievement if the incidence of child work is substantial low, or if children know how to arrange efficiently their time between schooling and working Educational achievement (or academic performance) is the short or long – term educational goals that one individual has obtained from their engagement of education Obviously, the level of educational performance of a child could highly determine his or her future income as well as living conditions, rather than his or her years of schooling Educational achievement is normally measured through scores from examinations or tests of cognitive skills, including verbal skills and mathematics skills For example, Heady (2000) employed results from an easy reading test, an easy mathematics test, an advanced reading test, and an advanced mathematics test in his survey as the measurement of educational achievement of children in Ghana Gunnarsson et al (2006), similarly, used the mathematics and language test scores of children on third and fourth year primary schools in nine Latin American countries Bezerra et al (2009) employed the school achievement tests in Portuguese and mathematics for students in Brazil Review of empirical studies 15 for agricultural households, the rise in revenue earned from crops this year requires the expansion of scales of production, leading to the increase in demand for child work Shocks affect household wealth, thus expectedly raise the possibility of labor force participation of children in family According to Mavrokonstantis (2011) and Beegle (2009), for children in urban, rice price and child work have a positive correlation following income effect But for those in rural, rice price expectedly performs both negative impact following income effect and positive impact due to the increase in cultivation of rice In practices, the first stage in two – stage least squares estimation is applied as follows: Wi = αi + Li + Ii + Fi + Si + ei (5) where Li represents instrumental variables, then the second stage is estimated after predicting the hours worked of children Ai = αi + Ii + β i : + Fi + Si + ei (6) Other problem exists in this regression is whether the residuals derived from estimation models are normally distributed and homoscedasticity For these issues, White test and ShapiroWilk test are employed to check the null hypothesis of normal distribution and homoscedastic assumption of residuals obtained from estimation models respectively The data: Young Lives Round – 2013 Young Lives is known as an international research project on childhood and their changing lives over specific period of time Using interview, group work and case studies with subjects of children, their family, their school and their community, researchers try to collect background information about lives, physical and mental health, and future prospects of children following different contexts These subjects have a long – term commitment when they agree to take part in this project This longitudinal database gathers information about 12000 children in four developing countries, namely Ethiopia, India, Peru and Vietnam during 15 years In each country, sample is divided into two age cohorts: 1000 children who were born in 1994 – 1995 and 2000 children who were born 2001 – 2002 The main goals of this project is to find out commonalities and differences in lives of children in four typical and different developing countries, then to build patterns and understandings about poverty transfer and poverty reduction policies 16 Data used in this study is from Young Lives Round implemented in 2013 for Vietnam The Round was conducted in 2002, followed by Round in 2006, and Round in 2009 The chosen sample is order cohort children who have age from 18 to 20 years old and younger cohort children at the age from 11 to 14 Each cohort includes two sub-samples First, the household data which covers household education, livelihoods and asset framework, household food and non – food consumption and expenditure, social capital, economic changes and recent life history, and socio – economic status Second, the child data which shows education, employment, earnings, and time – use, feelings and attitudes, anthropometry, health and nutrition of children in each cohort Data description Educational achievement: a dependent variable of study This factor is reflected by the score of cognitive development test: the mathematics test for each child in the sample Mathematic achievement test scores: children have to answer 27 exercises for older cohort and 34 exercises for younger cohort which containing additions, subtractions, divisions, multiplications, and problems related to math One point will be recorded for each correct answer The refused – to – answer and blank answers will be seen as incorrect ones As a result, the Math scores variable, pointing out mathematics performance of children, is counted by the proportion of correct answers Working state: is the main explanatory variable in this study This factor indicates the child work, and is measured by a number of hours that children spent their time on working on a typical day last week (typical weekday, not weekends or holidays) The definition of work, even children is paid or not, covers not only activities inside household such as tasks on family farm, cattle herding, other family business, shepherding, piecework or handicrafts done at home but also different activities outside their household If children both kinds of activity, the Hours of economic work variable are the sum of hours they spent on both locations Following my hypothesis, the correlation between working and education performance would be expectedly negative The number of working hours is drawn from Child employment, earnings and time – use sector in Young Lives Schooling characteristics: This factor attempts to catch up the education background and attitude of children about their school, such as school years, attitude to education of household, 17 education background of parents, and motivation to learning of children These variables are expected to be positively related to educational achievement of children School years: is the number of years of education that children have completed, denoted by Education years of child variable This data is obtained from household and child education sector Attitude to education of family: is the amount paid for educational expenditure in household, including school uniforms, schooling fees (registration and examination), donations to school, extra tuition, school books and stationary, and transport to school The sum of these expenses is denoted by Schooling costs variable Education background of parents: the parental education years of parent (father or mother) of children, represented by Parental education years variable This data is also contained in household and child education sector Educational motivation: there are six questions about feelings as well as motivation of children about their school These questions are built based on Likert scale from – strongly disagree to – strongly agree, including (1) being proud of clothes, (2) having the right books, pencils and other equipment for school, (3) being proud of shoes or of having shoes, (4) having correct uniform, (5) making plans for future studies and work, and (6) will be rewarded by a better job in future if study hard The Motivation about school variable is the average of recorded answer of six questions, which are extracted from feelings and attitudes section Individual characteristics: includes background information of children (age, gender, body mass index, number of siblings), innate ability of children (PPVT score), ant number of hours spent in housework Age: age of children, with the expectation that older children outperform younger ones Gender: is dummy variable whether children is female According to the definition of working, boys will be expected to outperform girls BMI: the body mass index of children, calculated by using weight in kilograms (kg) divided by height in meters squared This index defines a healthy body weight, including underweight (= 30) Number of siblings: is the number of siblings in family of children 18 Innate ability: this factor is measured by using the score from The Peabody Picture Vocabulary Test (PPVT) PPVT is described as a measurement for receptive vocabulary ability of children which is not affected by working or by education In this test, based on group of age, children are shown a series of four pictures that are numbered After hearing the “one – word” description of one picture, children have to say or to point to a number of described picture as their answer Expectedly, children with high innate ability will perform well in their learning achievement Time spent in housework: regarding the number of hours used in domestic activities within household like caring for younger or ill members, cleaning, cooking, washing, etc The Hours of domestic tasks variable is expected to harm the learning performance of children due to lacking of time for them to complete their homework Household characteristics: control for differences in background information of household lives, comprising region, area, household size, food consumption per capita, housing quality index, and access to services index Region: there are four regions in this sample, including Northwest, Red River Delta, South Central Coast, and Mekong River Delta Area: a dummy variable indicating household of children stays at urban It is expected that children in urban perform better than those in rural due to quality of education, convenient transportation, good standard of living, etc Household size: the amount of members in family Food consumption per capita: measures household welfare following Moratti and Natali (2012) The household spending is the total value in VND of expenditure of food consumed by family in the last 15 days, such as beans/ rice/ bread/ cereals, meat products, milk or milk products, fish and sea products, eggs, vegetables, fruit, spices, drinks, etc The food consumption per capita variable is obtained by dividing total spending by the amount of members in family Housing quality index: is defined as the average of rooms per person, floor, roof and wall Access to service index: is defined as the average of a set of dummies which indicating households having drinking water, electricity, toilet and fuel 19 CHAPTER IV: RESEARCH RESULTS This chapter discuss the general picture of child labor and education in Vietnam, then report all the results from empirical estimations: OLS and Two – stage Least Square, followed by the discussion of main findings Overview of child labor and education in Vietnam In Vietnam, education is a system of public and private education which is administrated by the Ministry of Education and Training This ministry is responsible for designing a long – term plan for education, following the requirements of labor market The formal education is twelve years begins at age and is divided into three levels: primary school (five years), intermediate school (four years), high school (three years) There are also pre-schools, vocational education and higher education (university, college, or institute) The primary education is compulsory, and students will learn typical subjects such as morals, Vietnamese language, math, nature and society, arts and physical activities in school at this level of education According to the Resolution of the 4th Plenum of the Central Committee of the 7th Party Conference (1993), the main target of education in Vietnam is "improving people’s general knowledge, training quality human resources, and nurturing and fostering talent." Based on the strategy of education reform, Vietnamese Government has continuously raised the pubic budget on education every year According to UNESCO, the share of GDP spent on education increased from 4.81 percent in 2011 to 5.66 percent in 2013 In the international Pisa test organized in 2012, Vietnamese students obtain impressive successes They achieve higher scores in reading, maths and science tests than other developed countries, such as United Kingdom and United States Following global ranking published by the OECD in 2015, the rank of Vietnam was 12th, compared to the United States of 28th However, following the report of World bank, although there is a remarkably increase in school enrollment rates in recent years, the quality and effectiveness of education, which are represented by educational achievement of students, continue to be low, especially of poor regions and provinces The quality in Vietnam education is still measured below the international standards because of poor teaching materials and methods, lacks of discussion and interaction between teachers and students, or interferences of the Vietnamese Ministry of Education and 20 Training As the results, many graduated students, who achieve high scores in their schools and universities, find themselves difficult to get a well – paid job, and needed to be retrained since they start working Meanwhile, the drop – out and repetition rates are also reported as high level, leading to the fact in which child labor still exists Although Vietnam Government attempts to reduce the incidence of child work by releasing many laws in which the employment of children under the age of 15 is prohibited, but they are not aggressively applied and enforced, thus the protection for children is still weak There was the survey implemented in 2012 by the General Statistics Office (GSO) of Vietnam, they report that child labor accounts for one – sixth (approximately 2.83 million children) of the whole child population, with about 42.6 percent of them are girls They engaged mainly in unskilled jobs including agriculture, construction, garments and restaurant services Among working children, there were about 32.4 percent of children which worked over 42 hours per week, and experienced decrease in time spent on schooling As a result, 96.2 percent of them drop their school because of work Child labor remains to be a serious problem in Vietnam due to some factual concerns which are discussed in turn Firstly, many families who have their traditional job across generations think that education is time – wasting and irrelevant They argue that their children only need to learn specific field enough in order to take over his or her family business, not spend almost time on enrolling school and get useless things As a result, they send their children to labor force after withdrawing them from their school In fact, families with farming tradition, children have a trend to take over and maintain the agricultural assets from their relatives Secondly, some underestimate the negative impact of working at the early ages on education as well as educational achievement On the one hand, they claim that for a particular job market, children who work and improve the required skills will have more competitive advantages compared to those without working background One the other hand, in case of households which are in poor conditions, children should take part in employment for sharing workload and supporting their families As a consequence, they have to spend their daily time on both learning and working, or drop out of school in a specific time for working This situation is popular and conventional in rural areas in which children typically work on farms or agricultural lands, therefore, they were not permitted to attend school by their parents during the seasons of harvest and planting Lastly, poor families found themselves unable to pay for school charges and fees, 21 including registration, uniform, books and stationery, or extra tuition In case of private schools, these costs are generally much higher and beyond the affordability of their conditions In addition to this, they are inaccessible to any support or aids due to the failures of policies and campaigns of education reforms of Vietnamese government Consequently, many of children have no other option than dropping out of schools Summarize the data This study covers data for 839 children at the age from 11 to 20 which are separated in two sub samples, including 449 (53.52 percent) that stay in rural areas and 390 (46.48 percent) that lives in urban areas Table shows the descriptive statistic for all variables of two samples For individual characteristics, it is surprisingly that children in rural sample achieve the higher average ppvt scores than those in urban sample (111.47 compared to 85.41) The average body mass index (BMI) of children in both areas are within the normal range of healthy (from 18.5 to 24.9) The gender distribution is balanced adequately, especially the percentage of female is about 50 in rural sample and 50.8 in urban sample Children in rural households spend more time on doing domestic activities than children in urban households, the average number of hours per typical day for these kinds of tasks for two sample are approximately 1.39 and 0.82 respectively Regarding household section, the comparison between two samples is fairly complicated Urban households predictably spend more money for food consumption than rural households It is also no doubt that households in urban area have more opportunities to access services rather than those in rural area However, the average housing quality index of rural households is surprisingly higher than of urban households There is no difference between the number of members per household in each area (from to people in a family) About region characteristics, the full sample covers children from four regions: Mekong River Delta (10.3 percent), Northwest (6.2 percent), Red River Delta (29.6 percent), and South Central Coast (54 percent) In term of schooling attributes, the level of parental education in rural households is lower than in urban households due to difficulties in living conditions in the past, specifically the average education years of parents in both areas are 8.3 and 9.7 respectively The scenario is contrastive for the level of education of children, the average education years of children are 8.9 22 in rural and 7.5 in urban However, it is obvious that households in urban spend more funds on schooling than those in rural For the educational achievement, children in urban areas perform predictably better than those in rural areas, but not much Specifically, the average percentages of correct answers of math test which children in rural and urban households achieved are about 45.5 and 50 percent respectively For the level of working, it is worth noting that children residing in rural areas spend approximately 4.6 hours per typical day on engaging in economic work, compared to about 2.3 hours of those staying in urban households on these kinds of activities Further, activities outside household account for a high proportion in economic work in both areas 23 Table 1: Descriptive statistics of the sample Full sample (839 obs) Variable Mean Std Dev Rural (449 obs) Urban (390 obs) Mean Std Dev Mean Std Dev Individual characteristics PPVT score 99.356 46.840 111.472 45.779 85.408 44.140 Age (months) 184.670 41.555 197.071 40.418 170.392 38.166 0.503 0.500 0.499 0.501 0.508 0.501 BMI 18.760 3.069 Domestic tasks (hours per 1.125 1.042 typical day) Number of siblings in 1.327 0.965 household Child labor measure (hours per typical day) 18.452 2.715 19.113 3.400 1.390 1.054 0.820 0.941 1.232 0.894 1.436 1.032 Activities inside household 0.816 2.063 1.069 2.316 0.524 1.683 Activities outside household 2.705 3.984 3.523 4.185 1.765 3.517 Economic work (total) 3.521 4.269 4.592 4.318 2.289 3.867 Ln(consumption/capita) 6.895 0.878 6.457 0.747 7.400 0.737 Housing quality index 0.629 0.118 0.650 0.138 0.605 0.083 Access to services index 0.721 0.245 0.553 0.181 0.915 0.148 Household size 5.253 2.123 4.675 1.609 5.918 2.429 Area (Urban = 1) 0.465 0.499 - - - - Region: Mekong River Delta 0.103 0.303 0.185 0.389 0.008 0.087 Region: Northwest 0.062 0.241 0.109 0.312 0.008 0.087 Region: Red River Delta 0.296 0.457 0.530 0.500 0.026 0.158 Region: South Central Coast 0.540 0.499 0.176 0.381 0.959 0.199 Education years of child 8.308 2.914 8.967 2.860 7.549 2.792 Parental education years 8.971 3.366 8.334 3.116 9.705 3.496 Motivation about school 3.808 Schooling costs (million 12.950 VND) Educational achievement measure 0.471 3.758 0.463 3.866 0.474 22.599 8.024 18.962 18.620 25.017 Mathematics test scores 47.620 17.492 45.541 17.983 50.013 16.613 11448.560 41899.220 20010.870 54971.310 1590.923 10938.560 0.524 0.500 0.494 0.501 0.559 0.497 11.572 1.296 11.105 1.192 12.110 1.200 Gender (Female = 1) Household characteristics Schooling characteristics Instrumental variables Income from crops (million VND) Household shocks Community - level rice price (thousands VND) 24 The estimation results for Whole sample Table presents discretely the OLS results arrived at three estimation models in different columns for the mathematics performance To summarize, column I indicates the relationship between educational achievement and working state without taking schooling attributes into consideration (the total effect) according to model (1) Column II reports the results which add the schooling characteristics (direct effect) following model (2) Column III describes the results which exclude the working state according to model (3) The estimated coefficients are significant at 10 per cent or less will be marked * Similarly, they will be marked ** if they are significant at per cent or less, and will be marked *** in case they are statistically significant at per cent or less The values in parentheses show the standard errors of each coefficients reported As column I of table shows, there is negative correlation between hours worked and mathematics scores of children following expected sign With significance level of 1%, the significant coefficient of working state variable represents the direct effect of working on mathematics achievement by itself plus the indirect effect via schooling Moreover, the set of characteristics comprising ppvt score, age, body mass index, food consumption per capita, and access to services index together appear to be determinants that affect the mathematics performance of children in this sample It is interesting to note that these factors affect mathematics scores following expected signs, except age variable Specifically, the coefficient of age turns out to be negative, leading to the statement in which younger children surpass older children in the math test Column II indicates the results after adding schooling characteristics into regression model, the score of mathematics test remains to be affected by the intensity of working, this negative impact is lower than the total effect (in absolute value) at column I because in this case, the coefficient of hours worked no longer capture the effect of working on mathematics achievement through schooling The difference implies that the indirect impact of child work on educational achievement exists To explain, with significance level of 10%, when the number of working hours increases by hour, the scores of the mathematics test of children will decrease directly by 0.72 percentage points, other things constant, together with the reduction by 0.28 (1 – 0.72) percentage points in math scores caused by the effect of working on schooling attributes 25 However, the schooling attributes including parental education years and education years of children predictably increase the outcome of math test at the level of 1% Body mass index (BMI) factor is no longer affect the mathematics performance, meanwhile the other significant determinants still statistically relate to math scores of children, following same signs as column I Column III reports the estimation model which drops the working state of children The results confirm the positive impact of schooling attributes, measured by education years of parents and children, on the mathematics achievement, not be biased by the omission of child work from the estimation model Table 2: OLS estimation results for mathematics test scores (whole sample) (I) (II) Variables -1.006*** -0.723*** Hours of economic work (0.26) (0.24) 0.337*** 0.231*** PPVT score (0.04) (0.04) -0.325*** -0.425*** Age (months) (0.05) (0.05) 1.017 -0.382 Gender (Female = 1) (1.09) (1.04) 0.352* 0.313 BMI (0.20) (0.19) 0.044 0.066 Hours of domestic tasks (0.56) (0.53) -0.863 0.645 Region: Northwest (2.85) (2.69) 4.439** 4.409** Region: Red River Delta (2.21) (2.09) -0.161 1.073 Region: South Central Coast (2.47) (2.33) 2.781*** 2.053** Ln(consumption/capita) (0.85) (0.82) -1.737 -1.356 Area (Urban = 1) (2.13) (2.02) 6.757 4.160 Housing quality index (4.83) (4.57) 12.170*** 6.076* Access to services index (3.48) (3.38) -0.488 -0.167 Household size (0.32) (0.31) (III) 0.223*** (0.04) -0.487*** (0.04) -0.508 (1.05) 0.316 (0.19) 0.392 (0.52) 1.474 (2.69) 4.927** (2.10) 1.647 (2.34) 2.190*** (0.82) -1.614 (2.03) 3.722 (4.59) 6.432* (3.39) -0.158 (0.31) 26 Table continued Number of siblings -0.307 (0.66) Education years of child Parental education years Motivation for education Schooling costs (million VND) Constant 41.043*** (10.02) 839 0.215 14.984 117.650 0.712 0.9972 0.157 -0.646 (0.62) 3.221*** (0.37) 0.582*** (0.17) 1.029 (1.13) 0.018 (0.02) 43.038*** (10.66) 839 0.307 19.125 209.250 0.330 0.998 0.375 -0.657 (0.62) 3.338*** (0.36) 0.586*** (0.18) 1.028 (1.14) 0.019 (0.02) 50.044*** (10.44) 839 0.300 19.509 187.770 0.350 0.997 0.127 Observations R2 Joint F test White test p-value Shapiro-Wilk W test p-value Note: Values in parentheses represent standard errors *** Notes significance at the 1% level ** Notes significance at the 5% level Tải FULL (60 trang): https://bit.ly/3R6Pntl * Notes significance at the 10% level Dự phòng: fb.com/TaiHo123doc.net Further estimation Due to the large differences between rural and urban areas in types of work, perception of parents, quality of school which cannot be captured in the model, the further estimation in which mathematics achievement are regressed for rural sample and urban sample respectively is suggested Table presents distinctly the OLS results for math scores following three empirical models in each column: I, II, III for rural sample, and IV, V, VI for urban sample with the same structure The work intensity still reduces the score of math tests of children in both areas after controlling the effect of schooling characteristics, but the direct impact of working in urban is higher than in rural Specifically, as column II and column V point out, the math score decreases directly by 0.55 percentage points for children in rural at the 10% level and by 1.03 percentage points at the 1% level for children in urban since the number of hours that they participate in 27 working increases by hour This result could be explained by the fact that as characteristics of rural areas, children tend to be familiar with working, especially they often engage in simple jobs such as farmer, construction worker, or factory worker Consequently, children in rural suffer less damage induced directly by working than those in urban Further, it is possibly that there are some outside activities children in urban spend on which could not be captured in this study besides working and learning, like entertainment or sports As a result, they are easy to be exhausted after working, thus affect directly to their educational achievement However, it is interesting to note that the coefficients which represent indirect impact of child work on math scores in urban is 0.16 (1.19 – 1.03) and in rural is 0.34 (0.89 – 0.55) The higher indirect impact indicates that work engagement affect schooling attributes (including education years of children and their parents, schooling costs, motivation about school), then educational achievement of children in rural rather than those in urban It is possibly explained that participation in working force obliges children in rural area to drop their school, or lose their motivation of attending school and learning academic knowledge The other characteristics including ppvt score, age, and food consumption per capita still remain statistically significant with mathematics performance of children in both rural and urban areas following same sign as estimation for the whole sample It is exciting to figure out that in the mathematics test, boys outperform girls in rural area while girls achieve higher scores than boys in urban area The results continue confirm the strong positive influence of schooling attributes, such as education years of children, on mathematics achievement in both sub samples, even controlling the effect of child work or not To be more specific, according to the significant level of coefficients shown in column II and column V, the increase in one education year will rise the math scores by 3.42 percentage points of the children in rural and by 2.88 percentage points of those in urban Additionally, it is noted that the magnitude of coefficient in rural sample is unexpectedly larger than in urban sample although children in urban are provided higher education quality (in learning materials, teaching methods, physical infrastructure and facilities, human resources) than children in rural This finding contribute to the evidence that school engagement is highly influential factor which would enhance much the level of educational outcome of children, especially those in rural area Along with the education years of children, the number of years that parents of them have completed also relate to math scores in urban households at the level of 1% following expected sign Similarly, the schooling costs operate 28 positively with the scores of math test of children in rural households at 5% level On the other hand, it is reasonable to observe that the index of accessing to services shows the significantly positive relationship with the math score for only children on urban area, meanwhile the number of siblings in household turns out to be statistically significant for those in this sample because of large education costs in urban area, consequently the budget for education is allocated among Tải FULL (60 trang): https://bit.ly/3R6Pntl children in households Dự phòng: fb.com/TaiHo123doc.net Finally, column III and column VI suggest that the omission of child work does not change the statistically significance of schooling attributes as well as other determinants, indicating that the estimated returns to schooling remain constant without adding working into the equation Table 3: OLS estimation results for mathematics test scores (sub - samples) Rural Variables (I) (II) (III) (IV) -0.886*** -0.554* -1.191*** Hours of economic work (0.34) (0.32) (0.39) 0.320*** 0.210*** 0.204*** 0.358*** PPVT score (0.05) (0.05) (0.05) (0.06) -0.300*** -0.407*** -0.454*** -0.332*** Age (months) (0.07) (0.07) (0.06) (0.07) 3.722** 1.340 1.259 -2.130 Gender (Female = 1) (1.57) (1.52) (1.52) (1.52) 0.234 0.125 0.147 0.348 BMI (0.36) (0.33) (0.33) (0.24) 0.091 0.250 0.542 -0.527 Hours of domestic tasks (0.76) (0.72) (0.70) (0.85) -2.263 -0.539 0.132 10.174 Region: Northwest (3.09) (2.90) (2.88) (12.32) 5.932** 6.022** 6.474*** -12.939 Region: Red River Delta (2.55) (2.41) (2.40) (10.14) -0.045 2.069 2.527 -13.558 Region: South Central Coast (2.74) (2.60) (2.59) (9.51) Urban (V) -1.025*** (0.37) 0.236*** (0.06) -0.397*** (0.08) -2.760* (1.45) 0.320 (0.23) -0.634 (0.82) 13.519 (11.73) -9.717 (9.69) -14.942* (9.04) (VI) 0.232*** (0.06) -0.490*** (0.07) -2.978** (1.46) 0.306 (0.23) -0.274 (0.81) 12.129 (11.82) -12.234 (9.73) -16.238* (9.11) 29 Table continued Ln(consumption/capita) Housing quality index Access to services index Household size Number of siblings 2.329* (1.20) 4.001 (5.79) 7.354 (4.53) -1.083 (0.69) 1.333 (1.13) Education years of child Parental education years Motivation for education Schooling cost (million VND) Constant 45.552*** (13.79) 449 0.209 8.205 87.780 0.956 0.998 0.310 1.950* (1.14) 1.246 (5.44) 0.867 (4.39) -0.622 (0.66) 0.671 (1.07) 3.432*** (0.50) 0.356 (0.26) 1.047 (1.64) 0.086** (0.04) 47.780*** (14.92) 449 0.318 11.139 164.250 0.823 0.998 0.652 Observations R2 Joint F test White test p-value Shapiro-Wilk W test p-value Note: Values in parentheses represent standard errors *** Notes significance at the 1% level ** Notes significance at the 5% level * Notes significance at the 10% level 2.014* (1.14) 0.751 (5.44) 1.370 (4.39) -0.581 (0.66) 0.586 (1.07) 3.526*** (0.49) 0.363 (0.26) 1.023 (1.65) 0.088** (0.04) 52.901*** (14.66) 449 0.313 11.566 138.440 0.919 0.998 0.314 3.362*** (1.21) 9.810 (9.54) 22.983*** (6.02) -0.306 (0.35) -1.310 (0.82) 38.597** (17.22) 390 0.251 8.976 78.110 0.849 0.997 0.179 2.069* (1.21) 8.477 (9.28) 19.157*** (5.83) 0.012 (0.34) -1.405* (0.78) 2.883*** (0.56) 0.794*** (0.24) 1.086 (1.56) -0.028 (0.03) 41.030** (17.48) 390 0.335 10.364 149.510 0.496 0.998 0.610 2.313* (1.21) 9.765 (9.35) 18.681*** (5.88) 0.005 (0.34) -1.326* (0.79) 2.998*** (0.56) 0.789*** (0.25) 1.179 (1.58) -0.028 (0.03) 52.749*** (17.09) 390 0.321 10.353 132.510 0.520 0.998 0.253 Results of two – stage least squares regression Column II and V of table show the results of first stage estimation with the dependent variable of hours worked for children in two areas The instruments perform well in predicting child work The income from crops reduce expectedly the number of hours worked of children in both sample following income effect, but this factor is only statistically significant in rural sample Also, the community – level rice prices affects the intensity of work of children based on 6671266 ... measure the total effect (both direct and indirect) of child labor on educational achievement Indeed, the coefficient of the working variable will reflect the direct impact of employment on learning... study explores the direct and indirect impact of child work on educational achievement of children at the age from 11 to 20 across rural and urban areas in Vietnam, using the data of Young Lives...UNIVERSITY OF ECONOMICS HO CHI MINH CITY VIETNAM INSTITUTE OF SOCIAL STUDIES THE HAGUE THE NETHERLANDS VIETNAM - NETHERLANDS PROGRAMME FOR M.A IN DEVELOPMENT ECONOMICS THE DIRECT AND INDIRECT IMPACT OF