DETERMINANTS OF SECONDARY SCHOOL DROPOUT IN VIET NAM

Transcription

1 MINISTRY OF EDUCATION AND TRAINING UNIVERSITY OF ECONOMICS, HO CHI MINH CITY NGO HOANG THAO TRANG DETERMINANTS OF SECONDARY SCHOOL DROPOUT IN VIET NAM MASTER OF PUBLIC POLICY THESIS HO CHI MINH CITY

2 MINISTRY OF EDUCATION AND TRAINING UNIVERSITY OF ECONOMICS, HO CHI MINH CITY FULBRIGHT ECONOMICS TEACHING PROGRAME NGO HOANG THAO TRANG DETERMINANTS OF SECONDARY SCHOOL DROPOUT IN VIET NAM Public Policy Major Code: MASTER OF PUBLIC POLICY THESIS SUPERVISOR Dr. JONATHAN R. PINCUS HO CHI MINH CITY

3 CERTIFICATION I certify that the substance of the thesis has not already been submitted for any degree and is not being currently submitted for any other degrees. I certify that to the best of my knowledge any help received in preparing the thesis and all sources used have been acknowledged in the thesis. The study does not necessarily reflect the views of the Ho Chi Minh City Economics University or Fulbright Economics Teaching Program. Author Ngo Hoang Thao Trang

4 ACKNOWLEDGEMENTS I would like to express my deep gratitude to my parents who always encourages me in my life and my studying. I would like to express my sincere appreciation to my supervisor, Dr. Jonathan R. Pincus, who has help me in performing the thesis. With rich knowledge, experience and enthusiasm, he has effectively contributed to my thesis. I am graceful to Dr. Nguyen Hoang Bao, Mr. Nguyen Xuan Lam for thoughtful and valuable comments on the early version of my work. I would like to thank all teachers in Fulbright Economics Teaching Program, who have retransmitted a lot of their knowledge and experience to me. Last but not least, I express my thanks to all of my friends who help and motivate me in performing the study. Ngo Hoang Thao Trang Ho Chi Minh City - May, 2010

5 ABSTRACT The study examines the effects of individual, household, community, and regional level on the dropout behavior of the child in secondary school in Vietnam by using the logistic regression model. The results of the empirical model confirmed age, working hours per year, education of parents, regions have large effects on the probability of leaving school. Meanwhile, household expenditure, the number of siblings, the proportion of pupils with reduced contributions, the pupil to teacher ratio, the pupil to classrooms ratio and the proportion of classrooms with good blackboards have small effects on the probability of leaving school. Therefore, an effective policy recommendation to reduce the dropout rate in secondary schools in Vietnam is to focus on investing more human capital in the present generation as well as supporting the important role of parent in education, reducing working hours per year of children and concentrating on the Mekong Delta and South East. Key words: School dropout; secondary school.

6 ABBREVIATIONS GSO: General Statistic Office MOLISA: Ministry Of Labor Invalids and Social Affairs VHLSS: Vietnam Household Living Standard Survey

7 CONTENTS CHAPTER 1: INTRODUCTION Problem statement The scope and the purpose of the study Research questions of the study The structure of the study CHAPTER 2: LITERATURE REVIEW Concepts Theoretical background Demand for education Education production function The framework of determinants of the dropout Empirical studies of school leaving in Viet Nam Conclusion CHAPTER 3: OVERVIEW OF SECONDARY SCHOOL IN VIET NAM Schooling trends and dropouts from secondary school in Vietnam Analyzing opportunities to access secondary school in Vietnam Conclusion CHAPTER 4: METHODOLOGY AND ANALYTICAL FRAMEWORK Data Methodology

8 4.3 Empirical model Variables in the empirical model Conclusion CHAPTER 5: ANALYZING DETERMINANTS OF DROPOUT SECONDARY SCHOOL IN VIETNAM Descriptive statistics Regression results Interpretation and discussion Scenario analysis Conclusion CHAPTER 6: CONCLUSION Conclusion Policy recommendations Limitations of the study. 60 REFERENCES APPENDIX

9 LIST OF TABLES Table 5.1: Logistic regression of school dropout as a function of selected individual, household, school and demographic characteristics, and the children sample aged 11-18, VHLSS Table 5.2: Marginal effects of the probability of dropping out.. 44 Table 5.3: Predicted dropout probabilty of children by education level of parents...52 Table 5.4: Predicted probability of dropping out by hours working per year 54 Table 5.5: Predicted probability of dropping out by region..55

10 LIST OF FIGURES Figure 3.1: Number of pupils in secondary school in Vietnam from Figure 3.2: Secondary education dropout and repeat rate, Figure 3.3 Secondary education dropout and repeat quantities, Figure 5.1: Dropout rate in Vietnam in 2006, by gender and by age...38 Figure 5.2: Dropout rate in Vietnam in 2006, by hours working per year 39 Figure 5.3: Dropout rate in Vietnam in 2006, by household expenditure 39 Figure 5.4: Dropout rate in Vietnam in 2006, by parent educational level.. 40 Figure 5.5: Dropout rate of Vietnam in 2006, by urban and regions 42 Figure 5.6: Conditional effect plot of dropout probability according to educational level of parents..53 Figure 5.7: Conditional effect plot of dropout probability according to hours working per year...54 Figure 5.8: Conditional effect plot of dropout probability by regions

11 1 CHAPTER 1 INTRODUCTION 1.1 Problem Statement Education is widely considered as an engine of economic and social development. The power of schooling is to raise incomes, to increase productivity and to promote social equity. Therefore, educational expansion at all levels is important for poor countries and the poor because it is the most powerful instrument for a society to escape from poverty (World Bank, 1999). The importance of public investment in education for economic development is shown by the success of the East Asian economies such as Hong Kong, Singapore and Korea. (Chew et.al., 1999). Moreover, research on returns to education often finds that returns to the primary level are highest and returns to secondary level are the second highest (Psacharopoulos and Anthony, 2002). Vietnam has recorded achievements in education and developed a comprehensive education and training system. Nevertheless, Vietnam is facing a crisis in terms of the quality of education at all levels and a crisis of dropouts of pupils. For years from 2001 to 2005, more than four million pupils have left school: more than one million at the primary level, two million at the lower secondary level and a million at upper secondary level. According to statistics for the first semester of the year from the Ministry of Education and Training, among fifteen million pupils in Vietnam, eighty six thousand have dropped out of school: nine

12 2 thousand at the primary level, forty thousand at the lower secondary level and thirty eight thousand at the upper secondary level. The Mekong Delta had the highest number of dropouts at twenty five thousand. The Central Highlands had eleven thousand dropouts and the North-West 4,600 dropouts. The consequence of pupils dropping out include private and public costs. For individuals, they cannot get skilled employment so they earn little money and face many risks and discrimination in society. For the country, dropouts reduce the rate of growth of the skilled labor force. Moreover, dropouts make the gap between the poor and the rich larger. Motivated by this situation, this study analyzes the determinants of the dropout rate from secondary school in Vietnam. The contribution of the study is to find out which factors influence the dropout rate and to suggest policies to reduce it. 1.2 The scope and the purpose of the study The study focuses on analyzing empirically factors that affect dropout behavior of children at the individual, household and school level. The analysis is carried out on the case of dropouts from the secondary level in Vietnam in The study aims at identifying the degree of impact of factors affecting the dropout rate from secondary school. The study provides input for policy makers formulating policies to reduce the dropout rate at the secondary level in Vietnam.

13 3 1.3 Research questions of the study The study attempts to answer the central question: What are the most prominent factors affecting secondary school dropouts in Vietnam?. In order to answer the central question, some following sub questions should be addressed: What are the factors theoretically affecting dropouts? To what extent do these factors explain dropouts at the secondary level in Vietnam? What should be done to reduce dropouts from secondary school in Vietnam? 1.4 The structure of the study The study consists of six chapters. Following this introduction, the rest of the study is structured as follows. Chapter 2 is devoted to the literature review in order to provide the analytical framework for the thesis. It will present concepts, discuss the theoretical background, and then lay out a framework for studying the determinant of school dropouts and review empirical studies concerning dropouts in Vietnam. Chapter 3 is an overview of secondary schools in Vietnam. The methodology and empirical model are presented in Chapter 4. It will describe the dataset, set up the methodology, explain the empirical model and describe the variables used. Chapter 5 reports and discusses the empirical results of the model. Chapter 6 concludes the study with a summary of the main findings and policy recommendations. It also includes some remarks on the limitations of the study.

14 4 CHAPTER 2 LITERATURE REVIEW Chapter 2 provides a review of the conceptual, theoretical and empirical studies of the dropout behavior of children. This chapter starts its journey with the concept of school dropouts. Next, it lays out the theoretical background on educational demand at the individual level and the household level. Then it presents a survey of the literature on key dropout factors. It discusses empirical studies in Vietnam in order to show the effects of such factors on dropouts. 2.1 Concepts As a first step, it is useful to provide the concept of school dropouts used in the study. According to MOET, a child is viewed as a dropout if he or she did not continue his or her schooling. Vo Tri Thanh and Trinh Quang Long (2005) criticize this definition since it does not count those who do not continue to enroll in school after having finished a given grade. This may lead to an underestimation of school dropouts. In the study, a child is a dropout if he or she has not completely enrolled in school in the twelve months prior to the survey, given that he or she used to enroll in school sometime before.

15 5 2.2 Theoretical background This sub section is devoted to a review of the main literature relating to investment in education on the demand and supply sides. Then it discusses factors affecting dropouts in recent empirical studies. The subsection is not intended as a complete and extensive review, it merely sets the stage by presenting relevant considerations Demand for education An elegant theoretical framework regarding determinants of the optimal length of education has evolved from the human capital theory of Becker (1967) and Ben- Porath (1967) [cited in Ranasinghe (1999)]. Human capital theory argues that additional schooling generates benefits in terms of enhanced future earnings and entails direct costs such as expenditures on school tuition and opportunity costs associated with delayed entry into the labor market. The individuals will compare the direct and the opportunity costs of schooling with its future benefits. The investment will continue so long as the marginal rate of return to additional schooling equals the marginal rate of the cost of education. According to the theory of Becker (1967), differences in the length of schooling may be due to differences in the discount rate or due to differences in the rate of return to education or the covariance between the two. People with lower discount rates (and low costs) and those with higher rates of return will stay longer in school than others do. Another schooling model developed in the spirit of Becker is Ben-Porath (1967). Ben Porath (1967) takes into consideration the dynamic nature of human capital

16 6 formation to develop his model. In general, the human capital models of Becker (1967) and Ben-Porath (1967) generate similar testable hypotheses regarding the demand for education. However, in the Ben-Porath model (1967), some particular features are different from the Becker model. Ben-Porath (1967) posits a direct role for age whereas Becker (1967) does not included age as a determinant of the equilibrium length of school. Next, the role of ability and family wealth is also different in the two models. According to Becker (1967), ability affects the rate of return. More able people stay longer in full time education because they have a higher rate of return. However, according to Ben-Porath (1967), ability affects duration through costs. The human capital theory of Becker (1967) and Ben-Porath (1967), however, implicitly assumes no resource constraints. This assumption does not hold in the real world. Therefore, resource constraints place a limit on the amount of resources that can be allocated within a household. In household schooling decisions, parents are viewed as the principal and children are viewed as agents. Parents can view education as both a consumption and an investment good so parents decide the education level of their children to maximize their utility. According to the model of household decision making regarding investments in education of boys and girls (Glick and David, 1998), parents are considered to live over two periods. In the first period, they earn money spending; feeding their children and spending money for their children to go to school. Consumption in the first period includes personal spending and spending on the education of children. In the second period, the

17 7 consumption of parents will depend on remittances from their children. Nevertheless, remittances from their children depends on the returns on educational investments of their children as well as their specific characteristics. Parents will choose the duration of their children s education to maximize their utility subject to income, time constraints for the household members and the earnings of the production function for each child. The limitation of household models is that preferences of parents for their children are identical. To overcome the limitation of the model, bargaining models assume that parents have different preferences for their education of sons and for that of boys leads to gender specific demand functions for schooling. Differential preferences may be a response to the actual or perceived differences in the labor market returns to female and male schooling (Patton, 1993) [cited in Chu Bao Hiep (2008)]. In the study, human capital models are embedded into a model of household demand where both parents and the individual child are decision makers. This framework captures the close economic interdependence between the child and the family. The basic implication of such a framework is that the decision to attend school depends not only on market forces and public programs that determine the costs and benefits of education, but also on the preferences of the family, resource constraints and alternative uses of the time of children in non-school activities, such as work or leisure.

18 Education production functions The human capital theories and household schooling decision models assume that schools are homogenous. In practice, schools are different from each other in productivity. A systematic investigation of the relationship between school inputs and the output of pupils is condensed in the education production function. An education production function assumes that the optimal learning process in schooling can be approximated by production theory. Inputs into the education production process include school inputs, peer groups inputs and inputs of individual pupils. The outputs of the education production process are expressed as a single dimension or multiple dimensions. The empirical estimates of the education production function choose a single output as the dependent variable. Polacheck et al. (1978) [cited in Ranasinghe (1999)] use exam scores as the output of pupils, Card and Krueger (1992a and 1992b) [cited in Ranasinghe (1999)] use labor market returns and Mora (1997) [cited in Ranasinghe (1999)] use school dropout decisions. Furthermore, [Chizmar and Zak (1983), (1984) and Chizmar and McCarney (1984), cited in Athula Ranasinghe (1999)] choose more than one output at the same time sharing the given amount of inputs. 2.3 The framework of determinants of school dropouts Now that the common theoretical framework is already established, there is a need to know more specifically about the determinants of school dropouts. These determinants are organized into four groups and each determinant impacts positively, negatively or both on the probability of dropping out. Each determinant

19 9 is individually considered through the presentation of the most relevant results in theories and in empirical studies. Group 1: Individual characteristics Gender This subsection deals with the relationship between gender and school dropouts. On this point, there is a consensus that gender inequality affects the dropout rate of boys and girls. One line of reasoning is that parents predict that the returns to the education of boys is higher than girls (Schultz, 1993). In the labor market, girls may be discriminated against in terms of earning, so the future earnings of girls are lower than that of boys. This means that monetary benefits to invest in education for girls may be lower than for boys. Moreover, even if educated girls receive earnings on a par with men, income remittances to parents from married women may be lower than married men. Another line of reasoning is that the opportunity cost of educating girls is higher than for boys. Girls in developing countries and in rural areas have to perform more domestic responsibilities than boys, reflecting cultural or social attitudes toward the proper economic role of women and girls. This means that the marginal cost of time of girls may be higher than that of boys and consequently the demand for their schooling will be lower (Glick and David 1998). Furthermore, girls have a higher probability of dropout if budgets of parents are constrained (Deaton, 1989). On the other hand, gender disparities in education are different between countries. According to Filmer (2006) girls are at a great educational disadvantage

20 10 in particular regions such as South Asia and North, Western, and Central Africa, especially in poorer households. At the other extreme there are countries, mostly in Latin America, where there is no female disadvantage, and often a small female advantage in education. Age This subsection focuses on the effect of age on dropouts. On this point, there is a consensus that the age of the child has a positive impact on the probability of dropout. According to Ben-Porath (1967) [cited in Ranasinghe (1999)], high opportunity costs and lower marginal benefits for older children discourage parents from investing in full time education for their children. At the working age, children may enter the labor force to help their parents. Moreover, higher costs of school fees at higher levels may be a constraint on enrollments of children. Child work This subsection is concerned about whether the dropout rate is related to the working status of the child. Many studies explain factors affecting child labor participation and the relationship between dropout decisions and work. Most researchers find a positive relationship between school dropout and child labor participation. According to Admassie (2002), child employment interferes with schooling by absorbing too much time of the child. Work also requires a lot of energy so that children cannot have the necessary energy for school attendance or for effective studying. Even for those children who might be able to combine farm work or domestic work with schooling, long hours of work will leave them

21 11 exhausted. Suffering from fatigue, these children will have very little mental stimulation, the results of which will be neglect of their studies. Group 2: Household characteristics Education of parents The educational level of parents is considered as an important factor affecting the dropout rate of children. Extensive empirical studies have provided universal support for the above argument. The empirical study of Glick and Sahn (1998) supported the negative relationship between the educational level of parents and the dropout probability. They argue that educated parents are more likely to recognize the benefits of schooling than less educated parents. Moreover, educated parents are more able to assist in the learning of their children than less educated parents. On the other hand, there is evidence that the effects of parents educational attainment is different between boys and girls. According to Tansel (1997) an increase in the schooling years of parents decreases the probability of a child dropping out. Moreover, he found that there are differential effects for boys and girls. The effects of the educational level of both mother and father on the probability of a daughter dropping out was larger than on sons except for the education of the mother at the primary level. This may be due to several reasons. It is possible that relatively uneducated parents tend to be in locations where there are some barriers to the school attendance of girls. It is also possible that the education of parents has a big effect on their attitudes towards girls schooling.

22 12 Household income Like education of parents, the negative relationship between income and the probability of dropping out is strongly supported by theoretical as well as empirical studies. According to Glick and Sahn (1998), wealthier household are likely to be able to pay for schooling out of current income or savings and have easier access to credit. Children from such households are expected to be more likely to enroll and to stay in school longer. Otherwise, poor households may be unable to afford the direct or indirect costs of schooling and are constrained in their ability to borrow to cover the costs. Moreover, richer household can invest in the education of children by tutoring at home or improving the health of children, thus increasing demand for schooling (Le Van Chon, 2000). On the other hand, short run income income variability affects child labor and schooling in developing countries. According to Duryea, et al. (2006), an unemployment shock (like job loss) significantly increases the probability that a child enters the labor force, drops out of school, and fails to advance in school. This result suggests that some households are not able to absorb short-run economic shocks, with negative consequences for children. Number of children This subsection focuses on the relationship between the number of children in the household and the probability of dropout. Most studies find a positive relationship in developing countries, including Peru and Brazil (Psacharopoulos and Arriagada, 1989). The number of children increases the probability of dropout because of

23 13 budget constraints on households. Moreover, according to Parish and Willis (1993) girls more are more likely to dropout than boys because an increase in the number of very young children may raise the demand for the labor of girls in childcare in the home. However, Chernichovsky (1985) argued that the tradeoff between child schooling and the number of children is rejected by the data he analyzed. He found that means that the number of children in a household has a negative effect on child dropouts in Botswana. Chernichovsky (1985) argued that role assignment in the household permits specialization, so that some children are assigned household and farm activities, while others are sent to school and concentrate on this activity. Group 3: School characteristics School fees School fees are considered as an important factor affecting the probability of dropout. Many theoretical and empirical studies form a consensus on this issue. Becker (1975) [cited in Ranasinghe (1999)] argues that whether a child goes to school depends on a comparison between future benefits of the human capital of the child and the direct and opportunity costs of schooling. So higher costs of schooling increase the probability of dropout. Most empirical research concludes that school characteristics related to costs and benefits influence enrollment and attainment of children (Al-Samarrai & Peasgood, 1998; Chernichovsky, 1985; Dostie & Jayaraman, 2006; Wolfe & Behrman, 1984; Zimmerman, 2001). Furthermore, Glewwe and Patrinos (1999) state that if school fees are excluded from models of

24 14 determinants of school dropouts, it would limit our understanding of family decision making on investing in education. He concluded that high costs keep some children out of school and this is particularly important for countries where free education is limited by resource constraints. School quality This subsection focuses on dealing with the connection between school quality (infrastructure and teacher qualifications) and dropouts. There have been inconclusive results in the empirical literature. No consensus has been reached on the direction of the connection as many theoretical and empirical studies provide contradictory arguments and evidence. Coleman (1966) [cited in Kain and Singleton (1996)] report that school inputs such as class size, teacher quality and infrastructure have a marginal influence on the achievement of pupils. A study in China concluded that community resources and school provisions reinforced transmission of background effects from parents to children, due to effects of school availability and quality (Hanum, 2003). Other studies in developing countries have also reported that common causes of inequality in educational opportunities were regional differences and school availability, expenses, and quality, as measured by such factors as student teacher ratios, teacher qualifications, and school fees and contributions. These inequalities influence not only the probability of dropout but also test scores of pupils. Similar research, including Brown & Park (2002), has found that school quality contributed positively to enrollment decisions, but not to learning outcomes in rural China. Moreover, Oakland (1986a &1986b) emphasizes

25 15 that strong and influential administrative and instructional leadership enhances school effectiveness. Principals in schools with lower drop out rates tend to provide stronger leadership and to use their staff more effectively. Stronger leadership is evident in their ability to communicate expectations for and their enforcement of student and teacher discipline, in their management style, and in their formulation of clear educational mission statements for the school that include the belief that all students are capable of displaying academic success. However, Hanushek (1986) [cited in Ranasinghe (1999)] argued that although school inputs such as teacher to pupils ratios, teacher education, teacher experience, teacher salaries, expenditure per pupil and selected administrative and physical facility variables are presumed to have a positive effect on the school output, these variables are statistically insignificant and there are many cases where the regression coefficients are negative. This finding is rather similar for both developed and developing countries. Loeb and Bound (1966) [cited in Athula Ranasinghe (1999)], through an extensive review of the literature, have observed that, in general, school inputs are significant when the labor market variables are used as the education output, whereas they are insignificant when academic outcomes are used as the education output. Group 4: Regional characteristics Demographic differences This subsection examines the relationship between dropping out of school and demographic differences. They are suggested as factors affecting dropout of

26 16 children. According to Ono (2000), children living in urban areas have a lower probability of dropout than rural children. The reason is that urban children have more opportunities to attend school. Rural areas, for example, may face higher opportunity costs for child schooling due to farm work options, as well as higher costs for transportation. Next, schools in rural areas especially secondary schools are often far away from the homes of children. Therefore, children in rural areas have suffered disadvantages in attending school, such as walking long distances to go to school, spending more time and money, which discourages school age children from getting more education. Moreover, different regions in the country possess different natural, economic and social conditions. According to Vo Tri Thanh and Trinh Quang Long (2005), geographic disparities significantly contributed to differences in school enrollment. Poor regions tend to have higher dropout rates. 2.4 Empirical studies in school leaving in Viet Nam The theoretical and empirical studies reviewed above raise important issues. Studies conducted in Vietnam emphasize strong effects of individual characteristics, household characteristics, school characteristics and regions on educational opportunities of children in general education. Behrman and Knowles (1999) examined the role of household income contributing to child schooling in Vietnam by using data from the 1996 Vietnam Social Sector Financing Survey. They found a considerable and significant association between parental income and schooling of children on grades completed

27 17 per school year. Moreover, gender differences occur between boys and girls. Schooling of girls is treated as more of a luxury (less of a necessity) than is schooling of boys. They also report that school fee exemptions are a significant mediator to reduce the association between family income and schooling. However, this factor does not have a broad impact on inequality in educational enrollment, because school fees only account for one-third of total household expenditures on education. Truong et al. (1999) [cited in Mai Thi Hoang Yen (2002)] reported that children aged from eleven to fourteen are likely to enter the labor force instead of attending to school. They also found that education of parents and family wealth positively affect enrollment. Vo Thanh Son, et al. (2001) [cited in Mai Thi Hoang Yen (2002)] reconfirmed the controversial effects of family wealth and gender on educational outcomes using a sample of children aged from twelve to twenty from the VHLSS of He found that children from poor families and girls were likely to drop out of secondary school. Moreover, school fees and probability of school dropout are strongly correlated. Therefore, government subsidies for school fee are necessary for disadvantaged groups. Nguyen Nguyet Nga (2002) reported that geographic disparities significantly contribute to school enrollments. Poor regions have lower school enrollment rates than rich regions. Moreover, not surprisingly, she concluded that school enrollments were still not equally distributed across income expenditure quintiles. He also found

28 18 no gender differences in net school enrollments at either the primary or the secondary level. Vo Tri Thanh and Trinh Quang Long (2005) identify the underlying determinants of the school dropouts in Vietnam by using three Living Standard Surveys in Vietnam conducted in , and They found that the major determinants of school dropouts were child characteristics (such as age, working time), household background (such as parental education, household expenditure and cost of schooling) and geographical differences (such as urban vs. rural, seven regions in Vietnam). Among these major determinants of school dropouts, the school dropout probability is very sensitive to the changes in household expenditure per capita and the direct cost of school, whereas other determinants have had only a minor impact. Nguyen Linh Phuong (2006) examined the effects of parental socioeconomic status, school quality and community factors on the enrolment of children and achievement in rural areas in Vietnam by using data from VHLSS The research revealed that gender, parental education, number of siblings, and ethnicity have a significant association with school enrollments. Furthermore, school and community factors seem not to affect school enrollments.

29 Conclusion The main purpose of this chapter has been to review the literature related to demand for and supply of education. Moreover, chapter also reviewed empirical models of determinants of school dropouts. From the educational demand perspective, the human capital theory of Becker (1967) and Ben-Porath (1967) argues that a rational individual chooses the optimal length of school in order to maximize the net present value of lifetime wealth. But in the real world, people are constrained by resources. Therefore, in household decision making regarding investments in education, parents choose the time that children spend in school to maximize their utility subject to income and time constraints of the household and the returns from the production function for children. Human capital theory and household demand models assume that schools are homogenous but in fact, schools are different. Therefore, literature on the education production function is used to explain the relationship between school inputs and outputs of pupils. Dropping out of school is not a random occurrence. There are many factors which might affect the problem of dropping out prematurely. These factors are organized into four groups: individual characteristics, family characteristics, school characteristics and regional characteristics. Empirical studies on dropouts in Vietnam emphasize strong effects of parental socioeconomic status, family wealth and region on educational opportunities of children in general education.

30 20 CHAPTER 3 OVERVIEW OF SECONDARY EDUCATION IN VIETNAM Chapter 3 provides an overview of secondary education in Vietnam. Firstly, it introduces schooling trends and dropouts in secondary school in Vietnam. Then opportunities to access secondary school are analyzed from different perspectives. 3.1 Schooling trends and dropouts from secondary school in Vietnam The formal general educational system in Vietnam consists of three levels: primary, lower secondary and upper secondary. Children are supposed to spend five years in primary school, beginning from the age of six to eleven. Children spend four years in lower secondary school and three years in upper secondary school. Therefore, with this general formal progress, pupils complete their formal educational training by the age of seventeen. Recent trends in secondary education are shown in figure 3.1. At the lower secondary level, the number of pupils has increased year by year from 1999 to 2004 at the rate of 180 thousand pupils per year. However, in 2005 the number of pupils decreased by 212 thousand of pupils and by 240 thousand of pupils in At the upper secondary level, the number of pupils increased on average by 16 thousand pupils per year from 1999 to 2006 (Appendix 1).

31 pupils 21 Secondary school enrollment in Vietnam from 1999 to Lower secondary Upper secondary Figure 3.1: Number of pupils in secondary school in Vietnam from Source MOET (2006) From 1999 to 2004, the dropout rate in upper secondary school was higher than that from lower secondary school. The average dropout rate in lower secondary school in this period was about 6.41% and 7.5% in upper secondary school. At the lower secondary level, the dropout rate decreased from a peak of 8.51% in to 5.12% in In upper secondary school, the dropout rate has shown an increasing trend. The peak dropout rate was 8.29% in (Appendix 1). The average repeat rate in lower secondary school in this period was about 1.19 % and 1.28% in upper secondary school. At the lower secondary level, the repeat rate decreased from a peak of 1.93% in to 0.83% in At the upper secondary level, the repeat rate has shown an increasing trend. The peak repeat rate was 1.4% in (Appendix 1).

32 22 Trend in Dropout rate by level of education Repeat rate by level education % 2,5 2 1,5 1 0,5 % Lower secondary level Upper secondary level Lower secondary level Upper secondary level Figure 3.2: Secondary education dropout and repeat rate, Source MOET (2006) When we consider the dropout rate and repeat rate in absolute terms, we can see another picture of the dropout and repeat rate in Vietnam. Although, the dropout and repeat rates in upper secondary school are higher than in lower secondary school, absolute figures show that dropouts and repeats in lower secondary school are higher than those in upper secondary school. This situation can be explained by the fact there are more pupils in lower secondary school than upper secondary school. (Appendix 1). Figure 3.3 Secondary education dropout and repeat quantities, Source MOET (2006)

33 Analyzing of opportunities to access secondary school in Vietnam In this subsection, a picture of opportunities to access secondary school in Vietnam is drawn to state that there exists inequality in opportunities to attend secondary school between the rich and the poor and between different regions. Figures on enrollments, expenditure on education, learning capacity of pupils, quality of schools, government education expenditure and exemptions are used to analyze this situation. The Vietnam Household Living Standard Survey (2006) is used in this analysis. Firstly, enrollment rates differ between levels of education and regions. In general, gross enrolnment rates for lower secondary are higher than those of upper secondary (96% vs 73.6%). At the lower secondary level, gross enrolment rates in rural and urban areas are over 95% whereas at the upper secondary level, the gross enrolment rate in rural areas is only 70.1% versus 85.7% in urban areas. The Mekong Delta has the lowest enrollment rate (86% in lower secondary and 55.7% in upper secondary) (Appendix 2). Secondly, average education and training expenses by education level are different between expenditure quintiles and regions. Expenses for education at the upper secondary level are higher than those for lower secondary. Moreover, very poor households spend 382 thousand VND/year on lower secondary school and 831 thousand VND /year on upper secondary, whereas the figures for rich households are 1.4 million VND/year and 2.2 million VND/year. This means that poor households spend less than one-third the amount paid by rich households.

34 24 (Appendix 3). Expenses for education in urban areas are double those in rural areas in lower and upper secondary school (Appendix 4). Among different regions, the South East has the highest average expenditure on education at both levels of school, while the North East and North West have the lowest levels at both levels (Appendix 5). Analyzing average expenditure on education and training per person by expense item, we can conclude that expenditures of poor households make up a small percentage of total household expenditures. Average spending on education among poor households makes up 2.4% of total household expenditure, whereas that of rich households makes up 12.3% (Appendix 6). Moreover, poor parents have to bear not only the whole cost of tuition fees but also other school fees such as clothes, contributions to school funds, text books, study tools and extra classes. In poor households, school fees account for 21% of total school expenses and other school fees account for 79% of total school expenses (Appendix 7). Thirdly, the learning capacity of pupils is different between expenditure quintiles and regions. In very poor households, only 5.5 % of pupils have distinction grades and a majority of pupils have normal grades. In rich households, 28.1% earn excellent grades and a majority of pupils have good grades. In rural areas, a majority of pupils have normal grades (52%) while in urban areas, a majority of pupils have good grades (41%). The Red River Delta, South Central Coast and South East have the highest rate of pupils having excellent and good grades (over 15% excellent grades and over 30% good grades). The North East, North West and

35 25 Central Highlands have the highest rate of pupils having weak grades (4.5% in the Central Highlands) (Appendix 8). Fourthly, the evaluation of households of the biggest difficulties differs between expenditure quintiles and regions. Very poor households reported that the biggest difficulties are a lack of facilities in school (29.6% in very poor households and 34% in poor households), quality of construction (18.5% in very poor households and 15.2% in poor households) and quality of teachers (17.9% in very poor households and 18.1% in poor households). In rural areas, the biggest difficulties are lack of facilities (30.4%) and quality of teachers (17.8%). Lack of facilities is the biggest difficulty in the Red River Delta, North East, North Central Coast, South Central Coast, South East and Mekong Delta (over 30%). However, in the North West and Central Highlands, the biggest difficulties are quality of construction (20%) and quality of teachers (24%) (Appendix 9). Fifthly, government expenditure on education over total government expenditure has increased from 14% in 1998 to 18.6% in Expenditures on general education are increasing while training education is decreasing. The government prioritizes general education in remote or mountainous areas and for ethnic minorities (Appendix 10). In addition, educational benefits from government expenditures show are unequal between the rich and the poor especially at the upper secondary level. For example, at the upper secondary level, the share of education benefits of very poor households is only 9% whereas that of rich households is up to

36 26 26%. This situation can be explained by the fact that the poor do not go to upper secondary school but the rich do. (Appendix 11). 3.3 Conclusion This chapter has focused on schooling trends, dropouts and opportunities to access secondary school in Vietnam. In general, secondary pupils at both levels are increasing year to year. There are more pupils in lower secondary school are higher than those in upper secondary school. Although the dropout and repeat rates in lower secondary school are lower than those of upper secondary, a larger number of pupils drop out or repeat grades in lower secondary than upper secondary. There remain inequalities in accessing secondary school between the rich and the poor and among regions.

37 27 CHAPTER 4 METHODOLOGY AND ANALYTICAL FRAMEWORK This chapter plays an intermediary role in linking the theoretical background to the empirical analysis. It firstly gives information on the data set. It next develops the analytical framework in order to specify the empirical model. After presenting the specification of the empirical model used to study the factors of dropouts, the variables used are described and justified. 4.1 Data Data for the empirical analysis has been obtained from the Vietnam Household Living Standard Survey (VHLSS) in 2006 which was conducted by General Statistic Office (GSO). The survey provides basic information at the household level such as demography, education, health, employment, income and expenditure. At the same time, it also provides information at the commune level such as population, religion, education, credits and other indicators. The survey was conducted nation-wide, involving a sample scale of 45,945 households (36,756 households for the income survey, 9,1898 households for income and expenditure survey) in 3,063 communes, representative for the whole country, eight regions, urban areas and rural areas. Within the scope of the study, only households having children aged from eleven to eighteen that have completed the primary level are included in the study. The data were filtered to remove errors and inconsistencies. After checking the completion of information for each variable of interest, the sub-

38 28 data set comprises 4,369 observations of which 18.4% are dropouts. Individual, household and school variable descriptive statistics are shown in Appendix Methodology We are interested in whether a child aged from eleven to eighteen years old drops out from secondary school or not. Therefore, the dependent variable will be the dichotomous variable that will take the value one when a particular characteristic is present and zero otherwise. There are some models which can be used to examine and predict the probability of school dropout such as linear probability (LMP), logit and probit models. According to Gujarati (1995), the linear probability model suffers problems such as: the estimated probabilities range from zero to one in this model is not guaranteed, R 2 value is lower, the non-normality of, heteroscedascity of, and marginal effects remain constant. The logit and probit models are alternatives to the linear probability model. The difference between the logit and probit models is that the curve of the logit approaches the axes slower than that of the probit model. In practice, the logit model is employed in many studies because of its mathematical simplicity (Gujarati, 2003). Therefore, the logit model is used in the study. According to (Gujarati, 2003), the logit model is based on the cumulative probability function which is mathematically specified as: (4. 1) Where Y=1: dropout; Y=0: not dropout

39 29 It is quite easy to verify that as ranges from to, ranges between 0 and 1. If is the probability of dropout and (1- ) is the probability of not dropout, we can write: (4. 2) is simply the odds ratio in favor of dropout. Taking the natural logarithm of the equation, we obtain: (4. 3) Meaning of the coefficients in the model Stating is the odds ratio in favor of dropout, is the beginning dropout probability. From equation (4.3) we obtain: (4.4) Assuming that the other variables in the model are constant, when is increased by one unit, will be: (4.5) (4.6) (4.7) When increases by one, the dropout probability of a child increases by:

40 30 (4.8) 4.3 Empirical model In the logistic model, the determinants of the incidence of dropout are described as: Sample regression function (SRF): (4.9) in which : is a probability of dropout : variables concerning individual characteristics : variables concerning family characteristics : variables concerning school characteristics : variables concerning regional characteristics 4. 4 Variables in the empirical model Dependent variable The dropout of a child is represented by a dichotomous variable. This variable is an observable event taking the value of one if the child drops out and zero if he or she does not. Independent variables Group 1: Individual characteristic variables Gender represents the gender of the child. It is a binary variable that takes the value of one if gender is male; otherwise if gender is female, its value is zero. Gender is expected to have a negative effect on the probability of dropouts.

41 31 Child age is a continuous variable, which is measured in years. It is expected to have a positive effect on the probability of dropouts. Older children have a higher probability of dropping out of school because of higher costs of schooling as well as higher opportunity costs forcing him or her to leave school sooner to enter the labor force. Hours worked per year of the child is chosen to proxy for the working status of the child. It is a continuous variable which is measured by the total number of working hours per year of the child (hundreds of working hours per year). The expected sign is positive meaning that the more working hours per year of the child, the higher the probability of school dropout. However, the coefficient of this variable may be overestimated because a higher number of hours devoting to working may be the outcome of leaving school, not simply the cause of dropping out. Group 2: Household characteristic variables Household expenditure is included to capture the effect of income on dropouts. It is measured by total expenditure per year (in million VND) of the household. We use annual expenditure per year rather than annual income because household income data in developing countries such as Vietnam are usually not accurately reported. The expected sign is negative meaning that household expenditure has a negative correlation with the probability of dropouts. Education of parents is included to capture the effect of parent s experience in education on dropouts. Since data on the education of parents are not available, the

42 32 level of the head of household is used as a proxy for education of parents. The probability of dropouts is likely to differ depending on the of parents. Therefore, education of parents is classified into four levels: primary level, lower secondary level, upper secondary level and higher level (college or university or higher education). These levels are coded as three dummy variables. The primary level variable is used for reference. The lower secondary level variable, upper secondary level variable and higher level variable are four binary variables and constructed in order to capture the educational level effects, one for each of the levels of education of parents. Each variable takes a value of one if education of parents belongs to that particular level, otherwise it takes a value of zero. The expected sign is negative meaning that a child living with parents who have higher educational qualifications has a lower probability of dropping out than a child living with parents who have just completed primary education. Number of children is a continuous variable which is measured by the number of children aged from one to seventeen in the family. It is expected that the number of siblings has a positive effect on the probability of dropouts. Group 3: School characteristics The cost of school is a continuous variable which is measured by the cost of school (in million VND) per year that a child has to pay. For those who drop out of school, the cost of schooling is proxied by the mean cost of schooling of children at the same age and living in the same region. The expected sign is positive meaning that a higher cost of schooling increases the dropout probability.