INTRODUCTION

Although Vietnam is still one of the poorest and under developed countries in the world, it has achieved significant progress in educating its residents. Compared to other nations with the same level of economic development, Vietnamese people have higher level of education. The literacy rates is as high at 94% of the total population in comparison with other regional countries such as Laos (57 percent), and China (81 percent) (UNICEF Vietnam, 2000). UNICEF (2000), however, indicates that it is critical not only to get all the children into school but also to keep them in schools. According to a report of WB(1996), nationally in Vietnam in 1995, of 10,000 students entering primary school, only 6,810 finished the primary level.

Generally, high repetition and dropout rates reduce the overall education achievement level and increase the costs of education for government and families. However, the underlying determinants of school dropout and grade attainment are little covered in Vietnam. While aggregate data on completion levels, and on the age-grade distributions of students provide some overall sense of the dropout situation, these do not allow investigation of underlying behavioral factors or institutional structures that are driving high dropout rates. This thesis will use panel data derived from two Vietnam living standard surveys to explore the determinants of dropout behavior of children in rural Vietnam.

The Objective of Thesis

- Explore the educational attainment trend, in general and the dropout trend and highest grade attainment, in particular, during the period of 1992/1993-1997/1998

- Use  micro, meso and macro factors to explain these trends

- Try to find out  how differently girls and boys have to face with the same risks concerning their schooling.

The scope of the thesis

This thesis will seek to determine socio-economic factors that explain household’s decisions regarding dropping out of school and grade attainment of children. These socio-economic factors may be at the macro, meso, and micro level. The author, however, does not integrate macro and meso factors into the analysis determinants of dropping out and grade attainment at the micro level. Macro and meso determinants will be treated  exogenously in the household’s decision. These factors set the environment in which household will act to  maximize their own utility.

Main Research Questions

o         What was  the dropout trends in the period of 1992-1998?

o         How could these trends be explained by macro, meso policies during this period?

o         Which factors determined whether a child who was enrolled in school in 1992/1993 left the school in the following five years?

o         Are boys and girls affected by all the above factors differently in terms of their decision to stay in or to drop out of school?

o         What causes a child to leave school at a given grade? Is there any difference in determinants between girls and boys?

Data

There are several source of data. Some of them are qualitative data obtained from other researchers' studies. Some of them are official data. These data are utilized to explore the education trends and other aspects of education in Vietnam and to explain the educational trends, in general and the dropout trend, in particular. Most of the data used for analyses come from two VLSSs.. Besides, the other sources of data will also be utilized to explore the education trends and other aspects of education.

Structure of the thesis.

The thesis is divided into four chapters. The first chapter lays out the theoretical foundation for the whole analyses of this thesis. In the second chapter, secondary data and qualitative work will be utilized to identify the educational trends over the last five years. How government policies translate to children’s outcome also will be assessed in this chapter.  Chapter three will present results of multivariable regressions to reveal the factors determining children’s choice of dropout or choice of highest grade he/she will attend. The thesis ends with main findings and some recommendations.

Chapter 1

Theoretical Framework

In this chapter, a conceptual framework which tries to integrate factors at household level, the meso (community) or macro levels will be used to investigate the determinants of household's decisions regarding their children's schooling dropout.

I. Conceptual framework

In this conceptual framework, the attainments of children depend on three primary factors - the choice made by society (or government) that determine the opportunities available to both children and their parents, the choices made by parents regarding the quantity and quality of family resources devoted to children and the choices that children make themselves given the investments in and opportunities available to them. The conceptual framework is graphically presented below.

Figure 1: Conceptual Framework

This framework reflects a choice-based view. Government, parents and children will make decisions which serve their very own interests subject to their different utility and constraints. Of all these decision, those made by government and household are on behalf of children while those made by children themselves reflect their own utility function and constraints. These decisions will determine the children's educational attainment.

In the next sections, the role of government will be studied first, then followed by the theory of household's investment in children's education which try to integrate children's attributes into choices made by their parents. 

II - The role of government in education

II.1 - Foundation for government to participate in educational sector

Although the arguments for government's involvement differ from individuals to individuals (Hanusheck, 2000), the government's intervention in the education sector can be justified by three common arguments. The first rationale is the imperfection in the capital market. The imperfect capital market causes it to fail to provide the social desirable level of education investment (Taylor, 1999). The second argument is that education can produce positive externalities, that are benefits to society that exceed the benefits to the students and their families themselves. However, when parents make educational choices, they tend to invest less in education than would be socially optimal. (Taylor, 1999). The third rationale is that there exist a parentalistic concerns for their children and other irresponsible individuals (Friedman, 1955). Education is considered as a way for government to redistribute its resources in their own direction.

II.2 - Government Intervention and Children's Education

Given that government involvement in the education is to deal with three broad issues stated earlier, its participation in education is undeniable. However, the issues arising are related to the extent and the way that the government intervenes in the education.

To absorb positive benefits from education, government has required that each child must receive a minimum amount of education. However, difference in resources per child amongst household will make this policy hardly feasible. Thus, payments of the costs made by government are the only feasible means of enforcing the required minimum. In addition, the assumption that society will have more benefits from educating those who have great ability and interests requires the government to invest more in the higher level of education. These lead to a situation that has been criticized by Tan and Mingat (1992). Children coming from poor households are more disadvantaged than their friends coming from rich households who can absorb more public resource.

Government's participation in education through provision of educational services have been thought to assure that the common core of values required for social stability would be achieved or to keep education a unifying force. However, as Friedman (1955) has argued, government's participation in education does not mean that it should administer the school system.

Decentralization is viewed  not only as  a way to deal with the problem of imperfect capital market but also as a way to assure the efficiency of the resources allocated for education. Although decentralization cannot completely solve those problems, it does reduce the number of households that are liquidity constraints to just those who are constrained because lifetime income, not merely current income is too low. (Hoxby, 1996).

Decentralization will make government play only a subordinate role in education. In addition, it makes families become the primary educational financiers since decentralization also means parents also have to pay a proportion of cost for educating their children. However, it should be noted that the family's education charge would act as a tax and thus it will hinder parents to send their children to schools.

In sum, government and society will benefit from educating young people. However, these benefits will not be generated if government and society cannot partly afford these young people's education. Even if they can, inefficiency in allocating the resources cannot ensure the education will be spread equally amongst people.

Although increase in efficiency of public spending on education can make existing funds more productive, resource allocation amongst household members is considered one of the most important determinants of a child's education attainment. In the next section, how household decides about their children's education will be further studied.

III - Education investment theory.

Education is widely considered both a consumption good and an investment good. Parents educate their children because they enjoy having literate and well-educated children. On the other hand, parents also invest in their children to ensure that their offspring will support them later in life. In the next section, the standard models of parental and household decision-making regarding children's education will be studied.

III.1. Theoretical Model of educational investment.

In this model, parents are viewed as the principle while their children will be considered as agents. Parents will determine the educational level of their children will obtain. To begin, the "unitary" household model will be investigated. In the unitary household, preferences of parents are identical. If their preferences are not the same, the household is assumed to act as if they were maximizing a single utility function (Glick & Sahn, 2000).

Consider a household consisting of a father, a mother, and N children, among them m children are daughters and n children are sons (of course, m+n=N). Parents are considered to live in two periods. In the first period, they work and raise their children. They will retire in the second period. In the first period, household consumption is total income after subtracting a proportion to invest in their children's schooling. Consumption in the second period depends on the remittances of their children's income which, in turn, partly depends on these children's level of education obtained in the first period. In their schooling decisions, parents must trade off their current consumption (reflected by investment in children's schooling) against their future consumption and their children's wealth. The identical preference of parents will be formally represented by a utility function as follow:

U=U(C₁, C₂, Y_d1, ..., Y_dm, Y_s1, ..., Y_sn) (1.1)

In which, C₁, C₂ denotes household (parental) consumption in the first and second period respectively. S_d,i with i = 1...m and S_s,j with j = 1...4, correspondingly denote the education of the i^th daughter and j^th son.  This model can be specified in more detail as follows:

 (1.2)

If the parents do not explicitly prefer one gender-specific child to the others, it is assumed that for any k^th and l^th children  and  when Y_k = Y_l.

As stated above, parents' consumption in the second period will be generated by transfers from their children. It is assumed that the amount of income remitted to parents is proportional to each child's wealth. Children's income in the second period in turns depends on the level of schooling attained in the first period as well as child-specific variables Z such as sex, birth rank, etc. Thus, transfers from child to parents in the second period will vary by child characteristics through their effects both on income and on remittance propensities. Thus, we have a formal second period parental consumption:

          (1.3)

where b_i is the rate of transfer per unit wealth from the i^th daughter and g_j is the rate of transfer per unit wealth from the j^th boy.

Children's wealth depends on their level of schooling attained as well as their specific characteristics and is formally presented as below:

  and             (1.4)

in which b and g are the respective monetary returns to investment in human. The monetary returns to schooling investment also reflect gender of the child.

It is assumed that in the first period, parents spend their total available time in the labor market while the time of children is divided between work and schooling. We have the household's full income constraint as below:

                   (1.5)

The left-hand side represents the total income of household in the first period, while the household's expenditure is on the right hand side. V denotes unearned income. T_m, T_f are total time that mother and father, respectively, work in the labor market. T_i, T_j are the time that the i^th daughter and j^th son devote to income generating activities. W_m, w_f, w_s^*, s_d^*  are their wage rate respectively. P is the direct cost of education which includes tuition fees, books, uniforms, etc. In this thesis, the cost of education is assumed to be identical for all grades and for both sexes.

Total time of children is assumed to be unity so, for all children, working and studying time are competing with each other. Furthermore, for simplicity, the total income of parents and unearned income will be equal to Y i.e. Y = V + T_mw_m + T_fw_f. Thus, we can re-write the full income constraint of household in the first period as:

          (1.6)

or                           (1.7)

The above equation shows that the cost of education consists of  two components : the first  (for ith daughter or for the jth son) is the direct cost and the second is the indirect cost that children have foregone when attending school. 

Parents choose S_di and S_si to maximize their utility subject to the full income constraint and the constraints relating earnings to schooling and parental consumption to child earnings. By substituting equations (1.3), (1.4), (1.6) into equation (1.2), the following maximization problem is derived:

Solving this maximization results in a reduced-form equation that shows the demand for quantity of daughters' and sons' schooling as following:

S_di = S_di(w_m,w_f,V,P,S_m,S_f,Z_di, Z_sj, H) ; and S_sj = S_sj(w_m, w_f, V, P, S_m, S_f, Z_di, Z_si, H)     (1.9)

These equations are functions of the price of education, wage rates of parents, child-specific characteristics of all children (Z_di and Z_sj with i = 1...m, j = 1..n) , unearned income, parents' education and other household and community factors (H).

III.3 Model implication

The maximization problem will be used to deeplier examine how "unitary" parents invest in their children's education. To begin, we will re-present the maximization problem:

If an interior solution is assumed (S_d,i and S_s,j > 0 with i = 1..n and j = 1...n), the first order conditions could be derived. For each daughter i, the first order condition is

  ( 1.11)

For each son j, the first order condition is:

 ( 1.12 )

Thus, parents invest in their children's education to the point that the marginal cost in terms of consumption in the first period equals the marginal benefit tomorrow. These marginal benefits are equal to the marginal utility of second period consumption multiplied by the remittance rate per unit of education (reflected in the children's wealth) plus the utility the parents derive from a marginal increase in the children's wealth, hence children's education.

Turning to the implication of the model for the resource allocation between children, in general, and between genders, in particular. If the market returns to the one child are higher than those to the other children, parents will invest more in that child's education since the utility of parents will increase due to increase in their consumption in the second period and due to higher total wealth of all their children. A similar conclusion could be derived when the rate of remittance of any child is larger than the others, or if parents are more concerned with any child's wealth than with other children's wealth.

If the wage rate of daughters and sons are equal, parents will invest in children's education up to the point at which the marginal benefit of education are the same for every children. Two children k^th and l^th, whose sexes do not matter in this analysis, will be examined. The parents will invest in the education of them to equalize their marginal benefit, i.e.

                   (1.13)

In order to remove the gender identity in this analysis, we re-denote a_k, d_l as rates of remittances, while k and l represent the monetary returns to investment in education of the k^th and l^th children respectively. If k>l, the left hand side of the above equation will be greater than the right hand side when evaluating at the same level of education. Since the marginal benefits to education are a decreasing function of education, the above equation is satisfied at a point at which S_k>S_l. When a_k>d_l or when , the marginal benefits from the k^th child's education is higher than the marginal benefit from the l^th child's education, given that the educational levels of these two children are equal. By a similar argument, investment in the k^th child's education is higher than investment in the l^th child. 

Some assumptions are made before studying the effects of changes in family factors. Firstly, it is assumed that there are two children  for studying, the l^th child and the k^th child. These children are either girls, or boys or boy and girl. Secondly, the monetary returns to education of the l^th child are higher than that to education of the k^th child, i.e. l>k.

Price elasticity will be investigated first. Given that the wage rates for the l^th child and k^th child are unchanged and equal, an increase in the direct costs of education will raise the marginal cost of education investment - that is, the left hand side of first order condition (FOC) equations for these two children. In this case, the assumption that l>k implies that the change in S_k that restores the equilibrium condition in the FOC equation for child k for each level of S_l is larger than the change in S_l that satisfies the l^th child's FOC equation for each level of S_k. That is the demand for education of the child with lower monetary returns to education is more price elastic than that of the child with higher monetary returns. Alternatively stated, when l>k, the marginal benefit from the education of child l decreases faster with S than does that from the k^th child's education. Hence the adjustment necessary to restore equilibrium is less with child l than child k. Thus when the price of education increases, the child with lower monetary returns will suffer from larger reduction in educational investment than the child with higher monetary returns. Similarly, in the case that the rate of remittance in the second period of child l is higher than that of child k, the demand for education of child l is less price elasticity than that of child k.

In the case that the current wage rate of child k increase while that of child l and direct cost of  education are unchanged and l is still higher than k (in addition, keep in mind that  wage rates of these two children are assumed to be identical). The marginal cost of education of child k is now higher than that of child l. In this case, changes in S_k is required to restore the equilibrium condition of the FOC equation for child k while the equilibrium condition of the FOC equation for child l is in comparison with demand for education of that child before increase in current wage rate .

Things become more difficult when (1) the current wage rate of child l increase and that of child k is unchanged, or (2) current wage rates of both children increase but that of child l is higher than that of child k. If the monetary returns to education of child k is still lower than that to education of child l, the demand elasticity with respect to changes in current wage rates will depend partly on how large the differences between the monetary returns to education of child l and child k and partly on how different between the remittance rates of child l and child k. If these differences are large enough to offset the effects of changes in current wage rate, there will be not affect the demand for educational of both children.

Similarly, an increase in the income affects the first order conditions through the marginal cost of educational investment. i.e. the left-hand side of the first order condition equations. This increase reduces the marginal utility of current consumption and therefore lowers the marginal cost. This fall in marginal cost is similar to a reduction in direct cost of education. Therefore, a rise in income leads to a greater increase in the investment in education for children who have lower monetary returns to education. Alternative stated, the demand for education of children with lower monetary returns to education is more income elastic that that of children who have higher monetary returns.

This model also implies that the price elasticity of demand falls with income and that a price elasticity of children with lower monetary returns to education falls faster with income than that of children with higher monetary returns to education. An increase in income lowers the marginal utility of current consumption and therefore, the marginal cost of human capital. This result implies that a rise in prices represents a larger increase in marginal cost at lower levels of income than at higher levels of income. Hence, increased prices will generate larger reductions in educational investment among families with lower incomes than among families with higher incomes.

VI. Chapter Summary

In this chapter, the theoretical framework have laid foundation for analyzing the determinants of dropout behavior in the next chapters. The analysis using micro data will be presented in the last chapter. While the chapter 2, how the government's intervention in the education sector affect children's education in general and children's decision regarding leaving schooling will be investigated.

Chapter 2

Economic reforms and impacts on education trends

In this chapter, we will study: (i) Vietnam's economic reforms and education sector reforms in recent years (ii) and to what extent these changes explain changes in Vietnam's education, in general, and the school drop out trend and grade attainment of children, in particular.

I - Economic Reform.

From the late 1980s, Vietnam began reforming its economy after a long period of poor performance It had adopted a variety of macroeconomic policies to stabilize its economy which led to a strong, broad-based growth.  This economic development is reflected in high GDP growth in many years, increase in government revenue, hence raise in total government investment, and dramatic success in poverty reduction. (See table 2.1)

Table 2.1: Some economic indicators over the period 1992-1997

Source: IMF (1999); NA: Not Available

Although economic growth can lead to substantial reductions in poverty and improvements in living standards, these consequences of growth are not necessarily automatic. In fact, the impact of growth on living standards depends on (1) how the benefits of growth are distributed between people and (2) the extend to which the growth supports social services. In the following section, we will try to explore how economic development contributes to recent trends in the Vietnamese education system.

II - Education System

In this section, the structure of the Vietnamese education system will be examined briefly. Current educational trends, in general, and the school dropout trend and children's grade attainment, in particular, will be explored in this section.

II.1. Vietnam's education system

Together with the historical changes of the country, Vietnam's education system has experienced many changes from the traditional system through the colonial schooling organized by French to the socialist system of the present day.

Since 1945, education activities were broadened over the whole country at every level. Vietnam had established a comprehensive network of educational institutions throughout country and laid the foundation for universal primary education by placing primary schools in every commune. Therefore, it had achieved high levels of literacy and school enrollment relative to its per capita income level (WB, 97).

These efforts to stamp out illiteracy and to increase the educational attainment level were reinforced during the doi moi period (NCSSH, 2001). In 1989, a completely standardized education system was finally crystallized and consists of six levels ranging from voluntary pre-school education including kindergarten and nursery school to college and university level education (Duong, 2001).

Children usually enter primary school at the age of six, finish their primary education by the age of eleven and leave secondary school at 15 years of age. However, many children start school later at age seven, eight or even older and some students are required to repeat grades so it overall requires more than 9 years to finish the basic education.

II.2.  Recent trends in Education

In this section, emerging trends in Vietnamese education system will be studied by exploring core educational indicators such as enrollment, dropout and repetition indicators.

II.2.a. Literacy and Enrollment rates.

Since independence, Vietnam has achieved a great success in improving the education status of nearly all its citizens. The literacy rate increased from only 15% in 1954 (Duong, 2001) to 91% in 1999 (Bhushan et. al., 2001). The literacy rate in recent years, however, shows an uneven distribution between social classes.  In 1992/1993, the distribution of literacy was relatively equitable in the middle of each expenditure quintile. By 1997/1998, the distribution was quite a bit steeper indicating a greater concentration of the illiterate among the lower income population and an increase in inequality (Bhushan et al, 2001).

Table 2.2 Changes in the literacy rate in Vietnam, 1989 to 1999 (%)

Source: Bhushan et. al. 2001

Although a sharp decline in enrollment rate at all levels of education occurred from 1986 to 1991, this trend did not last long. From 1992, the enrollment rates again began to increase at all school levels by all measurements. The gross enrollment rate (GER) of children aged 6-10 increased from 105.6% in 1992/1993 to 114.8 % in 1997/1998 and the net enrollment rate (NER) of this group of children also increased. However, at this age range, the gap in GER between the richest and the poorest was slightly wider. This tendency was confirmed by the NER measurement.(GSO, 1994,1999)

II.2.b. Dropout and grade repetition

Nationally, the dropout rates by grades have reduced year by year. The proportion of children enrolled in grade 1 then dropping out of school reduced from 12.93 % in the 1989-1990 school year to 7.5% in the 1997-1998 school. The dropout rate for children enrolled in other grades also showed a large decline (see table 2.3). Dropout rates by grade in the lower secondary schools are even higher. The dropout rate for grade 6, 7, 8, and 9 in the 1993/1994 school year were 10.4%, 10,7%, 9.3% and 95.6 % respectively.

Table 2.3 : Dropout rate by grade in primary education in the period 1993-1998

 Source:  National Committee for EFA Assessment, 1999

The dropout rate trend also shows a significant regional variation. In the school year of 1994/1995, the Mekong Delta region had a dropout rate of 16% for grade 1, against the national average of 9% and the South East region of 6%. For grade 4, the proportion of children dropping out of school in the Central Highlands was higher than national rate (11.6% vs. 17.4%).  In lower secondary schools, the dropout rates were exceptionally high in the Central Highland. By grade 6, 7 and 8, the dropout rates were 28.8%, 27.6%, 29.5% respectively, much higher than the national average dropout rate at these grades (table 2.4)

Table 2.4 : Dropout rates in basic education in 1994/1995 by region (percent)

Source: Beyond 20/20 (1998)

The dropout rate is much higher for female students. Tam et al. (1998) found that girls accounted for a much larger proportion of dropouts, although the total number of dropouts had decreased.. Cuc’s (1998) findings support this pattern. She finds that girls are more likely to be withdrawn from school than boys are although girls' achievement scores do not lag behind their male friends.

Relatively high dropout rates lead to low survival rate. Nationally in 1995, of 10,000 students entering primary school, only 6,810 would finish the primary level. Of this group of children who finished primary school, 5,482 would enter lower secondary schools, and 3,722 would complete their basic education cycle. Even, in the Central Highlands, only 878 students per 10,000 initial enrolments would finish basic education.

III. Explaining the current education trends from the supply-side perspective

Since doi moi, a variety of laws, acts, and decrees setting the legal framework for education activities in Vietnam have been issued. In this section, how education reforms have affected supply-side factors, which in turn influence educational trends, in particular the school dropout trend, will be studied.

III.1. Financial Provision

Recently, there have been  three main sources to finance the education sector: the Central State budgets, provincial budget and ODA loans. Of which the government contributes most of the funds. Informal sources of funds such as the ODA loans for education and training, commune authorities also contributed a substantial finance for education. These informal sources especially funds from commune authorities, however, were not well reported.

In order to understand how the financial mechanism affects the recent school dropout trend, it is better to examine the equity of state spending for education between educational levels and between provinces and regions.

Although primary level gets the largest share of the state budge for education (33% in 1995), the pattern of expenditures for education favors the rich. In general, the poorest 20% receive just over 10 percent of the total government subsidies (delivered mostly through primary education) while the richest 20% receive more than 35% (delivered through lower, upper secondary and higher education). 

Furthermore, the structure of government spending presents a irrationality in financial allocation. In 1994, 61 percent of the total education sector expenditures were spent on "regular" expenditures while the remaining 39% is left for spending on expansion and improvement of education such as purchase of text books, teaching aids, equipment or school constructions, etc (Duong, 2001). This generally leads a low quality of education, which is jointly  determined by teachers and education inputs. Low quality of education, in turn, causes people to begin to doubt the actual contribution of educated children to family income, thus leads to decision to withdraw their children from schooling system. Unfortunately poor parents seems to have higher degree of suspicion relating to the contribution of education to their total household income. This, together with heavier burden of education, the poor is more likely to withdraw their children from schools.

Differences in resources allocated to provinces and regions may lead to disparity in enrollment and dropout incidence between regions. Government resource for education should spent to areas with lower incomes or highest incidence of poverty to assure the equity of the government spending. Although the South East and the Red River Delta rank first and third on per capita GDP, these regions rank first and second on per capita subsidies going to education and training (all sectors) respectively. Of course, it should be considered that most of universities for which the MoET and other line ministries are financially responsible are located in these two areas. Overall, the pattern of public spending across the sector does not benefit those areas of Vietnam with the lowest incomes and the highest incidence of poverty.

III.2. The emergence of the private sector.

In this section, how policies such as cost recovery and private sector involvement affect the educational trends in recent years are investigated.

III.2.a. Cost recovery.

In an effort to reduce financial burden, since 1989 the government introduced tuition fees with only exemption given to the primary education (Duong, 2001). Generally, households have to pay two types of charge/fee: official tuition fee and informal charge. The official charges are flat and calculated on a per child basis. This form of flat charge represents a proportionally greater burden per child for poor families than rich ones since the disposable income of the poor is less than the rich. The regressive nature of the charges will increase as the amount of the charges increases(Beyond 20/20)..

While official tuition fees are low, the informal charges, set either at the central level or at the local level of school or commune , are  quite high for parents. However, this charges differs among families living in rural and urban areas and among educational levels. In rural areas, the educational expenditures are lower than that in the urban areas, but they are still higher than the official charges.(see tables 2.5).

Table 2.5: Costs of studying in a public primary school and lower secondary school.(VN dong)

Source : Vietnam Social Services Survey, 1996  in World Bank 1996

In addition, the "burden of education" ratio, the ratio of household's expenditure for education to household's total expenditure (WB, 1996) are much heavier for the poor than the rich. Nationally, the burden of education increases with the increase of educational levels. A child going to lower secondary school could consume up to nearly a third of a poor household's yearly total expenditure while only 12% of total expenditures of one household in the fifth quintile flies with that child yearly. Heavy burden of education has deterred many households sending their children to schools especially at the lower secondary education and upper secondary education.

Table 2.6: Quintile specific education burden ratio of sending children to public school (%)

Source : Vietnam social service survey 1996, in World Bank 1996

III.2.b. The emerging role of private sector in provision of education

After doi moi, the government has allowed the presence of private-founded schools. Currently, there are several types of non public schools: semi-public schools, “people-founded" schools and purely private schools.

However, in primary and lower secondary education level, private schools enroll only a limited number of students (respectively, 0.2% and 0.8% of total primary students and lower secondary students), most of them are from rich households. Furthermore, most of these primary and lower secondary private schools are located in urban areas. The minor role of private schools in providing primary and lower secondary education especially in rural areas could not release the burden of providing educational services of public schools.

IV. Quality of education and community's and families' educational values

In this section, issues related to quality of education and  parents' views on value of education will be investigated.

IV.1 Quality of education.

Given that the ability of each student are the same, factors affecting the quality of education can be categorized in two groups: the first group includes factors facilitating the learning process of student such as the number of teachers, school facility, etc while the second group relates to teachers' quality.

IV.1.1 Curriculum

The total amount of time a primary children spent in school are still low. A primary cycle of 5 years in Vietnam's education system is much lower than that of other developing countries with a primary cycles of six or seven years. In addition, school year of thirty-three weeks is very short by international standards. Moreover, just only 20% of children in Vietnam attend schools in a full day instruction of  5 to 6 hours . This may result in fewer knowledge a child can gain from a primary school.

The pedagogy is now emerged as an alarming problem. Much of the curriculum have been unchanged for the past 22 years (Oxfam UK/I, 1997. Primary teachers agreed that Grade 1 is too easy, grade 4 is too difficult while the basic language skills are not emphasized especially in minority areas (Swansea, 1998). Teachers also claimed that vocabulary is not appropriate to the level of student ability (Oxfam UK/I, 1997).

In addition, teaching strategies seems to be an problem. They are inflexible and teacher-centered. Teachers focused on finishing the lessons as scheduled. In the meantime, teacher-centered approach hinders the development of the analytical  skills of children and the development of a capacity for creative thought.

All these issues may lead to low quality of education. Only 8.9 percent of dropouts agree that lessons learnt at school in the past were useful to their everyday life, while 44.6 percent said that it was unidentified about the usefulness of lessons learned at school, and the remaining expressed that the lessons learned at schools were totally useless (Tam et al, 1998). In general, current curriculum is unsatisfactory and inadequate to provide students with knowledge and skills needed and the quality of education remains a serious problems to be solved (Duong, 2001).

IV.1.2. Facilities and resources for education

Low quality of education can also be attributed to (1) poor school infrastructure, (2) lack of teaching aids and equipment. Many schools lack basic facilities for education such as library, playground for students, even toilet, sanitary facilities, etc. (Tam et al, 1998.  In addition, after introducing doi moi, provinces, districts and communes are themselves responsible for fundamental spending for school infrastructure. This leads to differentials amongst provinces, regions and schools. However, generally, the money generated is often lagged behind the actual needs (Duong, 2001). 

Lack of teaching and learning aids also is a serious problem. Some teachers mention that the lack of equipment as a constraint to improving their teaching (Oxfam UK/I, 1997) Most of them mention simple, inexpensive items as the ones that they need most - pictures, maps, simple toys and games, more books, etc. The lack of teaching and learning materials makes the lesson more boring and more difficult to understand for children. These, in turn, may discourage the learning of children. Consequently, the possibility of dropout will increase.

IV.1.3. Quality of teaching.

Concerns about the quality of teaching have been raised recently. Many parents dissatisfy with what their children were learning in schools and the behavior of teachers. (CECI & RDSC, 2001). Many pupils complained about the characteristics and behavior of teachers. Teachers have low qualifications and most of them are not interested in teaching. Besides, many teachers teach without concern for whether or not their pupils understand the lessons.

Generally the teaching quality of teacher directly depends on (1) knowledge and skills that they obtained and (2) teaching incentive system. It has been increasingly realized that the knowledge and skills that teachers received in both pre- and in-service training is not adequate to ensure their teaching quality. The number of teachers with a lower secondary education certificate and 3 years of teaching training is much higher than the number of teachers with a upper secondary education and 2 years of teaching training (Oxfam GB, 2000)

The quality of teachers is even more serious if the question of the quality of those who enroll in teaching training schools and colleges is raised. Special characteristics of children aged  6- 11 require primary education teachers to be very qualified. However, due to the lack of enough incentives, teacher training schools and colleges throughout the country received only enrolled students with low learning capabilities comparing to those enrolled in other colleges and universities (Duong, 2001). In addition, time for teaching practice of these young teachers is not sufficient. Most of them teach only 2-3 lessons before graduating as a teacher (Swansea, 1998).

Quality of teacher even worsens when the teaching incentive is low. The problem posed by insufficient salary remains chronic over time (Duong, 2001). On average, a young teacher in rural areas can earn from VND 4,240,000 to 4,570,000 per year. Some teachers have to get support from their parents (Oxfam GB, 2000). These underpaid teachers have to divert their efforts to make extra earnings to supplement for their living needs (Duong, 2001). Consequently, there is a strong impetus for teachers to shift their focus from education to income generation. In the long term, their teaching skills, therefore, will inevitably be degraded.

In an ideal class, teacher acts as a facilitator in order to make the classroom to be a place where children can and want to learn. This not only means that children have enough basic supplies but also means that teachers should have relevant skills and be motivated to use them. Once Vietnamese parents feel that their children can be valuable persons from learning in schools, they will try their best to make their children stay in school.

IV.2. Other exogenous factors leading to dropout of children.

Factors including (1) production characteristics; (2) decline in the value of knowledge in parents' views and (3) attitudes of people in the community to dropouts strongly influence the decisions of whether or not to withdraw them from schools

In rural areas, doi moi does truly have very little impact on the production technologies which are mostly in the manual form. These technologies require labors who have practical experiences and/or have expertise which is transferred generations by generations. This leads to a perception that "agricultural production does not require a high education".

In addition, many people's perspectives on education are that education is only necessary for persons who intend to work in the industrial sector and who want to escape from the village. Thus, having high education level becomes less meaning to families and dropout becomes an ideal option. Furthermore, people are witnessing unemployed situation of many persons with high education. This fact discourages many parents to continue to send their children to higher education level after finishing a certain level of education.

Besides families, community is very important factor affecting motives, goals and results of a child's studying (Tam et al., 1998). Tam et al (1998) finds surprisingly that 31,5 percent of teachers, 75 percent of people living in three communes in the Red River Delta, and 69.3 percent of authorities in these communes stay neutral towards girls' dropout.

These factors make children think that education is not necessary in the current rural production and it is completely indifferent for them to go to school.

V. Chapter summary.

This chapter provides explanations for schooling dropouts from the supply-side perspective such as government policies for education including budget, educational policies; school characteristics as well as quality of education and social stigmas. However, the dropout decision is in the household's hand. What leads household to let their children drop out or even to withdraw their children from schools would be studied careful in the following chapters.

Chapter 3

Determinants of dropout of school and grade attainment of

children under 16 in rural Vietnam

This chapter will employ the model presented in the previous chapter to analyze how Vietnamese households make their decisions regarding school dropout and the highest grade attained of their children.

I - Empirical Approaches

I.1 Dropping out of school

The subjects for studying determinants of dropping out are 14-16 years old children in 1997/1998. These children must take part in both VLSSs and they have been enrolled at the years of 1992/1993 (it is the years that the first VLSS was conducted)

The determinants of schooling drop out are estimated using probit regression. The dependent variable will take the value of 1 if the subject reported that he/she are no longer enrolled in school while it will be equal to zero if the subject reported that he/she still enrolled in the school in 1997/1998. Using this sample allows investigating the effects of households' shocks on the process of making decisions of household regarding to educational investment. With this sample, it is also possible to assess the effects of dynamic factors such as the presence of siblings under 5 years, i.e. the number of brothers and sisters born during the last 5 years.

I.2. Grade attainment

The determinants of grade attained will be estimated on the sample of children aged 9-16. This grade attainment is an indicator of the cumulative investment in an individual's education. Because it represents the outcome of a series of ordered discrete choices -whether to go on to the next grade or withdraw from school and because the distribution of years of schooling is usually not normal around the mean, so OLS method is not appropriate to estimate the determinants of grade attainment but the ordered probit regression. The ordered probit model treats grade attainment as the outcome of ordered discrete choices and does not assume a normal distribution for the dependent variable.

II. Model Specifications

II.1 Dependent variables.

For the dropout model, the dependent variable (denoted as DROPOUT) is a binary variable which equals 1 if the observed child aged 14-16 have dropped out of school in the last 5 years and sets to zero otherwise, both conditional on enrollment at the beginning period.

For the grade attainment model, the value of the dependent variable is equal to the highest completed grade at the survey time. In this case, the dependent variable will equal zero if the child has not enrolled in school.

II.2 Independent Variables.

II.2.a. Child Characteristics.

·          Sex:  Child's gender will determine his/her opportunity cost of education. This is a binary variable, which take value of 1 if the child is male and of 0 if the child is female.

·          Age: Age determines the opportunity cost of a child.  In addition, in the analysis of grade attainment, age of a child will partly determine his/her current grade attainment.

·          Age squared: Child's age variable sometimes does not reflect the true relationship between children's age and children's grade attainment.

·          Presence of older siblings who have dropped out of schools or have never enrolled in the schools (still living in the household): The presence of older sisters will assumedly improve the continuation of going to school of girls since domestic works are now shared.

·          Presence of older siblings who are still enrolled in schools (still living in the household). This variable also is a binary one.

·          Presence of younger students: This variable has two outcomes, which take value of 1 for presence of younger sisters and of 0 otherwise.

·          Presence of younger siblings who are not enrolled in school at the time interviewed also negatively affects the chance to promote into higher grade of girls given the parents' different values on boys and girls.

II.3.2.b. Parents' characteristics

·       Parental Education: Mother's education and father's education will be studied separately since it is widely accepted that same-sex effect of parental education is much stronger than cross-sex effect.

·       Head's sex: Female-headed household is negatively affect the educational outcomes

II.3.2.c. Household Characteristics

·       Residence: in which region this household resides. Regions affect the opportunity cost of education.

·       Presence of children under 5 in the household. This variable will capture the dynamic characteristics of these two model.

·       Economic status of household: Resources that parents devote to their children should be viewed in a dynamic context. Four binary variables describe the dynamic characteristics of household's economy:

*                Households which belong to POOR group in both VLSSs

*                The value of the MOVEOUT variable will take the value of 1 if a household which was in poor group in 1992/1993, but is no longer in the poor group in 1997/1998.

*                If a household was in the non-poor group in VLSS 1992/1993, but is in the poor group in 1997/1998, it is called to fall into poverty.

*                The value of the NOSHOCK variable will be 1 if a household does not experience any big economic shocks that decreased their economy during last five years.

Since a household is categorized in a specific quintile based on its per head expenditure, thus changes in expenditure can proxy for other dynamic characteristics.

For analysis of grade attainment, the logs of real expenditure per capita will proxy for economic situation of the household.

III. Descriptive analysis

III.1 Dropout model.

There are 951 observations aged 14-16 in 1997/1998 in this database (489 observations are boy and 462 observation are girl). These children were 9-11 years old in 1992/1993 and were enrolled in schools in 1992/1993. These children are living in the rural areas in 1997/1998. If a family has more than a child in this studied age range, one of them will be selected randomly while the others will be dropped from the sample. Of these 951 children, 328, accounting for 34.5% of all children, dropped out of school in the last five years. The dropout rate for girls is higher than that for boys, i.e. 40% for girls and 29.24% for boys.

Whether or not a child involves in economic activities also play an important role in deciding his/her schooling continuation. Nearly 68% of girls who participate in economic activities drop out of school in the last five-year while this figure is 58% for boys (see table 3.1). On average an enrolled girl work 12.66 hours per week while a schooling boy works 11.09 hours per week.

Table 3.1: Child working and dropout in the last five years.

Note: Children are considered to participate in economic activities if working more than 3 hours a day.

Source: Author's calculation based on VLSS 97/98

Parental education is strongly correlated with children's decision regarding dropout of schooling. Only 10% of children whose mothers graduated from upper secondary schools dropped out of schools while 52% of children whose mothers did not have any education would be likely to drop out. This trait is also true for father's education. 48% children of fathers who did not have any education dropped out of school. (see table 3.2)

Table 3.2 : Parental education and children's dropout. (%)

Source: Author's calculation based on VLSS 97/98

Living in long run poverty is one of the main reasons for dropping out of school. 46% children who live in poverty in both two rounds of VLSS 1992/1993 and VLSS 97/98 dropped out while only 8.7 % of children of the rich household in both round of Vietnam Living standard survey are dropouts. Children whose families' living standard deteriorated have higher opportunity of dropout (See table 3.3). The gender gap is strongly reflected by this table.

Table 3.3: Economically dynamic characteristics and dropping out of schools.(%)

Source: Author's calculation based on VLSS 92/93 and VLSS 97/98.

The above descriptive analysis has displayed an overall picture of dropout issues during the economic development time. Before turning to a more comprehensive analysis on this issue, an descriptive analysis of factors affecting the grade attainment  will be studied.

III.2 Grade Attainment Model

The dataset for this analysis contains 2196 observations who are children aged 9 - 16 in 1997/1998. Of which 1054 observations are female children. All of these children live in rural areas. Each one lives with one or two biological parents. There are some cases in which a household has more than a child in the studied age range, only a child is selected randomly as in the dropout model.

A majority of children (60%) started their school at the legitimate age for schooling enrollment. Some of them started school at 5. Amongst late enrollees, 60% started school at 7. And the remaining 15% of total children started school at the age of 8 and over.

As children become older, they advance through school at slower pace. Given that a child starts school at 6 years old, in the first three or four years of school, he/she can complete 3.48 grades. However, in the next 5 years, an average child could only finish 3.67 more grades (mean grade attained of a child by the age of 15-16 is 7.48).

Child’s gender still matters in school access and completion. It is clear from the table that by the age of 13-14, the mean grade attained by girls is reduced in comparison with boys although girls’ mean grade repeat and the proportion of repeaters amongst girls is much lower than those of boys.

Household standard of living positively relates to child's grade attainment and his/her pace at which he advance through school A 9 or 10 year-old child from a rich household can complete grade 4 while those who come from poor household could only complete grade 3. At the age of 15 - 16, children from rich families can attain grade 8 or more while those from poor families could only finish primary school or obtain very few lower secondary education.

Maternal and paternal education are crucial factors in a child’s education. The mean grade attained by a child with poorly educated mother and/or father is  3.75 which is much lower than the 6.3 grades that an average child whose parents have at least lower  secondary certificate.

Disparity in grade attainment amongst children living in different regions is also worthy of consideration. On average, the children living in the Mekong Delta or in the Central Highlands complete grade 5 or grade 4 while those living in the Red River Delta could finish 6.6 grades.

The above analyses have shown that child's gender, child's age, his/her working participation, his/her family's income, maternal and paternal education level and region where he/she lives play major roles in determining whether she/he would withdraw from educational system in the last five years and in determining grade that a child can accomplish. In the next section, by using multivariable analysis, determinants of schooling dropout and highest-grade attainment will be further discussed.

IV. Results of regression models.

IV.1 Dropout model.

The results of the probit estimation for determinants of dropout of school amongst children aged 14-16, who took part in both Vietnam Living Standard Survey, are presented in the table 3.4. The column 2 shows regression results of probit estimation using pooled data. The results of probit estimation of boys are shown in the next three columns and the remaining columns present results of probit estimation for girls. In addition, for ease of interpretation, predicted probability of schooling dropout will be separately calculated while holding other variables in the model constant at their means.

Child's sex is positively associated with dropout although this relationship is not statistically significant. Being a girl will increase the likelihood to drop out of school by 16.9 percentage point. Predicted probability of schooling dropout  of a girl is 0.402 while that of a boys is only 0.287. Some explanations for this gender inequality in accessing to the formal education include (i) the differences in economic opportunities opened to boys and girls, hence differences in the values of their working time in the labor market; (ii) the possibly higher returns to education of boys when they both reach the adult age; (iii) likely overburdened housework of girls; (iv)parental preferences differ from boys to girls.

Child's age also plays an important role in household's decisions regarding education investment. As age increases by one year, the likelihood of dropout increases by 7.64 percentage points. As child's age increase, their opportunity cost of  education will also raise. In addition, the children's school performance is worse as the child's age raise.

The importance of increasing age regarding schooling dropout is different for girls and boys aged 14-16. The likelihood of girls rises by 8.6 percentage points while that of boys increases by 7.1 percentage points if their ages increase by a year. This implies that girls are more likely to drop out of school  than boys if they both reach these ages.

Child's working status also statistically significantly determines the decision of dropping. 39.35% of the children in the study sample works more than 21 hours per week. The likelihood of dropping out of school increase to  52 percentage points for a child labor. If holding other variables constant at their means, predicted probability of dropping out of a child labor is 0.63 while that of a non-working child is only 0.16. 

And as child's age, female child labors also face a higher risk to drop out in comparison with boys. The likelihood of dropout of a male child labor is 47 percentage points while that of a female child labor is 60.5 percentage points. The predicted probabilities of dropping out of a working boy and a working girl, which are calculated when all variables in the two gender-disaggregated models are constant at their sample means, are 0.56 and 0.7 respectively. This evidence further confirms that girls' opportunity costs of time are much higher than boys.

Table 3.4 - Marginal effects

Source: Author's calculation based on VLSS92/93 and 97/98

Increases in father's education also reduce the probability of dropout of their children. The probability of dropout reduces from 0.515 for children of  no-education fathers to 0.134 for children of upper-secondary-graduate fathers. And increases in education level of fathers also lead to reduction of dropout probability gap between girls and boys. Paternal education has a larger impacts on girls than on boy regarding dropout decision.

Table 3.5: The predicted probability of dropout by maternal and paternal education.

Source: Author's calculation based on VLSS VLSS92/93 & 97/98

The probability of dropout depends much on the previous and current household's living standard. For children whose household's living standard in 1992/1993 was categorized as a poor, the predicted probability of dropping out is high at 0.47 if his household's living standard did not improve in the last five years, while the probability of dropping out reduces to 0.32 for those whose household has escaped poverty in the last five years. The predicted probability of dropout is 0.41 for children whose household experienced an economic shocks which made them become poor during 1992/1993 - 1997/1998. Children who live in household with a stable wealth, i.e. they did not suffered from any special shocks that caused them to be categorized as a poor household in 1997/1998, have lowest probability of dropping out. Their predicted probability of dropout is only 0.24. However, not all the effects of changes in household's living standard are statically significant.

Table 3.6 - Predicted probability of dropout by economic changes

Note: Calculated at the means of other variables in these models

Source: Author's calculation based upon merging data from VLSS92/93 - 97/98

Changes in household's living standards affects boys and girls differently. Under any circumstances, boys are clearly more advantage than their female friends. Boys' predicted probability of dropout is low at 0.198 but that of girls is 0.292. From the gender disaggregated analyses, the likelihood of dropout of boy is 2.6 percentage points if their households had experienced a chronic poverty while that of girls from household with the similar characteristics in terms of living standard is 15.7 percentage points. This can be examined using pooled data with an interaction variable. It is not surprised that the largest gap in predicted probability of dropout between boys and girls happened amongst households with chronic poverty. The gender-gap in this case is large at 0.144 while the gender gap is only 0.1 in household with quite stable wealth. Benefits of escaping poverty also favored boys with the probability of dropout being 0.279 while that of girls is 0.361. 

The strong relationship between household' living standard and dropout confirms the increasing importance of family resources for children.

The presence of other children in the household. The likelihood of dropout increases by 4.4 percentage points if the number of the biological siblings under five living in the household rises by one. The effects of the increasing number of siblings under five are statistically significant at 1%. However, the significance of this effect differs from boys to girls. Girls are more likely to drop out of school than boys if the numbers of children under five increases. This gender-specific impact supports the view that girls generally take more responsible for housework than boys which, in turns, will induce them to drop out whenever the work burden is so heavy.

The effects of the presence of other siblings aged 6 and higher on children's education are mixed and very different for boys and girls. The presence of a currently-enrolled sib, who may older or younger, reduce the children's probability of dropping out. However, the impact is reserved if a child has at least a sib who have dropped out or have never enrolled. Having at least a younger sib who dropped out or never enroll will increase the likelihood of dropping out by 3.94 percentage points and having at least an older sib who dropped out or never enrolled increases the likelihood of dropout by 5.43 percentage points.

The impacts of the presence of sibs aged over 6 also differ largely from boys to girls. The likelihood of dropout of boys reduced under any circumstances regarding the sibling compositions and whether these sibs have ever enrolled or dropped out while that of girls follows the patterns that found out when analyzing the pooled data.

The results suggest that there may exist a  so-called "sib pressure". Living in a household having more "learning environment" may induce children to continue their schooling. In addition, going-school siblings can help each other learning process. It also could not exclude the roles of educated parents in creating a "learning environment" in their households.

Regional effect. Infrastructure an economic development, availability and adequacy of school facilities, attitude of people about children's education, production technologies, regional vulnerability to natural disaster, etc may affect a child's and his household's choice between schooling continuation or dropout. The predicted probability of dropout of children living in different regions of the country is presented in the table below.

Table 3.7. Predicted probability of dropping out by regions and by child's gender.

Note: calculated at the mean of other variables in the 3 models.

Sources: Author's calculation based on merging data from VLSS92/93 and 97/98.

Household head. After controlling for all other variables including household's economic status, and child's contribute, etc., children living in female head households is less likely to drop out of school. It is irony that the relative benefit from living in a household headed by women is lower than for girls. Within economic status categories, female household heads are more likely to invest resources, including time, money, and emotional support in facilitating the education of children living in their household.

Highest level of education. The highest level of education on children's dropout strongly affects the decision of dropout. If the children age 14-16 have yet finished primary education, the likelihood of dropout is 41.22 percentage points while that of  those who have finished primary education or have some lower secondary education is only 3.27 percentage points. The effects of having not finished primary is significant at 1%.

IV.2. Grade Attainment Model.

It is quite clear from the descriptive analysis presented in the previous section that the grade attainment of boys and the girls are not much different and that girls and boys expose to risk factors differently.

For both girls and boys, there exists a positive relationship between children's age and children's grade attained although these relationships is quite different from boys to girls. The relationship between child's age and child's grade attainment, however, is not a linear relationship. The grade attainment increases with age and decrease with its square, suggesting an inverted-U shape relationship. As child's age increase to a certain point, the average mean grade attainment that a child can achieved will be reduced. The higher coefficient of the age square of girls suggest that the cut-point occurred to girls is earlier than boys. This means as the a girl ages, their likelihood to have a higher grade attainment will be reduced more than that of a boy.

Table 3.8 : Results of the grade attainment regression

Source: Author's calculation based on VLSS 1997/1998

Regarding household composition factors including the educational status of the sibs, birth rank, the presence of the under five years old children in the household, the effects of these factors on child's grade attainment also differ depending on his/her sex.

If the child has sibs who are currently enrolled in school, his grade attainment will likely to be higher than that of children who have not any  currently-enrolled sibs while the likelihood of achieving a high grade is negatively affected if the child has sibs who have never enrolled in school or has dropped out of school. These results are compatible with the results gained from the analysis of dropping out presented above. They also are compatible with the result of Nielsen(1998) which states that "ceteris paribus, the presence of older school attending siblings increases the probability of school attendance". This is likely to reflect that households have different traditions and norms regarding the use of education. (Nielsen, 1998).

The higher birth rank that the child is, the lower grade she/he would achieve. The effect of birth rank differs from boys to girls with the bias against the girls whose higher birth ranks will lead to much lower likelihood to attain a high grade than that of boys.

The more years a child repeats, the more likely is that this child will achieve a low grade. It is interesting that if the effect of the number of years that a girl has to repeat on her grade attainment is much lower than that on the boys' grade attainment.

Working has a negative effect on child's grade attainment but the effect on the boys' grade attainment is lower that on the girls' grade attainment.

Mother's education and father's education also play vital roles in determining the highest grade that their children could attain. However, the effect pattern is not as similar as that in the dropout model. The effect of mothers' education on children's highest grade attainment is the same for both boys and girls. In addition, this effect is as strong as the effect of father's education on the girls' highest grade attainment. The effect of the mothers' education, however, is stronger than the effect of fathers' education on boys' highest grade attainment.

Region where the child resides also affects his/her highest grade attained. If the child lives in the Mekong Delta, the highest grade attained will be more likely lower than that of children living in other regions. If he/se lives in the Red River Delta, she/he will have chance to get higher grade than their counterpart friends in other regions. And it is not surprising that girls are always disadvantage than their male friends in grade attainments given that they are living within a region.

The ethnic of a child will, somewhat, have impacts on child's grade attainment. If the child's ethnic is Kinh, their likelihood of attaining a higher grade is much large than that of children from other ethnic. The household's economic status has a positive impact on the children's grade attainment . The boys also are bias against girls when the household's expenditure per capita increases.

V. Chapter summary

The above analyses have presented a comprehensive picture about which factors cause children to drop out of school during 1992/1993 and 1997/1998. These determinants include household economic situations during this period measured by changes in expenditure quintiles; parental education level; regions where the child resides;   household's learning environment including parental education and the sibs' learning process. In addition, child's attributes also play decisive roles such as age, genders. The analysis also reveals that these determinants affect boys and girls differently. Some factors biases girls but some other bias against boys. However, most of determinants are favorable for boys than for girls, for example the presence of infants (child under five years old) in the household or even effects of household's "learning environment" . The gender gaps is still an issue.

Chapter IV

Conclusion and Recommendation

I - Conclusion

There are many factors, both at the micro level and at the meso and macro levels, contributing to the household and children's decision of whether or not to drop out of school and at which grade these children will leave school.

At the meso and macro level:

-          In terms of financial decentralization, cost recovery policy has placed a heavy burden on the poor household. This policy acted as a tax on human capital accumulation of the poor.

-          Allocation of the education expenditure has shown a pro-rich pattern.

-          Mis-allocation of the resources for education also leads to low quality of education which are reflected by the fact that teachers show an undermined enthusiasm in teaching.

-          Lack of teaching aids and equipment makes education less efficient..

-          Diversification in the provision of education services does not benefit all the children

At the household levels

-          Long run poor indeed affects the parents' decision regarding leaving school of their children..

-          Parents education always plays major roles in children's outcomes. The higher educational level parents have, the less likely a child drops out of school.

-          Differences in regional-specific characteristics also plays a role in children and household's decisions of dropping out.

-          "Good" household's environment is another determinant of retention in school. Living in a household with a "learning environment" encourages children's staying in the schools. Or in other words, sibs' learning process has positive impacts on child's learning process.

-          Labor participation plays an important role in determining the children's dropout although participation in the labor force may be the consequence of the dropout not the cause of dropout. 

-          Higher age also make a child more likely to be withdrawal from schools..

-          Boys are less likely to drop out than girls in nearly all circumstances such as the presence of children under five in the households; the economic downturn of the household.

-          Concerning the determinant of highest grade attainments, factors that affect the children's decision to leave school also have the same effects on the grade attainment of the children.

-          Similar as dropout decision, factors determining the highest grade attainment affect boys and girls differently. Boys seem to get higher priority in schooling advancement than girls although there is little evidence to support the idea that boys' performance are better than girls' at primary and lower secondary schools.

II. Policy recommendation

From these findings, the following tentative policy implication can be drawn.

1-       Making public expenditure more equitable. This comprises several tasks including rationalizing resource allocation between education levels and between sub-sectors; better targeting geographically.

2-       The Government should be prepared to deal with the consequences of shocks suffered by the poor and ensure that the temporary shocks do not put them into poverty and thus lead to long-term reductions in educational investment to their children. To do so, Government should intervene to improve the income generation potential of the household through the creation of income-generating activities and through the provision of productive assets to create a more stable economic base. In addition, developing safety nets against short-term shocks will help reduce the irreversible harm they can cause to investment in education of household..

III. Suggestion for further research

A wide range of issues can be better analyzed. How differently household's decision regarding withdrawing their children affect boys and girls living in a same household or how price fluctuations affect the dropout decision of children could be studied. Integrating macroeconomic variables into this panel data could give a broader picture on how economic change affects the household's decisions. All these issues are suggestions for future research.

Bibliography

Al-Samarria, S. & T. Peasgood (...) Educational attainments and household characteristics in Tanzania, Working paper No 49, Institutes of Development Studies, University of Sussex, UK

Bauer, T. K, I. Gang (2000), Sibling Rivalry in Education Attainment: The German Case, Discussion Paper No. 180, The Institute for the Study of Labor (Bonn)

Behrman, J. R, J. C. Knowles (1999), Household Income and Child Schooling in Vietnam, The World Bank Economic Review, Vol 13, No. 2, 211-36, World Bank, Washington D.C.

Berhman, J. et al. (1999). Schooling Investments and aggregate conditions: a household-survey-based approach for latin America and the Caribbean. Working paper No....., Inter-America Development Bank, New York, the US

Bhushan, I., et al. (2001). Human capital of the poor in Vietnam, Asia Development Bank, Manina, Philippines.

Dreze, J. & G.G. Kingdon (1999). School participation in rural india, Review of development economics, Vol. No.

Duong, Le Bach, (2001). Education and Human development in Vietnam, Background paper for the National Human Development Report (2001) published by National Center for Social Science and Humanities. Unpublished

Ejrnes M., C. Portner (2000) Birth Order and the Intrahousehold Allocation of Time and Education (?)

Foster, A., M. Rosenzweig (2000), Household Division, Risk and Rural Economic Growth, The Center for Research on Economic Development and Policy Reform Working Paper No(..) Stanford University.

Friedman, M. (1955). The role of government in education. Available at. http://www.schoolchoices.org

Gard, A., J. Morduch (199?), Sibling Rivalry, Resource Constraints and the Health of Children, HIID working paper No…., Harvard Institute of International Development.

Garg,A. , J. Morduch (1998), Sibling Rivalry, Discussion Paper No 630, HIID, Harvard University.

Glick, P., D. E. Sahn (2000), Schooling of girls and boys in a West Africa countriy : the effects of parental education, income, and household structure, Economics of Education Review 19(2000), 63-87

Haddad et all, eds, (1997), Intrahousehold Resourse allocation in developing countries: models, methods, and Policy. Baltimore : The John Hopkins University Press Published for the International Food Policy Research Institute.

Hanushek, E. & J. Luque (2002). Efficiency and equity in schools around the world, NBER Working Paper No. 8949, The National Bureau for Economic Research, The US.

Hanushek, E. (1995) Interpreting recent research on schooling in developing countries, The WB Research Observer, Vol. 10. No. 2, August 1995

Haveman, R. & B. Wolfe (1995). The determinants of children's attainments: a review of methods and findings, Journal of Economic Literature, Vol. XXXIII, Dec 1995, pp 1829-1878

Holmes, J. (1999), Measuring the determinants of scholing completion in Pakistan: Analysis of censuring and selection bias., Economic Growth Center Working Paper No 794, Yale University.

Hoxby, C. M. (1996). Are Efficiency and equity in School finance substitutes or complements. Journal of Economic Perspectives, Vol 10. No 4 - 1996, pp 51-72

Leclercq, F. (2001). Patterns and Determinants of elementary school enrollment in rural North India, Working paper, TEAM-CNRS, Universite de Paris 1, Paris, France

Leclerq, F. (2001b), Child work, schooling and Household resources in Rural North India, TEAM-CNRS, Universite de Paris I (Sorbonne), France

Llord, C. B, A. Blank(1996). Children's Schooling in Sub-Saharan Africa: the role of fathers, mothers, and others. Population and Development Review. Vol. 22, No. 2, June 1996, 265-298

Nielsen, H.S. (1998). Low demand for primary education: traditions or economic incentives?, Working paper 98-05, Centre for Labour Market and Social Research, Department of Economics, University of Aarhus, Denmark.

Prescott, N. (1997). Poverty, Social services, and Safety Nets in Vietnam. WB Discussion Paper No. 376, The World Bank.

Son, Vo Thanh et al. (2001) School Enrollments and Drop-outs. in Living Standards during an Economic Boom ed. by D. Haughton et al., Published by Statistical Publishing House,

Strauss, J. and D. Thomas (1995), Human Resources: Empirical Modeling of Household and Family Decisions, chapter 34 in Jere Behrman and T.N. Srinivasan, eds., Handbook of Development Economics volume 3A.

Strauss, J., D. Thomas (1995), Human Resource : Empirical Modelling of Household and Family Decisions, in Berhman J. and Srinivasan eds, Handbook of Development Economics, Vol III. Amsterdam : Elsevier Science B.V.

Tansel, A. ( 1998), Determinants of school attainment of boys and girls in Turkey, Economic Growth Working paper No 789, Economic growth center, Yale University.

Taylor, L. (1999). Government's role in primary and Secondary Education, Economic Review- first quarter 1999, Federal Reserve Bank of Dallas, US

UN Vietnam (1998), Expanding choices for the rural poor - Human Development in Vietnam, Hanoi

UNICEF (1995), Education is a right. Available at http://www.unicef.org