Introduction

I. Research topic

One of the greatest concerns to developmental economists and economic policy makers is the interaction between population growth and economic development. Both have consequences for the natural world, the political order, and human wellbeing. The key element behind the change in population, particularly in a country like Vietnam where the growth is so rapid, is the level and pattern of fertility. Although mortality and migration also contribute to the size, structure and growth of population, and are hence important areas of study in their own right, population dynamics are strongly molded by fertility. Fertility is also important because it is inextricably bound up with many aspects of the economic and social milieu. A better understanding of fertility behavior may provide insight into a wide range of social and economic, patterns and changes such as labor force participation, income distribution and education aspirations for children.

As we know that while the economic growth of Vietnam is still at low level, Vietnam's population continues to growth rapidly. This population growth is mainly as a result at the fertility level. Yet the fertility has declined continuously, to probably more than two children per women in 1999, though substantial differences remain between rural and urban areas and among regions (see Nguyen Huu Minh, 2000).  Fertility level in Vietnam has changed fast in recent years. Based on data sets of census population in 1960, 1979, 1989, Dr Dang Thu calculates that the total fertility rate (TFR) in the North was 5.47 in the period 1946-50 and 5.36 in the period 1950-54. TFR increased rapidly during 1955-60 by 6.8, during the following ten years it was steady and then tended to decline gradually. TFR of the whole country is 5.78 in 1970-74, declining by 4.17 in 1985-1989. According to the final preliminary report of census population (see Chu Thi Loan,2000), the population of Vietnam was 76,327,919 persons in 1999. An increase of 11.9 million people in 10 years. Currently  Vietnam's population is equivalent to the population of a middle country. Average increasing population rate in the period 1989-1999 was 1.7%. It fell to 0.5% (compared with 1989). In 1998, the total fertility rate of the whole country was 2.5 children per woman at child-bearing age. It fell to 35% compared with the ten past years. However, there are many differences among the regions. For example, the fertility rate of ethnic minorities in remote areas is still at high levels. High fertility rates in minority areas has proved to be great obstacle to the implementation of poverty alleviation programs, socio-economic development and improvement of living conditions.

Although important, there have been very few researches on determinants of fertility in Vietnam. People just want to mention it but do not really focus on explaining it. Therefore, in the thesis, I will concentrate on pointing out the major socio-economic determinants of fertility in Vietnam. By this study, I hope to make some contribution to policy maker on the family planning program of reducing fertility and promoting economic development.

II. Focus and scope of the thesis

1. Focus of the Thesis

The research will focus on

· Reviewing theoretical and empirical issues concerning the fertility definitions, fertility measurement and factors affecting fertility level.

· Analyzing the fertility phenomenon of Vietnam and major determinants of fertility level based on data obtained from Vietnam Living Standard Survey (1997/1998).

· Recommending policies to reduce the fertility level in Vietnam.

2. Scope of the Thesis

This thesis, will review theoretical and empirical issues relating to factors affecting fertility. These will be analyzed in socio-economic terms that relate to the mother's characteristics and this 'social' aspect implies education, and culture. In addition, the fertility differences between rural and urban, among regions of the whole country will be analyzed in this study. Some other differences such as socio-economic characteristic of father, biology, migration, some proximate variables including contraception, abortion, post-partum infecundability, spontaneous intrauterine mortality and permanent sterility ..are omitted from this thesis.

III. The research question

As already  mentioned, the purpose of this study is to concentrate on analyzing the determinants of fertility in Vietnam, so the main research question is:

What are the major socio-economic determinants of fertility in Vietnam?

To answer this main question, the research will examine the following sub-question:

1.  Is the fertility equal  across regions ?

2. What are the major socio-economic  determinants of fertility in Vietnam and how important are they?

3. What is the role of various factors in explaining the differences or indifferences of fertility between rural and urban/ among regions/ and ethnic minorities?

4. What are the recommendations for the Government to undertake its policies on these factors?

iv. Structure of the thesis

Besides the Introduction, Appendix and Bibliography, the thesis includes four chapter:

Chapter 1: Analytical Framework

Chapter 2: Fertility status in Vietnam

Chapter 3: Determinants of fertility in Vietnam: Regression and

Chapter 4: Conclusion and Recommendation

Chapter 1: Analytical Framework

This chapter focuses on a selection of the more influential theories of fertility.

I. Concepts and Definitions

Fertility is real fertility ability of a population. It is different from fecundity. Fertility or survival births is affected by fecundity, marriage ages, status available and contraceptive use, economic development, status of women, and structural age - sex. Fecundity is the number of children a woman would have in the absence of all fertility-inhibiting factor.

Similar to above authors, GSO (1997) in Vietnam refers to live - born in term of fertility which is defined that live birth is the complete expulsion or extraction from its mother, a product of conception, irrespective of the duration of pregnancy, which, after such separation, breathes or shows any other evidence of life, such as beating of the heart, pulsation of the umbilical cord, or definite movement of voluntary muscles, whether or not the umbilical cord has been cut or the placenta is attached; each product of such a birth is considered live birth.

II. Measurement of fertility

1. Macro measures of fertility

The "macro" measures which refer to the fertility of population or a sub-group and the "micro" measures which cover individuals.

Aggregate measures of fertility are either period - specific or cohort-specific. The period-specific are cross-sectional measures which show the current or perhaps past, fertility level of the population under study. These measures are defined over a specific period of time( including crude birth rate, child-woman ratio, general fertility rate, age specific fertility rates, total fertility rate, gross reproductive rate, and net reproductive rate.), typically of a one year or a five year duration. In contrast, cohort fertility measures represent the fertility experience of an actual birth or marriage cohort as it moves longitudinally through time. In practice, cohort measures are rarely used because of the large amount of data required for calculating such measures.

2 Micro measures of fertility

Micro fertility behavior is commonly studied in terms of the cumulative fertility and in terms of current fertility. The most frequently used variable for cumulative fertility is the number of children ever born.

2.1.  Cumulative fertility or Children ever born (CEB)

CEB is the cumulative number of live births a woman has had. Although cumulative fertility can be an appropriate dependent variable for some analytical purposes and easiest fertility measure on which to collect data. However, as in the case of the aggregate measures of cumulative fertility, underreporting can result from age misreporting and omission of children who have died or have left the household. Therefore, measures of recent fertility, such as the number of children born in the previous five years, whether the respondent is currently pregnant, or whether she has given birth in the previous year, are often more useful for policy analysis.

2.2. Recent fertility

This indicator not only does better reflect fertility behavior at the present time, it also avoids the misreporting problems associated with cumulative measures. It is perhaps more frequently possible to use a measure of fertility which covers the past year because it is common in demographic surveys to collect information on live births by the age of mother during the previous 12-month period. This information is widely used for calculating aggregate annual fertility levels and for monitoring changes therein over time.

Another category of fertility measures used in micro-level research consists of various measures of desired fertility or family size preferences.

2.3. Desire family size (DFS)

DFS is defined as the number of existing children plus the number of additional children desired by parents. However, DFS and additional children desired are also beset by both measurement and theoretical problems. Whether or not to include these cases in the analysis and if so, how to include them, and if not, how to deal with the potential resulting bias, are difficult questions.

III. Theoretical Models of fertility determinants

The economic theory of fertility, there are two broad strands of influence. One such strand can be found in the works of Becker, Mincer and Willis that is the household "demand" model. The other strand is seen through the works of Easterlin that is synthesis model. What is the work's of each strand. We can analyze as follows:

1. The household "demand" model

1.1. The economic theory of fertility before Becker

This theory stems from the views of Leibenstein and Okun. According to Leibenstein, he said that our central notion is that people behave in the same way as they would if they applied rough calculations to the problem of determining the number of birth they desire. And such calculations would depend on balancing the satisfactions or utilities to be derived from an additional birth as against the "cost". He pointed out that we have to distinguish among three types of cost.

Firstly: The utility to be derived from the child as "a consumption", as a source of personal pleasure to the parents; secondly: The utility to be derived from the child as a productive agent, that is the child may be expected to enter the labor force and contribute to family income; finally: The utility derived from the prospective child as a potential source of security, either in old age or otherwise;

In the Lebenstein model fertility he pointed out that fertility fell as per capita income rose for two main reasons. First, as per capita income increases there are associated changes in the structure of economic activities which tend to reduce the value of children to their parents. Leibenstein suggests that the main causes of this decline are the decreasing utility of children of a given  order in providing old-age security to their parents and the decreasing potential of children contributing to family income through work activities. On the other hand, Lebenstein views the costs of children of a given order as increasing as per capita income increases because "the style in which a child is maintained depends on the position and income of parents" (Leibenstein (1957), p.161) Leibenstein also thought that the indirect costs of children of o given order would rise because he considered "opportunities activities as likely to grow as income increase." (Leibenstein (1957), p.162). That are views which were widely accepted at the time of his writing and contribute to useful things for many researchers of fertility afterwards.

1.2. Becker's Economic theory of fertility

Modern economist's modeling of fertility behavior began with two related insights: that "production" - of meals, healthy children and so on, i, e. of the whole range of goods and services not normally traded in the market - occurs in the household as well as the factory; and that an important input to the production (and consumption) of household-produced commodities is the scarce time of households as optimizing decisions representing economic choices, made fertility, an apparently non-economic area of human behavior, a valid subject for study by economists.

In this household model, children, or more correctly child services, are a normal good; with rising income parents will, all other things the same, want more child services. How then does the model explain the association of high and rising income with low and falling fertility? In two possible ways. First is a price effect. A principal component of the cost of children is the time parents devote to childbearing and childrearing. In higher income societies, wages and thus parents' opportunity cost of time are relatively higher. Insofar as children are more time intensive than other consumption commodities, their relative price rises with rising wages.

A second explanation for the association of rising income levels with falling fertility is the possibility that parents will substitute "quality" of children for high numbers of "quantity" in the production (rearing) and consumption (enjoying) of children - for example, preferring a few well-educated children to many who are not educated..

2. The synthesis model

A second approach of economists in the so-called "synthesis" model of fertility (Easterlin(1978), Easterlin. Pollak and Waachter (1980)). Like the household demand model, the synthesis model posits a utility maximizing household which faces a set of market and shadow prices, a particular household production demand-oriented analysis with demographers' modeling of the supply of children, thus "synthesizing" economic and demographic approaches. It also incorporates some aspects of sociologists' emphasis on the endogeneity of tastes (as opposed to the economists' emphasis on utility maximization constrained only by prices and income), thus "synthesizing" economic and sociological approaches.

In this model, variables which affect supply are based on the proximate determinants of fertility from the Bonggarts (1978,1982) model. This model states that fertility is function of sexual exposure, deliberate fertility control, and fecundability (the possibility to become pregnant)

Although both models have many useful things for research on fertility in recent year, they also have their limits, generally well recognized by their designers, both conceptually and as tools for empirical work. Their limitations are as follows:

Ø      The assumption of utility maximization. Although it is true in the larger field of economics, this assumption has been challenged. The critique is that an increasing degree of rationality in society as a whole is ignored as a possible explainer of fertility decline. For example, a decline in the influence of religion could reduce fertility even in the absence of price changes.

Ø      The "family" utility function. In these models, a single utility function is assumed to embody the preferences of husband and wife for number (and quality) of children. Yet there is evidence that husband and wife do not always agree

Although the theoretical work has not well-established theory, and has often not yielded specific, testable hypotheses, some common themes have emerged which have provided general guidelines concerning the variables to be examined in empirical studies of fertility. The following section will provide some factors that affect fertility.

IV. Empirical research on determinants of fertility.

Existing studies often show that factors including culture, religion, age, education level, income, occupation, status of women, and infrastructure have a strong effect on fertility. Although there are many factors affecting fertility, only those mentioned above will be attributed in this study.

1. Culture and religion

Religions beliefs and cultural pattern usually have strong associations with the roles that women and men play in society. As Cleland 1983 said that polygamy is one of the factor of traditional lifestyle that typically inhibits the number of children born per women. Differences in religion also have an impact on fertility. As a religious group, the Muslims have the lowest fertility (4.17) and those with "other" religions (mostly Zione, a local religion) have the highest (6.33). (Ceccato, 2000, p9).

2. Age

Age is one of the most important variables characteristic individual participants in the reproductive process and thus is included in some form in almost all studies of fertility. Many researches also find that age of the women has a positive effect on fertility. According to Desai 1996, when age of the woman increases by 1, the number of children ever born will increase 0.170 children (that is the case of Vietnam). In addition, the relationship between age and fertility in Cote d'Ivoire is the same as Vietnam. If age of the woman increases 1, the number of children ever born will increase 0.4296 children (Martha Ainsworth, 1989).

3. Income

Empirical studies have reached mixed conclusions concerning the relationship between income and fertility. If the analysis is restricted to the developing countries, income and fertility are sometimes positively related. For example, In Vietnam, the relationship between expenditure per adult (proxy for income) is positive. There results show quite clearly that in rural Vietnam the poor have fewer children than the better-off (Jaikishan Desai, 1995) and according to Ainsworth (1989), the positive relationship between income and fertility is also seen in Cote d'Ivoire. However, if one examines the relationship at the aggregate level for a cross-section of all countries, including industrialized countries, the expected negative relationship seems to hold such as Japan, Germany... This is suggestive of a non-linear relationship at the aggregate level, with fertility initially increasing with higher positive relationship between income and fertility posited above, but the difficulty of empirically isolating this theoretical concept has led to varied results and no firm conclusions (Ghazi M. Frarooq and Deborah S. DeGraff, 1988,p37).

4. Education

In comparisons of couples in 20 developing countries, lower fertility is almost always exhibited when the wife has secondary education (Rodrigue and Cleland, 1982). Summaries of household multivariate studies show that the effect of wife's education is negative more often, and more likely to be statistically significant. In addition, there is evidence the negative effect does not appear until higher levels of education are reached. Some primary education, in fact, appears to increase rather than decrease fertility. (in Sierra Leone) (Cohanrane, 1983, p607). This positive effect of education at low levels is especially marked in rural areas (Hermalin and Mason, 1980) and in less urbanized and poorer developing countries, where per capita in come is below US$500 (Cochrance, 1983,p578) It generally does not appears in more advanced developing countries. However, data analysis of 'the survey in changes of family and fertility 1990' (Nguyen Thi Van Anh, 1993), indicates that education level does not affect desire for more children significantly in Vietnam and according to Nguyen Minh Thang, Charles Hirschman, Nguyen Huu Minh, 1995 said that education level is not a deciding factor in fertility decisions because the impact of factors that strongly affect fertility, such as education level of women, has been reduced over time. In addition, the results in Cote d'Ivoire show that schooling has a negative effect on fertility  (Ainsworth, 1989).

5. Occupation

   Studies in developed countries indicate a negative relationship between female work hours and fertility. But studies in less developed countries have not always indicated a negative relationship and in the empirical work is presented by McCabe and Rosenzweig who obtain strong positive association between female wage and fertility

6. Infrastructure

            The empirical evidence in Vietnam suggested that fertility of people who have electricity or have a radio/television is lower than for women who do not have electricity or who do not have a radio/television. Moreover, fertility rates of groups of people with a high fertility history have experienced a distinct drop in fertility rate with the introduction of television/radio to their area (Nguyen Minh Thang, 2000). Another study showed that Communes have a road passing through have lower fertility.

Chapter II: Fertility status in Vietnam

The thesis make extensive use of the VLSS 1997-1998, the extremely rich datasets for analyzing

The children ever born by current fertility (including the children were born since 1993 up to 1998 ) will be used to analysis in this thesis because this indicator not only does better reflect fertility behavior at the present time, it also avoids the misreporting problems associated with cumulative measures.

I. Characteristics of fertility in Vietnam

1. Fertility trends

The fertility level in Vietnam has been declining since in the 1970s (Jones, 1982: Alman et al 1991, Banister, 1993, Nguyen Van Phai et al, 1996, Nguyen Minh Thang, 1999). Especially, it declines rapidly from 1993. We can see as follows in Table 1.

Table 1.        Total fertility in the period of 1969-1999 in Vietnam.

Unit: number of children

Sources: ESCAP; population census 1979, 1989, 1999: VNDHS 88, VNDHS 94, VNDHS 97, *author's calculation based on VLSS 1997-98

It is difficult to exactly compare data from different sources as the nature of the questions vary from survey to survey. Also, the technique of estimating fertility rates is different in each survey. As Table 1 shows, the fertility rate in Vietnam has decreased rapidly from 6.1 children in the 1969-1974 period to 4.8 children in the 1974-1979 period to 4.2 children in the 1986-1987 period to 4.0 children in the 1987-1988 period to 3.8 children in the 1988-1989 to 3.9 children in the 1991-1992 period to 3.5 children in the 1992-1993 period to 3.1 children in the 1993-1994 to 2.7 children in the 1992-1996 period and to 2.6 children in the 1997-1998 in the period.

2. Features of fertility in Vietnam

As I presented above, because TFR, ASFR ....indexes are difficult to calculated and to understand so that in this study the CEB index will be chosen to analyze the fertility differences in Vietnam.

2.1. Deferential fertility among regions

The general fertility patters and trends can be analyzed for the whole country and different sub-groups of the population. Table 2 presents the number of children ever born of a woman of the nation as a whole and its seven geographical regions.

Table 2.        The number of CEB for the 5 years preceding the survey by Region,  1997-98

Unit: person

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

It has been shown that the highest fertility levels are observed for the Central Highlands (region 5). In Central Highlands, a woman has average of 4.01 children while a woman in Vietnam has average of 3.10 children. However, the Red River Delta and the Southeast have the lowest marital fertility. Why are there different among regions?

Almost two-third of Vietnamese state that they are Buddhists. In this section  we divide religion into 3 groups like Buddhists, Christians and none both of them. There are significant differences among religions. According to the Vietnam living standard survey in 1997-98, which did obtain fertility rates by religion for Christian, children ever born per a woman age from 15-49 have 4.10; for Buddhists 3.02 and the other 2.91 (Author's calculation based on VLSSs 1997-98). From mentioned above, we can see that religion also impact on fertility, for Christians, number  of children born is higher than that of the rest of the religion.

2.2. Deferential fertility among Ethnic structures

Vietnam has 54 ethnic groups. The Kinh (Vietnamese) group accounts for 80 percent of the population. The Kinh live largely in the river deltas and cities. Other ethnic groups like Nung, Muong, E de, Gia Rai, and Thai live in highlands, mountains and plateaus. In this thesis, It is considered two groups that are Kinh and non-Kinh group.The fertility level of each ethnic group is quite different, the number of children ever born of Kinh is 2.75 children per woman while that of non-Kinh group is 3.37 children per woman( Author's calculation based on 1997-98 VLSS).

II. Socio-economic development and Fertility

1. Women's education

For women, education has a great effect on knowledge, attitude and practice in the field of family planning and reproductive health. Table 3 represents the number of children ever born of married women by education.

Table 3.        Relationship between Fertility with women's education

Unit: number of children

Source: Author's estimation based on 1997-98 VLSS

We can see that the relationship between women's education and fertility is inversely associated. If a woman has no schooling, she will have 4.61  children increasing by 1.51 times compared to CEB in entire country in the same period. In contrast, with women's education at university or higher, the number of children ever born will be reduced significantly, it declines by 1.63 times compared to CEB in a whole country. The higher education of women is, the lower fertility level is.

2. Women's occupation and fertility

Fertility level across kinds of occupation is also observed in Figure 1. As one expects, "White collar workers" had the lowest number of children ever born per women (2.13 children per a woman), "Agriculture" had the highest number of children ever born per women (3.27). Looking at the Figure 1 we can see that if women's job is Agriculture, the number of children ever born is 3.27 children. It increases by 1.17 times compared to that one of women having no professional  job, 1.54 times compared to that one of "white collar workers", 1.1 times compared to that one of "sales workers", and 1.36 times compared to that one of the "Blue collar workers". It is quite suitable for Vietnamese case. This relation  can be presented more detail in among regions and between rural and urban.

Figure 1: Fertility differences by women's occupation

Figure 1.      Fertility differences by women's occupation

Source: Author's calculation based on VLSS 1997-98.

3. Income and fertility

For the convenience of analysis, we divided 2444 observation into five quintiles and each quintile contains 20 percent of observations according to their income. The first quintile includes 20 percent of the lowest income level and the fifth one contains 20 percent of the highest income level[2] Table 4 presents the results of fertility level by quintile. As expected, households of the lowest income level have the children higher than that of the rest of the quintile. The average number of children per a woman in the lowest income quintile - quintile 1 is 3.75 persons while this figure in quintile 2-the second lowest income quintile is less than 1.18 times compared to that one of quintile 1. The number of children in quintile 5 - the highest income quintile is 1.65 times lower that one of quintile is 1 and 1.41 times lower than quintile 2 and lowest compared to that the whole sample.

Table 4.        Fertility changes between rural and urban by quintile

Unit: people

Source: Author's calculation based on 1997-98 VLSS

Chapter III: Determinants of fertility in Vietnam

1. Methodology

The thesis is interested in qualitative determinants of fertility level. The decompositions start with standard regression function, which was widely used to analyze determinants of fertility (Cochrane 1979, T.P, Schultz 1974, 1981 and T.W. Schultz, 1974 and Ainsworth et al 1996). The empirical model of fertility determinants regresses a measure of cumulative fertility children ever born to each women - on a set of independent variables that are assumed to be exogenous to fertility decision but that influence either demand for or supply of children. This reduced-form model of determinants can be written as

Y = a +bX +u

Where: Y is the cumulative fertility children ever born, a is intercept, b is a vector of coefficient, u is a random disturbance term reflecting unobserved characteristics distributed with mean Zero and constant variance, X is vector of individual characteristics. The model for CEB and posited signs can be summarized as:

Y = f (x₁⁺, x₂^-, x₃^- , x₄^-, x₅⁺, x₆^-, x₇^-, x₈^- , x₉^-, x₁₀^-, x₁₁⁺)

where x₁ is the age of women; x₂ is education level; x₃ is occupation of women; x₄ is income; x₅ is religion of women; x₆ is the woman's ethnic group; x₇ is region variable; x₈ is area of residence such as urban or rural; x₉ is marital status; x₁₀ indicates accessing to electricity; x₁₁ is frequency of cutting electricity.

The regression model will be estimated by single - equation regression techniques assumption that explanatory variables are exogenous. The analysis here will be mainly estimated by OLS. However, OLS technique does not reflect the family with two children or less they want to have more children while probit[3] technique needs the demand for that purpose.  Therefore, I will use both of them to estimate the children ever born or probability of having more children in order to compare between the two techniques. So that in this my thesis I will run some regression models. The selection of variables and their expected signs are used for equation regression that will be explained more detail in table 6:

II. Estimated results and Explanation

1. Ordinary Linear Squared model

1.1 Fertility model in the whole country

The following discussion outlines general trends in the parameter estimates. Initially, the regression model was estimated with explanatory variables as mentioned above. However, when this was done the coefficients on linear term for some regional variable s, for tribe variable and one of the job variable were not statistically significant at the 10 percent significance level. Therefore some of that variables were omitted and the regression model was re-estimated. In this study, I decide to omit the tribe variable for all regressions in later because it is not only statistically significant but also unexpected sign.

Examining first the regional location variables as Table 5 shows. Relative to Northern Uplands (the omitted regional dummy), the average number of children ever born in The Central Highlands, North Central Coast, and Central Coast are significantly higher. At the 10 percent significance level, the coefficients on linear term for Red River Delta, North Central coast, Central Coast and the Southeast are not statistically significant. However, as expected, the sight of Red River delta, Central Coast, Southeast and Mekong areas variables with fertility negative while in other parts of the country is positive. There results are interesting for some main reasons. First, they show that discussions of fertility in Vietnam should not be cast simply in terms of North-South differences. Even though the Northern Uplands, the North-Central Coast and the Red River Delta have similar political pasts, the economic conditions in the  Northern Upland and North-Central region are quite different from those in the Red River Delta. Second, these results indicate that the highest fertility observed in the Central Highlands, as discussed in chapter two, is related to lower education and living standard.

Considering next the educational variables on the fertility level. Table 5  also shows that the education level of the married women has a statistically significant effect on the fertility. As expected, the relationship between education and fertility is negative. Particularly, the number of children ever born of a woman having primary-education was 0.840 lower than the benchmark of no-education. This figure decreased with higher education attainment 1.208 for lower -secondary, 1.533 for upper-secondary and 1.931 for university and higher. In generally this is quite true for the most of the nations in the world and particularly in Vietnam. This result is interpreted as follows:

There are a variety of hypotheses about why expanding educational opportunities are inversely associated with fertility (Cochrance 1979, and Casterlin 1983).First, because women are unlikely to marry while they are still in school, schooling is associated better-educated women and better-educated populations are likely to have a greater knowledge of contraception and the ability to practice it more effectively. Second, for women with advanced education, particularly university degrees, childbearing involves large opportunity costs in the form of foregone income. Third, within a budget constraint there is always a trade - off between quantity and quality when it comes to children; better-educated women may opt for fewer children with high-quality lives, choosing, in a manner of speaking, to reproduce themselves socially rather than simply physiologically. Finally, and perhaps most importantly, mandatory school attendance, expanding educational expectations in the younger generation, and rising educational norms increase the cost of children and reduce the value of child labor for the entire, not only the educated, population. Thus, in this last view, expanding educational opportunities affect the fertility decisions of all couples.

Regression results also shows that women’s job is negatively associated with the number of children ever born. However, only three kinds of occupations were statistically significant. The number of children ever born of all occupation were significantly lower than the no-job benchmark. The average number of children ever born of a woman having white-collar occupation decreased 0.521 lower than the benchmark of no-job. This figure decreased with the other occupation 0.246 for sales occupation, 0.273 for blue-collar workers and 0.150 for the agricultural workers. This negative relationship can be explained as follows:

As in the case of education attainment. Potential causal linkages between work and fertility include the increased opportunity cost of women's time, the incompatibility of child-care and work, exposure to smaller family size norms and different attitudes towards women's role, greater access to information particularly concerning family planning, delayed marriage, and reduction in breastfeeding. Although women having agricultural sector is also negatively associated with the number of children ever born (comparison with women's no-job), this result is not statistically significant and the same as the descriptive section. It may be resulted from multivariate analysis  that lead to sample bias due to merging many variables. Therefore, our expectation is not satisfied.

The next variable, age of women has significantly positive effects on the number of children ever born. Holding all other explanatory constant, if the women has an additional year of age, the average number of children ever born would increase by approximately 0.181 children. This understandable that age plays an important role in fertility outcomes due to its association with menarche, marriage, widowhood, divorce, frequency of intercouse, fecundity and  menopause.

Expenditure impacting on fertility is seen as Table 5 (proxy for income). In general, fertility is negative associated with per capita expenditure in the whole country, it happens as expected. The result in model shows that the average number of children ever born by 0.0001children (controlling for all other explanatory variables). As Huw Jones (1984) said that 'the rich get richer and the poor get children'. This result indicates that there are trade off between quantity and quality. A rise in income was said to increase aspirations for social advancement, which may be though of an increased desire for other goods that compete with children for family resources. This desire would act to reduce the number of children people would have. In addition, rises in income and economic development imply such as changes as increase in education, shifts from rural to urban employment and life expectancy, and increased availability of contraceptive information. Each of these phenomena is by itself related to lower fertility. We also see that rise in income can lead to reducing in the roles of children. High income implies the rich family so that they do not need supporting of children for old-age, illness or disabilities...both of these possibilities are likely to reduce fertility.

   Marital status of women variables are showed that there is negatively associated with fertility level as expected. The average number of children ever born of widowed women was 0.613 children lower than the benchmark of without widowed and separated women, 1.409 children for separated or divorced women. It can be explained as follows: Marital status is intended to represent exposure to sexual intercouse as discussion above. In Vietnam, remarriage proportion of widowed and separated or divorced women is very low. Moreover, in recent years there are many divorces or separates of the couples. Therefore, the reducing in fertility is not avoidable.

Another variable, that is religious variables, variables in this dummy group are Buddhist with coefficient 0.184 and Christian with 0.487. This coefficients are highly statistically significant. One can argue that "Christian" has more number of children ever born than "Buddhist" and both of them has more number of children ever born than benchmark of none-both of them. The difference in fertility level of "Christian", "Buddhist" and none-both of them can, therefore, mainly be explained by the difference in custom. Moreover, fertility is equated with virtue and spiritual approval, while reproductive failure and self-imposed fertility restriction are seen as punishment and evil respectively.

The last, but not least important, variable in this dummy group is areas of residence with coefficient -0.341. This coefficient is also highly significant. Indicating that "women in urban" has less children born than "women in rural". It is understandable that because women in urban has access to better education, a wider spectrum of work opportunities a more  improved public health environment, and generally more avenues for self-improvement and social mobility. They also face higher coasts in raising children so that fertility of urban tends to be substantially lower than rural fertility.

Table  5: Regression results for CEB

Dependent variable is Children ever born after 1992

Source: Author's calculations based on data of Vietnam Living standard survey 1997-98

1.2. Children ever born Model in Rural area.

Turning to examine the regression results from the children ever born model in rural area. The model replicated the approach used in the children ever born  model in the whole country. The dependent variables is still number of children ever born. The independent variables are the same as used in the whole country of children ever born model. However, in the model of rural area, community variables is added .

Table 5 summarizes the regression for children ever born in rural area. Firstly we will look at the region variables, the Red River Delta, the South Central Coast and South East do not appear to be statistically significant and sight of the South East in rural is different with that of the whole country. The coefficients for regional variables in Red River Delta, Northern Central Coast, Central Highlands estimated here higher than those estimated in the whole country, in the other hand, the coefficients for regional variables in South Central Coast and Mekong areas estimated here decreasing in fertility is stronger than those of the whole country. It can be explained that some regions has discrimination.

Effects of women's education on fertility level estimated in the children ever born in rural area model nearly the same as those estimated in the whole country model. With the women's education level in primary, lower, upper, the decreasing in the average number of children ever born in rural area is weaker than those of the whole country while in level of university this figure is stronger than urban area. It reflects that in rural area the number of women having university education less than those of urban area or the whole country so that they usually have high position in the society or work opportunities are more than the urban after graduating from university.

Occupation variables are negatively associated with the number of children ever born as the same those estimated in the whole country model. However, in the model of rural area, the blue-collar worker variable is not statistically significant while the agricultural worker variable is statistically significant. It is recognized that the women having the blue collar occupation accounts for a small proportion while agricultural occupation dominated. In addition, the recent years, having many newly agricultural implements substituted for manual working and happened the lack of land in many area which  lead to abundance of labor and reducing in the value of children. Therefore, many couples do not want to bear more children. This results is also different from the analysis of chapter 2, due to this multivariate analysis.

As Table 5 also shows that the age, marital status (widowed, separated or divorced women), religion, and expenditure (proxy for income) coefficients in the rural model is nearly the same as those of the whole country model.

The next variable that is community variable such as having electricity or frequency of cutting electricity in a community. The regression give us a coefficient of -0.005 for having electricity variable, implying a negative relationship between the number of children ever born and the having electricity in a community. This negative  relationship can be explained as follows: Because having electricity is one of the index that reflects a good condition of living in an area while the frequency of cutting electricity is positively associated with children ever born. Holding all other explanatory variables the same, an additional frequency of cutting electricity, the average number  of children ever born increased by 0.076 children.  

2. Probit model

2.1. Model of desiring more children in the whole country

Table 7 presents the probit model of the probability of having more children in Vietnam 1997-98. Rather than reporting the parameter estimates", Which are difficult to interpret on their own. The marginal effects indicate the more children and respective explanatory variable. For dummy variables, the marginal effect is calculated as the change in the dependent variable associated with a move from a value of 0 for the dummy, to 1.

 Examining first the region variables. Table 7 indicates that the probability of having more children in 1998 is 13 percent point lower at the 10 percent significance level if women live a Red River Delta and is 17 percent point lower at that significance level if women live a Mekong area. But this relationship is not monotonic, women residing in all other regions, except the South East is likely to have more children at the 10 percent significance level.

The more variable important factor is education variables. This education level of women has strong negative relationship with fertility. Relative to the no-education (the omitted educational dummy), women in all other education level are less likely to have more children at the 10 percent significant.

Although all occupation of women variables are negatively associated with fertility, they do not appear to statistically significant at the 10 percent significance level.

Next variables, age of women, an additional year of age of the women is associated with a 6 percent point higher probability of fertility for that women. In addition, religions of women has also positive relationship with fertility. If religion of women is Buddhist, it will be likely to have more children (with a marginal effect of 0.072 and 0.097 for Christian). This results show that there are the impacting of religion on fertility level.

Considering next the sets of variables such as marital status of women, tribe, income, and area of residence (urban) except tribe and widowed women variables, appear to be statistically significant at the 10 percent significance level and reduce the probability of having mare children. However, relationship between income of women and fertility is not strong. If income increases by 1 unit, the probability of not having more children is not significant (with marginal effect of 0.000047).

In Summary, the probit model has indicated that a woman having less than 3 children is more likely associated with higher education level, stable job, living in area with good socioeconomic condition.

Source: Author's calculations based on data of Vietnam Living standards Survey 1997-1998

Table 7: Probability of bearing children per woman in the whole country

Number of obs  2444

Prob >chi2         0.000

Pseudo R²          0.3653

Note: (*) dF/dx is for discrete change of dummy variable from 0 to 1

Source: Author's calculations based on data of Vietnam Living standard survey 1997-98

Z and P|z| are the test of the underlying coefficient being 0

2.2. Model of desiring more children in the Rural area.

This model also replicated the approach used in the whole country model. However, these is a difference between 2 models. In rural area model, community variables (accessing to electricity and frequency of cutting electricity) are added as discussion above.

Table 8 summarizes the regression results for the model of rural area. Firstly we will look at the region variables. In the Red River Delta, South East and Mekong area are inversely related to fertility (desiring more children) (with a respectively marginal effect -0.062, -0.001, and -0.261). It is the same model of the whole country, that relationship is also not monotonic.

Although effects of education on having more children are estimated those estimated in the model of the whole country. The marginal effect of education in rural is lower than that of the whole country.

Like the whole country model, all of the women's occupation variables are not statistically significant at the 10 percent significance level, access to while-collar worker, sales, and agriculture reduce the probability that a woman have more children. But this relationship is also not monotonic, if the women are blue-collar workers, the probability of having more children increase again that is an exception case.

Next some variables such as marital status of women, tribe, income and religion of women are estimated in this model nearly the same estimated variable in the whole country model by both sight and the statistically significant at the 10 percent of those variable.

The last, but not least important, variables in this community group is electricity using and frequency of cutting electricity. Accessing to electricity variable reduces the probability of having more children at the 10 percent significant level (a marginal effect of 0.002) while frequency of cutting electricity is inversely related to accessing to electricity, it does not appear to be significant at the 10 percent significance level and increases the probability of having more children.

Table 8: Probability of bearing children in rural area (Probit Estimates)

Number of obs  1996

Prob >chi2         0.000

Pseudo R²          0.3609

Note: (*) dF/dx is for discrete change of dummy variable from 0 to 1

Source: Author's calculations based on data of Vietnam Living standard survey 1997-98

Z and P|z| are the test of the underlying coefficient being 0

Chapter 4: Conclusion and recommendation

Basing on the analysis of data and information obtained from a Vietnam Living Standard Survey, several findings and policy recommendation have been drawn from the previous chapters.

Ø      The analysis finds that an increase in the wife's education of the couples is important in reducing the number of children ever born or desired completed family size in Vietnam. Therefore we must continue to invest in the education of women. Education is the cornerstone to improving the status of women, both for the skill set it provides for gaining entry to the labor force. Moreover, education is one of the strongest determinants of fertility level of any socioeconomic variables in the data set. At least, mass education for women has to be implemented in the whole country, especially in Northern Uplands, Central Coast and Central Highland. Because the basis literacy skills that are conferred bring the concept of the written word into the realm of consciousness and enable the individual at the very least, to hear the message about family planning or health practices. In addition, with higher levels of education, the woman's attitudes toward marriage and childbearing become more important, yet her behavior will also be affected by the norms of the social groups to which she belongs. No individual's behavior can ever be divorced from the social, economic, and cultural context in which she is situated, but as educational attainment rises, the individual moves away from community based childbearing norms toward a more individual set of beliefs and behaviors and, perhaps, into social networks with lower fertility norms.

Ø      The finding that the poor belonged to mainly in rural areas and women working in non-modern areas, especially in agriculture sector and women without job have more children than those of modern areas. Thus, the government should be made to increase the opportunities for female employment in off-farm activities in the formal labor market. Because if farm women get good opportunities to work in off-farm or non-farm activities outside their own farm, the value of their time will be increased which also increase the cost of their children. Thus, they would prefer to have fewer children than before.

Ø      In this thesis finds that the number of children ever born or "more than two child family" in the Northern Uplands, North Central Coast, and Central highlands tended to be higher than in Red River Delta and Mekong areas. Therefore, the policy measures to support the development of such regions (such as provision of credit, technical assistance, and investing in education) are necessary. In addition, Government or local officials should introduce broad social programs to raise status of women, which is essential to prevent early marriage, and the use of incentives and disincentives to encourage preferences for smaller families.

This study shows that cultural and religious factors also impact on fertility level. Therefore, all of the policy measures listed above should be applied.