1.      Introduction:

Inequalities between the poor and the rich are always a cause for concern, in both the developed and developing worlds. Inequalities are revealed in many aspects of socio-economic life, from political to economic, from cultural to social. Inequalities in the health sector, expressed as inequalities in health outcomes, in the utilization of health services, and in the benefits received from public expenditures on health services are very large (Van Doorslaer et al. 1997; Castro-Leal et al. 1999; Castro-Leal et al. 2000; Gwatkin et al. 2000; Sahn and Younger 2000; Wagstaff 2000). Many governments, international organizations and bilateral aid agencies have committed to reduce poor-nonpoor inequalities in the health sector (World Bank 1997; Department for International Development 1999; World Health Organization 1999) but they still occur and are becoming even more serious.

Child malnutrition is one of the health sector problems. It attracts attention of not only nutritional analysts but also economists and politicians because that directly affects the country’s human resources. There is much work on child malnutrition. Among them, child malnutrition inequality emerges as one of the hottest issues. Like other health inequalities it appears in the middle phase of development process. Fewer opportunities are given to the poor. That is not only attributable to the limited resources but also the discriminated allocation itself.

Vietnam has a relatively high prevalence of child malnutrition although it is falling (World Bank 1999). It shows that the rate of underweight children in Vietnam is 33.1%, stunting 36.4% and wasting 5.6% in comparison to 19.1%, 25.8% and 7.3% of the world level (UNICEF Global Database on Child Malnutrition 2001). With great efforts of the Government, in recent years these numbers have lowered considerably. In 1985 nearly 60% of Vietnamese children under five suffered from stunting (NIN 1985). In 1996-97, only 44% of under five children were stunted and this proportion has dropped to 36.5% in the year of 2000 (Vietnam Children Indices 1998, 2000). However, another problem occurred? While the rate of child malnutrition decreases, inequality in child malnutrition increases. This inequality in 1998 was calculated to be 1.36 times higher than in 1993 (Wagstaff, 2001). It reflects an inevitable tendency of the first step of development. But it may overlap the growth achievement unless it is controlled and supervised in the right way. The questions are raised: What are the causes of child malnutrition inequality Vietnam and How to reduce it? According to the above reasonings, there is a need to analyze “Inequalities in Child Malnutrition in Vietnam”.

2.      Theoretical framework:

2.1  Concepts and definitions:

2.1.1        Malnutrition:

Nutritional status of one person is the condition of the body resulting from the intake, absorption and utilization of food (FAO, 1982, p.53).

Malnutrition is defined as nutritional disorder or condition resulting from faulty or inadequate nutrition.

Malnutrition is classified into four types (Mayer, 1976):

Over nutrition: this is a condition of one person who consumes too many calories.

Dietary deficiency: this is the result of consuming a diet that lacks sufficient amounts of a particular nutrient such as vitamin or mineral.

Secondary malnutrition: this is a consequence that an individual suffers when he or she is unable to digest or absorb food successfully. Secondary malnutrition usually stems from diarrhea, respiratory illness, measles and intestinal parasites.

Under nutrition: when an individual does not consume enough food he will get under malnutrition. He lacks of the calories or protein necessary for normal growth, body maintenance and physical activities.

In the context of this thesis, malnutrition is understood as under-nutrition. Because it is a popular phenomenon in developing countries, Vietnam is an example, where some people still suffer from poverty.

Malnutrition is measured by anthropometric indicators, which base on growth and body composition such as height and weight. There are three main anthropometric indicators (Koch, S. and Bui Linh Nguyen, 2001).

Stunting is used to describe a condition in which children fail to grow in sufficient height, given their age.

Underweight is used to describe a situation where a child weight less than expected, given his or her age.

Wasting refers to a situation where a child has failed to achieve sufficient weight for height.

These anthropometric indicators are calculated by Z score. The Z score is a statistical measure of a distance between the actual anthropometrical (height or weight) value of a child and the median (average) value of the population over the standard deviation.

The most common cutoff point is -2Z-score, i.e., two standard deviations below the median values of the international reference. This is the cutoff risk level used to differentiate malnourished children from those adequately nourished. Children whose Zhfa, Zwfa and Zwfh scores fall below this point are therefore considered stunted, underweight and wasted.

2.1.2        Malnutrition inequality:

Malnutrition inequality is defined as the difference in malnutrition rates between groups, in terms of age, income, education, ect.

Malnutrition inequality is measured by concentration curve graphically and concentration index mathematically.

Concentration curve plots the cumulative proportion of malnourished children on the y-axis against the cumulative proportion of children on the x-axis ranked by household consumption, beginning with the most disadvantaged child.

Figure 1.1: Malnourished concentration curve

Source: Wagstaff, A. and N. Watanabe (2000)

If L(p) coincides with the diagonal, all children, regardless of their household consumption, suffer from the same malnutrition rates. If, and is more likely, L(p) lies above the diagonal, inequalities in malnutrition favor the better-off children. There are less better off children getting malnutrition and we will call such inequalities prorich. If L(p) lies below the diagonal, we have inequalities in malnutrition disfavor the better-off. There are less malnutrition among the worse-off children and we call inequalities propoor The further L(p) lies from the diagonal, the greater the degree of inequality in malnutrition across quintiles of living standards.

Concentration index - denoted by C (Wagstaff, A. and N. Watanabe, 2000) - is the mathematical expression of concentration curve and defined as twice the area between L(p) and the diagonal.

Where n is the sample size, m is the mean of y and Ri is the fractional rank of the i^th person in the income distribution.

C takes a value of zero when L(p) coincides with the diagonal and is negative (positive) when L(p) lies above (below) the diagonal.

2.2  Health production function (Glewwe, P. et al. 2002):

Glewwe, P. et al. assumed that nutritional status of a child is a health production function and is determined by three kinds of variables: health input (HI), the local health environment (E) and the child’s genetic health endowment (e):

H = f (HI, E, e)

Health inputs are determined by household income level (Y), the educational levels of both parents (MS, FS), their “taste” for child health (h), the local health environment and the child’s genetic health endowment:

HI = g (Y, MS, FS, h, E, e)

Applying this function to econometric model, Glewwe, P. et al. has run regression and found that it is not household consumption which has significant impact on child malnutrition in Vietnam but commune health services. Therefore he emphasized improving health services is the best way to reduce child malnutrition in Vietnam.

2.3  Decomposing malnutrition inequality (Wagstaff, A. et al. 2001):

Suppose we have a linear regression model linking malnutrition variable, denoted by y, to a set of k determinants, x_k:

(1) where a and b are coefficients and e_i is an error term. We assume b_k is unchanged for everyone. There are links between the concentration index and the inequalities in the underlying determinants of y:

(2)

where m is the mean of y, `x_k is the mean of x_k, and C_k is the concentration index for x_k. GC_e is a generalized concentration index (GCI) for the error e_i and h_k is elasticity of y with respect to x_k. GC_e is easily computed as a residual from equation (2):

(3)      

Thus inequality in malnutrition can be decomposed into explained and unexplained inequality. The former is made up of k terms corresponding to the k covariates. Together, these terms are a weighted average of the concentration indices for the covariates, where the weight for the k^th covariate is its elasticity. Thus, the larger the elasticity, and the more unequally distributed across income groups x_k is (i.e. the larger is C_k), the greater the importance of inequality in x_k in accounting for inequality in malnutrition.

2.4  Decomposing the changes in malnutrition inequality:

Applying Oaxaca’s method, the above concentration index function continues to be decomposed to understand the changes in malnutrition inequality.

(4)              

or

(5)              

Basing on these functions, we are not only able to evaluate the changes of malnutrition inequality but also determine what causes these changes overtime.

These previous studies have found income/household consumption the most important determinant of malnutrition inequality. However what causes the change in malnutrition inequality overtime are still in dout, which needs further research in the future.

3.      Child malnutrition inequality in Vietnam:

In 1984, Vietnam was the second poorest country in the world (WB Report, 1986). The “Doi moi” has promoted and changed almost every aspects of Vietnam socio-economic life. GDP growth rate increases at average of 9 to 10%; inflation is controlled, more employment are created; international trade is opened and diversified; savings and investment are encouraged (Vietnam Economic Times, 2001). That helps to improve the living standard and reduce social problems in society.

The following figure shows that the rate of low height-for-age children has fallen continuously over time from 59.7% in 1985 to only 36.5% in the year of 2000.

Source: Le Thanh Ngoc (2001)

However with 36.4% of moderate child stunting and 11.9% of severe child stunting in 2001, Vietnam still have a rather high child malnutrition rate in the world (UNICEF, 2001). In Guinea those rates are 26.1% and 10.1%, Lao’s are 24.8% and 6.0% and Peru’s are 25.8% and 8.0%. The reason is that GDP per capita of Vietnam is lower than others. So is public expenditure for health. In 1999, Vietnam GNP per capital was only 370 USD per year and public expenditure accounted for 0.8% of GDP in comparison to Guinea’s 800 USD and 2.5% (Vietnam Committee for Child Protection and Care, 2001).

While child malnutrition reduces, there exists child malnutrition inequality in Vietnam.

3.1  Child malnutrition inequality by gender:

It seems that child malnutrition is slightly higher for the boy than the girl, with approximately 1.02 times higher in almost month groups, except for 0-5 month group with malnutrition rate of girls is 1.65 times higher than boys.

Source: Le Thanh Ngoc (2001)

3.2  Child malnutrition inequality by age:

Difference in child malnutrition by age needs to pay attention. In the following figure, child malnutrition increases when month group is higher in both years. Children belonging to 11-23 and 47-59 month groups are found to have highest prevalence of malnutrition, with 4.4 times higher than children belonging to 0-5 month group, which have the lowest, in 1992-93 and increases to over 10 times higher in 1997-98.

Source: Le Thanh Ngoc (2001)

3.3  Child malnutrition inequality by urban- rural area:

The demonstration of child malnutrition inequality is found between urban and rural areas. Moreover child malnutrition inequality is higher when the time goes by; with the gap is 1.6 times in 1993 and 2 times in 1998. This is the result of socio-economic inequality which produces larger and larger differentiation between urban and rural.

Source: Glewwe, P. et al. (2002)

3.4  Child malnutrition inequality by ecological region:

Child malnutrition inequality between regions is illustrated well in the following figure. In both years, South East is the region having the lowest child malnutrition proportion. In 1993, that rate is highest in North East and North West at nearly twice of South East. In 1998 the region having the highest is Central Highlands and the gap rises to 2.4 times.

Source: Glewwe, P. et al. (2002)

Child malnutrition rate has decreased over time at different levels for different regions. Red River Delta has that rate reduced most. Central Highlands has least reduction. Thus from a rather high position in child malnutrition rank in 1992-93, Red River Delta becomes the second lowest rate of child malnutrition in 1997-98 while Central Highlands becomes the highest.

3.5  Child malnutrition inequality by parents’ height:

Child malnutrition inequality is associated with parents’ height. Haughton, D. and Haughton, J. (1997) has clarified that the taller parents seem to have taller children and therefore the probability of having stunted children is lower. In the above figure, adequate nutritional children belong to taller parents than malnutrition ones. This presents more strongly from time to time.

Source: Author’s calculation basing on VLSS 92-93 and 97-98

3.6  Child malnutrition inequality by mother’s education:

The better the mother is educated, the less malnourished children she has. That is because when mother’s educational level is higher, she knows better about how to feed and take care her babies. This statement is justified in 1998 with child malnutrition rate of primary and lower secondary mothers are 1.63 times higher than upper secondary and university ones.

Source: Author’s calculation basing on VLSS 92-93 and 97-98

In 1993 mothers with lower secondary educational level reveal highest prevalence of child malnutrition. However the gap in malnutrition rates between mother groups of lower education, including iliteral, primary and lower secondary, and higher education, including upper secondary and university, is reasonable.

3.7  Child malnutrition inequality by father’s education:

The similar situation are exposed to father’s education but less severe level with primary fathers have child malnutrition rate 1.55 times higher than university fathers in the period of 1998.

Source: Author’s calculation basing on VLSS 92-93 and 97-98

Once again in 1993 fathers with lower secondary educational level introduce highest rate of child malnutrition. The gap in malnutrition rates between father groups of lower and higher education is still meaningful.

3.8  Child malnutrition inequality by living standard:

3.8.1        Income:

Child malnutrition inequality in Vietnam is prorich. It means the poorest have the highest rate of malnutrition and vice versa. In the following figure, child stunting inequality demonstrates obviously in both years with the rate of the poorest is twice of the richest  in 1992-93 and increases slightly to 2.43 times in 1997-98.

Source: Le Thanh Ngoc (2001)

Inequality in child malnutrition related to income is very popular and is found in nearly 20 developing countries such as Bangladesh, Indonesia, Philippines, Zambia, etc and often inequality disfavors the poor.

3.8.2        Housing status:

The following figure has illustrated child malnutrition inequality with respect to housing status. It is easy to realize that children living in semi-permanent house have a reasonable higher rate of malnutrition than those living in permanent one. Specifically, the stunting rate of children living in semi-permanent house is 1.23 times higher than in permanent one in 1993 and rises to 1.64 times in 1998.

Source: Author’s calculation basing on VLSS 92-93 and 97-98

3.8.3        Safe drinking water:

Children drinking water from well and other sources seem to have a significant higher proportion of malnutrition than children drinking water from tap. In 1993 malnutrition rate of children drinking water from well is highest at 3.1 times of from tap. In 1998 malnutrition rate of children drinking water from other sources is highest at 2.8 times of from tap.

Source: Author’s calculation basing on VLSS 92-93 and 97-98

3.8.4        Modern sanitation:

Similar to housing status and safe drinking water source, children living in house with semi-modern sanitation have obviously higher rate of malnutrition than those with modern one. It shows that the stunting rate of children with the earlier condition is 2.7 times higher than with the latter condition in 1992-93 and enlarges to 3.1 times higher in 1997-98.

Source: Author’s calculation basing on VLSS 92-93 and 97-98

If putting income, house, drinking water and sanitation together, we have a combined variable that tells us clearly how different child malnutrition rates between living conditions.  This variable is called living standard. One can see that child malnutrition proportions between living standards are really different in both years. With lower living standard, children have to suffer from more danger of getting malnourished. Thus their rate of malnutrition is higher. The data has justified that the malnutrition rate of low living standard children is 3.2 times higher than of high living standard in 1993 and 4.97 times higher in 1998.

Source: Author’s calculation basing on VLSS 92-93 and 97-98

In conclusion: (i) Child malnutrition has reduced over time but there exists child malnutrition inequality in Vietnam; (ii) There is difference in child malnutrition rates between groups in terms of age, urban-rural-mountainous area, ecological region, parents’ height, parent’s education and living standard and (iii) Difference in child malnutrition rates between living standards, determined by income, housing status, drinking water and sanitation, seems most obvious.

4.      Model, data and methodology:

4.1  Model:

This thesis takes advantage of health production function and decomposing methods, which are mentioned in theoretical framework, and using the following model to analyse child malnutrition inequality in Vietnam.

where y_i is malnutrition indicator, heigh-for-age Z score; a and b are coefficients; x_k is the k^th determinant and e_i is an error term. We assume that everyone faces the same coefficient vector, b_k.

4.2  Variables definition: Variables are grouped into three levels:

Child's level:

Child’s age: calculated by months, stretching from 1 to 60 months

Child’s sex: 1 for male and 0 for female child

Household's level:

Father and mother’s education: assessed by the highest diploma obtained by the father and mother.

House: there are two kinds: permanent and semi-permanent. The value of zero is attached to semi-permanent.

Safe drinking water: this is a dummy variable with three categories: water from tap, from well and from other resources. The based category is water from other resources.

Sanitation: also a dummy variable with modern and other types. Other types will take the value of 0.

Household consumption: this consumption per capita is calculated by thousand VND. It covers consumption for food, for daily needs and durable goods. The prices are adjusted so as to be able to compare between two periods with the origin is at January of 1998. This variable is taken logarithm to eliminate the skew of income/expenditure curve.

Commune level:

Region: this is dummy variable with 0 for rural and 1 for urban.

Tribe: this dummy variable presents for ethnic determinant with two values 1 and 0 with 1 for Kinh and 0 for others.

Religion: two categories for this dummy variables: religious and non-religious. Religious is the base with the value of 0.

4.3  Data sources:

Data source comes mainly from VLSS 92-93 and VLSS 97-98. Other sources such as GSO, NIN, WHO and WB are also useful.

In VLSS 92-93, there are 4799 households to be examined with 2173 observations of children under five. These data cover almost all regions and areas in the country as well as income quintiles. So I hope that it is good enough to represent for the whole population. Similarly there are 5999 households with 1707 observations of children in VLSS 97-98 to be researched. Both two samples are large enough to make reliable of any calculations and tests. The data of VLSS 92-93 and 97-98 are extracted and collected so as to make them compatible to be compared.

4.4  Methodology:

The thesis focuses on children under five collected from all regions in Vietnam in 1993 and 1998. Stunting Z score is used with the reference standard coming from WHO/NHCS recommendation to determine whether one child is malnourished or not.

The linear regression is run by Ordinary Least Squared method (OLS) to find out the relationship between child malnutrition and its determinants in STATA software.

Concentration index is calculated for each variable to measure inequality level.

Decomposing methods are made use to analyse the causes of child malnutrition inequality and the causes of its change overtime.

5.      Econometric results and interpretation:

5.1  Determinants of child malnutrition in Vietnam:

Eventhough household consumption is not the most important determinant because its coefficient is smaller than of religion in 1993 and sanitation in 1998, the contribution to reduce child malnutrition is seen obvious. Each 1% increase in household consumption will lead to an increase of around 0.3 units in child malnutrition indicator in both years.

Urban also affects child malnutrition by much. This variable accounts for 0.265 in 1992 and reduces to 0.192 in 1998. So a child living in an urban area will have Zhfa 0.265 and 0.192 units greater than whom living in a rural area, meaning less malnourished.

Source: Author’s calculation basing on VLSS 92-93 and 97-98

In regards to religion and ethnic determinants, they do not demonstrate a clear role though sometimes they account for significant high shares to child malnutrition in Vietnam. In 1992 coefficient of tribe is so large at 0.367 and accepted at 5% level of significance. This suggests that malnutrition level of Kinh children is 0.367 unit less than that of other ethnic minority children. Religion variable is not accepted due to insignificant p-value.

Conversely, the role of religion is confirmed in 1998 with coefficient of 0.168 and quite small p-value (p-value=0.013) while tribe variable is omitted. This means non-religious children seem to get less malnourished than religious ones. In conclusion we should be cautious about the contribution of ethnic and religion to child malnutrition in Vietnam.

Child’s age also affects nutritional status of children. The total effect of child’s age and child’s age squared in 1998 are negative, meaning during the period from 0 to 5 years, a child is more likely to be malnourished when he is older.

Father and mother’s heights are also attributable to child malnutrition even though their impacts are not so large. Coefficients of father and mother’s height are 0.037 and 0.040 in 1992-93 and changes to 0.046 and 0.033 in 1997-98. The relationship between parents and children’s height has been confirmed in some previous studies such as Haughton, D. and Haughton, J. (1997) as a genetic characteristic. So it is easy to understand that the taller parents seem to have less malnourished children.

Sanitation variable demonstrates a relative large share to child malnutrition in Vietnam. Its coefficient is 0.192 in 1992-93 and rises to 0.385 in 1997-98, showing the much larger effect on nutritional status of children over time.

The variables such as house, safe drinking water in theoretical also determine the level of child malnutrition. With better conditions such as permanent house and water from tap or well, the probability to get stunted of a child is low. However since the p-values of these variables are insignificant, I have no conclusion about their roles.

Total effect of mother's education is positive, meaning when educational level is higher, it helps provide more necessary knowledge about life for the mothers and that makes the rate of child malnutrition fall.

Among assumed variables, the role of child sex is hardly seen. So it is omitted from the model and will not be used in the following section to calculate concentration index.

5.2  Determinants of child malnutrition inequalities in Vietnam:

Household consumption inequality has increased during the period 1993-1998 because CI of lnex is 0.037 in 1993 and rises to 0.043 in 1998. This number is rather small compared to other variables such as moedu2 (CI=0.555) and toilet (CI=0.644) in 1998. However, look into “Contribution to C” column, we regconize that it plays the most important part to malnutrition inequality with the share being –0.052 in 1993 and –0.071 in 1998. The reason is malnutrition inequality related to household consumption is affected by three elements: coefficient, average household consumption and average value of child malnutrition indicator, stunting Z score. Eventhough CI of lnex is not very large but coefficient and mean value is reasonable higher than the others.

Source: Author’s calculation basing on VLSS 92-93 and 97-98

The next factor taking much concern is child’s age, represented by chiage and chiage2 variables. Although both their CI and coefficients are quite small, they have advantages of high mean values. Therefore they account for 0.038 and –0.026 in Contribution to C in 1993. However the net effect of child’s age is to reduce child malnutrition inequality and its value even declines in 1998 with nearly no impact on child malnutrition inequality (Contribution to C is 0.002 and –0.002). Therefore this variable is not taken into account in the following analysis. 

In regard to sanitation, we can see that it is not only inequality in modern sanitation accounting mainly for child malnutrition inequality but its relatively large impact on the variable of interest as well. Therefore its Contribution to C is rather high at –0.03 in 1998.

Other variables attributable to child malnutrition inequality are region, mother’ education and parents’ height.

The column “Change” in “Contribution to C” suggests how the changes in determinants affect the change in malnutrition inequality. Look at Table 5.2, we can see that almost variables have increasing effect on child malnutrition inequality in Vietnam. Sanitation accounts for the biggest share, 45%, while the second belongs to household consumption, 36%. So sanitation seems to have the most important role in rising malnutrition inequality overtime.

Owing to Oaxaca’s method, it is easy to realize that the role of the change in elasticity of malnutrition inequality with regards to sanitation is bigger than the role of the change in sanitation inequality in causing the change in malnutrition inequality. Moreover while rising elasticity increases child malnutrition inequality, declining sanitation inequality reduces it (Dh*C=-0.028 and DC*h=0.004).

Source: Author’s calculation basing on VLSS 92-93 and 97-98

In contrast, the role of elasticity and inequality are nearly equal in the case of household consumption (DC*h=-0.01 and Dh*C=-0.009 if we apply function 4 and DC*h=-0.009 and Dh*C=-0.01 if we apply function 5) and its proportion reveals the biggest compared to other variables. So that rising inequality in household consumption is attributable most to rising child malnutrition inequality in Vietnam during 1993-1998.

6.      Conclusions and recommendations:

6.1  Main results of the research:

There is a large set of determinants of child malnutrition in Vietnam, including household consumption, region, sanitation, child’s age, parents’ height and mother’s education. Children of poor families, those locating in rural and mountainous area with inferior living conditions, those inheriting shorter and lower educated parents have highest prevalence of child malnutrition.

Household consumption has a consistent role in reducing child malnutrition in Vietnam both in 1993 and 1998. Each 1% increase in household consumption in Vietnam will reduce the stunting Z-score of a child of around 0.3 unit in average.

Although child malnutrition has decreased overtime, child malnutrition inequality in Vietnam has increased.

Household consumption is the most important factor affecting child malnutrition inequality in Vietnam and this effect increases overtime. Its contribution to child malnutrition inequality is -0.052 in 1993 and -0.071 in 1998.

Increase in household consumption inequality is the main cause of the increase in child malnutrition inequality in Vietnam during the period 1993-1998.

6.2  Policy recommendations:

In order to reduce child malnutrition inequality in Vietnam, there is a need to implement two groups of solutions: to improve nutritional status of children and to improve living standard. The followings are policy recommendations:

Firstly, propagate and encourage every people, every national and international organization to take part in and support for children funds and programs such as Child Protection Fund, Child Schooling Grants, Smile Operation Program, Magazines and Libraries for Children, etc.

Secondly, establish and spread common education and health care services to ensure both physical and mental development for all children, such as free primary schooling for mountainous and minority ethnic children, nationwide vacination programs; vitamin A injection for children under 3 and under 6 months and program using iodine salt to prevent children from blind and goiter.

Thirdly, it worths considerations to carry out nutrition, health and other social programs to provide necessary knowledge for every people, especially experience on child bearing and feeding for the mother. This can be done with the support of telecommunication mediums or doing fieldtrips or some short-run courses in health communes, in schools or even in each village.

Forthly, by investing in social and economic infrastrutures, the Government encourages business activities of investors and economists and motivates the economy. The amount of goods and services is larger and more diversified whereas the price is relative lower. Then the living standard increases.

Finally, programs such as housing for Government staffs, housing for the poor, clean water for rural, rural urbanization and modernization are necessary for substantial development and helpful to reduce social inequalities. The thing should be keep in mind is not to create new programs but how to make them efficient because the the benefit of such programs may pour in the rich, who is might and has the right of distribution, instead of the poor.