Opportunities for Children in Vietnam

Transcription

1 Public Disclosure Authorized Public Disclosure Authorized Public Disclosure Authorized Public Disclosure Authorized Opportunities for Children in Vietnam Vietnamese Academy of Social Sciences (VASS) * The World Bank ** Abstract This paper examines opportunities for children in Vietnam by estimating and analyzing the Human Opportunity Index (HOI). Employing data from different rounds of the Vietnam Household Living Standard Survey (VHLSS), the Multiple Indicator Cluster Survey (MICS) , and the Study in Grade 5 Student Achievement in Mathematics and the Vietnamese Language in the school year, three domains of opportunity education, health, and basic housing infrastructure are investigated. The paper demonstrates both international comparisons and analyses of issues specific to Vietnam. The international comparison exercise finds a mixed picture of Vietnam s performance vis-à-vis other countries in the world. It does quite well in opportunities of school attendances and access to electricity, but falls behind comparator countries in opportunities of accesses to piped water and flush toilets. A mixed picture is also seen from results of analysis of the country s specific issues. The HOI is different from the coverage rate for almost all opportunities, thus indicating the existence of inequality in opportunities. Analysis of the relative importance of circumstances finds that education of household head, the household s level of well-being, and its regional location are consistently the most important determinants of the dissimilarity of allocation for almost all opportunities. Sample estimation of regional HOI indicates that there is a significant degree of heterogeneity across regions with regard to the opportunity of access to improved sanitation facilities. * Centre for Analysis and Forecasting (CAF) ** East Asia Region; and Poverty Reduction & Equity Department, PREM Network

2 Acknowledgements This report was prepared by Vu Hoang Dat of the Centre for Analysis and Forecasting (CAF), Vietnamese Academy of Social Sciences (VASS) under direct supervision of Nguyen Thang (CAF) and with research assistance and logistics support from Nguyen Thi Thu Phuong and Vu Van Ngoc (CAF). Throughout the research process, the CAF team received technical guidance and detailed comments on various drafts by Alejandro Hoyos Suarez and Ambar Narayan (PREM Poverty Reduction and Equity Group, the World Bank) and Nguyen Nguyet Nga (the World Bank in Vietnam). The team received data from Tran Thi My An (the World Bank in Vietnam) and the Social and Environmental Statistics Department of the General Statistical Office of Vietnam (GSO), and substantive comments from Laure Pasquier-Doumer (Development, Institutions and Globalization Unit (DIAL) of the French Research Institute for Development (IRD) and Paris Dauphine University), and numerous participants of a technical workshop held in Hanoi on March 8, 212, to discuss the draft report. Financial support for this work came from the World Bank, including resources from a Nordic Trust Fund grant for the project Equality of Opportunity and Economic and Social Rights. 1

3 Table of Contents ACKNOWLEDGEMENTS... 1 TABLE OF CONTENTS... 2 TABLES AND FIGURES... 3 LIST OF ACRONYMS AND ABBREVIATIONS INTRODUCTION THEORETICAL BACKGROUND SPECIFIC OPPORTUNITIES, CIRCUMSTANCES, AND DATA OPPORTUNITIES DATA SOURCES CIRCUMSTANCES INTERNATIONAL COMPARISON HUMAN OPPORTUNITY INDEX: ISSUES SPECIFIC TO VIETNAM ESTIMATION OF THE HOI CHANGES IN THE HOI OVER TIME RELATIVE IMPORTANCE OF CIRCUMSTANCES MOST AND LEAST VULNERABLE GROUPS WITH REGARD TO OPPORTUNITIES REGIONAL HOI CONCLUSIONS AND POLICY IMPLICATIONS REFERENCES APPENDIX 1 SUMMARY STATISTICS OF THE COVERAGE RATES AND THE HOI APPENDIX 2 OTHER HEALTH OPPORTUNITIES APPENDIX 3 THE SHAPLEY S DECOMPOSITION

4 Tables and Figures TABLE 1 LIST OF OPPORTUNITIES IN EACH DOMAIN AND CIRCUMSTANCE... 1 TABLE 2 SPECIFIC OPPORTUNITIES IN EACH ASPECT OF THE HEALTH DOMAIN TABLE 3 VULNERABILITY PROFILE FIGURE 1 A SIMPLE GRAPHICAL INTERPRETATION OF HOI... 7 FIGURE 2 HOI FOR SCHOOL ATTENDANCE OF CHILDREN AGES FIGURE 3 HOI FOR ACCESS TO ELECTRICITY FIGURE 4 HOI FOR ACCESS TO PIPED WATER FIGURE 5 ACCESS TO FLUSH TOILETS FIGURE 6 HOI ESTIMATION FOR EDUCATION... 2 FIGURE 7 ACCESS TO HOUSING INFRASTRUCTURE FIGURE 8 HEALTH OPPORTUNITIES FIGURE 9 DECOMPOSITION OF CHANGES IN HOI FOR EDUCATION OPPORTUNITIES OVER TIME FIGURE 1 DECOMPOSITION OF CHANGES IN HOI FOR HOUSING INFRASTRUCTURE OPPORTUNITIES OVER TIME... 3 FIGURE 11 RELATIVE IMPORTANCE OF CIRCUMSTANCES FIGURE 12 RELATIVE IMPORTANCE OF CIRCUMSTANCES AND OPPORTUNITY OF EDUCATION QUALITY FIGURE 13 RELATIVE IMPORTANCE OF CIRCUMSTANCES AND HEALTH OPPORTUNITIES FIGURE 14 RELATIVE IMPORTANCE OF CIRCUMSTANCES AND NUTRITION STATUS FIGURE 15 REGIONAL HOI AND ACCESS TO IMPROVED SANITATION FACILITIES... 4 FIGURE 16 REGIONAL HOI AND SCHOOL ATTENDANCE OF CHILDREN AGES

5 List of Acronyms and Abbreviations 27 Grade 5 Assessment BCG CAF DPT D-index DIAL EVN GDP GSO HOI IRD LAC MICS MOET MOLISA NIN OECD PISA PREM UNDP UNICEF USD VASS VHLSS WHO Study in Grade 5 Student Achievement in Mathematics and Vietnamese Language in the school year Bacillis-Cereus-Geuerin Centre for Analysis and Forecasting Diphtheria Pertussis Tetanus Dissimilarity index Development, Institutions and Globalization Electricity of Vietnam Gross Domestic Product General Statistical Office Human Opportunity Index French Research Institute for Development Latin American and Caribbean countries Multiple Indicator Cluster Survey Ministry of Education and Training Ministry of Labor, Invalids, and Social Affairs National Institute of Nutrition Organization for Economic Co-operation and Development OECD Program for International Student Assessment Poverty Reduction and Economic Management United Nation Development Program United Nations Children's Fund United States Dollar Vietnam Academy of Social Sciences Vietnam Household and Living Standard Survey World Health Organization 4

6 1. Introduction Vietnam has made impressive achievements in socioeconomic development since the country embarked on Doi Moi reforms. Notably, according to the World Bank s World Development Indicators, the country s per capita GDP rose substantially to more than US$1,2 in 21, from a low of less than US$1 in 199. As a result, Vietnam has crossed the World Bank s US$1, per capita income threshold to officially become a lower middle-income country. Non-income indicators also have shown considerable improvement, particularly in the provision of basic services for the country s entire population. Specifically, 96 percent of households had access to electricity by the end of 21 (EVN, 21) and 95 percent of children of primary school age attended primary school or higher in 26 (GSO, 26). However, there are areas where achievements are not as impressive. For instance, more than one-third of children under 5 suffered from stunted growth, while 38 percent of children under 2 were not fully immunized in 26 (GSO, 26). Furthermore, different groups of the population have benefited from economic growth to different degrees. While the poverty rate of the Kinh-Hoa group was less than 1 percent in 28, more than half of ethnic minority people (other ethnics groups different to Kinh-Hoa) still lived in poverty that year (VASS, 21). In 26, only 4 percent of children between 12 and 15 in the Red River Delta failed to complete primary school, as opposed to 31 percent of those in the North West (MOLISA and UNICEF, 28). These pieces of evidence on limited and unequal access of different groups of the population to certain types of basic services, along with recent work by the World Bank on the Human Opportunity Index, have motivated this study to investigate opportunities basic services that society considers must be universal, critical for people to improve and fully realize their human capital in multiple dimensions, and essential to determining whether a person will be able to live a life of her choosing. This index incorporates both the availability of the opportunities to the whole population and the inequality of allocation across groups with their own inherent circumstances characteristics that an individual is born into and has no influence over (gender, place of birth, ethnicity, etc.). This paper is organized as follows: After this introduction, section 2 briefly describes the theoretical background to be used in subsequent analyses. Section 3 discusses opportunities and circumstances under investigation and data sources. Section 4 presents results of international comparison. Section 5 discusses results of analyses of issues specific to Vietnam. Section 6 concludes. 5

7 2. Theoretical Background 1 The World Bank s World Development Report 26 indicates that 1 percent of difference in economic status between two families would predict about 4 to 7 percent of the difference in the families next generation. This evidence implies that transmission of inequality exists from generation to generation, and that household backgrounds would significantly affect opportunities of children when they reach adulthood. Moreover, the report also indicates a positive association between equality and efficiency at the economy-wide level: the higher the level of equality within an economy, the higher its efficiency. Meanwhile, opportunities the basic goods and services that are accepted to be available to all population groups to help them attain their potential capacities are quite far from being completely universal, especially in developing countries. Given the above- mentioned evidence, it is an appealing argument that access to basic services should be equal across subgroups of the population. This means that the circumstances people are born into, but not responsible for, should not affect the likelihood of their access to services. However, this principle is rarely met in practice, as there are not enough resources to provide services for all. The measure of coverage is used to evaluate the status quo and track progress with regard to availability of basic goods and services over time. This measure, however, could fail to take into account a very important aspect related to unequal access across different population groups. While incidence analysis could deal with this issue partially, it proves ineffective when it involves a number of criteria for grouping that are often associated with one another. For example, the ethnic minority group may also have a lower level of educational attainment, and therefore the low access of this group to basic goods and services may be caused not only by the ethnic characteristic itself but also by their low education. Thus, other measures are needed to monitor both coverage and distribution of access to basic goods and services across population groups, each of whom has specific inherent circumstances. To this end, the World Bank has recently developed the Human Opportunity Index (HOI). The HOI measures how far a society is from universal provision of basic services and goods, and the extent to which those goods and services are unevenly distributed across circumstance groups. To put it differently, the HOI takes into account both coverage of and inequality in access to a specific good or service (opportunity). Measuring HOI Primarily, the HOI is defined as the overall coverage rate of the opportunity (C) subtracted by a penalty for inequality (P). HOI = C P. P is constructed with a primary property that it is zero 1 This section is adapted from Molinas et al. (21), Barros et al. (29) and Abras et al. (211) 6

8 when there are no coverage differences between circumstance groups and it is positive and increases when differences (or inequalities) in coverage between groups increase. As such, HOI reflects both coverage and inequality in access. Alternatively, HOI could be expressed as: HOI = C (1 D) with D = (P / C), with D being the dissimilarity index or the D-index in short. The D-index reflects the proportion of existing opportunities that are misallocated across circumstance groups under the principle of equality of opportunity. Error! Reference source not found. Figure 1 illustrates the meaning of these concepts. Figure 1 A Simple Graphical Interpretation of HOI Source: Adapted from Molinas et al. (21); and Barros et al. (29). In Figure 1, the coverage rate is 51 percent, which means that about half of the pupils finish sixth grade on time. However, the probabilities are different for pupils of different circumstances, and 9 percent of the opportunities are allocated unevenly, thus the HOI (blue region) is only 42 percent. Empirically, procedures for estimating HOI are relatively straightforward and can be briefly described as follows. One first estimates a logistic model with the opportunity under investigation, for example access to clean water, as the dependent variable and circumstances as independent variables. Predicted probabilities of access,, for individual children are then obtained from the estimated logistic model. The coverage rate is then calculated as:, where w j are sampling weights. The D-Index,, is given by: 7

9 The HOI is then calculated using the above formula: The HOI has a number of important properties. First, it is an inequality-sensitive coverage rate in the sense that it is improved when inequality decreases without changes in number of opportunities in a society, or when the number of opportunities increase given no change in inequality. Second, it is consistent with Pareto efficiency, i.e. when coverage of a group increases without harming that of other groups, the HOI increases. Third, if the coverage of all groups equally increases by a proportion, the HOI increases by the same proportion. Changes in HOI over time can be decomposed into three components: 1) Composition effect, which reflects changes in HOI due to changes in distribution of circumstances; 2) Scale effect, which reflects changes in the coverage rate of all groups; 3) Equalization effect, which reflects changes in likelihood of access of vulnerable groups groups with lower-thanaverage coverage rate. The interpretation of these components is straightforward. For example, a positive change in the composition effect implies that the circumstances have become more favorable, thereby raising the HOI. There are some caveats that are important to take into account when interpreting the results. First, the HOI is not sensitive to inequality within vulnerable groups (or within non-vulnerable groups). Second, the HOI is not subgroup-consistent, which means that the HOI for the entire population cannot be estimated by aggregating HOIs of population subgroups. This caveat can be avoided using an alternative measure of the HOI, namely the geometric HOI. Third, the HOI is a function of the set of circumstances considered. However, there is a built-in property in the HOI that helps to mitigate this problem: The HOI will not be higher if more circumstances are added to the existing set of circumstances. In other words, the estimation of the HOI with a particular set of circumstances is an upper bound of the real HOI. Since the HOI was first published in 29, the index has been applied to Latin American and Caribbean countries (LACs). In the first report for LACs in 29, the HOI for five basic goods and services are estimated, including access to clean water, sanitation facilities and electricity, completion of sixth grade on time 2 and school attendance for children ages 1 to 14. In the second report for these countries in 21, the quality of education based on the OECD Program for International Student Assessment (PISA) is added to further differentiate between the LACs and developed countries. There are also a number of ongoing reports for African countries that extend analysis to include some health indicators on nutrition and/or immunization. However, important aspects of child development such as safe life or mental development opportunities are not found in any reports, possibly due to the lack of relevant data. Thus, the selection of basic goods and services to be included in analysis depends to a considerable extent on the objectives of the study at hand and also the availability of data. 2 This is the last grade of primary education in most Latin American countries. 8

10 Another feature of HOI reports is that they focus solely on children. One may presumably offer several reasons. First, by limiting analysis to children, one can avoid a technical challenge associated with dealing with the issue of efforts and choice because circumstances being studied are normally beyond children s control. Second, there is a general perception that opportunities in terms of access to basic services during childhood are essential for raising the human capital of children, which in turn is critical for them if they are to attain decent jobs in the future. Furthermore, as improving equality in human capital is more difficult than raising income (Bhushan et al., 21) and employment is the main, sometimes even the only, source of income for poor and low-income people, improving the human capital of children in poor households is an effective way to break the intergenerational transfer of poverty. 3. Specific Opportunities, Circumstances, and Data It is not possible to cover all potential opportunities or circumstances in practice. Therefore, in this section, we discuss in detail opportunities and circumstances that are analyzed in the current report. Data sources that are used for the analyses are also briefly described Opportunities To select the basic goods and services to be analyzed, we start from common domains of basic goods and services that are used in other studies for LAC and African countries, including education, infrastructure and health. Individual opportunities within each domain are then selected on the basis of two considerations (i) data availability, and (ii) the relevance for the current context in Vietnam. The left column of Table 1 presents specific opportunities to be investigated. A child is defined as attending school if he or she had gone to school within 12 months prior to the survey. The opportunities of the two groups of children, ages 7 to 11 and 12 to 15, are estimated separately. We use the age of 7 for the first group of children in estimating the attending school opportunity in order to avoid the effects of survey-time lags on the schooling status of children 6 years old. For instance, if the survey were conducted in June, 6-year-old children would be reported to be not in school if they did not attend any pre-primary school. More importantly, if the survey was conducted at different times for different regions or between urban and rural areas, since the geographic location is included in analysis as a circumstance, the results could be biased against certain regions. Finishing primary school on time is calculated for children between 13 and 14 years of age, and is defined as completing grade 5 the last grade in primary school in Vietnam s education system. We consider these opportunities as measures of quantity. As will be discussed later, Vietnam has performed quite well in providing educational services by standard measures. However, these measures cannot capture the quality of the service, which has recently been a topic of policy debate in Vietnam. In an attempt to fill these gaps, this report 9

11 takes an additional step to investigate an opportunity of the quality of education. An appealing argument for taking into account the quality of education, as well as suggestions for measuring this quality, is found in Molinas et al. (21), according to which, the measurement of coverage of basic services may be alternatively restricted to only those who have access to goods and services provided that outcome of such an access is higher than a required minimum level. Specifically, the authors note: The best practical measure of effective access to quality education could be the proportion of children of a given age with learning proficiency above a minimum level. Put differently, the pupils outcome is a good measure for studying the quality of education. We shall discuss in detail how we specify this opportunity later in this report. Table 1 List of Opportunities in Each Domain and Circumstance Domains Circumstances Education School attendance of children ages 7-11 School attendance of children ages Completion of primary school on time Ability to advance to lower secondary school without help for children in the last grade of primary school. Health Nutrition opportunities Immunization opportunities Prenatal care and services for birth-givers Basic housing infrastructure Access to electricity Access to clean water Access to sanitation facilities Gender of children Gender and education of household head Number of siblings ages -16 years in the household Whether household head is living with the spouse Expenditure per capita (log for international comparison and quintiles for the country s specific analysis) Living areas (urban vs. rural) Regions (country-specific analysis) Ethnicity (country-specific analysis) Quintiles of wealth index (for the quality of education and health opportunities) Highest education of parents (for the quality of education opportunity) Circumstances of the school (for the quality of education opportunity) Three aspects of the health domain are used for analysis: nutrition, immunization, and access to services of prenatal care and birth delivery of mother. Specific opportunities and their definitions in each aspect are given in Table 2. There are a number of opportunities that are investigated in other reports, such as whether the household has a bed net or children without chronic illness, but these opportunities are not relevant in the current context of Vietnam. 3 3 Very few households report not having a bed net and having no bed net is more of an urban phenomenon in Vietnam, which is not necessarily a bad thing, thus it is more or less dependent on choice of households. Meanwhile, less than 2 percent of children ages 6-16 suffer from a chronic illness (authors own estimation based on data from VHLSS 26). 1

12 Table 2 Specific Opportunities in Each Aspect of the Health Domain Opportunities Age group and definition Nutrition Children ages -5 Stunting (chronic undernutrition) Underweight (low weight for age) Wasting (low weight for height) Prenatal care and services for birth-givers Receiving antenatal care by skilled persons Skilled attendant at delivery Immunization Children ages 1-5 Immunization against tuberculosis (BCG vaccine) Complete immunization against polio Complete immunization against DPT Immunization against measles Complete immunization A child is said to be stunted, underweight, or wasted if his/her measures of height/length or weight are less than two standard deviations from the median values of the reference groups provided by the WHO Women giving births within two years of the survey start Skilled persons include doctors, nurses/midwives, elementary midwives/nurses A skilled attendant is defined the same as above Three injections Three injections All required vaccinations against six mentioned diseases Regarding housing infrastructure, three opportunities of access to clean water, sanitation facilities, and electricity are analyzed. For each of the two opportunities access to clean water and sanitation facilities two definitions are used, with one for international comparison and one for analysis of issues specific to Vietnam. For international comparison, access to clean water takes value 1 if the household s main source of drinking water is tap water either private or public and access to sanitation facilities is defined as the presence of flush toilet(s) in the house. 4 For analysis of issues specific to Vietnam, UNICEF s broader definitions of improved safe water or improved sanitation facilities are used for estimation of the opportunities. Under these definitions, improved safe water is private tap water from inside and outside the house, deep drill wells, hand-dug and reinforced wells, hand-dug, non-reinforced and covered wells, protected springs, rain water and bought water, and improved sanitation facilities refers to flush toilet, sulabh and double vault compost latrine (MOLISA and UNICEF, 28). Finally, access to electricity is defined as having direct connection to the electricity network; this definition is used for both international comparison and analysis of issues specific to Vietnam. These opportunities are studied for children ages to 16. As the age of children is frequently used in defining opportunities, it is important to clearly define the age used for estimation. Except for the case of nutrition, which requires the 4 While this indicator is relevant for the LAC in which urban population dominates, it may not be so for largely agrarian Vietnam. 11

13 population age exactly based on the date of birth, the age of children in other opportunities is calculated as the number of years from the birth year to the year of calculation, regardless of the month of the year in which the child was born. For example, the age of children born in 22 is 6 in 28. This age is known as education age Data Sources Data that are used for analysis in this report come from three main sources. The first source is the Vietnam Household Living Standard Surveys (VHLSSs) for the survey years 22, 24, 26, 28, and 21. The data is used to estimate quantitative measures of educational and infrastructural opportunities. The second data source is Multiple Indicator Cluster Survey in (MICS IV) for the estimation of health opportunities. The third data source is drawn from the Study in Grade 5 Student Achievement in Mathematics and Vietnamese Language in the school year (hereafter referred to as the 27 Grade 5 Assessment), which is used to analyze the quality of education. With a sample size of approximately 3, households for the 22 round; 9,189 households for the 24, 26, and 28 rounds; and 9,299 households for the 21 round, the VHLSSs are representative of the whole country and for either eight or six regions, 5 each is broken down further into urban and rural areas. The surveys provide rich information on household demographics, education of household members, livelihoods, and expenditure. It should be noted that the four rounds of the survey series from 22 to 28 share a common sampling frame, which is based on the 1999 population census, while the 21 round was sampled with a new sampling frame drawn from the 29 population census. With regard to sampling weights, we can access documentation describing how weights were calculated only for VHLSS 22 and VHLSS 24. The VHLSSs were sampled with a three-stage-procedure, according to which communes, enumeration areas, and households were selected at the first, second, and third stages, respectively. As a result, information on the populations of the whole country, the communes, and the enumeration areas at the time of each VHLSS round should be incorporated when the weights were estimated. However, according to Phung and Nguyen (27), only information on the population of the enumeration areas at the time of survey was updated when weights were calculated for the two rounds mentioned above, and other information was still sourced from the 1999 population census. If the information for the weight calculation was the same for the subsequent VHLSSs of 26 and 28, we can conclude that the information for weight calculation and the fixed sampling frame for the period of make the rounds in the period more representative than the sampling frame itself. Therefore, it is not fully plausible to draw comparisons between VHLSS 21 and older VHLSSs. 5 In 21, eight regions are integrated into six. Name of regions are given later in the paper. 12

14 The MICS IV survey includes 3,729 children under 5 years of age and 12,115 women ages 15 to 49 from 11,874 households, and as such, the survey adequately represents the whole country and six regions, with each being broken down into urban and rural areas. The 27 Grade 5 Assessment conducted by Ministry of Education and Training (MOET) in 27 is a test-based survey of nearly 6, pupils in all provinces of the country. There are two separate tests, one for mathematics and one for Vietnamese, each of which includes 4 multiplechoice questions. The assessment procedures are somewhat similar to those of PISA. One of the outcomes of the 27 Grade 5 Assessment is that pupils at grade 5 are classified into three groups: pre-functional, functional, and able to advance to lower secondary school without help in mathematics and the Vietnamese language separately. Pupils are classified as prefunctional in mathematics or the Vietnamese language if they do not meet minimum standards perceived to be required for applying corresponding skills into daily life. As a result, these pupils are not allowed to study these subjects at the starting grade grade 6 of lower secondary school. Functional pupils are those who meet the minimum standards but are believed to need some help in grade 6. The last group includes pupils who can manage the subjects in grade 6 without help. The thresholds for classifications as well as the level of difficulty of individual questions posed in each subject are constructed and evaluated by two panels of education specialists. The evaluation procedures are independent of the results assessment process. The first benchmark may be comparable to the second level of proficiency in the PISA, which is commonly regarded as basic skills of the subjects needed in the daily life. As mentioned earlier, in this study, results of the 27 Grade 5 Assessment will be used as the best practical measure of effective access to quality education. Furthermore, as Vietnam strives to achieve universal lower secondary education, a primary education with adequate quality should be able to provide pupils by grade 5 the last grade of primary school with necessary skills to move up to lower secondary school. Therefore, we analyze three opportunities: ability to advance to lower secondary school without help in each of the two subjects separately mathematics and the Vietnamese language and both subjects simultaneously. Another issue with the 27 Grade 5 Assessment is nonrandom missing values. The household background information is provided by pupils themselves. As grade 5 pupils may vary in awareness of their backgrounds such as with regard to their household assets, their parents education, and so on; one may raise a question of a negative association of children s capacity and missing values. We have performed a check and found out that the relationship between the missing values and achievements of children are not random. Therefore, we include a series of dummy variables to capture the missing values of each circumstance. For example, the missing value dummy variable for parental education takes 1 if the parental education is missing. This series of dummy variables are then included as circumstances in the HOI estimation. By this 13

15 inclusion, we can retain all observations to keep the representativeness as well as have more precise effects of other circumstances on the opportunities Circumstances The construction of circumstances starts with the gender of children, the gender and level of education of household heads measured in terms of degrees obtained, the number of siblings ages to16 in the household, whether the household head is living with a spouse, and the per capita expenditure of the household. These circumstances are also used in the above-mentioned World Bank studies for LAC countries, thus facilitating international comparison between Vietnam and those countries. This set of circumstances is extended to facilitate analysis of issues specific to Vietnam. The set varies across opportunities depending on availability of information collected in the surveys. For opportunities estimated based on data from VHLSSs, two circumstances of the ethnicity of the household head and regional location are added. Although there are 54 ethnic groups in Vietnam, due to limited sample size of VHLSSs, we classify the population into only two groups: Kinh and Hoa as the majority group, and other ethnic groups as ethnic minorities. 6 As per common classification, there are eight geographic regions in Vietnam, 7 and they are used as a circumstance to capture differences in opportunities across regions. Finally, per capita expenditure is divided into quintiles. The same set of circumstances, with the exception of the one regarding per capita expenditure, is used when HOI is estimated with MICS IV data. Data on expenditure are not available in the MICS surveys; consequently, we use the wealth index as an alternative measure of household well-being. The household wealth indexes of children and women are also classified into quintiles. The 27 Grade 5 Assessment also includes background information on households of assessed pupils. We first try to define circumstances to be as close as possible to those applicable to estimations based on data of the VHLSSs or the MICS IV. Therefore, gender, ethnicity, living areas, and regional location of the pupils are the same as those defined above. As we cannot define the household head, the maximum level of education of parents or guardians is used 6 Kinh (Viet) is the majority group in Vietnam, accounting for about 86 percent of the country s population. Hoa is actually a minority group but their level of development is comparable to that of the Kinh group and therefore the two are commonly grouped together in analytical work. Other ethnic minorities are found to lag considerably behind as compared to the Kinh and Hoa group. 7 The eight regions are North East, North West, and Red River Delta in the north; North Central Coast, South Central Coast, and Central Highlands in the center; and South East and the Mekong River Delta in the South. Since 21, the North East and the North West have been consolidated into one region, Northern Uplands. North Central Coast and South Central Coast have been combined into a single region, Central Coast. As a result, there have been just six regions since 21. However, we retain eight regions for analyses that used VHLSS

16 instead, and gender of household head is not included in analysis. Quintiles of the wealth index 8 are also used as a measure of well-being, similar to the approach in estimations based on MICS IV data. In addition, we include travel time to school of more than 3 minutes as a circumstance at the children s level. Other circumstances involving teachers and schools are clearly relevant as we are concerned with the quality of education. However, data availability does not allow us to include them in the analysis. Specifically, information on teachers is available for only 57 percent of pupils assessed. Although the selection of teachers is random within a school, the probability of being included in the assessment varies among teachers from different schools as it depends on the number of teachers in the school. As a consequence, estimations will be biased if we use only the subsample of pupils with information on their teachers (Vietnam National Institute for Education Science, 28). Therefore, it is possible to include only some circumstances at the school level, those where information is available for all pupils in the assessment. Specifically, the ethnicity and qualification of the headmaster in terms of both education and pedagogical training, the nature of the school such as being a mixed school of both primary and lower secondary education, the number of teachers per class, and the school s access to electricity are the circumstances at the school level that are used for analysis. 4. International Comparison In this section, we compare HOI for four basic services in Vietnam with those of Latin American and Caribbean (LAC) and African countries for which corresponding HOI figures are readily available. These HOIs include school attendance of children ages 1 to 14, as well as access to electricity, clean water, and sanitation facilities defined in a way suitable for international comparison as noted above. The circumstances include the gender of children, the gender and the level of education of household heads, the number of siblings ages to 16, whether the household head lives with a spouse, and the per capita expenditure of the household. These circumstances are also used in the World Bank s studies on LAC countries. Although data of VHLSS 21 are already available, we use the VHLSS 26 data for analysis in order to ensure comparability with HOI figures readily available for the LAC countries. HOIs for Vietnam for the four basic services compared to those of LAC and African countries are presented in Figures 2, 3, 4, and 5 respectively. 8 We employ the same method of wealth index estimation as it is applicable to MICS survey data. See Rutstein and Johnson (24) for detailed description of the method. 15

17 Figure 2 HOI for School Attendance of Children Ages 1-14 HOI (percent) Niger Mali Ethiopia Senegal Liberia Nigeria Sierra Leone Mozambique Madagascar Congo Rep. Dem. Cameroon Tanzania Rwanda Ghana Guatemala Malawi Honduras Zambia Ecuador Zimbabwe Uganda Nicaragua El Salvador Paraguay Vietnam Namibia Panama Kenya Colombia Mexico Uruguay Costa Rica Dominican Rep. Jamaica Peru Venezuela Argentina Brazil Chile Source: Authors calculations based on data from VHLSS 26 and the World Bank. There is a mixed interpretation of Vietnam s performance vis-à-vis the comparator countries. Vietnam does well in school attendance and access to electricity. Specifically, HOI for school attendance is higher than that of most African countries and several LAC countries, while the HOI for access to electricity is higher than all African countries and only slightly lower than most LAC countries. Notably, the HOI for access to electricity is only about 45 percent for Ghana, the best performing African country in the sample, as opposed to more than 9 percent for Vietnam. Figure 3 HOI for Access to Electricity HOI (percent) Rwanda Uganda Malawi Ethiopia Tanzania Niger Mozambique Sierra Leone Madagascar Kenya Congo Rep. Dem. Mali Zambia Zimbabwe Namibia Senegal Cameroon Nigeria Ghana Honduras Nicaragua Panama Peru Guatemala El Salvador Vietnam Colombia Ecuador Paraguay Dominican Rep. Jamaica Brazil Costa Rica Uruguay Venezuela Argentina Chile Mexico Source: Authors estimation based on VHLSS 26 and the World Bank. However, Vietnam does not perform well in terms of access to piped water. The HOI for access to flush toilets is somewhat better but is still relatively low. Vietnam s HOI for access to piped water is higher than only some African countries with low performance in this area, and it is lower than all LAC countries. The HOI for flush toilets is in the middle of the whole range of 16

18 African and LAC countries. However, Vietnam falls considerably behind top-performing countries in both of these basic services. Figure 4 HOI for Access to Piped Water HOI (percent) Liberia Nigeria Uganda Malawi Niger Zambia Mozambique Sierra Leone Mali Vietnam Congo RD Ethiopia Zimbabwe Madagascar Kenya Cameroon Rwanda Tanzania Ghana Nicaragua Peru Senegal El Salvador Paraguay Jamaica Vietnam urban Bolivia Namibia Dominican Republic Guatemala Ecuador Honduras Costa Rica Panama Mexico Uruguay Venezuela, R.B. de Argentina Brazil Chile Source: Authors estimation based on VHLSS 26 and the World Bank. However, access to piped water and flush toilet is more relevant in urban areas. The better performance of LAC countries in these two aspects thus can be partially explained by the high proportion of urban population, which reaches 89 percent in Chile (the country with the highest HOI for clean water and sanitation facilities). For a better comparison with LAC countries, we also estimate the HOI for these two opportunities for urban Vietnam only. These estimates are also presented in Figure 4 and Figure 5. Figure 5 shows that urban Vietnam compares comfortably with LAC countries in terms of HOI for access to flush toilets, being outperformed by only a few top countries. With regard to access to piped water in urban areas, Vietnam is in the top half of the whole sample, but falls behind most LAC countries and the African nation of Namibia. This may highlight a priority area for urban planning in Vietnam. 17

19 Figure 5 Access to Flush Toilets HOI (percent) Uganda Rwanda Niger Mozambique Malawi Tanzania Madagascar Sierra Leone Mali Ethiopia Congo RD Cameroon Kenya Zambia Ghana Liberia Nigeria Zimbabwe Nicaragua Namibia El Salvador Vietnam Senegal Bolivia Guatemala Honduras Panama Jamaica Paraguay Mexico Dominican Republic Costa Rica Brazil Ecuador Peru Vietnam urban Uruguay Argentina Venezuela, R.B. de Chile Source: Authors estimations based on data from VHLSS 26 and the World Bank. 5. Human Opportunity Index: Issues Specific to Vietnam International comparisons as presented in the previous section help to visualize Vietnam s place among other countries. However, they do not reveal the penalty for inequality in the distribution of opportunities across circumstance groups, which is the most important property of HOI. This section looks into opportunities and circumstances that are specific to Vietnam, and attempts to shed light on five questions: (i) what is Vietnam s progress in moving toward universal access for the opportunities and their standing statuses as of 21; (ii) how important is inequality and each of the three factors composition, scale, and equalization in terms of contribution to changes in HOI over time; (iii) what is the relative importance of each circumstance or each group of circumstances in dissimilarity in access; (iv) what are characteristics of the most and least vulnerable groups with regard to an opportunity; (v) does the HOI vary across regions? 5.1. Estimation of the HOI As noted in Section 3.2, the VHLSS 21 is not fully comparable with older VHLSSs. Therefore, progresses as well as decompositions of the relative importance of composition, scale, and equalization in terms of contribution to changes in HOI over time of opportunities, of which analyses are based on data of VHLSSs, are analyzed only for the period of The VHLSS 21 is used to investigate the status quo of the opportunities as of the latest data availability. But this gap has not changed much in the case of the opportunity of school attendance of children ages 12 to 15. In Figure 6 and other figures presenting the results of HOI estimation, we also present the 95 percent confidence interval of HOI (see Appendix 1 for statistics of the confidence intervals) to 18

20 see whether HOI is statistically significantly different from coverage rate. Since 24, both coverage rate and HOI for quantity education opportunities have been high, resulting in small differences between these indicators. However, the confidence intervals of the HOI estimates are also quite small, showing statistically significant differences between the two indicators for all quantity education opportunities. Overall, the standing status as of 21 of quantity opportunities of education is relatively good. The coverage rates as well as the HOI of the opportunities of completing primary on time and school attendance of children ages 7 to 11 are both above 9 percent. Moreover, the small gaps between the coverage rate and the HOI indicate minor inequalities in accessing the two opportunities. However, the indicator showing school attendance of children ages 12 to15 is slightly less impressive. Although the coverage rate is high at 89 percent, a modest but statistically significant gap between the coverage rate and the HOI implies a certain level of inequality in accessing this opportunity across groups of different circumstances. Figure 6 a, b, c presents the coverage rate and the HOI of three education opportunities and their changes in the period and the standing status as of 21 (See Appendix 1 for statistics of the HOIs and the coverage rates). All figures show relatively good performance in education provision in terms of quantity, with discernible improvements in both the coverage rate and the HOI over the period. Of these three opportunities, the indicator on the completion of primary school on time shows most significant improvements, with the HOI rising from 84.6 per cent in 22 to 93.5 percent in 28. The school attendance of children ages 7 to 11 improves only slightly, because both the coverage rate and the HOI were already high in 22, at 95.9 and 94.1 percent respectively. The gap between the coverage rate and the HOI for the completion of primary school on time, which was discernible in 22, has steadily narrowed over time. But this gap has not changed much in the case of the opportunity of school attendance of children ages 12 to 15. In Figure 6 and other figures presenting the results of HOI estimation, we also present the 95 percent confidence interval of HOI (see Appendix 1 for statistics of the confidence intervals) to see whether HOI is statistically significantly different from coverage rate. Since 24, both coverage rate and HOI for quantity education opportunities have been high, resulting in small differences between these indicators. However, the confidence intervals of the HOI estimates are also quite small, showing statistically significant differences between the two indicators for all quantity education opportunities. Overall, the standing status as of 21 of quantity opportunities of education is relatively good. The coverage rates as well as the HOI of the opportunities of completing primary on time and school attendance of children ages 7 to 11 are both above 9 percent. Moreover, the small 19

21 gaps between the coverage rate and the HOI indicate minor inequalities in accessing the two opportunities. However, the indicator showing school attendance of children ages 12 to15 is slightly less impressive. Although the coverage rate is high at 89 percent, a modest but statistically significant gap between the coverage rate and the HOI implies a certain level of inequality in accessing this opportunity across groups of different circumstances. Figure 6 HOI Estimation for Education Figure 6.a Completing Primary School on Time 1 percent HOI Coverage HOI_low (95%) HOI_high (95%) year Figure 6.b School Attendance by Children Ages percent HOI Coverage HOI_low (95%) HOI_high (95%) year Figure 6.c School Attendance by Children Ages

22 1 percent HOI Coverage HOI_low (95%) HOI_high (95%) year Figure 6.d Quality of Education 1 8 per cent 6 4 the "best" HOI the "comparable" HOI Coverage 2 Math Vietnamese Both Subjects subject Source: Authors estimations based on data from VHLSSs and 27 Grade 5 Assessment. As the starting point, the HOI of the quality education opportunity is estimated with a set of circumstances including those of the children s school as part of an effort to generate a more complete picture of the inequality; we name this the best HOI of the opportunity. However, this set of circumstances makes the HOI incomparable with the HOI of quantity education. Thus, we also estimate the HOI of quality education opportunity with the same set of circumstances 9 as that of the quantity education opportunities, and we name this the comparable HOI. From Figure 6.d, which reports indicators on the quality of education, 1 we can observe that the two HOIs are only slightly different from one another. These results 9 Information from the 27 Grade 5 Assessment does not allow us to define exact circumstances as do those of the quantity education opportunities. For example, the wealth index is used instead of the expenditure per capita. In addition, as an effort to make the HOIs of the quality and quantity opportunities of education as comparable as possible, we also exclude missing observations when the HOI is estimated with this set of circumstances. The exclusion of missing observations increases the coverage rates by nearly one percentage point. 1 The coverage rates in the figure correspond to the best HOI. As noted in the above footnote, the coverage rates corresponding to the comparable HOI increase by nearly one percentage point. 21

23 indicate a relatively modest role of the children s school in influencing the access of children to the opportunity. Therefore, in following analyses, we focus only on the best HOI and its corresponding coverage rate. Indicators in Figure 6.d show considerably more limited availability of quality education opportunities for children: only 62.2 percent of pupils in grade 5 would be able to advance to the lower secondary school without help. As the completion of primary school does not fully signal the ability of pupils to successfully study in the lower secondary school, the considerable difference between this indicator and the indicator of quality of education reiterates a frequently raised question about quality vs. quantity in the education system in general and primary school in particular. The apparent and statistically significant difference between coverage rate and HOI indicates the existence of inequality in access to quality education between various groups, especially when outcome indicators on the ability to advance to lower secondary school without help in both subjects are considered. With regard to access to housing infrastructure, changes in the coverage rates and the HOI in the period from 22 to 28 as well as the standing status as of 21 with regard to access to electricity, improved safe water, and improved sanitation facilities are presented in Figures 7.a, 7.b, and 7.c. It can be seen that progress was made with regard to access to electricity and improved sanitation facilities, but access to improved safe water did not change much. It should be noted that it was not straightforward to collect information on improvements in the cleanliness of the water, because the codes of the question on drinking sources in the VHLSSs questionnaire changed significantly between VHLSS 22, 24 and 26. There is only one code in the VHLSS 22 for both improved and unimproved sources of water and we decided to exclude the year 22 when analyzing this opportunity. Generally, both coverage and HOI of access to improved safe water is relatively high, particularly in comparison with access to piped water only. In 21, the coverage rate was 88 percent, and the gap between the coverage and HOI, which was modest but statistically significant, indicates that circumstances affect access to the service at a certain level. For electricity, the coverage rate is over 95 percent in 21, thus revealing an area in which Vietnam did particularly well. A closer look at these indicators reveals that the coverage rate significantly improved from 22 to 24, and the gap between the coverage and HOI also narrowed down during these years. Figure 7 Access to Housing Infrastructure Figure 7.a Access to Electricity 22

24 1 Percent HOI HOI_low (95%) HOI_high (95%) Coverage Year Figure 7.b Access to Improved Safe Water 1 Percent HOI HOI_low (95%) HOI_high (95%) Coverage Year 23

25 Figure 7.c Access to Improved Sanitation Facilities 1 Percent HOI HOI_low (95%) HOI_high (95%) Coverage Year Source: Authors estimations based on data from VHLSSs. Indicators showing access to improved sanitation facilities are the worst ones among the three types of housing infrastructure. Although there was a significant progress during the period from 22 to 28 and a further improvement in 21, the coverage rate reached just about 64 percent by 21. Furthermore, the big gap between the coverage rate and HOI indicates a remarkable inequality with regard to access to this service. We now turn to health-related opportunities. As noted in Section 3, we investigate a wide range of health opportunities, using information from MICS IV. Indicators on pre-delivery and delivery care for mother, vaccination against selected basic diseases for children ages 1 to 5, 11 and nutrition status of children ages to 5 are given in 11 We exclude children under 1 year old to ensure that children under the study are old enough for all kinds of commonly required vaccines. 24

26 Figure 8.a and 8.b respectively. Although the completion of immunization requires vaccinations against six diseases, Figure 8.a does not present all of them; it shows only the complete immunization and immunization against measles, which requires one injection, and polio, which requires three injection. These indicate patterns of the coverage and HOI with regard to other diseases that are very much similar to these two diseases (see Appendix 2 for indicators on other diseases). 25

27 Figure 8 Health Opportunities Figure 8.a Care for Mothers and Immunization for Children Ages percent HOI Coverage HOI_low (95%) HOI_high (95%) Receiving antenatal care by skilled person Skilled attendant at delivery Immunization against Measles opportunity Complete immunization against Polio Complete immunization Figure 8.b Nutrition for Children Ages percent HOI Coverage HOI_low (95%) HOI_high (95%) No stunting No underweight No wasting opportunity Source: Authors estimations based on data from MICS IV. The results presented in Figure 8.a indicate that the coverage rate of receiving antenatal care and assistance at delivery by skilled persons are high, at 94 and 93 percent respectively. The coverage rate of immunization against measles is also high, at 92 percent. However, the indicator on the completion of immunization against polio is considerably less impressive, at just less than 71 percent. This may partly be accounted for by the difference in the number of injections required for complete vaccination against these diseases. Specifically, immunization against measles requires only one injection, compared with the three injections required for polio. There might also be a possibility of underreporting complete immunization against the 26

28 latter disease because a large proportion of vaccination cards are kept at communal health stations. As a consequence, this particular question in the survey often was answered purely based on the mother s recollection, which could be inaccurate as the mother might miss one or two dates of injection or vaccination against the disease (GSO, 26). The coverage rate of the complete immunization opportunity is, as expected, lower at approximately 61 percent which might provide an underestimation for the reason noted above. However, this relatively low coverage rate could not be completely attributed to the issue of recording methods. Therefore, it should be a priority to improve the recording methods first. Subsequently, implementation of the immunization program should be enhanced. Among the above-mentioned health-related opportunities, there are modest inequalities in the opportunities of skilled attendance at delivery and immunization against polio, as we can observe certain gaps between the coverage rates and the HOIs. The gap is more obvious and so is inequality in the opportunity of complete immunization. As observed in Figure 8.b, status varies across different nutrition opportunities for children under 5. The coverage rate of no acute undernutrition (wasting) is very high, at 95.9 percent, but that of no chronic undernutrition (stunting) is significantly lower, at 77.3 percent. Our results here are quite different from figures of the National Institute of Nutrition (NIN), 12 the official organization in charge of tracking the nutrition situation in Vietnam. They report that the non-stunting rate of children ages to 5 was only 7.7 in 21. In more detail, this figure lies outside the 95 percent confidence interval of our estimates. 13 Indeed, we are puzzled by the nutrition data reported by NIN. It is hard to explain a sine-shaped variation in the stunting rates of children under 5 years old in the period from 1999 to As the stunting status measures the long-term development of children, it would be reasonable to expect a smoother trend, especially because there has been a strong downward tendency on the rate in the period. In addition, there are conflicts in figures between NIN s two sources of information titled Statistics of children s nutrition over years 15 and Decreasing trends of malnutrition of mothers and children under 5 years old during the 2-29 period. 16 Specifically, the stunting rate of children was 29.5 percent in both sources in 25. However, the rates for 26 differed considerably across these sources, 31.9 percent in the former and 35.2 percent in the latter. We are not in a good position to evaluate the creditability of NIN s figures because we are unable to access the documentation that was used to estimate these different figures. On the 12 National Institute of Nutrition and UNICEF (212) 13 The 95 per confidence interval of our estimation of non-stunting is between 75.8 per cent and 78.9 per cent 14 A graph with figures for stunting and underweight rates of children aged to 5 years in Vietnam is available at: [accessed: 1 May, 212]. 15 See note 14 above. 16 Available at: [accessed: 1 May 212] 27

29 other hand, the MICS IV survey has followed the same procedure as the VHLSSs, which are quite well-known for their creditability. However, our analyses relating nutrition with MICS IV data in the current report should be taken as indicative only as long as the differences between our figures and those of NIN are not fully explained. Coming back to our results, the non-stunting coverage rate of 77.3 percent is even slightly higher than the official objective of the country being set for the year 22 at 77 percent. 17 However, the gap between the coverage rate and the HOI of the opportunity is discernible and statistically significant, indicating a certain degree of inequality in the opportunity. The gap, however, is statistically insignificant for the no-wasting opportunity, implying that circumstances do not affect the access to this opportunity Changes in the HOI Over Time As discussed in the theoretical background section, changes in HOI over time can be decomposed into changes in (i) composition, (ii) scale, and (iii) equalization. In this subsection we employ this decomposition technique to analyze the relative importance of these components in terms of their contribution to changes in HOI in each of three two-year sub-periods of the period. The rounds of the VHLSS series in the period, as noted in section 3.2, well represent a unique population. Therefore, they provide a good sample for this decomposition exercise. However, VHLSS 21 used a different sampling frame and therefore it is not comparable with the older VHLSSs. As a consequence, it is not possible to combine the VHLSS 21 and older VHLSSs for the decomposition as the exercise employs an assumption that each of the surveys at the two points in time represents a unique population. The results of the previous subsection should also be recalled that the change in HOI for school attendance of children ages 7 to 11 is very modest; therefore for this particular indicator, the decomposition would be unnecessary. It should be noted again that the study period for access to improved safe water is 24-28, which means that the year 22 is left out. 17 Decision No. 226/QD-TTg of the Prime Minister, dated February 22,

30 Figure 9 Decomposition of Changes in HOI for Education Opportunities Over Time Figure 9.a Completion of Primary School on Time 6 Figure 9.b School Attendance of Children Ages percent percent Composition Scale Equalization Source: Authors estimation based on data from VHLSSs. Composition Scale Equalization The contribution of each component to the improvements of HOI for the two remaining education opportunities is given in figures 9.a and 9.b. The biggest improvements were observed, as noted in the previous section, in the sub-period from 22 to 24. Progress on the completion of primary school in due time was steady in the sub-period from 24 to 28, with the scale effect taking the lead in driving the improvement. Given the very high coverage rate of 96 percent in 28, priority regarding this opportunity should be given to maintaining the current pace. However, there are signs of concerns with regard to the school attendance rate of children ages 12 to 15. Discernible progress was observed only in the sub-period from 22 to 24, while some regress was seen in the 26-to-28 sub-period. This may indicate that achieving the goal of universal completion of lower secondary school in the coming years will be a very challenging task. Figures 1.a and 1.b present results of the decomposition exercise for access to electricity and sanitation. As with education, the most striking improvement of HOI for access to electricity was achieved in the sub-period from 22 to 24, estimated at more than 1 percent. Progress for this opportunity is also observed for the sub-periods of 24 to 26 and 26 to 28, but the extent or magnitude of improvement was remarkably smaller. Both scale and equalization effects have contributed to the progress of this opportunity, with the former playing a larger role than the latter, thus indicating that both coverage of and equality across groups have been improved. Figure 1.b indicates that progress in access to sanitation was steady in the period of 22-28, being largely driven by the scale effect and to a lesser extent by the equalization effect. This implies that with respect to access to electricity, both coverage of all groups and equality across groups were improved. This means that policy should aim to simultaneously create this opportunity for all groups and maintain the pace of equalization. 29

31 Figure 1 Decomposition of Changes in HOI for Housing Infrastructure Opportunities Over Time Figure 1.a Access to Electricity Figure 1.b Access to Improved Sanitation Facilities percent percent Composition Scale Equalization Composition Scale Equalization 6 Figure 1.c Access to Improved Safe Water 4 percent Composition Scale Equalization Source: Authors estimation based on data from VHLSSs. As shown in Figure 1.c, access to improved safe water actually diminished slightly in the subperiod from 24 to 26. As noted earlier, the coding of water sources underwent drastic changes between 24 and 26, making it problematic to analyze data on this indicator. The codes, however, did not change further between 26 and 28, so the result for this period is a reliable sign that significant progress was made between 26 and 28, thanks to positive effects of scale and equalization Relative Importance of Circumstances We have found that there is a certain level of inequality in access to certain basic goods and services. A related question that arises is which circumstances play a more important role in determining the dissimilarity, i.e. the relative importance of contribution of misallocation to the overall coverage? The answer to this question could indicate priorities of intervention for those 3

32 who want to reduce inequalities across various groups of the population. In this subsection, we do a decomposition exercise following Shapley s idea of relative importance of various factors (see more details on the Shapley s decomposition in Appendix 3). The decomposition provides us with relative role of each or a group of circumstances in affecting the inequality. It should be noted, however, that low contribution of one circumstance or a set of circumstances should not be interpreted as a sign that these are not relevant for the opportunities of the child, but rather that these circumstances alone do not determine the inequality of opportunities observed in the data (Abras et al., 211). Figure 11.a presents results of decompositions for six opportunities sharing the same set of circumstances in 21. We also recall the dissimilarity level of the opportunities and if the dissimilarity level of an opportunity is small, it is then not worth discussing among the driving circumstances for inequality. For this reason, we will not discuss results for electricity, school attendance of children age 7 to 11, or the completion of primary school on time. For the remaining opportunity of education the school attendance of children ages 12 to 15 years the most important circumstance is the education of household head, and that is followed by the household s well-being. These two circumstances account for more than 5 percent of the dissimilarity. Region is also an important circumstance; it accounts for approximately 21 percent of the dissimilarity. With regard to access to improved safe water and improved sanitation facilities, the region circumstance plays the most important role followed by the household s well-being, ethnicity, and education of household head. The location circumstance, which consists of two categories urban and rural plays a more important role than education opportunity. Still, its role is modest relative to the region circumstance. 31

33 Figure 11 Relative Importance of Circumstances Figure 11.a 21 percent Figure 11.b 28 percent Figure 11.c 22 percent School attendance Source: Authors estimation based on data from VHLSSs School attendance Completion of primary school on Clean water Sanitation Electricity time opportunity Child gender HH composition Location Ethnicity HH head education Region Well-being School attendance School attendance Completion of primary school on time opportunity Sanitation Electricity Child gender HH composition Location Ethnicity HH head education Region Well-being School attendance School attendance Completion of primary school on time opportunity Sanitation Electricity Child gender HH composition Location Ethnicity HH head education Region Well-being 32

34 To determine whether the relative importance of these circumstances changed over time from 22 to 28, we do the same exercise for the year 28 and the year 22. The results are presented in Figures 11.b and 11.c. As noted above, clean water opportunity is excluded. Generally, the relative importance of circumstances did not change significantly: Region, the household s well-being, and the educational level of household head are consistently the most influential circumstances in the sanitation facility opportunity as well as the education opportunities of school attendance of children ages 12 to 15. Comparing these results with the results for 21 discussed above, we can confirm the relative importance of three circumstances: region, the household s well-being, and the educational level of household head. Meanwhile, the ethnicity circumstance seems to grow in relative importance in 21. While little can be done with the educational level of household head, targeted support provided to disadvantaged regions and/or individual especially ethnic households, if implemented effectively, may help reduce the D-index and thereby raise the respective HOIs. Figure 12 Relative Importance of Circumstances and Opportunity of Education Quality School Well-being Region percent Parents' education Ethnicity Location HH composition Time to school Child gender Math Vietnamese Both subjects subjects Source: Authors estimation based on data from 27 Grade 5 Assessment. For the quality of education, we have a different set of circumstances and they are grouped as presented in Figure 12. Although education of parents still plays an important role in determining access to quality education, it is ranked second, not first as shown above, when the school attendance of children ages 12 to 15 opportunity is considered. The household well-being is consistently the most important contributor to the dissimilarity index. Apart from these circumstances, the ethnicity and the region are also relatively important circumstances. The two circumstances individually account for 16 to 2 percent of the dissimilarities. Meanwhile, the 33

35 circumstance at school level plays a relatively minor role; this is consistent with above results that the exclusion of this circumstance reduces the HOI only marginally. It should be noted, however, that this circumstance might be captured with less precision than other circumstances in that survey. In particular, the qualification of pupils teachers is not included in the survey. Figure 13 Relative Importance of Circumstances and Health Opportunities percent Receiving antenatal care by skilled person Skilled attendant at delivery Immunization against Measles 21 Complete Immunization against Polio 24 3 Completion Immunization opportunity Child gender HH composition Location Ethnicity HH head education Region Well-being Source: Authors estimation based on data from MICS IV. Figure 13 shows differences in the relative importance of the ethnicity in opportunities of cares for mother and immunizations for children. It is the most important circumstance, accounting for one-quarter of the dissimilarities of the opportunities for mothers but playing only a modest role in the opportunities for children. Meanwhile, the roles of the household s well-being and the region are quite consistent across opportunities. Both of the circumstances explain approximately 2-3 percent of the dissimilarities, with household well-being playing a slightly more important role. Another relatively important circumstance is the educational level of the household head, which accounts for about 15 to 2 percent of the dissimilarities, not including the opportunity of a skilled attendant at delivery. 34

36 Figure 14 Relative Importance of Circumstances and Nutrition Status percent No stunting No underweight No wasting opportunity Well-being Region HH head education Ethnicity Location HH composition Child gender Source: Authors estimation based on data from MICS IV. With regard to the nutrition status opportunities, we only focus on the opportunity of no stunting as the gap between the coverage rate and the HOI of the two remaining nutrition opportunities is small or even statistically insignificant. However, analysis for the opportunity of no stunting is also valid for the opportunity of no under-weight, because we can observe in Figure 14 that the patterns of circumstances importance are somewhat similar across the two opportunities. As shown in the left bar in Figure 14, the household s well-being individually explains more than one-third of the dissimilarity of the no-stunting opportunity, followed by the region as the second most important circumstance. The location is another circumstance that can explain the dissimilarity at a certain level Most and Least Vulnerable Groups with Regard to Opportunities The previous subsection has revealed the relative importance of circumstances or groups of circumstances in determining dissimilarity. However, that exercise does not provide us with information on who has the best and worst access to a particular kind of basic goods and services, and thus termed the least and the most vulnerable group with regard to a particular opportunity. In addition, it cannot tell us whether there are groups that are the most vulnerable with regard to all kinds of basic goods and services. To answer these questions, we can construct profiles of these groups. 35

37 Table 33 presents summary statistics of circumstances of the lowest and highest quintiles in terms of predicted probability for selected opportunities in 21. Instead of discussing all opportunities, we focus on opportunities that exhibit large inequalities in each domain, because patterns of other opportunities in the same domain are similar to those that are presented here. Generally, the results presented in this subsection are consistent with those obtained from the decomposition exercise presented in the previous subsection. Specifically, there are discernible differences in circumstances that are important in determining the dissimilarities of the most and the least vulnerable groups. Living in poor and rural households with a less educated head of household, and belonging to the ethnic minority group are common characteristics of the most vulnerable groups with regard to all opportunities. In term of distribution of opportunities across regions, children living in the Red River Delta account for extremely high proportions in the least vulnerable groups in all opportunities, and they are completely absent in the most vulnerable group except the opportunity of nonstunting. On the contrary, children living in the Northern Uplands have significantly higher proportions in the most vulnerable groups and less representation in the least vulnerable group. Similar situations except for the opportunity of completing immunization are also observed for the children in the Central Highlands. Meanwhile, children living in the Mekong River Delta face the same but more difficult situations for the opportunities of school attendance of children ages 12 to 15, access to improved sanitation facilities, and completed immunizations. The study of only two quintiles as well as of correlations between circumstances results in some extreme cases. Children of the ethnic minority group often live in poorer households with a less educated head, and the combination of these circumstances commonly results in the ethnic minority children being in the most vulnerable group, while being almost absent in the least vulnerable one with regard to all opportunities. Similarly, children living in households with less educated household heads dominate the most vulnerable groups. Notably, all children between 12 and 15 in the most vulnerable group with regard to attending school live in households with a head who did not complete lower secondary school; this circumstance is absolutely absent among the least vulnerable group. 36

38 Table 3 Vulnerability Profile Education of household head Circumstances Child is male No primary school Primary school Lower secondary school Upper secondary school Professional Training College/ University School Attendance (12-15 years old) Most Vulnerable 2% Least Vulnerable 2% Access to improved sanitation facilities Most Vulnerable 2% NA Least Vulnerable 2% NA No stunting (-5 years old) Most Vulnerable 2% Least Vulnerable 2% Complete immunization (1-5 years old) Most Vulnerable 2% Least Vulnerable 2% Skilled attendant at delivery Most Vulnerable 2% NA Least Vulnerable 2% NA Circumstances HH Head is male HH Head living with the spouse Number of siblings Expenditure/ Wealth Index Quintile Ethnic Minority Urban School Attendance (12-15 years old) Most Vulnerable 2% Least Vulnerable 2% Access to improved sanitation facilities Most Vulnerable 2% Least Vulnerable 2% No stunting (-5 years old) Most Vulnerable 2% Least Vulnerable 2% Complete immunization (1-5 years old) Most Vulnerable 2% Least Vulnerable 2% Skilled attendant at delivery Most Vulnerable 2% Least Vulnerable 2% Circumstances RRD a NE NW NCC SCC CH SE MRD 37

39 School Attendance (12-15 years old) Most Vulnerable 2% Least Vulnerable 2% Access to improved sanitation facilities Most Vulnerable 2% Least Vulnerable 2% RRD Northern Uplands Central Coast CH SE MRD No stunting (-5 years old) Most Vulnerable 2% Least Vulnerable 2% Complete immunization (1-5 years old) Most Vulnerable 2% Least Vulnerable 2% Skilled attendant at delivery Most Vulnerable 2% Least Vulnerable 2% a: full names of regions are provided at the end of the table 38

40 Vulnerability Profile: Quality of Education Circumstances Child is male No primary school Primary school Highest education of parents Lower secondary school Upper secondary school College/ University Wealth Index Quintile Ethnic Minority Most Vulnerable 2% Least Vulnerable 2% Circumstances Head teacher is ethnic minority Upper secondary school or lower Education of head teacher Professional/ Pedagogical Pedagogical College/ secondary College school Pedagogical University/U niversity or higher Shared school between primary and secondary Teacher per class Most Vulnerable 2% Least Vulnerable 2% RRD NE NW NCC SCC CH SE MRD Most Vulnerable 2% Least Vulnerable 2% RRD: Red River Delta; NE: North East; NW: North West; NCC: North Central Coast; SCC: South Central Coast; CH: Central Highlands; SE: South East; MRD: Mekong River Delta Source: Authors estimation based on data from VHLSS 21, MICS IV and the 27 Grade 5 Assessment. Urban 39

41 5.5. Regional HOI The results of analysis of the relative importance of circumstances show that the region circumstance consistently plays an important role in determining dissimilarities. This implies a considerable variation in opportunities for children across regions, which is further confirmed by the results of analysis of vulnerable profile with regard to various opportunities. This issue can be analyzed further by employing an alternative version of HOI geometric HOI to estimate the opportunities of access to improved sanitation facilities and school attendance of children ages 12 to 15 across eight regions. These opportunities have the largest inequalities among those that can be investigated over time, given data availability, in each domain of housing infrastructure and education. The geometric HOI has a desirable property of subgroup consistency. When HOI for all subgroups increases, the overall geometric HOI always increases, which may not be the case for the HOI version employed in previous (sub)sections. Figure 15 Regional HOI and Access to Improved Sanitation Facilities 1 8 per cent Red River Delta North East North West North Central Coast South Central Coast Central Highlands South East Mekong River Delta Source: Authors estimation based on from VHLSSs In Figure 15, the bars depict HOI, while the Ts above the bars depict coverage rate. The figure shows that HOI for access to improved sanitation facilities consistently varies across regions throughout the period as well as in 21. In addition, the low coverage rate tends to accompany higher inequality. With regard to changes in HOI over time, HOI of all regions increased in the same period, but at different paces. Specifically, the largest and most consistent increase is observed in the South East. At the other extreme, the North West had extremely low 4

42 HOI in 22, and improvement there was modest and inconsistent. The South Central Coast exhibited a small decline in HOI in the sub-period of due to an increase in inequality, although the coverage rate increased. However, HOI in that region rebounded in 21 due to improvement in coverage as well as a reduction in inequality, as the gap between the coverage rate and the HOI of the region decreased. Figure 16 Regional HOI and School Attendance of Children Ages per cent Red River Delta North East North West North Central Coast South Central Coast Central Highlands South East Mekong River Delta Source: Authors estimation based on data from VHLSSs. Figure 16 shows that the North West and the Mekong River Delta are ranked lowest in terms of HOI for the opportunity of school attendance of children ages 12 to 15; the former also experienced the highest degree of inequality. Furthermore, with the exception of the North Central Coast, all regions experienced improvements in HOI in the sub-period of 22-24, but there have been no clear trends across regions since 24. Specifically, decreases are seen in North East, North West, Central Highlands and Mekong River Delta in the subperiod. This might present a serious problem for North West and Mekong River Delta, because these regions with low HOI rankings might find it even harder to catch up. The North West also experienced an increase in inequality in the sub-period. The standing status as of 21 seems to be worse in some regions than it was in 28, especially the North Central Coast and the South East. However, as the VHLSSs between the two years are not fully comparable, 41

43 we cannot reach a reliable conclusion about the trend. Meanwhile, the standing status as of 21 still shows the North West and the Mekong River Delta with low rankings. 6. Conclusions and Policy Implications This report investigates opportunities for children in Vietnam, using the recently developed Human Opportunity Index, which captures both coverage rate and equality of coverage across subgroups of the population, defined by circumstances exogenous to the individual s control. The report covers three domains education, health, and basic housing infrastructure with a number of specific opportunities in each domain, and it carries out both international comparison and analysis of issues specific to the Vietnamese context. Results of international comparison show a mixed picture of Vietnam s performance relative to LAC and African countries. It does relatively well vis-à-vis these countries in terms of school attendance and access to electricity, but falls behind them in terms of access to clean water and sanitation facilities. Analysis of issues specific to Vietnam also shows a mixed picture. In addition to above-mentioned opportunities, Vietnam performs reasonably well in the provision of cleaner water and care for mothers at the time they give birth. However, the country does not do very well when the quality of education, sanitation facility, and immunization opportunities are considered although the result of the last opportunity should be analyzed with caution as it may be underestimated due to the method used to collect information. Furthermore, the HOI of almost all opportunities is statistically significantly different from the coverage rate, indicating the existence of inequality in access across different groups of the population. The highest level of inequality is found in access to improved sanitation facilities, and certain degrees of inequality are found with regard to the quality of education and immunization, and no stunting of growth. For opportunities that can be investigated over time, progress is observed for all opportunities, especially the completion of primary education on time, access to electricity, and improved sanitation facilities. Analysis of the relative importance of circumstances reveals that the level of well-being as measured by household s per capita expenditure or the wealth index, education of household head, and the household s regional location are consistently the most important determinants of dissimilarity in almost all opportunities. The ethnicity circumstance is the most important circumstance in determining access to maternal care, and it is also relatively important for access to improved sanitation facilities and safe water. Furthermore, a comparison made between 22 and 28 data for some opportunities indicate that the ranking of circumstances is stable over time. 42

44 Results of analysis of vulnerable groups with regard to specific opportunities (i.e. people with the most limited access to a specific opportunity) show that living in poor and rural households where the head of household has limited education, and belonging to the ethnic minority group are common characteristics of the most vulnerable group with regard to all opportunities. Moreover, children of the ethnic minority group often live in poorer households headed by a person with low level of educational attainments, and these circumstances tend to reinforce one another to result in the common presence of ethnic minority children in the most vulnerable group; these factors are almost absent in the least vulnerable one with regard to all opportunities. In terms of geographic location, children living in the Northern Uplands and the Central Highlands face a higher probability of being in the vulnerable groups in almost all opportunities. Children living in the Mekong River Delta also account for significantly high proportions of the vulnerable groups of the opportunities of school attendance of children ages 12 to 15, access to improve sanitation facilities, and complete immunization of children ages 1 to 5. Analysis of the regional HOI finds that there is a significant degree of heterogeneity across regions with regard to the opportunity of access to improved sanitation facilities in terms of status in the initial year, 22, over the period from 22 to 28, and status in 21. The South East shows the largest and most stable increase, while the North West in 22 had a very low HOI that was improved in a slow and unstable manner during the period from 22 to 28. For the school attendance of children ages 12 to 15, the North West and the Mekong River Delta are ranked lowest in both 22 and 21. In terms of changes over time, all regions except the North Central Coast experienced improvements in both HOIs and coverage rates in the subperiod, but the HOI declined in the sub-period for North East, North West, Central Highlands, and Mekong River Delta. The North West also experienced a rise in inequality during this sub-period. A number of policy implications have come out of the analyses. Because improving opportunities for children in terms of both expanding coverage and reducing inequality plays a critical role in helping Vietnam sustain rapid and equitable growth and catch up with more advanced countries, policies should be designed and effectively implemented to make improvements in areas where the HOI is lowest. Specifically, policies should be geared towards expanding access to quality education, improved sanitation facilities, and immunization for children under 5 years old. As the regional circumstance consistently raises inequality in the majority of opportunities, support programs in these opportunities should effectively target regions that are considerably behind the others in these areas. In addition, they should provide better support to specific groups of children who are found by this study to be vulnerable with regard to almost all opportunities. 43

45 References Abras, A., Narayan, A., Cuesta, J. and Hoyos, A. (211). Opportunities for Children in a Post- Conflict Country: the Case of Liberia. The World Bank. Barros, R. P. D., Ferreira, F. H. G., Vega, J. R. M., and Chanduvi, J. S. (29). Measuring Inequality of Opportunities in Latin American and the Caribbean. The World Bank. Bhushan, I., Boom, E., Nguyen, H. H. and Nguyen, M. T. (21). Human Capital of the Poor in Vietnam. The Asian Development Bank, Manila, The Philippines. Electricity of Vietnam (EVN) (21). Báo cáo đánh giá tình hình thực hiện kế hoạch 21, nhiệm vụ và các giải pháp triển khai kế hoạch 211 (Evaluation Report on the Implementation of the 21 Plan, and Tasks of the 211 Plan and Measures to Implement It). Hanoi, Vietnam. General Statistics Office of Vietnam (GSO) (26). Vietnam Multiple Indicator Cluster Survey 26, Final Report. Hanoi, Vietnam: General Statistics Office. Available at Ministry of Labor, Invalids and Social Affairs (MOLISA) and UNICEF Vietnam (28). Children in Vietnam Who and Where are the Poor? The Development and Application of a Multidimensional Approach to Child Poverty. Vietnam. Molinas, J. R., Barros, R. P. D., Saavedra, J. and Giugale, M. (21). Do our Children Have a Chance? The 21 Human Opportunity Report for Latin American and the Caribbean. The World Bank. National Institute of Nutrition (NIN) and UNICEF (212). Summary Report: General Nutrition Survey Available at Rutstein, S. O. and Johnson, K. (24). The DHS Wealth Index.. DHS Comparative Reports No. 6. Available at Phung, D. T. and Nguyen, P. (27). Vietnam Household Living Standards Survey (VHLSS), 22 and 24 Basic Information. An Annex in Market, Policy and Poverty Reduction in Vietnam, Nguyen, T. and Hansen, H. (eds). Culture and Information Publishing House, Hanoi, Vietnam. Shorrocks, A. F. (212). Decomposition Procedures for Distributional Analysis: A Unified Framework Based on the Shapley Value. Journal of Economic Inequality. DOI 1.17/s z 44

46 The World Bank (25). World Development Report 26 Equity and Development..The World Bank and Oxford University Press. Vietnam Academy of Social Sciences (VASS) (21). Vietnam Poverty Assessment Poverty Reduction in Vietnam: Achievements and Challenges. Hanoi, Vietnam. Vietnam National Institute for Education Science (28). Study in Grade 5 student achievement in mathematics and Vietnamese in the school year. Hanoi, Vietnam. 45

47 Appendix 1 Summary statistics of the coverage rates and the HOI Appendix 1.a Opportunities with data sourced from VHLSSs Unit: percent Year HOI Confidence interval (95%) Lower bound Upper bound Coverage rate Finished primary school on time School attendance of children ages School attendance of children ages Access to electricity Access to improved safe water Access to improved sanitation facilities

48 Appendix 1.b Education quality Unit: percent Confidence interval (95%) Subjects HOI Upper Coverage rate Lower bound bound The "best" HOI Math Vietnamese Both Subjects The "comparable" HOI Math Vietnamese Both Subjects Appendix 1.c. Health opportunities Unit: percent Confidence interval (95%) Opportunities HOI Upper Coverage rate Lower bound bound Receiving antenatal care by skilled person Skilled attendant at delivery Immunization against measles Complete immunization against polio Complete immunization Ages -5 no stunting Ages -5 not underweight Ages -5 no wasting Source: Authors estimations based on data from VHLSSs, MICS IV, and the 27 Grade 5 Assessment. 47

49 Appendix 2 Other Health Opportunities 1 8 per cent HOI Coverage HOI Low (95%) HOI High (95%) Immunization against BGC Compete immunization against DPT Source: Authors estimation based on data from MICS IV. 48

50 Appendix 3 The Shapley s Decomposition 18 Suppose that we have varying circumstances, and we want to know how much of the inequality of opportunity is explained by each one of these circumstances. The approach proposed by Shorrocks (1999) is to employ the Shapley value concept that is used in cooperative games to distribute an outcome among players. Here we apply this idea to the inequality of opportunity. As discussed in Section 2, we can identify the inequality of opportunity by the penalty or by the dissimilarity index. When the penalty or the dissimilarity is equal to zero, it means that there is not an inequality of opportunity. These two measures depend on the set of circumstances considered and have an important characteristic: Adding additional circumstances always increases the penalty or dissimilarity. If we have a set A and a set B, and set A and B do not share common opportunities, then: Or, alternatively, the dissimilarity index: The idea of the Shapley decomposition is to identify how much a measure changes when we add a circumstance to different sets of circumstances. Because the circumstances are correlated, the change in the dissimilarity by adding a circumstance depends on the initial set of circumstances. Then, we can measure the impact of a circumstance in the total dissimilarity as the average of all the possible changes. The effect of a circumstance A is: D A = { A} S! ( n S 1! ) S N \ n!.. ( D( S { A} D( S ))) Where N is the full set circumstances, which includes n circumstances in total. S is a subset of not containing the circumstance A. is the dissimilarity calculated with a set and the circumstance. is the dissimilarity estimated with the set. We can define the percentage of contribution as:. And the sum of the contribution of all the circumstances is equal to one: 18 Adapted from a note provided by Alejandro Hoyos of the World Bank. 49