A multidimensional approach to child poverty in vietnam

UNIVERSITY OF ECONOMICS HCMC
VIETNAM – NETHERLANDS PROGRAMME
FOR M.A IN DEVELOPMENT ECONOMICS

A MULTIDIMENSIONAL APPROACH TO
CHILD POVERTY IN VIETNAM

A thesis submitted in partial fulfillment of the requirements for the degree of
MASTER OF ARTS IN DEVELOPMENT ECONOMICS

By
LE THI KIM NHUNG

Academic Supervisor:
Dr. TRAN TIEN KHAI

Ho Chi Minh, December 2014


Le Thi Kim Nhung

Master’s Thesis

VNP19-2015

ABSTRACT

In this study, multidimensional child poverty in Vietnam is approached using
Young Lives Round 3 dataset. In total, 2898 children were analyzed after missing
observations were taken out. Six dimensions with 16 indicators were selected based
on previous researches, public consensus and data availability. These dimensions

are: Education, Health, Nutrition, Shelter, Child Work and Leisure. The Alkire and
Foster (2011a) counting approach is applied to create multidimensional child poverty
index and break down application is used for measuring the depth and intensity of
child poverty. When the poverty cut off point is set at level k=0.3, 30% of children
were identified to be in multidimensional poor. The shelter dimension contributes
most deprived indicators to overall multidimensional child poverty rate while child
insurance draws slightly attention due to its contribution to overal rate followed the
shelter indications. With respect to regional disparity, this is specially critical in
upland northern region where children are greatly affected by multidimensional
poverty. Using poverty line from the cut off point, logistic regression model was
carried out to determine major factors that causes overall multidimensional poverty.
Household characteristics including parents’ education, employment and household
income have significant impact on poverty rate and effect high transition probabilities
for the child moving into and out of multidimensional poor condition.

Key words: Multidimensional poverty measurement, Child Poverty, Deprivation,
Alkire & Foster Methodology, Young Lives.

2


Le Thi Kim Nhung

Master’s Thesis

VNP19-2015

ACKNOWLEDGEMENTS

Completion this thesis is a both painful and enjoyable experience. Fortunately,

I received a lot of help and support from many people for making this thesis possible.
I would like to use this opportunity to express my gratitude to all of them.
First and foremost, I would like to give my special appreciation and thanks to
my academic supervisor, Dr. Tran Tien Khai, for always being helpful, patient and
encouraged. His comments and instructive criticism advices helped me a lot in writing
this thesis.
Similarly, I sincerely thank to the VNP Scientific Committee and staffs of
Vietnam-Netherlands Programme for their helpfulness and willingness in providing
useful information during VNP thesis process.
Lastly, I am also hugely appreciated to my family and my classmates at VNP
K19 who supported me generously for completing this thesis.

3


Le Thi Kim Nhung

Master’s Thesis

VNP19-2015

TABLE OF CONTENTS

LIST OF FIGURES ................................................................................................................................... 7
LIST OF TABLES ..................................................................................................................................... 7
LIST OF APPENDICES ............................................................................................................................ 7
ABREVIATIONS ....................................................................................................................................... 8
CHAPTER 1: INTRODUCTION .............................................................................................................. 9
1.

Problems Statement ........................................................................................................................ 9

2.

Research Objective ....................................................................................................................... 10

3.

Main Research Question .............................................................................................................. 11

4.

Research Scope ............................................................................................................................. 11

5.

Research Methodology ................................................................................................................. 11

6.

Thesis Structure ............................................................................................................................ 11

CHAPTER II: LITERATURE REVIEW ............................................................................................... 13
1.

Definition ...................................................................................................................................... 13
1.1.

Child Poverty ...................................................................................................................... 13

1.2.

Measuring Child Poverty .................................................................................................... 14

2.

Dimensions and Indicators in Multidimensional Child Poverty ................................................. 16

3.

Child poverty Profile .................................................................................................................... 20

4.

Conceptual framework ................................................................................................................. 22

4


Le Thi Kim Nhung

5.

Master’s Thesis

VNP19-2015

Summary ....................................................................................................................................... 22

CHAPTER III: METHODOLOGY ........................................................................................................ 23

1.

DATA ............................................................................................................................................ 23

2.

METHODOLOGY ....................................................................................................................... 24

2.1. Calculation Multidimensional Child Poverty Index (MPI) ......................................................... 24
2.2. Rationale of dimensions and indicators selection ........................................................................ 26
2.3. Weights & Poverty Cutoff ............................................................................................................ 33
2.4. Determinants of child poverty ...................................................................................................... 35
2.4.1. Child Characteristics .......................................................................................................... 35
2.4.2. Household Characteristics .................................................................................................. 35
2.4.3. Community Characteristics ................................................................................................ 38
2.5. Econometrics model ...................................................................................................................... 38
3.

Summary ....................................................................................................................................... 42

CHAPTER IV: EMPIRICAL ANALYSIS.............................................................................................. 43
1.

Overview Child Poverty in Vietnam ............................................................................................ 43

2.

Indicators Deprivation.................................................................................................................. 45

3. Poverty cut off and MPI estimation ................................................................................................. 49

4. Censored Headcount and Related contributions to MPI ................................................................. 51
5. Determinants of Child Poverty......................................................................................................... 52
6. Summary ........................................................................................................................................... 56

5


Le Thi Kim Nhung

Master’s Thesis

VNP19-2015

CHAPTER V: CONCLUSION ................................................................................................................ 58
1.

Conclusion ...................................................................................................................................... 58

2.

Key Lesson learned and policy options .......................................................................................... 59

3.

Limitation and future research ...................................................................................................... 61

REFERENCE ........................................................................................................................................... 62
APPENDICES .......................................................................................................................................... 67

6

Le Thi Kim Nhung

Master’s Thesis

VNP19-2015

LIST OF FIGURES
Figure 1: Conceptual Framework ................................................................................................ 22
Figure 2: Monetary and multidimensional child poverty in Vietnam 2008 divided by Urban,
Rural and Ethnics ......................................................................................................................... 44
Figure 3: Monetary and Multidimensional child poverty in Vietnam, 2008 divided by Regions 45
Figure 4: Proportion of children deprived in each indicators ..................................................... 46
Figure 5: Indicators deprivation by gender based on Young Lives round 3 ............................... 47

LIST OF TABLES
Table 1: Selected indicators and deprivation threshold (Alkire & Roche, 2012) ........................ 18
Table 2: Selected Domains, Indicators and Deprivations Cutoff ................................................. 33
Table 3: Determinants of child poverty and model specifications ............................................... 40
Table 4: Multidimensional child poverty estimate on various cut off point ................................ 50
Table 5: Contribution of indicators to MPI ................................................................................. 51
Table 6: Decomposition of Multidimensional Poverty indices by region..................................... 52
Table 7: Descriptive statistics of Child Poverty determinants variables ..................................... 53
Table 8: Logistic Regression estimates of determinants of child poverty .................................... 54

LIST OF APPENDICES
Appendix 1: Indicators for measuring multidimensional child poverty ...................................... 67
Appendix 2: Logistic regression.................................................................................................... 74
Appendix 3: Regression result with Odds Ratio .......................................................................... 75

Appendix 4: Marginal Effect ........................................................................................................ 76
Appendix 5: Wald test ................................................................................................................... 77

7


Le Thi Kim Nhung

Master’s Thesis

ABREVIATIONS

GSO

General Statistics Office

HDI

Human Development Index

MDG

Millennium Development Goal

MPI

Multidimensional Poverty Index

VHLSS

Vietnam Household Living Standard Survey

MOLISA

Ministry of Labor, Invalids and Social Affairs

UN

United Nations

UNGA

United Nations General Assembly

8

VNP19-2015


Le Thi Kim Nhung

Master’s Thesis

VNP19-2015

CHAPTER 1: INTRODUCTION
1.

Problems Statement
Child poverty approach has become an area of focus for last two decades

resulting from widely acknowledgement of child focused perspective in poverty
eradication (Gordon, 2003; Minujin, Delamonica, González, & Davidziuk, 2005).
Major reasons contributed to this increasing acknowledgement include these: Firstly,
children are at a higher risk in poverty than adults due to their dependence on
distribution of resources from their parents, household and communities in their
direct environment (White, Leavy, & Masters, 2003); Secondly, if children grow up
in poverty, they seem to be in poor condition in their adulthood as well (Corak,
2006a). Next, children’s basic needs are specific and totally different from adult’s,
especially needs for nutrition and education in early stages, and the deprivation of
these needs in childhood can cause permanent effect on their later outcome (Duncan
& Brooks-Gunn, 1999). As a result, academic and policy makers attempt to target
these needs effectively, based on empirical evidence of child poverty’s experiences
(Boyden, 2005) and the use of multidimensional approach is necessary to capture
these basic needs in identifying indicators for measuring child poverty (UNICEF,
2007b).
In Vietnam, the rates of children who are considered in poverty condition
remain high despite the rapid economic growth in the last decade since the
renovation in late 1980s has large impact on country poverty reduction. Hinsdale et
al. (2013) illustrate that one-third of Vietnamese children, estimated around 7 million
children are considered in multidimensional poor. In addition, the high deprivation
that children, are facing draws attention of further studies on children’s basic needs
for effective policy intervention. Despite the interest, there have not been
comprehensive analyses of poverty towards children, as well as research studies
investigating the causes of child poverty, presenting the knowledge gap for policy

9


Le Thi Kim Nhung

Master’s Thesis

VNP19-2015

implication policy and target directed towards children. In fact, numerous studies
were conducted towards child poverty and deprivation through multidimensional
approach in Vietnam (Roelen, 2013; Roelen & Gassmann, 2012; Roelen, Gassmann,
& de Neubourg, 2010; UNICEF, 2008). However, these studies only focused on
views of choosing methods measuring child poverty and deprivation through
multidimensional approach, rather than investigating the characteristics of children
who fall into poverty conditions.
There are many debates in previous researches on child poverty indicators for
measurement in both monetary and multidimensional approach (Gordon, 2003).
Therefore, choosing an appropriate methodology to measure child poverty and define
factors impacting child poverty may result in effective policy implications and
interventions, which better identify specific poverty situations (Roelen, Gassmann, &
Neubourg, 2009). In addition, children’s living conditions depend mostly on their
parents or caregivers’ well-being, thus it is necessary to understand causes underlying
household poor’s conditions, such as economic and demographic factors that drive
individual earning capacity. The objective of this study is to demonstrate the choices
of dimensions, indicators for measuring child poverty in Vietnam through
multidimensional approach and estimate probability of a child being poor by using
data from Young Lives Survey. In particular, this study expects to investigate poverty
situation among children aged under 15 in Vietnam and determine factors that
decrease or increase the probability of child poverty.
2.

Research Objective
The two research objectives are specified as follows:

(i)

Identify the dimensions, indicators of child poverty in Vietnam through
multidimensional approach.

(ii) Identify the determinants of child poverty

10


Le Thi Kim Nhung

3.

Master’s Thesis

VNP19-2015

Main Research Question
(i)

What are the extents and characteristics of childhood poverty in Vietnam
through multidimensional approach?

(ii) What factors determine the probability of a child being poor?
4.

Research Scope
In this study, under-fifteen children are the units of analysis. In Young Lives

dataset, they are divided into two groups: younger cohort (eight years old) and older
cohort (fifteen years old). Their deprivations in selected domains and indicators are
used as the proxy to measure multidimensional poverty. Data in this study is observed
from Young Lives data round 3, year 2009.
5. Research Methodology
This research will apply Alkire-Foster methodology to measure child poverty in
Vietnam for answering the first research question. Logistic regression model is used
to define significant factors for planning an effective intervention policy in reducing
child poverty.
6. Thesis Structure
The thesis is structured into four chapters, following the introduction chapter; the rest
content is outlined as below:
Chapter 2: Literature Review. This chapter will illustrate the literature of child
poverty as multidimensional, starting with definition of child poverty and
multidimensional approach. Theoretical studies as well as empirical results of child
poverty will be reviewed then. Finally, the profile of child poverty is demonstrated.
Chapter 3: Data and Methodology. This chapter analyses the characteristics of the
data used and Alkire-Forster methodology adapted to measure child poverty in
Vietnam. Arguments as well as grounds for selecting domains and indicators will be

11


Le Thi Kim Nhung

Master’s Thesis

VNP19-2015

reviewed then. In this chapter, weight and poverty cut off follow Alkire-Foster are

also outlined. The analytical framework and logistic regression will be further
discussed to identify determinants of child poverty.
Chapter 4: Empirical Analysis of Child Poverty in Vietnam. This chapter starts with
the analysis of the percentage of indicators of overall deprivation. Next, the
Multidimensional Poverty Index, logistic regression model and their interpretation are
emphasized. In this chapter, a summary of findings as well as recommendation of
relevant policies to reduce child poverty bases on statistical results are also presented.
Chapter 5: The overall conclusion is reviewed in this section. Based on the result
estimation, key lessons and policy options are recommended for development and
application of multidimensional approach for children, followed by the limitations of
the study and suggestions for future research.

12


Master’s Thesis

Le Thi Kim Nhung

VNP19-2015

CHAPTER II: LITERATURE REVIEW
Child poverty concepts, its definition, measurement approaches and profile are
discussed in this chapter. In the first part, this chapter starts with the definition of
child poverty and multidimensional approach for measuring. Literature review
towards selecting dimensions, indicators to measure multidimensional child poverty
in previous studies is presented in section 2. Next part, a profile of child poverty will
be reviewed in section 4. And then, based on sections 3 and 4, the conceptual
framework is suggested in the following section. Finally, literature review chapter is
summarized in section 5.

1.

Definition

1.1. Child Poverty
The United Nations General Assembly (UNGA), in 2006, adopted the
international definition of child poverty for the first time as follows: “Children living
in poverty experience deprivation of the material, spiritual and emotional resources
needed to survive, develop and thrive, leaving them unable to enjoy their rights,
achieve their full potential or participate as full and equal members of society”
(UNGA, 2006, para 460). This definition suggests that the concept of child poverty
should be developed based on children’s basic rights and needs approach in order to
capture

multidimensional

forms

children

are

experiencing

poverty.

This

internationally agreement and definition provided a major step forward for
researchers, organizations and policy makers who are concerned with child well-being

and child poverty.
Until 1999, child poverty matter had not drawn proper attention for deserved
special emphasis development studies. Poverty still meant adult and household
poverty; neither data nor theoretical or empirical researches on child poverty were
available. In 1999, The UN Expert with leading researchers such as Peter Townsend

13


Le Thi Kim Nhung

Master’s Thesis

VNP19-2015

and Alberto Minujin outlined the case of child poverty based on data of children
among the poor and monetary approach application. By the end of 1999, UNICEF
experts attempted to emphasize the important of tackling children poverty in order to
eradicate overall poverty (UNICEF, 2000), this report was published in World Bank
afterward. In 2003, the first ever of global child poverty estimates was established by
Professor Peter Townsend and David Gordon (Gordon, 2003). The data result from
this report later was presented in the State of the World Children 2005 (UNICEF,
2004), illustrated that over one billion of children were deprived in one or more basic
needs, has immediately created a widespread acknowledgment of tackling child
poverty all over the world. A few years later, child poverty has been becoming the
centre of international stage and there has been a significant rise in the number of
studies on this field (Boyden, Hardgrove, & Knowles, 2005; Corak, 2005; Minujin et
al., 2005; Noble, Wright, & Cluver, 2006).
In short, under this human-rights approach, children experience poverty through
various dimensions and this issue should be considered as a multi-faced phenomenon.

As such, selected dimensions and indicators in child poverty should reflect basic
needs for children in their daily lives.
1.2. Measuring Child Poverty
A considerable number of approaches have been applied in literature for child
poverty measurement; however they could be categorized into two major groups: unimultidimensional and multidimensional approach.
Monetary approach belongs to one dimension approach group, which is widely
accepted and considered as the traditional method for measuring poverty. This method
uses income-based poverty line as the tool to identify the poor and non-poor.
However, due to multi-faced issues under child poverty perspective, this approach
received various criticism since it cannot capture all aspect as well as their

14


Le Thi Kim Nhung

Master’s Thesis

VNP19-2015

interactions (Sen, 1985). Sen (1985) claimed that tackling child poverty should not
only concern with raising income and utility but also expand to others child’s
capabilities.
Following that concept, Sen (1985) and Corak (2006b) basically developed a
multidimensional method called capability approach with focus on survival (child
health), that is the basic capability of child well-being. The Bristol approach, on the
other hand, was developed based on human-rights following key statements from the
Convention of the Rights of the Child (CRC) (UNICEF, 1989). This approach was
first introduced by Gordon (2003) and applied continuously by UNICEF as the major
measurement for Global Study on Child Poverty and Disparities. In fact, Bristol

approach was aimed to create a worthwhile methodology for comparing child poverty
between countries and UNICEF regions. Dimensions used in this approach require a
set of basic services and standards for children regarding their living condition and
ability to participate in society. In another perspective, Young Lives approach is
similar to Bristol approach, which also aims to capture child’s deprivation aspects;
however it does not intend to build dimensions and indicators for measuring child
poverty. In fact, the project would like to underlie deeper analysis of the causes and
consequences of child poverty.
For the purpose of this study, the definition from the United Nations Children’s
Fund will be applied for measuring child poverty in Vietnam: “Measuring child
poverty can be no longer be lumped together with general poverty assessments which
often focus solely on income levels, but must take into consideration access to basic
social services, especially nutrition, water, sanitation, shelter, education and
information” (UNICEF, 2007a, p. 1).

15


Le Thi Kim Nhung

2.

Master’s Thesis

VNP19-2015

Dimensions and Indicators in Multidimensional Child Poverty
It is well acknowledged that the selection of dimensions and indicators is the

essential issue to measure multidimensional poverty. For this reason, the justification

on how these dimensions are being selected is the key matter in multidimensional
poverty to create a meaningful and reliable poverty index.
Basically, data availability is the most considerable issue when dimensions and
indicators are selected in multidimensional approach towards poverty; however, there
are a large volume of published studies arguing that there are theoretical resources
and universally standards can be used in selecting process. For example, according to
Sen (2004), the selection of dimensions and indicators in multidimensional approach
should follow public critical observation and discussion. Particularly in his paper
published in the year 2004, he illustrated several of criteria for choosing dimensions;
with highlight the most important priority criteria is the purpose of the valuation such
as targeting, monitoring, measuring quality of life. Besides, he also identified two
additional criteria, indicating that dimensions should be argued belief in some value,
priority for relevant group and they should be social influence (Sen, 2004).
Furthermore to Sen’s argument, Alkire (2008) identified five instinct reasons used to
justify dimensions: theory, public consensus, ongoing deliberative participatory
process, data assessment and empirical evidence regarding people’s value or expert.
This view was also supported by Biggeri and Mehrotra (2011) who argued that
dimensions and indicators should be conceptualized, justified by public consensus and
selected through different priorities.
In terms of child poverty, Bristol approach adopted by Global Child Poverty
(UNICEF) attempted to measure child poverty based on international justification and
agreement, in which selected dimensions and indicators should be justified by
international consensus; as such Children Convention Right and MDGs are two

16


Master’s Thesis

Le Thi Kim Nhung

VNP19-2015

important standard measurements. Following this concept, Gordon (2003) argued that
beside the universal justification factor, dimensions also should be child-focused
oriented and be informative towards children’s related problems. In his research,
based on children’s rights from CRC, he defined levels of deprivation through
thresholds in order to approach child deprivations including: shelter, sanitation, water,
information, food, education and health. In his follow-up study with Dr Nandy
Shailen, he pointed out more detail of rationale for choosing dimensions and
indicators in Bristol approach and also provided some criticized arguments regarding
the commonly used measurement such as World Bank’s popular “US $1/day”
indicator,

the

Asset

Ownership-based

Wealth

Index

(World

Bank)

and

Multidimensional Poverty Index developed by Alkire and Foster (2011a).
However, Gordon (2003) and major studies from Bristol approach only focus on
analyzing child poverty at the international levels, while national level measurement
of child poverty requires dimensions, indicators and cutoffs should be considered with
country context-specific for better policy intervention in specific country (Roelen et
al., 2009). As a result, the empirical study of Alkire and Roche (2012) suggested to
measure the depth, intensity and composition of multidimensional poverty applies for
country level. By using the data of under-five child poverty from Bangladesh and
applying Alkire-Foster methodology, they built up indicators and dimension based on
the Bristol Approach and drew intention of how the thresholds and cutoffs can show
the sensitivity of changes in both national and sub-national level. The study does not
illustrate the rationale and justification for selection of dimensions and indicators;
they just simply follow the deprivation thresholds that have been set up before from
Bristol Approach, which include nutrition, water, sanitation, health, shelter and
information. Despite of that, the methodology applied in this study provides useful
references regarding the multidimensional child poverty studies later. In another case
study on child poverty in specific country, Minujin and Delamonica (2005) provide

17


Le Thi Kim Nhung

Master’s Thesis

VNP19-2015

in-depth analysis of child poverty measurement and examine the disparities among
regions in Tanzania by applying deprivation developed by Gordon and Townsend.
The study points out how a country can apply the multidimensional approach to

measure the child poverty and come out with effective design package for tackling
child poverty in each region in the country. In addition, analysis over time was
adopted to measure the change the child poverty condition which was helpful for
programs design and targeting. Moreover, the depth and sensitivity of child poverty
was also carried out in this study by modifying the threshold of dimensions. With this,
the authors pointed out that the adoption of dimension threshold is important to be
flexible in each country and specific situation.
Table 1: Selected indicators and deprivation threshold (Alkire & Roche, 2012)
Dimension

Deprivation Thresholds

Nutrition

Children who are more than two standard deviations below
international reference population for stunting (height for age) or
wasting (weight for height) or are underweight (weight for age). The
standardization follows the algorithms provided by the WHO Child
Growth Reference Study (WHO, 2006)

Water

Children using water from an improved source such as open wells,
open springs or surface water

Sanitation

Children using unimproved sanitation facilities such as pit latrine
without slab, open for pit latrine, bucket toilet and hanging toilet.

Health

Children who have not been immunized by 2 years of age.

Shelter

Children living in a house with no flooring (i.e. a mud or dung floor)
or inadequate roofing

18


Le Thi Kim Nhung

Information

Master’s Thesis

VNP19-2015

Children who have no access to a radio or television

Source: Alkire and Roche (2012)
Concerning empirical studies of child poverty application for Vietnam, several
attempts have been made to approach child poverty underlying multidimensional
methodology. Roelen et al. (2009) demonstrated the importance of choosing
dimensions and indicators using in child poverty measurement approach. The paper
suggested a generic process which underlies definition and concept in order to
develop a sound and solid foundation in the purpose of avoid misunderstanding in
approaching child poverty analysis. The application of this process for Vietnam

context proved the consistency of the result child poverty estimation as well as
adequate findings and understandings with previous studies. In addition to empirical
analysis of child poverty in Vietnam, Roelen and Gassmann (2012) discussed the
results of child poverty in Vietnam using indicators and thresholds which were proved
to be most appropriate for Vietnam context. Most of the dimensions and indicators are
based on deprivation threshold from Bristol Approach with some adjustments
adoption in Vietnam’s situation. One interesting point of this research is that, beside
the disparity between rural and urban, the disparity of ethnicity was shown clearly in
the results, which provides useful information for policy makers to pay their intention
of targeting. In this study, the overlapping analysis of different dimensions was also
noticed. Likewise, in the report co-conducted by UNICEF and the Ministry of Labor,
Invalids and Social Affairs (MOLISA) in 2008, a new approach of multidimensional
child poverty was suggested and applied in empirical examples using VHLSS and
MICS dataset (Multiple Indicator Cluster Survey). Based on financial and technical
support from UNICEF, through several workshops from experts in order to formulate
relevant indicators specialized for Vietnam, the new approach to multidimensional
child poverty was developed, highlighted the specific nature of poverty among
children and provided strong base evidence for effective policy implications. Seven

19


Le Thi Kim Nhung

Master’s Thesis

VNP19-2015

dimensions were chosen based on the statement of CRC and basic needs for Vietnam:
education, health, shelter, water and sanitation, child work, leisure and social

inclusion and protection (UNICEF, 2008). The mechanism in selecting domains and
indicators was focused in this study, with background information and experts’
opinions, complemented by public consensus. The report provided a fundamental
concept as well as meaningful thresholds for setting up dimensions and indicators in
order to measure more sound and solid child poverty index as well as useful
application for Vietnam’s situation, especially the water and sanitation index.
3.

Child poverty Profile
Although child poverty multidimensional approach has brought important

methodology in terms of measuring and comparing multiple aspects of children
experience in poverty condition, identifying risk factors that cause the deprivation of
dimensions among children is raising a concern from an academic perspective.
The necessity of developing child poverty profile has been illustrated in previous
researches. For example, Notten, Neubourg, Makosso, and Mpoue (2012)
demonstrated the need for creating a child poverty profile in Congo Brazzaviller, in
the purpose of identifying the most vulnerable group and find out areas that policy
implications can be potentially applied for better intervention. In their study, instead
of measuring the overlapping of deprivation, the poverty headcount was examined.
The result, technically showed the group of children in the age 6 to 11 is the most
vulnerable group to be experienced in deprivation, especially in relation to the
enrollment-based indicators. In addition, Dercon (2012) also suggested that child
deprivation dimensions can be influenced by various factors in a very complex way,
hence correlating alternatives such as individual, household and community factors
towards child poverty is a considerate matter in order to avoid misleading simple
interpretation from comparison deprivation of dimensions. In his research, Young

20

Le Thi Kim Nhung

Master’s Thesis

VNP19-2015

Lives data was recommended since it can be helpful in capturing this concept. Young
Lives is a longitudinal project, analysis and can figure out the causes and
consequences of child poverty in which various risk factors affect child welfare
outcome (Boyden, 2005).
Conversely, very few empirical studies adopted this concept in particular country
or region. Adetola and Olufemi (2012) applied Alkire Foster methodology to measure
the multidimensional child poverty in rural Nigeria and after that, he used the
obtained MPI as a poverty line to identify children who are poor and non poor.
Dimensions and indicators were included: Safe Drinking Water, Sanitation, Housing,
Health, and Nutrition. In order to identify the determinants, they applied the logistic
regression to conclude the marginal effect of child characteristics, household
characteristics and community characteristics on multidimensional poverty. The
results revealed that health and sanitation are two dimensions which contribute mostly
to overall multidimensional index. Also, logistic model result suggested that parent’s
education, employment in service sector and presence of health facility reduces the
probability of multidimensional child poverty. In another perspective, Boyden et al.
(2005) provided a longitudinal study of children’s poverty in four countries (Vietnam,
Ethiopia, Peru and India) using Young Lives data. The most important finding and
result from this study is their conclusion on how economic growth does not solve the
child poverty issue completely; in contrast, it creates the inequalities among children
in some circumstances. In addition, they also concluded that the deprivation during
childhood would have long term impact when the children grow up and this requires
policy implications for designing social policies for minimize these impacts on

children.
Overall, studies are presented thus far in this section identify that micro-level
determinants including child, household and community characteristics are really
effective in capturing the impacts that determine the children are in and out poverty

21


Master’s Thesis

Le Thi Kim Nhung

VNP19-2015

line. As such, based on empirical results, policy implications are more focused and
effective in reducing child poverty.
4.

Conceptual framework

The conceptual framework is developed based on child poverty dimension from the
study of UNICEF (2008) apply for Vietnam and relevant characteristics literature,
which is demonstrated as follows:
Figure 1: Conceptual Framework
Child Poverty
Child Characteristics

Household Characteristics

Community Characteristics

5.



Education



Health



Shelter



Water & Sanitation



Child Work



Leisure

Summary
This chapter presents the literature review of child poverty definition,

multidimensional measurement, choice of dimensions and child poverty profile. Six
dimensions of enrollment, health, shelter, water & sanitation, child work and leisure
are selected for measure multidimensional child poverty. In general, this
multidimensional poverty index is determined by child characteristics, household
characteristics and community characteristics. The next chapters will present further
analysis of methodology to measure child poverty in multidimensional approach.

22


Le Thi Kim Nhung

Master’s Thesis

VNP19-2015

CHAPTER III: METHODOLOGY
1. DATA
All the data, indicators and variables will be obtained from the dataset of Young
Lives Round 3 (2009). In fact, Young Lives is generally a rich dataset when it comes
to information that is relevant for children. This is a longitudinal project that aims to
investigate the causes and consequences of child poverty in four countries (Ethiopia,
India, Peru and Vietnam). The project was set up in 2001, tracking the continuing
poverty changes and the development of 12,000 children, their family and community
in selected countries with 5 rounds in total, 3 years for each round. In Vietnam, five
provinces were selected for the representative of each region: Lao Cai, Hung Yen, Da
Nang, Phu Yen and Ben Tre. Data of Round 3 (2009) in Vietnam was collected
between the middle of September 2009 and January 2010, in which 2,939 children
and households were visited and directly interviewed (1,963 of the 8-year-old and 976
of 15-year-old). Quantitative data has been collected to illustrate wide range

information related to household characteristics, infant health and nutrition, education,
anthropometry; child labor and social capital. Even though the project aims to capture
the same measure across four countries, however, the sample size is quite small
compared with the whole population. Therefore, the sample is not appropriate for
monitoring child poverty at the country level. The full sample was published and can
be

downloaded

from

UK

Service

Data:

/>Missing value is usually problem that occurs in survey data. Whenever there is
missing information in indicators or variables, we excluded that household from the
computation. The final total number of observation for computing is 2898.

23


Master’s Thesis

Le Thi Kim Nhung

2.

VNP19-2015

METHODOLOGY
2.1.

Calculation Multidimensional Child Poverty Index (MPI)

In order to answer the first research question for measuring multidimensional
child poverty, the study will apply the method Alkire & Foster (Alkire & Foster,
2011b) in which headcount and dual cut off measurement is suggested. In assumption
that D = [dij] is the n x d matrix achievements, where n is the number of children, d is
dimensions and dij> 0 represent ith children’s achievement in dimensions jth. A vector
zj = z (1, 2, 3,…n) of deprivations cutoffs (for each indicator) will be created to
determine whether the child is deprived at that dimension. In particular, if the child’s
achievement is below the cutoff then that child is considered to be deprived in that
dimension. From the matrix of achievement D, we can define a matrix of deprivation
go = [goij] in which goij=1 when diconstruct column vector c of deprivations counts ci = [goij] whose ith entry ci represent
ith child deprivation. According to methodology of Alkire and Foster (2011b), an
“intermediate” approach is suggested where an individual will be considered as being
poor if cik, where k is some intermediate cutoff lying between 1 and d. In another
word, this method can confirm the child is poor if that child is deprived more than k
dimensions. Eventually, children are identified living in absolute poverty when they
have at least two or more deprivations (k=2) and in severe poverty for those who
suffer at least one deprivation (k=1). The next step is calculating the proportion of
children who is identified in multidimensional poor by counting headcount ratio.
H 

q
n

(1)

Where q=q (y;z) is the number of children in the set Ek, as identified using the Zk
dual cutoff method. Average shared deprivation is found by following formula:
A  c(k ) /(qd )

(2)

24


Master’s Thesis

Le Thi Kim Nhung

VNP19-2015

The adjusted headcount ratio Mo (y,z) is then measured by:
MPI= Mo= HxA

(3)

H is the proportion of children who are poor, illustrates the incidence of
multidimensional poverty, while A is the average proportion of weighted deprivations
children suffer at the same time, it represent for the intensity of multidimensional
child poverty.
In addition, MPI can be decomposed into the contribution of sub-groups of
population (Alkire & Foster, 2011b). For example, let’s assume that x and y are the
distribution of two sub groups respectively; (x, y) be the matrix of total population

achievement. Denote n(x) as number of children in sub-group 1, n(y) as the number of
children in sub-group 2 and n(z)=n(x)+n(y):

Income Education

X=

4

4

Person 1

3

5

Persion 2

8

6

Person 3

The population subgroup decomposability can be calculated follow this
formula:
P( z ) 

n( x )

n( y )
P( x) 
P( y )
n( z )
n( z )

(4)

Then one can calculate the contribution of each group to overall poverty:
C ( x) 

n( x ) P ( x )
*
n( z ) P ( z )

25

(5)