UNIVERSITY OF ECONOMICS

ERASMUS UNVERSITY ROTTERDAM

HO CHI MINH CITY
VIETNAM

INSTITUTE OF SOCIAL STUDIES
THE NETHERLANDS

VIETNAM – THE NETHERLANDS
PROGRAMME FOR M.A IN DEVELOPMENT ECONOMICS

THE IMPACTS OF HEALTH SHOCKS ON
CHILD LABOR: EVIDENCE IN VIETNAM

BY

NGUYEN THI HA GIANG

MASTER OF ARTS IN DEVELOPMENT ECONOMICS

HO CHI MINH CITY, December 2017


UNIVERSITY OF ECONOMICS
HO CHI MINH CITY
VIETNAM

INSTITUTE OF SOCIAL STUDIES
THE HAGUE

THE NETHERLANDS

VIETNAM - NETHERLANDS
PROGRAMME FOR M.A IN DEVELOPMENT ECONOMICS

THE IMPACTS OF HEALTH SHOCKS ON
CHILD LABOR: EVIDENCE IN VIETNAM
A thesis submitted in partial fulfilment of the requirements for the degree of
MASTER OF ARTS IN DEVELOPMENT ECONOMICS

By

NGUYEN THI HA GIANG

Academic Supervisor:
DR. LE VAN CHON

HO CHI MINH CITY, December 2017

i


DECLARATION
I hereby declare that my thesis entitled “Impacts of health shocks on child labor: An
evidence from Vietnam” is the result of my own work and includes nothing which is the
outcome of work done in collaboration except as declared in the Preface and specified in
the text.
My dissertation is not substantially the same as any that I have submitted, or, is being
concurrently submitted for a degree or diploma or other qualification any other University
or similar institution except as declared in the Preface and specified in the text.

I further state that no substantial part of my dissertation has already been submitted, or, is
being concurrently submitted for any such degree, diploma or other qualification at any
other University or similar institution except as declared in the Preface and specified in the
text.

Date: September 20, 2017
Signature

Full name: Nguyen Thi Ha Giang

ii


ABSTRACT. Base on household decisions relating to child labor, this paper employs the
Young Lives Dataset and Heckman’s selection model to explore the impacts of health
shocks on child labor in Vietnam during the period of 2006-2009. This study also
considers whether the buffering effect of asset holdings and access to credit are existence
to cushion the impact of health shocks on child labor. The main findings indicate health
shocks only impact on the decision send the child to work, meaning increase the
probability of the child labor participation. Asset holdings is the significant mechanism
to households coping with health shocks. Also, the buffering effect of assets on the child
labor participation is found. However, the access to credit is not significant in both
functions. Additionally, poverty still remains as the crucial factor to determine child labor.
JEL Classification: D13, J13, J22, O12
Keywords: child labor, health shocks, buffering effect, heckman’s selection model
Abbreviations:

December 2017

iii



Table of Contents

ABSTRACT. ................................................................................................................................... iii
CHAPTER I: INTRODUCTION ..................................................................................................... 1
1.1.

Problem Statement and Significance of Research ............................................................ 1

1.2.

Research Objectives and Research Questions .................................................................. 4

1.3.

Scope of the study ............................................................................................................ 4

1.4.

Structure of Thesis Design ............................................................................................... 4

CHAPTER II: LITERATURE REVIEW ......................................................................................... 5
2.1.

Economic Child Labor ..................................................................................................... 5

2.2.

Impacts of Health Shocks on Household Outcome ........................................................ 10


2.3.

Response of Household with Health Shocks .................................................................. 12

2.4.

Health Shocks and Child Labor ...................................................................................... 15

CHAPTER III: RESEARCH METHODOLOGY .......................................................................... 19
3.1. Research Methodology ........................................................................................................ 19
3.1.1. Analytical Framework .................................................................................................. 19
3.1.2. Econometric Model ...................................................................................................... 22
3.2. Vietnam Young Live dataset Overview .............................................................................. 28
CHAPTER IV: RESULTS AND DISCUSSION ........................................................................... 29
4.1. Descriptive statistics ............................................................................................................ 29
4.2. Regression results ................................................................................................................ 32
CHAPTER V: CONCLUSION AND POLICY IMPLICATIONS ................................................ 40
REFERENCES ............................................................................................................................... 42
APPENDIX .................................................................................................................................... 46

iv


The list of Table and Figure
Table 1: Variable Definition ........................................................................................................... 27
Table 2: The child’s work hour following gender and the type of site........................................... 29
Figure 1: The graphical relation between the child work hour and the child age........................... 29
Table 3: Distribution of child labor following the child age .......................................................... 30
Table 4: Description of health shocks ............................................................................................ 30

Table 5: The statistical description health shocks and child labor ................................................. 31
Table 6: Description of using variables .......................................................................................... 32
Table 7: Results of Heckman’s selection model ............................................................................ 33
Table 8: Marginal effect on the child labor participation function ................................................ 34

v


CHAPTER I: INTRODUCTION
1.1.

Problem Statement and Significance of Research

Child labor and health shocks are interesting topics in both pieces of research as
well as policy aspects, especially in developing countries. Most of all governments admit
the important role of children protection and of constructing a good environment for the
child development (UNESCO, 2008). Therein, child labor reduction is one of the main
goals that numerous international organizations and governments are trying to achieve the
improvement. Regarding health shocks, many studies indicate its wide range of negative
impacts on household outcome including children’s life (Beegle et al, 2004; Dilion, 2012;
Alam & Mahal, 2014). Understanding deeply both research areas and linking them
together create a valuable research direction and still useful to contribute to the child
development. Related to this connection, this paper puts health shock’s consequence and
child labor work together in order to explore, interpret and build an improvement for the
child development in Vietnam and developing countries as well. The below presentation
will bring a whole picture to access this issue as well as provide grounds that this research
is to deserve your attention.
Regarding health shocks, these are the important sources of household risks in
developing countries, which are able to raise negative potential impacts on household
outcome such as income reduction, out of expenditure and unbalancing labor supply. In

addition, psychological problem and the persistence of serious disease may be the longterm effects on the life of families (Alam & Mahal, 2014). Health shocks are considered
as idiosyncratic risks that are difficult to predict as well as being costly household budget
(Wagstaff, 2007; Bandara et al, 2015; Mitra et al, 2016). Following World Bank (2017),
the crude death rate (per 1000 people) in Vietnam increases significantly during the period
of 2002-2014, from 5.521% to 5.815%. The one main cause of death comes from the
cardiovascular diseases and diabetes (WHO, 2015). According to the report of Hanoi
School Public Health (2016), the total health expenditure in Vietnam increases from 5.2%
of GDP to 6.9% of GDP, at nearly 191,000 billion VND in 2014, where the private
healthcare spending constitutes more than 52% of the total healthcare expenses. In
particular, the healthcare expenditure of each household experiences a rise of $11.4 in
2002-2014 per month, reaching about 116 USD each year. It is noticed that the health care
cost still remains as the catastrophic expenditure which the rate of out of pocket
expenditure and the private expenditure on health accounts for more than 80% (2014),


compared with 45.5% of that of the world (World Bank, 2017). In other words, the high
healthcare spending can lead a burden that can push households to impoverishment.
Besides, results from other studies in Vietnam provide various empirical evidences about
negative impacts of health risks such as earned income reduction, out of pocket
expenditure, reducing labor productivity and falling the individual's BMI (Wagstaff, 2006;
Van Minh et al, 2012, Bales, 2013). Therefore, it is not difficult to admit that health shocks
create disturbances in the household life and should be noticed in a developing country
like Vietnam.
Under pressure of health shocks, households will apply different strategies in order
to protect their life become more stable. There is a wide range of different coping
strategies that families can consider to respond the health shock’s consequence such as
using savings, trading livestock, selling assets, accessing the credit, changing household
labor supply, especially including the using of child labor (Bandara et al, 2015; Bonfer &
Wright, 2016). Under the shortage of budget and labor force, parents may decide to send
their children to enter into the labor market without other mechanisms (Basu &Van, 1998).

Children may spend more time to work to find income for solving the smooth consumption
problem or substituting employment for people occurring the health risks and caring
patients as well. Besides, when household members occur the death or illness, children are
also aware of their hard circumstances and desire for helping their families. In another
aspect, the demand for unskilled labor as child labor might remain in a high level of lowincome and middle-income countries, where the agricultural sector plays an important
role. Furthermore, if households occur health shocks, using child labor in these areas is
expected to increase more than others (Brown et al, 2002).
The involvement of children in the workforce can have the negative impacts on
childhood as well as children’s future life (ILO, 2013). In other words, when children
spend more time for work activities than study or leisure, their human capital are difficult
to remain in the good status. Besides, child labor can relate to the poor nutrition and
survival of children as well. In several cases, children even have to work the much hours
and overtake their ability, and this can generate the huge vulnerable sequels on children.
Some children even cannot attend school, others seem to lose the life as normal children.
A research of Beegle et al (2004) find the significant negative effect of child labor on
school attainment in the Vietnam. As the results, child labor also can reduce the human

2


capital of the child while the child’s contribution plays a vital role in raising economic
development in the future.
Indeed, child labor still exists although it is prevented by many international
conventions as well as domestic laws. For the legislation of child labor, Vietnam
government gives some decisions related to lessening the employment of children. In
distance, Vietnam sanctioned the international conventions for children rights such as the
Worst Forms of Child Labor Convention (Ratifications of C182) in 2000 and Convention
Concerning Minimum Age for Admission to Employment of ILO (C138) in 2000. In the
internal nation, the Vietnamese government passed the amended Labor Code in 2012
prohibits the child labor under the age of 15, excluding some exceptions; Circular

No.10/2013/TT/LDTBXH (2013) listed occupations and locations where adolescent labor
is prohibited. Besides, it takes into account for Decision No. 1023/QD-TT issued by
Vietnam's Prime Minister in June 2016 about the national project of preventing and
minimizing child labor. This program will take place over the period 2016-2020, with
main targets are to attack illegal child labor, lessen child labor as well as protect the
childhood before the sequel of child labor by providing various opportunities development
to children. However, the number of children attending economic activities in Vietnam
still remains at around 2.83 million individuals, responding nearly 86% in rural area, and
1.315 million children belonged to hazardous labor (Viet Nam National Child Labor
Survey, 2012). Also, the survey indicates that children aged from 5-17 those who do not
attend school result in doing the salaried work and chores as well as incapable of education
investment, around 21% and 9.2% respectively. Therefore, it cannot be denied that
although there are various efforts of the government to protect children rights, child labor
is still persisted.
Above overview poses interesting stories to study in both topics of child labor and
health shocks in Vietnam. Many specific studies also focus on the consequences of risk
events on the household outcome, especially impacts of health shocks and coping
strategies (Wagstaff, 2007; Mitra et al, 2015). Besides, many researches relate to children
and child labor (Rosati & Tzannatos, 2000; Edmonds & Pavcnik, 2002; Beegle at al,
2004). However, the relationship between health shocks and child labor seems to be a gap.
This paper will connect both child labor and health shocks work together. Besides, this
study also considers whether other mechanisms such as household own asset and access
the credit can help households coping with these risks to instead of using the child
3


employment. In other words, whether the existence of the buffering effect of asset holdings
and access to credit on child labor when households experience health shocks. From there,
the paper can give some recommendations to assist families in protecting childhood
through mitigating child labor in Vietnam.

1.2.

Research Objectives and Research Questions
The study focuses on the main objective is to consider the impacts of health shocks

on child labor in developing countries, using Vietnam's Dataset as an example.
Particularly, the paper will explore the change of children's working time when households
experience one or several shocks related to health such as the death or illness of the family
member. Besides, the paper considers some other mechanisms that households can use to
reduce the impacts of health shocks on child labor (if it exists). In order to achieve
objectives, the study will find the answer to two research questions.
(i)

The first question: if a family experienced health risk events such as death,
illness of household members, the child labor will increase or not?

(ii)

The second question: whether other mechanisms such as accessing the credit,
asset holdings can reduce the negative effects of health shock on child labor?

1.3.

Scope of the study
The study focuses on the relationship between child labor and health shocks, both

subjects are interesting stories to explore as well as improve for developing countries.
Using a longitudinal dataset for Vietnam from Young Live project of UK, the paper
employs data from 2006 and 2009 responding with child labor ranking from 4 to 16 years
old.

1.4.

Structure of Thesis Design
This study is organized as following manners. Chapter 2 presents some related
theories and empirical studies about economics child labor, the consequence of health
shocks on the household outcome, household responses with health shocks and the
relationship of health shock and child labor. Chapter 3 shows the research methodology
including the theoretical model and estimate model as well as an introduction to Vietnam
Young Live dataset. The data statistical description, estimation results, and discussions
will be illustrated in Chapter 4. The final Chapter will give some remarkable conclusions.

4


CHAPTER II: LITERATURE REVIEW
2.1.

Determinants of Child Labor
Child labor is the familiar term in the economic development and is defined with

widen extents. In common definition, child labor includes the employment of children that
work in some areas that is able to have the harmful ability to the child welfare such as
business, agricultural or industrial activities. In some cases, it also contains domestic
activities if these activities account for the large time of the child. The age of child labor
puts below a standard level depend on the views of organizations as well. Following the
definition of International Labor Organization (ILO) where is known as a background
organization related to labor in the world, "Child labor is often defined as work that
deprives children of their childhood, their potential, and their dignity, and that is harmful
to physical and mental development". Children aged 5 to 17 years old can be classified as
child labor. However, the interpretation of "harmful" to the child's development is not

consistent among organizations as well as empirical studies. Edmond (2008) argues that
the harmful activities are activities creating the negative effect on children's welfare and
the harmful degree of activities is depended on the age and the characteristic of each child.
In fact, various papers consider child labor with the widen scope containing both
economic and domestic activities. It is not denied that even the time allocation for work
activities which is not followed by definition of ILO, it also can drop the time spending
for study and leisure as well, meaning potential reduction on the child development.
Hence, various previous papers employ the extent term of child labor (Bazen & Salmon,
2008; Dillon, 2012; Bandara et al, 2015; Alam, 2015). This paper integrates the wider
definition of child labor as well. In particular, this study focuses on children belonging the
age interval between 4 and 16, and engaging to activities including the outside work
(unpaid and paid wage) and the domestic work (caring other members and do chores).
Regarding the fundamental theory of child labor, Becker (1965) presents the
standard work – leisure model about the allocation of the child's time based on the
household decision-making. In particular, parents would make decisions with the
assumption of maximizing their household utility, based on the perfect market, initially at
least. Backer's function includes the quantities of the child, the child education, the leisure
of the child and parents, and household expenditure. This model argues that if non-labor
income increases, the leisure time is expected to increase as well. Developing another basis
child labor model, Basu and Van (1998) approach in both labor demand and labor supply
5


side with the assumption that parents push their children to labor market since they have
to face the concern about their family's survival. Besides, the study argues that parents will
make the decision to substitute between adult labor and child labor through comparing the
market wage with the adult wage and children wage. In other words, parents will make
decisions to maximizing utility between consumption and child labor under constraint of
household budget.
In the whole picture about child labor determination, child labor can be affected

by the crucial factors such as poverty, the imperfect market, and household characteristics
(Basu & Van, 1998; Baland and Robinson, 2000; Bandara et al, 2015).
Numerous studies suggest that the main source of child labor might be from
poverty. Basu and Van (1998) argue that altruistic parents willing to send their children to
work only if households have to face the shortage of income. In other words, poverty is
the main cause in order to parents make the decision use child labor. In a basic model of
child labor, Basu and Van (1998) use two assumptions and named as the term of "Luxury
Axiom" and "Substitution Axiom". In particular, "Luxury Axiom" is stated that children
will be sent to the labor market only if the household income not coming from child labor
stands in the very low level. While the "Substitution Axiom" means that child labor and
adult labor can substitute each other under the market view. In short, under the financial
pressure, families can use child labor as a substitution for adult labor, and child labor can
contribute income to remain the survival of households as well. This is similar to the
results of Edmonds and Pavcnik (2005), showing that when households are wealthier,
child labor also performs a decrease. Similarly, Ray (2000) finds the positive relation
between poverty and child labor both in Peru and Pakistan. The richer household is
expected to invest more than in the child's human capital, then use less more the child
work (Dumas, 2013). In addition, several risks such as crop shocks and health shocks, also
are the potential sources of poverty for households, can affect dramatically to the declining
household income. From that, households may reallocate the time to work, leisure and
education of children unless other coping strategies are available at the required scale. In
other words, households which occur shocks are able to use more child labor than others.
Some empirical papers have presented the evidence that child labor is considered as the
response of households to cope with shocks with the lack of other coping mechanisms
(Cain, 1982; Murduch, 1999; Gertler & Gruber, 2000; Dillon, 2012). Jacoby & Skoufias

6


(1997) find there is different between male and female child work under the pressure of

credit constraints
Dumas (2013) considers land as an important wealth resource of households, not
only creating the household income but also used as a collateral item to access to credit,
especially in the rural area. However, the effect of landholding is dynamic depending on
the degree of the interaction between the wealth effect and the substitute effect. Studying
in India, Basu et al (2010) illustrate the relationship of child labor and land size as an invert
U-curve, meaning that the higher degree of own land will raise the time children spend to
work until the maximum point, and then child work time will reduce gradually.
Using the VHLSS 2008 in Vietnam, Hien (2012) studies the impacts of land size
to child labor, focusing on the child from 10-14 years old. In conclusion, the main finding
is found that farm land plays an important role that causes child labor under the failure of
labor market. Besides, the result also indicates the significant non-linear relationship
between land size and child labor in participation decision. It is should be noted that this
analysis considers child labor as a two-stage of decision-making. Thus, Heckman selection
model and double-hurdle model present the better than tobit model.
The other crucial factor which also interprets child labor is the imperfection of the
market. The existence of the imperfect market departs primarily from the problem of
information asymmetry. Dumas’s study (2013) finds that the imperfection of the labor
market makes an increase of child labor. For examples, if households face agriculture
shocks where make change their output of production, they are difficult to find the
additional tenants. This makes an increase of the transaction cost, following that hiring
employers is not preferred. Therefore, the internal labor might be used more, of course
including child labor. Besides, the seasonal effect can create the temporary shortage in the
labor market where people seem to work at the similar time, and this prevents families
from exchanging their employment, and then using child labor is prior. Bar and Basu
(2009) find that in the imperfect labor market, even the household’s land ownership leads
to an small increase of child labor in the short run, child labor still remains in the long
term.
Similar studies for the credit imperfect market, the child work increases if the credit
market is the less competitive (Baland and Robinson, 2000; Ranjan, 2001). Using

Tanzania data, Beegle et al. (2003) indicate that child labor experiences a decrease if
households can access the credit market. However, in the reality, lenders are defined in
7


the limited range including people who can have ability to pay debt. The bank commonly
bases on the collateral asset to give loans and then, it seems to limit loans for poor
households especially in the rural area. Poor households have still stood far way with
access credit with the existing of the credit supply shortage. In some specific cases, when
households experience some risk events and without credit accessing, child labor can be
used as a coping mechanism as well. From this point of view, financial support packages
such as the microcredit program can be provided to households without the collateral asset
to help them coping with risks. This is really the positive method to the development
organizations and governments bolster poor households (Beegle et al, 2006; Dumas,
2013).
Basu and Tzannatos (2003) study about the linkage of child labor, schooling cost
and the quality of education. Following the result, the support of the public education
system can encourage the schooling attendance of children. Besides, a good education
system can take more beliefs from parents about benefits that education can bring for their
children such as better skills and knowledge, and might ensure the better life in the child
future. Therefore, the degree of child labor is expected to reduce in the high quality of
education system. Besides, in the modern life, there are extra schooling expenditures such
as school fees, school uniform, book, extra tutor and transportation cost to the school. This
is really a significant problem for low-income families where they are difficult enough to
the financial source to invest for children's schooling. Noted that children who are less
schooling's attendance might work more others. Moreover, children can still remain their
participation in the school, but in turn, they may have to spend more time to work, even
hard working to cover the education cost.
Cigno and Rosati (2000) indicate that when households meet budget constraints
and cannot enter the labor market, they do not spend their income for the education

expenditure of children. If the older children work, this can mitigate the financial
constraint and force for investing in the education of younger siblings. In addition, poverty
and shocks are potential causes which households might forego the education investment
for children. Baland and Robinson (2000) indicate that although parents can perceive the
higher earned income from education’s children in the future, low-income households still
employ the child labor and deplete encourage children to school when they meet resource
constraints. A similar research of Ranjan (2001) also finds that poor households with credit
constraints may not borrow from other sources, using the child labor become an optimal
8


decision without no bans, and when they can access to credit, they are willing to support
for children's schooling. This result also is similar to the other papers which also show that
accessing on credit can help to decrease child labor and increase the child schooling as
well (Beegle et al, 2006; Islam & Maitra, 2012; Badara et al, 2015).
Parental characteristics also attribute to the determinants of child labor, because
parents make the direct decision relate to the child living in usual aspects. Basu and Van
(1998) show that parents are altruism and they send their children to work if only the
household meet the very basic need. Additionally, parents are more altruistic might
understand the important vital of education on the children's future. This means that more
altruistic parents prefer to invest in the development of human capital than find income
from child employment. In the case of selfish parents, the decision sending their children
to work based on comparing the gained return between child labor and the expected
income from education in the future. Benefits of education in the future may be difficult
to control than that of child work in the present. Besides, it is clear that the expense of
education is only recovered in the future and if households are facing with the budget
shortage, parents may have to deliberate about education investment for their children.
Additionally, Basu (2006) argues that there are differences between the decision-making
of the mother and father about children's education. Another common factor also contained
in the child labor‘s model is parental education. The higher education of parents has

tendency increase in education of children, meaning they may prefer to send their children
to school than to work (Ray, 2000; Strauss & Thomas, 1995; Gebru & Bezu, 2014).
The household size can impact on the child work (Edmonds, 2007). Particularly,
if a family include some dependent members, it can create the requirement of caring as
well as make the high expenditure. In Vietnam, it is common that older children have to
care their siblings or grandparents and take more time to do housework, even work outside
together adult labor.
Ahmed (1999) finds that child labor in the rural area is more common than in urban
due to the large proportion of agricultural sector where low-skill labor demand stands at
the high level. From the demand side, child labor is preferred by cheaper employment,
even with no payment. Besides, because children are considered as the unskilled labor,
they often do not require other non-benefits such as health and social assurance.
Furthermore, in developing countries where may have the large population living in the
rural area as well as have the lower technology, this presents the higher demand for child
9


labor (Bachman, 2000; Brown, 2002). Additionally, the higher time of the child work in
the rural area is explained by the lower income and lower education of parents in this area
(Edmonds & Pavcnik, 2002). In some cases, children also show greater performance than
adults, for example, the carpet job (Levison et al, 1998). The social norm of household
and community also affect the child labor function (Emerson & Knabb, 2007).

In sum up, budget constraints (e.g poverty) and the imperfect market (labor, credit)
are prone to force the child to the labor market. Relating to shocks inside and outside of
families, they might generate negative effects on the household income as well as
household labor supply, and then, can raise the attendance of the child to work directly or
indirectly. Also, the household characteristics and the region factor also attribute to the
determinants of child work.
2.2.


Impacts of Health Shocks on Household Economics
In terms of impacts of health shocks on household outcomes, it is clear that the

poor health is associated with various negative effects on the economic and non-economic
side (O'Donnell et al, 2005; Wagstaff, 2007; Alam & Mahal, 2014). Above section
presents that the negative effects of shocks such as the lower income and unbalance labor
supply, can be more likely to increase child labor. Therefore, this section will highlight a
deeper insight on the impacts of health shocks on households in whole, with aiming to
understand fulfill aspects in the health shock’s consequence and coping mechanisms to
move forward the linkage of the child labor and health shocks in the next section.
Health shocks defined as negative events which relate to the health problem of
family members including the death of family members and the illness or injury of those
(Alam & Mahal, 2014). Some researchers indicate that a health event which categories
"shock" has to create strong negative impacts. For examples, "shock" has to include the
death of adults or catastrophic treatment illness such as terminally ill, fatal diseases and
incidents (Bandara et al, 2015). However, it is not denied that when any family member
gets sick, even just the mild illness, this also forces people to spend more time for looking
after each other or work more to substitute for illness individuals. Therefore, children will
probably be assigned one or several jobs and the fact, they might spend more time to work.
Actually, it is not surprising that there are a lot of previous papers studying the
impacts of health shock on household outcomes with large combinatorial aspects. Alam
10


and Mahal (2014) show a study for reviewing the empirical literature about economic
impacts of health shocks in low - and middle-income countries (LMICs) with the
household level. Following this paper, health shocks can impact on household through (1)
out-of-pocket (OOP) health payments, measures of catastrophic spending and
impoverishment; (2) household labor supply and household income; (3) non-medical

expenditure. Heath shocks are measured by several indicators, for example, parental death,
adults death, illness or death of member family, the measure of disability, change in selfreported health, the specific illness (e.g, cancer, HIV).
The main results of empirical studies show that health shocks in LMICs can are
likely to higher OOP health payments which against the smoothing consumption of
households, and this leads families to impoverishments, especially for poor families. For
distance, with poor families, the share of OOP health payment in the total income is larger
than richer families, and therefore, poor families occur more serious effects. The more
public services provided to patients or the health insurance program can support for poor
households covering some expenses when they have to go the hospital, and this may tend
to lower OOP health expenditure. For Vietnam data, Van Minh et al (2012) find an
increase of 2.5% in poverty if households meet health risks. Also, the result of Wagstaff
(2007) argues that the death of adults makes a rise of the medical expenditure in the last
month in Vietnam at around 27%.
Health risks can be the cause of the time-work loss of family members as well as
employment income reduction. When households face a death or illness of any family
member, this can make their labor supply unbalanced, especially for adult mortality. The
decreasing labor force results in losing people, the low productivity of illness individuals
and other members sent to caregivers, and then, drop the wage, farming returns or business
earnings. In most case studies, the death of adults leads to a lower labor supply in the
family. Beegle (2005) studies for Tanzania finding that there are 66 – 75% of men's wage
within 6 months are decreased when families experience the death of an adult at age of 15
– 50 due to AIDS. Additionally, the death of a household member pushes a decrease of
hour worked by over 8.63% in the past week in Bangladesh. Using data in Vietnam, Bales
(2013) measure the health shock as the variable of adult member bedridden due to illness
for 14 days or more in 12 months, and find that health risks lead a lower annual workday
by 7.7%. Results from a study of Wagstaff (2007) show that if Vietnamese households
experience the death of working age member in urban areas in two years, lead to drop by
11



26% of the total income and 36.5% of earned income. Bales (2013) uses the VHLSS data
to research impacts of severe illness, adult death and the onset of disability on household
welfare in Vietnam. Results indicate reduce in labor supply as well as non-farm
employment income. Although some papers argue the effects of health risks on household
income are negative, while others do not find the connection between them. Results are
able to depend on the measures of health shocks as well as kinds of the labor force
employed. For examples, Yomano & Yayne (2014) indicate an insignificant relation
between any adults death and off-farm income in all, while the result shows a significant
decrease in the death of the male household head. The ambiguous effect also presents in
results of Khan (2010).
Other health shock effects are on non-medical consumption. Using data from
Vietnam, Wagstaff (2007) finds the negative effect on food expenditure. However, Bales
(2013) employs the VHLSS data and find a reduction in labor supply as well as non-farm
employment income while the non-medical payment is not impacted by health risks.
Another conclusion, households do not use all of the consumption insurance to cope with
health shocks when they can access the credit market. Household characteristics also main
factors affecting consumption smoothing.
In sum up, health shocks can make negative effects on the household’s
consequence such as income reduction, out of pocket health expenditure, unbalanced labor
supply. It takes into account that child labor also is existence as the result of health shocks.
The next section will present the household behavior to coping with health shocks. From
that, it can help to understand more detail about the decision-making process for child
labor latterly.
2.3.

Response of Household with Health Shocks
Several main mechanisms to cope with health shocks include the reduction of

expenditure (food, non-food), selling assets or livestock, using savings, borrowing from
formal or informal sources, intra-household labor substitution (Yilma et al, 2014; Alam &

Mahal, 2014; Mitra et al, 2014; Bonfrer & Wright, 2016). It takes into account that under
pressure of health shocks, child labor is also a coping strategy without other mechanisms
(Basu & Van, 1998).
Beegle et al (2006) argue that using assets is an important way to cope with
negative impacts of transitory income shocks. Besides, using data in Vietnam, Wainwright
12


and Newman (2011) find that households employ the liquid asset holding to reduce
impacts of economic shocks, while the negative impact is depleted the household savings,
livestock holdings, and insurances if households experience idiosyncratic shocks like
health shocks.
Baland and Robinson (2000) indicate that with the perfect financial market
assumption, households can access to credit, which have the competitive interest rate, can
smooth their consumption. Using data from Indonesia, Gertler et al (2009) find that
households can access the formal capital market to coping with illness. In particular,
households which have the close distance with commercial banks or microfinance
institutions can cope with illness better others. Another research for Bangladesh of Islam
& Maitra (2012) indicate that microcredit packages mitigate the effects of income
fluctuations on household consumption. The financial support from the government or
credit organizations may be helpful to the household in need, and they are necessary to
support to households approaching these credit packages. In India, Mohanan (2013)
indicates that there are more than 70% of families which troubled with illness prefer to
borrow money for the medical payment. In addition, household's wealth plays a significant
role in ensuring household against health shocks. Families are more likely to pay the
medical cost or smoothing consumption through selling assets (or livestock) as well as
using assets as collateral for formal or even informal credit. The higher income households
who have the strong financial background have more ability to adapt with the negative
effects of health risks, while the lower income counterparts are difficult to enough wealth
for paying the medical expense or hiring added employees. Following Bandara et al

(2015), assets play a vital role in mitigating the impacts of death in households on total
work hour of children when households occur both income and non-income shocks.
Households also employ some informal coping mechanisms to adapt to health risks
without sending their children to work such as receiving support of extended families,
friends and neighbors both financial and work aspects. For examples, households can
receive the loans with low-interest rate, even no-interest rate from nearly relatives or
neighbors, transfers, or through in-kind support such as food, seeds and introducing
employment (Yilma et al, 2004). This is especially popular in rural where is less likely to
access to the other types of formal strategies and having the strong social network.
Families which are not affected by health shocks can help others troubled by health

13


fluctuation. In countries like Vietnam, where the tight community seems common,
especially in the rural area, it is difficult to deny supports from the community.
In Vietnam, Mitra et al (2015) found that to cope with health shocks, Vietnamese
households tend to the enhancing strategies containing making loans, selling assets and
reducing education spending. Following Wagstaff (2007), the labor supply adjustment
plays an important role in coping with health shocks. In the rural, trading in livestock is
an important coping mechanism for smoothing household consumption. Besides, if
families can access the microcredit, they do not sell their livestock. It helps them remain
the tools for production (e.g remain the livestock farming to create the future income)
(Islam & Maitra, 2012).
However, some coping strategies might be difficult to apply in reality, poor
households are less likely to against health shocks with based on savings or assets (ILO,
2013). In developing countries, providing the formal credit are not sufficient and slightly
inaccessible, especially in rural areas and for poor families. Besides, some poor
households even are not enough to means to apply the coping mechanisms when they face
with health shocks, for examples, selling land and assets, or using land and assets to

mortgage in the bank. In addition, because households experiencing health shocks may
show the hard status to pay debt, can be more difficult to achieve the requirement to make
loans from financial institutions (Wagstaff, 2007). Similar results, Gertler et al (2009)
also indicate that the adult death or illness may reduce the ability to make loans due to
diminishing the belief that lenders can pay these loans on time. Besides, health shocks
may drop assets holding through selling them to pay the medical cost or maintain the
consumption smoothing, and this makes the lack of collaterals in the future (Beegle et al,
2006). Therefore, households may access to informal credit with the higher interest rate
than the formal credit because this can provide for them the immediate income as well as
be easier to approach. The cost of the informal borrowing is high interests, even become
a debt accumulation in the long-term which households have to bear, even some poor
families have to forgo treatment for their illness. The pressure of the medical payment is
large, and the ability to access credit is uncertain, households may reduce their
expenditure, or diversity their income from farm or business, push children to work, even
cut down the investment in the education of children.
Another alternative strategy which households apply to deal with health shocks is
adjustment their labor supply. Bazen and Salmon (2008) use the “added worker effect
14


hypothesis” to explain the effects of health shocks on the labor supply of children and
spouse in the household level. The result shows that father’s illness in the short-lived time
or required treatment lead to higher child work. Meanwhile, only if the father experiences
a chronic illness, it is observed an increase of the mother work. There is the similar result
of a rise in labor supply when other family members face the health shocks. In Indonesia,
Gertler & Gruber (2002) indicate the change of the household labor supply when the head
person meets a health risk, this is higher work hour of other household members. Using
the data from Ethiopia, Kadiyala et al (2009) show that Prime Age Adult Mortality (PAM)
can make negative impacts on welfare’s children relied on the unbalance labor supply and
out of expenditure. Besides, PAM leads to lose the working time of people that become

caregivers, and can reduce the performances of childcare as well as is more likely to push
children to attend the inside and outside activities.
In sum up, health shocks have widen impacts on households, not only create the
negative problem to the household standard living but also generate the unbalanced labor
supply. Application some mechanisms such as asset holdings or access to credit can help
households responding with negative effects of health shocks. In a less narrow aspect,
child labor can contribute as a source of coping with health shocks without other
mechanisms both finding income and substitute adult labor. The following section will
mention more detail about this relationship.
2.4.

Health Shocks and Child Labor
Health shocks are considered as idiosyncratic shocks, they depart from internal

households and have to impact on the income and labor force of households at the same
time. This sections will discuss the impacts of health shocks based on the indirect and
direct effects on child labor. The relationship between health shocks and child labor is
illustrated through four main channels (Dinku, 2017). Firstly, health shocks can rise
directly the number of child labor. Children have to spend their time to take care of
diseased members. Secondly, another households can experience the unbalance of the
family labor supply when they meet health shocks, through (1) Loss of individuals (labor),
especially adults (main labor); (2) A drop in productivity due to the less healthy of labor;
(3) Other member may leave her/his job to care for the diseased person. Children might
have to engage to some work activities such as farming work, chores and collecting wood
to substitute for their parents or other members (when they spend time with illness people).
Substitution effects lead children transfer from study or leisure to work. Thirdly, health
15


shocks can generate a decrease of the household smoothing consumption because of an

increase of the medical spending, especially costly treatments. This can raise the financial
pressure on households, and parents may have to send their children to work more
frequently. Fourthly, health risks can make a loss household income through reducing
wage's employment and farming profit. Moreover, the trading assets (even land) or making
loans can create the loss of the future earning such as benefits obtained from assets or
gained interest as well as future payables. As the result, the children may face a rise of
work time when their family experiences health risks. In the other words, to adopt income
loss, parents may send their children to work to find supplement income. Additionally,
lower-income leads families to reduce investment in education's children, and therefore,
the future human capital is expected a lower level as well as attendance of children to the
labor market increase.
Under health risks, child labor can be employed as a coping strategy to substitute
the adult labor in labor market, however, it is not necessary to understand this is a perfect
substitution (Basu &Van, 1998; Raijan, 1999). The feature of works, as well as applied
technologies, would cover the ability of substitution. In particular, the participation of
children in the types of tasks on the labor market also is relied on their skills and health
status. Child work may consistent with domestic household activities such as house
chores, collecting the firewood, water and caring younger siblings. Although child labor
is banned in some fields, especially works in the formal labor market, farming and business
activities are still the potential sources of the child work. Additionally, children can
support for adult works together in some part of the work process. Therefore, children can
spend more time for studying in school or enjoying the leisure, while they can help their
adults in the family some works as well. For examples, study's Ray (2000) presents results
related to the complement between mother and daughters in household tasks in Pakistan.
Additionally, Peru's data give another interesting result, the higher wage of male leads a
lower work hour of girls. Diamond and Fayed (1998) also argues a similar result for female
adults and children in the household level.
In the types and the degree of health shocks also indicate the various effects on
child labor (Alam and Mahal (2014). If family members occur the mild illness, they may
still remain to work some activities. Naturally, the labor productivities might reduce.

Meanwhile, members which have the illness in bed may be difficult to attend any activities
in households, even need caregivers to caring. This is more likely to increase
participation's children in the labor market rather than the mild illness case (Bazen and
16


Salmon, 2008). However, when families experience the death of members following longtime treatment, especially the main labor, the effect of health shocks on families may
exponential increase (Dillion, 2012). Dependent people as children may have to take part
into the labor market, even out of school due to the shortage budget which comes from the
loss huge income from wage's labor as well as the medical payment.
As the review in section 2.2, although many empirical studies focus on the impacts
of health shocks on household labor supply, not seem to be much focus on child labor. In
Bangladesh, Bazen and Salmon (2008) study the impacts of parental health shocks on
child work in the short term and in the long term. Using the bivariate probit model, results
show that father experiences illness leading an increase of child work in overall.
Meanwhile, the proportion of children works increase with the illness of the mother during
the last month. Similarly, Dillon (2012) also finds the significant impacts of health shocks
on child work. In particular, if the illness occurs in the other children of the family, in
overall, the child time allocation of agricultural work will increase around 4 hours each
week. Adult female’s illness leads an increase of 1.6 hours per week which children spend
for caring of younger siblings, while adult male counterpart is associated with 2.6 hours
per week for the child work on the business. Using Tanzania’s data, Bandara et al (2014)
consider health shocks include not only death parents but also other members and find that
the impact of a death in the family on the total work hours is positive significantly both
the male and female children samples. This paper also studies the different types of work
which children enter into. As the results, health shocks generate an increase of the
agriculture work hour, whereas there is a decrease of the inside work. Additionally, the
results also indicate that assets create the buffering effect on health shocks.
In the basic assumption, if households face with the income or non-income shocks,
child labor is employed without other coping mechanisms (Basu & Van, 1998; Gertler &

Gruber, 2000; Dillon, 2012). However, in the wider approach, even when households have
the ability to access to the credit or retain the asset, child labor is still used as the useful
mechanism to households coping with shocks. In other words, the impact of health shocks
on child labor might depend on the degree of asset holding and access to credit as well.
Beegle et al (2003) find that access credit might mitigate the negative effects of income
shocks on child labor using an interaction variable of collateral asset and the crop loss
event to measure the buffering effect of credit on child labor. The similar result is also
found in the research of Bandara et al (2015) where the significant result is found in the
mitigating impact of asset holding on child labor in both income shocks and health shocks.
17


Furthermore, as the results of health shocks, child labor makes the negative
impacts on childhood as well as the child human capital. Another aspect, in terms of child
background, when households occur the health risks, families may reduce their abilities to
provide necessary conditions for children. For examples, mother's death may the lack of
caregiver for children, may lead the less fulfill growth for them. Besides, the exhaustion
of household budget can lead to the less investment in the education and the nutrition as
well. The psychological impacts also important for the child development. Using data in
Vietnam, O'Donnell et al (2005) indicate the evidence of the long-term effects of child
work on children's health.
Omitted variables bias can appear by the simultaneity relation between child labor
and health shocks due to similar negative events such as disease, drought, and flood. For
examples, when families experience drought event, they may not enough fresh water to
use. This may affect negatively to the health of member households due to lack of water
or even excessive anxiety, and also increase time children spend to collect water or
material for livestock. Therefore, it is necessary to use the additional variable related to
other shocks to capture this problem (Bandara et al, 2015). Additionally, mortality of
member in households also considers as a permanent shock, some effects are unobserved.
Furthermore, some unobserved household characteristics can be influent simultaneously

on the health of numbers in families and child labor. In distances, Farrell and Fuchs (1982)
show that parents who are less expectation of return in future from investment on the
education of children (e.g shift child time to work) may also not invest their health (e.g
lead health risks). In this case, using variables of parental characteristics can help control
the bias problem of omitted variables (Bazen & Salmon, 2008; Bandara el al, 2015; Dinku,
2017).

18


CHAPTER III: RESEARCH METHODOLOGY
3.1. Research Methodology
3.1.1. Analytical Framework
In terms of household decisions for child labor, the paper employs a basic model
of Basu and Van (1998) and is developed by Kruger et al (2007) and Bandara et al (2015).
Accordingly, parents would make household decisions in all, also including the child work
as well as the child schooling, and decisions based on the maximizing utility of the
household. The aim of the paper is to consider whether households give the decision send
their children to work with the shortage of budget which comes from the health shocks.
This study will begin with the utility function of consumption and human capital
development of children in households with assuming families of 1 parent and 1 the child.
𝑈(𝑐𝑖𝑡 , ℎ𝑖𝑡 )𝑖,𝑡 =

𝜎
𝑐𝑖𝑡

𝜎

+ 𝛼ℎ𝑖𝑡


(1)

Where 𝑐𝑖𝑡 is denoted as consumption of household i in period t; ℎ𝑖𝑡 is the human
capital of children; σ is the elasticity of substitution and constant with 0 < σ < 1; α is
constant parameter, α > 0.
The paper assumes that consumption of households comes from income which was
earned from adult and child labor, and parents will attend fully the labor force. For the
given the utility function, consumption c will be positive if family income is positive.
𝑐𝑖𝑡 ≤ 𝑤𝑐𝑖𝑡 𝑙𝑐𝑖𝑡 + 𝑤𝑝𝑖𝑡 𝑙𝑝𝑖𝑡
Where 𝑙𝑐𝑖𝑡 , 𝑙𝑝𝑖𝑡 are the time which the child and parent spend to work respectively;
𝑤𝑐𝑖𝑡 , 𝑤𝑝𝑖𝑡 are wages of the child and parents respectively.
The paper will start with the initial simple function with the assumption that
households have neither asset holdings nor access to credit. In the next steps, the study
will add asset holdings and access to credit in turn to find their influence in the relationship
of child labor and health shocks.
(1)

Assume that households have neither asset holdings nor access to credit
The paper begins with households with the assumption the absence of asset

holdings as well as access to credit. Parental income is measured as function of three main
19