The impacts of households characteristics on the occupation opportunities of women the case study in vietnam

UNIVERSITY OF ECONOMICS
HO CHI MINH CITY
VIETNAM

ERASMUS UNVERSITY ROTTERDAM
INSTITUTE OF SOCIAL STUDIES
THE NETHERLANDS

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

THE IMPACTS OF HOUSEHOLD’s
CHARACTERISTICS ON THE OCCUPATION
OPPORTUNITIES OF WOMEN:
THE CASE STUDY IN VIETNAM

BY

NGUYỄN THỊ TỐ VY

MASTER OF ARTS IN DEVELOPMENT ECONOMICS

HO CHI MINH CITY, JANUARY 2018


CERTIFICATION
“I certify the content of this thesis has not already been submitted for any degree and
is not being currently submitted for any other degrees.

I certify that, to be the best of my knowledge, any help received in preparing this
thesis and all source used, have been acknowledged in this thesis.”

Signature

Nguyễn Thị Tố Vy
Date 18th January 2018

i


ACKNOWLEDGEMENT
I would like to send my thanks to my lecturers in Vietnam Netherlands Program who
helped me to achieve the knowledge through interesting lessons, usefull assignment,
utility seminars and new information during my master studying.

I greatly express my special thanks to Dr. Truong Dang Thuy for all his academic
recommendations and supports through finishing thesis process.

I am gratefull to all the staffs in the Vietnam Netherland Office have delivered me
books and necessary documents during the studying process. My thanks are also
extended to all my classmaters, who have sympathised and shared learning
experiences with me through over last two years.

I deepest thank to all of my family’s members who have congrated and given me a lot
of supports during last three years. They have transfered my strong motivation into
the practical actions to help me finish this course on time.

ii


ABBREVIATIONS
VHLSS …………………………………………………..Viet Nam Households Living Standard Survey

VND …………………………………………………………………………………………………….Viet Nam Dong
GSO ……………………………………………………………………….General Statistic Office of Vietnam

BLS …………………………………………………………United State Bureau of Labour Statistics
CPS……………………………….. The March Current Population Survey of United States
SOEP ………………………………………………………..…. German Socio-Economic Panel GSO
obs …………………………………………………………….……………………………………..…. observations
WEs ……………………………………………………………………………….women
status

employment

FBS ………………………………………………………………Pakistan Bureaus of Statistics House
VWU ………………………………………………………………………………. Vietnam Women Union

iii


ABSTRACT
This study purposes to examine the factors affect Vietnamese women’s
decision to participate in workforce and labor supply. Our samples were taken from
dataset of Vietnam Household Living Standard Survey in 2014 including women from
15 to 55 years of age in 63 provinces and cities in Vietnam. We found that the
married women’s employment was strong determined by their individual’s
characteristics (age, head of household, particular by Vietnam Women Union). The
age and head of household have strong negative impacts on the married women’s
occupation opportunity. The Women Union has strongly positive impacts on both
married women and female group. We also found that household size does not have
any impact on female group. Inversely, it has strong positive impacted on married
women group. The bigger household size increasing the opportunities of married

women’s labor force participation, while the higher number of children reducing this
opportunities down. The rural married women have much more occupation
opportunities than urban married women’s but the urban married women have to
work with longer working hours. Although increasing husband’s wage was driven the
labor force participation of married women down but it can not support to improve
the working hours of this women group. The married woman’s occupation
opportunity was not impacted by her and her husband’s education levels. The
phenomenon as married women more likely participate in to the economic activities
than the unmarried women was also determined in our study.

iv


TABLE OF CONTENTS
CERTIFICATION ................................................................................................................... i
ACKNOWLEDGEMENT........................................................................................................ ii
ABBREVIATIONS ............................................................................................................... iii
ABSTRACT ........................................................................................................................ iv
LIST OF TABLES ................................................................................................................. vi
CHAPTER 1: INTRODUCTION ............................................................................................. 1
1.1.

Problem statement ..................................................................................................................1

1.2.

Research objectives .................................................................................................................2

1.3.

Scope of the study ...................................................................................................................3

1.4.

Research structure...................................................................................................................3

CHAPTER 2: LITERATURE REVIEW AND .............................................................................. 4
EMPIRICAL STUDIES ........................................................................................................... 4
2.1.

Literature review of labour supply models ............................................................................4

2.2.

Review of empirical studies ....................................................................................................8

CHAPTER 3: RESEARCH METHODOLOGY AND DATA ........................................................ 17
3.1.

Research methodology ......................................................................................................... 17

3.2.

Data ....................................................................................................................................... 18

3.2.1.

Data Source ................................................................................................................... 18

3.2.2.

Variables description ..................................................................................................... 20

3.3.

Collective household labor frame-work .............................................................................. 24

3.4.

Approached models.............................................................................................................. 25

CHAPTER 4: RESEARCH RESULTS ..................................................................................... 30
4.1.

Summary Statistics ............................................................................................................... 30

4.2.

Research results .................................................................................................................... 58

4.2.1.

Results of Probit, Tobit regression, marginal effects for the full sample ...................... 58

4.2.2.
Results of Probit, Tobit Regression, marginal effects for the sub-sample of married
women 63
4.3.

Discussions ............................................................................................................................ 69

CHAPTER 5: CONCLUSIONS AND POLICY IMPLICATIONS .................................................. 82
5.1.

Conclusions ........................................................................................................................... 82
v


5.2.

Policy implications ................................................................................................................ 83

5.3.

Limitations of the study ....................................................................................................... 84

REFERENCE ...................................................................................................................... 85
APPENDIX ....................................................................................................................... 93

LIST OF TABLES
Table 1: Definitions Variables Used in the Model .................................................................... 21
Table 2: Summary statistics for the full sample (N = 9,171) .................................................... 32
Table 3: Summary statistics for the sample of married women (N = 4,077) ........................... 35
Table 4: Employment status ..................................................................................................... 38
Table 5: Employment status and household size ..................................................................... 40
Table 6: Employment status and number of children .............................................................. 42
Table 7: Employment status and number of old persons ........................................................ 43
Table 8: Results of Probit and Tobit regression with the full sample (N = 9,171).................... 60
Table 9: Results of Probit and Tobit Regressions for Married Women (N = 4,077) ................. 66
Table 10: Compare Strongest Impacts on Women’s Employment status ............................... 77

vi


CHAPTER 1:

INTRODUCTION
1.1.

Problem statement

In the past, Vietnamese family size appeared to be much bigger than that in the
present. During the decades of 1980s, 1990s, one could easily find a family with 8-10
members or even 12-14 members. Nowadays, there are only 3-4 members in an
urban family as also in a rural one. The over 6 members of households are rarely in
the modern Vietnamese society. In generally, the traditional extended families were
really substituted by popular nuclear families. The previous house-wife women in
these traditional families were becoming workers, officials, teachers, engineers,
business or even leaders, government official. However, according to the results of
almost labor force investigations in Vietnam and all over the world, the participation
rate of female is always lower than that of male in the labor market. In the
investigation of Vietnam General Statistics Office (GSO) in 2014, the participation rate
of male of labor force is 82.5% and participation rate of female of labor force is
73.3%. While the ratio of unemployment women is always higher than the ratio of
unemployment man: the ratio of youth female unemployment is 7.15% while the
ratio of youth male unemployment is 5.51% only. These show that the participative
opportunities of women in labor force market should be paid attention by
government, policy makers, offices and everyone.
The results from the study of Rahmah Ismail & Noorasiah Sulaiman (2014)
showed that even though the entrance of female labor in almost economic sectors is

increasing over years, their participation rate is still low in Malaysia. It is the similar
situation in Sudan (Julliffe, 2004) in which presented female career is less than 50 %
out of the total career in these countries. The age of married women and the number
of children are deeply influenced on the supply of female labor. As the rate of
entrance of women labor attributed to several reasons, an investigation of

1


determinants of female labor supply is really responsible, as they are important for
policy design and the further benefits of females.
In several researches as Francois (1998), Basu (2006), Atal (2010) have
investigated the relationships between the supply of female labor and the process of
decision by the household. They found that the participation of female in labor
workforce is decided by the household and the distribution of power is formed
endogenously among the household members. The occupation opportunity of
women indicates that the contribution of women income to their families will
determine the distribution of her power. This will increase in the household when she
could receive more benefits. It means that she has more opportunities to get another
job but not house-working only. Thus, the decision of woman whether work outside
or stay at home as a house-wife is determined by the collective utility of the
household and the market wage is as an economic incentive only. Moreover,
participation of females to the labor force market is always dependent on several
factors.
1.2.

Research objectives

The aim of this research is to examine the impacts of household characteristics,
especially household size on married women’s occupation opportunities in Vietnam.

Model of Probit, OLS and Tobit are used to analyze the decision to participate in
workforce and the number of working hours of women.
Several variables are used as independent variables.
Household size is defined as the number of members in the household in our
study. Beside, the other characteristics of the household and woman are also
analyzed, including: household’s ownership asset, household’s total income, number
of children in household, household poverty status, household locality status or
woman’s marriage status, woman’s education, woman’s age, her husband’s wage,
her husband’s education, her husband’s employment status and other variables.
This study attempts to answer one main reseach question as follow:

2


 Which factors affect Vietnamese women’s decision to participate in the
workforce and labor supply?
1.3.

Scope of the study

We attempt to examine the employment of female and married women in 63
provinces in Vietnam and base on the dataset taken from Vietnam Household Living
Standard Survey (VHLSS) in 2014. Subjects of the study include women from 15 to 55
years of age.
1.4.

Research structure

The proceeds of this study as follows: Chapter one is the introduction. Chapter two
presents the literature review and empirical theories. Chapter three introduces the

collected dataset and the methodology we used to analyze the data in this study.
Chapter four presents the estimation results and discussions parts. The conclusion,
policy implication and limitation are presented in the section five.

3


CHAPTER 2:

LITERATURE REVIEW AND
EMPIRICAL STUDIES
2.1.

Literature review of labour supply models

The labour force is defined by US Beureau of Labour Statistics (BLS) as: “All persons
classified as employed or unemployed. Employed persons are persons aged 16 years
or older, who work for an employer or are self-employed, and excludes people who
work as volunteers and people engaged in self-service/homemaking. Unemployed
persons are those jobless persons who are aged 16 years or older, were available for
job in last four weeks, and had made specific efforts to find a job at any time during
that period”. This definition can show clearly the total of labour force of an economy.
However, it does not show how to exclude the persons who are labour force
participation but they do not work actually at the investigated period because of their
special characteristics as: students, retires, the disabled, and the voluntarily idle as
well as it has no limited ceiling age.
According to the definition of macroeconomists, the labor force is defined as
the people who are employed and unemployed called working-age persons. The
working-age persons are defined as people between the ages of 16-64 but not
include the students, retirees, the disabled people, house-workers and the voluntarily

idle. The advantage of this definition is it has the limited ceiling age. Although, it can
also category between employed, unemployed and students, retirees, the disabled
people, the voluntarily idle, but it does not include the self-service/homemakers in
the labor force. Actually, students, retires, the disabled people, voluntarily idle, the
self-services and homemakers also join economic activities. They directly or/and
indirectly make the goods, materials or services for the economy and get back
income from their works.

4


There are some disadvantages among the definitions here. Both of them seem
difficultly examine or determinate works of house workers, homemakers, the fulltime or/and part-time jobs of students, retires as well as the disabled in Vietnam.
According to Labor Law of Vietnam No. 10/2012/QH13, the working-age persons who
are aged 15 or older, for men aged 15-60 and for women aged 15-55. Therefore, in
this study, we concern about the female and married women who participate in
Vietnam labor market aged 15-55 years.
There are three labour supply theoretical models in the traditional
microeconomic. They are individual labour supply model; household model included
the unitary household labour model and collective household labour model.
Individual labour supply model

The individual labour supply model is a basic model follows the neoclassical theory. It
is known as the basic model of a trade-off of a consumer between consuming more
goods or more leisure. In 1920s, Marshall wrote that, more working hours mean
earning more income is limited by time spending for leisure and reverse which had
found in his research (Marshall, 1920). Theoretically, it is analyzed the individual
labour supply decision is assumed that each unit of labor maximizes his/ her utility
signified to income and entertainment. The result of works creates income, after that
consumer use income to buy goods and/or services. The choosing of goods and/or

service that is making the unilities. Entertainment as service itself is understood as a
kind of goods and any person who consumes it is also given back the utility by their
consumption activities. This theory proves that limited amount of time available,
more working hours more income earning but less time for leisure. These presented
clearly in the studies of Mincer and Becker (Mincer, 1962; Becker, 1965).
The individual labour supply model absolutes the individual labour decision
that have no any relative to other members in his/her family or household. The
participation decision of this individual in the labour market is a unique although
ignore the impact of household factors. In the fact, the participation of a member in

5


the family could be impacted by any his/her relative in their family. That is the reason
why the individual labour supply model was not approached widely now.
Household models

The individual model of a choice between entertainment and consumption ignores
some of element factors as household production or family income etc. which may
influence labour supply. Actually, an individual participation decision in labor market
not only depends on his or her income and time, but also depends on other factors as
his or her wage, the number of family’s members, the house-working hours that
other members can share in their family, other members’ income and the changing of
goods or leisure’s prices. Therefore the household model seems more suitable than
the individual model. Nowadays, both individual labour supply model and household
model have certain restrictions viewed from many corners. Individual labour supply
model ignore the various relatives in the household. Household model ignore the
individual characteristics of a person, reverse.
Family’s characteristics and structures have considerable influence on the
behaviour of its members. Hence, they can directly and indirectly influence on

individual’s choices. The studies of family choices have been improved along two
different lines. The first one is the rule that a single factor can be balanced with its
family through their utility function. It called unitary household labour supply model.
The second one known that the collectively household labour supply model that
makes choices which based on something individual does. By this way, the choices of
each member is not be enlarged (or constrained) by familial framework only, it may
be also influenced by the others.
Unitary household labour model

Fortin and Lacroix (1997) proved that in the unitary household labour supply model,
under a budget constraint, a household has its participation decision in the labour
market in order to it can maximize the household’s utility. An individual’s labour
decision might be a factor that impacted by his or her own choices or by other

6


difference factors come from his or her family’s members conditions (so called
household’s characteristics, for example: the number of members in the household,
total income of the household, the number of children are there in the household,
ect.) as also come from the outside factors (so called society factors, for example: the
unemployment and employment rate in the market, rural or urban condition, ect.). In
a long run, there are some different versions of household models developed and
used to estimate the household’s behaviours in two aspects: the household
behaviours in the general aspect and the household behaviour in the special aspect.
Athough, after the numbers of evidences in the researches of Blundell and Meghir
(1986), Blundell and Walker (1986) as Lundberg (1988) Fortin and Lacroix (1997), the
unitary model is rejected under its literature influences.
Collective household labour model

Although the unitary model was rejected, its limitations also are the premises of the
further model development after that. The studies of Apps and Rees (1988),
Chiappori (1992), Chiappori (2002) proved that there is a better model which can
consider individual’s preferences and bargain to Pareto efficient outcome of all
household’s members in a family. That is the collective household labour model. The
household’s labour market participation decision has two stages under the collective
labour supply model, the first stage is that non-labour’s income is balanced among
household’s members by sharing principle, and the second one is that each member
of the household makes his or her own labour supply decision in the way can
maximize his or her own utility under the match budget constraint. Fortin and Lacroix
(1997), Blundell et. al. (2007), Bloemen (2004) have also concluded that the collective
labour supply model is more advantage than unitary household labour models
through the household labour supply decision behaviours. By adding the income
sharing factor, Fortin and Lacroix (1997) tested and developed the collective model
on the basic of unitary model. The study of Chiappori et. al. (2002) inherited the
previous results and improved the collective model with working spouses as well as

7


other distribution factors. Further, Donni (2003) took the non-participation of this
aspect into the household members’ unrestricted working hour decision.
2.2.

Review of empirical studies

This section review previous empirical studies on the decision to participate in labor
force and labor supply. Almost all studies about the participation in labor fore apply
the Logit/Probit model, while those about labor supply usually apply Tobit model or
the Heckman selection model which analyze both decision on the participation and

labor supply simultaneously. In the labour market, the women labour supply is rising
yearly all over the world. Specially, the labour force activities of women are higher in
developing countries not only in number but also in kinds of work. There are major of
researches in the yield of women labour market participation were studied by lots of
authors in the application science.
The results in Khan and Khan (2009)’s study show that the participation
decision of women in the labour market associated with literature implications. The
independent variables are pointed to 75% of the variation in labour force
participation is accounted for the explanatory variables. They found that the
probability derivative of mother’s age is positive but the age square reverse is
opposite. This proved that when the age of women rise up the probability of labour
force participation of married women grows up. The maximum age of labour force
participation is at 39.49 years old. The authors estimated the average age their study
at the age of 41.61 years old. This result is as the same as the national statistics in
which shown that the women group of age from 40 to 45 years old is the women
group that participate in the labour force with highest rate (Pakistan Bureaus of
Statistics House - FBS, 2003:16).
In this research, the dependent variable – married women’s labor force
participation (MLFP) that is binary has two values: it is 1 value if the married woman
participated in the labour market; it is 0 if the married woman did not. The authors
use normal probability with Probit model to estimate the non-linear maximum

8


likelihood of labor force participation of married women. There were limited amount
of explanation variables in the MLFP function. It is defined that the labor force
participation of married women is the activities done as paid-employment, selfemployment, or in a family enterprise for at least one working hour during the
researching periods (usually one week).
The equation is: 𝑌 = 𝑓(𝑋)

where, Y is the decision of married women to participate labor force. X is a vector of
exogenous socio-economic variables influencing the female’s decision of labor force
participation (also called independent variables or explanation variables). The authors
study the decision participation of married women at the age of 16-60 years old in
labor force activities. There are 3,911 observations in this research that studied in
two districts of Punjab province in Pakistan through the survey was conducted in
2004-2005. They valuated that one woman if she has more one year of education
their labour force participation may increase by 7.9 %. The results also show that the
probability of labour force participation of mothers as head of households is 3.7 %.
D. Blau and M. Kahn (2006) increased the sample size. In addition, the effect of
selected endpoints was minimized by their model. They aimed to make clearly the
arguments of the changing in labour supply. They used three sets of three year
groups: they are including 1979-1981 (call “1980” group), 1989-1991 (call “1990”
group), and 1999-2001 (call “2000” group) to focus on factors that effect on the
participation decision of married women (in the age group of 25-54 years old) in
labour market. The March Current Population Survey (CPS) data used to observe the
labour supply behaviour of United States married women of above three groups.
There are 64,001 observations in “1980” group, 58,987 observations in “1990” group
and 48,733 observations in “2000” group.
The authors used cross-sectional data on individuals to estimate functions such as the
static labour supply models. Firstly, the authors counted the changes of married
women in the labor supply market from 1980 to 2000. They treat the models as a

9


linear model and Tobit model to compare the results for robustness. Secondly, they
used dis-aggregation by Subgroups including: Education groups; Mothers of Young
Children groups; Age Groups. Thirdly, they adjusted the extensive versus the
intensive margin. Fourthly, the authors pointed to the explanatory variables as the

key of the regression process by testing robustness after finding. The robustness of
their results was tested by re-avaluating used models after tax incomes, wages and
salaries. And then, they changed the selectivity of Married Couples. When faced to
omited variable bias, they used grouped data to solve these problems. The results are
presented by differences reports for each of the three periods, the elasticity are
shown together for each period. The elasticity of women wage declined and the
trend of the long-term of spouse’s wage elasticity also decreased continuously in the
period of 1980-2000, it is really highest decreasing in the 1980s. The results of this
study showed that the wives’ labor supply was impacted negatively by the effects of
husbands’ wage and the absolute value of these significant negative effects become
weaker by the time. At the mean part of labour supply, the absolute value as well as
the elasticity of the husbands’ wages was declined considerably compare to the raw
hour impacts.
The paper of Ismail & Sulaiman (2014) focused on evaluating the impacts
which influenced to the participation decision of married women in the general
labour supply in Malaysia. They used a dummy variable to compute the women
labour force. It is understood as 1 if a woman has a job and as 0 if she has not. The
authors considered the logistic regression approach as the method to estimate the
categorical dependent variable. There were 4,000 households in Peninsular Malaysia
were interviewed and randomly chosen in the field survey in 2011. The authors
collected data from this survey contained communications about heads of families,
spouses, heads of household, education, occupation. This study aimed to focus on
the participation decision of married women through the data of 3,520 married
women who were collected in the survey. The utility maximization as follows:

𝑈 = 𝑈(𝑋, 𝐿, 𝑄) (1)

10

The estimation of the logistic model utilizes the likelihood ratio test (LRT) as an
indicator for fitness of the model and the t-test for identifying the significant of the
parameters. The estimation model for this study is as follows:
LS𝑖
= β0 + β1 HWi + β2 FWi + β3 FEDUi + β4 FAGEi + β5 NLYi + β6 NUMCi + β7 GLOBi
+ μ2

(12)

Firstly, the authors use logistic binomial form to investigate. The secondly, the
probability method was choosen to find out the changing jobs probability of a worker
and the other. Then, the logarithm model is used to produce the equation model and
transform itself. Finally, to analyzing the logistic model the author used the likelihood
ratio test (LRT) as a fitness indicator of the model and to identifying the utilized
parameters’ significant they used t-test as a utilized model. The results presented
that there are three kinds of variables which are educational attainment, age of
women and number of children in their family are main determinants of the married
women labour supply. In contrast, husbands’ wage and own wage are insignificantly
determinants the supply of married women labour in this study. The years of
schooling and ages are important determinants in this estimation. Years of schooling
positively affect the supply of married women labour whereas age has an adverse
effect. Wages (either own or husband’s wage) were not proven to be significant from
this study. The importance of education is supported by the more likely effect of
women with secondary and tertiary education to be in the labour market compared
to those with primary and no schooling. Women’s participation in the workforce is
most welcome to obtain benefit, especially those with higher level of education.
Merz (2006) collected the data from German Socio-Economic Panel (SOEP) to
concentrate the undercurrent trends that there are some women groups showing
their subgroup and heterogeneous development in these. They used the
representative sample as the individual household and private who are taken from

Federal Republic of Germany. The authors use the collective theory to study their
research. They defined three different categories: full-time, part-time and non11


employment in the 20 years period of time from 1984-2004. They calculate the real
rate of wage hours of an individual through this annual cross-sectional dataset as the
rate of gross monthly labour income and the working-hours of this person in the
researched months. This ratio calculated by using the worked-hours that signed in
their contracts. These results are not including the real working-hours and others also
are not presented in the contracts. The price index comes from German SocioEconomic Panel (SOEP) as the real working hours was used to interpret the results of
wage rate in the real terms (2000=100). They found that the gross incomes has been
decreased under the effects of several individual characteristics as sex, age, years of
education, marital status, nationality after the regressions of the log on three dummy
variables as full-time employment, occupation and industry. The samples were
divided into two groups as Western are of Germany and Eastern are of Germany. The
results were explained that the ratio occupation employment of Germany married
women nonstop rising and distinguishing with high-middle-low education levels or if
they have young children or not, this employment rate even was highest with women
who have kids or who have the high education level. In contrast, the women who
have to work more hours weekly their husbands have less wage rate. They also find
out that the women who have higher-wage husbands could participate in the labour
market as the similar as the other women. They even could employ more part-time
hours or employ fewer weekly hours in the labour market per week than women who
have lower-wage husbands.
Berulava and Chikava (1983) used the collective frame-work to resolve the
behavior of individual in the household labour supply in several labour market
frames. At the researched regions – Georgian, French and Romanian – the authors
based on the dataset of two years as 2002 and 2005 of Generations and Gender
Survey, they contemporaneous applied unitary and collective household labour
supply models to prove the restrictions of the parametric. They also applied the

collective lobour supply model and conducted comparative research to define the
main determinants of household labour supply. In particular, household preferences

12


are described by unique, well-behaved utility function (Fortin and Lacroix, 1997). The
unitary household labour supply model, The collective labor supply model were
applied to estimating the labour supply equations, the authors applied the Chiappori,
Fortin and Lacroix (2002) methods and the semi-log function form but not apply the
collective model as given before. The authors established a restrictions set from
several function to measure the validity of the models as unitary and collective labour
supply models, as the methods that Chiappori, Fortin and Lacroix (2002) used in their
study. Beside, the authors also examine the unitary model according through testing
the null hypothesis on non-existence of distribution factors (DFI). To test the
relevance of the collective model they measure the validity two elements as the
validity of cross-derivative condition and the validity of Pareto affection.
The results show that the main determinant of the household labour supply in
all of researched region in this paper including: the women and their spouses’ wages,
cross-wage terms, and distribution factors. In the researched regions of French, the
significant at the level of 1% presented that the working-hours of labour supply are
impacted negatively. In contrast, in the researched regions of Georgia and Romania,
the women’ wage impacted positively on the working-hours of labour supply and it
shows a statistically significant. The findings operated the correlated differences of
spouses in these three researched regions. The numbers of lower income spouses in
Georgia are much more substantiates than the lower income couples in others,
France and Romania. Relative with working hours in the labour supply, Georgia is also
the country which has the widest gender gap in this study. It explained that these
factors cannot solve and substitute the differences in preference structures between
countries. The authors presented the results which appropriated with the collective

model in the researches in France and Romania. At any justifiable significant level,
these results nearly always are positive with the restrictions of Pareto affection as
well as cross-term condition of the collective model. While the data of Georgian
rejected almost of restrictions of this model. In contrast, for the unitary model (of
non-existence of distribution factors) applied successfully to reject the restrictions of

13


this model in all these three countries. This certify that the unitary model is valuable
apply in showing the behaviors of households labour supply transitional countries,
particularly in developed economics. It may have similar positive impacts comparing
with previous results came from other studies.
In other study of Dostie and Kromann (2012), the elasticity of labour supply
through the own-wage, couple-wage, non-labour income were measured by using
the data of married women who live in Canada from 1996 to 2005 (the Survey of
Labour and Income Dynamics - SLID). The data of Canadian Tax and Credit Simulator
(CtaCS) and all income levels of household also calculated in this research. Under
investigating at the level of federal also at the level of province, working-hours that
gave household back the real earnings during 10 years (from 1996 to 2005) have
been taxed at the fourth level instead of the third level in 2001 by their government,
rising one level in the fact. It is also akin in several recently studies as the variable CPS
used to evaluate the elasticity of labour supply (Heim, 2007; Blau and Kahn, 2007). To
evaluate the elasticity of the labour supply the authors used the econometric method
with four important steps alike to Heim (2007).
Firstly, the Heckman’s two step produce was applied to investigate the inverse Mills’s
ratio (λit). The chosen equation was estimated by their Probit model to calculate at
the time called t for per member called i (means - it) in the labour supply market:

̂=

λit

̂0 + δ
̂3 Iit +
Φ(δ
̂0 + δ
̂3 Iit +
ϕ(δ

δ̂ Xit )
δ̂ Xit )

(3.1)

where, Iit is total non-labour income of per family and Xit includes the other variables
that show an opportunity of an individual who can participate in the labour force.
They also adjust the endogenous participation of an individual in the labour market
by using the inverse Mills’s ratio in this step. Secondly, the authors impute the nonobserved wages using the traditional regression imputation method. The wage was
computed by using the characteristics of per person adding human capital. Thirdly,
they base on the working-hours that always are positive to evaluate the regression of

14


give working-hours. And other variables explaining the number of worked-hours
involved the non-labour income and the inverse Mills’s ratio. Fourthly, the
participation of an individual in the labour supply market was investigated in this final
step by applying the Probit model. Beside, the dependent variables are the wages of
participation labour force women as well as the women who can not participate in
the labour force but they could receive the wages by applying the equation (3.2):

𝑃 (Participationit
= 1) = Φ (δ0 + δ1 lnw
̂ w it + + δ2 lnw
̂ h it + δ3 Iit + δXit ) (3.4)
In above the equation, the wages of wife, husband and other income were calculated
by the participation of per member in their family. The husband’s consumption
behaviour is difference from the wife’s consumption behaviour by difference income
sources in the similar family. In the period of times as 10 years (from 1996-2005),
they recognized that the labour supply elasticity really declined by applying the
income tax policy of Canada government, the results showed that the elasticity of
labour force move to nearly zero. The most reasonable evidence that is the elasticity
of Canada labour force is lower than the elasticity of the United State. That is why
they found the dramatic evidences in US proved that the declination of this changing
in Canada is too small to calculate. They analysed that the fewer working-hours of
wives the larger elasticity of women participation in the labour force. They called
three hours of working-woman are the limitation units of worked-hours that if the
working-hours of a woman more than this unit the elasticity of these are nearly zero.
In the case the taxes were accounted when the income of a woman is lower this unit
and the elasticity of the hours of work are also zero. The chain of results of their
study presented that the impacts of living public policies have weaker influence than
itself one decade before in Cananda. While, the women who have less working-hours
in a week, have more influences to their spouse’s wages or even to their wages.
Therefore, these low working-hours women should be concerned on if the
government want to rise the working hours for the women labour force.

15


Smith and Stelcner (1988) used released data from the 1981 Census to re-test

the behaviour of married women in labour force in Canada. The authors used
Heckman's procedure to ensure for selectivity bias and incorporating the influence of
income taxes, this study finds small wage and income effects on Canadian wives'
labour supply. Data are come from the Statistics Canada (1985) showed the Public
Use Sample Tape, Household / Family File in 1981. The sample is restricted to 2,851
married couples, with wives twenty to fifty-four years of age. While the tape provides
information on labor earnings and on actual worked weeks in 1980, data on hours of
a worked week are presented only on the week ending 3 June 1981 for the reference.
The hourly wage rate is calculated by dividing annual earnings in 1980 by the product
of weeks worked in 1980 and hours worked in the reference week in 1981. The
division of annual working-hours between working- hours per week and worked
weeks in a year, while potentially important, was not examined in this study. In this
paper, the authors also report labour supply wage elasticity estimates that are small
and not significantly different from zero. Their data are drawn from the 1981 Census
and used the Heckman’s technique (Heckman, 1980) to put down the bias of
selectivities and calculate the factors that impact on labour supply cause of income
taxes.

16


CHAPTER 3:

RESEARCH METHODOLOGY AND DATA
3.1.

Research methodology

Work force participation

Our study follows the paper of Khan and Khan (2009) in which the authors have
evaluated the decision of labor supply participation of married women for the normal
probability using Probit model. One of our two dependent variables is called as
women’s labor force participation is showed by a function of Pr = F(x) = WEs. It
receives value 1 if a woman participated in the labor force and receives value 0 if she
did not.
The function is: Pr (Participation =WEs = Y=1) = F(x)
where, Participation = Y = WEs is the decision of women labor force participation;
x1………xn are the exogenous socio-economic variables influencing the female’s
decision of labor force participation.
Labor supply

Moreover, we use Tobit model to evaluate the factors which may impact on the
number of working hours of a woman (WORKHOURS). The second dependence
variable presents the working time of a woman in a month at the investigated period.
The unit of working time is calculated by her real working hours limited from 0-480
hours per month. The model is presented as follows:
𝑦𝑗∗ = 𝑥 ′ 𝛽𝑗 + 𝜀𝑗
𝑦𝑗 = 𝑥 ′ 𝛽𝑗
𝑦𝑗 = 0

if
𝑖𝑓

𝑦𝑗∗ > 0
𝑦𝑗∗ ≤ 0

where 𝑦𝑗 is the WORKHOURS
We study the decision participation of female and married women at the age of 1555 years old in labor force activities. There are 9,171 observations in this study that is

17


collected from 63 provinces of Vietnam in VHLSS in 2014. It is called as the full sample
of female group. We run the two regressions presented above with this full sample
which has nineteen independent variables. They are involved by household
characteristics (the first characteristic group) and individual’s characteristics of a
woman (the second characteristic group) (presenting in detail in table 1). The Probit
model is used to run this full sample (female group) with dependent variable is WEs
with 9,171 observations. In the Tobit model, we run with dependent variables is
WORKHOURS with 9,161 observations only because there are 100 observations in the
female group which have not enough information to running this model.
In addition, we also run the two models above with sub-sample of married
women group. The married women sample with 4,077 individuals who are married at
the investigated period collected from the full one (so call married women group or
sub-sample). Beside the similar eightteen independent variables as they are in full
sample (not including marriage status variable), we add husband’s factors as the third
characteristic group of a woman (presenting in detail in table 1). This sub-sample is
also used to run the two models: Probit and Tobit with the similar 4,077 observations
separated two dependent variables as WEs with Probit model and WORKHOURS with
Tobit model.
3.2.

Data

3.2.1. Data Source

As mentioned, we study the decision of women (in the age group of 15-55 years) to
participate in labor force activities and the factors that impact on the decision of
women (also in the age group of 15-55 years) to participate in labor force activities.

The women are understood as the married women (in the sub sample) or the female
(in the full sample) as well. The research study in 63 provinces and cities in Vietnam
through the data of Vietnam Household Living Standards Survey (VHLSS) was
conducted in 2014. In addition, we also investigate the impacts of household
characteristics on the number of working hours of women.

18