WORK INTENSITY AND INDIVIDUAL WELL-BEING:
EVIDENCE FROM THAILAND
by
Anant Pichetpongsa
submitted to the Faculty of the College of Arts and Sciences of American University in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy in Economics
%lA fi'iyC ~ Chair: Professor Maria S. Floro
Professor John Wiljou^hby
Professor Thomas Hj
Dean, College of Arts and Sciences 3 / > W . Date
2004
American University
Washington, D.C. 20016 AMERICAN UNIVERSITY LIBRARY
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. UMI Number: 3122640
Copyright 2004 by Pichetpongsa, Anant
All rights reserved.
INFORMATION TO USERS
The quality of this reproduction is dependent upon the quality of the copy submitted. Broken or indistinct print, colored or poor quality illustrations and photographs, print bleed-through, substandard margins, and improper alignment can adversely affect reproduction. In the unlikely event that the author did not send a complete manuscript and there are missing pages, these will be noted. Also, if unauthorized copyright material had to be removed, a note will indicate the deletion.
® UMI
UMI Microform 3122640 Copyright 2004 by ProQuest Information and Learning Company. All rights reserved. This microform edition is protected against unauthorized copying under Title 17, United States Code.
ProQuest Information and Learning Company 300 North Zeeb Road P.O. Box 1346 Ann Arbor, Ml 48106-1346
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. © COPYRIGHT
By
Anant Pichetpongsa
2004
ALL RIGHTS RESERVED
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. This work is dedicated to my wonderful parents,
Surasak Pichetpongsa
and
Sohhua Pichetpongsa
Who set me on my path.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. WORK INTENSITY AND INDIVIDUAL WELL-BEING:
EVIDENCE FROM THAILAND
BY
Anant Pichetpongsa
ABSTRACT
This research examines the measurement of individual well-being based not
only on money income, but also those aspects of individual capabilities and functioning
that are not acquired through the market nor solely the result of market participation. This
research focuses especially on the incidence of work intensity aspect (based on time use
information), which is believed to convey an important qualitative dimension of
individual well-being that the conventional measures do not. Specifically, individual
well-being is determined by three main components -personal income, educational
attainment, and work intensity.
An individual-level survey of urban poor home-based workers in Bangkok,
Thailand collected in 2002 is used to develop the well-being index. The survey shows
that there are significant differences in well-being between men and women workers.
Women respondents are worse off in all respects of quality of life -money, education,
and time.
The empirical tests of the well-being component index and subjective well
being show that the inverse work intensity index is a good predictor of individuals’
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. subjective well-being. It also solves the puzzle that income alone cannot significantly
explain individuals’ perspective of their well-being. Income only brought happiness to
individuals if they did not have to work incredibly long hours to earn it. The empirical
study of subjective well-being and well-being index components also suggests that
economic policies and development strategies that tend to increase low-wage jobs, e.g.,
low-wage export oriented development, will not raise the well-being of individuals,
specifically poor workers Finally, the empirical results indicate that the traditional
methods of measuring well-being that do not take into account time use information tend
to omit crucial well-being information, and are likely to give an incomplete picture.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ACKNOWLEDGEMENTS
This research would not have been completed without the essential efforts of
a multitude of people. First and foremost, I wish to thank Dr. Maria Floro who offered
hours of advice and constructive criticism. This project bears the mark of her expertise in
countless ways. She has been a selfless mentor, and much appreciated source of
information and encouragement throughout my graduate experience. I also wish to thank
Dr. Thomas Hertz who not only provided excellent training in econometrics, but also
tremendous guidance and support. In particular, Dr. John Willoughby has patiently
reviewed the draft and shared his invaluable critique and advice on this project.
The field research was performed with the assistance of Ms. Rakawin
Leechanavanichphan, Ms. Daonoi Srikajon, Ms. Jirapom Changtong, Mr. Wason
Reesomwong, Ms. Aphichaya Nguanbanchong, and Dr. Yada Praparpun. Ms.
Leechanavanichphan and Ms. Srikajon from HomeNet provided necessary information
and granted an access to local communities. This survey work would not have been
possible without the cooperation of HomeNet. Ms. Changtong and Mr. Reesomwong
accompanied me on rigorous, physically demanding home visits. Their patience with the
interview process was, in a word, admirable, and they never complained about the work
schedule or challenge of extensive, daily travel. Ms. Nguanbanchong was an excellent
colleague. She helped with survey design and translation. Together, we shared our
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. V
frustration and delight during the fieldwork. Dr. Praparpun offered her valuable
experienced on field study and encouraged my research.
I would like to thank Ms. Marjorie Miles for her outstanding editorial skills.
Her effort to polish and perfect my work was very much appreciated. I am also
particularly blessed by the friendship and encouragement of Ms. Piyapom Piampratom.
She listened to my thoughts, endured my complaints, and always provided me with
unconditional support throughout this project.
Finally, I am indebted to my parents, Surasak and Sohhua, who inspired me
to believe that success can be achieved, either the hard or easy way, when enough energy
and determination has been dedicated to it. They offered reassurance when I lost faith
during this research study, and convinced me that the world was, in fact, not falling apart.
They also gently asked of my progress and gave continual affirmation, even when I was
undeserving.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. TABLE OF CONTENTS
ABSTRACT...... ii
ACKNOWLEDGEMENT ...... iv
LIST OF TABLES ...... x
LIST OF FIGURES...... xii
CHAPTER
1. INTRODUCTION ...... 1
1.1. Quality of Life Measurement ...... 5
1.2. The Informal Sector, Home-Based Workers, and Well-Being. . . 7
1.3. Potential Contribution ...... 9
1.4. Organization of Dissertation ...... 10
2. REVIEW OF THE LITERATURE...... 12
2.1. The Well-Being Measurement Literature ...... 12
2.1.1. Well-Being Measurement ...... 14
2.1.2. Determinant of Well-Being ...... 21
2.2. Time Allocation Literature ...... 32
2.2.1. Recall/Estimated Method ...... 35
2.2.2. Time Diary M ethod ...... 36
2.2.3. Direct Observation M ethod ...... 39
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. vii
2.3. Concluding Remarks ...... 39
3. ANALYTICAL FRAMEWORK ...... 41
3.1. Determinant of Well-Being ...... 43
3.2. Analytical Framework of Well-Being ...... 48
3.2.1. Components of Well-Being ...... 50
3.2.2. Constraints of Well-Being ...... 58
3.2.3. Well-Being Maximization Process ...... 60
4. METHODOLOGY OF CONSTRUCTING AN INDIVIDUAL WELL-BEING INDEX AND COLLECTING TIME USE D A TA ...... 64
4.1. Construction of an Individual Well-Being Index ...... 64
4.1.1. The Level of Educational Attainment Component Index 66
4.1.2. The Personal Income Component Index...... 67
4.1.3. The Inverse Incidence of Work Intensity Component Index .. 69
4.1.4. The Individual Well-Being Composite Index ...... 75
4.1.5. The Subjective Well-Being Indicator ...... 78
4.2. Time Use Data Collection Methodology ...... 80
4.2.1. The Simplified Time Use Diary Approach ...... 83
4.2.2. The Circle of Trust Approach...... 84
5. EMPIRICAL ANALYSIS: THE CASE OF URBAN-POOR HOME-BASED WORKER IN THAILAND ...... 86
5.1. Background ...... 86
5.2. The Condition of the Urban-Poor Home-Based Workers in Bangkok and Sample Selection Survey M ethod ...... 94
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. viii
5.3. Pertinent Characteristics of Households and Individuals in the Time Allocation Subsample ...... 96
5.3.1. Characteristics of Survey Household Sample ...... 96
5.3.2. Characteristics of the Survey Respondents ...... 100
5.3.2.1. Education and Migration ...... 100
5.3.2.2. Employment and Earning ...... 104
5.3.2.3. Allocation of Time to Market Work, Domestic Work, and Other Activities ...... 114
6. EMPIRICAL ANALYSIS OF INDIVIDUAL WELL-BEING...... 126
6.1. Analysis of the Well-Being Index for Thailand Home-Based Workers ...... 126
6.1.1. The Analysis of Well-Being Index (Model 1 ) ...... 128
6.1.1.1. Gender ...... 130
6.1.1.2. Employment Type ...... 133
6.1.1.3. The Level of Community Organization ...... 137
6.1.1.4. Life Cycle (Age)...... 139
6.1.1.5. Social Support and Household Structure (Presence of Dependent Members) ...... 141
6.1.2. The Analysis of Inverse Work Intensity Index (Model 2) ...... 143
6.2. Subjective Well-Being Indicators and the Validity of the Individual Well-Being Index ...... 145
6.3. Individual Well-Being (Borda) R ank ...... 153
7. SYNTHESIS AND CONCLUSIONS ...... 159
7.1. Thai Home-Based Worker’s Time Allocation and Well-Being Levels ...... 161
7.2. Policy Implication and Concluding Remarks ...... 163
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ix
APPENDIX
A. MATHEMATICAL APPENDIX ...... 167
B. SURVEY COMMUNITY DESCRIPTION...... 171
B.l. Nawamin Community ...... 171
B.2. Nomklao Community ...... 172
B.3. Udomsuk Community ...... 173
BIBLIOGRAPHY ...... 174
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. X
LIST OF TABLES
TABLE PAGE
2.1. Basic Time Diary Format ...... 37
4.1. Classification of the Most Common Overlapping Activities ...... 70
4.2. Minimum and Maximum Values of Well-Being Attributes ...... 77
5.1. Thailand Poverty and Income Disparity 1988-2002 ...... 90
5.2. Number of Households and Homeworkers Aged 13 Years and Over, by Sex, Area, and Region ...... 92
5.3. Selected Characteristics of Subsample Households ...... 98
5.4. Selected Characteristics of Individual Respondents ...... 101
5.5. School Achievement, by Age Group and Sex (Percentage) ...... 104
5.6. Type of Industry (Occupation), by Employment Type ...... 105
5.7. Employment and Income, by Employment Type and Sex ...... 106
5.8. Daily Income, by Employment Type and Sex ...... 110
5.9. Hour Work per Day (Primary Market Paid Work Activity Only) by Employment Type and Sex ...... 112
5.10. Average Time Allocation in All Activities by Employment Type (Minute per Day) ...... 116
5.11. Comparison of Varied Measures of Time Use, by Female and Male Home-Based Workers (Minutes per D ay) ...... 120
6.1. Coefficients Estimates from OLS, Model 1 ...... 129
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. xi
6.2. Individual Well-Being Index and Component Indices, by Sex ...... 132
6.3. Individual Well-Being Index and Component Indices, by Employment Type ...... 134
6.4. The Individual Well-Being Index and Component Indices, by Sex and Employment T y p e ...... 136
6.5. The Individual Well-Being Index and Component Indices, by Community’s Size and Level of Organization ...... 138
6.6. The Individual Well-Being Index and Component Indices, by Stage of Life-Cycle ...... 140
6.7. The Individual Well-Being Index and Component Indices, by Social Support and Household Structure ...... 143
6.8. Coefficients Estimates from OLS, Model 2 ...... 145
6.9. The Summary of Individual Subjective Well-Being Indicators ...... 147
6.10. Coefficient Estimates of Well-Being Component Indices ...... 149
6.11. Coefficient Estimates from OLS, Models 5 to 8 ...... 151
6.12. Spearman Correlations: The Individual Well-Being Index, Component Indices, and Individual Subjective Well-Being Indicator. . . 152
6.13. The Calculation of the Well-Being Aggregated Score (Borda Score/Rank) of Male Self-Employed Workers ...... 155
6.14. Well-Being Rank of Respondents, by Sex and by Employment Type... 156
6.15. (Spearman) Correlation Matrix of Well-Being Ranks ...... 158
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. LIST OF FIGURES
FIGURE PAGE
2.1. General Satisfaction and Domain Satisfaction Explanation ...... 20
3.1. The Effect of Personal Income to Individual Well-Being ...... 45
3.2. The Effect of Education to Individual Well-Being ...... 46
3.3. The Effect of Work Intensity on Individual Well-Being...... 47
3.4. The Individual Well-Being Model under Time and Budget Constraints.. 63
5.1. Thailand’s Annual Growth Rate from 1975-2003 ...... 88
5.2. The Location of Survey Community Sites ...... 96
5.3. Geographic Location of Thailand’s Region ...... 103
5.4a. Kernel Density Estimation of Individual Monthly Income by S e x 108
5.4b. Kernel Density Estimation of Individual Monthly Income, by Employment Status ...... 108
5.5. Kernel Density Estimation of Individual Earning per Hour, by Sex .... 114
5.6. Average Time Spending by Category on both Primary and Secondary Activity of All Survey Respondents ...... 118
5.7. Men and Women Time Allocation: Primary Activities Only...... 123
5.8. Men and Women Time Allocation: Primary and Overlapped Activities (Based on Assumption 1: Equal W eight) ...... 124
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. xiii
5,9. Men and Women Time Allocation: Primary and Overlapped Activities (Based on Assumption 2: Half Weight) ...... 125
6.1. The Kernel Density of Estimated Individual Well-Being Index, by Sex ...... 133
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1
CHAPTER 1
INTRODUCTION
The meaning of the quality of life has been discussed for ages. Classical
Greek Philosophers concluded that the good life resides in virtue. Similarly, the
Confucian school in ancient China described the good life in terms of an orderly society
in which individuals correctly performed their roles and responsibilities (Diener and Suh
2000). More recently, utilitarian and neoclassical economists define it in terms of
maximizing individual happiness and pleasure, referred to as “utility”; with this concept
of utility playing a very important role in mainstream economic theory until now. In an
economic sense, individual satisfaction with his/her well-being is a board concept that
includes not only material achievement (income), but also other aspects of life, such as
self esteem, employment status, education and health. However, in recent years, the
economic debate on individual well-being has been deeply renewed by the Sen’s
capability approach -well-being is seen in term of a person’s ability to do valuable acts or
reach valuable states of being (Sen 1985,1992, 1993). In other words, he suggested that
well-being is considered in terms of human functionings and capabilities. Human
functionings relate to what a person may value doing or being, while capabilities concern
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2
the ability of an individual to achieve different combinations of his/her functionings, and
define the freedom to choose the life that he/she prefers (Sen 1993).1
Economic globalization and financial and economic crises in recent decades
have exacerbated instabilities the markets and financial systems in Asia and throughout
the world. The processes of uneven economic restructuring have created new patterns of
wealth, poverty, and shifts in the patterns of employment and paid /unpaid work
(Bardhan and Klasen 1998, Beneria and Floro 2003). Firms and enterprises laid off
employees as a survival strategy to cope with their financial difficulties. Informal
employment, as a result, increasingly plays an important role in the survival strategies
adopted by poor urban workers. These changes have brought greater attention to the
heightened vulnerability and the deteriorating quality of life of those poor urban informal
workers. To address their poverty and raise their living standard in a more systematic and
comprehensive manner, the study of these informal workers’ well-being should be done
on the individual level.
The critical question is how are we able to appropriately measure the well
being or quality o f life? Dasgupta (1999,11) defined two possible ways to measure the
well-being of an individual:
One is to study the constituents of well-being (e.g., health, happiness, freedom to be and do; more broadly, basic liberties); the other is to value the commodity determinants of well-being (goods and services which are inputs in the production of
These two categories are complementary but distinct -a functioning is an achievement, whereas a capability is the ability to achieve.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3
well-being; for example, food, clothing, potable water, shelter, and resources devoted to national security).
Not only economists, but other social scientists including sociologists,
anthropologists, and psychologists are also interested in developing a proper method of
measuring well-being. For example, sociologists and psychologists have relied on the
notion of subjective well-being (SWB) and General Satisfaction (GS) since the late
3 1960s. The SWB approach studies the constituents of well-being, revealing an
individual’s level of satisfaction/happiness. It is based upon two underlying assumptions
namely: that people can clearly evaluate their own situation and that interpersonal
comparisons are possible (Cantril 1965, Ross and Willige 1997). If these conditions are
not satisfied, then SWB can inaccurately measure quality of life. In contrast to the
methods used in the other fields of study, economists try to measure the quality of life
with proxies primarily based on the person’s access to material goods and services. In
particular, emphasis is given to income. However, some economists such as Ferrer-i
Carbonell (2002a, 2002b) and Van Praag (1991) also utilize the subjective well-being
concept in their work.
This dissertation explores further the measurement of individual well-being
and addresses the issue of well-being comparability between individuals. More
specifically, it develops a well-being index that takes into account not only money based
income, but also those aspects of individual capabilities and functioning that are not
2 The different in these two studies/methods will be discussed later in the well-being index construction methodology section. SWB is based on the subjective, self-reported measure of well-being that is extracted from individual answers to a life satisfaction question or with respect to various domains of life, such as job, housing, or health.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4
necessarily acquired through the market nor solely the result of market participation.
With data on informal home-based workers in Thailand, this dissertation will examine
how time use patterns can serve as indicators o f quality o f life. This dissertation also
focuses on two main objectives. The first is calculating the level of individual well-being
by incorporating time-use related indicators in addition to well-being determinants. Time
use patterns are associated with individual well-being in that the individual’s incidence of
work intensity and differential length of the working day (above the average) are 4 inversely related to that person’s well-being. Work is defined here as activity that
creates disutility, as many economists acknowledged it. Certain type of work (both paid
and unpaid) can, however, yield satisfaction such as baseball players -they get paid to
work as baseball players but they also enjoy playing the game. Another example,
childcare can also yield satisfaction to individual although it is a chore (unpaid work). To
achieve the first objective, an individual well-being index is developed, utilizing sample
time use data collected during fieldwork in Bangkok, Thailand, from June to October
2002, which included 110 home-based worker interviews. This well-being index takes
into account the person’s educational attainment and earned income, in addition to the
incidence of work intensity. The second objective is developing a new methodology for
collecting time use data to make it more appropriate for informal sector workers in
4 Time use information is to be determined whether it is a good predictor of individual well-being (based on subjective well-being indicator).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 5
developing countries.5 This is due to the unique situation faced by home-based workers
with regards to both time constraint and their perception of time.
The scope of our sample population is limited to informal home-based workers in
an urban low-income community in Thailand, as home-based workers are likely to be
both “time-poor” and “money-poor.” These include shopkeepers, market venders,
subcontracted workers, and retailers. Due to their high incidence of both time and
monetary poverty, we believe their level of well-being is seriously affected by their time
use patterns. Our home-based worker sample does not include professionals who work at
home such as computer programmers, executives, freelance architects, sale associates,
etc.
1.1 QUALITY OF LIFE MEASUREMENT
In terms of quality of life measurement, the most well-known indicator is the
Human Development Index (HDI), developed by the United Nations Development
Program (UNDP). The HDI is based on three key components: longevity, educational
attainment, and standard of living.6 It presents the simple average, with equal weight, of
the three attribute indices. This index has been criticized by many economists in that it is
based exclusively on the socio-economic sphere of citizenship, especially on GNP per
5 High quality time use data are difficult to collect. Quality can be affected by such things as the interviewers, and systematic differences in time-use due to the day of the week and season of the year in which data are collected. These problems are best handled by improving the data collection methodology. 6 Longevity is measured by life expectancy at birth. Educational attainment is measured by a weighted average of adult literacy (two-thirds) and combined first, second, and third-level gross enrollment ratios (one-third). For standard of living, the adjusted measure used is the purchasing power parity of the income equivalent in US dollars. A more complete discussion on the Human Development Index is presented in Chapter 2.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 6
head, which is both limiting and potentially misleading (Dasgupta 1999; Dasgupta and
7 Goran Maler 1999). Also, other economists suggest that other variables or factors
should be taken into account in order to evaluate a person’s well-being. For instance,
Dasgupta’s index of social well-being suggested that the well-being of people is also g affected by their civil and political liberties (Dasgupta 1999, 2001). However, these
kinds of well-being measurement, including Human Development Index, Social Well-
Being index, GDP per head, and poverty lines, do not take into account of the amount of
time required to earn the money. This is an obvious flaw that rarely recognized by many
economists.
Juster and Stafford (1985,1991), Flora (1995a, 1995b), and Flora and Hungerford
(2001) among others, argue that time use is also a determinant of well-being. One
important dimension of time use that affects a person’s quality of life is time or work
9 intensity, performing two or more activities simultaneously. The secondary and tertiary
activities that are performed in addition to primary ones (considered by the individual as
the main activity) are also referred to overlapped activities. For example, cooking can be
done simultaneously with listening to music or child minding can be performed while
cleaning the house (Flora 1995a; Folbre 1997; Flora and Miles 2003; Hungerford and
7 The most discussed problem in using GNP as one of the indices is that GNP doesn’t include the depreciation of capital assets, so it is incapable of reflecting future prospects. Moreover, GNP is not the flow equivalent of wealth. (Dasgupta 1999) 8 Dasgupta (1999,2001) added the civil and political liberties attribute indices in to his quality of life index calculation. 9 The notion of work intensity was first discussed by Karl Marx who considered lengthening or intensifying the work day by capitalists as a means of extracting surplus value from workers (Marx 1990). However, the notion of work intensity in this dissertation is more restricted in the sense that it refers to the performance of any tow or more overlapped work activities. It is also different from the Marx’s notion in that it includes both wage work and unpaid work.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7
Floro 2001). This incidence of work intensity reveals an important qualitative dimension
of time use that the conventional measures of living standards do not. In other words,
welfare measurement must involve not only wealth (the stream of income earned to
acquire goods and services) or socio economic status, but also how time is spent by
people. Studies by Roldan (1985), and Wolfe and Haveman (1983) have also shown the
link between time use patterns and health. Their studies show that high levels of work
intensity create stress and lead to eventual deterioration of health.
1.2 THE INFORMAL SECTOR, HOME-BASED WORKERS, AND WELL BEING
Before going further, it is necessary to briefly discuss what is meant by the
informal sector and home-based workers. When economists use the term “informal
sector,” they are usually referring to a dualistic pattern of development present in labor-
surplus economies (Schaefer 2001).10 The criteria used to define the formal sector and
the informal sector varies between legal, technical, and financial differences. The
informal sector definition used in this dissertation is based on the ILO Report of the
Fifteenth International Conference of Labor Statisticians (1993), which defined the
informal sector as “all unregistered or unincorporated enterprises below a certain size,
including : micro-enterprises owned by informal employers who hire one or more
employees on a continuing basis; and own-account operations owned by individuals who
may employ contributing family workers and employees on an occasional basis (ILO
10 The size of the informal economy is reported by the ILO (2002) to be one half to three-quarters of non- agricultural employment in developing countries, e.g., 65 percent in Asia and 72 percent in sub-Saharan Africa. If data were available for additional countries in Southern Asia, the regional average for Asia would likely be much higher.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 8
2002a, 11).” The informal sector, therefore, consists of small-scale, self-employed
activities (with or without hired workers), typically at a low level of organization and
technology, with the primary objective of generating employment and incomes (ILO
2002b). Home-based workers are defined by the ILO as those who produce a product or
provide a service to a contractor or an employer through the selection of their own work
place (ILO 2002a, 2002b).
Time use patterns can provide important information on the well-being of the
poor home-based worker in the informal sector as they struggle to meet their survival
needs. Several studies illustrate that home-based workers, the majority of whom are
women, take this form of employment in order to combine market (paid) work with
domestic activities (unpaid work), such as cooking, cleaning, house keeping, and
childcare (ILO 2002a, Bajaj 1999, and Lazo 1992). These home-based workers can be
classified into two groups, namely: a) home-based workers who are self-employed, and
work in their own business; b) home-based workers who are paid by others to work
under the sub-contract agreement, also known as homeworkers (ILO 2002a). Given
women’s multiple roles as income earners, primary caregivers and household managers,
their “time poverty” often takes the form of high work intensity, which directly affects
the health and productivity of these workers (Floro 1995a). Given there are only 24
hours in each day, women end up juggling many chores and overlapping their activities
to deal with this lack of time (Floro 1999). While the necessity to overlap activities can
greatly impact an individual well-being, few economists have researched the relationship
between quality of life and time use patterns. This task is complicated by the scarcity of
data on time use that include overlapping activities.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 9
1.3 POTENTIAL CONTRIBUTION
Building on the work of Floro (1995a), Dasgupta (1993,1999,2001), UNDP
(2001,2002,2003), Juster and Stafford (1985,1991), Robinson and Godbey (1997), and
Floro and Hungerford (2001), a new comparable well-being measurement approach is
developed in this dissertation. It takes into account the contribution of material goods and
services accessed through the market on individual well-being, along with the labor
process involved in acquiring these purchased goods and services and in producing non-
market goods and services for one’s own consumption. By incorporating the manner in
which the individual acquires her entitlements into the well-being index, the shortcoming
present in current indices that emphasize only income or consumption of basic goods and
services is lessened. Work intensity is also expected to provide crucial information on
individual well-being. It could reveal the fact that money income needs to take into
account the level of work intensity (or time spent to earn that money) to explain the level
of well-being. This will benefit policymakers by allowing them to design more direct and
effective policies in order to raise individual quality of life.
This study also proposes a new methodology for gathering data on time use
that is more appropriate for individuals who are likely to be time-constrained in
answering recall questions of the current time use diaries or are likely to have little or no
conception of clock-time while doing their daily routine. Since the focus of this study is
on informal sector workers in low-income communities in Thailand, there is a need to
address these data collection issues. Chapter 4 discusses this new method in greater
detail.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 10
Research on time use patterns is very timely since time use patterns,
especially overlapping activities, can convey information about an individual’s quality of
life that conventional measurements currently do not (Floro 1995a, Floro and Miles 2003,
Folbre 1995). Overlapping activities can increase productivity since work and leisure can
be combined. However, by combining work and leisure time, we might actually see a
dramatic decline in a “pure” leisure time (Floro and Miles 2003). The effect of unpaid
labor on non-market production, social reproduction, human development, and economic
growth is gaining concern by the researchers, and the time use pattern (especially, on
overlapping activity) can provide us a more accurate contribution of each individual on
non-market production of goods and services (Folbre 1997, Floro and Miles 2003).
1.4 ORGANIZATION OF DISSERTATION
The dissertation chapters proceed as follows. Chapter Two introduces
important findings in both the theoretical and empirical literature that are relevant for this
study and have been advanced by other authors. The literature review surveys the
literature on well-being measurements, determinants of well-being, and the time
allocation data collection. Chapter Three develops a theoretical model of well-being
measurement to explain the relationship of individual well-being and its determinants.
This chapter begins with a model of a household well-being function by Floro (1995a).
This model is then expanded from a single person household to a multi-person household,
incorporating factors such as individual self-esteem and social status that also affect well
being. Chapter Four presents the methodology for constructing the well-being index
based on three determinants of well-being. It also presents the time use data collection
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 11
methodology, particularly the newly adapted time use data collection techniques
appropriate to conditions faced by workers in poor households in developing countries.
Chapter Five gives background information on the economic and social
situation in Thailand and outlines the pertinent characteristics of households and
individuals in the sub-sample drawn from the Bangkok Urban Poor Home-Based Worker
Survey. This section also provides a framework for understanding the results of the well
being index detailed in Chapter Six. Chapter Six focuses on the analysis of the well-being
index associated with this theoretical model. This chapter also introduces an alternative
well-being aggregation method that yields normative significance.
To conclude, Chapter Seven synthesizes the main findings of the thesis and
discusses the implications of the well-being measurement to the home-based workers in
Thailand.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 12
CHAPTER 2
REVIEW OF THE LITERATURE
The level of an individual well-being could be measured directly through its
constituents, e.g., happiness, or indirectly though its determinants, e.g., income. Some
researchers also believe that a person’s well-being is affected by the availability of
individual time and with the set of factors that determine the manner in which that time is
used. This chapter discusses the theoretical and empirical literature on well-being
measurement and the determinants of well-being, which includes how time is used. This
chapter also discusses the recent literature on time use data collection and its
measurement.
2.1 THE WELL-BEING MEASUREMENT LITERATURE
What is well-being? How can we determine the level of well-being, welfare,
or the quality of life of each individual, group of people, or people in one country? These
kinds of concerns have prompted many social scientists to study and analyze the sources
and determinants of well-being. The earliest economists believed individual and social
welfare were directly associated with the production of goods and services through the
market. Adam Smith (1776), Pigou (1952), and Marshall (1961) stated that increased real
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 13
output will increase both individual and social welfare (Juster and Dow 1985).1 However,
in recent years, economists have begun to include a broader range of goods and services
flow, including non-market activities such as human capital in their analysis of welfare.
Sen (1993), Dasgupta (1999,2001), Floro (1995a), and Juster and Stafford (1985) have
argued it is necessary to include non-market activities such as time allocation, work
intensity, and civil and political rights and freedoms of welfare. Floro (1995a, 10) defines
the notion of well-being as “ ... the physical, social, and mental development of human
capabilities of its members by means of access to and consumption of basic commodities
such s food, health care, education, shelter, etc., as well as through participation in
activities.” Some economists have also borrowed and integrated tools from other fields
of social science into their work. Ferrer-i-Carbonell (2002a, 2002b) and Van Praag
(1991) utilize the subjective well-being (SWB) concept in their work, for instance. In
addition, there have also been attempts to supplant the traditional concept of welfare
measurement (GDP per capita, for example) with a more comprehensive index of welfare
such as the Human Development Index (HDI) and Gender-Related Development Index
(GDI). Dasgupta (1999, 2001) also constructs the index of social well-being which is
based on several individual components such as per capita income, life expectancy at
birth, infant survival rate, adult literacy rate, and indices of political and civil rights.
1 Here, only material goods and services constitute the basis for assessment of well-being and economic development. For example, the real gross domestic product per capita has been used as the common measure of development progress.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 14
2.1.1 Well-Being Measurement
This section begins with discussion of the Human Development Index and
continues with other research in the area of well-being measurement. The Human
Development Index (HDI) developed by United Nations Development Programme
(UNDP) is widely used as an over-all measurement of well-being. The HDI has been
2 published every year since 1990 in the UNDP’s Human Development Report. The HDI
measures how well a country, region, or province has performed, not only in terms of real
income levels, but also in terms of the social indicators of people’s ability to lead a long
and healthy life, and to acquire knowledge and skills. The Human Development Index is
based on three attribute indices -life expectancy at birth, educational attainment, and
standard of living. Educational attainment is measured by combining adult literacy (two-
third weight) and combined primary, secondary, and tertiary enrolment ratio (one-third
weight). The standard of living is measured by real per capita income at purchasing
power parity, PPP. To construct each attribute index, fixed minimum and maximum
values have been established for each of these indicators as:
a) Life expectancy at birth: 25 years and 85 years
b) Adult literacy rate: 0% and 100%
c) Combined gross enrolment ratio: 0% and 100%
d) Real GDP per capita (PPP$): $100 and $40,000.
For any component of the HDI, attribute indices can be computed according
to the general formula:
2 The UNDP’s Human Development Report also includes several other composite development indices, which are the human poverty indices and indices of gender empowerment and inequality.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 15
Actual xt - Minimum x, ^ ^ Maximum xl - Minimum jc(
Where / = Index;
x, = Value of component x for individual i.
However, the income component index is a little more complex. The income
component index is a proxy for a decent standard of living. The basic approach in the
treatment of income has been driven by the fact that achieving a respectable level of
human development does not require unlimited income. To reflect this, income has
always been discounted in calculating the HDI (UNDP 2001). Therefore, the index is
calculated as:
W(y) = logy~ feg.-flfr.- , y = real GDP per capita (2.2) logTmax -l° g Tmi„
The HDI is widely accepted as a good measurement of well-being. However,
researchers such as Dasgupta (1999,2001) and Folbre (1997) have criticized it by
pointing out that it focuses only on a subset of important dimensions of human
development; its measurement of those components is arbitrary; it is extremely non-linear
in adjusted income, i.e., there are trade-offs implicitly built into the measure; it is too
strongly correlated with GDP per capita; and the use of GDP per capita as a measurement
of standard of living is questionable since it does not include the depreciation of capital
assets. Therefore, they argue, it is incapable of reflecting future prospects.
Adapted from the Human Development Index (HDI), the Gender-Related
Development Index (GDI) was introduced by United Nations Development Programme
(UNDP) in 1995. The GDI measures similarly to the HDI, but it also captures inequalities
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 16
between women and men (UNDP 2001). It is based on the same three components as the
HDI, which are a) female and male life expectancy at birth (longevity), b) female and
male adult literacy rates and combined primary secondary and tertiary enrolment ratios
(knowledge), and c) estimated female and male earned income, reflecting the difference
between women’s and men’s command over resources (decent standard of living). If
there were gender equality in human development, the GDI and the HDI of any given
country would be the same. The greater is the gender disparity in basic human
development, the lower is the country’s GDI compared to its HDI. The methodology of
developing the GDI is based on the Gender-Equity Sensitive Indicators (GESI or xede) to
correct each of the HDI variables for the gender inequality and aggregating them (Anand
3 and Sen 1995). However, the procedure of aggregating different focus variables in the
inequality correction process is considered to be somewhat deceptive, “since the different
variables might work in somewhat opposite directions, moderating the influence of each
other in the inequality between individuals” (Anand and Sen 1995, 10).
Other well-being measurements have been developed in recent decades
including one by Dasgupta (1999,2001). His version of well-being measurement is based
on the same components as the HDI. However, His measure also includes political and
civil rights variables. The political rights attribute index measures the rights of citizens to
play a part in determining who governs their country, and what the laws will be
(Dasgupta 1999, 2001). The index of civil rights measures the extent to which people are
able to openly express their opinions without fear of reprisals (Dasgupta 1999,2001).
3 The further information on the calculation and certain properties of the Gender-Equity Sensitive Indicators (GESI) can be found in Gender Inequality in Human Development: Theories and Measurement (Anand and Sen 1995).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 17
Both the political rights and the civil rights indices are coded in a range from 1 (highest
degree of liberty) to 7 (lowest degree). The methodology used in computing this well
being measurement is based on the Borda rule as each of these indices are strictly ordinal
in contrast to the rest of the well-being measurement variables which are based on
4 cardinality measurement. The Borda rule provides a method of rank-order scoring, the
procedure is to award each alternative a point equal to its rank in each criterion. For
instance, suppose a given country ranks a, b, c, d, and e for each of the well-being
component variables A, B, C, D, and E. Then, the Borda score is calculated by summing
all of the ranks, or (a+b+c+d+e). This rule provides a complete ordering of alternatives.
The Dasgupta’s well-being measurement has two underlying assumptions, namely: a)
that the components of well-being, as well as overall well-being, are comparable between
both individuals, and groups and b) that all of the well-being components are aggregated
based on a strictly ordinal measure. According to him, life expectancy at birth has
become the closest approximation of quality of life.
Quality of life or well-being can also be influenced by satisfaction gained
from the time spent on specific activities such as leisure activities (Juster and Dow 1985,
Juster and Stafford 1985, Floro 1995a, Floro and Hungerford 2001, and Floro and Miles
2003). Juster and Dow (1985) and Juster and Stafford (1985) point out that time use data
provide vital information about well-being. They argue that “time plays a crucial role not
only as an input into a variety of market and non-market production activities, including
leisure, but that time use is equally important as a direct source of satisfaction” (Juster
4 For example, the measure of private consumption in dollars or in cents does not matter, as long as we remember that the latter is one-hundredth of the former. This enables us to say that someone consumes twice as much as another person, or that someone today consumes three times as much as another person.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 18
and Dow 1985, 339). Each individual is considered to have a unique preference on how
he/she uses his/her time, and the outcome of that unique preference is the satisfaction
from the activities themselves, which could be a measure of well-being. Their
methodology on developing their well-being measurement called Process Well-Being
(PWB) is based on responses from the survey participants. Respondents were asked to
rank, on a scale of 0 through 10, a set of 22 pre-specified activities, e.g., work, cleaning
the house, watching television, playing with children, etc. (Juster and Dow 1985). PWB
is then based upon the time spent in various activities as stated above, weighted by the
process benefits for each activity, or mathematically,
P W B ^w /, (2.3) <=1
where wt indicates a measure of satisfaction with activity i, and ti is the number of hours
during a certain accounting period. However, wt is fundamentally subjective based on
each respondent’s perspective. To correct this problem, w: was transformed by Juster and
Dow (1985) into a mean-adjusted data process well-being (PWB-M) and a mean-and-
variance-adjusted data process well-being (PWB-V) as shown in the following equation:
1 -A *,=y, — 2 > / (2-4)
The variable yt denotes the raw score stated by the respondent for activity i and m is the
number of activities for which a raw score was available for a particular respondent.
Hence, equation (2.4) becomes:
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 19
1 M yt—m~{ Zjo W: = (2.5) v<*(yi - ym)
Thus, all of the respondents were given not only a uniform mean preference of zero, but
were also standardized by the personal variance in responses (Juster and Dow 1985). By
using the PWB developed from time spent on each activity as a proxy for well-being,
Juster and Dow (1985) reached the conclusion that females have lower process well
being score than males for all three methods. Further, age is positively related to the
process well-being, and having children tends to lower the level of process well-being.
The assumption that well-being is comparable between individuals still underlies this
measurement.
Other methods of developing a well-being measurement involve life
satisfaction or subjective well-being question techniques. This method has mostly been
utilized by sociologists and psychologists, including Cantril (1965), Wilson (1967), Ross
and Willige (1997), and Diener and Suh (2000). However, some economists were also
adopted the concept of subjective well-being, e.g., Oswald (1997), Pradhan and Ravallion
(2000), Graham and Pettinato (2002), Ferrer-i Carbonell (2002a, 2002b), Ferrer-i
Carbonell and Frijters (2002), Van Praag, Frijters, and Ferrer-i Carbonell (2002). The life
satisfaction concept can be separated into two themes namely, satisfaction with life as a
whole and domain satisfaction. Domain satisfaction relates to individual satisfaction in
different domains of life such as health, financial and employment situation. Satisfaction
with life as a whole can be seen as an aggregate concept, which can be folded into its
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2 0
domain components as illustrated in figure 2.1 (Van Praag, Frijters, and Ferrer-i
Carbonell 2002; and Ross and Willige 1997).
Figure 2.1: General Satisfaction and Domain Satisfaction Explanation
^ Financial Satisfactioii ^ House Satisfaction—- ^ Health Satisfaction General Satisfaction Leisure Satisfaction'' Environment Satisfaction
Sources: Van Praag, Frijters, and Ferrer-i Carbonell (2002, 31)
Most studies on the question of subjective well-being or life satisfaction share
the following structure. Respondents are asked how satisfied they are with their life as a
whole or with a specific domain within it. Participants can respond in terms of verbal
response categories, e.g., “dissatisfied”, “satisfied”, and “very satisfied”. Alternatively,
their responses can be categorized numerically from 0 to 10, where 0 correspond to “most
dissatisfied” and 10 refers to “most satisfied”. This methodology of well-being
measurement is based on two assumptions that responses are comparable and that they
adhere to strict ordinal measurement. Therefore, when two respondents give the same
answer, it is assumed that they enjoy similar satisfaction levels, and ordinal interpersonal
comparability is permitted. Economists utilize ordered probit or logit models with control
variables such as age, income, gender, and education to best analyze this type of data
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 21
(Van Praag, Frijters, and Ferrer-i Carbonell 2002). For a better understanding, the model
of domain life satisfaction can be shown mathematically as
G S - f (DSl DSj; z) (2.6)
where, GS = General life satisfaction;
DS = Domain satisfaction; and
Z = Common unobservable variables or part of the error term.
The domain of satisfactions is also a set of functions as presenting in equation
(1.7) where jc . stands for the subset of jc variables in domain j.
DSj = f(Xj,z) y= l,2,...,J (2.7)
The empirical tests conducted by Ferrer-i Carbonell (2002a, 2002b) and Van
Praag, Frijters, and Ferrer-i Carbonell (2002) on subjective well-being reached stable
significant and intuitively interpretable results. Hence, they conclude that the assumption
of interpersonal ordinal comparability of satisfactions cannot be rejected.
2.1.2 Determinant of Well-Being
The various methods of measuring well-being discussed in the previous
section have recently caught the attention of economists, including, Dasgupta (1999,
2001), Folbre (1997), Wolff and Zacharias (2003), Juster and Dow (1985), Anand and
Harris (1994), Anand and Sen (1995), and Sen (1993). A broader understanding of well
being beyond the notion of income has evolved in economics. Factors such as education
are also being recognized as important in affecting an individual’s quality of life. The
following section discusses the determinants of well-being that are utilized in the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2 2
analytical framework/model. They are personal income, the level of educational
attainment, and the incidence of work intensity.
2.1.2.1 Personal Income
Income is the most common measurement of an individual’s standard of
living. Since income is not desired for its own sake, any income-based notion of well
being must refer -directly or indirectly- to those basic ends which are promoted by
income as means (Sen 1993). Individual well-being does not depend on income (as
means) only in absolute terms (e.g., directly related to consumption of goods and
services), but also on the subjective perception of whether income is adequate to satisfy
one’s needs (e.g., social dimension). In general, income affects individual well-being by
its direct effect on the level of consumption of purchased goods and services. It also
impacts the individual’s bargaining power within household, and relates to psychological
well-being through its effect on social status and self-esteem.
a.) Material Well-Being
The effect of income on the consumption of goods and services is straight
forward. Economists focusing on goods and services have associated larger real output
with increased individual and social welfare since the time of Adam Smith (1776/1993),
Pigou (1952), and Marshall (1961). Standard economic models assume increased on
income affects material well-being positively, under the premise that more is better and
increases in income will be desirable from an individual’s perspective. Higher incomes
allow people to consume not only more, but also higher quality of goods and services
allowing individuals a higher standard of living or well-being (at least in a material sense).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 23
In other words, personal well-being is generated by access to and consumption of these
purchased commodities (Sen 1993). However, some economists concentrate not only on
what people get from goods and services, but also on the individual mental/intellectual
reaction to those commodities and services (Cohen 1993, Sen 1993). These economists
are concerned not only with how much physical nourishment a person gets from food, but
also with how much enjoyment or happiness he/she derives from that nourishment.
b.) Household Bargaining Power
What might be less obvious is the effect of personal income on household
bargaining power. Manser and Brown (1980), Lundberg and Poliak (1993), McElroy and
Homey (1990), and Agarwal (1997) have explored the effect of intra household relations
on household and individual behavior and well-being. Several models have been
developed in recent decades. These include the consensus and altruist model, the
cooperative bargaining model, and the non-cooperative bargaining model. Many studies
suggest that the well-being of an individual within a family depends on the income
distribution between household members (Lundberg and Poliak, 1993). The bargaining
power of individual household members is related to their fall-back position or threat
point5 The threat point is determined by how well one would be able to sustain
independent living if cooperation fails. Individual earning or income is considered one of
the most important determinants of household bargaining power. According to the
cooperative bargaining model, each family member has input into the decision making
5 The fall-back position or the threat point (the outside options which determine how well off one would be if the cooperation failed) is mainly focused on determinants, such as income and education, which can sustain the independent living.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 24
process, where the weight given each person’s input depends upon his or her threat point
or the set of opportunities outside the household. In this model, household decisions are
made as result of a cooperative game.6 Hence, given an exogenous increase in the income
of one household member, household demand for goods and services should shift to more
closely reflect that person’s preferences, resulting in a higher state of well-being of that
household member (Manser and Brown 1980).
Agarwal (1997) also analyzed to the determinants of household bargaining
power. She notes that for developing countries, the determinants of household bargaining
power might be different from those existing in developed countries. These include
economic assets, access to resources, state and local support systems, social norms, and
individual perception about need and deservedness. She also suggests that the fully
specified bargaining model, may not be appropriate since “the outcomes of bargaining
need not result from an explicit process negotiation between the parties; they could even
result from implicit differences in bargaining power” (Agarwal 1997,10).
c.) Social Dimensions
Following earlier discussions, individual well-being is affected by the
personal income not only in absolute terms, but also by subjective perception. Income
perception is based on current and past income, as well as income in relation to that of
other people. The latter reflects the importance of the relative position of individuals in
society for their satisfaction with life. This is often referred to as the comparison income
Typically Nash equilibrium, which assumed that each household member can make binding, costlessly enforceable agreements with one another
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 25
or relative well-being effect (Ferrer-i Carbonell 2001a). The comparison income or
relative well-being effect is influenced by the individual’s social status and self-esteem,
with higher social status or self-esteem increasing individual well-being. According to
some sociologists, status is defined as evaluated social standing, the overall evaluations
people make of each other, with status being evaluated on the basis of various “status
claims” such as education, occupation, and income. By utilizing the status attainment
model with respect to various status claims, sociologists such as Campbell and Henrentta
(1980,1978) and Colman and Rainwater (1978) have shown that income or earnings has
the strongest effect on maintaining one’s status. However, Campbell and Henrentta
(1980) also found that it is unlikely that people primarily seek income in an attempt to
maximize their status, but rather to maximize their short and long run consumption
(Campbell and Henrentta 1980).
2.1.2.2 Education: Determinant of Well-Being
Education is another important determinant of well-being. Higher educational
attainment is associated with lower unemployment and higher wages, higher family
income, and better health for adults and their children. Many studies have shown that a
substantial portion of the gap in well-being among racial and ethnic groups can be
attributed to differences in educational opportunities and attainment. For example,
Donohue and Heckman (1991) found that improvements in the economic status of
African American in the 1960s and early 1970s resulted in part from improvements in
educational attainment and school quality. According to psychological and sociological
studies, education gives people access to non-alienated paid work and economic
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 26
resources that increase the sense of control over life, explaining much of education’s
positive effect on psychological well-being (Ross and Willigen 1997). Education,
therefore, helps enhance personal welfare through various mechanisms both monetary
and non-monetary based, including increased earnings potential, improved health, and
personal development (i.e., self-esteem).
a.) Income Effect o f Education
The relationship between education and income has long been studied by
economists. Mincer (1958,1974), Arrow (1962), Merrett (1966), Lucas (1988) and
Romer (1986) consider education as human capital. Hence, increased investment in
education or human capital yields economic benefits and social return, in terms of higher
labor skills, higher earning potential and greater enjoyment of life. There is ample
evidence that shows education improves the labor productivity of workers, resulting in
higher earnings (Mincer 1958,1974; Light 1998,2001; and Heckman, Farrar, and Todd,
1996). A study by Andolfatto, Ferral, and Goome (2000) found that individuals who
invest heavily in education or human capital tend to experience higher levels of earnings
and income throughout most of their life-cycle.
Some economists, however, have argued that studies have overestimated the
return of education on income since other factors - like the ability of workers, race and
sex, and work experience during school years have been ignored (Light 2001; Wiess
1995; Klein, Spady, and Weiss 1991; and Griliches and Mason 1972).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 27
b.) Health Aspects
Economists, psychologists, and sociologists, are also analyzing positive
associations between education and the health of individuals (Ross and Wu 1995,1996;
Ross and Willigen 1997; and Ferrer-i Carbonell 2002b). Education can affect the health
of individuals in many ways, in that higher levels of education can lead to better
understanding of nutrition, resulting in a healthier life. For instance, Ross and Wu (1996,
719) pointed out that:
Well educated people experience better health than the poorly educated, as indicated by high levels of self-reported health and physical functioning and low levels of morbidity, mortality and disability. In contrast, low educational attainment is associated with high rates of infectious disease, many chronic noninfectious diseases, self-reported poor health, shorter survival when sick, and shorter life expectancy (Feldman, Makuc, Kleinman, and Comoni-Huntley 1989; Guralnik, Land, Fillenbaum, and Branch 1993; Gutzwiller, LaVecchia, Levi, Negri, and Wietlisbach 1989; Kaplan, Haan, and Syme 1987; Kitagawa and Hauser 1973; Liu, Cedres, and Stamler 1982; Morris 1990).
Education also affects health through the types of work and economic
conditions that educated employees have access to (Ross and Wu 1996). Well educated
individuals are more likely to be employed as full-time workers than poorly educated
ones. Their incomes are also expected to be higher, making them less likely to experience
economic hardships. Economic hardship negatively impacts the health of the individual
since it raises the level of depression, stress, and hopelessness, which decrease the
resistance to disease (Syme and Berkman 1986). Another positive relationship between
education and health can be found in personal healthy life style choices (Ross and Wu
1996). Well educated people tend to engage in a more healthy behavior than those less
educated. For example, well educated individuals are more likely to never have smoked
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 28
or to have quit smoking (Winkleby et al. 1992 and Shea et al. 1991). Educational
attainment is positively correlated to higher levels of physical activity as well, creating
better health outcomes (Ross and Wu 1996 and Shea et al. 1991). The well educated also
tend to drink more moderately than the poorly educated. Lower levels of educational
attainment are more likely to be associated with alcohol abuse (Midanik, Klatsky, and
Armstrong 1990). Alcohol, in general, adversely affects individual health and well-being.
c.) Personal Development
Education is also positively associated with individual well-being through its
effect on personal development. With high level of educational attainment, individuals
have greater ability to make informed decisions, and to develop better attitudes toward
themselves, thereby enhancing self-esteem. Education helps individuals successfully
adapt to the challenges facing them and attribute their successes to themselves, thereby
viewing themselves as capable and worthy and definitely improving their self-esteem
(Owens, Mortimer, and Finch 1996). Sociologists separate the concept of self-esteem into
global self-esteem and specific self-esteem, where global self-esteem refers to the
individual’s positive or negative attitude toward himself/herself as a totality. Specific-self
esteem deals with one dimension of an individuals’ self concept (Rosenberg et al. 1995,
Owens 1994, Hoelter 1986). People may have attitudes toward an object as a whole or
toward a specific aspect of the object. For instance, students may have attitudes toward
the university as a whole, while having different attitudes toward specific departments or
professors in the university. Thus far, only the global self-esteem has been found to be
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 29
7 strongly related with psychological well-being (Rosenberg et al. 1995). Maintenance of
self-esteem leads to self-protective motives, self enhancement processes, and a variety of
coping processes, reducing the risk of depression and distress (Rosenberg 1985 and
Rosenberg et al. 1995).
Since education helps develop personal ability through better decision
making, improving self-esteem, etc., this personal development also affects the
bargaining power of the individual. The effect of improving bargaining power on an
individual’s well-being was discussed earlier.
2.1.2.3 Incidence of Work Intensity
In the past, economists tended to associate well-being and utility mainly with
“leisure” time. Leisure time has been utilized in many theoretical models of labor force
participation, and household production. The process of performing work, however, also
conveys information on the individual quality of life. The incidence of work intensity is
an important aspect of the work process that has received little attention in economics
research until recently. A possible reason for this is the difficulty in measuring and
g documenting this dimension of work time (Floro 1995a, and Juster and Dow 1985). As
discussed previously, conventional measures of living standards do not include any
considerations of length or intensity of work time performed by an individual. For
purpose of this analysis, work intensity is defined as two or more tasks performed
7 Psychosocial well-being includes depression, anomie, general anxiety, resentment, anxiety-tension, irritability, life satisfaction, happiness, negative affective states, etc. 8 This is different from the broader notion of time intensity, which involves performing any two activities at the same time, since some of these activities can be pleasant and enhancing activities.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 30
simultaneously by an individual such as mending clothes while caring for children, or
assembling shoes while waiting for the rice to cook (Floro 1995a; Floro and Miles 2003;
9 and Floro and Hungerford 2001). The level of work intensity affects an individual’s
quality of life in many ways through its effect on health/stress, work productivity, and
income.
The correlation between the level of work intensity and the level of
individual well-being is not always either positive or negative. Increased work intensity
can increase an individual’s well-being by increasing his/her output. This increase in
output can then result in a higher quality of life for the individual through an income
effect. As discussed in Floro (1995a, 6), “any gain in the amount of output produced per
unit of time is due to an increase in the energy (mental or physical) expended by the
worker by undertaking two activities simultaneously.” However, this increase in output
can also come with an associated cost such as a deterioration of output quality. When an
individual needs to shift attention frequently from one task i.e., -cooking- to another i.e.,
-cleaning- it might be more likely that food will get burned or that some cleaning chores
are missed.
An example of how increased work intensity is negatively correlated with
well-being is that of stress. There can be significant health/stress effects associated with
increased work intensity. Performing two or more tasks over prolonged periods can
create stress, sometimes without the individual even noticing. This increase in stress often
9 However, if we base the productivity with the unit of time (linearity in time and ignoring the individual effort), then the productivity increase since based on the same amount of time, the output increases.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 31
leads to an overall deterioration of both physical and metal health (Wolfe and Haveman
1983, Roldan 1985, and Floro 1995a).
The effect of work intensity on individual well-being depends both on the
length of time an individual multitasks and the nature of the activities performed, along
with the amount of energy and the level of concentration required (Floro 1995a, Floro
and Hungerford 2001).10 The longer the time an individual performs two or more
simultaneous tasks (paid work or unpaid work), the greater is the amount of stress
generated from the work process. The nature of activities is also important. If an
individual needs to combine tasks, which both require considerable energy (physical or
mental) or uninterrupted attention, or have a tight dead line, the negative effects of this
increased work intensity on well-being is greater. On the other hand, if an individual
combines two pleasant activities which do not require continued high levels of
concentration or energy -such gardening while performing passive child care- the well
being of the individual is not necessarily negatively affected by this overlapping (Floro
1995a).
The length of the work day, which includes the time spent on both paid
market work and unpaid household work, also affects well-being.11 The length of the
work day can affect individuals both positively and negatively. At the lower end of the
spectrum; increases in the length of the work day are positively related to well-being as
this increased work leads to greater commodity creation and higher consumption with a
10 For example, individual tends to be under greater stress than normal in working on two or more tasks simultaneously when the uninterrupted attention, a due date, or some other conditions need to be fulfilled. Note that the measurement of the working day becomes more complicated recently since individuals tend to do more multitasking.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 32
corresponding higher standard of living. However, when the length of the working day
increases beyond a critical point, time spent on leisure and social activities are necessarily
reduced with the corresponding decline in well-being becoming unavoidable.
2.2 TIME ALLOCATION LITERATURE
Time allocation or time use data were first collected by the Union of Soviet
Socialist Republic (USSR) in the early 1920s, and followed later in the early sixties by
several European countries (Floro 1999). Time use data were initially collected mainly by
12 developed countries since the process was associated with high collection costs. The
first time allocation survey in United States was conducted in 1965 by researchers at the
University of Michigan . More than 1,300 American adults aged 18-65 participated. In
that time allocation survey, time use diaries were used to collect the data. Since then
several time use surveys have been conducted in US -in 1975 by the University of
Michigan and in 1985 by the Survey Research Center at the University of Maryland.
Although time allocation data are quite challenging and costly to collect, these surveys
were conducted in order to arrive at a better estimate of the value of goods and services
produced. This information on non-labor force use of time such as commuting, leisure
and recreation, and household work and child care helps give a fuller picture of quality of
13 life and aids in constructing more complete national income accounts (Folbre 1997).
An important question when utilizing time use data is whether or not they are
comparable in terms of sampling design and data collection method so that the patterns of
12 These, generally, are associated with both tangible and intangible cost such as training cost and time. 13 Folbre (1997) also suggests a term for these two main motivation as “Gross Domestic Product (GDP)- Oriented” and “Welfare-oriented” approaches, which widely accepted by most economists.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 33
time allocation between countries or even between cities in the same country are
comparable. Thus far, there has been no standardized time use classification system
although this issue is gaining more attention and several European countries’
governments and researchers are working towards this goal. Standardization of time use
data has proved difficult, partly because the socio-economic and cultural context of data
collection varies widely -both within and between countries. For example, even within
India it is difficult to develop a classification of time use that would be equally relevant
for middle-class women in Delhi and tribal women living in a largely subsistence
economy. Standardization of the classification system is also a gender issue because both
the pattern of time use and the meaning of specific categories of time use differ for
women and men, even within the same socio-economic and cultural context. For example,
a time use survey in the Philippines found that men reported heating food previously
cooked by their wives and boiling water to make coffee as “cooking”. Analysis of the
1992 Australian time use survey also found that the majority of simultaneous activities
reported by women involved two “work” activities, such as childcare and housework, or
cooking and housework, while those for men involved rest and recreation combined with
childcare (Australian Bureau of Statistics 1994). Further difficulties arise from the fact
that the primary motivation for conducting time use surveys varies among countries.
However, many international organizations, such as the United Nation’s Statistic
Department, have been working to address these problems. United Nations Development
for Women (UNIFEM) and Economic and Social Commission for Asia and the Pacific
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 34
14 (ESCAP) also tried to address the problem of time use data standardization as well. For
some time, at both the global and regional (Asia-Pacific) levels, statisticians and
researchers have been working to develop a common set of classifications of time use
data. In particular, the UNDP regional gender project, APGEN, has worked with India,
the Philippines, the Republic of Korea and several other Asia-Pacific countries that have
undertaken or plan to undertake time allocation surveys to provide a standard set of
guidelines for the implementation of such surveys and a standard classification system to
facilitate national and cross-country data comparisons.
The methods currently used to collect time use data in both developed
countries and developing countries include the recall/estimated method, the time diary
method, the direct observation studies, and the experience sampling method (ESM) -
which is a random beeper method. However, many researchers still conclude that the
time diary method, which records the chronology of various activities in a given time
period (day), is the most valid and reliable method (Juster and Stafford 1991, Robinson
1985, Robinson and Godbey 1997, Kalton 1985). The following section gives brief
descriptions of how time use data are collected with each method.
14 From my experience as an observer at the UNIFEM and ESCAP project meeting on Gender Approach to the Collection and Use of Statistics for the Informal Sector, Homework and Time Allocation in Bangkok, Thailand.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 35
2.2.1 Recall/Estimated Method
While data is easiest to collect with the recall method, it is associated with
high measurement error.15 Data is collected using a stylized activity list, which asks
questions like the following: How many hours did you work yesterday? On average, how
many hours do you watch TV per day? or How much time do you spend cooking? It costs
less to collect data with this method than it does with others, although one always needs
to remember that the results are much less reliable. The other drawback is that the
stylized activity list might not be able to capture some activities which happen
infrequently or rarely.
This methodology has also shown a bias toward over-reporting or
overestimating (Robinson 1985, Robinson and Godbey 1997, Juster and Stafford 1991,
and Juster 1985). When survey respondents are asked to provide daily and weekly
estimates of several activities, most survey respondents give estimates that add up to
considerably more than the 168 hours of time available each week.16 Robinson and
Godbey (1997, 59) give the following as example of this problem: “ average
estimated weekly times totaled 187 hours, and their list of activities did not include time
for church-going, shopping.... In Hawes et al.’s (1975) national survey, estimated weekly
activities averaged at more than 230 hours, and our own studies of college students, the
totals reached more than 250 hours.” Godbey (1997) also argued that time allocation data
The recall/estimated method relies heavily on human memory of past events, which has a high tendency to be inaccurate when the events happened in the past (e.g., three days or a week). 16 The activities in this survey include most common activities such as work, leisure, volunteer work, personal care, sleep, etc.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 36
on labor force participation collected with this method can also give researchers
17 inaccurate results when questions are asked in the following manner:
On average, how many hours do you work per week?
Your answer: hours
2.2.2 Time Diary Method
The time diary method is considered a special type of recall/estimated
technique since the events or activities are reported in a diary shortly after the event or
activity has occurred. This method takes into account the fact that human memory is
short, so in order to get the most accurate data, memory must be transferred from the
human brain to the diary as quickly as possible. The time diary method is not only likely
to give more accurate results, but also has other advantages such as being comprehensive.
It enables the survey respondent to report all of the activities in self-descriptive terms
which can then be subjected to uniform coding decisions (Juster 1985). An example of
the basic open interval time diary is given in Table 2.1.
This is again due to the fact that recall/estimated method relies on human memory, which is most likely to be inaccurate when the events happened in the past. This question has a high tendency to yield an overestimating or underestimating answer (Godbey 1997).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 37
Table 2.1: Basic Time Diary Format
Principal Start End What else Where Who For whom Activity: what Time Time were you were were with were you were you doing? doing? you? you? doing this? Midnight
Midnight
The design of the time diary format in Table 2.1 is able to capture both
primary and the secondary activities (overlapping activities). The “with whom are you
doing this” question can help indicate whether this activity is leisure or work. This free
form time diary can be adjusted naturally to suit the purpose of any time use project, e.g.,
the 1994 Australia time use survey. It can be argued that information from a single day
time diary might not give us accurate statistical results (based on the time diary data)
since the day when the recording takes place may not be typical. For instance,
respondents may spend more time than usual working or they might be sick and stay in
bed all day. One day time diaries might also suffer from seasonal changes. An example of
this might be the farmers whose work has to be done in the spring with less work being
done in the winter time. These difficulties can be overcome with large and randomly
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 38
18 drawn samples. When these two conditions are met, we can, statistically, expect that
respondents who record more time than usual at work or in home on their diary should be
balanced by others who are spending less time than usual (Robinson and Godbey 1997).
The time diaries are considered to be the most effective and reliable method
of collecting time allocation data, although they are costly for the surveyor and time
consuming for the survey respondent (Juster and Stafford 1991, Robinson 1985,
Robinson and Godbey 1997, Kalton 1985). It is costly to train the interviewers so that
they are able to communicate and explain exactly what the survey respondents have to do
with the time diary. Also, the interviewers need to make sure that all of the respondents
have the same understanding on important terms such as work, leisure, etc. It is time
consuming for the survey respondents to keep a journal all day long. In some cases, the
respondents might refuse to cooperate with the project due to their own time constraints.
Robinson (1997, 7) also points out that “we know no more than what people are able or
willing to tell us in this reporting framework. If they want to distort their accounts, we
have only limited ability to control or correct them. Thus, only a few respondents report
engaging in sexual or other personal biological activity in their diary accounts.” To test
reliability and validity, Kalton (1985) looked at time use data collection methodologies
used on the samples from Jackson and Ann Arbor, Michigan. He concluded that the time
diary approach does not appear to generate seriously inflated or deflated aggregate time
figures for any particular set of activities; nevertheless, the other methodologies do.
Kalton (1985,95) suggested that “First a probability sample of households must be selected; then individuals within households must be selected at random, and the dates for which time diaries are to be obtained from each respondent must be selected by a probability sampling method.”
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 39
Moreover, the time diary method is likely to produce highly replicable and reliable results
(Kalton 1985).
2.2.3 Direct Observation Method
Theoretically, the direct observation approach should be one of the best
techniques to collect time use data since the interviewers have a chance to directly
observe the activities which survey respondents do, and take note it in the time diaries.
This technique is similar to the time diary methodology. The difference is that the diary is
kept by the interviewers, not the survey respondents. The direct observation method is
costly and has some serious flaws. Because the observer has to be stationed within the
household for an extended period, it can be costly. The observers’ presence might also
alter the nature of the activities of the respondent since an observer would indeed be an
intruder into the household’s privacy. Moreover, Juster (1985) found that the direct
observation method is associated with a relatively high non-random refusal rate. In sum,
the direct observation method is preferred theoretically, but it does not work well
practically.
2.3 CONCLUDING REMAKRS
In sum, most of the recent well-being measurements such as the Human
Development Index by UNDP (2001, 2002,2003) and the index of political and civil
right by Dasgupta (1999, 2001 ) are based on the determinants of well-being. This is due
to the fact that constituents of well-being are difficult to measure. Even if measurable,
certain assumptions must be satisfied to make measures comparable between individuals.
The most well-known determinant of well being is individual income. However, Juster
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 40
and Dow (1985), Juster and Stafford (1985), Floro (1995a), and Floro and Hungerford
(2001) believe that time allocation is also important determinant since it conveys
information of individual well-being that traditional measures do not.
The incidence of work intensity calculated from time use patterns could be
one of the important determinants of well-being, yet both the theoretical and empirical
literatures do not incorporate it into current well-being measures. The following chapter
proposes an expanded model of individual well-being that incorporates individual time
allocation as a well-being determinant. The expanded model also addresses individual
well-being in relationship to other individuals in the same household. Most important,
this model of well-being measurement is then adapted to construct the well-being index,
using data from a home-based worker survey in Thailand.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 41
CHAPTER 3
ANALYTICAL FRAMEWORK
Since the locus of sensation, perception, and the feeling of well-being is at
the personal or individual level, it is more appropriate to develop a quality of life
measurement starting from that point. Along with the sensation and perception of well
being, socio-economic characteristics and time use are also experienced at the individual
level. Therefore, the well-being index developed in this dissertation is based on
individual level determinants.
In general, the idea of constructing a well-being measurement is to rank each
state of affairs numerically - the higher the state of affairs, the higher the number
awarded to it. “The state o f affairs includes in its description the allocation of resources
(who gets what, when, where, and why) and anything else deemed to be relevant for
personal and social choice” (Dasgupta 2001,14). Let us call the value which each
individual attaches to his/her personal circumstances in each state of affairs his/her
welfare (Dasgupta 2000,2001).1 Assume Y is the set of all possible states of affairs, and
Y consists of elements of a, b, c, etc ., Y = {a,6,c,...}. Assume further that we have N
1 This is referred by Dasgupta (2001,14) as “the goods and services he enjoys, his personal relationships, and those other aspects of life that affect him.”
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 42
individuals, labeled /, in the society. Let Ui {a) be the index of /’s personal well-being in
social state a . Further, for an individual to be in state a , the level of each individual’s
well-being determinants and/or constituents, which are a,P,S , and so on, must meet the
standard. U, (a ) can be thought of as the numerical index, and Ut (a) is built from the
personal level of the well-being constituents and/or determinants. Mathematically, we can
say that
£/,(a) = ir(a„/M,...) (3.1)
The greater the index number, the higher the well-being of individual i.
Ul (a) >- Ui (/>) signifies that f s personal well-being is greater in state a than in state/).
When discussing well-being constituents and well-being determinants, one must
remember that conceptually, they are different ways of measuring of individual well
being. Measuring an individual’s well-being based on the constituents of well-being
(experiential or desire-satisfaction), is basically measuring the output components of
well-being, e.g., health, happiness, enjoyment, freedom to be and do. One can also study
the determinants of human well-being, which are commodity inputs in the production of
well-being -such as food, clothing, clean water, shelter, access to information and
resources.
In sum, the constituents and determinants of well-being can be thought of as
being ends and means, respectively. Obviously, the constituents of well-being are hard or
impossible to observe by others. Because constituents are reported by the individuals,
how each individual actually measures them is important to researchers. Sociologists and
economists have measured constituents based well-being with a general life satisfaction
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 43
or well-being subjective questionnaire, as discussed earlier. Again, the subjective well
being/life satisfaction method relies heavily on an “interpersonal comparable
assumption,” which might not be appropriate in some cases. It is even difficult to
distinguish sometimes whether a characteristic or variable is a constituent or determinant
of well-being.
3.1. DETERMINANT OF WELL-BEING
As the constituents of well-being are difficult to observe, the subjective well
being methodology has relied on some very restrictive assumptions. Because of this, the
welfare studies based on the determinants of well-being are currently more in favor by
researchers, such as Dasgupta (1999,2001), Folbre (1997), Ferrer-i Carbonell (2002a),
Floro (1995a), Floro and Hungerford (2001), Juster and Dow (1985), Anand and Harris
(1994), Anand and Sen (1995), Sen (1993), Slottje (1991), and UNDP (2001,2002,
2003). Studying the means of well-being not only results in a well-being measurement,
but also conveys information on how to improve quality of life. This might be the case
where the well-being of individuals could significantly improve if policymakers provide
3 them with a better working/living environment. The well-being index developed in this
dissertation, therefore, is based on the determinants study, given it is likely to be more
measurable and efficient. A subjective well-being technique will also be utilized to obtain
2 Economists who have explored the notion of subjective well-being include Gibbard (1996), Clark (1997), Oswald (1997), Pradhan and Ravallion (2000), Ferrer-i Carbonell (2002a, 2002b), Ferrer-i Carbonell and Frijters (2002), Van Praag, Frijters, and Ferrer-i Carbonell (2002), Ferrer-i Carbonell and Vanpraag ( 2002). However, other social scientists including sociologist and psychologist have developed measurement and methodology of subjective well-being. To name a few, Cantril (1965), Wilson (1967), Ross and Willige (1997), and Diener and Suh (2000). 3 Some group of people such as homeworkers in informal sector, the working or living environment might be the main determinant of their well-being.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 44
a measure of quality of life based on constituents. This will then be used later to examine
the validity of the well-being index based on the determinants of well-being.
Material goods and services have long been used to assess peoples’ standard
of living. However, more comprehensive welfare measurements developed recently are
based on more than only material goods and services, such as life expectancy at birth,
adult literacy rate, and the level of political participation. Time allocation of individuals
also conveys important information about their well-being that conventional determinants
do not. Time use, particularly in home production activities yields utility not only through
the production and consumption of material goods but also through the intrinsic
satisfaction or enjoyment of activities themselves (Juster and Stafford 1985,1991;
Robinson and Godbey 1997; Floro 1995a; Floro and Hungerford, 2001; and Juster and
Dow 1985). Building on the work on household well-being by Floro (1995a), this
dissertation measures the quality of life for a given individual based on the following
determinants: personal income (or material goods and services), educational level, and
incidence of work intensity (based on time use patterns). These determinants could
diminish or enhance human capabilities and functionings, which finally result in a higher
or lower level of individual well-being. In the following section, I will explain the
rationale for choosing these three components and show their causality to individual well
being.4
Because income has long been the main indicator of ones standard of living,
personal income in this model will influence well-being via the level of consumption of
4 The complete discussion on the determinants of well-being could be found in chapter 2: The review of literatures.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 45
purchased goods and services; the individual bargaining power within the household; and
its impact in terms of social status and self-esteem. Figure 3.1 summarizes the effect of
personal income on well-being as discussed in Chapter 2. Personal income determinant
is positively correlated with the consumption of purchased goods and services, individual
bargaining power, and the individual’s social status and self esteem. As a consequence,
an increase in income enhances human capacities of individual, and leads to a
corresponding increase in individual well-being.
Figure 3.1: The Effect of Personal Income to Individual Well-Being
(+) Level of goods and (+) services consumption
(+) Household bargaining (+) Individual Personal income power well-being
Social status and self (+) indicates esteem of individual a positive (+) (+) relationship
As previously discussed, educational attainment is also related to an
individual’s quality of life. Similarly to personal income, higher levels of educational
attainment also enhance human capabilities, which lead to an increase in person’s quality
of life level. Well-being is affected through nutritional intake and health status; personal
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 46
development, i.e., self esteem; and the level of purchased goods and services consumed.
This is presented in Figure 3.2, which shows the positive relationship between
educational attainment and well-being.
Figure 3.2: The Effect of Education to Individual Well-Being
(+) Health Aspect (+)
Personal development: (+) Individual Educational Self esteem well-being attainment
Personal income: The (+) indicates level of consumption a positive (+) (+) relationship
Work intensity is also associated with individual well-being, but the
relationship is not so straight forward. Work intensity can be positively associated with
well-being through a corresponding increase in income and/or output as presented in
Figure 3.3. Work in this context refers to both paid market works and unpaid household
work, including childcare cleaning, cooking, house repairing, etc. However, an increase
in output due to higher work intensity can be accompanied by a deterioration of output
quality. The level of individual well-being can also decline as a result of health/stress
factors associated with high work intensity.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 47
This dissertation studies home-based workers, individuals who are likely to
be both “time-poor” and “money-poor.” They need to simultaneously work long hours on
paid market and unpaid household activities, with very little remuneration in return. An
increase in their output and/or income resulting from higher work intensity tends to only
slightly affect their quality of life. One reason is that an increase in output is often
accompanied by a deterioration of output quality, causing some of the increased output to
not pass quality control standards. As a result, income is not increased. Also, home-based
workers are likely to combine their paid and unpaid work in order to complete necessary
household chores. Their poverty necessitates them doing as much market work as they
can while their household obligations do not decrease. Therefore, it is likely to be
appropriate to assume that the negative health/stress effect outweighs the positive
output/income effect in the case of most home-based workers. Consequently, the
incidence of work intensity is negatively related to the level of individual well-being.
Figure 3.3: The Effect of Work Intensity on Individual Well-Being
(-) (-) ------Health/stress effect
Incidence of Individual work intensity well-being
------k . Output/income effect* (+) (+)
(+) indicates a positive relationship (-) indicates a negative relationship
*Note that an increase in output is usually accompanied by a deterioration of output quality.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 48
However, these three determinants are not the only determinants that affect
well-being of individuals. Other economic demographic and social factors, such as
presence of dependent members, receipt of social support from government agencies, and
living in a well organized and supportive community, could also affect individual well
being as well. The latter may be crucial and help enhance a person’s well-being since
living in a well organized community enables him/her to have access to a better market
opportunity and basic services such as health, water, sanitation, electricity, and public
transportation.
3.2 ANALYTICAL FRAMEWORK OF WELL-BEING
The theoretical model of individual well-being is based on the
5 household/individual well-being function by Floro (1995a). In standard household well
being function models, households are treated as micro-firms that combine market-
purchased inputs and household labor and time to produce commodities. The
commodities acquired either through market purchases or home production activities
have direct effect on the well-being of each household member. The household
production function can be explained with Zmi representing the market purchased
commodities consumed by the household, and Zhi representing home produced
commodities. Zhi is obtained by combining the household market purchased “goods”
(market inputs) with the amount of time in a household production function. Furthermore,
Floro (1995a) added time allocation -such as the length of the working day- work intensity into the basic household behavior model.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 49
with constant returns to scale in household technology and the absence o f joint
production, the household production function is in the form;6
Zhl=Zhl(X,Lh) (3.2)
Where X representing the market goods required to produce the commodities,
and Lh is the time input needed to combine with market goods in order to produce the
commodities. Both the market purchased commodities, Zmi, and the home production
commodities, Zhi, not only provide direct benefits by raising household satisfaction
through their consumption, but they also satisfy the necessary commodity needs of the
household for its reproduction and maintenance. Nevertheless, for simplicity,
consumption is not separated into home production or market purchased goods in this
model.
In reality, the household is not a unitary unit of consumption and production
as household members in most cases are engaged in some from of bargaining and
negotiation. This analytical model, however, does not address the issue of intra household
bargaining and decision making. This is a limitation of this dissertation that will be
addressed in future research. This analytical model, therefore, will assume that the
household is composed on one working adult member who makes certain decisions on
6 Poliak and Wachter (1975) argue “If these assumptions are not satisfied, commodity prices depend on the household’s consumption pattern. Hence, price differences among households reflect differences in tastes as well as technology. Since commodities prices are not determined solely by goods prices and technology but also reflect the consumption pattern chosen by the household, it is misleading to treat the demand for commodities as a function of “commodity prices.” Juster and Dow (1985, 401) give us the example of the joint productivity problem as “ ... The activity of cooking may be seen as having two outputs. One is a tangible meal and the other is ‘the subjective experience of having cooked for x minutes.’ If both of these outputs from the single activity ‘cooking’ influence utility, the shadow price for the produced good, the meal, cannot be determined independently of the preferences of the household for goods and time use.”
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 50
household resource allocation and one dependent household member. In short, we have
a multi-person household with only one decision maker in the model. It is further
assumed that the working adult is an altruist whose well-being increases when the well
being of other household members increases. Thus, a rise in the well-being component of
a dependent member would indirectly enhance the well-being of the adult member.
3.2.1 Components of Well-Being
The components of well-being for the working adult in the household with
the dependent can be divided into three categories: 1) the flows of commodities and
services consumed by the individual from both market and household production; 2) the
physical, social, and mental aspects of well-being, i.e., health status, and self-esteem and
social status; and 3) the well-being of dependent members in the household.
Mathematically, the well-being function of individual i in household n can be explained
as
Win =W(Cin,H i„,Sin,WJn) (3.3)
Where Wm = The well-being of a working individual i in household n\
Cin = Goods and services, which can be produced by home production or
purchased in the market, that are consumed by individual i in household
«;
Hm = The health status of individual i in household n, including physical,
7 This is suitable with the survey sample since most of the respondents are female with children. She makes some decision in the household, given the prevailing gender relations and social norms.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 51
emotional, and mental aspects;
Sm = The self-esteem of individual i in household n, i.e., the individual’s
attitude toward himselfTherself, and the social status of this individual;
and
WJn= The well-being of dependent member j in household n.
For simplicity, the household subscript, n, will be drop from the analytical
model. The well-being of the dependent member j, Wjn, can be defined further by tat
individuals’ component of consumption level and health status, or
(3.4)
Where Wj = The well-being of dependent member j;
Cj = Goods and services consumed by individual j that can be produced by
home production or purchased in the market; and
Hj = The health status of individual j, including physical, emotional, and
mental aspects.
Therefore, by substituting equation (3.4) in to the equation (3.3), the well
being of the working adult i in the household with dependents can be rewritten as,
(3.5)
By refining all of these components/aspects of well-being further, they are
functions of determinants of well-being, including personal income, educational
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 52
attainment, and work intensity. Each component of individual well-being could be
rewritten in the form of these determinants of well-being as follows:
Consumption ofgoods and services o f working adult member i, C,, is a
function of the personal income and the incidence of work intensity, or
C ,=c(y„k,) ;fi>0and^>0 (3.6) dyi dk,
Where: C, = Goods and services consumed by individual i that are produced by home
production or market purchased;
y = Personal income of individual i; and
£, = The incidence of high work intensity, measured by the length of time the
individual performs two production activities simultaneously.
Consumption and personal income are positively related since an increase in
income increases the ability of an individual to purchase more final goods and services,
g or production inputs for the household production process from the market. The
incidence of work intensity and individual consumption are related to each other in two
ways. First, work intensity indirectly affect the consumption of individuals through the
This might not be true in some cases for an altruist household member. An altruist’s consumption of goods and services might not increase at all when his/her income rises. This is due to the fact that the altruist might prefer to transfer all of his/her income to the dependent members. His/her well-being can be increased by transferring income to others.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 53
9 income effect. Second, by putting more effort (physical, mental, and time) into the
production process, the productivity tends to increase, leading to a rise in consumption.
Health status o f working adult i, //,, is related to personal income, the
incidence of work intensity, and the level of educational attainment, or
Hi =h(yi,ki,edui) (3.7)
® >0,^<0,^>0 (3.8) dyt dk, dedui
where Hi - The health status of working adult member i, including physical,
emotional and mental aspects;
y = The personal income of working adult /;
kt - The incidence of high work intensity for working adult /, measured by
length of time the individual performs two or more production activities
simultaneously; and
edu: = The Level of education attainment of working adult member /.
An individual starts with his/her own specific endowments of “healthiness”.
The health status of an individual is enhanced by consumption of food (the calorie
consumption or calorie intake) and better knowledge of food intake (associated with the
better education of an individual). It is harmed by an increase in the incidence of work
intensity. Therefore, income, which leads to an increase in consumption, is positively
9 This follows the same logic as the case of personal income and consumption. The altruist might work harder or work more intensively in order to allocate all of the extra goods and services produced to the dependent members in the household.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 54
related to an individual’s health status, while increased work intensity adversely affects
health. This is due to the higher possibility of the household member facing stress or
affected by health problems, due to the nature of overlapped work activities performed
under time constraints or other kinds of limitations. As a consequence, the first-order
partial derivative of health status with respect to the incidence of work intensity is
negative. The relationship between individual health status and the level of education
attainment was discussed earlier. Higher educated individuals have more knowledge
about nutrition intake, greater self-esteem, and a greater ability to make informed
decisions. This helps promote good health. Therefore, a positive relationship between
health and education is expected.
In this analytical model, the self-esteem and social status of working
individual i, Sj, is determined by the personal income and the level of education
attainment, or
Si = s(yi,edui) (3.9)
fr>0’oyl i dedui d ~>0 (110)
Where St = The self-esteem of working adult i, i.e., individual’s attitude toward
himself/herself, and the social status of individual i;
y,= The personal income of working adult i; and
edu: = The level of educational attainment of working individual i.
In sum, education and personal income are positively related to the self
esteem and social status of the individual. Individuals with high levels of educational
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 55
attainment tend to make more informed decisions and to develop better attitudes toward
themselves, enhancing their self-esteem. Consequently, the first-order derivative of self
esteem and social status with respect to personal income and education is greater than
zero.
The last two well-being components of working individual i are the
consumption and health status o f dependent member j. The consumption of dependent
member, CV, is assumed to be a function of personal income and the incidence of work
intensity of working individual i. In other words, a change in either income or work
intensity of the independent member is associated not only with his/her well-being level,
but also with the dependent member’s consumption and well-being. The health status of
dependent member, H j, is also a function of the individual income and the level of
educational attainment of individual i. Since the dependent member cannot make
decisions in this model, his/her health status depends on the decisions made by working
member i. These decisions in turn rely on individual Vs level of education and income.
Each component can be written as
dC, SC, C =C .(>>,,&,); where—->0 a n d —->0 (3.11) dyi dk,
and
dH, dH Hj =Hj (yi,edui)-, where — ->0 a n d — > 0 (3.12) dy, dedui
The consumption and health status of dependent member j, therefore,
depends on the characteristics of the altruistic member i. As stated in Equation (3.11), an
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 56
increase in the altruist’s personal income and incidence of work intensity would raise the
consumption level of the dependent member. For instance, a mother of dependent
children would push herself to work more intensively in order to produce more goods and
services so that her children’s level of consumption could increase. According to
Equation (3.12), an increase in working adult /’s personal income and level of
educational attainment is positively related to the health status of dependent member /. As
in the case of consumption, when the altruist’s personal income is higher, he/she can
provide better nutritional intake for the dependent members. Also, with a higher
education, the working adult would be more knowledgeable in caring for the dependent
member’s health both physical and emotional.
The incidence of work intensity function requires more discussion, given its
dimensional effect on individual well-being.10 The incidence of work intensity, k, is
related to the length of time spent in joint work activities.
(3.13)
(3.14)
(3.15)
where
kt- The incidence of high work intensity for working adult i, measured by
The incidence of work intensity is treated as a well-being determinant, not a constituent.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 57
the length of time the individual performs two or more production
activities simultaneously;
x = A scalar that denotes a composite index of the pertinent characteristics or
the nature of the activities combined, e.g., the amount of attention or
energy required, with higher work intensity associated to work where
both the main and secondary work activities require constant attention;
4 , = The time spent in market paid work in combination with another
activity by working individual i; and
4 = The time spent in unpaid household work in combination with another
activity by working individual i.
An increase in time spent by an individual on overlapping work activities
positively affects the incidence of work intensity as indicated in the first-order partial
derivative. Moreover, the longer the time spent in performing two tasks simultaneously,
the more an intensification of work time occurs. The second-order partial derivatives are
then positive, indicating that the incidence of work intensity progresses at an increasing
rate. A more intuitive explanation of how overlapping activities affects work intensity is
already discussed in chapter two. The working person is driven to work more intensively
to reach a subsistence or targeted well-being level. In the case of our survey, working
individuals worked hard in order to send their children to college, in order to improve the
children’s in the long term.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 58
3.2.2 Constraints of Well-Being
In our analytical framework, a working individual faces two main constraints.
Those are a time constraint and a budget/income constraint.
Time constraint : The total time available each day to working adult member i,
denoted Tt, can be spent on only two dimensions, work, Lwi, or leisure, Lsi, as presented
in Equation (3.16).
T,=LWI+LS (3.16)
However, since overlapping activities or joint production are allowed in this
conceptual model, total time spent by an individual each day in equation (3.16) can be
modified to reflect this change, as follows:
^=(i«+4+i«,+4)+i, (3-i7)
where
Tt- Total available time each day to working individual /;
Lmi - Time singularly spent in paid market work by working individual i\
t mi = Time spent in paid market work while in combination with another
activity by working individual i;
Lhi = Time spent only in unpaid household production/work by working
individual /;
L*hi - Time spent in unpaid household work overlapped with another activity
by working individual i; and
Lsl - Time spent in leisure activities by working individual /.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 59
Hence, the amount of work performed by the individual is not limited by the
length of working day. The individual can decide not only how to allocate his/her time to
each activity, but also whether to perform that activity singularly or in combination with
other activities (Floro 1995a). An individuals’ work time, Lwi, can be decomposed into
four aspects of time use as follows:
1) Time spent in a single paid market work activity, Lm;
2) Time spent in a single unpaid household work activity, Lh;
3) Time spent in paid market work in combination with another
activity, 4 ; and
4) Time spent in unpaid household production/work in combination with
another activity, 4 (Floro 1995a; Juster and Stafford 1985; and Floro and
Hungerford 2001).
Budget/income constraint on purchasing commodities and services: This
model assumes that an individual has two potential income sources: monetary-based and
asset-based. The individual’s full-income constraint can be written in the form of wage or
non-wage labor earnings per unit of market work time, wi, and/or an exogenous asset-
based income, 4 >as presented in Equation (3.18).
44 =Yj = Ai +wi [Lmi + 4 , ) (3.18)
where
Pm = The market-determined goods and services price vector of the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 60
individual’s consumption goods and services;
C, = The goods and services consumed by working individual /; and
yt = The personal income of working individual /.
Combining the two constraints and substituting (Tt - Lhj - t hi - Lsj) , in
Equation (3.18), we get:
P.C, =Yt = Ai + wt{T -L hi - 4 - 4 ) (3.19)
3.2.3 Well-Being Maximization Process
By combining Equations (3.5) and (3.19), the maximization problem (the
well-being process that is subject to the budget/income and time constraints) faced by
working individual i can be written as
Max W, =W (C], Ht, 5,, C}, Hs) Subject to (3.20) V ;=^ = 4 +W/(7;-4-z;(-4)
We now substitute the determinants of well-being from Equations (3.6), (3.7),
(3.9), (3.11), and (3.12) in to maximization process in Equation (3.20), so that the
individual welfare maximization problem becomes:
Max Wt = w [c(yl,ki),h{y„kl,edui),s(y„edul),cJ (y ,,^ ),^ O '/»«**<)] Subject to (3.21) 4c,=i; = 4+w,(7;-4-z;-4)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 61
The first-order partial derivatives of this individual welfare maximization
process with respect to the well-being determinants are as follow:11
(3.22) dyt ’ dedui ’ dkt
Note that the effect of the incidence of work intensity on individual well-
being is ambiguous. It depends on whether the positive income effect is greater or smaller
than the negative health effect. The incidence of work intensity, therefore, may enhance
personal overall well-being when the positive income effect outweighs the adverse health
effect. As previously discussed, however, the home-based workers in our survey are
expected to be below or at the poverty line threshold. It is likely that an increase in
produced output corresponds with a deterioration of output quality. The income tends to
be only slightly affected by work intensity factor. We can assume that a change in work
intensity affects the well-being of an individual more greatly through the health status
than through the income/consumption aspect. As a result, the sign of the first-order
partial derivative of individual welfare with respect to the incidence of work intensity is
given to be negative.
This dissertation will not try to solve the maximization problem so that the
necessarily condition for optimality will be reached. The above conceptual framework,
however, give us an idea about how personal income, the level of educational attainment,
and time allocation affect the well-being of a working individual. It is also worth
mentioning that this well-being maximization problem violates two main assumptions of
the theory of household production. These are that a household has a “constant return to
The mathematical proof of this result is presented in Appendix A.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 62
scale in household technology” and that there is “an absence of joint production”. As a
result, the commodity demand function derived from this optimization problem no longer
12 exhibits all the properties of the demand functions of traditional demand theory.
Moreover, the Slutsky equation derived from this optimization problem no longer gives a
meaningful result, since the price of commodities is no longer related to commodity
demand. This point is also raised by Poliak and Wachter (1975) in their derivation of
joint production functions in which certain activities yield not only services or goods
13 output, but also the process itself provides direct satisfaction or utility to the individual.
Figure 3.4 shows the relationship among the different determinants of well
being and the level of individual well-being discussed in this analytical framework.
Basically, the well-being of the independent working household member depends on
his/her level of constituents of well-being, C,, Hl ,S,, as well as on the dependent
member’s constituents, Cj,Hj . The well-being constituents are determined by the well
being determinant, specified in this model as yi,ki,edu i. Personal income and the level of
educational attainment are positively associated with the well-being constituents and also
the level of well-being. The effect of changes in work intensity can be either positive or
negative on an individual’s well-being depending on the size of the income and health
effects. However, it is assumed to adversely affect individual well-being in this analysis.
Lastly, this analytical framework is under both time and budget constraints. The next
12 Specifically, commodity prices do not to explain commodity output. 13 They also suggested an alternative solution as using market “goods” price instead of using commodity prices in commodities demand function. As a consequence, they concluded that the commodities demand is a function of goods price, fixed wage rate, and non-labor income, and the time associated to activity r is a function of goods price, fixed wage rate, and non-labor income.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 63
chapter presents the construction methodology of the individual well-being index based
on this analytical framework as well as the methodology needed in time use data
collection.
Figure 3.4: The Individual Well-Being Model under Time and Budget Constraints
Consumption of dependent j Health Endowment Health Status H, = h(y.,k.,edu.)
Health Status of dependent j H. = hj ( y i,edui)
Health Endowment Individual Well-being function
c i (T / ’ > h i O', > K , edu t), W ; ~ W s, ( y ,, edu t) , Cj (y, ,k ), hj (y,, edu,)
Self Esteem and Social Status Consumption of goods and services S, =s(yi,edui) C =c(y,A,)
-► Positive Effect
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 64
CHAPTER 4
METHODOLOGY OF CONSTRUCTING AN INDIVIDUAL WELL-BEING INDEX AND COLLECTING TIME USE DATA
4.1 CONSTRUCTION OF AN INDIVIDUAL WELL-BEING INDEX
This chapter discusses the methodology of developing an individual well
being index as well as a methodology for time use data collection. In analytical
framework, three determinants of well-being were discussed, i.e., personal income, the
incidence of work intensity, and educational attainment. To appropriately link those well
being determinants with the data collected from our home-based workers survey, we
must assign a suitable proxy for each of the determinants. Monthly personal income is
used as a proxy for our personal income determinant. Monthly individual income refers
to gross regular income measured in Thai baht from all sources, including informal wages
and salaries, business income, profits, and other benefits. Intra-family transfers such as
housekeeping or personal allowances are not included. The inverse incidence of work
intensity is measured by the length of the working day (in minutes), and time spent on
multitasking (in minutes) each day.1 The level of individual education will be the proxy
for educational attainment. Educational levels include primary, secondary, vocational,
certificate of study, and university. Each of these components will be discussed below.
1 Multitasking here refers to the overlapping of work activities when at least one of the activities is related to unpleasant work. The categorization of overlapped work/time intensity and some underlying assumptions is discussed later in this chapter.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 65
Based on the methodology used in calculating the Human Development
Index (HDI) (UNDP 2001,2002,2003), our individual well-being composite index is
developed by taking a simple average of the three individual well-being component
2 indices. There are the personal income index, the inverse work intensity index, and the
level of education attainment index. By taking a simple average, these three component
indices are being given equal weight. There is no a priori assumption that this is not the
case. Furthermore, the individual subjective well-being index, based on the respondents’
responses, will be developed for comparison and testing purposes.
Most of the quality of life indices based on the socio economic components
are strongly cardinal, i.e., they are scale invariant. For example, there is no difference
between two individuals, one earning $5,000 and the other $4,000, and two other
individuals, one earning $4,000 and the other $3,000. An individual with an inverse
work intensity index of 9 is also considered strictly better than one with an inverse work
intensity index of 7. If the cardinality assumption is taken for those indices, then the
satisfaction difference between an inverse work intensity index of 9 and one of 7 is the
same as the satisfaction difference between 6 and 4. This is not the case with an ordinal
measure. In this dissertation, the incidence of the work intensity index and the level of
educational attainment index are strictly ordinal. Since the composite well-being index
combines both cardinal and ordinal indices together, in order to properly rank the well
being of each individual and yield its normative significance, we adopt the “Borda Rule”
These three indicators are calculated based on the determinants of well-being. Moreover, since these indices are based on the determinants in the analytical framework in chapter three, the causality and the relationship among these variables are as stated earlier.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 66
used in other studies (Dasgupta 1999,2001; UNDP 2002).3 This dissertation will first
assume the cardinality assumption for well-being index calculation. Then, the cardinality
assumption will be relaxed and the ordinal aggregate approach will be used so that a
more accurate well-being ranking system is achieved.
The first step in developing the individual well-being index is the calculation
of each of the well-being component indices. These include personal income index (y),
the inverse work intensity index ( k), and the level of educational attainment index (edu).
The methodology used in this case is derived from the construction of the Human
Development Index. Let the quality of life of an individual be assessed on the basis of M
attributes (indexed by /), and there are N individuals in the economy (indexed by j). Let
X tj be the index of attribute i for person j. The index of attribute i for individual j, ItJ is
then calculated as
(4.1) (max; {x,}-miny{^})
4.1.1 The Level of Educational Attainment Component Index
Higher educational attainment not only enhances the personal income
(through the income effect of education), but it also bestows other non-monetary base
benefits to individuals. These can include greater self-esteem, a greater ability to make
informed decisions, and better nutrition and diet. The level of educational attainment for
3 To illustrate how the Borda Rule works, in our case, the criteria of ranking are our main component indices, which are time/work intensity, wealth, health, educational attainment, and SWB. Suppose an individual has the ranks of a, b, c, d, and e, for the five component indices, respectively. Then, his Borda score is (a+b+c+d+e). This rule will give us a complete ordering of alternatives.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 67
individual j ranges from 0 (no formal education) to 16 (undergraduate level of education).
According to the formula of the attribute index calculation stated above, the level
educational attainment component index for individual j, eduj , is developed as,
{ ^ e d u j ~ m^nj { ^ e d u j } ) edUj = ■> ;------— £—pr (4.2) (maxy \Xedu] | - min^ \X eduJ j)
where eduf = The level of education attainment component index for individual j;
X edu j = The current level of educational attainment of individual j;
miny \X e(tu = The minimum values for the level of educational attainment
for a given individual, indexed j, within the entire sample;
and
maxy {Xeduj} = The maximum values for the level of education attainment
for a given individual, indexed j, within the entire sample.
The educational attainment component index ranges from zero to one, with
the higher index value indicating a higher level of educational attainment. An
individual’s well-being also increases as the value of educational attainment attribute
index moves toward a value of one.
4.1.2 The Personal Income Component Index
Income is a means to attain human development, not an end by itself. The
UNDP (2001, 341), therefore, treats the income component of the HDI as “a surrogate for
all dimensions of human development not reflected in a long and healthy life and in
knowledge - in a nutshell, it is a proxy for a decent standard of living.” There is also the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 68
notion that one does not need unlimited income for achieving a respectable level of
human development (UNDP 2001,2003). Given how income is transformed, even with
little income, a lot can be achieved in human development. To reflect this, the income
needs to be adjusted as it enters the HDI. The logarithm of income is, therefore, used in
calculating the HDI. In other words, the personal income component index of individual
j is calculated as income increases, its value is adjusted downwards through mathematical
treatment. It is also assumed that the past month’s personal income reflects the
normalized income earnings of the individual, taking implicitly into account the
variations in earnings over a specified period of time. In this study, the calculation of the
personal income component index is in the following form.
(4.3) (log {max, {XJ7}} -logjm in, {X„}})
where log{XyJ j = The discounted current personal income of individual j;
logjmin, = The minimum discounted values for the level of personal
income for a given individual, indexed j, within the entire
sample; and
log{ max,{X y, j j = The maximum discounted value for the level of personal
income for a given individual, indexed j, within the entire
sample.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 69
Values of the individual personal income component index also range from
zero to one. The higher the value of the personal income component index, the higher is
the level of individual well-being.
4.13 The Inverse Incidence of Work Intensity Component Index
The calculation of the inverse incidence of work intensity index requires
several steps. First, we need to properly classify the different overlapping work activities.
Second, this component index need to take into consideration both the time spent on
overlapped activities and the length of the working day. Our overlapping work activity
classification is based on that developed by Floro and Hungerford (2001). Floro and
Hungerford (2001) divided overlapping activities into four categories. These were
pleasant work, unpleasant work, pleasant non-work, and unpleasant non-work. Work and
non-work activities are classified based on whether or not that specific activity can be
hired somebody to do it as an option of doing yourself (Floro and Hungerford 2001). For
instance, paid market work is counted as a work activity since one can pay someone to do
it. Another example is childcare activity since if we do not want to do it, we can hire
somebody to do it. On the other hand, non-work activity is the activity that no one can
take your place - one needs to do that activity by oneself. For example, watching
television (leisure activity) is categorized as non-work since it is an activity that we
cannot hire someone to do it for us. Sleep is also an example of non-work activity since
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 70
we cannot pay someone to sleep for us. The most common combinations, resulting from
4 this classification process of overlapping activities are presented in Table 4.1.
Table 4.1: Classification of the Most Common Overlapping Activities
Primary Activity Secondary Activity A Unpleasant work Unpleasant work B Unpleasant work Pleasant work C Pleasant non-work Unpleasant work D Pleasant work Unpleasant work E Unpleasant work Pleasant non-work F Pleasant work Pleasant non-work G Pleasant non-work Pleasant non-work Source: Floro and Hungerford (2001)
The overlapping combinations presented in table 4.1 refer to all forms of time
use intensity, not all of which equate to work intensity. Therefore, for purposes of this
study, only the combinations involving unpleasant work are utilized in measuring work
intensity. These include categories A (two unpleasant forms of work combined), B
(unpleasant work combined with pleasant work), C (pleasant non-work combined with
unpleasant work), and D (pleasant work combined with unpleasant work). While this
classification is based on the work by Floro and Hungerford (2001) using Australian time
use data, there are also subjective considerations in our classification process. In our
analysis, unpleasant work time include time spent on all activities associated with labor
market works, e.g., work at main job, travel and communication time associated with the
labor market works, time associated with cleaning and maintaining work tools and work
space. Both active and passive childcare activities such as physical care of children and
For more information on the classification process of overlapping activities, see Floro and Hungerford (2001, 12-15).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 71
playing with children are arbitrarily classified for purposes of this study as pleasant
works, although this may not be the case for certain individuals. The same is true for
many of the domestic work activities such as cooking, house repairing and gardening,
which are classified as pleasant work. Again, some individuals may not perceive them to
be such. Other domestic work activities such as washing dishes, laundry and ironing, and
house cleaning are classified as unpleasant work.
The next question is how to measure time spent engaged in overlapped work
(secondary) activities. Juster and Stafford (1991,482) suggest that “the primary and
secondary activities may be performed one at a time or sequentially rather than in
parallel.” They argue that what we observe as overlapped or secondary activities are
actually just sequential switches between the various tasks. Floro and Hungerford (2001,
12) also argue that “overlapping of activities may just be frequent switches between
activities and if the time grid were fine enough, the issue of secondary activities would
then effectively disappear.” However, according to some psychologists like Ruthruff,
Pashler and Klaassen (2001) and Meyer and Kieras (1997), these tasks actually could be
performed in a parallel fashion. This, however, might create bottlenecks, which decrease
the individual’s overall ability or the attention paid to overlapping or secondary
activities.5 In their view, the time spent on the overlapped activity happens
simultaneously with the primary activity, and it is counted as an extra hour (time) for the
individual. Based on the sequential argument, there is no extra hour gained from
performing overlapped activity. There is, however, under the assumption of parallel
5 This bottleneck is created due to some neural limitations (Ruthruff, Pashler and Klaassen 2001) or due to strategic postponement of the less important task (Meyer and Kieras 1997).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 72
activities. If an individual spends 90 minutes on paid market work as a primary activity
and 90 minutes on child care as a secondary task, should the total time be counted as 180
minutes (90 minutes of market paid activity plus 90 minutes of child care) or only 90
minutes of time engaged with overlapping activities (since there is no extra time gained
from performing overlapped activity)? One could argue for either of these options or a
third of taking the characteristic of bottlenecks into account giving the overlapped
activity a weight of 50 percent, the total time would be 135 minutes -90 minutes of
primary activity plus 45 minutes of overlapped activity.6 In this dissertation, we adopt the
assumption of parallel activities and measure the total time spent engaged in overlapping
activities by the total number of minutes of the two activities together.
To develop the inverse work intensity index, the amount of time spent on the
overlapped activity (extra hours) is used. For example, one hour is added to category B
classification when respondents spend an hour working on paid market work combined
with household chores. Then, time spent on overlapping activities from all different
categories of work intensity shown in Table 4.1 was added to get the total amount of time
spent on the overlapping activities that classified as work intensity, given an equal weight.
For example, let nt is the number of minutes spent in overlapping activities category /,
where i is the overlapping categories A, B, C, or D. In other words, the overlapped work
activity index measures the time spent by individual i in doing any one of the following -
A) unpleasant primary (paid or unpaid) work activities combined with unpleasant (paid or
unpaid) work activities; B) unpleasant primary (paid or unpaid) work activities combined
6 The overlapping activity has been given a weight of 50 percent to capture the bottleneck characteristics of performing two or more tasks simultaneously.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 73
with pleasant (paid or unpaid) work activities; and C) pleasant primary non-work
activities combined with unpleasant (paid or unpaid) work activities; and D) pleasant
primary (paid or unpaid) work combined with unpleasant (paid or unpaid) work activities
Then the total time spent of overlapping work activities is ]T n,.
We also take into account the length of the working day in calculating the
incidence of work intensity. The notion of the average working day has both social and
biological attributes, and is measured here in relative terms. The length of the paid market
working day in this study is bounded ultimately by the total number of hours in the day
that the individual spends on paid market work. Some studies suggest that human
physical, emotional, and intellectual capacities do not allow an endless extension of work
effort in a given day (Green 2002). The biological and physiological needs of the body
require some minimum renewal time such as sleep and personal care. Hence, the
incidence of work intensity rises when the length of the working day exceeds a
reasonable time, limited by those human capacities.
To be consistent with other component indices, the inverse work intensity
component index is computed as an inverse so that the value of the index gives the same
result patterns as other component indices. The higher the inverse work intensity index
7 value, the higher the level of individual well being will be. In other words, a high value
in the index indicates reduced or minimal intensification of the work day and corresponds
to a higher level of individual well-being. The logarithm form is utilized in our
calculation of the inverse work intensity index in order to capture the fact that the
7 The personal income component index and the level of education attainment index are represented in the way that a higher the value of the index, a better off the individual well-being is.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 74
incidence of work intensity increases at an increasing rate when individuals perform
extended periods of overlapped work activities and/or lengthen their work day. The
inverse work intensity component index is developed by taking an inverse of a simple
average of the length of the paid market working day sub-component and the work-
overlapped activity sub-component, with each given equal weight. Mathematically, the
inverse incidence of work intensity index for individual j can be calculated as shown in
Equation (4.4).
(log\X wdj} - log jminy (log {X0Vj} -logjm in, {Xov,}})
M max, min, I*-.)}) M max, -logjmin, {X0Vj}}) (4.4)
where Xwf/J = The length of paid market working day (in minutes) of individual j;
X ov j = The length of the overlapped (paid or unpaid) work activity (in
minutes) of individual j;
min, {Xwd j | = The minimum values for the length of paid market working
day for a given individual, indexed j, within the entire
sample;
miny [Xov /1 = The minimum values for the overlapped (paid or unpaid)
work activity performed, for a given individual, indexed j,
within the entire sample;
max; [Xwd J | = The maximum values for the length of paid market working
day for a given individual, indexed j, within the entire
sample; and
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 75
maxy {Xov j | = The maximum values for the overlapped (paid or unpaid)
work activity performed, for a given individual, indexed j,
within the entire sample.
The inverse incidence of work intensity component index ranges from zero to
one. When the value of the index moves toward zero, this indicates a high level of work
intensity. A low level of work intensity is implied by an index value close to one. Hence,
a high value in the work intensity component index indicates corresponds to a high level
of individual well-being.
4.1.4 The Individual Well-being Composite Index
The final step of constructing the individual well-being composite index is to
take a simple average of the sums of each of the component/attribute indices, with each
given an equal weight. The computation process of the well-being composite index for
individual j, WBIj, is presented in Equation (4.5).
where 0 < WBI] < 1 (4.5) m
where WBI. = The well-being index for individual j;
ltJ - The component/attribute indices of the individual well-being index; and
m = The number of components/attributes of individual well-being.
The value of the individual well-being index also has a range of zero to one.
The higher the well-being index value, the better individual j is in terms of the m
attributes of well-being. These attributes are the personal income index, y; the level of
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 76
educational attainment index, edu; and the inverse work intensity index, k. Therefore, the
individual well-being index for individual j, WBIj, is calculated as,
WBIj = \[y j +kj+ eduj ) (4.6)
Substituting the values obtained from Equations (4.2), (4.3), and (4.4) into
equation (4.6), we derive the following individual well-being index (WBI)
(!°g {X yJ} - log { min y{ X yj}} ) (logjmaxj - log{min,
(log[Xwd J } - log jminy \Xwd]}})
(lQg{ maxy min/ W B I ^ \ +1 — (log {X qvj } - iog |miny } | ) (l°g{ maxy miny
(maxy -miny {-W )
The maximum and minimum values for each well-being component required
in our well-being index computation method are calculated based on the 2002 Bangkok
urban poor home-based workers survey. All of the minimum and maximum values
associated with well-being attributes are presented in Table 4.2. The poorest individuals
from our survey earn only 1,000 baht, while the richest individuals make 60,000 baht per
month. These numbers are used as the lowest and highest values in order to calculate the
personal income attribute index. Similarly, for educational attainment, 0 implies an
uneducated individual, while 16 refers to a college graduate. The length of the work day
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 77
ranges from 1 minute to 920 minutes, while time spent in work related overlapped
activities ranges from 1 to 620 minutes per day.
The ability to compare well-being indices between individuals is based on the
premise that individuals can make interpersonal utility comparisons, given the possibility
of emotional connections that encourage empathy (England 2003). The latter point raises
the possibility of translating between one’s own and another person’s metric for well
being in which case, comparisons of well-being between individuals is a reasonable
exercise. This is contrary to the argument in neoclassical theory that interpersonal utility
comparisons are impossible.
Table 4.2: Minimum and Maximum Values of Well-Being Attributes
Minimum Value Maximum Value Time/work intensity attribute 1.) Work and unpleasant overlapping 1 620 activities (Minutes) 2.) Length of the working day 1 920 (Minutes) Personal income attribute Income (Baht)1 1,000 60,000 Educational attainment attribute The level of educational attainment 0 16 Subjective well-being SWB answer (score) 0 10 1 The current exchange rate is roughly 42 baht per dollar 2. Subjective well-being attribute is used to calculate the subjective well-being index in order to check the robustness of the individual well-being index.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 78
4.1.5 The Subjective Well-Being Indicator*
An individual well-being index based on observable (and measurable)
behavior is obviously not a perfect measure of individual well-being Focus on observable
behavior ignores the fact that a person’s behavior or well-being is driven not only by
higher levels of income, other socio economic variables, or time use, but it is also driven
by feelings of revenge and jealousy, imitation of others, social norms and institutions, and
legal prohibitions (Gibbard 1996; Pradhan and Ravallion 2000; Ferrer-i Carbonell 2002b;
and Van Praag; Frijters; and Ferrer-i Carbonell 2002). These components of well-being
9 will not be revealed by using only revealed preference theory. The subjective well-being
indicator attempts to capture these unobservable determinants of well-being since
individuals will be able to reveal their general level of life satisfaction without feeling the
need to hide it.
As previously discussed, one of the restrictions of the subjective well-being
question is that individual responses are assumed to be mutually comparable.10 In other
words, the subjective well-being question relies heavily on the premise that individuals
understand and respond to subjective questions in similar ways (Ferrer-i Carbonell 2002b
and Van Praag; Frijters; and Ferrer-i Carbonell 2002). This is a reasonable assumption as
long as there is common base used by individual respondents, allowing them to judge
their own well-being in a similar manner. Studies have found evidence that within the
8 Note that this separated index is constructed to test and compare with the individual well-being composite index. 9 The generalized axiom of revealed preference stated that if x‘ is revealed preferred to xs, then xs cannot be strictly directly revealed preferred to x*. For more information on revealed preference theory see Varian (1992) chapter 8. 10 Respondents are also assumed to be able to clearly evaluate their own present situation.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 79
same language community, individuals have a similar understanding of concepts such as
welfare, well-being, and happiness (Van Praag 1991; Ferrer-i Carbonell 2002a, 2002b).
Our sample of Bangkok home-based workers was collected from the same geographical
area and common language community. To develop the subjective well-being indicator,
we use the respondents’ responses to the subjective general life satisfaction question,
patterned after, the German Socio Economic Panel (GSOEP) survey. The SWB question
asked follows:
We would like to ask you about your satisfaction with your life in general.
Please answer by using the following scale. 0 means totally unhappy and 10 means
totally happy. How happy are you at present with your life as a whole?
The subjective well-being indicator for individual j, swbj , is developed based
on the responses from the above question as,
(4.8)
Where swbj = The subjective well-being indicator of individual j;
X swbj ~ The level of subjective perception of life satisfaction in general of
individual j;
mmin7 {XwAJj = The minimum values for the level of subjective perception of
life satisfaction in general for a given individual, indexed /',
within the entire sample; and
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 80
max, J = The maximum value for the level of subjective perception of
life satisfaction in general for a given individual, indexed j,
within the entire sample.
This indicator based on the subjective well-being score (given by the
respondents) ranges from zero fro totally unhappy to one for the highest level of
satisfaction.
4.2. TIME USE DATA COLLECTION METHODOLOGY
This section discusses the methodology used in collecting the time use data
among home-based workers living in low-earning communities in Bangkok, Thailand.
The three time allocation data collection techniques most commonly used are the
recall/estimated method, the time diary method, and a direct observation method. These
three conventional methods unfortunately were not appropriate for our survey
respondents: as all three are based on a western concept of clock-time. This is not the
case with our respondents where some of them still wake up in the morning when the sun
rises rather than using alarm clocks. Moreover, their work in the informal economy is
especially fluid where they shift from one activity to another without being bound by a
rigid schedule. Juster and Stafford (1991) recognize that a valid measurement of time use
data is both difficult and costly to obtain, especially in the informal sector. This suggests
that an appropriate methodology to obtain reasonable quality time use data requires
modification of existing methods in order to minimize measurement error, time, and cost.
This study makes use of two methods that seems appropriate and well-suited to the
characteristics of our survey respondents. They take into account the following:
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 81
1.) The concept of time: Home-based worker respondents’ concept of time
differs from that typically used in formal labor markets and in
developed countries. Our survey respondents base their days around
routine activities rather than on a watch or clock. For example, when
asked “what time do you start working?”, a typical answer by formal
sector workers will give an exact time like 8.30 or 9.00 a.m. In contrast,
informal home-based workers in developing countries, commonly,
answer “after I drop my children at school,” “after I have finished my
breakfast,” or “after I’ve finished all of my morning domestic chores."
2.) Uneven literacy rates: It is common in developing counties that a
proportion of the population is illiterate. Due to the uneven in literacy
rate of survey respondents, certain time use collection techniques such
as diaries or recall might not work well. As a result, a new
methodology needed to be developed so that time allocation data from
illiterate, as well as literate, respondent would be similar without
needing to shift between two methods and creating some distortion in
11 the survey.
3.) Flexible work schedule: Home-based workers do not have rigid or
well-defined work patterns as many workers in the formal sector do.
They work whenever they have work to do, or whenever they get new
work from the contractor. Respondents might work every day for a
11 We might experience some distortion/bias if we utilize two different methodologies of time use data collection for the same survey.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 82
week, and then do nothing for a week or two before they get another
job contract.
4.) Flexible work patterns: Home-based women workers typically combine
their paid market work with unpaid household work. For instance, a
home-based grocery store owner might cook dinner while waiting to
serve customers if they appear. Hence, the time use data collection
method has to be designed so that it can capture both the main activity
and the secondary activity.
Therefore, a modified method was developed so that cost and efficiency were not
compromised. Basically, home-based workers tend to have busy lives. Time use
questionnaires, therefore, must be designed so that they are both easy to understand and
less time consuming to complete. There are different ways to accommodate these
concerns. One is to shorten and simplify time diary formats. Another is to use of a third
party who is trusted by the respondent (to avoid flaws in the direct observation approach,
12 discussed earlier). This dissertation modifies the traditional approach of time use data
collecting by, simplifying the time use diary format, and combining it with the recall
method. Direct observation by the respondent’s family or relatives is then also used
13 whenever respondents are either illiterate or too busy. These two approaches will be
explored in turn.
This approach will eliminate the time constraint of the respondents. This interesting approach is the result of the discussion with Maria Floro (my dissertation advisor).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 83
4.2.1 The Simplified Time Use Diary Approach
This methodology is an adaptation from the “tomorrow” approach, which
14 was used by the University of Michigan in the first US time use survey in 1965. A
shortened and simplified time use diary (to enter tomorrow’s activities) is given to the
home-based worker respondents on the first interview session. General information,
including employment and gender related data is also collected then. The interviewer also
explain the concept o f time, and related definitions such as which kinds of activity are
considered work, leisure, etc, to the survey respondents. They gave an instruction and
orientation to the respondents.15 The interviewers need to be sure that survey
respondents clearly understand all definitions and concepts to ensure consistency across
the sample. On the interviewers next visit (the day after “tomorrow”), a short recall
interview with the survey respondents is conducted to ensure that the time use diary has
been filled out correctly. ^ During this session, the interviewer can add any missing
information that was not given. This multi-visit approach also affords the interviewers an
opportunity to know the survey respondents better and to gain their confidence over time.
17 As a result, more accurate data will be obtained.
14 The interviewer conducted a brief “warm-up” interview on the first day and left the diary for the respondent to enter the next day’s activities. The interviewer returned to the respondent’s home on the subsequent day (that is, the day after “tomorrow”) to ensure that the diary had been filled out correctly and to fill in any missing parts if it had not. (Robinson, 1997, 72) The interviewer has been given the instructions and orientation on the data collecting methodology and all of the definitions. This explanation ensures that the survey respondents know about the time concept so that they can fill out the time diary correctly. Since the recall method works well only when the questions were asked with in 24 hours after the activities. 17 Some compensation was given to the survey respondents as an incentive and for their time spent on completing the questionnaires.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 84
These time diaries were designed free form with no pre-assigned time
18 periods. The survey respondents were asked to fill the diary in every hour. Since they
were so busy with their work, they were allowed to complete the diary every two or three
hours if necessary. This was easier for them since they did not need to fill in the diary
every time they encountered a new activity. The other advantage of the free form diary
was that the respondents could write down activities at any time period. Moreover, the
simplified time diary further reduced the refusal rate of survey respondents and
psychologically motivated the respondents to enter the data in the time diary.
This methodology, however, might result in a bias toward an overestimated
time allocation for each activity (e.g., rounding up the activity that took 13 minutes to 20
minutes). Nevertheless, with the tomorrow recall sessions, interviewers were able to
correct some of these problems. In sum, by combining the “tomorrow” approach with the
simplified time diary, an acceptable level of accurate data with lower survey costs and
refusal rates were made possible.
4.2.2 The Circle of Trust Approach
This approach modifies the direct observation method and addresses an
important issue related to its approach. The direct observation method, while proven to be
effective in time use data collection, raises the possibility of bias given its intrusive
19 character. This intrusive aspect limitation can be partially addressed by altering the
observer. Interviewers might be viewed as household intruders simply because the
18 In the 1992 Australian time use survey, the time use diary involved 5 minute-time slots. 19 The household intrusion aspect refers to the change in respondent behavior due to the fact that he/she feels that someone is watching them.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 85
respondents are not familiar with nor trust them. However, if the observers are persons in
the family or ones familiar with the survey respondents, then one may reduce the so-
called intrusion bias.
Consequently, instead of hiring an outside observer we asked respondent’s
family members such as sons, daughters, and other relatives who lived under the same
roof as the respondent to do a direct observation. The trusted observer can also utilize the
20 recall method to capture all of the respondent’s activities. For instance, the trusted
observer might keep observing the respondent all morning, but the trusted observer might
not be able to stay home and keep watching the respondent in the afternoon. The trusted
observer, then, can use a recall method with the respondent for the period that he/she is
not home. In this case, the respondent would be willing to give the trusted observer a
more accurate amount of time spent on each activity since the trusted observer is a family
member. This methodology helps to address the cost and accuracy issue of time use data
collection.
This dissertation used both approaches in order to collect time use data given the
unique respondent limitations. The simplified time diary approach is utilized as the
primary technique. The circle of trust approach is used as the secondary method when
respondents refused to keep time diaries for themselves (and/or they were illiterate) and
21 one of the household members volunteered to be the trusted observer.
20 The observers who are the family member will be called as “trusted observers”. 21 In several cases, the respondents were very busy with their work thereby hesitates to fill out the time diary. Some compensation was given to the trusted observer as an incentive.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 86
CHAPTER 5
EMPIRICAL ANALYSIS: THE CASE OF URBAN-POOR HOME-BASED WORKER IN THAILAND
5.1 BACKGROUND
This chapter presents the Bangkok, Thailand urban-poor, home-based worker
survey, and the pertinent characteristics of the households and individuals in the time
allocation sub-sample. Thailand is the only Southeast Asian country never occupied by a
foreign power, except in war. The country was an absolute monarchy from 1782 until
1932, when rebels seized power in a coup and established a constitutional monarchy.
Since then, Thailand has come under the rule of many governments, both civil and
military. Thailand, similar to other Asian countries, started the process of economic
development from a rural agricultural-based economy, and then gradually moved toward
an industrializing and service-based one. By the mid-1980s, many foreign investors
transferred their production plants to Thailand due to the availability of a large surplus of
cheap labor and it was at the threshold of being a newly industrializing country (NIC). Its
annual growth rate peaked at 13.3 percent in 1988, but had declined sharply to 8.5
percent when the Gulf War started in 1991 as shown in Figure 5.1.1 The economic boom
at that time was driven by an increase in exports due to the devaluation of Thai baht.
During that successful period, Thailand was described in the World Bank’s 1993 Miracle
Report as a model for economic development (Lauridsen 1998). However, this prosperity
1 This is based on the GDP growth rate (constant at 1988 price).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 87
has not done much to alleviate the poverty of the Thai people. The average growth rate of
Thailand (constant at 1988 price) from gulf war period until before the Asian Financial
Crisis was 8.04 percent (NESDB 2003c). Thailand entered its most severe economic
crisis since World War II in midsummer of 1997. As it so happened, the Thailand
financial crisis triggered the more serious financial crisis throughout Asia, which later
spread around the globe. The Asian financial crisis caused deterioration of the overall
economy of the country, and many medium and large industries went out of business.
The crisis also extensively affected people in both urban and rural areas. Not only
unskilled workers have been laid off, but also skilled labor, educated employees, and
many executives. The GDP growth rate hit its bottom at -10.5 in 1998, and the inflation
rate rose to 8.1 percent in the same year. Between August 1998 and August 1999, the
Thai government launched three fiscal stimulus packages, including tax and tariff
reductions, to boost domestic demand (see Figure 5.1). Consequently, it was running a
large deficit which, on a cash balance basis, reached a cumulative 79.4 billion baht in the
first half of fiscal year 1999/2000. Thailand's economy is beginning to recover from the
global economic slowdown in 2001. The country's real gross domestic product (GDP)
grew only 1.8 percent in 2001, and jumped to 4.9 percent in 2002. The country also ran a
merchandise trade surplus of 8.6 billion dollars in 2001 and a current-account surplus of
6.2 billion dollars (equivalent to 5.4% of GDP) in the same year. Thailand has a
diversified export base including agricultural commodities and manufactures. However,
80 percent of exports are now manufactured goods, and they are highly dependent on
imported inputs.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 88
Figure 5.1: Thailand’s Annual Growth Rate from 1975-2003*
k. a.
CO 00 O CM co CO o CM 00 0 0 0 5 0 5 0 5 CD 0 5 o O O) 0 5 0 5 0 5 0 5 0 5 o> O) o o OJ CM
-10
-15 Year
* This is annual GDP growth rate (constant at 1988 price) from Bank of Thailand.
Thailand is considered a "medium HDI" country in the Human Development
Report in the rank of seventy-forth in 2001 (dropping from seventieth in 2000) from a
2 total of 175 countries (UNDP 2003). Life expectancy at birth was 66.9 years in 2001,
which is roughly average for the East Asian and Pacific region. The adult literacy rate
(age 15 and above) is 95.3 percent (increasing from 92.4 percent in 1990), much higher
than the regional average of 85.3. By 2001, 61.8 million people lived in Thailand, and the
average growth rage (1997-2001) was 1 percent. Thailand ranked sixty-first in 2001 in
the Gender Related Development Index (UNDP 2003). The life expectancy at birth was
73.2 and 64.9 years, respectively for women and men. There are large income differences
2 The human development rank decline even though the Human Development Index value keeps increasing throughout the year. This implies a slower development than other countries in term of HDI.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 89
between men and women, with men earning $7,975 per year in 2001 compare to only
$4,875 for women.
Poverty and income inequality still persists in Thailand. With the onset of the
financial crisis in 1997, an additional 3 million people became poor (6.8 million were
already below the poverty line). The incidence of poverty climbed to 16 percent (head
count ratio) in 1999, as shown in Table 5.1. The national average of the official poverty
line was 878 baht ($20.90) per person per month in 1998 and 886 baht ($21.10) per
person per month in 1999 (NESDB 2003a). This means that the poverty line (converted
to a daily basis) was significantly lower than the minimum wage, constituting less than
one-fourth of the minimum wage in 1999. In current dollar terms, the official Thai
poverty line in 1999 was equal to approximately $0.75 per day. The official Bangkok
poverty line was roughly 12 percent higher than the national average in 2001, and the
head count ratio for the Bangkok metropolitan area was zero as reported by NESDB
(2003a). Incomes are distributed more unequally in Thailand than in most middle income
countries. The crisis has also caused the gaps between the rich and poor to widen. The
GINI index of income inequality increased from 51.1 to 53.3 between 1998 and 1999
(NESDB 2003a). The poorest 20 percent of the population received only 3.8 percent of
aggregate national income while the richest 20 percent received about 59.0 percent in
1999 (NESDB 2003a). Due to improvements in the economy since the crisis, inequality
is lessening, but it is still high, with the poorest 20 percent of the population receiving 6.1
percent of aggregate national income while the richest 20 percent received about 50.0
percent in 2000 (UNDP 2003).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 90
Table 5.1: Thailand Poverty and Income Disparity 1988-2002
Poverty Poverty line: Head line (baht Bangkok Count Poor per Metropolitan Ratio people** GINI 20% 20% Year month) area (Percent) (million) Index Poorest* Richest* 1988 473 506 10.4 17.9 48.5 12.5 54.2 1990 522 604 27.2 15.3 52.4 4.2 57.8 1992 600 666 23.2 13.5 53.6 3.9 59.0 1994 636 658 16.3 9.7 52.7 4.0 57.7 1996 737 774 11.4 6.8 51.5 4.2 56.7 1998 878 935 13.0 7.9 51.1 4.2 56.5 1999 886 942 15.9 9.9 53.3 3.8 58.5 2000 882 972 14.2 8.9 52.5 6.1 50.0
2001 916 1,027 12.96 8.2 ---
2002 922 1,021 9.79 6.22 51.11 - "
* This is the percentage of aggregate national income. ** This is number of people under the poverty line. Source: NESDB Poverty Analysis 2003a and UNDP 2003
The economic crisis of 1997-1998 created an economic decline and many
companies and factories went out of business. Many workers lost their jobs, and
unemployment rates increased dramatically from 1.6 percent in 1997 to 4.36 percent in
3 1998 (NESDB 2003b). The formal sector was hit hardest by the crisis. During the crisis,
many factories tried to reduce production costs through subcontracting work. The
4 informal sector absorbed many of these unemployed workers from the formal economy.
Real wages and household incomes drop significantly as a result of the crisis. This
situation forced other household members, especially women, to enter both the formal
However, the rate dropped to 3.23 percent in 2001. 4 Some unemployed workers were established their own-account work in the informal sector such as food vendor. However, due to the high interest rate, there was less loan capital available from formal financial institution and higher cost of borrowing from informal sources. More workers tend to enter the subcontracting arrangement.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 91
and informal labor market.5 In 1999, the Ministry of Labor and Social Welfare estimated
that at least 16 million Thai people were working in the informal economy. However,
according to the first National Statistical Office Homework Survey in February 1999,
more than 20 million workers worked in informal markets, accounting for two thirds of
the Thai labor force (of roughly 30 million people). Moreover, approximately three
hundred thousand of these informal workers are homeworkers (Table 5.2). It is obvious
that this number seems to be relatively small compared to the number of informal
workers (only 1.5 percent of the total informal work force). The National Statistic Office
(1999) did note that the number of wage-contracted workers depends on business
conditions in the economy. Since the survey was conducted during the economic crisis,
some employers reduced their production capacities, work was decreased or there was no
work. Interestingly, more than 80 percent of homeworkers -roughly 250,000- are female.
These female wage-contracted workers, with little or no skills, contend that poorly paid
and long hour work is better than no work at all. Table 5.2 also shows that only 55,000
wage-contracted workers were reported in the Bangkok Metropolis area. Unfortunately,
this latest homework survey does not represent the total number of Thailand’s home-
based workers since it was conducted on wage-contracted workers only, ignoring self-
employed workers.
Due to the limitation of formal market jobs, most of these women are more likely to enter the informal sector. 6 The National Statistic office’s definition of homeworkers refers to an individual at least 13 years old with the following main characteristics: 1.) The place of work can be anywhere other than the employer’s work place. 2.) The homeworker does not produce the goods for his/her sale but must return them to the employer. 3.) There is an agreement on pay between the homeworker and the employer. Either employer or homeworker, or both, can fix the pay rate. 4.) The work must be done as required by the employer.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 92
Table 5.2: Number of Households and Homeworkers Aged 13 Years and Over, by Sex, Area, and Region
Number of Homeworkers Region and Area Households Total Male Female Whole Country 226,473 311,790 62,250 249,540 Municipal Area3 32,150 64,297 23,813 40,484 Non-municipal Areab 194,323 247,493 38,437 209,056 Bangkok Metropolis 26,603 54,812 20,414 34,398 Central Region3 50,877 78,132 18,153 59,979 Municipal Area 2,840 5,381 2,021 3,360 Non-municipal Area 48,047 72,751 16,132 56,619 Northern Regiond 49,922 57,604 7,760 49,844 Municipal Area 707 911 351 560 Non-municipal Area 49,215 56,693 7,409 49,284 Northeastern Region3 72,910 89,778 11,954 77,824 Municipal Area 896 1,513 542 971 Non-municipal Area 72,014 88,265 11,412 76,853 Southern Region' 26,151 31,464 3,969 27,495 Municipal Area 1,104 1,680 485 1,195 Non-municipal Area 25,047 29,784 3,484 26,300
Source: The National Statistic Office Thailand Homework Survey (1999) Note: a) A political unit, such as a city, town, or village, incorporated for local self-government; b) A political unit without a local self-government; c) A wide alluvial plain whose fertile soils are replenished by the Chao Phraya River and other rivers flowing out of the northern mountains. This central plain has been described as one of the “rice bowls” of Asia because of its high agricultural productivity. d) This is the second largest region of Thailand, which lying north of the central plain. e) This region is a poorest and also largest region of Thailand, which lies to the west along the border with Myanmar, is also marked by north-south trending mountains. f) A long peninsula—part of the greater Malay Peninsula—makes up the south of the country. Although dominated by north-south mountains, this region is also noted for its coastal beaches and its many islands.
However, some non-government organizations, dealing with home-based
workers and the informal economy in Thailand, argue that this survey result
underestimates the total number of the wage-contracted workers since some of the sub
contracted workers do not reveal their status (Yimyam, Susanha et al. 2001). They are
afraid that by revealing their illegal home-based work status, the government might force
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 93
them to register with the authorities, and end up paying taxes and fees to the government.
Those non-government organizations estimate that at least 900,000 people (this number is
actually three times higher than the official estimate by the National Statistics Office) are
sub-contracted workers in the Thai informal sector (Yimyam, Susanha et al. 2001).
Home-based work in Thailand occurs in various production activities in both
urban and rural areas. These include cloth making, quilt making, silk and cotton weaving,
wood carving, apparel making, artificial flower making, gem cutting, and shoe making,
etc. In urban areas, especially in Bangkok, home-based work is extensively found in the
production of ready to wear clothing, shoes, artificial flowers, food, and so on. Thailand’s
sub-contracted workers commonly consider their sub-contract work as their primary work
source, not as part time work. Formal written contracts between homeworkers and
contractors or sub-contractors are rarely made. Normally, informal verbal contracts are
accepted by the wage-contracted workers. Therefore, the trust between the workers and
sub-contractors is very important. This kind of trust is established through longevity of
acquaintance either directly or indirectly through relatives or close friends (Lazo 1992).
Work contracts can also be inconsistent with breaks of up to two or three consecutive
weeks.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 94
5.2. THE CONDITION OF THE URBAN-POOR HOME-BASED WORKERS IN BANGKOK AND THE SAMPLE SELECTION SURVEY METHOD
The data used for this analysis were collected during June 2002 and September
7 2002 by American University researchers in cooperation with HomeNet, Thailand. The
overall objectives of the survey were (1) to better understand the gendered effects of
labor market informalization and the growth of precarious jobs, and (2) to identify the
roles of credit, saving, and time allocation patterns in the ability of informal sector
workers and their households to respond to the conditions created by economic structural
changes (due to financial crises and economic recessions). A random sample of 170
households was collected for this purpose; with only 110 households participating in time
allocation module. This was due to the nature of the time allocation questionnaire (time
diary method) which is more time consuming and complicated than the other survey
questionnaires. The respondents who participated in the time use survey were selected for
our sub-sample. This survey study focuses on home-based workers in the informal sector.
This sector is divided into two main categories, wage-contracted workers and self-
employed workers. Households with wage-contracted workers (also known as
homeworkers) are classified as households with at least one household member engaged
in wage-contracted work. Self-employed workers can be based at home (e.g., grocery
store and restaurant), or can work outside the home (e.g., street vender).
HomeNet, Thailand was established in order to assist home-based workers in obtaining technical assistance, securing occupation and job contracts, raising the quality of life, etc. HomeNet supports its 108 local membership groups in 17 provinces, mostly in the North and North-East region, as well as Bangkok. HomeNet was also one of the organizations who help develop the eighth Thailand national development plan, specifically for home-based workers.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 95
The questionnaire, reflecting the multi-purpose nature of the survey, contains
information on household income, housing, credit, savings, education, and time
allocation. The time allocation questions are recorded in minutes for the last 24 hours for
housework activities such as cooking, cleaning, and childcare, and for income-earning
activities. A community questionnaire was also administered to gather information on
housing and access to social services such as electricity, water, transportation, health
facilities, and schools. The survey questionnaire consisted of 7 sub-modules, which
include the general household module, home-based worker and enterprise module, time
allocation module, credit module, savings module, household decision making module,
and urbanization module.
To better represent the entire Bangkok metropolitan urban poor area, the
survey was done in three separate community sites, Udomsuk, Nomklao, and Nawamin
communities as shown in Figure 5.2. These community sites were chosen randomly
based on the following characteristics: community size, geographic location, community
type, etc. There are many home-based and other informal workers residing in these
communities who are still struggling to make a living. For example, more than 80 percent
of the Udomsuk residents are associated with home-based work. Out of a total of 170
households, fifty samples were collected from the Udomsuk site, and sixty samples each
were collected from the Nomklao and Nawamin sites. The short description of each
community site is presented in Appendix D. This survey is purposive that only
households with at least one adult household member (household head or spouse) in the
informal sector were selected. Given the list of households with these characteristic, the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 96
sample was selected randomly, with one in every six households chosen. This was done
based on the community map or community list of households.
Figure 5.2: The Location of Survey Community Sites
Map of Thailand Enlargement of Bangkok Map
■ iw iiw m ® 89 . Ha I MYANMARy'-' CSuaW - . o J u o * %•““* V- «y*A. i IA11 A M ) ' i«nfnangecm ;3^»r>a ,r *a,h *9 . *nS«rtS’ '"5 iduman'daman secSec BanuWIt*'8a * * CAMBODIA j 7 ,. a® -* / Kamporcg . fSipuStS Y C h u k o t ? Ljr Kota BaiwtaJ .^Terengganu Community Sites IWOMfStA MAtAY^A 1. Nawamin Site 2. Nomklao Site 3. Udomsuk Site 5.3 PERTINENT CHARACTERISTICS OF HOUSEHOLDS AND INDIVIDUALS IN THE TIME ALLOCATION SUBSAMPLE 5.3.1 Characteristics of the Survey Household Sample The time use sub-sample data set involves 110 adult respondents from the 9 Bangkok household sample. Table 5.3 presents the composition/structure of the For instance, if the first household was chosen, the next household sample would be the seventh household. In cases where the seventh household did not match our characteristic, then the following household was chosen instead. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 97 households and shows the frequency distribution of household income. Eighty-five percent of the households in the sample were either couples, couples with dependents and/or non-dependents. Roughly 70 percent of the households had at least one dependent member living in the household. Dependents were defined as people 15 or younger, unemployed, or with disabilities. Sixty-six percent of the households had children aged 14 or younger, with 30 percent having children aged 4 or younger. The presence of dependent members in the household might be one of the reasons adults entered the labor market as home-based workers. This could also affect their use of time with work and the need to mind their children simultaneously, resulting in higher work intensity. Households were generally small, with 3.81 members per household.10 This number is consistent with the average household size collected by the Thailand National Statistical Office in the Household Socio-Economic Survey of 2001, with an average of 3.6 persons per household. Household size ranged from one to nine persons per household, with 77 percent of households having of 4 members or less. One explanation for this small size is that the majority of the urban poor migrated from rural areas, leaving parents and other family members behind. In some cases, children are sent to live with the family members still in the rural area, so that the home-based workers would have more time to work. 9 These respondents were instructed to record their primary, secondary activities, where they were, who was with them, and for whom did they do the primary activities. 10 A household was defined as a collection of individuals that lived together under the same roof and regularly ate together. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 98 Table 5.3: Selected Characteristics of Subsample Household Household Type Total Percentage Couples only 4 3.6 Couples + dependents1 58 52.7 Couples + dependents + non-dependents2 12 10.9 Couples + non-dependents 19 17.3 Female headed3 2 1.9 Female headed + dependents 3 2.7 Female headed + non-dependents 8 7.3 Female headed + dependents + non-dependents 4 3,6 Total 110 100.0 Household with children 0-4 years old 33 30.00 Household with children 5-14 years old 40 36.36 Rest of the household 37 33.64 Total 110 100.00 Average household size 3.81 - Geographic Location (sub sample sites) Udomsuk (Thanin) community site 14 12.7 Nomklao community site 50 45.5 Nawamin (Samukkee Patana) community site 46 41.8 Total 110 100.00 Monthly Household Income4 (in baht) 0 - 5,000 1 0.91 5,001 - 8,000 11 10.00 8,001 -11,000 25 22.73 11,001 -13,000 28 25.45 13,001 -15,000 13 12.73 15,001 - 18,000 11 10.00 18,001 -21,000 4 3.64 21,001 -25,000 6 5.45 25,001 - 30,000 5 3.64 30,001 and more 6 5.45 Total 110 100.00 Average Household Income (baht) 14,499 N o te : 1. All children under 15 years, not in the labor force, and disabled household members are considered dependents. 2. Non-dependent household members are classified as the household members who are 15 years old and older and/or in the labor force. 3. This is the household in which the household head is female. The marital status of the household head can be either married, divorced, widowed, or single. 4. This refers to gross regular income measured in Thai baht from all sources, including informal wage and salaries, business, government pension, grants/transfer money from any organizations and other sources. These incomes are calculated on a monthly basis. The current exchange rate is 42 baht per 1 US dollar. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 99 We turn next to the characteristics of housing and living conditions reported by the respondents. About 4.5 percent of households live in rented apartments and 95.5 percent in separate building structures (houses). Apartments consisted of a few rooms rented in houses or small buildings. Houses and apartments tend to be very small. Most of the houses were self-constructed on land that the respondents did not own. More than 80 percent of all households lived in houses with two or fewer rooms. None had land ownership rights and only 52 percent of households paid rent to landlords. Electricity was available to all households. However, about 2 percent of households had no access to piped water. Reflecting the poor state of infrastructure, roughly 30 percent of all households had no access to a public sewerage system. The average household income of our sample was somewhat higher than in the latest national average income survey. Households averaged 14,499 Thai baht per month, or the equivalent of $345 at the current exchange rate. The average national household income from the Socio-Economic Survey 2001 was 12,185 baht per month (Thailand, National Statistical Office 2001).11 The average household income in the Bangkok metropolitan area in 2000 was 24,690 baht per month, or $588 (Thailand, National Statistical Office 2000). The average income per month of the respondents in our sub-sample was a little more than half of the average Bangkok household income. This finding confirms their position among the urban poor. This refers to the nominal current income measured in Thai baht from all sources, including wages and salaries, profits (farm and non-farm), properties income, current transfers, income in-kind, and other money receipts such as insurance proceeds, lottery winnings and other windfall receipts. Since the definition of household income used in both surveys are similar, the reason for this difference is that the national level survey includes households in the poorest rural areas. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 100 5.3.2 Characteristics of the Survey Respondents 5.3.2.1 Education and Migration Thai home-based work is generally dominated by females since women are more likely to accept lower remuneration in order to be able to work at home. Another possible reason for female domination in this sector is that home-based work is traditionally performed by females as a supplementary source of income (Lazo 1992). Therefore, not surprisingly, 84 percent of our respondents were female home-based workers. Table 5.4 compares educational attainment, age, and region of birth of male and female participants. The average age of male and female participants was 45 and 41, respectively. There is marked gender disparity in educational attainment. About 13 percent of females had no schooling, with only 17 percent continuing beyond primary school. In contrast, 95 percent of males had some education and 42 percent had primary schooling and higher. Moreover, women averaged only 4.7 years in school, while men spent 6.2 years in school on average. In total, around one-tenth of all respondents had no 12 schooling. Only 45 percent finished the forth grade, even though Thailand has a six- 13 year compulsory education requirement by law. From the latest Population and Housing Census by National Statistical Office (1990), around 10 percent of Thai people have never entered into the schooling system, which basically the same number as our data set. 13 „ Before the 1970’s, compulsory education was up to the forth grade only. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 101 Table 5.4: Selected Characteristics of Individual Respondents Female Male Total Age Number %1 Number %2 Number %3 15-24 1 1.1 1 5.6 2 1.8 25-34 25 27.2 2 11.1 27 24.5 35-44 29 31.5 6 33.3 35 31.8 45-54 31 33.7 3 16.7 34 30.9 55-64 3 3.3 5 27.8 8 7.3 65 and above 3 3.3 1 5.6 4 3.6 Total 92 100.0 18 100.0 110 100.0 Average age (years) 41.3 44.9 41.9 HighestEducationalAttainment Bachelor degree or higher 0 0 0 0 0 0 Certificate of diploma4 0 0 1 5.6 1 0.9 Secondary school5 1 1.09 1 5.6 2 1.8 Grade 9 15 16.3 5 27.8 20 18.3 Primary school6 21 22.9 3 16.7 24 21.8 Grade 4 43 46.8 7 38.9 50 45.4 Not attending school 12 13 1 5.6 13 11.8 Total 92 100 18 100 110 100 Average years of schooling 4.7 6.4 5.0 Region of birth7 Northeastern region 43 46.7 6 33.3 49 44.5 Central region 12 13.0 5 27.8 17 15.5 Bangkok 13 14.1 4 22.2 17 15.5 Northern region 21 22.8 2 11.1 23 20.9 Southern region 3 3.3 1 5.6 4 3.6 Total 92 100.0 18 100.0 110 100.0 Note: 1. Percentage of female participants. 2. Percentage of male participants. 3. Percentage of all participants. 4. This includes respondents who finished secondary school before getting the certificate or diploma. 5. This is equivalent to high school in the United States. 6. This is equivalent to elementary school in the United States. 7. This shows the original birth place of respondents before they migrate to Bangkok. Detail of Thailand geographical region are presented in Figure 5.3. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 102 Poverty and unemployment in rural agricultural sectors are the main reasons young females and males migrate from rural villages to search for jobs in urban areas (Praparpun, Boonmathya, Leechanavanichpan 1999). Only 15 percent of all participants were bom in Bangkok. The other 85 percent migrated from the countryside. Fifty-three percent of migrants came from the northeastern region, considered the largest and poorest part of Thailand. 14 The others came from rural areas in the southern, northern, and central regions. 15 Figure 5.3 presents the geographic location of each region. Migrants move to Bangkok in the hope of finding a better place to live and a better job. Another reason that young women migrate from their rural villages is to live in the Bangkok that is of fancy shopping malls and opportunities depicted through televised media, popular magazines, newspaper, commercial advertising billboards (Praparpun, Boonmathya, and Leechanavanichphan 1999). 14 The Northeastern region lies on a dry plateau with harsh climatic conditions which cause aridity and drought throughout the region, alternating with floods in the wet season. The Southern region is bordered to the east by the Gulf of Thailand and to the west by the Andaman Sea. It is famous for its beaches, rubber and coconut plantations. The Northern region is a mountainous region with its natural forest and deep valleys. Thailand’s second largest city, Chiang Mai, is located in this region. The Central region of lush and verdant valleys is dissected by the main river of the country, the Chao Phraya. The region is often known as the “Rice Bowl of Asia”, and is the richest, most extensive rice growing area in the country. The eastern region is also included in this part. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 103 Figure 5.3: Geographic Location of Thailand’s Region V** LASS ; naxrmAimux r, AMHCCHA Mi SJf.feM muow m, MAI All & A Source: Tourism Authority of Thailand Table 5.5 presents the level of educational attainment for male and female respondents (head and spouse) by age groups. As previously discussed, there are substantial disparities between male and female level of education. Men in our sample had higher levels of education than women for all age groups. While only 24 percent of women aged 25-34 years had more than a primary school education, 50 percent of men did. Only 20 percent of women aged 35-44 years went beyond primary school, while to 33 percent of men the same age did. Lastly, only 13 percent of females aged 45-54 had any education beyond primary school, compared to 33.3 percent of males. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 104 Table 5.5: School Achievement, by Age Group and Sex (Percentage) Age Group (in years) 15-24 25-34 35-44 45-54 55-64 65and+ All Females No schooling 0.0 8.0 17.2 12.9 0.0 33.3 13.0 4th grade 100.0 16.0 34.5 74.2 100.0 66.7 46.7 Primary school 0.0 52.0 27.6 0.0 0.0 0.0 22.8 9th grade 0.0 24.0 17.2 12.9 0.0 0.0 16.3 Secondary school 0.0 0.0 3.4 0.0 0.0 0.0 1.1 Some University 0.0 0.0 0.0 0.0 0.0 0.0 0.0 Total 100.0 100.0 100.0 100.0 100.0 100.0 100.0 No. of females 1 25 29 31 3 3 92 Males No schooling 0.0 0.0 0.0 0.0 0.0 100.0 5.6 4th grade 0.0 0.0 50.0 66.7 40.0 0.0 38.9 Primary school 100.0 50.0 16.7 0.0 0.0 0.0 16.7 9th grade 0.0 50.0 16.7 33.3 40.0 0.0 27.8 Secondary school 0.0 0.0 16.7 0.0 0.0 0.0 5.6 Some University 0.0 0.0 0.0 0.0 20.0 0.0 5.6 Total 100.0 100.0 100.0 100.0 100.0 100.0 100.0 No. of males 1 2 6 3 5 1 18 5.3.2.2. Employment and Earning While a number of the workers in our sample were employed with varied work arrangements in various industries, all of them worked in the informal economy. Workers were either self-employed or wage-contracted (casual or short-term worker).16 The types of occupations (industries) are presented in Table 5.6. Three-fourths or 75 percent o f the wage-contracted workers were in the textile/garment industry (e.g., Casual workers are workers who have an explicit or implicit contract of employment which is not expected to continue for more than a short period, whose duration is to be determined by national circumstances. Workers in short-term employment are workers who hold explicit or implicit contracts of employment which are expected to last longer than the period used to define "casual workers" (ILO 2001) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 105 assemble some parts of clothing and/or shoes). Most of the self-employed workers worked in the services industry (e.g., home-based grocery stores) and food industry (e.g., food vendors and home-based restaurants), or 43 percent and 40 percent respectively. Table 5.6: Type of Industry (Occupation), by Employment Type Wage- Contracted Self-Employed Total (N=110) (N=45) (N=65) Number % Number % Number % Industry Occupation Food Industry 4 8.89 26 40.00 30 27.27 Construction/industrial worker 1 2.22 4 6.15 5 4.55 Textile/garment Industry 34 75.56 2 3.08 36 32.73 Services Industry 4 8.89 28 43.08 32 29.09 Other non-food manufacturing 2 4.44 1 1.54 3 2.73 Others 0 0.00 4 6.15 4 3.64 Total 45 100.00 65 100.00 110 100.00 The difference in employment between men and women is striking as indicated in Table 5.7. Around 53 percent of female participants were self employed workers, compared to almost 90 percent of male respondents. Individual monthly earnings among self-employed also differed sharply by sex. Only 35 percent of female self-employed workers earned monthly incomes of 8,000 baht or higher, compared to 50 percent of male self-employed workers. Although there may be a number of reasons why women are less likely to be self-employed, one possible reason is gender differences in educational attainment; men in Thailand on average have more schooling than women. The average earnings per month of male and female contracted workers, however, were similar at 5,002.50 baht and 4,867.79 baht respectively (or $119 and $115 at the current Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 106 Table 5.7: Employment and Income, by Employment Type and Sex Contracted Worker1 Self-Employed2 Total Monthly Individual Income3 (Baht) Number4 %5 Number % Number % Female 0-2000 7 16.28 1 2.04 8 8.70 2001 - 4000 15 34.88 8 16.33 23 25.00 4001 -6000 12 27.91 19 38.78 31 33.70 6001 - 8000 4 9.30 4 8.16 8 8.70 8001 -10,000 2 4.65 12 24.49 14 15.22 10,001 - 12,000 0 0.00 3 6.12 3 3.25 12,000 or more 3 6.98 2 4.08 5 5.43 Total 43 100.00 49 100.00 92 100.00 Female's average income (Baht) 4,867.79 6,640.31 5,811.85 Male 0 - 2000 0 0.00 0 0.00 0 0.00 2001 - 4000 1 50.00 1 6.25 2 11.11 4001 - 6000 0 0.00 3 18.75 3 16.66 6001 - 8000 1 50.00 4 25.00 5 27.78 8001 -10,000 0 0.00 5 31.25 5 27.78 10,001 -12,000 0 0.00 1 6.25 1 5.56 12,000 or more 0 0.00 2 12.50 2 11.11 Total 2 100.00 16 100.00 18 100.00 Male's average income (Baht) 5,002.50 12,092.97 11,305.14 All individual (Male and Female) 0 - 2000 7 15.56 1 1.54 8 7.27 2001 -4000 16 35.56 9 13.85 25 22.73 4001 - 6000 12 26.66 22 33.85 34 30.92 6001 - 8000 5 11.11 8 12.31 13 11.82 8001 -10,000 2 4.44 17 26.15 19 17.27 10,001 - 12,000 0 0.00 4 6.15 4 3.63 12,000 or more 3 6.67 4 6.15 7 6.36 Total 45 100.00 65 100.00 110 100.00 Average Individual Income (baht) 4,873.77 7,982.50 6,710.75 Note: 1. This group refers to those who produce a product or provide a service to a contractor or an employer, and select their own work place. 2. This refers to the self-employed workers who work in their homes. They include: food venders, small grocery stores owner, own barber shops, own beauty salons, own bike repair shops, etc. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 107 3. This refers to gross regular income measured in Thai baht from all sources, including informal wage and salaries, business, benefit, investment and other sources. Intra-family transfers such as a housekeeping or personal allowance are not included. These incomes are calculated as a monthly basis. The official poverty line for Bangkok Metropolitan was 1,021 in 2001. 4. This refers to the number of participants in the given category. 5. This refers to the percentage of participants in the given category. exchange rate). Self-employed respondents, on average, earned 64 percent more each month than wage-contracted respondents did. Almost 80 percent of wage-contracted participants earned 6,000 baht ($143) per month or less. This figure is consistent with the average earnings of workers in a Praparpun, Boonmathya, and Leechanavanichphan (1999) study of home-based workers, where 84 percent earned 6,000 baht or less. Almost 16 percent of wage-contracted workers earned only 2,000 baht ($48) or less per month as seen in Table 5.7. Moreover, two female wage-contracted workers earned less than the Bangkok poverty line of 1,021 baht per month per individual. The average individual monthly income of all respondents was 6,710 baht (or equivalent to $160). This income is relatively low when considering the time that the respondents spent working. For example, one contracted worker worked roughly 69 hours per week (approximately 10 hours a day, seven days a week) and earned only 2,400 baht (around $60) that month. These contracted workers knew that they were exploited by their subcontractors, but they accepted it because poorly paid work was better than no work at all. Figure 5.4a presents the kernel density estimation of individual monthly income by sex. It clearly shows the pattern that female income is distributed at lower levels of earning, and that male monthly income density is higher at higher levels of income. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 108 Figure 5.4a: Kernel Density Estimation of Individual Monthly Income, by Sex Wamen Men .0002 .00015 .0001 .00005 0 5000 10000 15000 20000 Individual Monthly Income Figure 5.4b: Kernel Density Estimation of Individual Monthly Income, by Employment Status Wage-Contracted Worker ------Self-Employed worker .00015 .0001 .00005 5000 10000 15000 20000 Individual Monthly Income Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 109 Figure 5.4b shows the kernel density estimation of monthly individual income by employment type. The kernel density estimation of wage-contracted respondents is skewed toward the left, showing their lower earnings compared to self- employed respondents. Table 5.8 presents the daily income by employment type and sex of respondents. Daily income was calculated from the income per period reported by respondents. The periods reported include day, week, two weeks, and month. Based on information obtained from the survey, we also assume that the respondents worked 6 days a week. Around 46 percent of females earned 165 baht ($4) per day or less, which is 17 the minimum wage rate required by law. Only 11 percent of men earned this amount or less. Women on average also earned roughly only half of the daily income generated by men. Sixty-two percent of wage-contracted workers also earned less than the minimum daily wage, while only 26 percent of self-employed workers did. Moreover, self-employed workers on average generated 38 percent more daily income than wage- contracted workers. Low remuneration was not the only problem among sample respondents. Others problems identified by participants included lack of child care facilities, long average working days of 10-15 hours, being cheated by intermediaries and factories, lack of accounting skills, lack investment funds, no access to credit from formal financial institutions, no social security, and no healthcare or other benefits. 17 This minimum wage was set by the Thailand, Ministry of Labor and Social Welfare in 2002. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 110 Table 5.8: Daily Income, by Employment Type and Sex Contracted Worker Self-Employed Total Monthly Individual Income3 (baht) Number % Number % Number % Female 0 -85 13 30.23 1 2.04 14 15.22 86 -165 14 32.56 15 30.61 29 31.52 166-215 7 16.28 12 24.49 19 20.65 216-265 4 9.30 3 6.12 7 7.61 266 - 325 2 4.65 9 18.37 11 11.96 326 - 400 0 0.00 6 12.24 6 6.52 400 or more 3 6.98 3 6.12 6 6.52 Total 43 100 49 100 92 100 Female's average income (baht) 172.16 234.85 205.55 Male 0-85 0 0 0 0 0 0.00 86-165 1 50 1 6.25 2 11.11 166-215 0 0 3 18.75 3 16.67 216 - 265 1 50 3 18.75 5 27.78 266 - 325 0 0 6 37.5 5 27.78 326-400 0 0 1 6.25 1 5.56 400 or more 0 0 2 12.5 2 11.11 Total 2 100 16 100 18 100 Male's average income (baht) 176.92 427.69 399.83 All individual (Male and Female) 0-85 13 28.89 1 1.54 14 12.73 86 -165 15 33.33 16 24.62 31 28.18 166-215 7 15.56 15 23.08 22 20.00 216 - 265 5 11.11 6 9.23 11 10.00 266 - 325 2 4.44 15 23.08 17 15.45 326 - 400 0 0.00 7 10.77 7 6.36 400 or more 3 6.67 5 7.69 8 7.27 Total 45 100 65 100 110 100 Average Individual Income (baht) 172.37 282.32 237.34 * Daily income is calculated from the income per period reported by respondents. The period includes per day, per week, per two weeks, and per month. Base on the survey information, we assume that the respondents work 6 days a week. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Ill Table 5.9 presents a comparison of male’s and female’s number of hours in market work by employment type. In general, women worked fewer hours than men in both wage-contracted and self-employment types of work. The difference is small for self-employed workers, but much larger for wage-contracted workers. Female respondents, on average, spent about 10 percent less time than men in paid work, at 8.97 and 9.93 hours, respectively. More than 30 percent of wage-contracted workers spent 9 hours or more per day in paid work, compared to 60 percent of self-employed respondents. This result is consistent with the findings of other studies on home-based workers in countries such as Mexico, the Philippines, India, Sri Lanka, Brazil, Bangladesh, Pakistan, and Thailand (Bajaj 1999; ILO 2002a; Pongsapich 1992; Praparpun, Boonmathya, and Leechanavanichpan, 1999; Lavinas et al. 1999). These also showed that home-based workers work long hours, sometimes more than 12 hours a day. Women home-based workers in the football (and leather) industry in Pakistan work approximately 10 hours per day (Bajaj 1999). The long working hours of respondents can be seen clearly in Table 5.9. About 47 percent of all participants spent more than 9 hours per day on income generating activities. In contrast to formal market work, where 8 hours per day is standard, only 39 percent of all participants allocate 8 hours or less on paid market work. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 112 Table 5.9: Hour Work per Day (Primary Market Paid Work Activity Only), by Employment Type and Sex Contracted Worker1 Self-Employed2 Total Hours Work per Day3 (hour) Number %4 Number % Number % Female 0 -4 2 4.65 0 0.00 2 2.17 4 -7 12 27.91 10 20.41 22 23.91 7 -9 15 34.88 11 22.45 26 28.26 9-12 12 27.91 17 34.69 29 31.52 12 or more 2 4.65 11 22.45 13 14.13 Total 43 100.00 49 100.00 92 100.00 Female's average working hour 8.06 9.77 8.97 Male 0 -4 0 0.00 0 0.00 0 0.00 4 -7 0 0.00 0 0.00 0 0.00 7 -9 1 50.00 7 43.75 8 44.44 9-12 1 50.00 6 37.50 7 38.89 12 or more5 0 0.00 3 18.75 3 16.67 Total 2 100.00 16 100.00 18 100.00 Male's average working hour 9.50 9.98 9.93 All individual (Male and Female) 0 -4 2 4.44 0 0.00 2 1.82 4 -7 12 26.67 10 15.38 22 20.00 l "vl CD 16 35.56 18 27.69 34 30.91 9-12 13 28.89 23 35.38 36 32.73 12 or more 2 4.44 14 21.54 16 14.55 Total 45 100.00 65 100.00 110 100.00 Average Individual working hour 8.12 9.82 9.13 Note: 1. This group refers to those who produce a product or provide a service to a contractor or an employer, and select their own work place. 2. This refers to the self-employed workers who work in their homes. They include: food venders, small grocery stores owner, own barber shops, own beauty salons, own bike repair shops, etc. 3. This is an individual’s time per day spent on paid market work and other related work. 4. This refers to the percentage of participants in that category. 5. Most of the individuals who work more than 12 hours per day are grocery store owners. Time spent on paid work was counted immediately after they got up and opened their store front until they closed their store front. This might take up to 15 hours. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 113 Because the work day is long, average hourly earnings are low. On average, wage- contracted respondents earned only 20.88 baht per hour (49 cents), while self employed 18 respondents earned only 31.02 baht per hour (74 cents). Unlike the mean, which can be influenced by a few outliers, the median earning of wage-contracted workers and self- employed workers in the sub-sample was even lower at 15.43 (36 cents) and 23.68 baht per hour (56 cents), respectively. Roughly half of all survey participants earned less than 20 Baht per hour. Fifteen percent of all respondents worked more than 12 hours per day. Respondents who worked more than 12 hours per day were mostly self-employed, and tended to own a home-based grocery store. Those respondents recorded their primary activity as labor market work even as they mind their home-based stores without helping any customers. Minding the store often overlapped with other secondary activities such as watching TV, reading, cooking, cleaning, etc. Also, these respondents opened their home-based stores almost immediately after waking up, and closed just a few minutes/hours before going to bed. Due to this, the reported working hours per day of some self-employed workers in the sub-sample was beyond 12 hours. Figure 5.5 presents the kernel density estimation of hourly earning by men and women. The result is similarly to the monthly individual income kernel density estimation, where male hourly earning is distributed at the higher level while female income is distributed at the lower levels. This income per working hour is calculated from individual daily income divided by the individual working hour per day. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 114 Figure 5.5: Kernel Density Estimation of Individual Earning per Hour, by Sex Women Men .04 - .01 - 50 100 Individual Hourly Income 5.3.2.3 Allocation of Time to Market Work, Domestic Work, and Other Activities The analysis of time allocation begins with the average minutes per day spent by the home-based worker respondents, shown in Table 5.10, for all types of activities (market work, domestic work, leisure, and others). It shows that self-employed workers, on average, spent more time on labor market work than wage-contracted workers in both primary and secondary activities, or 589 minutes compared to 487 minutes per day respectively (primary activity only) and 93 minutes, compared to 34 minutes (secondary activity). Primary labor market work activities accounted for almost 41 percent of self- employed workers’ 24-hours time available per day, and 34 percent for the case of wage- contracted workers. In contrast to labor market work, wage-contract workers allocate more time to unpaid household work, which includes domestic work (food preparation, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 115 house cleaning, laundry, etc), childcare, and shopping than self-employed workers. This might be due to the fact that some of self-employed workers work in places outside the home such as the street (e.g., food vendors). As a result, time available for household 19 work (as primary and secondary activities) declines sharply. Wage-contracted and self- employed workers in the sub-sample seem to allocate roughly the same proportion of their total primary time to leisure activities, about 11 percent and 12 percent, respectively. However, wage-contracted workers spent 36 percent more time than self-employed workers on overlapped leisure activities. Table 5.10 also presents average time spent on all activities by respondents in the sub-sample. Average time spent by all respondents on market paid work as a primary activity is roughly 9 hours a day. However, by taking secondary labor market work activities into account, the time spent in the labor market rises by 13 percent to 11 hours on average. For instance, wage-contracted workers overlap time on subcontracted work with a primary activity such as watching their favorite television show. If overlapped labor market work activities were not taken into account, the overall time spent on labor market work would be underestimated. On average, all respondents spent almost 2 hours a day on domestic work (primary activity only) such as food preparation and cooking, dish washing, house cleaning, laundry and ironing, house maintenance, and so on. Respondents spent an average of 53 minutes on childcare tasks. However, this number doubles when secondary Food venders cook their food at home, but sell that food on street or road. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 116 Table 5.10: Average Time Allocation in All Activities, by Employment Type (Minute per Day) Contracted1 Self-Employed2 Total average3 Percentage Primary Activities Mean %4 Mean % (min. per day) Distribution Primarv Work Activities Labor Market Work5 487.44 33.85 589.23 40.92 547.59 38.03 Household work Domestic6 142.11 9.87 92.31 6.41 112.68 7.83 Childcare7 64.11 4.45 45.53 3.16 53.14 3.69 Shopping8 45.56 3.16 27.15 1.89 34.68 2.41 Sub-total 251.78 17.48 164.99 11.46 200.5 13.92 Primarv Non-Work Activities Leisure Activities9 154.55 10.73 159.38 11.07 157.41 10.93 Other Activities Personal care10 142.67 9.91 139.15 9.66 140.59 9.76 Sleeping 403.56 28.03 387.25 26.89 393.91 27.35 Sub-total 546.23 48.67 526.40 36.56 534.50 37.12 Total 1440 100.00 1440 100.00 1440 100.00 Overlapped Activities11 Overlapped Work Activities Labor Market Work 34.44 5.68 93.23 18.07 69.18 12.08 Household work Domestic 78.89 13.01 81.92 15.88 80.68 14.09 Childcare 109.27 18.02 73.31 14.21 88.02 15.37 Shopping 9.33 1.54 6.92 1.34 7.91 1.38 Sub-total 197.49 32.56 162.15 31.43 176.61 30.84 Overlapped Non-work Activities Leisure Activities 349.33 57.60 256.35 49.69 294.39 51.41 Other Activities Personal care 25.22 4.16 37.54 7.28 32.5 5.68 Sleeping 0 0.00 0 0.00 0 0.00 Sub-total 25.22 4.16 4.15842 0.81 32.5 5.68 Total 606.48 100.00515.888 100.00 572.68 100.00 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 117 Note: 1. This group refers to wage-contracted workers who produce a product or provide a service to contractors or employers. 2. This refers to the self-employed workers who use their home as a base for their business e.g., food venders, small grocery stores, barber shops, beauty saloons, bike repair shops, etc. 3. This is the mean time (simple average) spent on all economic activities by all participants in a 24-hour period. 4. This refers to the percentage of each activity of the total available time each day. 5. This is the time spent by the survey respondents, which is associated with paid market work. This also includes travel time associated with paid market work. 6. This includes food preparation and cooking, dish washing, laundry, ironing, clothes care, house cleaning, other housework, home maintenance, household management, transporting adult household members, and travel associated with any of the above activities. 7. This includes physical care and minding of own and other children, care for sick or disable child, teaching own and other children, playing with own and other children, and travel associated with children. 8. This includes purchasing goods and services, and travel associated with purchasing goods and services. 9. This includes mostly passive leisure such as reading, watching TV, listening to the radio, and communicating with others. In addition, active leisure is included in this category such as exercising. Social life and entertainment, including visiting friends/family/kin, are also included in this category. 10. This includes personal care such as bathing, dressing, and eating. 11. This includes all reported minutes in secondary and tertiary activities at equal weight as the primary activities. activity time on childcare is counted. It is noteworthy that respondents with household members, six years or younger, spent approximately 500 minutes a day on childcare activities (both primary and overlapped). Most childcare activities were reported as secondary activities; often times, overlapped with primary work activities. Respondents spent an average of 200 minutes on all unpaid work activities including domestic work, childcare and shopping. (If secondary activities were included, the total time spent on unpaid work doubles.) Respondents in our Thailand subsample, therefore, allocate 52 percent of their 24-hour time period to work activities, both paid and unpaid. In contrast, a sub-sample of Australians from the 1992 National Australian Time Use Survey spent only 35 percent of their available time on both paid and unpaid work activities (Floro and Miles 2003). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 118 An average of 157 minutes a day was spent on leisure activities, which included mostly passive leisure (e.g., reading, watching television, listening to the radio, and communicating with others.) by all respondents. Only 3 percent of all respondents participated in active leisure activities such as sport and exercise, the rest of the participants engaged in passive leisure only. Passive leisure accounted for 90 percent of time spent on overall leisure activities. This may be a reflection of the time intensity of respondents. Overall, participants spent less than 11 percent of all available time each day on leisure activities. This is in contrast to the sub-sample of 1992 National Australian Time Use Survey, where Australians spent almost double that -16.5% or 235 minutes- 20 on leisure activities (Floro and Miles 2003). Figure 5.6: Average Time Spending by Category on both Primary and Secondary Activity of All Survey Respondents 700 600 500 <0 I 400 ■ Secondary Activity in □ Primary Activity § 300 c i 200 100 0 B Labor Domestic Childcare Shopping Leisure Personal Sleepping Market Work Care Work Activities This is to illustrate that the home-based workers in the sub-sample are generally both time and money p o o r. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 119 Table 5.11 provides a comparison of men’s and women’s average time spent in various activities when secondary activities are included on all activities. As discussed in chapter 4, two alternative methods are used in accounting for secondary activities. In the first, primary and overlapped activities are given equal weight (assumption 1). In the second, overlapped work activities are given half the weight of the main activity 21 (assumption 2). Table 5.11 shows women engaged in labor market activities to a lesser extent than men. On average, women spent approximately half an hour less than men under the first assumption, and roughly forty minutes under the second assumption. Secondary work activities contribute an additional 28 percent of total work time under the first assumption, with the amount done by women (267 minutes per day on average, 35 percent of the total work time) more than double that done by men (138 minutes per day on average, 20 percent of the total work time). These differences in our sub-sample are striking, and indicate the inequality in the extent of overlapped work done by men and women. A particularly important question regarding the overall burden of work is whether or not women, who contribute about the same amount of time in labor market activities as men, has a similar household work burden as men. It is interesting to note that women spent more than double the time that men did performing household work (216 more minutes under the first assumption, or 166 more minutes under the second assumption). This implies gender inequality in the household division of labor. The second method accounts for the bottleneck problem which reduces the work efficiency of secondary tasks due to lack of concentration or attention. In some cases, efficiency of primary activities also affected by the bottleneck, e.g., by frequently switching tasks between primary and secondary tasks, ability to concentration on both activities declines. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 120 Table 5.11: Comparison of Varied Measures of Time Use, by Female and Male Home- Based Workers (Minutes per Day) Women Deflated Primary and Primary Overlapped Primary and Overlapped Deflated Activities only Only Overlapped1 Only Overlapped2 Labor Market Work3 538.15 74.13 612.28 37.07 575.22 Household work3 Domestic 122.93 89.68 212.61 44.84 167.77 Childcare 58.91 94.1 153.01 47.05 105.96 Shopping 37.88 8.91 46.79 4.46 42.34 Sub-total 219.73 192.68 412.41 96.35 316.07 Leisure Activities3 151.9 293.42 445.32 146.71 298.61 Other Activities3 Personal care 139.02 32.94 171.96 16.47 155.49 Sleeping 391.2 0 391.2 0.00 391.20 Sub-total 530.22 32.94 563.16 16.47 546.69 Total 1440 593.17 2033.17 296.59 1736.58 Change 593.17 296.58 Men Deflated Primary and Primary Overlapped Primary and Overlapped Deflated Activities only Only Overlapped Only Overlapped2 Labor Market Work3 595.83 43.89 639.72 21.945 617.78 Household work3 Domestic 60.28 34.72 95 17.36 77.64 Childcare 23.61 56.94 80.55 28.47 52.08 Shopping 18.33 2.78 21.11 1.39 19.72 Sub-total 102.22 94.44 196.66 47.22 149.44 Leisure Activities3 185.56 299.33 484.89 149.665 335.23 Other Activities3 Personal care 148.61 30.28 178.89 15.14 163.75 Sleeping 407.78 0 407.78 0 407.78 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 121 Sub-total 556.39 30.28 586.67 15.14 571.53 Total 1440 467.94 1907.94 233.97 1673.97 Change 467.94 233.97 Note: 1. This is the sum of time (in minutes) spent in each activity, whether primary or overlapped. Primary and overlapped activities are given equal weight. 2. In summing the total time spent in each activity, overlapped activities are given half (0.50) the weight of primary activities. This is based on die alternative assumption that individuals focus less energy and/or attention on those activities that are considered secondary and/or tertiary (overlapped). 3. See notes to this variable in Table 5.10 Women bear a much higher burden of the unpaid household work (domestic work, childcare, and shopping), reflecting Thai norms and culture that women are always in charge of household work, whether they are engaged in market activity or not. These double duties could be one of several factors which may directly affect women’s work intensity and health. Table 5.11 also shows that childcare is a household activity that is often combined with other activities. When both primary and secondary childcare activities are taken into account, the original average time of 58.9 minutes spent by women in childcare increase to 153 minutes (assumption 1) or to 105.9 minutes (assumption 2), an increase of 160 percent and 80 percent respectively. Men’s average total childcare time increased by 241 percent from 23.61 minutes to 80.55 minutes (assumption 1) or by 120 percent to 52.08 minutes (assumption 2). Childcare is another activity that reveals gender differences. Whether as a primary or overlapped activity, women spent more time caring for children than men. Most unpaid household work activities were done as secondary activities, e.g., working on wage-contracted assignments and minding children simultaneously. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 122 Gender inequalities appear in the non-work categories as well. For instance, under assumption 2, men spent 18 percent more time than women on leisure activities, and 4 percent time more sleeping. Accounting for secondary activities stretched both men’s and women’s time spent on work and leisure beyond 24 hours. By performing overlapping activities, women stretched their time by 593 minutes (41 percent increase) using the first assumption, and 297 minutes (21 percent increase) using the second assumption. Men, on the other hand, stretched their time by 32 percent (assumption 1), and 16 percent (assumption 2). This suggests that there is an underestimation of the amount of unpaid labor used in the non-market production of goods and services when secondary activities are omitted. Also, women have a greater tendency to combine one or more activity per unit of time and carry a higher work burden than men. This is because women have already been assigned their job functionality (by norm and/or culture) to be in charge of household unpaid work. Hence, women often stretch their time beyond 24 hours in order to accommodate both labor market and unpaid household work. Figures 5.7 to 5.9 illustrate the information on Table 5.11. Figure 5.7 shows the time spent on all primary activities by men and women. It indicates clearly that women spent more time on unpaid household work, while men spent more time on paid labor work and other non-economic related activities such as leisure. However, when time use data takes into account overlapped activities, a different picture emerges. The overall work burden increases dramatically for female respondents, especially labor market work, domestic work, and childcare activities, as shown in Figure 5.8 (assumption 1) and Figure 5.9 (assumption 2). With respect to leisure and other non-work activities, women in the sub-sample seem to allocate slightly less time to these activities than men. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. poue wt pr sin fte oyih onr Frhrrpouto poiie ihu pr ission. perm without prohibited reproduction Further owner. copyright the of ission perm with eproduced R Minutes per day Primary Activity Primary Figure 5.7: Men and Women TimeOnlyand Women Primary5.7:Allocation: Men Activities Figure akt ok Care Work Market ao Dmsi Cidae hpig esr Proa Sleepping Personal Leisure Shopping Childcare Domestic Labor Work 0 40 0 80 00 20 40 1600 1400 1200 1000 800 600 400 200 Minutes per day per Minutes Activities E3 Male ■ Female ■ Male ■ Female □ 123 poue wt pr sin fte oyih onr Frhrrpouto poiie ihu pr ission. perm without prohibited reproduction Further owner. copyright the of ission perm with eproduced R Overlapped Activity Overlapped Minutes per day Women Allocation:Time PrimaryOverlapped and Figure and Men 5.8:Activities 0 0 6 Primary and Primary 00 7 akt ok Care Work Market ao Dmsi Cidae hpig esr Proa Sleepping Personal Leisure Shopping Childcare Domestic Labor Work 80 80 90 1950 1900 1850 1800 (Based on Assumption 1: EqualWeight)(Based on Assumption Activity Minutes per day per Minutes 2000 2050 Male ■ El Female Female □ Male ■ 124 poue wt pr sin fte oyih onr Frhrrpouto poiie ihu pr ission. perm without prohibited reproduction Further owner. copyright the of ission perm with eproduced R Overlapped Activity Overlapped Minutes per day Primary Women Allocation:Time andand OverlappedMen Figure 5.9:Activities akt ok Care Work Market ao Dmsi Cidae hpig esr Proa Sleepping Personal Leisure Shopping Childcare Domestic Labor Work 60 60 60 60 60 60 70 70 70 70 70 1750 1740 1730 1720 1710 1700 1690 1680 1670 1660 1650 1640 FT' T F T (Based on Assumption 2: Half 2:(BasedHalf on Weight)Assumption I Activity Minutes per day per Minutes ...... I _____ I .... . ■ Male ■ 0 Female 0 125 ■ Male ■ Female 126 CHAPTER 6 EMPIRICAL ANALYSIS OF INDIVIDUAL WELL-BEING 6.1 ANALYSIS OF THE WELL-BEING INDEX FOR THAILAND HOME - BASED WORKERS The analysis of individual well-being is empirically investigated in this chapter. A composite index based on individual level information such as the level of educational attainment, personal income, and inverse incidence of work intensity, is developed and then applied to the 2002 Thailand sub-sample survey data. The composite well-being index ranges from 0 (zero) to 1 (one), with a higher number corresponding to a higher quality of life. For estimation and investigation purposes, two tests are conducted on the well-being index of individuals and the inverse work intensity index. Both the well-being index of an individual (WBIt) and the inverse work intensity index (k) is expressed as a function of the exogenous economic demographic and social factors that can diminish or enhance a person’s well-being and incidence of work intensity. We estimate the following models using ordinary least squares regression analysis: MODEL 1: WBI, (y,k,edu) = xijpj + e, (6.1) MODEL 2: k, = xl}p} + e, (6.2) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 127 WBI, = A + FEM,fi2 + EMT& + SS,fiA + ORG,fr + DEPfo + (p .3 J AGEfi1 + AGE]/?8 + e, Jfc, = + FEMlp2 + EMT& + SS,fiA + ORGfi, + DEPtj36 + (6.4) AGEtP7 + AGEffa + et where WBI, = Estimated well-being index of individual i; kj = The inverse work intensity index of individual /; FEMi = A dummy variable for gender of individual i where female =1 and male =0; EMTi = A dummy variable for employment type of individual i where wage-contracted = 1 and self-employed = 0; SSt = A dummy variable for social support received by individual i from various sources, where receiving social support =T and not receiving social support = 0; ORGi = A dummy variable for organizational capacity of the community where individual i resides, where a high level of organization = 1 and a low level of organization = 0; DEI] - A dummy variable for presence of dependent members (children aged under 15 years old) in the household where present in the household =1 and present in the household = 0; AGE, - The age of individual i, defined in years; and AGE' =The square of age of individual i, defined in years. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 128 6.1.1 The Analysis of Well-Being Index (Model 1) The sex of the individual ( F E M i ) reflects the prevailing social and gender norms in Thailand and influences the household division of labor. The community-based environment ( O R G l ) affects an individual’s well-being in several ways: a) provisioning or availability of basic services such as health, water, sanitation, and electricity; b) access to market opportunities outside the community in terms of proximity to large centers or good public transportation; and/or c) the extent of social capital and community-based networks of support in terms of presence of active community organizations, formal as well as informal. To test the effect of the community environment on the individual well being index, we classify the communities where respondents reside into two categories based on their organizational capacity, social cohesion, and access to support services. The community type are classified as well organized and very supportive (with O R G j dummy =1), and less organized and less supportive communities (with O R G j dummy = 0). The dummy variable for social support ( S S i ) indicates whether or not the individual and his/her household receive social support from civil society groups or organizations. These include nongovernmental organizations (NGO), saving clubs, workers’ associations or labor groups. Social support also includes support from government agencies in the form of scholarships, loans, trainings, and collective bargaining. Similarly, the dummy variable for household structure ( D E P t ) indicates whether or not a dependent member is present in the household where the respondents reside. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 129 The regression results for this model are given in Table 6.1. The individual and household information explains roughly 28 percent of the individual well-being index. Overall, these individual and household variables are reliably predicting the well- Table 6.1: Coefficients Estimates from OLS, Model 1 Well-Being Index Coefficients Robust Standard Errors Constant 0.3264**** 0.1096 SEX -0.1221**** 0.0295 EMT 0.0148 0.0166 SS 0.0301** 0.0167 ORG 0.0387** 0.0219 DEP -0.0139 0.0156 AGE 0.0043 0.0049 AGE2 -0.0000 0.0000 Number of observation 110 F( 7,103) 4.31 R2 0.2828 **** Significant at 1 percent level *** Significant at 5 percent level ** Significant at 10 percent level * Significant at 20 percent level being level of individuals. Given the small size of the sample, the residuals have been tested to ensure that they are distributed normally.1 The test results indicated that the p - 2 values for the /-tests and F-test in our analysis are valid and accurate. The model was also tested for heteroskedasticity. By utilizing the robust standard errors, however, our 1 Normality is not required in order to obtain unbiased estimates of the regression coefficients. OLS regression merely requires that the residuals (errors) be identically and independently distributed. Furthermore, there is no assumption or requirement that the predictor variables be normally distributed. 2 The p-value from Shapiro-Wilk W test is based on the assumption that the distribution is normal; therefore, a very large /j-value (.81), indicates that we cannot reject the hypothesis that the residual is normally distributed. We also conducted a standardized normal probability (P-P) plot and a plot of the quantiles of a variable against the quantiles of a normal distribution, and the plots show no indications of non-normality in residuals. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 130 model would still be valid even if heteroskedasticity exists. Also, a regression 4 specification error test indicated the non-existence of model specification error. Each of the explanatory variables in the model will be discussed separately, starting with the gender effect on the well-being of the individual. 6.1.1.1 Gender From Table 6.1, as expected, the gender coefficients show that well-being drops significantly if the individual is female. This significant welfare difference between men and women respondents could be explained by prevalent Thai social norms and customary views regarding the role and status of women . 5 Although the role of Thai women has never been solely restricted to the household or domestic sphere, their labor market activities tend to be limited to labor intensive tasks requiring little or no skills - such as food processing, sewing clothing, and so on (Praparpun, Boonmathya, and Leechanavanichpan 1999). Because their assigned primary role is to raise children and care for the family, higher levels of education is perceived as unnecessary (Pramualratana, Havanon, and Knodel 1985; Thailand, National Commission on Women’s Affairs, Office of the Prime Minister 1995a; and Yoddumnem-Attig, Bencha et al. 1992). This translates The Cook-Weisberg test for heteroskedasticity using fitted values of well-being index indicates the problem of heteroskedasticity in the model. 4 A model specification error can occur when one or more relevant variables are omitted from the model or one or more irrelevant variables are included in the model. If relevant variables are omitted from the model, the common variance they share with included variables may be wrongly attributed to those variables, and the error term is inflated. On the other hand, if irrelevant variables are included in the model, the common variance they share with included variables may be wrongly attributed to them. The males are traditionally breadwinners and the heads of the family, while women care for the home, the children, and the family. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 131 into low levels of educational attainment, with women respondents finishing only compulsory education or less. The problem of low levels of educational attainment for Thai women is also raised in other studies on women workers by Lazo (1992), UNDP (2003), Praparpun, Boonmathya, and Leechanavanichpan (1999), Thailand, National Commission on Women’s Affairs, Office of the Prime Minister (1995b), and National Statistical Office Thailand (1999). Women also tend to have lower earnings and higher levels of work intensity. This high work intensity was frequently observed in our fieldwork among urban couples in the low-income neighborhoods of Bangkok. Both spouses typically worked as home-based workers. When they finished their work in the evening, men typically rested immediately afterwards. Women were unable to rest, however, as they needed to clean up the workplace, put the working tools back in place, and then prepare the evening meal. This lower level of earnings and higher level of work intensity for female home-based workers is not only observed in Thailand, but also in other Asian countries. Bajaj (1999,39) indicates that “female wages were found to be 6 6 percent of male wages in the garment industry in Bangladesh.” The relatively higher incidence of work intensity alongside lower educational attainment and earnings helps to explain the lower value of the women’s well-being index. Table 6.2 presents a statistical summary of individual well-being for Thai 7 men and women in our sub-sample. The result is consistent with the regression estimation. On average, the well-being index for Thai men is higher than that of women 6 UNDP (2003) reported in the Human Development Report 2003 that the adult literacy rate of Thai women is roughly 3 percent lower than the adult literacy rate of Thai men. 7 This subsample includes 92 female respondents and 18 male respondents. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 132 (0.411 compared to 0.302, respectively). This indicates that men respondents from our sub-sample experienced a higher quality of life than women respondents. Table 6.2: Individual Well-Being Index and Component Indices, by Sex Female (Na92) Male (N=18) Individual Well-Being Index1 Mean 0.302 0.411 Std. deviation 0.072 0.120 Minimum/Maximum 0.102/0.455 0.167/0.640 Decomposition of Well-Being Index 1) Educational Attainment Index* Mean 0.296 0.403 Std. deviation 0.161 0.220 Minimum/Maximum 0.000/0.750 0.000/0.875 2) Personal Income Index3 Mean 0.384 0.516 Std. deviation 0.153 0.169 Minimum/Maximum 0.000/0.732 0.234/0.999 3) Inverse Work Intensity Index4 Mean 0.226 0.315 Std. deviation 0.167 0.155 Minimum/Maximum 0.010/0.601 0.101/0.555 Note: Full details on these calculations can be found in chapter 4. 1. The individual well-being index is calculated as jygj ------where 0 £ WBI £ 1 and I * edu , v, k 1 J i *‘r* m 2. The educational attainment component index is calculated as 3. The personal income component index is calculated as y (l0gK,}-l0g{min;{^,}}) (log {max y {X j} -log{m in {x j}) 4. The inverse work intensity component index is calculated as k l 0 ° g {X**J } " [°g {min; {X ^ . J }}) , (l0g { ^ J-l0g{min;{^,}}) ] 2 [(logfm ax.{xwy}} -log{miny{xw .}}) (log{maxy{*„.}} -log{miny{ * ,,} } )/ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 133 As can be seen in Table 6.2, women also had a lower mean in every individual attribute index. The educational attainment index for women was 0.296 compared to 0.403 fro men. The personal income component index was 0.384 for women and 0,516 for men, while the inverse work intensity component index was 0.226 for women and 0.315 for men. Figure 6.1 presents the kernel density distribution of the individual well being index by sex. It shows that more women are found at the lower end of the individual well-being index. Figure 6.1: The Kernel Density of Estimated Individual Well-Being Index, by Sex • Women Men Individual Well-Being Index 6.1.1.2 Employment Type We now turn to the effect of employment status on individual well-being. Our sub-sample is limited to wage-contracted workers and self-employed workers. The Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 134 results from the OLS test (Table 6.1) show that the level of well-being of a sub contracted worker is slightly higher than that of a self-employed worker. However, this result is not statistically significant. If one does not control for other factors such as gender and age, the mean of the individual well-being index of self-employed worker respondents (0.327) is slightly higher than that of the wage-contracted worker respondents (0.309) as shown in Table 6.3 . 8 Table 6.3: Individual Well-Being Index and Component Indices, by Employment Type Wage-Contracted Self-Employed (N~45) (N=65) Individual Well-Being Index1 Mean 0.309 0.327 Std. deviation 0.082 0.096 Minimum/Maximum 0.167/0.532 0.102/0.640 Decomposition of Well-Being Index 1) Educational Attainment index* Mean 0.297 0.324 Std. deviation 0.170 0.179 Minimum/Maximum 0.000/0.563 0.000/0.875 2) Personal Income Index3 Mean 0.328 0.460 Std. deviation 0.166 0.136 Minimum/Maximum 0.000/0.732 0.167/0.999 3) inverse Work Intensity Index* Mean 0.302 0.198 Std. deviation 0.191 0.135 Minimum/Maximum 0.010/0.601 0.047/0.592 Note: see Table 6.2 note. The median of the individual well-being index by employment type is 0.325 and 0.304, for self-employed worker respondents and wage-contracted worker respondents respectively. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 135 A further decomposition of the well-being index shows that wage-contracted home-based workers tend to earn lower personal income and have a lower levels of educational attainment, reflected by the lower values in these component indices (Table 9 6.3). As mentioned in Chapter 5, it is important to note that none of the wage-contracted workers received social security or benefits; they worked longer hours under poor working conditions; and they earned very low wages. The inverse work intensity component index shows that the mean value for wage-contracted workers is higher than that of self-employed workers, at 0.302 and 0.198 respectively. This implies that for our sub-sample, wage-contracted workers worked less intensively than self-employed workers. Possible reasons for this were discussed in Chapter 5. Table 6.4 shows the individual well-being index of our sub-sample by employment type and sex. Male respondents in both types of employment yielded higher well-being indices than female respondents . 1 0 This implies that males, whether wage- contracted or self-employed, experience a better quality of life than females in both types of employment. It is also revealed that women score lower on every component index in both types of employment. This result also reconfirms the difference in well-being between men and women. This supports the regression analysis results that employment type does not explain men and women well-being differences. 9 Wage-contracted workers are often exploited in remuneration and working hours by the contractor. 10 Note that the number of observations for male wage-contracted workers is only two. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 136 Table 6.4: The Individual Well-Being Index and Component Indices, by Sex and Employment Type Wage-Contracted Self-Employed Female Male Female Male (N=43) (N=2) (N=49) (N=16) Individual Well-Being Index Mean 0.307 0.350 0.298 0.419 Std. deviation 0.073 0.259 0.071 0.107 Variance 0.005 0.067 0.005 0.011 Decomposition of Well-Being Index 1) Educational Attainment Index Mean 0.298 0.281 0.293 0.418 Std. deviation 0.162 0.398 0.160 0.205 Variance 0.026 0.158 0.026 0.042 2) Personal Income Index Mean 0.326 0.361 0.436 0.535 Std. deviation 0.168 0.180 0.118 0.164 Variance 0.028 0.032 0.014 0.027 3) Inverse Work Intensity Index Mean 0.297 0.406 0.164 0.304 Std. deviation 0.192 0.198 0.110 0.153 Variance 0.037 0.039 0.012 0.023 Note: see Table 6.2 note. While Table 6.3 indicated that self-employed workers generally had a higher quality of life than wage-contracted workers, this is not the case for self-employed female workers. As shown in Table 6.4, female wage-contracted workers actually had a higher well-being index number than female self-employed workers. A decomposition of the individual well-being indices reveals that the incidence of work intensity is much higher for self-employed female workers (0.164) than wage-contracted female workers (0.297).11 This result reflects the fact that higher work intensity deteriorates the well- This is the number obtained from the inverse work intensity index. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 137 being of individuals through health effects (Wolfe and Haveman 1983 and Floro and Hungerford 2001). The decomposition also shows that female wage-contracted workers who earn the least, has the lowest personal income index. This corresponds to an ILO study which states that “women homeworkers in manual jobs are among the lowest paid workers in the world” (ILO 2002a, 43). 6.1.13 The Level of Community Organization Individual well-being is also indirectly affected by the community environment and provisions available where respondents reside. All three communities were similar in terms of availability of services. All had access to water, electricity, and public transportation. However, more than 85 percent of the people living in Udomsuk 12 had no access to a sewage system. Also, Udomsuk community is farther from the main road and lacks sufficient public transportation. Nomklao and Nawamin also had more active community organizations and support groups, such as home-based workers organizations, saving clubs, women groups, and occupational groups. The saving clubs of these two communities were so well organized that the residents had even collectively negotiated with the land owners to rent the land instead of illegally occupying the area; thereby reducing the associated uncertainty and risk of being forced out. Further, these two communities were able to obtain support from government organizations such as the Community Development Office and the House Associations and Training Center for Urban Poor Development for training and employment assistance. While less than 30 percent of individual in Nawamin community have no access to public sewage system and less than 1 percent in the case of people who live in Nomklao community. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 138 As expected, the coefficient of community organization capacity ( ORGi) in Table 6.1 shows that individuals who reside in well-organized and supportive communities are more likely to have a higher level of well-being. This can be explained Table 6.5: The Individual Well-Being Index and Component Indices, by Community Size and Level of Organization Well Organized Less Organized and Very and Less Supportive Supportive Com m unity Com m unity (N=96) (N=14) Individual Well-Being Index Mean 0.322 0.301 Std. deviation 0.089 0.101 Minimum/Maximum 0.102/0.640 0.167/0.532 Decomposition of Well-Being Index 1) Educational Attainment Index Mean 0.309 0.344 Std. deviation 0.173 0.191 Minimum/Maximum 0.000/0.875 0.000/0.563 2) Personal Income Index Mean 0.424 0.283 Std. deviation 0.152 0.185 Minimum/Maximum 0.000/0.999 0.020/0.592 3) Inverse Work Intensity Index Mean 0.235 0.278 Std. deviation 0.167 0.275 Minimum/Maximum 0.010/0.601 0.070/0.563 Note: see Table 6.2 note. by the fact that well-organized and supportive communities are able to better support residents by enhancing their bargaining power with contractors or manufacturers; developing innovative savings and credit facilities such as savings clubs and labor groups; and mobilizing residents to demand better access to infrastructure, social services Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 139 and support from the government and other institutions (Thailand, HomeNet 1999 and Economic and Social Commission for Asia and the Pacific: ESCAP 2001). Table 6.5 shows the statistic summary of the respondents’ well-being index by the level of community organization and support. This confirms the OLS analysis that respondents from well-organized communities generally have a higher well-being index value, with 0.322 for those respondents compared to 0.301 for respondents from less organized communities. 6.1.1.4 Life Cycle (Age) We now explore the relationship between individual life cycle stages and the individual’s well-being index. Life cycle theory divides human life into 5 stages: infancy, learning, maturity, retirement, and the prospect of death (Sutch 1991 and Modigliani 1986). Individual age (AGE) and the square Of that age value (AGE2) are used in the regression to measure the effect of life cycle on individual well-being. The square value of age is used to capture the non-linear relationship of life cycle and well-being. The positive coefficient of AGE in Table 6.1 indicates that as an individual ages (progressing through the more intensive work stages of the lifecycle), his or her well-being increases slightly (as seen by the small size of the coefficients). The negative sign of AGE 2 indicates that the well-being of an individual increases at a decreasing rate. However, neither age nor age-squared are statistically significant in the model. They are also insignificant when we do not control for other independent variables. Table 6 . 6 presents the summary of the respondents’ well-being index by individual life cycle stage (age group). The life-cycle data are categorized into 3 stages, the first maturity stage, AGEi, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 140 (20-37), the second maturity stage, AGE 2 , (38-55), and the semi-retirement stage, AGE 3 , (56 and up). The mean value of the well-being index for the first and the second maturity stages of individual life-cycle differed only slightly from 0.320 to 0.316. However, the well-being index increased dramatically to 0.338 at the semi-retirement stage of the life cycle. This result can be explained in part by the age hierarchy in cultural norms of Thai people. Older persons are not required to work as much in the household, leaving these to Table 6 .6 : The Individual Well-Being Index and Component Indices, by Stage of Life-Cycle First Second Maturity Maturity Retirement Stage: Age Stage: Age Stage: Age of 20-37 of 38-55 of 56 and up (N=44) (N-54) (N=12) Individual Well-Being Index Mean 0.320 0.316 0.338 Std. deviation 0.070 0.094 0.140 Minimum/Maximum 0.181/0.455 0.102/0.640 0.167/0.628 Decomposition of Well-Being Index 1) Educational Attainment Index Mean 0.354 0.280 0.313 Std. deviation 0.174 0.153 0.246 Minimum/Maximum 0.000/0.750 0.000/0.750 0.000/0.875 2) Personal Income Index Mean 0.383 0.418 0.433 Std. deviation 0.155 0.161 0.197 Minimum/Maximum 0.000/0.712 0.065/0.999 0.167/0.831 3) Inverse Work Intensity Index Mean 0.223 0.248 0.269 Std. deviation 0.168 0.170 0.165 Minimum/Maximum 0.010/0.601 0.042/0.577 0.056/0.560 Note: see Table 6.2 note. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 141 be completed by younger household members. The length of their working day also declines due to some physical limitations. The personal income index rises from 0.383 in the first maturity stage, to 0.418 in the second maturity stage, and to 0.433 in the semi- retirement stage. This outcome is consistent with the typical life-cycle model, where individuals or households accumulate wealth or saving to provide for retirement (Hubbard, Skinner, and Stephen 1994; Modigliani 1986; and King and Dicks-Mireaux 1982). The inverse work intensity index also tends to be positively related to life-cycle stages. The inverse work intensity index rises from 0.223 at the first maturity stage, to 0.248 at the second maturity stage, and to 0.269 at the semi-retirement stage due to age hierarchy in Thai cultural norm. 6.1.1.5 Social Support and Household Structure (Presence of Dependent Members) The variable coefficients of social support (SSt) and household structure (DEP,) in the OLS test yields interesting results (Table 6.1). An individual who receives social support is more likely to have higher well-being than a person who does not, which is as expected. The significance of this variable may be explained by the fact that with social support from various sources, the individual could have gotten a better job contract, attended skill training workshops, accessed low interest loans, gotten a higher education, and so on. This social support variable does not only affect individual well-being directly, but can indirectly affects individual well-being by enhancing the well-being of dependent members. An example of this may be a scholarship for the child of a respondent which Under Thai norms and culture, all unpaid household tasks are fulfilled by younger household members, especially sons and daughters, when parents are older. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 142 allows her or him a better education leading to an eventual increase in the well-being of the entire family and society. The coefficient of the dummy variable for household structure ( DEPj) suggests that with the presence of dependent members in the household, the well-being of the adult member tends to decline. However, this finding is not statistically significant at any level of significance. The explanation of this negative relationship might be that the incidence of work intensity increases (a decline in inverse work intensity index) since the tendency to overlap activities increases due to the need to care for children. Floro and Hungerford (2001) suggest that individuals in different stages of the life cycle have different demands on their time. Parents with young children probably experience greater time pressure than childless individuals. This increase in time pressure is likely to result in multitasking and increase work intensity. Juster and Dow (1985) also found similar evidence from their process well-being measurement where having children tended to lower the level of well-being of individuals. Table 6.7 presents the summary of the respondents’ well-being index by social support received and household structure. The mean value of the well-being index for the individuals who received social support from various sources is higher than who 14 do not received social support, 0.349 to 0.305. The mean value of the well-being index by household structure, however, indicates that there is indifference whether or not dependent members are presented in the households. This indifference is also confirms 14 This difference is also statistically significant as suggested by the t-statistic of 2.468. This result is consistent with the regression result in Table 6.1. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 143 by the value of t-statistic of 0.009, which is corresponding with the regression results in Table 6.1. Table 6.7: The Individual Well-Being Index and Component Indices, by Social Support and Household Structure Social Support Household Structure With Without Receive Not Receive Dependent Dependent (N=36) (N=74) (N=72) (N=38) Individual Well-Being Index Mean 0.349 0.305 0.320 0.320 Std. deviation 0.097 0.084 0.086 0.098 Minimum/Maximum 0.201/0.640 0.102/0.628 0.163/0.640 0.102/0.628 Decomposition of Well-Being Index 1) Educational Attainment Index Mean 0.354 0.293 0.326 0.289 Std. deviation 0.188 0.166 0.182 0.161 Minimum/Maximum 0.000/0.750 0.000/0.875 0.000/0.750 0.000/0.875 2) Personal Income Index Mean 0.452 0.384 0.411 0.396 Std. deviation 0.154 0.163 0.158 0.172 Minimum/Maximum 0.000/0.830 0.103/0.999 0.000/0.999 0.065/0.831 3) Inverse Work Intensity Index Mean 0.243 0.239 0.223 0.274 Std. deviation 0.124 0.186 0.157 0.183 Minimum/Maximum 0.054/0.577 0.010/0.601 0.010/0.601 0.051/0.577 Note: see Table 6.2 note. 6.1.2 The Analysis of Inverse Work Intensity Index (Model 2) Model 2 investigates more closely the relationship of individual and social factors on individual’s incidence of work intensity (inverse work intensity index). Table Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 144 6 . 8 shows the result of the OLS test. 1 5 In Table 6 .8 , it can be seen that the inverse work intensity index drops significantly if an individual is female. This finding supports the results of the well-being regression analysis shown in Table 6.1 since women’s health and well-being levels deteriorate when there is a high incidence of work intensity. Employment type also plays a significant role in this model. The positive sign of the coefficient indicates that wage-contracted workers have less incidence of work intensity than that of self-employed workers. As discussed before, both wage-contracted workers and self-employed workers have long working days and tend to overlap their tasks. However, self-employed workers are more likely to be faced with a higher incidence of work intensity due to the fact that they work longer hours (especially those who own home-based grocery stores), and they are more likely to overlap paid market work and unpaid domestic work, as suggested in Chapter 5. One of the most interesting findings in this model is that the individual’s incidence of work intensity increases (reduction in the inverse work intensity index) when a dependent member is present in the household. The significant increase in work intensity mainly comes from an increase in multitasking, especially on the childcare activities. This result corresponds to the Floro and Hungerford (2001) argument that childcare increase work intensity. This finding is also consistent with the results of the well-being regression analysis in Table 6.1. The normality test suggested that the residual in this model seems to be not normally distributed. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 145 Table 6 .8 : Coefficients Estimates from OLS, Model 2 Inverse Work Intensity Index Coefficients Robust Standard Errors Constant 0.6284 0.2475 SEX -0.1515**** 0.0427 EMT 0.1416**** 0.0352 SS 0.0057 0.0268 ORG 0.0256 0.0491 DEP -0.0438* 0.0307 AGE 0.0124 0.0124 AGE2 -0.0001 0.0001 Number of observation 110 F( 7,102) 4.720 R2 0.210 **** Significant at 1 percent level *** Significant at 5 percent level * * Significant at 10 percent level Significant at 20 percent level 6.2 SUBJECTIVE WELL-BEING INDICATORS AND THE VALIDITY OF THE INDIVIDUAL WELL-BEING INDEX We now investigate the validity of the individual well-being index by using a comparative analysis with respondents’ own perception of their well-being. Using the answers to the following survey question: “How satisfied are you at present with your life as a whole?", we developed a parallel indicator, the subjective well-being indicator, based on their responses . 1 6 The most dissatisfied response corresponds to 0 and the most satisfied response corresponds to 10. The subjective well-being indicator reveals a subjective perception of the quality of life, which could be used as a reasonable proxy of The calculation and the methodology of subjective well-being indicators were discussed in chapter 4. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 146 17 individual well-being. While the individual well-being index was developed based on the observable determinants of well-being: personal income, the level of educational attainment, and the incidence of work intensity, the individual subjective well-being indicator is constructed based on both unobserved and measured: a respondent’s feeling about life in general. Individual behavior or well-being is driven not only by the achievement of higher levels of income, socio economic variables, or the manner of time spent during a day, but also by feelings of revenge and jealousy, imitation of others, social norms and institutions, and legal prohibitions (Gibbard 1996; Pradhan and Ravallion 2000; Ferrer-i Carbonell 2002b; and Van Praag; Frijters; and Ferrer-i Carbonell 2002). Table 6.9 presents the individual subjective well-being indicators for all respondents on average by sex and by employment type. As with the individual well being index, on average, the individual subjective well-being indicator of men is higher than that of women (0.728 and 0.593, respectively). This suggests that women from our sub-sample feel less satisfied with their life in general than men do. This well-being difference (mean values) is also confirmed by the result of t-statistic. Table 6.9 also gives the individual subjective well-being indicators by employment type. On average, self- employed workers had a higher subjective well-being indicator than wage-contracted workers. However, the difference is small at 0.623 for self-employed workers compared This is due to the fact that individuals tell us about their true level of well-being based on all personal information available, some of these which cannot be accessed or observed by others. Moreover, the use of this gauge of individual well-being requires two assumptions as discussed earlier: a) individuals are able to evaluate their own situation, and b) responses among individuals can be compared. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 147 to 0.604 for wage-contracted workers. However, the t-statistic suggests that this welfare difference is statistically insignificant at any level. Table 6.9: The Summary of Individual Subjective Well-Being Indicators Subjective Well-Being Indicator1 Mean Std. deviation T-Statistic2 By Gender of Respondents Women (N=92) 0.593 0.183 2.873 Men (N=18) 0.728 0.171 “ By Employment Type Wage-Contracted Worker (N=45) 0.604 0.192 0.511 Self-Employed Worker (N=55) 0.623 0.185 Note: The M l details on this calculation can be found in chapter 4, 1. swb = f e * ~JT- 2. This is t-test where null hypothesis is mean values are the same for these two groups. To see how well our well-being component indices explain the respondents’ feeling of their well-being, we test the following two models of subjective well-being indicators and the well-being component indices MODEL3: SWB, = ft+ EDU,J32 + Y,&+e, (6.5) MODEL 4: SWB, = /?,+ EDU,/32 + Y fo + k,(3A + e, (6 .6 ) 18 Table 6.10 shows the results from the OLS regression analysis. In the first model where educational attainment and income alone are control variables, our well being component indices can explain 34.10 percent of the subjective well-being indicator. Both of our models have been tested for the normality distribution in residuals, and the results indicate a normality distribution. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 148 The personal income component index has a statistically insignificant effect on the subjective well-being level of individual. The inverse incidence of work intensity index is then added as an explanatory variable in the second model. The personal income component index and the other two indices are now statistically significant, with the expected signs. This may indicate that income buys happiness only when people do not have to work incredibly long hours and more intensively to earn it. In other words, money income alone cannot explain the well-being of our respondents. Well-being is based on both income and how long the individual works to earn that money (including work intensity). What appears to matter for subjective well-being of individuals is not 19 their money income, but their hourly earnings. This finding is important since it supports our hypothesis that the time use factor (incidence of work intensity) can convey crucial well-being information that traditional factors do not. Hence, ignoring the time use factor in the estimation of well-being could result in a serious estimation bias. Also, by taking into account the incidence of work intensity, the explanatory power of the well being component indices for the subjective well-being indicator jump dramatically to 46.20 percent. Further, the educational attainment index has a highest effect to the subjective well-being as the difference between the least and best educated person translates to an increase of 0.59 of the subjective well-being indicator, on the scale where 0 is the lowest level of well-being and 1 is the highest quality of life level. 19 This implies that low-wage jobs with long work hours do not make people happy. This result gives a substantial benefit to policymakers, which will be discussed in policy implications section. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 149 Table 6.10: Coefficient Estimates of Well-Being Component Indices (robust standard errors in parentheses) Subjective Well-Being Indicator Model 3 Model 4 Constant 0.3857“* 0.2276*** (0.0425) (0.0512) EDU 0.6006“* 0.5903*** (0.0847) (0.0701) Y 0.1027 0.2541*“ (0.0929) (0.0837) k 0.4156“ * (0.0758) Number of observation 110 110 F( 8,101) 34.30 36.01 R2 0.341 0.462 **** Significant at 1 percent level *** Significant at 5 percent level ** Significant at 10 percent level * Significant at 20 percent level We now investigate the gender effect which is found to be strongly related to both the well-being index and the subjective well-being indicator in our study. Four models of the subjective well-being indicator have been tested by controlling for the gender of the individual (FEM) and the education index ( EDU), the personal income index (F), and the inverse work intensity index (/t), as follows: MODEL 5: SWBi = ft + FEM, ft + e, (6.7) MODEL 6 : SWB, = ft + FEM,ft + EDU,ft + e, (6 .8 ) MODEL 7: SWB, = ft+ FEM, ft + EDU, ft + y ,ft + et (6.9) MODEL 8 : SWB, = ft + FEM, ft + EDU, ft + y,ft +k,ft+e, (6.10) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 150 20 Table 6.11 presents the results of the OLS tests of Models 5 to 8 . The results of Model 5 reveal the negative relationship between the gender effect (FEM) and respondents’ general perception of well-being. This gender effect is very significant with a large coefficient. By controlling for the educational attainment level (EDU), Model 6 indicates that the strong gender effect from earlier model starts diminishing both in coefficient size and in the level of significance. This means that the level of educational attainment can explain some parts of the negative relationship between gender and subjective well-being. The explanatory power of the independent variable also jumps from 7.1 percent to 35.2 percent. When controlling for income (y), Model 7 is consistent with Table 6.10, demonstrating that money income alone does not fully account for life satisfaction and happiness. However, it is able to explain some parts of the gender effect on the subjective well-being indicator. The most interesting findings are in Model 8 when we control for the incidence of work intensity (k). The gender effect on subjective well-being now has a very small coefficient and is statistically insignificant. The results from Model 8 imply that the gender effect is derived from the fact that women in our sub-sample have lower income per hour along with higher work intensity (as explained by the inverse work intensity index and the personal income index) and lower educational attainment levels. The explanatory power (R2) increases dramatically from 35.5 percent in Model 7 to 46.1 percent in Model 8 . This reveals that the incidence of work intensity is a very important factor in determining a women’s level of subjective well-being. Similarly to Table 6.10, 20 The residuals for all of these models have been tested whether or not they are normally distributed, and all models here were past the test. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 151 the educational attainment index substantially affects the subjective well-being of individual -the difference between the least and best educated person translates to an increase of 0.59 of the subjective well-being indicator. Table 6.11: Coefficients Estimates from OLS, Models 5 to 8 (robust standard errors in parentheses) Subjective Well-Being Indicator Model 5 Model 6 Model 7 Model 8 Constant 0.7278**** 0.4931**** 0.4633**** 0.2286**** (0.3949) (0.0409) (0.0571) (0.0687) SEX -0.1343**** -0.0718*** -0.0642*** -0.0007 (0.0439) (0.0296) (0.0308) (0.0324) EDU 0.5825**** 0.5760**** 0.5900**** (0.0812) (0.0838) (0.0694) Y 0.0630 0.2535**** (0.0974) (0.0919) k 0.4151**** (0.0778) Number of observation 110 110 110 110 F-value 9.36 34.92 24.29 27.26 R2 0.071 0.352 0.355 0.462 **** Significant at 1 percent level *** Significant at 5 percent level ** Significant at 10 percent level Significant at 20 percent level Besides the explanatory power of the independent variables (indicated by R2), the relationship between the estimated well-being index and individual’s subjective perception of their well-being is also statistically tested by using the Spearman 21 correlation coefficient. The correlation matrix, given in Table 6.12, suggests that the Spearman's rank correlation coefficient does not use the actual observed data, but the ranks of the data, to compute a correlation coefficient. That is, replace the smallest X value with a 1, the next smallest with a 2, and so on. Repeat the same procedure for the Y values. Therefore, it makes no assumption about the distribution of the values. (Miller and Miller 1999) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 152 well-being index is likely to be highly and significantly correlated with the subjective well-being indicator. In other words, there is a fairly strong relationship between these two measurements. We can conclude that these two measurements are correlated to each other. The education component index is also highly correlated and significant for the Table 6.12: Spearman Correlations: The Individual Well-Being Index, Component Indices, and Individual Subjective Well-Being Indicator Inverse Work Education Personal Individual Individual Intensity Attainm ent Income S ubjective W ell-Being Component ComponentCom ponentW ell-Being Index Index Index Index Indica tor Individual Well-Being 1 Index Inverse Work Intensity 0.429** 1 Component Index 0 Education Attainment 0.642** 0.016 1 Component Index 0 0.987 Personal Income 0.404** -0.318** 0.065 1 Component Index 0 0.0007 0.5 Individual Subjective 0.616** 0.35** 0.528** 0.135 1 Well-Being Indicator 0 0.002 0 0.16 Note: 1. the values in the table are listed as follows: Correlation value Probability (P > |t|) Sample size 2. The Spearman’s rank correlation Coefficient is calculated from following formula Where, r = the Spearman Rank Order Correlation Coefficient, d - the difference between a subjects ranks on the two variables, n = the number of subjects. 3. ** Correlation is significant at the .01 level (2-tailed) 4. The null hypothesis is that the two variables are independent. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 153 subjective well-being indicator. Table 6.12 reveals that the educational attainment index is the most highly correlated index to the subjective well-being indicator. Interestingly, the negative relationship between the personal income index and the inverse work intensity index suggests that higher personal income comes at the expense of working longer hours with a higher incidence of overlapped work activities. This seems to explain why the inverse work intensity index is positively correlated to the amount of satisfaction (and lack of stress) as measured by the subjective well-being indicator. By looking at the comparative analysis with the respondent’s own perception of their well-being, individual well-being (and its components) seems to be highly correlated with the respondents’ subjective perception. Therefore, the individual well-being index is likely to be an appropriate well-being measurement for our sub-sample data. 6.3. INDIVIDUAL WELL-BEING (BORDA) RANK The individual well-being index measures the welfare of individuals based on three different determinants of well-being, and on the cardinality assumption. However, this kind of index (methodology in aggregation) lacks normative significance as suggested by Dasgupta (1993,1999,2001) and the UNDP (2002). A possible method of aggregation that yields normative significance is the Borda rule (Dasgupta 1993,1999 and UNDP 2002). UNDP (2002, 138) describes the Borda rule as one providing “a method of rank-order scoring, the procedure being to award each alternative a point equal to its rank in each criterion of ranking, adding each alternative’s scores to obtain its Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 154 aggregate score, and then ranking alternatives on the basis of their aggregate score.” Hence, the Borda rule invariably yields a complete ordering of alternatives, or a well- 23 being index in our case. By utilizing the Borda rule for our sub-sample data, we can find comparative ranking of each individual’s welfare. Generally, a higher individual well-being index score corresponds to a lower Borda well-being score. However, due to the result of complete ordering of alternatives, the respondent with the highest well-being index value will not necessarily yield the lowest Borda well-being aggregate score, or the highest well-being rank. Table 6.13 illustrates the calculation of the well-being aggregated scores of self-employed males under the Borda rule. The Borda score/rank has been given to each component of the index of well-being for each individual, and the aggregated well-being score is obtained by adding all of the scores, or (a+b+c) in the table. Then, the individual well-being rank is calculated by ranking all of the well-being scores. The lower the well-being score, the higher the well-being rank will be. The higher well-being rank refers to higher well-being for each individual. Table 6.13 shows that respondent who has the highest (Borda) well being rank (in the male self-employed category), but he is actually ranked third in the individual well-being index. This is due to the fact that the gap between each ranking component of well-being is very large for this respondent. For instance, he ranks first in the personal income rank, but he does poorly in the inverse work intensity rank (ranked For example, suppose that a respondent earns ranks of a, b, and c, for his/her component indices in the well-being index. Then, his/her Borda well-being score is obtained by simply adding all of his/her component indices ranking together, i.e., (a+b+c), in this case. 23 Dasgupta (1993) viewed it as a “social welfare function." Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3 2 2 6 13 11 14 16 Rank3 Borda W ell-Being 31 1 66 5 65 4 90 7 62 131 101 8 120 12 Borda Well- Aggregated Being Score2 8 138 18 19 60 30 51 33 115 59 76 36 74 63 37 22 23 109 10 20 132 14 61 143 = Number of Number =component/of well- individual Intensity (c) Borda Score: Inverse Work andm 0.36 0.14 0.27 0.282 0.363 0.515 0.101 88 132 W ork Index inverse Intensity 9 0.531 2 0.18 34 0.176 20 64 Borda Score: Personal Incom e (b) 0.999 1 0.172 0.478 34 0.552 0.537 17 0.521 0.522 0.423 46 0.246 0.522 20 0.25 Index Income Personal 1 0.831 2 4 0.508 31 4 0.592 44 0.423 0.339 46 78 0.196 18 24 24 0.522 20 24 0.522 20 48 0.475 38 4848 0.393 48 Education Borda Score: Attainment (a) m 0.75 0.25 0.25 48 0.478 0.25 0.25 0.25 0.25 0.563 0.875 0.375 0.563 0.375 Index Education A ttainm ent = — —— - where , /.. index =well-being ofComponent individual index, WBIj 2. Aggregated well-being 2. Aggregated score aofis score (a+b+c). sum Borda to each component index,assigned equal or 3. Well-being rank is based on the aggregated well-being score: The lower the score, the higher die rank is. well-being die the rank aggregated on the is score, lowerbased higher The the score: rank 3. Well-being 0.306 0.341 0.4060.390 0.188 0.5630.362 96 0.478 0.394 0.386 0.640 0.378 0.451 Index1 Individual W ell-Being Note: well-being individual The is index as calculated follow: 1. Table 6.13: ofthe Calculation Aggregated 6.13: The Table (Borda Well-Being Score Score/Rank) Male ofSelf-Employed Workers 1 5 8 3 0.5627 0.563 9 6 4 2 0.628 15 14 0.333 10 13 11 12 0.346 0.375 16 0.301 Rank W ell- Index Being being. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 156 63rd). This demonstrates how the Borda well-being rank gives us a more complete and comparable ranking system for individuals. Table 6.14 presents the average well-being rank by gender and employment type of the respondents from our sub-sample. On average, men are five places ahead of women in well-being rank at 50.89 compare to 55.96. This difference is also statistically significant as suggested by the t-statistic. Similarly, the average well-being rank of self- employed workers (45.48) is higher than the average rank of wage-contracted workers (69.07). The difference in mean values of the well-being rank between self-employed and wage-contracted workers, however, is not statistically significant. The results in Table 6.14 are consistent with the estimated well-being index and the subjective well-being indicator. Table 6.14: Well-Being Rank of Respondents, by Gender and by Employment Type Well-Being Rank Mean1 Std. deviation T-Statistic2 By Gender of Respondents Female (N=92) 55.957 32.112 4.495 Male (N=18) 50.889 31.889 By Employment Type Wage-Contracted Worker (N=45) 69.067 32.485 1.628 Self-Employed Worker (N=55) 45.477 28.030 ~ 1. The lower the average well-being rank, the better the well-being of individual is. 2. This is t-test where null hypothesis is mean values are the same for these two groups. We also conducted a regression of subjective well-being against the Borda score for personal income (y), education attainment (edu), and work intensity ( k), and the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 157 results confirmed the findings in Table 6.10. Similar regression tests of the Borda score 24 against the variables in Table 6.1 confirm the results given in that table as well. Dasgupta (1999) suggested the Spearman correlation coefficient to obtain the relationship among the Borda well-being ranking, each ranking based on the other well being components, and the subjective well-being ranking. This would show how close the Borda ranking is to the other components. Table 6.15 provides the correlation coefficient for each pair of rankings from the four rankings in Table 6.13. It indicates that the correlation coefficients between the Borda well-being ranking and the others are 0.445 with inverse work intensity; 0.5714 with educational attainment; 0.5033 with personal income; and 0.6248 with subjective well-being. The results suggest that the educational 25 attainment rank would be the closest measure to the quality of life ranking. Further, the present findings imply that if we had to choose a single indicator of aggregate well-being, either educational attainment or personal income would do. The matrix also suggests that the inverse work intensity rank is negatively related to the personal income rank, which is similar to the correlation matrix of the subjective well-being indicator. In other words, higher personal income comes at the expense of working more intensively, including longer working hours and overlapped work activities. Results are available upon request. 25 One possible explanation might be that education is associated with most of domains of life such as earnings, health status, social status, etc. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 158 Table 6.15: (Spearman) Correlation Matrix of Well-Being Ranks Education Personal Subjective Borda Well- Inverse Work Attainm ent Income W ell-Being Being Rank Intensity Rank Rank Rank Rank Borda Well-Being 1 Rank Inverse Work 0.4445** 1 Intensity Rank 0 Education 0.5714** 0.0016 1 Attainment Rank 0 0.9871 Personal Income 0.5033** -0.3175** 0.065 1 Rank 0 0.0007 0.5 Subjective Well- 0.6248** 0.3498** 0.5276** 0.135 1 Being Rank 0 0.002 0 0.1595 Note: 1. the values in the table are listed as follows: Correlation value Probability (P > |t|) Sample size 2. The Spearman’s rank correlation Coefficient is calculated from following formula 0%t Where, r = the Spearman Rank Order Correlation Coefficient, d - the difference between a subjects ranks on the two variables, n = the number of subjects. 3. ** Correlation is significant at the .01 level (2-tailed) 4. The null hypothesis is that the two variables are independent. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 159 CHPATER 7 SYNTHESIS AND CONCLUSIONS This final chapter summarizes the contributions of this dissertation to the theoretical and empirical literature on well-being measure in general and in the Thai context in particular. This summary also raises the implications of these contributions to our current understanding about well-being measurement in the context of a growing informal sector, especially the number of home-based workers. A synthesis of policies is also provided with the particular objective of improving home-based workers’ well being. This study highlights the fact that time allocation factors convey qualitative information about individuals’ well-being that conventional well-being measures do not. The time and work intensity aspects (derived from time spent on overlapped work activities) affect individual well-being in many ways, including work productivity, mental health, and physical health. Overlapping activities are defined here as secondary and tertiary activities that are simultaneously performed with primary ones. Examples include cooking and listening to music or child minding while cleaning the house. An increase in the time spent on overlapping activities represents the way that individuals utilize their time more intensively. To the extent that overlapping activities can intensify work and affect a person’s level of stress, health, and discretionary leisure, the omission of overlapped activities leads to an inaccurate assessment of an individual’s well-being. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 160 Following the Asian economic crisis, many large and small factories in Thailand went out of business. Thousands of laid off workers gradually moved toward informal sector. Many became home-based workers, majority of them are in the informal sector having to accept low wage rates and working with no labor protection. For women workers, combining both paid market work and unpaid domestic work became a necessity, creating a higher incidence of time and work intensity. The theoretical framework for the well-being measure developed in this study build on the analytical framework on well-being by Floro (1995a), and extends to incorporate multi-person household . 1 Using the index construction methodology adopted from the Human Development Index (HDI), the well-being measure here consists of three components; personal income, the level of educational attainment, and the incidence of work intensity. These determinants directly and indirectly affect the constituents of individual well-being such as health and mental status, self esteem and social status, and the level of consumption. This dissertation also provides two modified time use data collection techniques: the simplified time use diary method and the circle of trust approach, which is more appropriate for the informal economy and developing countries than are traditional methodologies. Both approaches were developed to lower costs and capture overlapping activities while still maintaining the accuracy of the data collection. 1 However, this multi-person household well-being function is limited to one independent adult and one dependent household member. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 161 7.1 THAI HOME-BASED WORKER’S TIME ALLOCATION AND WELL BEING LEVELS Using our sub-sample data from the 2002 Bangkok Urban Poor Home-Based Worker Survey, we showed that the effects of overlapped activities on the pattern of time allocation between men and women are substantial. Omission of overlapping activities results in a serious underestimation of economic contributions of individuals, especially in non-market production. For example, time spent on unpaid household work by female 2 workers rose almost 1 0 0 percent when overlapping activity information was included. Time spent on childcare by women also increased almost 3 times when overlapping information was included. Our results also indicate that the time use patterns of our home-based worker respondents varied significantly by gender and employment. For instance, with overlapping activities included, men spent only half of the total time that women spent on household work, and self employed workers spend roughly 30 percent 3 more time on paid market work than contracted-workers. Further, the average work day of our survey respondents was 9-10 hours. Women also earned less than men for the same hours of work. Our time use survey and regression analyses produced interesting findings. First, an individual’s well-being is significantly and negatively related to gender. Female home-based workers in our sub-sample tended to have a lower quality of life than men. Further, decomposition of the individual well-being index found that Thai women are Note that primary and overlapping activities are given equal weight in this case. 3 Note that men also spent roughly half of the total time that women spent on household work, overlapped household work not included. However, without overlapped activities, self employed workers spent only 17 percent more time on paid market work than wage-contracted workers. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 162 worse off in all respects. They were money poor, time poor, and education poor. The difference in overall welfare between men and women informal sector workers could be partly explained by prevalent Thai social norms and customary views regarding the role and status of women. Women are expected to do most household chores even though they also needed to work in a paid market job. Women end up spending their time more intensively by working long days and multi-tasking. Their role and status also affected their level of education and earning, both directly and indirectly. Second, individuals in our sub-sample who resided in well organized and supportive communities had a higher well-being level. Third, social support received from various sources is positively related to the level of individual well-being. We also found that the well-being level of self employed workers was better than wage-contracted workers. One the most interesting findings of this study is that the inverse work intensity index statistically predicts subjective well-being of our respondents. It solves the puzzle that income alone cannot do, namely, explain significantly the individuals’ perspective of their well-being. What mattered for respondents’ subjective well-being 4 was not money per se, but earnings per hour. Income only brought happiness to respondents if they did not have to work incredibly long hours to earn it. The inverse work intensity index also substantially explains the difference in subjective well-being between men and women as suggested in Chapter 6 . These results indicate that the traditional methods of measuring well-being based on education and income omit crucial information, which time use (work intensity) data captures. This qualitative measure of 4 In other words, it was income and time spent to earn that amount of income that affect individuals’ perceptions of well-being. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 163 individual well-being captured by the information on overlapping activities in our time use factor of well-being as suggested by Juster and Stafford (1985,1991), Robinson and Godbey (1997), and Floro (1995a), was discussed in the review of literature. By including overlapped activities information in our time use survey, a more accurate image of individuals’ economic contribution and coping strategies was provided. Understanding how individuals organize their daily life can provide us a better assessment of the effects of economic and social policies on individual well-being. 7.2 POLICY IMPLICATION AND CONCLUDING REMARKS Information on home-based workers and the informal economy is especially important in the developing countries since the number of informal workers in some countries is even greater than the number in the formal sector. Despite their large economic contribution, their quality of life leaves much to be desired yet policymakers tend to neglect them. However, the Thai government has started to realize the informal sector’s contribution to the Thai economy. The Thailand Eighth National Development Plan actually referred directly to these workers. Nevertheless, specific laws and regulations that protect home-based workers’ rights and enhance their welfare have yet to be made. As evidenced by research in this dissertation, policymakers will likely continue to ignore an important aspect of their quality of life unless time and work intensity data analysis is included in policy formulations and evaluations. Unfortunately, the scarcity of relevant data reinforces the lack of understanding by Thai policymakers about the lives and work situations of these informal sector workers prove to be a real obstacle for policy formulation and implementation. Moreover, the lack of coordination among Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 164 policymakers, academics, related NGOs, and informal workers themselves has also slowed the progress towards developing adequate and effective labor and social policies. The well-being index developed in this dissertation benefits policymakers and the general public in the way that it allows for comparability. It enables policymakers to identify those who need the most help. The components of the well-being index also allow policymakers to identify the type of programs, policies, or public assistance needs to enhance well-being of these individuals. From the study of subjective well-being and well-being index components, the personal income and inverse work intensity component indices indicates that economic policies and development strategies that tend to increase low-wage jobs, e.g., low-wage export oriented development, will not raise the well-being of individuals, specifically poor workers. A gender-blind policy that raises individual income by means of increasing individuals’ market work hours will not do its job in enhancing well-being. A more effective policy or strategy to enhance quality of life is to raise incomes via higher wages, without having to increase market work hours. The results of our study raise serious doubts on any policies that assume any jobs are good jobs. All jobs are not the same, and only jobs with higher per hour wages have a significant positive well-being effect. The inverse work intensity component index also indicates that policymakers can enhance a person’s well-being, especially women’s, by reducing her work intensity via programs that provide child care services, health care services, and high-wage employment creation and decent working conditions by means of upholding and protecting workers’ rights. Furthermore, the educational attainment component index, which is found to be highly correlated to individuals’ subjective well being, suggests that providing universal primary education can effectively increase Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 165 human capabilities and well-being. This finding is also pointed out in other studies, including those of the World Bank (King and Hill 1993). Finally, this study demonstrates a strong need for time use data to be collected regularly since it can be utilized by policymakers, non-governmental organizations, and researchers to access the differential effect of economic and social policies, not just on income level or on market prices, but also on the manner by which people spend their time. This dissertation demonstrates that time use information is informative and theoretically coherent to convey individual well-being information. It adds to our understanding of well-being and gender differences in well-being by allowing comparisons between individuals’ well-being. At the micro level, a better understanding the factors that promote or lower well-being enables policymakers to target vulnerable individuals or households that need the most assistance. The well-being index developed in this dissertation is different from other well-being measurements such as Human Development Index and GDP per capita in the sense that it includes time use information. This study shows that time use information is a good predictor of individuals’ subjective well-being; money income alone does not explain the quality of life on experiences. Hence, any well-being measurement that does not include information on time use is likely to give an incomplete picture. The well-being index in this dissertation is also different from the subjective well-being indicator in the sense that it allow for comparability . 5 The absence of comparability in subjective well-being indicator is a limitation that sociologists and some economists admit. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 166 The well-being index can be incorporated into other economic models that take into account some proxy of well-being. For instance, Weicher (1999) and Ravallian (1996) argue that poverty should be measured not only based on income or consumption expenditure, but also on non-cash benefits or non-income determinant. Foster (1994) also emphasizes the need for a welfare-based approach for poverty measurement. Other decision making models like the household saving model by Engen et al. (1999) which relates saving decisions to life-cycle and individual welfare could also benefit from its inclusion .6 The incorporation of an individual well-being index in economic models allows one to explore a new dimension; however, this is beyond the scope of this dissertation . 7 The (time use related) well-being index developed in this dissertation is restricted to the individual level in theory and construction; nevertheless, it yields richer qualitative information than existing conventional well-being measurements. Due to the survey data limitations, it is beyond the scope of this study to examine the likely effect of any policy regime change on the well-being of individuals. Future research could pursue this interesting thread and allow for a richer analysis of policy and development strategy. The individual well-being index could be applied to the economics model for both dependent and independent variables. 7 This is also due to the limitation in our sub-sample data set. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 167 APPENDIX A MATHEMATICAL APPENDIX The first-order partial derivative results in equation (3.22) can be solved by using the Largrange multiplier method with a partial derivative on the conditioning well being maximization process. A positive effect of personal income to individual i ’s well being can be proved as shown in Equation (A.l), (A.2), and (A.3). 1 O' dW, ~dWi dC~ ' dWt QH~ ~dWt dS~ ~dWt ~dWt dHj . 1 + + + -f* (A.l) & [ac, Sy,\ [dH, dy,_ [as; dyt \ ay, _ dHj Sy, _ Further, signs of each set of partial derivative in Equation (A.l) have already l been solved in chapter 3, as follows: dWt dC, dWt dH, > 0; > 0: ac, ' Sy, _ [an, _ dWt as,] dWt a c ; >0; >0; (A.2) as, [acj dy, _ dWi dHj >0 dH] ' ay, Consequently, by substituting these signs in Equation (A.2) into Equation (A.1), the relationship of personal income and individual well-being of a working adult is found as shown in Equation (A.3). The intuitive explanations behind these results have already been extensively discussed in Chapters 2 and 3. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 168 dm -- ( + ) + (+) + (+) + (+) + (+) & (A.3) dW, *•>0 An increase in individual f s personal income increase dependent j ’s well-being, Wj, through dependent f s consumption, C ,, and health, H}. These will eventually 2 enhance well-being of the working individual i, W, . Therefore, an increase in individual f s income, yn directly affects working individual fs well-being via individual i’s constituents, C,, H, , and S’,; and indirectly through an increase in dependent member’s well-being, Wy. An effect of the level of educational attainment of working adult i, edu t, to individual well-being, Wt, can be solved as shown in Equation (A.4), (A.5), and (A. 6 ). dm ' dm dH> " ~QWt dm, dHj + as, ] + (A.4) dedU' [dH, dedUi [as, dedut dHj dedu , Signs of each derivative component in Equation (A.4) are shown in Equation (A.5). These results are based on explanation in the review of literature. dm dH. dm ds. dm dHj > 0; >0; >0 (A.5) dHi dedui dS, dedu. dHj dedu, As a consequence, Equation (A. 6 ) presents the effect of the level of educational attainment on individual well-being by substituting the Equation (A.5) into Equation (A.6 ). This is true since working adults are assumed to be altruist members in this particular model. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 169 dW, dedu,M+M+M+) (A.6 ) dW, ^->0 dedu, The positive relationship of individual z’s level of educational attainment, edui , and individual i’s well-being, W,, is found. This positive effect comes directly from an increase in constituents of individual / and indirectly from a rise in well-being of dependent member, Wy. A higher adults’ education enhance dependent member’s well being through a better nutrition intake, a better prevention of illness, etc. From Equation (3.13), the incidence of work intensity is a function of overlapping work activities on both market and domestic work, TV and L'ht . The first- order derivative of work intensity with respect to TV and I*, are positive. Therefore, the first-order derivative of individual well-being with respect to the incidence of work intensity is solved as shown in Equations (A.7), (A. 8 ), and (A.9). dW, ~dw, \ fSC, dk, dCt dk, dW, dH, dk, dH, dk, dk, _3C,'[ ISk, dtm,mi J, dH. ySk, dL>»u y d k,-d L lj (A.7) ~dW, fdCj dk, " + f { dk, d£ W i mi J Relationships of each derivative component in the Equation (A.7) are discussed in the review of literature and analytical framework. Signs of each component are shown in Equation (A. 8 ). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Consequently, substituting signs of each derivative component from Equation (A.8 ) into Equation (A.7) yields the relationship between the incidence of work intensity and well-being of working adult. d W i / w w \ n ( A ' 9 ) — - >or < 0 , dk, The first-order derivative of individual f s well-being with respect to the incidence of work intensity is ambiguous as shown in Equation (A.9). Work intensity is negatively related to well-being through both metal and physical health of individual. On the other hand, an income effect generated by higher work intensity enhance individual V well-being. Therefore, the effect of work intensity on working adult’s well-being can be positive or negative, depending on a size of the health effect and income effect on well being. If the health effect dominates the income effect, then the incidence of work intensity adversely affects well-being. In the case of home-based workers, combining long work days with high overlapping work activities, the adverse effect via mental and physical heath seems to be greater than the enhancement effect, increasing in work productivity and income. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 171 APPENDIX B SURVEY COMMUNITY DESCRIPTION B.I NAWAMIN COMMUNITY Nawamin community is also known as “Samakke Pattana” community (Unity Development community in English). This community is located in the northern part of Bangkok metropolitan area, and the size is around 4 acres (roughly 16,000 square meters). The land that this community located is a property of the “Buddhist Monk Hospital Foundation.” The people in this community have started to overrun into the Buddhist Monk Hospital Foundation’s property since 1989. After living illegally (without paying any rent) for 12 years, the Foundation’s representative and the community committees have finally reached to a rental agreement. A rental period was set to three years, started from December 2000 to November 2002, with an option to renew the rent agreement for another three years. Nawamin community accepted to pay a monthly rent in the total of 24,264 Baht, which is equivalent to $578. In 1996, NGOs, the Housing Cooperative, and the House Associations and Training Center for Urban Poor Development were in charge of grouping up community committees and a leader. In 1999, the committees were formally registered to the municipal office, Bangkok and a meeting is hosted by mapping a project team on a monthly basis to set up policies to serve up community developments. Since 1996, the committees and other organizations had continued their works on many Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 172 on-going campaigns such as requesting for water supply, constructing pavements, building a nursery, renting land, requesting for temporally house-licenses, and forming up saving and occupation groups. Nawamin community consists o f498 households, in total of 1,909 residents. Most of them have migrated from other regions of Thailand, especially from northeastern and northern part. In detail, 45 percent of migrated household are from northeastern part, 25 percent are from northern region, 15 percent are from central region, and 15 percent are from southern part of Thailand. Large numbers of the community members are Buddhist and a few of them are Muslim and Christian, respectively. B.2 NOMKLAO COMMUNITY Nomklao community is located in the property of the Crown Property Bureau and there are approximately 200 households at the present time . 1 The first group of intruders had migrated to this area since 1977. Before 1987, there were roughly 10 households. At the end of 1995, The Crown Property Bureau, the Community Development Office (Bangkok) and NGOs collaborated to provide basic infrastructures, children educations, and voting rights to this community. The community has organized a saving group (informal financial institution) for the purpose of buying land, which is supported by the Community Development Office. There are currently over 150 members who anticipate the group activities as well as habitually make some deposits to serve their aims under an arrangement of 18 1 The Crown Property Bureau is an organization that manages all King’s assets. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 173 community committees. Most of Nomklao community’s residents work as self-employed workers such as dressmakers, home-based grocery store owners, and street vendors. B.3 UDOMSUK COMMUNITY Only 130 households reside in this community. The majority of people here are home-based workers who assemble the upper part of shoes by using hand stitching with low compensation per pair. They are facing some troubles such as lower wages per piece, dishonest sub-contractors or factories, and healthiness problems. This community is fairly small so that everyone knows each other in a good relationship. Some of their residents have no access to the sewage system and proper housing. Also, due to the far distance from the main road, the access to the public is somehow limited. Nevertheless, the difficulty of finding permanent residence or having their own land manipulates these people to arrange a saving group and cooperative aiming for buying land, backing up by NGOs and related organizations. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 174 BIBLIOGRAPHY Arrow, K.J. (1962). “Economic Welfare and the Allocation of Resources for Inventions.” In RR. Nelson, ed., The Rate and Direction o f Inventive Activity: Economic and Social Factors. New Jersey: Princeton University Press. Agarwal, Bina (1997). “Bargaining and Gender Relations: Within and Beyond the Household.” FCND Discussion Paper No. 27, International Food Policy Research Institute, Food Consumption and Nutrition Division. Anand, Sudhir and Amartya Sen (1995). “Gender Inequality in Human Development: Theories and Measurement.” Background Paper for the Human Development Report 1995, UNDP. Anand, Sudhir and Christopher J. Harris (1994). “Choosing a Welfare Indicator”. The American Economics Review 84(2), 226-231. Andolfatto, David, Christopher Ferrall, and Paul Goome (2000). “Life Cycle Earning, Income and Wealth.” Discussion paper, Federal Reserve Bank o f Cleveland. Australian Bureau of Statistics (ABS) (1994). How Australians Use Their Time, Catalogue No. 4153.0, ABS, Canberra. Becker, Gary S. (1965). “A Theory of the Allocation of Time.” The Economic Journal 75(299), 493-517. . (1974). “On the Relevance of the New Economics of the Family (in Is Economic Theory with It?).” The American Economic Review 64(2), 317-319. Bajaj, Manjul (1999). “Invisible Worker, Visible Contributions: A Study of Homebased Women Workers in Five Sectors across South Asia.” Background Paper for Regional Policy Seminar, WIEGO. Biddle, Jeff E. and Daniel S. Hamermesh (1990). “Sleep and the Allocation of Time.” The Journal o f Political Economy 98(5 Parti), 922-943. Blau, Francine D., Marianne A. Ferber, and Anne E. Winkler (1998). The Economics of Women, Men, and Work. New Jersey: Prentice Hall. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 175 Campbell, Richard T. and John C. Henretta (1978). “Net Worth as an Aspect of Status.” American Journal o f Sociology 83,1204-1223. . (1980). “Status Claims and Status Attainment: The Determinants of Financial Well-Being.” American Journal o f Sociology 86(3), 618-629. Cantril, H. (1965). The Pattern o f Human Concerns. New Brunwick: Rutgers University Press. Charmes, Jacques, (2000). “Informal Sector, Poverty and Gender: a Review of Empirical Evidence.” Background Paper for the World Development Report 2001. Chen, Martha, Jennerfer Sebstad, and Leslie O’Connell, (1999). “Counting the Invisible Workforce: The Case of Homebased Workers.” World Development 27(3), 603- 610. Clark, A.E. (1997). “Job Satisfaction and Gender: Why are Women so Happy at Work.” Labor Economics 4(4), 341-372. Cohen, G. A. (1993). “Equality of What? On Welfare, Goods, and Capabilities.” In Martha Nussbaum and Amartya Sen, ed., The Quality o f Life, pp. 9-29. New York: Oxford University Press. Coleman, R. and L. Rainwater (1978). Social Standing in American. New York: Basic. Dasgupta, Partha (1993). An Inquiry into Welfare and Destitution. Oxford: Clarendon Press. . (1999). “Valuation and Evaluation: Measuring the Quality of Life and Evaluating Policy.” Background Paper prepared for the World Bank’s annual World Development Report 2000. World Bank, Washington, DC. . (2001). Human Well-Being and the Natural Environment. New York: Oxford University Press. Dasgupta, Partha. And Karl-Goran Maler (1999). “Net National Product and Social Well- Being.” Background Paper prepared for Symposium Issue on Net National Product in Environment and Development Economics. Diener, Ed and Eunkook M. Suh (2000). Culture and Subjective Well-being. Massachusetts: The MIT Press. Donohue, John J. and James Heckman (1991). “Continuous Versus Episodic Change: The Impact of Civil Rights Policy on the Economic Status of Blacks.” Journal o f Economic Literature 29,1603-1643. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 176 Economic and Social Commission for Asia and the Pacific: ESCAP (2001). Social Safety Net. Bangkok, Thailand: ESCAP. Engen, Eric M. et al. (1999). “The Adequacy of Household Saving.” Brookings Papers on Economic Activity 2,65-187. England, Paula (2003). “Separative and Soluble Selves: Dichotomous Thinking in Economics” In Julie. Nelson and Marrianne Ferber, ed., Feminist Economics Today. Chicago: University of Chicago Press. Ferrer-i Carbonell, Ada (2002a). “Income and Well-being: An Empirical Analysis of the Comparison Income Effect.” Discussion Paper No. TI2002-019/3, Tinbergen Institute, Amsterdam, Netherlands. . (2002b). “Subjective Questions to Measure Welfare and Well-Being: A survey.” Discussion Paper No. TI2002-020/3, Tinbergen Institute, Amsterdam, Netherlands. Ferrer-i Carbonell, Ada and Paul Frijters (2002). “How Important is Methodology for the Estimates of the Determinants of Happiness.” Discussion Paper No. TI2002- 024/3, Tinbergen Institute, Amsterdam, Netherlands. Ferrer-i Carbonell, Ada and B.M.S. Van Praag (2002). “The Subjective Costs of Health Losses due to Chronic Diseases.” Discussion Paper No. TI2002-023/3, Tinbergen Institute, Amsterdam, Netherlands. Floro, S. Maria (1995a). “Women’s Well-Being, Poverty, and Work Intensity.” Feminist Economics 1(3), 1-25. . (1995b). “Economic Restructuring, Gender and the Allocation of Time.” World Development 23(11), 1913-1929. . (1999). “Double Day/Double Shift.” In Janice Peterson and Margaret Lewis ed., The Elgar Companion to Feminist Economics. Cheltenham, UK: Edward Elgar. Floro, S. Maria and Maijorie Miles (2003). “Time Use, Work, and Overlapping Activities: Evidence from Australia.” Cambridge Journal o f Economics 27(8), 881-905. Floro, S. Maria and Thomas L. Hungerford (2001). “What We Can Learn From Time Intensity About Individual Well-Being.” Paper for the Levy Economics Institute conference on “What has happened to the Quality o f Life in America and Other Advanced Industrialized Countries. June 2001. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 177 Folbre, Nancy (1995). “Engendering Economics: New Perspectives on Women, Work, and Demographic Change.” In Michael Bruno and Boris Pleskovic ed., Annual World Bank Conference on Development Economics, pp. 127-66. World Bank, Washington, DC. Folbre, Nancy (1997). “A Time (Use Survey) For Every Purpose.” Paper presented at conference on Time Use, Non-Market Work, and Family Well-Being, November 1997. Washington, DC. Foster, James E. (1994). “Normative Measurement: Is Theory Relevant?” The American Economic Review 84(3), 365-370. Gibbard, A. (1996). “Interpersonal comparisons: Preference, good, and the intrinsic reward of a life”. In Alan P. Hamlin, ed., Ethics and Economics, Chapter 18., Cheltenham, UK: Edward Elgar Publishing Company. Graham, Carol and Stefano Pettinato (2002). Happiness and Hardship: Opportunity and Insecurity in New Market Economies. Washington, DC: Brookings Institution Press. Green, Francis (2002). “Work Intensification, Discretion and the Decline in Well-Being at Work.” Paper prepared for the Conference on Work Intensification, November 2002. Greene, William H. (2000). Econometric Analysis. New Jersey: Prentice-Hall. Griffin, James (1986). Well-Being: Its Meaning, Measurement, and Moral Importance. Oxford: Clarendon Press. Griliches, Zvi and William M. Mason (1972). “Education, Income, and Ability.” The Journal o f Political Economy 80(3), S74-S103. Grossbard-Shechtman, Amyra (1984). “A Theory of Allocation of Time in Markets for Labor and Marriage.” The Economic Journal 94(376), 863-882. Grounau, Reuben, (1976). “The Allocation of Time of Israeli Women.” The Journal of Political Economy 84(4), Part 2. ______. (1977). “Leisure, Home Production, and Work—the Theory of the Allocation of Time Revisited.” The Journal o f Political Economy 85(6), 1099-1124. Heckman, James, Anne Layne Farrar, and Petra Todd (1996). “Human Capital Pricing Equations with an Application to Estimating the Effect of Schooling Quality on Earnings.” The Review o f Economics and Statistics 78(4), 562-610. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 178 Hoelter, Jon W. (1986). “The Relationship Between Specific and Global Evaluations of Self: A Comparison of Several Models.” Social Psychology Quarterly 49(2), 129-141. Hubbard, R. Glennk, Jonathan Skinner, and Zeldes P. Stephen (1994). “Expanding the Life-Cycle Model: Precautionary Saving and Public Policy.” The American Economic Review 84(2), 174-179. International Labour Office: ILO (2002a). “Women and Men in the Informal Economy: A Statistical Picture.” The statistical report for the 9(fh Session of the International Labour Conference, Geneva, Switzerland. . (2002b). “Decent Work and the Informal Economy.” The background report for the 9(fh Session o f the International Labour Conference, Geneva, Switzerland. Jackson, Cecile and Richard Palmer-Jones (1998). “Work Intensity, Gender and Well- Being.” Discussion Paper No. DP 96. United Nations Research Institute for Social Development, Geneva, Switzerland. Juster, F. Thomas (1985). “Conceptual and Methodological Issues Involved in the Measurement of Time Use.” In F. Thomas Juster and Frank P. Stafford, ed., Time, Goods, and Well-Being , pp. 19-31. Ann Arbor: Institute for Social Research, University of Michigan. Juster, F.T. and Greg K. Dow (1985). “Goods, Time, and Well-being: The Joint Dependence Problem.” In F. Thomas Juster and Frank P. Stafford, ed., Time, Goods, and Well-Being , pp. 397-413. Ann Arbor: Institute for Social Research, University of Michigan. Juster, F.T. and Frank P. Stafford (1985). Time, Goods, and Well-Being. Ann Arbor: Institution for Social Research, University of Michigan. . (1991). “The Allocation of Time: Empirical Findings, Behavioral Models and Problems of Measurement.” Journal o f Economic Literature 24,471-522. Kalton, Graham (1985). “Sample Design Issues in Time Diary Studies.” In F. Thomas Juster and Frank P. Stafford, ed., Time, Goods, and Well-Being, pp. 93-112. Ann Arbor: Institute for Social Research, University of Michigan. King, Elizabeth M. and M. Anne Hill (1993). Women’s Education in Developing Countries: Barriers, Benefits, and Policies. Baltimore: The John Hopkins University Press. King, M. A. and L-D. L. Dicks-Mireaux (1982). “Asset Holdings and the Life-Cycle.” The Economic Journal 92(366), 247-267. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 179 Klein, Roger, Richard Spady, and Andrew Weiss (1991). “Factors Affecting the Output and Quit Propensities of Production Workers.” The Review o f Economic Studies 58(5), 929-953. Korrenam, Peter and Arie Kapteyn (1987). “A Disaggregated Analysis of the Allocation of Time within the Household.” The Journal o f Political Economy 95(2), 233- 249. Lauridsen, Laurids S. (1998). “The Financial Crisis in Thailand: Causes, Conduct and Consequences?” World Development 26(8), 1575-1591. Lazo, Lucita S. (1992). “Homeworkers in Thailand: Bane or Boon of Industrialization?” Homeworkers o f Southeast Asia: The struggle for Social Protection in Thailand. pp. 1-18. Bangkok, Thailand: International Labour Organization. . (2000). “Analyzing the Policy Implications of Time Use Data.” Framework Paper for ESCAP Workshop for High Level Policy Makers on Counting Policy Implications o f Time Use Survey Data. Bangkok, Thailand: ESCAP. Lavinas, Lena; Bila Sorj, Leila Linhares, and Angela Jorge (1999). “Home work in Brazil: New Contractual Arrangements.” SEED Working Paper No.7. International Labor Office, Geneva, Switzerland. Light, Audrey (1998). “Estimating the Returns to Schooling: When Does the Carrer Begin?” Economics o f Education Review 18, 291-309. . (2001). “In-School Work Experience and the Returns to Schooling.” Journal of Labor Economics 19(1), 65-93. Lucas, R.E. Jrs. (1988). “ON the Mechanics of Economic Development.” Journal o f Monetary Economics 22, 3-42. Manser, Marilyn and Murray Brown (1980). “Marriage and Household Decision-Making: A Bargaining Analysis.” International Economic Review 21(1), 31-44. Marshall, A. (1961). Principles o f Economics (9th ed., with annotations by C. W. Guillebaud). London: Macmillan. Marx, Karl (1990). Capital: Volume 1. New York: Penguin Books. (Original work published 1867) McElroy, Maqorie B. and Homey, Mary Jean (1990). “Nash-Bargained Household Decisions: Reply.” International Economic Review, Vol. 31, Issue 1, pp. 237-242. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 180 Merrett, Stephen (1966). “The Rate of Return to Education: A Critique.” Oxford Economic Papers 18(3), 289-303. Meyer, David E. and David E. Kieras (1997). “A Computational Theory of Executive Cognitive Processes and Multiple-task Performance: Part2. Accounts of Psychological Refractory-period Phenomena” Psychological Review 104, 749- 791. Midanik, Lorraine T., Arthur L. Klatsky, and Mary Anne Armstrong (1990). “Changes in Drinking Behavior: Demographic, Psychosocial, and Biomedical Factors.” International Journal o f the Addictions 25,599-619. Miller, Irwin and Miller Marylees (1999). John E. Freund’s Mathematical Statistics (Sixth Edition). New Jersey: Prentice Hall. Mincer, Jacob (1958). “Investment in Human Capital and Personal Income Distribution.” Journal of Political Economy 66(4), 281-302. . (1962). “Labor Force Participation of Married Women in Aspects of labor economics.” National Bureau Conference Series, No. 14, 63-105. NBER, Cambridge, MA. . (1974). Schooling, Experience and Earnings. New York: Columbia University Press. Modigliani, Franco (1986). “Life Cycle, Individual Thrift, and the Wealth of Nations.” The American Economic Review 76(3), 297-313. Morgan, James and Martin David (1963). “Education and Income.” The Quarterly Journal o f Economics 77(3), 423-437. Oswald, A.J. (1997). “Happiness and Economic Performance.” The Economic Journal 107(445), 1815-1831. Owens, Timothy J. (1994). “Two Dimensions of Self-Esteem: Reciprocal Effects of Positive Self-Worth and Self-Deprecation on Adolescent Problems.” American Sociological Review 59(3), 391-407. Owens, Timothy J., Jeylan T. Mortimer, Michael D. Finch (1996). “Self-Determination as a Source of Self-Esteem in Adolescence.” Social Forces 74(4), 1377-1404. Paetkau, Mark (1999). “Micro and Small Enterprise Development & Poverty Alleviation in Thailand.” In Gerry Finnergan ed., ILO/UNDP Working Paper 4: Financial Support For Micro and Small Enterprises (MSEs) in Thailand, THA/99/003. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 181 Pigou, A. C. (1952). The Economics o f Welfare (4th Ed.). London: Macmillan. Pradhan, M. and Ravllion M. (2000). “Measuring Poverty Using Qualitative Perceptions of Consumption Adequacy.” Review o f Economics and Statistics 82(3), 462-471. Poliak, A. Robert and Michael L. Wachter (1975). “The Relevance of the Household Production Function and Its Implications for the Allocation of Time.” The Journal o f Political Economy 83(2), 255-278. Pongsapich, Amara (1992). “An Exploratory Survey on Rural Women Homeworkers in Thailand.” Homeworkers o f Southeast Asia: The struggle for Social Protection in Thailand, pp. 19-113. Bangkok, Thailand: International Labour Organization. Pramualratana, Anthony, Napapom Havanon, and John Knodel (1985). “Exploring the Normative Basis for Age at Marriage in Thailand: An Example from Focus Group Research.” Journal o f Marriage and the Family 47(1), 203-210. Praparpun, Yada, Rattana Tosakul Boonmathya, and Rakawin Leechanavanichpan (1999). “The Situation of Women Subcontracted Workers in the Garment Industry in Bangkok Thailand.” A Research Project o f Women’s Economic and Legal Right (WLER) Program, The Asia foundation, San Francisco, CA. Pryer, J. (1989). “When Breadwinners Fall 111: Preliminary Findings from a Case Study in Bangladesh.” IDS Buletin 20(2), 49-57. Pugliesi, Karen (1995). “Work and Well-Being: Gender Differences in the Psychological Consequences of employment.” Journal o f Health and Social Behavior 36(1), 57- 71. Ravallion, Martin (1996). “Issues in Measuring and Modeling Poverty.” The Economic Journal 106(438), 1328-1343. Robinson, John P. (1985). “The Validity and Reliability of Diaries versus Alternative Time Use Measures.” In F. Thomas Juster and Frank P. Stafford, ed., Time, Goods, and Well-Being, pp. 33-62. Ann Arbor: Institute for Social Research, University of Michigan. Robinson, John P. and Geoffrey Godbey (1997). Time for Life: The Surprising Ways American Use their time. Pennsylvania: Pennsylvania State University Press. Roldan, Martha (1985). “Industrial Outworking, Struggles for the Reproduction of Working Class Families and Gender Subordination.” Beyond Employment: Household, Gender and Subsistence. Oxford: Basil Blackwell. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 182 Romer, P.M. (1986). “Increasing Returns and Long-Run Growth.” Journal o f Political Economy 94,1002-1010. Rosenberg, Morris (1985). “Self-Concept and Psychological Well-Being in Adolescence.” The Development o f Self Book Chapter, pp. 2Q5-246. Florida: Academic Press. Rosenberg, Morris, Schooler Carmi, Carrie Schoenbach, and Florence Rosenberg (1995). “Global Self-Esteem and Specific Self-Esteem: Different Concepts, Different Outcomes.” American Sociological Review 60(1), 141-156. Ross, Catherine E. and Chia-ling Wu (1995). “The Links between Education and Health.” American Sociological Review 60(5), 719-745. . (1996). “Education, Age, and the Cumulative Advantage in Health.” Journal o f Health and Social Behavior 37(1), 104-120. Ross, Catherine E. and Marieke Van Willigen (1997). “Education and the Subjective Quality of Life.” Journal o f Health and Social Behavior 38(3), 275-297. Ruthruff, Eric, Harold E. Pashler, and Alwin Klaassen (2001). “Processing Bottlenecks in Dual-Task Performance: Structural Limitation or Strategic Postponement.” Psychonomic Bulletin and Review 8(2), 73-80. Samuelson, Paul A. “Social Indifference Cruves.” The Quarterly Journal o f Economics 70(1), 1-22. Schaefer, Kendall K. (2001). Macroeconomic Implications o f an Urban Informal Sector: A Theoretical Model and a South African Case Study. Dissertation, American University, Washington, DC. Sen, Amartya (1985). Commodities and Capabilities. Amsterdam: North-Holland. . (1992). Inequality Reexamined. Cambridge: Harvard University Press. . (1993). “Capability and Well-Being.” In Martha Nussbaum and Amartya Sen, ed., The Quality o f Life, pp. 30-53. New York: Oxford University Press. Shea, Steven, Aryeh D. Stein, Charles E. Basch, Rafael Lantingue, Christopher Maylahn, David S Strogatz, and Lloyed Novick (1991). “Independent Associations of Educational Attainment and Ethnicity with Behaviral Risk Factors for Cardiovascular Disease.” American Journal o f Epidemiology 134, 367-382. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 183 Smith, Adam (1993). An inquiry into the Nature and Causes o f the Wealth o f Nations: A Selected Edition . New York: Oxford University Press. (Original work published 1776) Slottje, Daniel J. (1991). “All Things Reconsidered: The Life-Cycle Perspective and the Third Task of Economic History.” The Journal o f Economics History 51(2), 271- 288. Sutch, Richard (1991). “Measuring the Quality of Life across Countries.” The Review of Economics and Statistics 73(4), 684-693. Syme, Leonard S and Lisa F. Berkman (1986). “Social Class, Susceptibility, and Sickness.” The Sociology o f Health and Illness 73(4), 684-693. Thoits, Peggy A. (1986). “Multiple Identities: Examining Gender and Marital Status Differences in Distress.” American Sociology Review 51,259-272. Thailand, HomeNet. (1999). Organizing Homeworkers: Three Case Studies o f Women in the Garment Sector. HomeNet, Bangkok, Thailand. Thailand, National Commission on Women’s Affairs, Office of the Prime Minister. (1995a). Thailand’s Report on the Status o f Women and Platform for Action. NCWA, Bangkok, Thailand. . (1995b). “Women of Thailand.” A special issue of the newsletter o f the National Commission on Women’s Affairs. NCWA, Bangkok, Thailand. Thailand, National Economic and Social Development Board: NESDB (2003a). “Poverty Indicator and the analysis of poor people.” Viewed on December 7,2003 http://www. nesdb.go. th/Mainmenu/Macro/Devindex_data/dev6page. html . (2003b). “Thailand Labor Force, Employment, and Unemployment statistics: 1998-2003” Viewed on December 14,2003 http://www.nesdb.go.th/dataJndex/social/data/1/dataO IJ3/laborJll.xls . (2003c). “Quarterly Gross Domestic Product” Viewed on December 14,2003 http://www.nesdb.go.th/Main_menu/Macro/GDP/menu.html Thailand, National Statistical Office (1999). “The first National Thai Homework Survey.” Viewed on December 12,2002 http://www. nso.go. th/eng/stat/homewk/homewk htm Thailand, National Statistical Office (2000). “Average Household Income By Region: 2000.” Viewed on November 26,2003 < http://www. nso.go. th/eng/stat/subject/subject. htm#catal 1> Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 184 Thailand National Statistical Office (2001). “Household Socio-Economic Survey.” Viewed on May 26,2003 < http://www. nso.go. th/eng/stat/socio/socio. htm> United Nations Development Programme (UNDP) (2001). Human Development Report 2001. New York: Oxford University Press. United Nations Development Programme (UNDP) (2002). The Arab Human Development Report 2002: Creating Opportunities for Future Generations. Kingdom of Jordan: Hashemite. United Nations Development Programme (UNDP) (2003). Human Development Report 2003. New York: Oxford University Press. Van Praag, B.M.S. (1991). “Ordinal and cardinal utility: an integration of the two dimensions of the welfare concept.” Journal of Econometrics, pp. 28-34. New York: St. Martin’s Press. Van Praag, B.M.S., Pual Frijters, and Ada Ferrer-i Carbonell (2002). “The Anatomy of Subjective Well-Being.” Discussion paper No. TI2002-022/3. Tinbergen Institute, Amsterdam, Netherlands. Varian, Hal R. (1992). Microeconomic Analysis. New York: W.W. Norton & Company. Weicher, John C. (1999). “Some Income-Measurement Issue and Their Policy Implications.” The American Economic Review. 89(2), 29-33. Weiss, Andrew. (1995). “Human Capital vs. Signaling Explanations of Wages.” The Journal o f Economic Perspectives 9(4), 133-154. Wilson, W. (1967). “Correlates of Avowed Happiness.” Psychological Bulletin 67, 294- 306. Winkleby, Marilyn A., Darius E. Jatulis, Erica Frank, and Stephen P. Fortmann (1992). “Socio-Economic Status and Health: How Education, Income, and Occupation Contribute to Risk Factors for Cardiovascular Disease.” American Journal o f Public Health 82, 816-820. Wolfe, Barbara and Robert Haveman (1983). “Time Allocation, Market Work, and Changes in Female Health.” American Economic Review 73(2), 134-139. Wolff, Edward N and Ajit Zacharias (2003). “The Levy Institute Measure of Economic Well-Being.” Working paper No. 372. The Levy Economics Institute, Annandale- on-Hudson, New York. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 185 Yimyam, Susanha et al. (2001). “Overview of the Thailand informal worker.” The Journal ofHomeNet Thailand 2,12-15. Yoddumnem-Attig, Bencha et al. (1992). Changing roles and statuses o f women in Thailand: a documentary assessment. Institute of Population and Social Research, Mahidol University, Nakhonpathom, Thailand. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.