Trade Liberalization, Social Policies and Health: A Theoretical and Empirical Exploration

Download as pdf or txt
Download as pdf or txt
You are on page 1of 340

Trade Liberalization, Social Policies and Health:

A theoretical and empirical exploration

Courtney McNamara

Doctor of Philosophy

University of York

Social Policy and Social Work

August 2014
ABSTRACT

This thesis theoretically and empirically explores how trade liberalization and
social protection policies interact to influence health. It is conducted from a political
economy of health perspective. A systematic review of the literature finds that trade
liberalization is often conceptualized in public health work in problematic ways.
Because the health impacts of trade liberalization are especially under-
conceptualized in the context of labour markets, this thesis empirically explores the
2005 phase-out of the Multi-Fibre Arrangement (MFA). This phase-out abruptly
liberalized the textile and clothing (T&C) sector and triggered substantial shifts in
employment across a wide range countries. Data from 32 T&C reliant countries were
analysed in reference to the pre- and post-MFA periods of 2000-2004 and 2005-2009.
Fuzzy-set qualitative comparative analysis (fsQCA) was used to examine the
association between: a) T&C employment changes, b) countries’ level of labour
market and welfare state protections, and c) changes in adult female and infant
mortality rates. FsQCA is a cross-case method which enables logical statements to be
made about the combinations of causal conditions that are sufficient for an outcome.
Process tracing was used to further investigate these fsQCA results through in-depth
case work. Findings suggest that the worsening of adult female mortality rates is
related to T&C workers’ lack of social protection in the context of trade
liberalization. Overall, it is found that social protection is often inaccessible to the
type of workers who may be the most vulnerable to processes of liberalization, and
that many workers are particularly vulnerable due to the structure of social protection
policies. Social protection policies can therefore both moderate pathways to health
and influence the type of health-related pathways resulting from trade liberalizing
policies. This work contributes to our understanding of the complex pathways
between trade liberalization and health and sets the stage for further investigations.

2
TABLE OF CONTENTS
ABSTRACT .........................................................................................................................................2

LIST OF FIGURES .............................................................................................................................7

LIST OF TABLES...............................................................................................................................8

LIST OF ACRONYMS........................................................................................................................9

ACKNOWLEDGEMENTS .............................................................................................................. 11

AUTHOR’S DECLARATION ......................................................................................................... 12

CHAPTER 1 INTRODUCTION TO THE THESIS ..................................................................... 13

1.1 Background ......................................................................................................... 13


1.1.1 Research Objectives and Questions ..................................................................... 14
1.2 Methodology ....................................................................................................... 15
1.2.1 Political Economy of Health Perspective ............................................................. 15
1.2.2 Research Design .................................................................................................. 17
1.2.3 Analytical Frameworks and Methods .................................................................. 18
1.3 Contributions to Knowledge ............................................................................... 19
1.4 Organization of Thesis........................................................................................ 20

INTRODUCTION TO PART 1 ..................................................................................................... 21

CHAPTER 2 LITERATURE REVIEW ......................................................................................... 22

2.1 Introduction ........................................................................................................ 22


2.2 Methods: ............................................................................................................. 23
2.2.1 Inclusion Criteria ................................................................................................. 23
2.2.2 Exclusion Criteria ................................................................................................ 23
2.2.3 Searching for Studies ........................................................................................... 23
2.2.4 Selection of Studies ............................................................................................. 24
2.2.5 Preliminary Search and Refined Search Strategies .............................................. 24
2.2.6 Defining Conceptual Frameworks ....................................................................... 26
2.2.7 Data Extraction .................................................................................................... 28
2.2.8 Critical Appraisal and Narrative Synthesis .......................................................... 30
2.3 Results ................................................................................................................ 30
2.3.1 How is Trade Liberalization Understood in Analytical Frameworks
Relating Trade to Health ...................................................................................... 33
2.3.2 How is Health Conceptualized in Analytical Frameworks Relating Trade
to Health .............................................................................................................. 36
2.3.3 How Do Researchers Theorize the Mechanisms and Pathways Mediating
the Trade Liberalization and Health Relationship? .............................................. 39
2.4 Preliminary Summary of Literature Review....................................................... 88

CHAPTER 3 MOVING THE RESEARCH AGENDA FORWARD ............................................ 90

3.1 Towards a Common Definition of Trade Liberalization .................................... 90


3.1.1 Definitional Insights from Development Economics ........................................... 93
3.1.2 Defining Trade Liberalization for Public Health .................................................. 95

3
3.2 Considering Health Conceptualizations ............................................................. 95
3.2.1 Overall Conceptualizations .................................................................................. 95
3.2.2 Health Inequalities Versus Health Inequities ....................................................... 96
3.2.3 Framing Health .................................................................................................... 96
3.3 Developing our Understanding of the Links Between Trade
Liberalization and Health ................................................................................... 97
3.3.1 Early and Influential Frameworks: Setting the Stage ........................................... 97
3.3.2 Across the Contexts: Strengths, Weaknesses, Overlap and Gaps ......................... 97
3.3.3 General Considerations ...................................................................................... 101
3.4 Towards a Conceptual Framework ................................................................... 103
3.4.1 Sociopolitical Theories ...................................................................................... 105
3.4.2 Social Production of Disease/Political Economy of Health ............................... 105
3.4.3 Social Determinants of Health, Population Health, and Fundamental
Causes................................................................................................................ 106
3.4.4 Latin American Social Medicine/Collective Health ........................................... 107
3.4.5 Health and Human Rights .................................................................................. 107
3.4.6 Psychosocial Theory .......................................................................................... 107
3.4.7 Ecosocial Theory ............................................................................................... 108
3.5 Trade Liberalization and Health: An Ecosocial Framework ............................ 111
3.5.1 Framework Limitations...................................................................................... 116
3.6 Summary of Literature Review & Progress Towards Research Objective
1 ........................................................................................................................ 117

INTRODUCTION TO PART 2 .................................................................................................. 120

CHAPTER 4 THE TEXTILE AND CLOTHING SECTOR AND THE MFA PHASE-
OUT ............................................................................................................................................... 122

4.1 Introduction ...................................................................................................... 122


4.2 The T&C Supply Chain .................................................................................... 122
4.3 Trade Preference Schemes and Regional Trade Agreements ........................... 123
4.4 Protection and Liberalization of the Sector ...................................................... 125
4.5 Employment Impacts of the MFA Phase-Out .................................................. 126
4.6 A Framework for Exploring the Health Impacts of the MFA Phase-Out ........ 130

CHAPTER 5 METHODS: AN EMPRICAL CROSS-CASE ANALYSIS.................................. 133

5.1 Introduction ...................................................................................................... 133


5.2 Methodological considerations ......................................................................... 133
5.3 Fuzzy-set Qualitative Comparative Analysis ................................................... 134
5.4 Operationalizing fsQCA ................................................................................... 136
5.4.1 Stage 1: Outcome Indicators, Case Selection and Causal Conditions ................ 137
5.4.2 Stage 2: Examining Necessity and Sufficiency .................................................. 175
5.4.3 Stage 3: Logical Reduction ................................................................................ 177

CHAPTER 6 RESULTS AND DISCUSSION ............................................................................ 179

6.1 Introduction ...................................................................................................... 179


6.2 Analyses of Necessity ....................................................................................... 180
6.3 Analyses of Sufficiency .................................................................................... 181
6.3.1 General Inspection of the Truth Table ............................................................... 184
6.3.2 Overall Truth Table Results ............................................................................... 186
6.4 Logical Reduction............................................................................................. 187
4
6.4.1 Solution Paths for Adult Female Mortality ........................................................ 188
6.4.2 Solution Paths for Infant Mortality Rates .......................................................... 189
6.5 Discussion ......................................................................................................... 189
6.6 Summary ........................................................................................................... 193

INTRODUCTION TO PART 3 .................................................................................................. 194

CHAPTER 7 METHODS: AN EMPIRICAL WITHIN-CASE ANALYSIS ............................ 196

7.1 Introduction ...................................................................................................... 196


7.2 Process Tracing Overview ................................................................................ 196
7.3 Defining Process Tracing and Causal Mechanisms ......................................... 198
7.4 Steps in Theory-Building Process Tracing ....................................................... 199
7.4.1 Case Selection.................................................................................................... 200
7.4.2 Collection of Empirical Material and Building Causal Mechanisms ................. 205

CHAPTER 8 CASE STUDIES AND THE SEARCH FOR CAUSAL MECHANISMS............ 207

8.1 Introduction ...................................................................................................... 207


8.2 Solution Path 1: Brazil ...................................................................................... 207
8.2.1 Introduction ....................................................................................................... 207
8.2.2 Collection of Empirical Material and Causal Mechanism .................................. 208
8.2.3 Findings ............................................................................................................. 209
8.2.4 Summary ........................................................................................................... 211
8.3 Solution Path 2: Italy, the Slovak Republic, Korea and Portugal ..................... 211
8.3.1 Introduction ....................................................................................................... 211
8.3.2 Collection of Empirical Material and Causal Mechanism .................................. 212
8.3.3 Findings ............................................................................................................. 214
8.3.4 Summary ........................................................................................................... 224
8.4 Solution Path 3: Bangladesh ............................................................................. 226
8.4.1 Introduction ....................................................................................................... 226
8.4.2 Collection of Empirical Material and Causal Mechanism .................................. 227
8.4.3 Findings ............................................................................................................. 228
8.4.4 Summary ........................................................................................................... 233
8.5 Solution path 4: The Kyrgyz Republic ............................................................. 235
8.5.1 Introduction ....................................................................................................... 235
8.5.2 Collection of Empirical Material and the Causal Mechanism ............................ 236
8.5.3 Findings ............................................................................................................. 238
8.5.4 Summary ........................................................................................................... 244
8.6 Solution Path 5: Korea ...................................................................................... 244
8.6.1 Introduction ....................................................................................................... 244
8.6.2 Collection of Empirical Material and the Causal Mechanism ............................ 245
8.6.3 Summary ........................................................................................................... 247
8.7 Solution path 6: China and Thailand ................................................................ 247
8.7.1 Introduction ....................................................................................................... 247
8.7.2 Collection of Empirical Material and the Causal Mechanism ............................ 248
8.7.3 Findings ............................................................................................................. 249
8.7.4 Summary ........................................................................................................... 256
8.8 Solution path 7: Sri Lanka, Bangladesh and Indonesia .................................... 257
8.8.1 Introduction ....................................................................................................... 257
8.8.2 Collection of Empirical Material and the Causal Mechanism ............................ 258
8.8.3 Findings ............................................................................................................. 260
8.8.4 Summary ........................................................................................................... 272

5
CHAPTER 9 DISCUSSION......................................................................................................... 275

9.1 Introduction ...................................................................................................... 275


9.2 Summary of Causal Mechanisms ..................................................................... 275
9.3 Limitations of Results:...................................................................................... 278
9.3.1 Data Availability ................................................................................................ 278
9.3.2 Precision of Causal Mechanisms ....................................................................... 278
9.3.3 Generalizability of Causal Mechanisms ............................................................. 279
9.4 Broader Insights ................................................................................................ 281
9.5 Combining fsQCA with Process Tracing ......................................................... 282

CHAPTER 10 OVERALL FINDINGS, CONTRIBUTIONS AND NEXT STEPS ................. 285

10.1 Introduction ...................................................................................................... 285


10.2 What are the Pathways Between Trade Liberalization and Health:
Insights from the Literature Review ................................................................. 285
10.3 Moving from the Theoretical to the Empirical ................................................. 291
10.4 The Phase-Out of the Multi-Fibre Arrangement and its Impact on Health
Outcomes in Countries Reliant on the Textile and Clothing Sector ................. 292
10.5 Moving from the Empirical back to the Theoretical ........................................ 295
10.6 Next Steps ......................................................................................................... 297

BIBLIOGRAPHY ......................................................................................................................... 299

6
LIST OF FIGURES
Figure 1 The continuum of frameworks, theories, and models (Carpiano and Daley 2006) ................ 26
Figure 2 Example of the connection between a framework, a theory, and a conceptual model
(Carpiano and Daley 2006) .................................................................................................. 27
Figure 3 Example of the difficulty in distinguishing between frameworks, theories, and conceptual
models ................................................................................................................................. 28
Figure 4 Overview of Study Identification Process ............................................................................ 32
Figure 5 Globalization and Health: a framework for analysis and action (Woodward et al. 2001) ...... 41
Figure 6 Globalization and Health: a framework for analysis and action (in detail)
(Woodward et al. 2001) ....................................................................................................... 42
Figure 7 Framework for Analyzing the Links Between Globalization and Health
(Labonté and Togerson 2003) .............................................................................................. 44
Figure 8 The health impacts of globalisation: a conceptual framework (Huynen et al. 2005) ............. 49
Figure 9 The health impacts of globalisation: a conceptual framework (in detail) (Huynen et al. 2005)50
Figure 10 Towards Health-Equitable Globalisation: Rights, Regulation and Redistribution
(Labonté et al. 2007)............................................................................................................ 51
Figure 11 Towards Health-Equitable Globalisation: Rights, Regulation and Redistribution
(Labonté et al. 2007)............................................................................................................ 52
Figure 12 Trade and social determinants of health (Blouin et al. 2009) .............................................. 54
Figure 13 Potential Impact of Adjustment Policies on Vulnerability of Women and Children to
HIV/AIDS in Sub-Saharan Africa (De Vogli & Birbeck 2005) .......................................... 75
Figure 14 Potential Impact of Adjustment Policies on Vulnerability of Women and Children to
HIV/AIDS in Sub-Saharan Africa (De Vogli & Birbeck 2005) .......................................... 76
Figure 15 Macro-level framework and policy entry points (Muntaner et al. 2010) ............................. 78
Figure 16 Micro-level framework and policy entry points (Muntaner et al. 2010) .............................. 79
Figure 17 Example of usefulness in recognizing overlap between contexts ...................................... 100
Figure 18 Ecosocial Theory (Krieger 2008) ..................................................................................... 112
Figure 19 Trade Liberalization and Health: An Ecosocial Perspective ............................................. 114
Figure 20 Employment changes after the MFA phase-out (Lopez-Acevedo & Robertson, 2012)...... 129
Figure 21 Macro-level framework and policy entry points (Muntaner et al. 2010) ........................... 131
Figure 22 Micro-level framework and policy entry points (Muntaner et al. 2010) ............................ 132
Figure 23 HDI fuzzy-set scores versus adult female mortality rates ................................................. 161
Figure 24 GNI per capita fuzzy-set scores versus adult female mortality rates ................................. 161
Figure 25 HDI fuzzy-set scores versus IMR .................................................................................... 162
Figure 26 GNI per capita fuzzy-set scores versus IMR .................................................................... 162
Figure 27 Case-Selection after a fsQCA from Schneider and Rohlfing (2013) ................................. 201
Figure 28 Solution Path 1 XY plot .................................................................................................. 208
Figure 29 Solution Path 2 XY plot .................................................................................................. 212
Figure 30 Solution Path 3 XY plot .................................................................................................. 226
Figure 31 Solution Path 4 XY plot .................................................................................................. 236
Figure 32 Solution Path 5 XY plot .................................................................................................. 245
Figure 33 Solution Path 6 XY plot .................................................................................................. 248
Figure 34 Solution Path 7 XY plot .................................................................................................. 258

7
LIST OF TABLES
Table 1 T&C Employment as a percentage of total manufacturing employment .............................. 144
Table 2 Final list of included countries and the type of mortality change investigated...................... 145
Table 3 Absolute Changes in Adult Female Mortality Rates ............................................................ 149
Table 4 Relative Changes in Adult Female Mortality Rates ............................................................. 150
Table 5 Absolute changes in infant mortality rates .......................................................................... 153
Table 6 Relative changes in infant mortality rates............................................................................ 154
Table 7 Fuzzy membership scores for the outcome sets of Health Improving and Health Worsening
adult female and infant mortality rates ............................................................................... 159
Table 8 Fuzzy set scores using the ‘estimated-averaged’ HDI rankings from 2004-2009 ................. 164
Table 9 Fuzzy-set membership scores in the set of Highly Developed Countries ............................. 165
Table 10 Fuzzy membership scores in the set of Protective Labour Market Policies ........................ 169
Table 11 Fuzzy-set membership scores in the set of Protective Welfare State Policies ..................... 172
Table 12 Fuzzy membership scores in the sets of Employment Growth and Employment Loss ....... 174
Table 13 Tests of necessity, consistency and coverage scores .......................................................... 180
Table 14 Tests of Sufficiency Summary Table ................................................................................ 182
Table 15 FsQCA truthtable results .................................................................................................. 184
Table 16 FsQCA Logical Reduction Results ................................................................................... 188

8
LIST OF ACRONYMS

ASSIA Applied Social Science Index and Abstracts


AOA Agreement on Agriculture
AGOA Africa Growth and Opportunity Act
CSDOH Commission on the Social Determinants of Health
DALE Disability-Adjusted Life Expectancy
EBA Everything but Arms
EMCONET Employment Conditions Knowledge Network
EPF The Employees’ Provident Fund
EPZ Export Processing Zone
ETF Employees’ Trust Fund
HALE Health-Adjusted Life Expectancy
HDI Human Development Index
ILO International Labor Organization
ITS Interrupted Time Series Analysis
IMF International Monetary Fund
IMR Infant Mortality Rates
FDI Foreign Direct Investment
fsQCA Fuzzy-set Qualitative Comparative Analysis
FTZ Free Trade Zone
GATT General Agreement on Tariffs and Trade
GATS General Agreement on Trade in Services
GDP Gross Domestic Product
GKN Globalization Knowledge Network
GNI Gross National Income
GPG Global Public Good
GSP General System of Preferences
LTA Long Term Agreement Regarding International Trade in
Cotton Textiles
MFA Multi-Fibre Arrangement
NBER National Bureau of Economic Research
OECD Organization for Economic Co-operation and Development
PRSPs Poverty Reduction Strategy Papers
RO Research Objective
RQ Research Question

9
SAPs Structural Adjustment Policies
SDOH Social Determinants of Health
SPI Social Protection Index
SPS Agreement on Sanitary and Phytosanitary Measures
STA Short Term Arrangement Regarding International Trade in
Cotton
T&C Textile and Clothing
TBT Agreement on Technical Barriers to Trade
TNC Transnational Corporation
TPP Trans-Pacific Partnership Agreement
TRIMS The Agreement on Trade Related Investment Measures
TRIPS Trade-Related Intellectual Property Rights Agreement
UDHR Universal Declaration of Human Rights
UN United Nations
UNIDO United Nations Industrial Development Organization
WOK Web of Knowledge
WHO World Health Organization
WTO World Trade Organization

10
ACKNOWLEDGEMENTS

With greatest thanks to my supervisor, Dr. Chris Holden, for always letting
me take my own path and extending invaluable advice, support and encouragement
along the way. Many thanks also to Dr. John Hudson for his generous input and
insightful questions. I’d also wish to thank Dr. Stefan Kühner for his helpful
feedback at the upgrade stage.

Portions of this work were presented at the 2nd Interest Meeting of the
European Society for Health and Medical Sociology and at seminars within the
Department of Social Policy and Social Work at the University of York. I am
grateful to participants at these events for stimulating discussions. I am especially
thankful to Professor Espen Dahl for comments on some of the empirical work of
this thesis.

Special thanks go to my family who have always endured and supported my


work, despite it taking me to increasingly distant corners of the world. Heartfelt
thanks are also due to an incredible group of friends who have always been there for
me, regardless of where there is.

Finally, the deepest of gratitude to Massimo. Thank you for never losing your
sense of humour, even when I’d lost mine. You’re the best.

11
AUTHOR’S DECLARATION

I declare that, except where explicit reference is made to the contribution of


others, this thesis is the result of my own original research and has not been
submitted for any other degree at the University of York or any other institution.

12
CHAPTER 1 INTRODUCTION TO THE THESIS
1.1 Background
The health impacts of trade liberalization have begun to receive more
attention in contemporary public health scholarship. However, this research is largely
focused on the direct impacts of healthcare privatization, changing dietary patterns,
tobacco use, alcohol consumption, and access to medicines. Receiving relatively
little consideration are the impacts of trade liberalization on the social determinants
of health (Blouin et al., 2009).

Generally speaking, the social determinants of health (SDOH) constitute the


resources by which people can control the circumstances of their life. They include
things like income, education and employment. Encompassing the reduction of
tariffs, quotas, and other barriers to trade, trade liberalization can directly affect the
distribution of these and other important SDOH (Blouin et al., 2009).

However, social policies also contribute to the resources available to people,


directly through transfers and services, and indirectly through policies which affect
people's opportunities to generate resources in the market (Lundberg, 2008).
Therefore, the extent to which trade liberalization impacts SDOH and health will
depend not just on the characteristics of trade policies but also on the characteristics
of states’ social policies.

A growing literature of comparative social epidemiological studies almost


uniformly demonstrates that health outcomes vary significantly across different
welfare state arrangements. The ‘welfare state’ is a term often used in public health
as shorthand to describe the characteristics of a state’s social policy (Bambra &
Eikemo, 2009). However, as Navarro and Muntaner (2004) point out, these studies
often assume that countries have complete control over the decisions that impact
broad determinants of health, such as levels of employment and income inequality.
Such research ignores the role of trade policy in limiting countries' policy space
surrounding economic, political and social decisions.

While a number of studies have explored the health impacts of trade


liberalization conceptually, few have empirically linked liberalizing processes to
determinants of health and health outcomes. Furthermore, since the effects of
globalizing processes like trade liberalization are extremely context dependent, it has

13
been suggested that public health research should now move to studying the
pathways between these macro-level processes and health via reference to case-
specific contexts (Doyal, 2005) .

1.1.1 Research Objectives and Questions:


With the above considerations in mind the first research objective of this work
(RO1) is as follows:

Research Objective 1 (RO1): To identify how trade liberalization and social policy
interact to influence health and its social
determinants.

Towards meeting this objective an initial research question is:

Research Question 1 (RQ1): How do researchers theorize the pathways and


mechanisms mediating the trade liberalization and
health relationship?

While the first research objective of this thesis relates to understanding the health
impacts of trade liberalization more broadly, the second and final research objective
of this thesis (RO2) relates to a specific empirical examination of how trade
liberalization interacts with social policy to influence health. It is as follows:

Research Objective 2 (RO2): To investigate and analyse how the phase-out of the
Multi-Fibre Arrangement impacted health in
countries reliant on the textile and clothing sector for
employment.

The research questions associated with this objective are:

Research Question 2 (RQ2): How did health outcomes change after the phase-out
of the Multi-Fibre Arrangement in countries reliant
on the textile and clothing sector for employment?

Research Question 3 (RQ3): What are the potential causal mechanisms


responsible for these changes?

This empirical work and the overall research design of the thesis will be
described in greater detail later in this chapter.

14
1.2 Methodology

1.2.1 Political Economy of Health Perspective


This thesis is conducted from a political economy of health perspective
(Bambra, 2011; Birn, Pillay, & Holtz, 2009; Nancy Krieger, 2011). As such it
crosses both social epidemiological and sociological boundaries (Beckfield &
Krieger, 2009; Muntaner et al., 2011). A political economy of health approach
regards economic, social and political structures as the ultimate determinants of
health.

A political economy of health approach can incorporate biomedical and


behavioural models of health by recognizing the importance of health care services
and lifestyle/behavioural factors, such as those related to diet, smoking and alcohol
consumption (Birn et al., 2009). Much of the work which investigates the health
impact of trade policy uses a political economy approach in this way, for example by
drawing attention to the role of political actors and processes in facilitating the
spread of unhealthy diets, people’s consumption of tobacco and alcohol, and the
privatization of health systems.

A political economy of health approach can also be used to explain how social
determinants of health come to shape health distributions. To date there is no
universally agreed definition of what the SDOH concept actually encompasses,
however reference is commonly drawn to the work of the World Health
Organization’s (WHO) Commission on the Social Determinants of Health (CSDOH).
Here attention is focused on the conditions under which people live and work
(CSDOH, 2008). That there is no universally accepted definition of SDOH is not
necessarily a problem since public health professionals may be conceptualizing these
determinants in ways which are useful for the populations they seek to consider and
the problems they seek to address. Nonetheless, it is important to recognize the
broader implications of how we discuss and conceptualize SDOH (Raphael, 2011).

Some researchers conceptualize the social determinants of health in ways that


overlap with biomedical and behavioural models of health by incorporating the
health care sector and lifestyle factors such as smoking and dietary considerations,
into the concept. While recognizing the health importance of these conditions, this
thesis draws on the sociological construct of fundamental determinants of health to

15
set aside these factors which are more biomedical and behavioural in nature. This is
done in order to achieve analytical clarity and also to recognize a distinction in health
determinants which takes us beyond descriptive accounts by assigning a higher
degree of priority to addressing certain health determining factors over others.

The fundamental determinants of health are perhaps best understood by


contrasting them with what researchers Link and Phelan (1995) describe as
‘contextualized risk factors’. Whereas the latter are seen to explain “how people
come to be exposed to individually-based risk factors such as poor diet, cholesterol,
lack of exercise, or high blood pressure” (p. 81), the fundamental determinants of
health are broadly conceptualized as relating to peoples’ social position and thereby
include factors like “money, knowledge, power, prestige, and the kinds of
interpersonal resources embodied in the concepts of social support and social
networks” (p. 87).

An important implication of this distinction is that even if we change the


contexts within which people are exposed to individually based risk factors, the
fundamental determinants of health will continue to shape population health profiles.
This is because factors related to peoples’ social position are associated with multiple
risk factors as well as multiple health outcomes; and as Link and Phelan (1995) note,
we live in a dynamic world system whereby new diseases and risk factors are always
emerging. As such, those with the most resources will always be best positioned to
achieve better health.

The financing of health care can itself be conceived of as a fundamental


determinant of health, as lack of access to publicly funded care can cause households
to spend significant amounts of money on health services. Lack of access to care can
also cause loss of earnings in cases of illness (Labonté & Schrecker, 2007). However,
in using a political economy approach, this thesis is predominately focused on how
economic and political processes impact determinants of health beyond medical care
and the contextualization of risk factors. Thus the use of SDOH as a concept in the
remainder of the thesis refers to this focus.

Emphasis is placed on these fundamental determinants of health since, as


previously noted, work which focuses on the relationship between trade and health
has paid them less attention and because, as the fundamental cause theory makes

16
clear, a focus on these types of causes will be the most effective in facilitating lasting
health improvements. With this focus, this work is aligned with population health
perspectives which recognize that while good medical care is vital, unless political,
economic and social drivers of health are addressed, people’s opportunities for well-
being are limited (although often population health approaches are also concerned
with contextualized risk factors) (Jamrozik & Hobbs, 2004; Kindig & Stoddart, 2003;
Roos, Brownwell, & Menec, 2006).

The implications of this methodological focus are found in the decisions made
in relation to the research design of this thesis, which are detailed in the following
section.

1.2.2 Research Design


An important aspect of the research design of this thesis is that the objectives,
research questions and methods were not fixed from the beginning, but rather
emerged and evolved as the work progressed. Two major research components
characterize the thesis.

In seeking to answer the initial research objective and question (RO1 and
RQ1), the first component of this thesis is a systematic literature review which aims
to understand the pathways characterizing the trade liberalization and health
relationship. For reasons outlined above, this review pays particular attention to
SDOH pathways.

Findings from the literature review informed the second component of this
thesis which is a case study of how trade liberalization and social protection policies
interact to influence distributions of health. The ‘case’ in this study is the 2005
phase-out of the Multi-Fibre Arrangement (MFA) which after nearly 50 years of
protection abruptly liberalized the textile and clothing sector. The units of analysis
are the many low- and middle-income countries for which the clothing and textile
industry is an important source of employment, especially for women. Since systems
of social protection in these countries are highly diverse, analysing how important
determinants of health and health outcomes changed in these countries after the MFA
phase-out can help to develop our understanding of how trade liberalization and
social policies interact to influence health (RO1 and RO2).

17
1.2.3 Analytical Frameworks and Methods
Two analytical frameworks are associated with this thesis. The first is an
adapted ecosocial framework which delineates the links between trade liberalization
and health. This framework is presented in Chapter 3 and was developed as a way of
organizing the findings of the literature review. A main feature of this framework is
the recognition that the pathways between trade liberalization and health overlap and
interact across various socio-political, historical and spatiotemporal considerations.

The second analytical framework is based on work undertaken by the


Employment Conditions Knowledge Network (EMCONET) of the World Health
Organization’s Commission on the Social Determinants of Health (Benach,
Muntaner, & Santana, 2007). This framework is first described in Chapter 2 and is
found to offer the most comprehensive outline of pathways to health in the context of
labour markets. The framework is conceptually separated at a micro- and macro-
level. Considered at the micro-level are a number of behavioral, psychosocial, and
physiopathological pathways. At this level, four main categories of risk exposure
(physical, chemical, ergonomic, and psychosocial) are seen to be mediated by social
mechanisms and influenced by six different types of employment conditions (full
employment, unemployment, precarious employment, informal employment, child
labour, and slavery & bonded labour). At the macro-level, a main consideration is the
role of labour market and welfare state policies in shaping these more micro-level
considerations.

Together these two frameworks helped to guide decisions about the methods
used in the case-study component of this thesis: fuzzy-set qualitative comparative
analysis (fsQCA) and process tracing. As will be made clearer in Chapter 5, fsQCA
is a cross-case method which enables logical statements to be made about the
conditions that are necessary and/or sufficient for an outcome. This method was
chosen over the typical cross-case method utilized in macro-level public health work,
logistical regression, since it is better able to consider the interactive effects of causal
conditions (as captured in the adapted ecosocial framework) and how they may vary
according to different macro-level contextual conditions (as laid out in the
EMCONET framework [Benach et al., 2007]).

A main limitation of fsQCA however, is that like other cross-case methods, it


identifies associations, not causation. It is for this reason that process tracing was

18
used to further investigate the fsQCA findings through in-depth case-study work of
twelve countries. Since the EMCONET framework (Benach et al., 2007) identifies a
comprehensive indication of the potential causal processes behind the fsQCA results,
it is this framework that is used to guide and causally interpret these process tracing
efforts.

1.3 Contributions to Knowledge

The relationship between trade and health is undeniably, enormously complex.


Seeking to untangle these pathways is therefore an ambitious endeavour. The
literature review undertaken in this thesis represents the first attempt at delineating
the health impacts of trade liberalization through a systematic narrative synthesis.
While focused on SDOH related pathways, this analysis exposes new considerations
for the trade and health research agenda. Of primary importance are the various, and
often problematic, ways in which trade liberalization is conceptualized and defined
in public health work. Drawing on the field of development economics, this thesis
provides a definition of trade liberalization which avoids such problems and can
begin to move the agenda forward. Critically analysing the trade and health literature
also exposes the utility of ecosocial theory, which is found to both capture the major
features of the trade liberalization and health relationship and provide a useful means
of advancing areas thus far under conceptualized.

In undertaking the case study of the MFA phase-out, this thesis applies an
original and innovative method to analyze how trade and social policies interact to
influence health: fuzzy-set qualitative analysis combined with process tracing
methods. In doing so, it offers some of the first empirical evidence of how trade
liberalization interacts with social policies in the context of labour markets to
influence health. While these results are tentative in nature, they do offer an
important consideration for both policy makers and researchers alike: that social
protection is often inaccessible to the type of workers who may be the most
vulnerable to processes of liberalization and that many workers are particularly
vulnerable to processes of liberalization due to the structure of labour market and
social protection policies.

19
1.4 Organization of Thesis
This thesis consists of 10 chapters which are conceptually organized into three
parts. Following this introduction, the first part of this thesis is covered in Chapters 2
and 3. These chapters deal with the first research component of this thesis: the
systematic literature review. While Chapter 2 describes the methods and results of
the review, Chapter 3 critically discusses these findings and presents the
aforementioned adapted ecosocial framework.

The second and third parts of this thesis are concerned with the second
research component: the case study of the MFA phase-out. The second part consists
of Chapters 4, 5, and 6 and deals with the cross-case component of this work.
Chapter 4 will introduce the MFA phase-out and further consider the analytical value
of the EMCONET framework (Benach et al., 2007). Chapter 5 will introduce fuzzy-
set qualitative comparative analysis as the ideal cross-case method for investigating
the health impacts of the MFA phase-out, and Chapter 6 will present the results of
the fsQCA and a critical discussion of these findings.

The third and final part of this thesis deals with the within-case analysis of the
MFA phase-out. It consists of Chapters 7, 8 and 9. Chapter 7 introduces process
tracing as useful method for looking further into the fsQCA results. Chapter 8
presents the results of the in-depth case study work, and Chapter 9 critically
discusses these findings. Finally, Chapter 10 will conclude with a consideration of
the main results, strengths, and limitations of this thesis, as well as where we are left
in terms of next steps for research.

20
INTRODUCTION TO PART 1
As mentioned in Chapter 1, this thesis is conceptually organized into three
parts. The first part of this thesis is comprised of Chapters 2 and 3. These chapters
concern the first research component of this thesis which is a systematic literature
review of conceptual frameworks which link trade liberalization to health. Chapter 2
will detail the methods used in this review and explore the general findings of this
work. Chapter 3 will then critically discuss these results in as far as they address the
challenges in understanding the links between trade liberalization and health.

21
CHAPTER 2 LITERATURE REVIEW
2.1 Introduction
This chapter will present findings from a systematic literature review of
conceptual frameworks which link trade liberalization to health. By identifying
elements which link trade liberalization to health, conceptual frameworks offer the
benefit of rendering manageable otherwise complex pathways. Conceptual
frameworks also lend support in the construction of theories and models and can
guide researchers’ choice of methods and research designs (Veselý & Smith, 2008).
Undertaking a narrative synthesis of frameworks can therefore help develop a
consensus on what the pathways between trade liberalization and health are,
highlight gaps in the literature, direct priorities for future research, and allow for the
creation of a delimited number of frameworks from which to work, the value of
which has been described by population health researcher, Barbara Starfield (2001):

“Common frameworks enhance the likelihood that the results of research


studies can be interpreted in a policy framework. When studies use the same
framework, difference in results will suggest the possible mediating or compounding
role of previously unidentified factors that need to be considered in subsequent
research” (p. 552).

This idea is paraphrased particularly well by Labonté and Togerson (2003) in


relation to their narrative synthesis of globalization and health frameworks:

“[R]esearch (or research synthesis) using a smaller number of frameworks is


more likely to generate findings that can be translated into policy ‘so what’s’? It is
more likely to generate novel and important questions requiring new empirical
answers. The absence of an analytical framework makes it more difficult to
adjudicate the full range of positive and negative affects of globalization, and
particularly trade liberalization, on health outcomes or health-determining contexts”
(p. 4).

This chapter is divided into three main sections. Following this introduction,
section two will provide the details and appropriateness of the methods used in this
review. Here explicit processes of study identification, study selection, data
extraction, critical appraisal and data synthesis will be detailed. In a preliminary
synthesis of the findings, the third section of this chapter will explore the general

22
results of the literature included in this study. Chapter 3 will then critically discuss
these results in as far as they address the challenges in understanding the links
between trade liberalization and health. Drawing on the points raised in this
discussion as well as social epidemiological theory, Chapter 3 will also present a new
conceptual framework which both captures the major features of the trade
liberalization and health relationship and provides a useful means of advancing areas
thus far under conceptualized.

2.2 Methods:

Explicit details of the protocol and logic used to identify, select, critically
appraise, and synthesize findings is outlined in the below sections. This makes clear
the links between the details of included studies and conclusions drawn.

2.2.1 Inclusion Criteria


For literature to be included in this review, a clear analytical framework for
conceptualizing pathways between trade liberalization and health must have been
explicated.

2.2.2 Exclusion Criteria


Conceptual frameworks that were not available in English were not included
in this review.

2.2.3 Searching for Studies


Locating articles that investigate the relationship between trade liberalization
and health involved two primary strategies:
1. The Applied Social Science Index and Abstracts database (ASSIA) database,
PAIS International database (CSA Illumina, Bethesda, Maryland;
http://www.csa.com), Econlit (OvidSP, New York, New York,
http://www.ovid.com), and the ISI Web of Knowledge (WOK) (CSA
Illumina, Bethesda, Maryland; http://www.csa.com) were searched between
December 12, 2010 and March 22, 2011.
The ASSIA database includes works published since 1987, the PAIS
International database includes work published since 1972, the Econlit
database includes work since 1969, and the WOK since 1990. These four
databases were chosen for their expanse of literature related to economics,

23
social policy, and health.
Search parameters included the following keywords “analytical”,
“framework”, “pathways” “linkages”, and “relationship”; in various
combinations with descriptions of liberalization including “trade”,
“globalization”, “globalisation”, “liberalisation”, “liberalization”, “global
economy”, “global economics”, “transborder”; and health outcome
descriptors such as, “health”, “well-being”, “living conditions”, “morbidity”,
“mortality”, “poverty”, “infant mortality”, and “life-expectancy”.
2. Google Scholar was also utilized to locate articles. Search keywords mirrored
those described above.

2.2.4 Selection of Studies


Literature identified through the above search strategies underwent a two
stage selection process. In the first stage of the selection process, titles and abstracts
were systematically scanned for their relevance to the topic. Studies which clearly
did not meet the inclusion criteria were classified as irrelevant. Where a definite
decision could not be made, the second stage of the selection process required that
full reports be obtained for a detailed assessment of the study against the inclusion
and exclusion criteria.

2.2.5 Preliminary Search and Refined Search Strategies


A preliminary search was undertaken in the ASSIA and PAIS databases to
pilot the efficacy of the search terms and selection process against the inclusion and
exclusion criteria. This led to a low recall of 158 articles. Studies which clearly did
not meet the inclusion criteria were classified as irrelevant (n=99). Where a definite
decision could not be made, the second stage of the selection process required that
full reports be obtained for a detailed assessment of the study against the inclusion
and exclusion criteria (n=59). Of these full reports only 6 were identified for
inclusion in the review.

To further increase recall and precision, a refined search strategy was


developed whereby reference lists of articles meeting the inclusion criteria were hand
searched to identify additional frameworks. Furthermore, lists of potential
frameworks were identified by using Google Scholar’s ‘cited by’ feature on
frameworks identified for inclusion.

24
However, frameworks identified through these supplementary strategies with
an exclusive focus on the health impacts of trade liberalization via reference to health
services, tobacco, alcohol, medicines or food, were not included. While these factors
are all important determinants of health, each of these areas already has a rich and
extensive research environment. Moreover, this review is interested primarily in the
understudied impacts of trade liberalization on the social (fundamental) determinants
of health.

In the preliminary search, three main issues, or a combination thereof, arose


which makes explicit the type of studies which are not included in this review. These
issues also highlighted the need for a more explicit understanding of what a
conceptual framework is. These issues and a more thorough explanation of what this
review deems a conceptual framework are discussed below.

First, it quickly became apparent that much of the literature identified by this
review discusses relationships between trade liberalization and a health determinant
(e.g. poverty) without addressing any clear pathway from the determinant to health.
In an article published by the International Monetary Fund (IMF), for example,
authors Berg and Krueger (2002) discuss the relationships between trade
liberalization, growth, and poverty (as income deprivation). It goes without say that
poverty is a major determinant of health. However, because we must otherwise rely
on other theoretical models of how and why being poor brings about health impacts,
articles that do not explicitly discuss the health impacts of poverty are not included in
this review.

It was also recognized that much of the identified literature discusses aspects
of trade or globalization not specifically related to trade liberalization. This was
likely to happen given the use of broad economic keywords in the search strategy.
Arestis and Caner (2004) for example, focus on the poverty impacts of financial
liberalization. While there are many important relationships and commonalities
between trade liberalization, financial liberalization and other processes of
globalization, this review sought to explore these issues from the singular perspective
of trade liberalization. Moreover, any framework which relates trade liberalization to
other aspects of globalization will be included in the review and expose such links.

25
Refining the search terms would have been one way to deal with these first
two issues, excluding ‘poverty’ or ‘globalization’ from the search keywords might
have increased the likelihood that identified frameworks were eligible for inclusion.
However, because of the already low recall numbers, any potential loss of
frameworks did not seem worth this risk.

Finally, much of the identified literature relates economic processes to health


outcomes with little explanation of the pathways by which these relationships are
produced and maintained. For example, Martens and colleagues (2010) use a
globalization index to test hypothesized health impacts of various dimensions of
globalization. While important for providing empirical evidence, analyses or models
that fail to elucidate specific channels of interaction between trade liberalization
dimensions and health, are not included in this review. This issue in turn, raised the
question of what should be considered a conceptual framework.

2.2.6 Defining Conceptual Frameworks

Population health researchers Carpiano and Delay (2006) distinguish between


conceptual frameworks, theories, and models in A guide and glossary on
postpositivist theory building. These constructs are seen to operate along a
continuum where their “scope decreases as logical connectedness and specificity
increases” (see figure 1) (p. 565).

Figure 1 The continuum of frameworks, theories, and models (Carpiano & Daley 2006)

26
The authors argue that a “conceptual framework identifies a set of variables
and the relations among them that are presumed to account for a set of phenomena”
(p. 565 emphasis in original). Frameworks are thus meant to organize assumptions
and act as a prerequisite for theory, which in turn “explicates a more dense and
logically coherent set of relationships, including direction, hypotheses, as well as
how variables may covary” (p. 565). Finally, the authors argue that models “make
specific assumptions about a limited set of parameters and variables” (Carpiano &
Daley, 2006, p. 565). It should be noted however, that this view of theory building
has been questioned on the basis that “the birth of new theory is a much more chaotic
process” (Vågerö, 2006, p. 573).

To make the differences between the three constructs more apparent, the
authors present the following illustration in figure 2.

Figure 2 Example of the connection between a framework, a theory, and a conceptual model
(Carpiano and Daley 2006)

Clarifying the differences between these three constructs is conceptually


useful in some regards, but perhaps overly simplified. While the authors
acknowledge some overlap between the three concepts (see figure 1), it seems
plausible that even greater intersection may occur, when for instance, more is known
about certain links than others. For example, authors might acknowledge the role of
the social environment in influencing health behaviours very broadly, but then draw
on more refined theory to explore the links between health behaviours and obesity in
greater detail, resulting in relationship which is illustrated in figure 3. This seems

27
especially more likely when a greater number of contextual levels are considered in a
largely nascent area, as in the case of trade liberalization and health. Therefore, for
the purposes of this review, authors drawing on any of the three constructs in a way
which delineates one or more pathway between trade liberalization and health will be
included and for simplification purposes will be deemed a ‘framework’.

SOCIAL
Framework:
ENVIRONEMENT

HEALTH
Theory:
BEHAVIOURS

Conceptual Model: OBESITY

Figure 3 Example of the difficulty in distinguishing between frameworks, theories, and


conceptual models

In order to help guide the decision of whether a framework should be


included, this study adopted a broad definition from social science research which
proposes that a conceptual framework can be understood as a tool for organizing a
set of ideas and interrelationships in a way which maps out elements of larger social
processes (Vesely & Smith, 2008). This conceptualization was then incorporated into
the search strategy by asking the question ‘Does this author provide a clear,
systematic structure, outlining possible pathways through which trade liberalization
potentially impacts health?’

2.2.7 Data Extraction


Once studies were identified for inclusion, a process of ‘data extraction’ was
undertaken. In order to answer the first research question of this thesis (RQ1), data
was extracted from identified studies to answer the following questions.

1. How is trade liberalization understood in analytical frameworks relating


trade to health?
2. How is health conceptualized in these frameworks?
3. How do researchers theorize the mechanisms and pathways mediating the
liberalization and health relationship (RQ1)?

28
The first two of these questions were posed since the ways in which trade
liberalization and health are understood can both conceal and illuminate pathways
and mechanisms important for health.

In line with the first question, a main consideration of the data extraction
process was determining how trade liberalization is understood by framework
authors. In her influential book on poverty, Ruth Lister (2004) describes the
differences between concepts, definitions, and measurements. Although some
overlap is acknowledged, concepts are understood to operate at the general level,
whereas definitions and measurements represent areas of respectively narrower
focus. Lister notes that the three constructs are often conflated in work on poverty,
despite their different purposes. She also notes that the constructs are highly
contentious due to their material effects. However, by distinguishing between the
three, Lister not only provides a more comprehensive account of poverty, but also
helps to elucidate on what level controversies are based.

Given the highly complex and often contentious relationship between trade
liberalization and health, this differentiation was found to be especially useful in
determining how trade liberalization is understood across conceptual frameworks
included in this review. Therefore, information on how authors conceptualized,
defined and measured trade liberalization was noted for each framework. For the
purposes of this review, trade liberalizing concepts were taken to be the broad
meanings authors conceptualized as lying behind trade liberalizing processes, for
example ‘market integration’ or ‘free trade’; as well as the larger contexts within
which authors see these processes taking place, for example ‘globalization’ or
‘neoliberalism’. Definitions by contrast, were characterized as offering more specific
explications of trade liberalizing processes (e.g. tariff reductions). And
measurements, in turn, were understood as ways of operationalizing these defined
processes.

In line with the second question of this review, the data extraction process
was also concerned with how framework authors conceptualize health. This review
characterized authors’ conceptualization of health based on whether overall health or
health inequalities were explored; whether trade liberalization was related to specific
health outcomes or health/health inequalities in general; and whether authors used a
readily identifiable 'frame' of health, where the latter draws on work by Labonté

29
(2008) which highlights five readily identifiable frames of health: health as security,
health as development, health as a global public good, health as a commodity, and
health as a human right. These frames constitute the ways in which health is
discussed in both global trade and global health discourses and thus has implications
for how well frameworks are received and acted upon. These frames also have
important implications for health equity. Therefore the implications of these frames
should be taken into consideration if a comprehensive understanding of how trade
liberalization impacts health is to be accepted by actors in the global arena with
positive impacts on health equity.

The final element of the data extraction process focused on the third question
of this review: how do researchers theorize the mechanisms and pathways mediating
the liberalization and health relationship. In reviewing included frameworks, four
contextual domains were readily identifiable within which trade liberalization was
seen to impact health: flow of goods; agricultural and food trade; structural
adjustment policies (SAPs) and Poverty Reduction Strategy Papers (PRSPs), and
labour markets. Characterizing frameworks on this basis represented the broadest
way of understanding how mechanisms and pathways were theorized.

2.2.8 Critical Appraisal and Narrative Synthesis


Studies were critically appraised on their ability to answer the three main
questions of this review. It has been argued that narrative syntheses are the best way
to describe and evaluate macro-level pathways to health (Labonté & Schrecker,
2006). The ability to explore and synthesize findings from frameworks included in
this review was made possible by extracting and organizing the data collected into
thematic categories that responded to the three questions characterizing this review,
as identified above.

2.3 Results
In total, 41 studies were identified for inclusion in this review. Figure 4 below
presents an overview of the study identification process. It should be noted that some
of the frameworks included in this review were drawn upon more than once across
different literature. This was most often the case with literature which drew on work
of the World Health Organization’s (WHO) Globalization Knowledge Network
(GKN) and Commission on the Social Determinants of Health (CSDOH). Duplicate

30
frameworks however, were counted uniquely in the 41 identified studies, as different
literature typically emphasized or deemphasized certain aspects of the same
framework.

31
Titles and Abstracts from Database Review

ASSIA/PAIS n=158
EconLit n=396
Web of Knowledge n= 625
Total n=1179

Stage I: Irrelevant

n=1004

Full Reports Assessed for Eligibility

ASSIA/PAIS n=59

EconLit n=40

Web of Knowledge n=76

Total n=175
Stage II: Irrelevant

n=158
Studies to be Included in Review

ASSIA/PAIS n=6

EconLit n=2

Web of Knowledge n=9

Total n=17

Additional Studies Identified by


Google Search
n=10

Additional Studies Identified


through Supplementary Search
Strategies
n=14

Total Studies in Review

N=41

Figure 4 Overview of Study Identification Process


32
The remainder of the results section will be divided into three major
subsections and organized as follows to answer the three questions of this review:
1. How is trade liberalization understood in analytical frameworks relating trade
to health?
2. How is health conceptualized in these frameworks?
3. How do researchers theorize the mechanisms and pathways mediating the
liberalization and health relationship?

2.3.1 How is Trade Liberalization Understood in Analytical Frameworks


Relating Trade to Health
This section will demonstrate how trade liberalization is understood in
analytical frameworks included in this review, first, by describing how trade
liberalization is conceptualized and then by exploring how it is defined. As
previously noted, in her influential book on poverty, Ruth Lister (2004) describes the
differences between concepts, definitions and measurements. Whereas concepts
operate at the general level, definitions and measurements represent areas of
respectively narrower focus. In terms of trade liberalization, concepts can be
understood as the broad meanings behind trade liberalizing processes as well as the
contexts within which these processes take place. Definitions by contrast, offer more
specific explications of the processes which characterize trade liberalization and in
turn, can be operationalized via measurements.

How is trade liberalization conceptualized in analytical frameworks relating


trade to health?
Frameworks were first categorized by the context in which trade liberalization
is discussed. Not surprisingly, the majority of framework authors contextualize trade
liberalization in relation to globalization. Others discuss trade liberalization
exclusively in reference to the related contexts of structural adjustment policies
(Breman & Shelton, 2007; De Vogli & Birbeck, 2005), aggregate shocks (Mendoza,
2009) development issues, (Singer, 2008; Stuckler & Basu, 2009) foreign policy
(Feldbaum, Lee, & Michaud, 2010), the General Agreement on Trade in Serivces
(GATS) (Woodward, 2005), and trade policy in general (Blouin et al., 2009; Grown,
2005; Rayner, Hawkes, Lang, & Bello, 2006; Thow, 2009).
Among framework authors who use globalization to contextualize trade
liberalization, many restrict their analysis to that of economic globalization

33
(Beaglehole & Yach, 2003; Cheru, 2002; Koivusalo, 2006; Labonté et al., 2007;
Polakoff, 2007; Smith & Signal, 2009; Woodward, Drager, Beaglehole, & Lipson,
2001, 2002) and draw on Woodward and colleagues' (2001) characterization of
economic globalization as “the fundamental driving force behind the overall process
of globalization” (p. 876).
Literature drawing on the framework devised by the Globalization Knowledge
Network of the WHO Commission on the Social Determinants of Health also adopts
the position that economic globalization is the fundamental driving force behind
globalization; this literature however, does not discount other domains of
globalization that may have an impact on health (e.g Labonté et al., 2007).
Other frameworks position globalization in relation to a variety of economic,
political, technological, cultural, social and environmental domains (e.g. Borghesi &
Vercelli, 2003; Doyal, 2002; Huynen, Martens, & Hilderink, 2005a, 2005b; Labonté
& Torgerson, 2003, 2005; Lee, 2000). Few offer more narrow understandings, such
as the framework presented by Cornia (2001) which defines globalization as “the
process whereby national and international policy-makers promote domestic
deregulation and external liberalization” (p. 834).
Many framework authors identify neoliberalism as a force shaping
globalization and development policies (De Vogli, Gimeno, & Mistry, 2009; Fox &
Meier, 2009; Koivusalo, 2006; Labonté et al., 2007; Meier, 2006; Muntaner et al.,
2010; Polakoff, 2007; Singer, 2008; Yaşar, 2010). The 'Washington Consensus'' is
also highlighted as shaping globalization and its related processes by a number of
authors (Corrigall, Plagerson, Lund, & Myers, 2008; De Vogli et al., 2009; Fox &
Meier, 2009; Labonté & Schrecker, 2006, 2009; Meier, 2006; Yaşar, 2010). Many of
these same frameworks, especially those drawing on the work of the WHO GKN,
also emphasize the role of power relations in shaping the globalization context (De
Vogli et al., 2009; Fox & Meier, 2009; Koivusalo, 2006; Labonté & Schrecker, 2007;
Labonté & Schrecker, 2009; Labonté et al., 2007; Labonté & Schrecker, 2006;
Meier, 2006; Muntaner et al., 2010).
Related to the above, some framework authors conceptualize globalization as
an inevitable force albeit sometimes necessitating complementary policies (e.g.
Thow 2009), whereas others acknowledge it as a political phenomenon, capable of
being challenged (e.g. Labonté et al. 2007).
Framework authors also conceptualize trade liberalization by appealing to
34
broad ideas of openness (Beaglehole & Yach, 2003; Blouin et al., 2009; Cornia,
2001; Diaz-Bonilla, Babinard, & Pinstrup-Andersen, 2002; Doyal, 2002; Labonté &
Schrecker, 2006; Labonté et al., 2007; M. B. Smith & Signal, 2009; Woodward et al.,
2001), market integration (Diaz-Bonilla et al., 2002; Huynen et al., 2005a, 2005b;
Labonté & Torgerson, 2003; Rene Loewenson, Nolen, & Wamala, 2010; Polakoff,
2007; Woodward, 2005) and trade flows (Polakoff 2007; Singer 2008; Huynen et al.
2005a, 2005b; Lee 2000). A range of institutions, agreements, and policies are also
drawn on and shape implicit explanations of trade liberalization. Institutions
commonly identified are the World Trade Organization (WTO), IMF and World
Bank. Agreements drawn on include the General Agreement on Tariffs and Trade
(GATT), GATS, and the Trade-Related Intellectual Property Rights Agreement
(TRIPS), along with various regional and bilateral agreements. Specific measures of
these agreements are also identified such as the Agreements on Sanitary and
Phytosanitary Measures (SPS), and Technical Barriers to Trade (TBT).
Finally while many authors clearly position financial flows and foreign
investment within conceptualizations of trade liberalization (Borghesi & Vercelli,
2003; Doyal, 2002; Polakoff, 2007; Grown, 2005; Labonté & Torgerson, 2003;
Muntaner et al., 2010; Smith & Signal, 2009; Woodward et al., 2001), others seem to
position these concepts in separate domains (De Vogli & Birbeck, 2005; De Vogli et
al., 2009).

How is trade liberalization defined in analytical frameworks relating trade to


health?
Trade liberalization itself is scantly explicitly defined in identified
frameworks. However, exceptions to this finding include frameworks by Grown
(2005), Hawkes (2006), Rayner and colleagues (2004), Labonté and Torgerson
(2005); and Thow (2008).
One of the most comprehensive explanations of trade liberalization is
provided by Thow (2009). The author outlines various aspects of trade liberalization
such as its aim: “to improve economic growth through allowing countries to
specialise in their production of goods and services, and trade with each other”; the
forums through which trade liberalization occurs: “multilateral WTO negotiations,
regional bilateral free trade agreements (FTA), negotiations for WTO accession
(countries acceding are required to liberalise trade policies in line with current WTO

35
agreements and negotiation with current members), and liberalization of polices as a
component of loan conditionalities of international financial institutions (particularly
the International Monetary Fund and the World Bank); as well as the areas where the
'liberalisation of policy regimes' takes place: in “physical trade, financial flows and
sectoral liberalisation”. Finally, Thow establishes that the “most obvious policy
changes relate to reductions in barriers to import of goods (tariff reductions), but also
includes “export promotion, reducing restrictions on company ownership, financial
flows and trade in services, implementing customs reforms and laws regarding
equality of treatment (both for firms and countries)” (p. 2151).
Other explicit definitions are less extensive. Labonté and Torgerson (2005) for
instance, define trade liberalization as “the removal of border barriers, such as tariffs,
on the flow of goods and capital” (p. 161). Grown (2005) defines trade liberalization
as “the progressive reduction of barriers to imports and exports” which “can occur
through an autonomous decision of a government to remove or reduce barriers to
exports, eliminate subsidies to domestic industries and firms, and privatize goods and
services, all of which are intended to result in the freer movement of capital, goods,
and labour across borders” (p. 28-29).
Definitions of trade liberalization will be further discussed in Chapter 3.

2.3.2 How is Health Conceptualized in Analytical Frameworks Relating


Trade to Health
Identified frameworks are almost equally split between those exploring
overall health status outcomes (n=21) and those exploring health
inequalities/inequities (n=20). Some frameworks explore both overall health status
outcomes and health inequalities/inequities; when this occurred, frameworks were
characterized under the thematic category of health inequalities/inequities for
simplification purposes.
Of those exploring overall health, chosen outcomes vary in specificity. Some
frameworks for example, employ very specific measures of health such as HIV/AIDs
(De Vogli and Birbeck 2005; Yaşar 2010), whereas others explore more general areas
of health outcomes such as nutrition related diseases (Loewenson et al., 2010;
Rayner et al., 2006; Thow, 2009), reproductive health (Grown, 2005), mental health
(Corrigall et al., 2008), and occupational health (Loewenson, 2001). The majority of
frameworks however, conceptualize health in broad and often undefined terms

36
(Bettcher, Yach, & Guindon, 2000; Blouin et al., 2009; Breman & Shelton, 2007;
Cornia, 2001; Diaz-Bonilla et al., 2002; Huynen et al., 2005a, 2005b; K. Lee, 2000;
Stuckler & Basu, 2009; Woodward, 2005).
Of frameworks exploring health differences between populations, the majority
frame these differences as 'health inequalities' (Beaglehole and Yach, 2003; Borghesi
& Vercelli 2003; De Vogli et al., 2009; Fox & Meier, 2009; Feldbaum et al., 2010;
Hawkes, 2006; Labonté and Schrecker 2006; Meier, 2006; Mendoza, 2009;
Muntaner et al., 2010; Singer, 2008; Smith & Signal 2009; Woodward et al., 2002).
However, others use the concept of 'health equity' to frame differences (Koivusalo,
2006; Labonté & Schrecker, 2006, 2007, 2009; Labonté & Torgerson, 2003, 2005).
Within these inequality/inequity frameworks, outcome measures also vary in
their specificity. The majority explore population health differences in very broad
terms often highlighting implications for inequalities/inequities both between and
within countries (Borghesi & Vercelli, 2003; Feldbaum et al.. 2010; Koivusalo 2006;
Labonté et al. 2007; Labonté & Schrecker, 2006, 2007 2009; Labonté & Torgerson,
2005; Woodward et al., 2001, 2002). Others explore more specific areas of health
differences such as nutrition related inequalities (Smith & Signal, 2009; Hawkes,
2006; Mendoza, 2009), inequalities in workers' health (Muntaner et al., 2010),
inequalities in non-communicable disease prevalence (Beaglehole & Yach, 2003),
and drug-related health inequalities (Singer, 2008).
Not only can frameworks be differentiated on the basis of their chosen overall
health outcome, but nearly a third of all frameworks (n=14) used a readily
identifiable 'frame' of health to contextualize chosen outcomes (Labonté, 2008).
Identified frames are: health as a human right (Cheru, 2002; Grown, 2005; Fox &
Meier, 2009; Koivusalo, 2006; Labonté et al., 2007; Meier, 2006; Polakoff, 2007),
health as a development strategy (Singer, 2008; Borghesi & Vercelli, 2003; Mendoza,
2009; Fox & Meier, 2009; Corrigall et al., 2008; Stuckler & Basu, 2009), and health
in reference to global public goods (Labonté et al., 2007; Labonté & Schrecker,
2006; Meier, 2006; Woodward et al., 2002). Two frameworks framed health both in
reference to global public goods and human rights (Labonté et al., 2007; Meier,
2006) and another contextualized health as a human right, a development strategy
and in reference to global public goods (Fox & Meier, 2009).
Of those framing health as a human right, Cheru (2002) and Labonté and
colleagues (2007) both draw on Article 25 of the 1948 Universal Declaration of
37
Human Rights (UDHR). However, Labonté and colleagues also highlight the
importance of the 1966 International Covenant on Economic, Social and Cultural
Rights. Cheru additionally draws attention to the specific right to education
embodied in Article 26 of the UDHR, as well as to the right to food in general. In a
review of globalization's impact on child labour, Polakoff (2007) frames child health
as a human right as codified in the 1989 United Nations (UN) General Assembly
adoption of the Convention of the Rights of the Child, the International Labor
Ogranization's (ILO) Convention Concerning the Prohibition and Immediate
Elmination of the Worst Forms of Child Labor (also known as the Worst Forms of
Child Labor Convention adopted in 1999) and the Optional Protocol to the
Convention of the Rights of the Child, adopted by the United Nations in May 2000.
Polakoff also highlights various domestic laws prohibiting child labor and protecting
children's rights. Grown (2005) in investigating the links between trade liberalization
and reproductive health draws on the importance of sexual and reproductive rights
which she defines as “the prevalence, nature, and distribution of both reproductive
health problems and needs within a given population, the quantity, quality, type, cost,
and distribution of reproductive health services, and finally, the 'fit' between
problems and services, for example, the extent to which women and men can access
the services they need” (p. 29). Koivusalo (2006) discusses the implications of
“commercial rights in the context of human rights and how this relates to the rights to
health and access to health care”. Finally, Meier (2006) and Fox and Meier (2009)
question the utility of framing health as an individual right and instead frame health
as a collective right to development drawing on the UN Declaration on the Right to
Development.
Frameworks which position health in reference to development either describe
health as a driver of economic growth (Bettcher et al., 2001), as a piece of a larger
development agenda (Fox & Meier, 2009; Stuckler & Basu, 2009) or both (Corrigall
et al., 2008; Meddoza, 2009; Singer, 2008; Yaşar, 2010). Two of these frameworks
drew connections between health and the Millennium Development Goals (Singer,
2008; Stuckler & Basu, 2009).
Frameworks relating health to the concept of global public goods (GPGs)
universally identify public health systems as a GPG. Frameworks by Labonté and
Schrecker (2002) and Woodward et al. (2002) both highlight the Framework
Convention on Tobacco Control as an example of how the regulation of health-
38
damaging products is also a GPG. However, while Labonté and Schrecker more
generally emphasize the role of global governance in supplying global public goods,
Woodward and colleagues emphasize the specific role of institutional frameworks
and international rules. Frameworks by Meier (2006) and Fox and Meier (2009) also
direct attention to the role of governance in supplying public goods. However, while
Labonté and Schrecker believe that “internationally, the lack of a global government
means that many kinds of public goods relevant to health are seriously
undersupplied, raising important issues of health equity”, frameworks by Meier and
Fox and Meier position health within a collective right to development, and see its
achievement as a key instrument in facilitating the necessary international
cooperation for GPGs (p. 33).

2.3.3 How Do Researchers Theorize the Mechanisms and Pathways


Mediating the Trade Liberalization and Health Relationship?
Few studies included in this review explore the exclusive impact of trade
liberalization on health. The exceptions are frameworks by Blouin and colleagues
(2009), Thow (2009), Rayner and colleagues (2007) and Grown (2005). While
Blouin and colleagues explore the health impacts of trade liberalization in general,
Grown explores the specific impact of trade liberalization on reproductive health,
and Rayner and colleagues, as well as Thow, explore the specific impact of trade
liberalization on nutritional health.
As discussed previously, the majority of frameworks included in this review
explore the health impacts of trade liberalization within broader globalization
frameworks. Other frameworks explore the pathways from trade liberalization to
health within broader frameworks relating to foreign policy, external shocks, and
trade policy in general.
The scarcity of frameworks which focus exclusively on trade liberalization,
and the abundance of those focusing on globalization, most likely reflects a logical
progression of research which first needed a broad understanding of contextual
factors before more specific pathways could be investigated.
Three early frameworks should be recognized for their role in setting the
foundation of later work, those by: Woodward and colleagues (2001), Labonté and
Torgerson (2003), and Huynen and colleagues (2005ab). A more recent framework
by Labonté and colleagues (2007) also acts as the basis for a number of frameworks

39
included in this review and should be highlighted.
Frameworks included in this review draw largely, but not entirely, on these
four works and begin to paint a more comprehensive picture of trade liberalization's
impact on health. While framework authors offer many interpretations of the
manifestations of trade liberalization, this review identified four, non-mutually
exclusive contexts through which trade liberalization is understood to impact health:
increased flows of goods and people; structural adjustment policies (including
poverty reduction strategy papers); agricultural and food trade; and labour markets.
Following this introduction, this section will first describe the four most
influential frameworks included in this review as mentioned above. In providing a
basis for future research, these early frameworks tend to provide very general
pathways between trade liberalization and health. Their strength however, lies in
their ability to contextualize trade liberalization within broader political processes.
Moreover, based on the framework provided by Labonté and colleagues (2007), later
frameworks are more likely to incorporate pathways which account for social
determinants of health. The framework by Blouin and colleagues (2009) will then be
presented as the only framework that provides a broad overview of the exclusive
impacts of trade liberalization on health. This framework outlines the impacts of
trade liberalization and health across four major pathways: income, income
inequality, insecurity and diet and nutrition. Finally, studies characterized by each of
the four contexts identified above will be discussed.

Five Early and Influential Frameworks

Early and Influential Framework 1: Globalization and Health: a framework


for analysis and action (Woodward et al. 2001)
The framework by Woodward and colleagues (2001) is one of the earliest
identified in the review. It is also one of the most drawn on by other frameworks.
Illustrated in figure 5 (in general) and figure 6 (in detail), one of the basic premises
behind this framework is that “economic globalization has been the fundamental
driving force behind the overall process of globalization” (p. 876). Another is that “a
genuinely health-centred process of globalization can be achieved only by ensuring
that the interests of developing countries and vulnerable populations are fully
represented in international decision-making forums” (p. 880).

40
Figure 5 Globalization and Health: a framework for analysis and action (Woodward et al. 2001
p.877)

41
Figure 6 Globalization and Health: a framework for analysis and action (in detail) (Woodward
et al. 2001 p.878)
As can be seen in figure 6, the authors conceptualize economic globalization
as three processes which exist within a positive feedback loop: “[i]ncreasing cross-
border flows stimulate the development of global rules and institutions, which
promote the opening of economies, which increase the scale and scope of cross-
border flows” (p. 876). Trade liberalization pathways can be seen as a major
component of the ‘opening of economies’ process, which is associated with the
“lowering of trade barriers, removal of capital controls, and liberalization of foreign
exchange restrictions” (p. 876).

42
Pathways to health are characterized as either direct or indirect. Direct
pathways include those which impact the health system or affect population level
health through international markets (TRIPS, for example, is seen to directly
influence health via its influence on pharmaceutical drug prices). Indirect pathways
are those which act through the national and household economies. Examples of
indirect pathways include for example, “the effects of trade liberalization and
financial flows on the availability of resources for public expenditure on health” or
for instance, “the effects on nutrition and living conditions resulting from impacts on
household income” (p. 876). However, trade liberalization’s independent
contribution to these pathways is not explicitly accounted for.
Finally, while not easily identifiable in the authors’ illustrations, the
importance of social policies “such as safety nets and the protection of health and
education spending” is noted, but not emphasized (p. 879). It is also noted, that the
“extent, timing, pace, sequencing and design of policies directed towards opening the
economy must be appropriate to each country’s particular circumstances” (p. 877).

Early and Influential Framework 2: Frameworks for Analyzing the Links Between
Globalization and Health (Labonté and Torgerson 2003)
The framework by Labonté and Torgerson (2003) also represents both an
early and influential work in the area of globalization and health. Commissioned by
the WHO, this framework is based upon a review and critique of other frameworks
relating globalization to health and “might best be considered the rudiments of a
smaller set of frameworks to guide future research and policy analysis” (see figure 7)
(p. 12).

43
Figure 7 Framework for Analyzing the Links Between Globalization and Health (Labonté and
Togerson 2003 p.13)
In this framework, trade liberalization is identified as a defining characteristic
of globalization, and conceptualized by appealing to trade flows and the idea of “free
trade”. The authors note that few of the frameworks included in their review “extend
this emphasis on trade to include an analysis of trade policies (e.g. tariff reductions,
export subsidies” (p. 9). They also note that “[m]ost framework authors regard trade
agreements as a driving force in increased trade flows” and that “[o]nly a few
distinguish the potential health impacts of trade liberalization from de-regulation of
global finance markets” (p. 9).
In a hierarchal fashion, the authors organize pathways between globalization

44
and health via super-ordinate, global, domestic, community and household contexts.
The super-ordinate context refers to countries' pre-existing endowments (such
as per capita income, natural resources, human capital, and demographic profiles)
and political systems and processes which together shape domestic policy responses
to national outcomes of globalization processes. Globalization can for example, be
mediated by a country's “acceptance of discrimination (on the basis of race, ethnicity
or gender), definition of public need and attitudes towards privatization,
determination of public policy, level of unionization and accountability of public
administration” (p. 14). Conflict and political instability, the extent of status
hierarchies and power relations are also seen to determine to what extent macro-
economic policies are accepted or complied with domestically.
The global context is seen to impact health through four key pathways:
macroeconomic policies; trade agreements, flows and institutions; intermediary
global public goods and; official development assistance. Aspects of trade
liberalization are identified within the pathways of macroeconomic policies and trade
agreements and flows, as described below.

Domestic Macroeconomic Policies


Discussion of macroeconomic policies is largely focused on the
conditionalities embedded within SAPs and PRSPs. Trade liberalization is identified
as one of five general reform areas of SAPs which “allow[s] free markets to establish
prices” through “tariff reductions, removal of import controls and [by] eliminating
restrictions on foreign investment/capital markets” (p. 19). Other areas of reform
include the reduction of state controls on prices, the privatization of state-owned
productive assets, domestic austerity measures such as reduced government
spending, and re-orientation of state processes “towards enhancing development of
the private sector” (p. 20).
The authors’ review identifies a range of impacts associated with SAPs such
as, inequalities in access to health services, negative environmental impacts,
increasing food prices, declining government expenditures, social polarization, and
increases in poverty and income inequality.
Policies associated with PRSPs are seen to rely on the privatization of “state
productive assets, increased trade liberalization” and “cost-recovery for health,
education and other social programs” (p. 23). In comparison to the work done on

45
SAPS, fewer studies included in the authors’ review investigate the impacts of
PRSPs. Highlighted criticisms however, outline important implications of PRSPs for
health: namely that PRSPs emphasize cost recovery mechanisms for healthcare
services such as user fees while ignoring the failures of fee-exemption programs for
the poor. PRSPS are also criticized on the basis that they lack clear commitments to
the provision of resources important for health and education, and because they fail
to consider health as an outcome rather than simply a means of development.

Trade Agreements, flows and institutions


Within this pathway trade liberalization is seen to have impacts across five
domains: the physical environment, the social environment, competitive pressures,
(the loss of) regulatory space, and capital markets. Within the physical environment
specific trade liberalizing policies are not explicated but noted are associated losses
in bio-diversity, deterioration of landscapes, as well as consumption related
environmental degradation. Within the sphere of the social environment, the authors
exclusively highlight decreased government revenues resulting from tariff
reductions. Also noted is the fact that many countries have been unable to implement
alternative sources of funding in the face of these declines. Tariffs are observed to be
of particular importance for developing countries, given that they “constitute a very
large portion of overall tax revenue in many developing countries, compared to an
average of only 4% for high-income nations” (p. 26).
Highlighted in the domain of “competitive pressures”, are the largely negative
impacts on domestic manufacturing brought about by the introduction of cheap
foreign imports. It is highlighted for instance that the introduction of cheap imports
has “weaken[ed] the local entrepreneurial base” in the Indian state of Kerala, which
“will erode the State’s ability to tax domestic wealth for purposes of income
redistribution, gender empowerment, maternal/child health and other low
wealth/high health outcomes” (p. 27).
In the fourth domain, trade liberalization is associated with a loss of domestic
regulatory space largely through the implementation of trade agreements.
Agreements highlighted within this domain are the WTO Agreement on Trade-
Related Investment Measures (TRIMS) which “for example, prevent countries from
placing performance requirements (such as requiring local content) on foreign
investment”, the WTO Agreement on Government Procurement which “requires

46
governments to take into account only “commercial interests” when making
purchasing decisions, specially banning preferences based on environment, human or
labour rights”, and the Agreement on Sanitary and Phytosanitary Measures which
“requires that a country’s food and drug safety (sanitary and phytosanitary)
regulations be based on a scientific risk assessment, even if there is no discrimination
between domestic and imported products” (p. 28).
It is noted that a loss of domestic regulatory space can have both positive and
negative implications. It can be positive for example, if it prevents governments from
providing support to environmentally destructive firms. It can be negative however,
if it prevents governments from enacting health protecting regulations.
Finally, costs of implementing trade agreements are described as exceeding
some of the least developed countries' development budgets, which along with low
levels of public sector employment, are seen to further restrict domestic regulatory
space.
In the final domain, the liberalization of capital markets, Labonté and
colleagues note other authors’ claims “that liberalization in capital markets has had
far more negative and very little positive impact than has liberalization in goods”
(see for example, Cornia (2001) and Labonté (2001)) (p. 28). Although it is also
acknowledged that capital market liberalization policies can be mediated by actions
in other areas.
Foreign direct investment (FDI) is observed to be largely the province of
developed countries with 90% of it taking place between North America, Europe,
Japan and China. Seventy percent of the world’s population shares the remaining
10%, with Sub-Saharan receiving only 0.5%. However, it is also noted that “[d]espite
the low proportionate flow, FDI still crowds out domestic investment in most African
countries, which means that many countries’ economies are now being driven almost
entirely by foreign investors who may have little interest in the country other than
how it might affect returns on investment, a condition referred to as
“compradorization” (p. 29). FDI is also noted to have “only been successful in
creating employment in developing countries when it has been in very large amounts
(e.g. Singapore, Malaysia, the Mexican maquiladoras), which precludes a fairer
distribution of FDI amongst all developing countries” (p. 29). It also “generally
takes advantage of low labour costs rather than developing new technological
capacities in the host country, rendering the employment and its associated economic
47
growth very fragile” (p. 29). Finally, both capital mobility and transfer pricing (“in
which companies engage in intra-firm trade with their own subsidiaries”) are noted
to impede the ability of governments to generate revenue, even when FDI is
substantial (p. 29).
Trade liberalization is not discussed explicitly in relation to domestic,
community and household contexts in this framework, however, these areas are
understood to be impacted by trade liberalization policies presented in the global
context. Domestic contexts influenced by global contexts include macroeconomic
policies, migration/refugee policies, labour policies, food security policies, public
provision policies, political power policies and environmental policies. Community
contexts influenced by these domestic policies are access to services and programs,
geographic disparities, community capacities and urbanization. Household contexts
are, in turn, influenced across the domains of income distribution, subsistence
production, health behaviors and social expenditures. Finally, environmental
pathways are seen to influence and be influenced across each of these contextual
levels.

Early and Influential Framework 3: The health impacts of globalisation: a


conceptual framework (Huynen et al. 2005ab)
Huynen and colleagues (2005ab) acknowledge the importance of the above
two frameworks, but argue that their focus on economic globalization and
international governance is too limited. The authors argue that “a conceptual
framework for the health effects of the globalisation process requires a more holistic
approach and should be rooted in broad conception of both population health and
globalisation” (p. 1). The authors highlight that “proximal factors act directly to
cause disease or health gains, and distal determinants are further back in the causal
chain and act via (a number of) intermediary causes” (p. 2).
Globalisation is acknowledged to be a highly complex, multifaceted
phenomenon which can be defined as “an intensification of cross-national cultural,
economic, political, social and technological interactions that lead to the
establishment of transnational structures and the global integration of cultural,
economic, environmental, political and social process on global, supranational,
national, regional and local levels” (p. 2). However, in order to focus their conceptual
model the authors decide to set apart “the following important features of

48
globalisation: (the need for) new global governance structures, global markets, global
communication and diffusion of information, global mobility, cross-cultural
interaction, and global environmental changes” (p. 2). Trade liberalization can be
seen as an aspect of global markets.
Within the authors’ conceptual model (see figure 8), these features of
globalization are seen to impact distal determinants of health such as “health (-
related) policies, economic development, trade, social interactions, knowledge, and
the provision of ecosystem goods and services. In a more detailed illustration of the
authors’ framework (figure 9), global markets are seen to impact economic
development and trade. Within the domain of economic development, the historical
debate over whether economic globalization benefits or harms countries is revisited.
Within the domain of trade, increased global trade flows are of primary concern.
Economic development and trade are seen to affect the more distal
determinants of health, including health services, the social environment, lifestyle
factors, the physical environment, as well as access to food and water. While the
specific mechanisms through which this occurs is not explicated, increased flows are
often the main culprit, except in the case of food security where “liberalization
policies are expected to have profound implications on food trade and, subsequently
food security” (p. 9).

Figure 8 The health impacts of globalisation: a conceptual framework (Huynen et al. 2005 p.3)

49
Figure 9 The health impacts of globalisation: a conceptual framework (in detail) (Huynen et al.
2005 p.5)

Early and Influential Framework 4: Towards Health-Equitable Globalisation:


Rights, Regulation and Redistribution (Labonté et al. 2007)
This framework was developed by the WHO Globalization Knowledge
Network (GKN) which was formed in 2005 to investigate the ways in which
globalization influences the social determinants of health. The authors underscore an
explicit concern for equity in access to these determinants. Given this primary
concern with equity, gains and losses of globalization are not weighed against each
other in the developed framework. This is because the authors believe that
“deterioration in access to [SDOH] for relatively disadvantaged members of society
cannot be balanced by gains elsewhere if the effect is to increase inequity” (p. 19).
A definition of globalisation is adopted from Jenkins (2004) where
globalisation is understood as “‘a process of greater integration within the world
economy through movements of goods and services, capital technology and (to a
lesser extent) labour, which lead increasingly to economic decisions being influenced
by global conditions’—in other words, the emergence of a global marketplace” (p.
17). Like Woodward and colleagues (2001) and Labonté and colleagues (2003), the
authors take economic globalization as primary importance and do so on the basis
that most “dimensions and manifestations of globalisation that are not obviously

50
economic in nature nevertheless, on closer examination, are best explained or
understood with reference to economic factors” (p. 16). However, the authors also
acknowledge that “economics is not the entire story of globalisation and its effects on
health” and argue that “the perspective on globalisation [they’ve] adopted does not
assume away various dimensions of globalisation that are not self-evidently
economic” (p. 17).
Both a simple and complex framework is presented (see figures 10 and 11
respectively). These frameworks draw on work by Didrichsen, Evans and Whitehead
(2001) which identifies “four main mechanisms—social stratification, differential
exposure, differential susceptibility, and differential consequences—that play a role
in generating health inequities” (Labonté et al., 2007, p. 14). As can be seen in figure
11, proxies of social stratification include factors related to income, education,
gender etc. Factors related to differential exposures and vulnerabilities on the other
hand are associated with things like living and working conditions, food quality and
security, and environmental conditions. Six factors are identified for mediating the
relationship between globalisation and health: “material deprivation, medical
progress, acute psychosocial stress, unhealthy lifestyles, stratification and lack of
social cohesion, and positive and negative shocks” (p. 26). These factors interact
across four main pathways to affect health: asymmetries in power and resources;
trade liberalization; aid and investment; and basic needs.

Figure 10 Towards Health-Equitable Globalisation: Rights, Regulation and Redistribution


(Labonté et al. 2007 p.18)

51
Figure 11 Towards Health-Equitable Globalisation: Rights, Regulation and Redistribution
(Labonté et al. 2007 p.87)
Trade liberalization is related to a limited number of issues namely economic
insecurity, declines in public revenues, trade in health services, and food security.
Specific trade liberalizing policies and their relationship to these factors however, is
not comprehensively explored.
That said, trade liberalizing policies which promote exports are, in general,
related to a high suicide rate among farmers in India via economic insecurity.
Reductions in tariffs and regulatory barriers are also seen to increase economic
insecurity by decreasing workers’ revenues and shifting employment. It is also
highlighted that trade liberalization, in conjunction with financial liberalization, has
meant “more volatile markets and increased frequency of external shocks (such as
financial crises, currency devaluations and rapid changes in labour markets and
employment) which translate into increased economic insecurity of individuals” (p.
41). Similar to claims made by Labonté and colleagues (2003), the relationship

52
between trade liberalization and economic instability on its own however, is
acknowledged to be less robust.
It should also be noted that social protection policies are highlighted within
this pathway for their role in mitigating the relationship between trade liberalization,
economic insecurity and health. For example, the reduction or elimination of tariffs
on imports is noted for decreasing public revenues and spending on “health, water,
social services and other public initiatives linked to [SDOH]” in developing countries
(p. 42).
GATS is associated with trade in health services and while it remains unclear
the impact this agreement will have on health, the authors highlight concerns
surrounding its potential to contribute to the commercialization of health services and
thus inequities in access to health services.
Finally, decreases in food security are seen to be impacted by trade
liberalization through greater competition between local producers and foreign
imports. Trade liberalization is also associated with decreasing food security through
policies which promote cash cropping and instabilities in global food prices. It is
acknowledged that “little evidence directly addresses the link between trade
liberalisation and food security” (p. 45), that “greater attention needs to be given to
the sequencing of [trade] reforms”, that “transitional compensatory measures targeted
to lower-income groups may be needed”, and that “[f]or countries with a large
proportion of low income and resource-poor people living in rural areas and who
depend on agriculture, reforms aimed at raising productivity and at non-agricultural
employment creation are essential for enhancing food security” (p. 46).

Early and Influential Framework 5: Trade and social determinants of health


(Blouin et al. 2009)
As previously noted, few studies included in this review explore the exclusive
impact of trade liberalization on health. The framework provided by Blouin and
colleagues (2009) is alone in providing a broad overview of the health impacts of
trade liberalization, though other studies do explore the specific impacts of trade
liberalization on reproductive health and nutritional outcomes (see Thow 2009,
Rayner et al. 2007 and Grown 2005).
In this framework the authors focus on pathways which lead from trade
liberalization to health via SDOH (see figure 12). Four major pathways are identified

53
on the basis of work done by Cornia and colleagues (2008): income, inequality,
economic insecurity and unhealthy diets. These factors are in turn seen to impact
health via material and psychosocial pathways.

Figure 12 Trade and social determinants of health (Blouin et al. 2009 p.503)

Income
In the first pathway, the debate between trade liberalization, growth and
poverty is recounted. The claim that trade enhances growth, reduces poverty, and
thus improves health is exchanged for a more nuanced assessment. The authors argue
that “trade liberalization alone is insufficient to boost the economy” and that
“[c]omplementary policies are needed to ensure that trade openness leads to a high
level of growth” (p. 503). Necessary policy conditions “include a stable
macroeconomic environment, competitive exchange rate, solid fiscal policies, well
functioning agricultural and labour markets, and physical infrastructure (port roads,
telecommunications)” (p. 503). The authors also note the potentially disequalizing
impacts of “a country reducing or removing its barriers to imports and foreign
investment and subsequent outcomes on the poverty level of its own population” due
to shifts in employment patterns and changes in prices of and external demand for
goods (p. 503).

Inequality
In the second pathway, the authors identify other common assumptions of
trade liberalization: “that developing countries, which have an abundance of

54
unskilled labourers, would gain from trade in products produced by unskilled labour”
and furthermore that “the position of unskilled labour in the labour market would be
enhanced vis-á-vis other factors of production, leading to a fall in the skills premium
and hence reductions in inequality” (p. 504). These claims are rejected however, on
the basis of evidence from poor countries which demonstrates a relationship between
increasing wage inequalities, trade openness in general and tariff reductions in
particular.

Economic Insecurity
Like Labonté and colleagues (2003 and 2008), in the third pathway, Blouin
and colleagues (2009) recognize that “trade liberalisation is usually accompanied by
enhanced openness to foreign capital and liberalisation of financial markets and
services”, and that this combination “is often associated with heightened economic
insecurity” (p. 504). Economic insecurity is seen to be influenced by these processes
via “financial crises, currency devaluations, and rapid changes in labour markets and
employment” (p. 504). However, it is acknowledged that there are challengers to this
view (for example, Bourguignon & Goh, 2003).
Without explicating specific trade liberalization policies, the authors focus
their attention in this pathway on the movement of labour from one sector to another
and reference a single study by the ILO (Torres, 2001). This study examined 77
countries and demonstrated that “high levels of international trade in a national
economy were associated with increased movement of workers between sectors” (
Blouin et al. 2009, p. 504). It is highlighted that “[t]his process--known as
churning—needs social safety nets and smooth employment transition mechanisms
to lessen material and psychological stress to workers and their families” (p. 504).

Unhealthy Diets
In the final pathway trade liberalization is related to diet and nutrition via
changes in food prices, “increased desirability and availability of unhealthy foods,
worsening asymmetry between consumers and suppliers of foodstuffs, and growing
urbanization and changes in lifestyle” (p. 504). However, while case studies are
drawn on to illustrate these relationships, specific trade liberalization policies remain
unnamed. In addition, financial liberalization is seen to affect food availability in
this pathway through the penetration of supermarkets and foreign investment in fast-
food outlets.
55
Summary
The above discussed frameworks identify a range of factors which are related
to trade liberalization and important for health. While all of the above frameworks
identify potential pathways through which trade liberalization impacts health,
specific liberalizing policies are only rarely named and never comprehensively
explored.
However, for the purposes of this review, one of the strengths of the early
frameworks is their thorough contextualization of trade liberalization within larger
processes of globalization. It has been argued elsewhere that a particular advantage
of the framework devised by Woodward and colleagues (2001) is that it “focus[es]
on the range of policy choices (by both governmental and private actors) that operate
at the supranational level to affect health” (Labonté et al., 2005, p. 8). The authors’
representation of economic globalization as a feedback loop is also a particularly
useful tool for understanding how trade liberalization is propagated. In a similar vein,
De Vogli, Gimeno and Mistry (2009) propose a conceptual framework which
demonstrates how public polices and economic inequality also exist as part of a
feedback loop. The authors argue that policy reforms representing the ‘Washington
Consensus’, including “financial deregulation, trade liberalization, privatization of
state enterprises, flexibilisation of labour markets and reductions in public
expenditures for health and social welfare”, have contributed to the growth of
transnational corporations (TNC) (p. 689). This growth in turn, is noted to have
allowed TNCs to further advance policies of the Washington Consensus. It is argued
that this “feedback between globalization and the accumulation of wealth and power
among TNCs is key to understanding the increase in economic inequalities between
and within countries”, as well as the resulting health inequalities (p. 689). However,
it is also noted that progressive welfare policies can counteract the described
feedback loop and limit increases in health inequalities. Together the two frameworks
provide a more comprehensive context within which trade liberalization can be
understood.
It has also been acknowledged that a key strength of Labonté and colleagues’
(2003) framework “is its explicit attention to globalization’s influences on the
“policy space” available to national and subnational governments” (Labonté et al.,
2005, p. 8). While the authors associate trade liberalization with a (loss of) policy
space via trade agreements, it is unclear how all the agreements identified by the

56
authors relate to trade liberalization, for instance, the WTO Agreement on
Government Procurement or the Agreement on Sanitary and Phytosanitary Measures.
The supra-ordinate context described in this framework is however, particularly
useful for the purposes of this review, as it identifies a range of factors that shape
acceptance of and domestic policy responses to globalization processes, including
trade liberalization.
A limitation of these two frameworks is their arguably limited focus on
SDOH. Woodward and colleagues, for instance, focus their description of
globalization’s impact on health around health systems (Labonté et al., 2005).
Labonté and colleagues (2003), while incorporating many social determinants of
health into their framework, fail to provide a detailed analysis of the mechanisms “by
which various causal pathways lead to changes in individual and population health
status” (Labonté et al., 2005, p. 8).
The frameworks by Labonté and colleagues (2007) and Blouin and colleagues
(2009) specifically respond to these issues. Labonté and colleagues do so by
exploring the health impacts of globalization through SDOH pathways and by
incorporating Diderichsen, Evans, and Whitehead’s (2001) health inequity model
into their framework. Blouin and colleagues (2009) do so by identifying four
pathways which are impacted by trade liberalization and lead to different
distributions of health: income, income inequality, economic insecurity, and
unhealthy diets.
One criticism of the framework by Blouin and colleagues (2009) is that it
doesn’t account for the positive impacts of trade liberalization on health (Bovet &
Paccaud, 2009). Based on the purposes of this review, other concerns with this
framework relate to its under-conceptualized pathways. In the income pathway for
example, the debate between trade liberalization, growth and poverty is recounted
but rests largely on the results of empirical tests versus a theoretical consideration of
potential pathways. Labour markets and wage differentials are the primary concern in
the income inequality pathway, although conceivably trade liberalization may impact
income inequality through other domains for example, food prices and government
spending. The same case may be made for the economic insecurity pathway.
Additionally, while the role of social policies is emphasized by the authors, a
consideration of the impacts of trade liberalization on government revenues, an
important determinant of social policy, is also absent.
57
In total, the above discussed frameworks have been influential in that they set
the precedent for much of the contemporary work on globalization and health.
Frameworks included in this review draw largely, but not entirely, on these works
and begin to paint a more comprehensive picture of trade liberalization's impact on
health.
The above discussed frameworks are also influential in that they very much
represent landmark investigations along a research continuum which sets the
precedent for understanding the relationship between trade liberalization and health.
While early frameworks establish the context in which trade liberalization can be
understood as an essential component of greater globalization processes, later
frameworks begin to account for a wider range of health influencing pathways by
incorporating a SDOH perspective. More recently, the framework by Blouin and
colleagues (2009) is the first attempt at comprehensively examining the specific
impacts of trade liberalization on health. Nonetheless pathways presented in this
latter framework could be further developed for reasons discussed above.
While framework authors offer many interpretations of the manifestations of
trade liberalization, this review identified four, non-mutually exclusive, contexts
within which trade liberalization is understood to impact health: (1) increased flows
of goods and people; (2) agricultural and food trade; (3) structural adjustment
policies (including poverty reduction strategy papers) and (4) labour markets.
However, it is worth reiterating that these contexts were identified in the context of a
selective search strategy which excluded studies which focused exclusively on health
services, tobacco, alcohol, medicines or food. Moreover, while these contexts are
non-mutually exclusive, pathways described by framework authors almost always
emphasize aspects of one of these contexts over another and were thus characterized
accordingly. However, when framework authors present a pathway in a way in which
there is strong overlap between two contexts, the pathway is included in discussions
of both contexts. Moreover, while these contexts were conceived to organize
pathways presented by frameworks, rather than the frameworks themselves, few
frameworks span more than one context.

Increased flows of goods and people


In this context, trade liberalization is often seen as playing a central role in
increasing the flow of goods and people. Framework authors who relate trade

58
liberalization to increased flows of goods often do so in relation to food. Because so
many authors explore the specific impact of food trade on health, these frameworks
will be explored in a separate section below.
Other framework authors exploring trade liberalization’s impact on the flow
of goods and people often do so in terms of communicable and non-communicable
diseases. Historically speaking, communicable diseases are one of the most well-
known health risks of increased trade (Bettcher et al., 2000). In a framework
exploring the impact of globalization on health, Lee (2000) identifies trade
liberalization as a driver of economic change, influencing communicable diseases
through a range of intermediary determinants of health. Without specifying specific
trade liberalizing mechanisms, communicable diseases are seen to increase with the
“greater worldwide mobility of people, through business, tourism, rural-urban
migration and displacement” (p. 256). Communicable diseases are also seen to be
impacted by increased flows of animals, plants and other goods.
Woodward and colleagues (2002) support Lee’s position and highlight the
association between communicable diseases and the increased movement of people,
animals, and animal products. The authors also note that increased flow of goods
have “led to new human diseases, for example, bovine spongiform encephalopathy”
(p. 6). Again specific trade liberalizing mechanisms are not named. Feldbaum and
colleagues (2010) also support Lee’s position but note the specific negative impact of
tariff reductions (in conjunction with the inadequate regulation) on the spread of
communicable disease.
Also explored by framework authors are the impacts of increased flows of
goods on non-communicable diseases. In one of the earliest frameworks included in
this review Bettchar and colleagues (2000) note that “the health risks and benefits
associated with the liberalization of trade in goods are highly dependent on the nature
of the commodities concerned” (p. 5). The authors identify four categories of goods:
“legal and beneficial (e.g. nutritive food and cost-effective technology); legal and of
doubtful benefit (e.g. technologies of low cost-effectiveness); legal and harmful (e.g.
tobacco, alcohol and weapons); illegal and harmful (e.g. illicit drugs)” (p. 5). The
authors focus their discussion on the health impacts of increased tobacco trade which
is seen to be facilitated by “significant reductions in tariff and non-tariff barriers to
trade” (p. 5). Woodward and colleagues (2002) supplement this conception by
acknowledging transnational tobacco corporations “as among the strongest
59
proponents of tariff reduction and open markets” (p. 7). Doyal (2002) additionally
highlights the impacts of tobacco trade on women, but while citing ‘liberalization’ as
the culprit behind gendered consumption trends, fails to name specific liberalizing
policies.
Beyond communicable and non-communicable diseases, Singer (2008) relates
import liberalization to an increased flow of drugs, noting a range of health impacts
as well as differential influences across “countries of production, countries of trans-
shipment, and countries of targeted consumption” (p. 469). Singer discusses the
impact of legal drugs such as tobacco and alcohol in relation to global use trends, but
focuses to a large extent on the health impact of illegal drugs, implicating
pharmaceutical companies as an increasingly important “source of illicitly consumed
drugs in developing countries” (p. 471). Here health impacts are framed as a
hindrance to development since they cause losses in productivity, both in terms of
supply and demand.
On the supply side it is noted that workers employed in the production of
illicit drugs are exposed to health threats via exposure to toxic chemicals. The
authors also note particularly poor social relations between workers and their
employers, an inability of lab workers to organize, and the inability of workers to
appeal for help from the government. On the demand side, Smith and Signal note
that “drug use lowers productivity through occupational injuries, the spread of
diseases, and drug overdose” (p. 473). Illicit drug use is also noted to
disproportionately impact the youth, and is associated with various physical and
mental health problems as well as the corruption of public officials and the
breakdown of public institutions and violence.

Summary
It is acknowledged by Bettchar and colleagues (2000) that trade liberalization
may increase the availability of products beneficial to health. However, framework
authors exploring the health impacts of trade liberalization via increased flows of
people and goods (outside of food trade) often cite negative consequences for both
communicable and non-communicable diseases. It is unclear trade liberalization’s
specific role in either of these contexts given specific liberalizing strategies are rarely
identified. Tariff reductions are identified by one framework author as facilitating
increases in communicable diseases, though a clear picture of the relationship is not

60
provided. Reductions in both tariff and non-tariff barriers are identified by another
framework author for their role in increasing non-communicable diseases, although
again, the relationship between the two remains vague at best.
The impacts of tobacco consumption are an area of common concern among
framework authors exploring the links between trade liberalization and
communicable disease. Authors emphasize the role of transnational tobacco
companies in advocating for further liberalization policies and highlight differences
in consumption patterns between men and women.
Finally, import liberalization is noted for its role in increasing the flow of
drugs, with differential impacts to be found at both the global and national level.
Pharmaceutical companies are highlighted as an increasingly important source of
illegal drug consumption.

Agricultural and Food Trade


Another context identified as central to the relationship between trade
liberalization and health is agricultural and food trade. One of the main economic
assumptions behind trade liberalization is that open markets will lower food prices
and lift the incomes of agricultural producers thereby reducing poverty and hunger in
developing countries. Framework authors however, provide evidence for a more
critical analysis of this theory, highlighting differential impacts of open markets on
consumers, farmers, corporate actors, women and the poor. In this policy area, trade
liberalization is seen to impact health through three main pathways: food security,
diet, and food safety.
A range of trade agreements are seen to affect food and agricultural trade, for
example, the WTO Agreement on Technical Barriers to Trade (which contains rules
relating to food quality standards and labelling); the Trade-Related Intellectual
Property Rights Agreement (TRIPS) (which protects seed patents); the agreement on
the application of Sanitary and Phytosanitary Measures (SPS) (which sets standards
for trade-related measures that can be taken to protect human health); as well as
various regional, and bilateral trade agreements (Rayner et al. 2007).
However, one of the most discussed agreements in reference to food and
agriculture trade is the 1994, pre-WTO Agreement on Agriculture (AOA). This
agreement was enacted with the purpose of leveling the playing field of agricultural
trade by requiring governments to eliminate quantitative import restrictions, lower

61
agricultural tariffs, reduce domestic support and eliminate export subsidies
(Leowenson, 2010). Whether the agreement has helped or hindered the elimination
of poverty and hunger in developing countries, and whether further negotiations will
improve or further compromise the attainment of these objectives in poor countries is
however, hotly debated (Diaz-Bonilla et al., 2002).
An oft cited concern is that despite the rhetoric of a level playing field,
industrialized countries continue to protect their agricultural producers while
developing countries face increasing pressure to further liberalize imports. This is
argued to limit the market opportunities for developing countries’ exports, thereby
reducing their economic output and productivity. It is also seen to negatively affect
the livelihoods of agricultural producers in developing countries who must compete
with cheap imports (Labonté et al., 2008). Beaglehold and Yach (2003) note for
example, that “US and European Union (EU) agricultural subsidies limit competition
from primary producers of fresh produce in developing countries and seriously
reduce these countries' national incomes” (p. 904). This reduction in developing
countries' national incomes is seen to indirectly increase their risk of non-
communicable disease epidemics through “changes in household income,
government expenditure, the exchange rate, and prices” (p. 904).
While there is strong consensus that the loss in market access for developing
countries is substantial, framework authors also acknowledge the complexity of the
issue and highlight for instance that “no one-to-one correspondence exists between
the value of subsidies, on OECD's definition, and income lost by agricultural
producers outside OECD” (Labonté & Schrecker, 2006). Both Koivusalso (2006)
and Loewenson and colleagues (2010) further highlight that if agricultural subsidies
are reduced in the developed world, the benefits for agricultural exporters in
developing countries will largely accrue to large transnational producers, especially
since small producers have less access to “capital investments in technology, and
other measures that increase output and lower prices” (Koivusalso, 2006, p. 17).
Koivusalo (2006) additionally highlights that if industrialized countries
reduced their agricultural domestic support subsidies, impacts would “differ between
different groups of developing countries, with the least developed countries
benefiting the least” and middle income countries in Latin-America and Asia,
benefiting the most (p. 17). This is because low income countries are largely net
importers of food and would be negatively affected by any increase in food prices.
62
Koivusalso supports this position by drawing on an analysis by the Food and
Agricultural Organization which shows that of the 46 least developed countries, 31
are net importers of agricultural products.
Beyond the impacts of agricultural trade liberalization in developed countries,
framework authors discuss the differential impacts of agricultural and food
liberalization policies via food security, diets and food safety. The next three
sections will discuss these areas in detail.

Food Security
While it is acknowledged that overall global food security improved between
1960 and 2000, framework authors note that this achievement masks important
regional differences such as increasing food insecurity in Sub-Saharan Africa where
the number of malnourished children under the age of five increased by 14 million
between 1970 and 1997 (Diaz-Bonilla et al., 2002). It is also highlighted that while
overall food security has increased, the proportional rate of decline in
undernourished children and adults in the world has been slow and even reversed
since 2000 (Labtone et al., 2007).
Labonté and colleagues (2007) note that there is little evidence to assess the
impact of trade liberalization on food security, however, the authors highlight a study
by the United Nations Food and Agricultural Organization which examined the
impact of economic reforms on food security in fifteen small and developing
countries. This study found that “trade reform can be damaging to food security in
the short to medium term if it is introduced without a policy package designed to
offset the negative effects of liberalization” (United Nations Food and Agriculture
Organization (2006, p. 75), cited in Labonté and colleagues (2007, p. 45).
Researchers conducting this study highlight the following as instruments of reform,
many of which can be considered as central aspects of trade liberalization:
1) exchange rate regime liberalization;
2) foreign exchange liberalization: elimination of restrictions on foreign
exchange earnings;
3) tariffication of quantitative restrictions on imports and removal or reduction
of import licensing requirements;
4) lowering of tariffs and reduction of their dispersion;
5) reduction or elimination of the use of export prohibitions, licensing

63
requirements and other export restrictions;
6) reductions of export taxes and surcharges;
7) loosening of controls on interest rates and, generally, an increase in real
lending rates. Financial sector reform has often been accompanied by a
widening of financial intermediation margins;
8) reducing the rate of expansion of the money supply through instruments of
monetary policy;
9) increasing the government’s revenue base, strengthening tax collection
efforts, and raising tax rates, especially tariffs on public services;
10) reducing real government outlays.
More specifically, liberalized international food and agricultural markets are
related to greater food insecurity in developing countries via instabilities in global
food prices (Labonté et al., 2008). In this sense, trade liberalization is thus
conceptualized as greater market integration. The authors highlight that between
1995 and 1996 “developing countries faced an average 40 percent increase in their
food import bills, due to poor harvests, demand in China, and the dramatic drop in
food aid levels as US surpluses were absorbed in commercial markets” (p. 45).
Mendonza (2009) however, notes that increased consumption subsidies, along with
import liberalization strategies, can temper the negative impact of food price shocks.
De Vogli and Birbeck (2005) cite evidence which shows how the removal of
food subsidies may negatively impact access to food, as well as increase women and
children's exposure to HIV/AIDs. The authors highlight for example, that “in
Zambia, after the removal of subsidies in 1985, the price of maize meal rose by 50%.
In Zimbabwe, after eliminating food subsidies, the cost of living for lower-income
urban families rose by 45% between mid-1991 and mid-1992” (p. 111). Increases in
food prices are not only noted for making the accessibility of food more difficult, but
also for reducing the amount of money families have available for other basic
commodities.
The authors highlight that as the primary providers of meals, women in Sub-
Saharan Africa bear the majority of this burden. Urban women in particular are seen
as disproportionately impacted by these policies since the majority of their food is
purchased rather than produced. Rural women are also seen to experience largely
negative impacts from these policies as they “barely cover their subsistence with the
food they produce, and with the increasing cost of food they can no longer afford to
64
purchase other foods” (De Vogli & Birbeck, 2005, p. 111).
With less money for food and basic commodities, De Vogli and Birbeck
highlight that women often rely on commercial sex as a survival strategy. This
increases women's exposure to sexually transmitted diseases and sexual abuse. It also
increases the vulnerability of infants to HIV/AIDS, and predisposes children of poor
mothers to HIV by forcing them to abandon school in search of work. It is noted that
children in impoverished families are likely to be abandoned and often find
themselves living and working on the street, “where they may be forced into
prostitution to exchange sex for money, goods, food or shelter” (De Vogli & Birbeck
2005, p. 111).
Finally, Labonté and colleagues (2007) relate the liberalization of agricultural
export markets in developing countries to food insecurity at the household level. The
authors highlight research done by the International Food Policy Research Institute in
the 1980s. This research “examined the nutritional impact of a series of cash
cropping schemes in ten developing countries” (p. 45). Without explicating the
particular liberalizing policies that promoted these schemes, the authors note that
“the findings suggested that cash cropping generally results in higher incomes and
spending on food, but has relatively small impact on energy intake, and in most
cases, little or no impact on childhood malnutrition” (p. 45). This study also
illustrates how the concept of food security should take into account nutritional
adequacy. Nutrition is discussed in greater detail in the next section.

Nutrition
Globalization is argued to have brought about changes in the “quantity, type,
cost and desirability of foods available for consumption”, altering both consumption
patterns and nutritional outcomes (Hawkes, 2006, p. 2). This section will explore the
pathways presented by framework authors which highlight trade liberalization's role
in these processes.
Within the agricultural and food trade literature, framework authors focus on
what is termed the 'nutrition transition' to explain much of the relationship between
trade liberalization and poor nutritional outcomes. Coined by Popkin (1998), the
nutrition transition can be understood as the shift from traditional staples, such as
cereals and complex carbohydrates, to more energy dense foods and refined
carbohydrates. Increased consumption of foods outside the home also characterizes

65
this transition. Rayner and colleagues (2007) additionally highlight that “[t]here may
be a case for unbundling the nutrition transition from one single process into three,
namely diet, the physical environment and culture, recognizing that each of these
transitions overlap, combine and amplify each to the other” (p. 70).
The most commonly identified pathways linking the trade liberalization to the
nutrition transition are, the increasing reliance of countries on food imports and the
rise of transnational food companies. While the former is seen to be a result of
various import and export liberalizing strategies, the latter is largely seen to be a
result of various FDI liberalizing policies. The nutrition transition is noted for its
differential impacts across low, middle, and high income countries as described
below.
In developing countries, Rayner and colleagues (2007) highlight the dual
burden of continuing malnutrition and simultaneous rises in diet-related chronic
diseases. Trade liberalization is seen to affect the food supply chain in developing
countries “at varying levels of complexity that can be characterized as follows: food
imports and exports, the local/global balance of the internal dynamics of the food
supply chain, FDI in food processing and retail and commercial promotion of food”
(Rayner et al., p. 70). Without identifying specific liberalizing policies, these factors
are in turn related to an increased reliance on food imports; a shift from local to
'value added' processed foods; an increased availability of highly processed foods
through FDI and supermarket enlargement; and a shift in cultural expectations via
advertising. The need for stronger food governance is emphasized.
Thow (2009) provides one of the most comprehensive understandings of trade
liberalization’s impact on the nutrition transition. Through a systematic review of
nutrition and liberalization related literature, the author uses a range of WTO
agreements to outline three broad policy areas through which trade liberalization may
impact diets and the nutritional transition in developing countries: trade in goods;
trade in investment and trade in services and support/protection for domestic
production and industry.
Within the trade in goods pathways, both import facilitation and export
promotion policies are identified as primary determinants of health related to trade
liberalization. Under import facilitation, WTO agreements are emphasized for their
role in liberalizing trade in goods through the reduction of both tariff and non-tariff
barriers to trade. Highlighted examples of non-tariff barriers to trade include
66
“quantitative restrictions, import licensing, variable levies, import quotas, and
technical barriers” (p. 2153). GATT is identified as the key agreement within this
pathway since the schedule of commitments attached to it indicates when countries
will reduce their tariffs and by how much. The SPS and TBT agreements are also
identified for their role in committing countries to remove “restrictions and import
regulations that are disguised protectionist measures” (p. 2151). Finally, the Most
Favoured Nation (MFN) principle is highlighted for dictating that “all nations should
be given equal treatment, meaning that countries cannot use trade barriers to
discriminate based on the source of imports” (p. 2152).
Thow highlights two key implications of the removal of barriers to food
imports: firstly, an increase in food imports, and secondly, a decrease in the costs of
importing food, and thus increased competition between imported and locally
produced food.
According to Thow, increases in food imports have meant a shift in countries'
food cultures through the increased availability of both processed and high value
goods, such as fruit and dairy products. Additionally, the availability and low cost of
animal feed, “(in many cases, from developed countries with subsidised
production)”, has meant greater consumption of meat products (p. 2151). While
Thow emphasizes the positive effects this has had in areas with problems of
undernutrition, the negative implications this can have in countries where
undernutrition is not a concern, are also highlighted. Thow also associates increased
openness in food markets with a 'dumping' of low quality foods in developing
countries. Finally, the decreased costs of imported food, especially processed foods,
and increased competition from local food providers, is seen to increase the
consumption of unhealthy, refined foods.
Thow highlights that “export promotion is also an important component of
trade liberalisation” (p. 2153). While the various forms of support that the author
relates to trade liberalization aren’t explicated, the International Trade Centre is
highlighted as a major promoter of export promotion policies. The promotion of
export industries is seen to be associated with an increase in land usage for cash
crops (crops for export) in developing countries, resulting in less production and
consumption of traditional domestic staples.
Currency devaluation is also highlighted as a trade liberalizing strategy and
seen to promote export industries by decreasing the costs of goods for purchasing
67
countries. It is noted that this strategy may increase the costs of imported goods and
both negative and positive diet-related implications of this process are highlighted.
Negatively, increases in the cost of food can result in reduced food consumption and
dietary diversity. On the other hand, “if domestic production has the capacity to
respond, currency devaluation can have positive dietary effects through increasing
the availability and consumption of locally produced goods” (p. 2154).
Next, Thow relates trade liberalization to investment and trade in services on
the basis that “policies associated with trade liberalization also act to encourage
investment, as a means to economic growth” (p. 2154). In this pathway, Thow
highlights three agreements for their roles in promoting investment and trade in
services through trade-liberalizing policies:
· GATS, “which includes changes in regulations such as reductions in
restrictions on foreign ownership of companies”,
· TRIPS which enhances related regulatory change via protection of
intellectual property rights, and
· TRIMS which “contains commitments to remove any restrictions on where
companies source their inputs (e.g. domestically rather than from imports)”
(p. 2154).
Policies related to these agreements are seen to effect food systems by
increasing food industry investment, increasing competition, increasing the
development of food technology (and technology transfer), creating more new foods
and food service establishments, and increasing food marketing. These food system
effects in turn, are seen to increase the availability of processed foods, stimulate the
local industry, improve food storage and safety, and increase the availability and
awareness of “high profit margin novel foods” (p. 2152). These mechanisms are then
described as increasing people’s consumption of processed, refined and pre-prepared
foods.
In Thow’s final pathway, the removal of farmer subsidies and the removal of
import tariffs on goods that are also produced locally are highlighted as key aspects
of trade liberalization. Agreements identified to be associated with these policies are
the AOA, under which “developing countries agreed to cut subsidies by 13%”, the
Agreement on Subsidies and Countervailing Measures which regulates the use of
subsidies and prohibited subsidies (“those that are attached to conditions such as the
use of domestic materials by industry”), and the SPS and TBT which also address the
68
conditions under which subsidies can be used as a form of protection for local
production and industry (p. 2155).
Since the impact of tariff reductions on food consumption is previously
addressed, the removal of subsidies is the focus of this pathway. First, the elimination
of prohibited subsidies is seen to reduce the overall cost of processed foods, and
increase the attractiveness of investment into the food industry, “thus magnifying the
effects of investment-related measures discussed earlier” (p. 2155).
The impact of the removal of agricultural subsidies is seen to vary across
countries. The removal of agricultural subsidies in the developed world for instance,
may increase the cost of imported goods elsewhere. However, the removal of these
subsidies, combined with supplementary trade measures, might reduce the problem
of low priced goods being 'dumped' on developing country markets. Subsidy
reduction might also impact food availability in developing countries by removing
incentives for production. The author notes however that “there is little available
literature on the outcome of such policy changes for diet” (p. 2155).
Using the concepts of 'dietary convergence' and 'dietary adaptation', Hawkes
(2006) explores how global market integration influences dietary patterns in middle
income countries. Hawkes adopts definitions of these concepts from Kennedy,
Nantel and Shetty (2004) where dietary convergence is understood as “increased
reliance on a narrow base of staple grains, increased consumption of meat and meat
products, dairy products, edible oil, salt and sugar, and a lower intake of dietary
fibre”, and dietary adaptation is defined as “increased consumption of brand-name
processed and store-bought food, an increased number of meals eaten outside the
home and consumer behaviours driven by the appeal of new foods available”
( Kennedy, Nantel, & Shetty, 2004, p. 9 cited by Hawkes, 2006, p. 3).
Three process are highlighted for their role in facilitating dietary convergence
and adaptation: “(I) the production and exchange of goods in the form of agricultural
production and trade; (II) the flow of investment across borders in the form of foreign
direct investment in food processing and retailing; and (III) the global
communication of “information” in the form of the promotional food marketing” (p.
3). A noteworthy strength of this framework is that each of these processes is
supported with evidence from case studies.
Within the production and exchange of goods pathway, trade liberalization is
understood as increased 'market-orientation'. Increases in market-orientation are seen
69
to have occurred in middle income countries within the context of structural
adjustment, through regional and bilateral trade agreements, as well as through
GATT and the AOA. Increased market-orientation in general is seen to have
increased flows of food trade, foreign investment and the size of transnational food
companies. These processes are in turn noted to “have altered the supply of foods
associated with the nutrition transition” (p. 3). The integration of vegetable oil into
Brazilian, Chinese and Indian markets is explored and used to exemplify the author's
claims. In this case study, liberalization policies in Brazil, such as lowered import
tariffs and export taxes, in combination with investment liberalization and currency
devaluation, are noted for their role in the convergence of vegetable oil consumption
in China and India.
Rather than being seen as a component of trade liberalization, increased flows
of investments across borders, is conceptualized as a globalization process, related to
but distinct from trade liberalization. However, within the global communication of
“information pathway, trade liberalization might be understood as the opening of the
communications market, which is seen to occur “due to some domestic deregulation
and trade agreements” (p. 10). Along with various other incentives, the opening of
the communications market is identified as driving the globalization of food
marketing which is in turn highlighted for its role in promoting energy-dense and
highly-processed foods, encouraging both greater consumption and production of
these products.
Hawke uses Thailand as a case study to explore processes in this pathway and
highlights the Thai advertising and promotions industry for its dynamism, which is
seen as “both related to the country's tradition of openness to trade and investment”
(p. 10) Foreign ownership of advertising agencies in Thailand, for example, is
described as unrestricted and advertising campaigns as only somewhat regulated. An
increased presence of foreign brands is seen to have been facilitated through free
trade agreements and created the need for product differentiation through advertising.
This market openness is then related to the dietary convergence of “processed
savoury and sweet snacks” (p. 10).
Turning to high income countries, Smith and Signal (2009) highlight how the
nutrition transition taking place in low and middle income countries has increased
global dairy demand. This is associated with increasing costs of dairy products and
negative nutritional health outcomes in New Zealand. While it is acknowledged this
70
trend has had positive impacts on New Zealand's dairy exporting farmers and
producers, and thus positive impacts on the national economy, these advantages have
come at the expense of local consumers, “especially those which are
socioeconomically disadvantaged” (p. 2).
Trade liberalization is conceptualized broadly by drawing on the notions of
market-orientation and openness. Reform measures associated with it are the
“removal of government subsidies, reduction of import tariff and non-tariff barriers,
removal of control on interest rates, wages and prices, restructuring and sale of
government assets and reform of tax structures including the application of a neutral
good and services tax” (p. 3).Drawing on Woodward and colleagues' (2001)
globalization and health framework, the authors highlight both direct and indirect
pathways between trade liberalization, and nutritional health outcomes related to
diary consumption patterns. Reform measures associated with trade liberalization, in
conjunction with the expansion of large supermarket chains, are seen to have
increased the price of milk in New Zealand, especially relative to cheap sugar-
sweetened carbonated beverages. This is in turn directly related to greater nutritional
health inequalities, with lower income New Zealanders more likely to choose
cheaper, nutritionally poor beverages. Changes in milk supply and limited purchasing
points for milk are also associated with the identified reform measures but indirectly
related to nutritional health inequalities via national and household economies.

Food Safety
Trade liberalization is broadly related to food safety concerns through
increased flows of food products and via food safety standards. Increased flows of
food are highlighted for increasing the susceptibility of developed countries to
dangers of food-borne illnesses. More uniform safety standards are seen as a
mechanism combating this threat. However, these standards are also seen as
hindering the exporting capacities of developing countries and especially the poor,
since they often lack the institutions and infrastructure necessary for compliance
(Diaz-Bonilla et al., 2002).

Summary
In the context of agricultural and food trade, framework authors challenge the
assumption that trade liberalization improves health through the reduction of food
prices and by lifting the incomes of agricultural producers. They do this first by

71
demonstrating how further reduction of agricultural subsidies in developed countries
may come at the expense of consumers in lower income countries in the form of
higher food prices. While it might be argued that the gains in market access for
agricultural producers outweigh the costs of increased food prices, framework
authors demonstrate the difficulty in making this claim given that the beneficiaries of
such reductions are expected to be large transnational companies.
In terms of food security, framework authors highlight that improvements
over the past three decades have been slow and in some regions reversed. While
import liberalization may lower food prices, food security may be hindered by
instabilities in global food prices and through the reduction of consumption
subsidies. Reduction in consumption subsidies also impacts how much money
families have available for other basic commodities and disproportionately impacts
women who may turn to commercial sex as a survival strategy, placing themselves
and their children at risk of sexual abuse, sexually transmitted diseases and
HIV/AIDS. Furthermore, despite reductions in food prices, framework authors
highlight the negative impacts trade liberalization has on health by shifting the type
and quality of food available to people with the poor suffering the greatest. Finally,
food and agricultural trade liberalization also impacts health outside of food prices
and agricultural wages through increased trade flows and related food safety
concerns.

Structural adjustment policies


Another key context through which framework authors discuss the health
impacts of trade liberalization is structural adjustment policies. Structural adjustment
policies are discussed in reference to loan conditionalities imposed by either the IMF
or World Bank and often in reference to neoliberal ideology and the Washington
Consensus. Both Labonté and colleagues (2007) and Labonté and Schrecker (2006)
acknowledge the similarities between adjustment policies and Poverty Reduction
Strategy Papers which emerged in 1999 as a new requirement for countries to receive
grants or loans from the World Bank or other development agencies. Framework
authors however, have yet to explore their impact on health with reference to trade
liberalization.
Trade liberalization is universally seen as a fundamental aspect of structural
adjustment policies, along with financial liberalization, government spending

72
reductions and various privatization and macroeconomic stability policies.
Many framework authors highlight the difficulty in linking structural
adjustment policies to health. One reason is because these policies are undertaken in
countries already in distress, making it difficult to know to what degree factors
outside of conditionalities are responsible for health outcomes. Stuckler and Basu
(2009) for example, highlight government corruption as one factor outside of
structural adjustment policies that might play a role in influencing health outcomes.
Labonté and Schrecker (2006) as well as Cheru (2002) additionally highlight the
difficultly in separating the specific impacts of adjustment policies from those of
market and globalization pressures in general. It is noted however, that if we are
interested in knowing how market forces impact health, it is less important to be able
to attribute how much of a policy is due to adjustment in particular, and how much to
market forces in general (Labonté & Schrecker, 2006).
Perhaps more relevant to the purposes of this review is the difficulty in
separating the impacts across the range of policies adopted under adjustment
(Labonté & Schrecker, 2006). This task becomes even more challenging when we
consider that effects of policies are often “indirect, acting through policy channels
such as privatization, liberalization, and stabilization, for which global health
evidence has been growing but is generally lacking (Stuckler & Basu, 2009, p. 774).
Despite these difficulties, studies have attempted to assess the impacts of
adjustment, though few have done so by isolating the health impacts of trade
liberalization. In terms of general health impacts Cheru (2002) highlights that while
some countries have witnessed growth in the context of adjustment, few have seen it
sustained. Meier (2006) elaborates on this by highlighting that states experiencing
growth “have often done so at the expense of widening inequality within societies
among the most poor and vulnerable” (p. 720).
Framework authors Stuckler and Basu (2009) review the evidence linking
IMF imposed adjustment to global health and find no evidence of positive health
effects. While they do find some evidence for neutral effects, most of the effects are
found to be largely negative (Stuckler & Basu, 2009). Other framework authors
relate adjustment policies in general to: the deterioration of public goods for health
(Fox & Meier, 2009); weakened health care systems (Stuckler & Basu, 2009);
“Impeded efforts to control tobacco, infectious diseases, and child and maternal
mortality” (Stuckler & Basu, 2009 p. 771); cuts in health budgets and the imposition
73
of user fees in education and health services (Cheru, 2002; Yaşar, 2010); increased
poverty and income inequality (Labonté & Schrecker, 2006; Yaşar, 2010); increased
rates of sexually transmitted diseases (Meier, 2006; Yaşar 2010); worsening
nutritional outcomes (Cheru, 2002; Labonté & Schrecker 2006, Meier, 2006;); and
increases in unemployment and deteriorating living conditions (Labonté &
Schrecker, 2006; Meier, 2006).
Few framework authors highlight the isolated health impact of trade
liberalization in the context of adjustment policies. However, both Cheru (2002) and
Labonté and Schrecker (2006) relate structural adjustment to reductions in
consumption subsidies. Cheru (2002) associates these reductions with reductions in
the realization of the right to food and Labonté and Schrecker (2006) note negative
impacts in terms of nutrition and household income.
In exploring the impact of adjustment policies on the vulnerability of women
and children in Sub-Saharan Africa to HIV/AIDS, De Vogli and Birbeck (2005)
provide the most comprehensive conceptualization how trade liberalization polices,
within the context of adjustment, impact health.
The authors’ framework is composed of five different pathways. Of these, the
first and the third incorporate elements of trade liberalization. The five pathways are
1) currency devaluation and the removal of food subsidies; 2) privatization; 3)
financial liberalization and trade liberalization; 4) user-fees for health services; and
5) user-fees for education.
The first pathway, currency devaluation and the removal of food subsidies, is
illustrated in figure 13. As discussed previously, in this pathway, the removal of food
subsidies has a largely negative effect on the prices of basic items such as food,
housing and transportation. This comes largely at the expense of women who are
responsible for ensuring the provision of food and basic commodities. The authors
note that impoverished women are forced to adopt risky survival strategies in an
attempt to acquire these goods, often through commercial sex market. Impoverished
women and children are also more susceptible to non-consensual sex and domestic
violence.
In the third pathway, illustrated in figure 14, trade liberalization, in
conjunction with the largely negative impacts of financial liberalization, is seen to
have mixed effects on the production capabilities of small farmers. Subsidy reform is
highlighted as having mostly negative impacts on the production of small farmers by
74
weakening their economic activity. It is noted for example that in Ghana, “[f]ertilizer
reform that involved the removal of subsidies increased the price of insecticides,
fungicides, and spraying machines make these inputs unaffordable for most small
farmers” (p. 113). Trade liberalization is also noted to have potentially positive health
impacts through import liberalization and export promotion policies which “may
reduce imbalances in the import-export ratio resulting in economic growth and
reduced poverty” (p. 113).

Figure 13 Potential Impact of Adjustment Policies on Vulnerability of Women and Children to


HIV/AIDS in Sub-Saharan Africa (De Vogli & Birbeck 2005)

75
Figure 14 Potential Impact of Adjustment Policies on Vulnerability of Women and Children to
HIV/AIDS in Sub-Saharan Africa (De Vogli & Birbeck 2005)
When the production of small farmers declines, often due to displacement by
large-scale producers, migration of males in search of employment to urban areas is
seen to increase. A handful of studies are drawn on which suggest that migrant men
are more likely to engage in risky sexual behavior. This in turn is seen to increase the
risk of women’s exposure to HIV/AIDS given they “are often unaware of the HIV-
related risks involved in consensual unsafe sex when their partners return home” (p.
114). Moreover, due to their financial dependency, it is also noted that women may
have little power to negotiate safe sex, and be at increased risk of domestic violence
and physical abuse. Finally, women abandoned by their migrating partners may enter
the commercial sex market as a survival strategy, increasing their exposure to sexual
transmitted disease.

Summary
It is widely acknowledged that trade liberalization is a fundamental aspect of
adjustment policies, although the relationship between trade liberalization and PRSPs
is less explored. Attempts at linking adjustment policies to health are rife with
difficulties, especially when trying to separate the impacts of trade liberalization
from other adjustment related policies.
It is also acknowledged that while structural adjustment policies in general
may lead to growth, sustainability of this growth is uncertain, moreover, increases in
growth have often been accompanied by increases in income inequality. While
adjustment policies in general are related to a range of health concerns, only a few
76
framework authors isolate specific trade liberalizing policies within the context of
adjustment and health. Among those that do, reductions in consumption subsidies are
a shared concern. Such reductions are related to matters of food accessibility,
nutrition, household income, and women and children’s vulnerability to HIV/AIDS.
It is also recognized that trade liberalization may have a positive impact on health
when import liberalization and export promotion policies reduce imbalances in the
import-export ratio and thereby increase growth and reduce poverty.

Labour Markets
Descriptions of trade liberalization’s impact on health through its influence on
labour markets vary in complexity based on the intention of the framework.
Drawing on the work done by the Globalization Knowledge Network of the
WHO CSDOH, Labonté and colleagues (2007) provide a broad description of the
relationship between labour markets, trade liberalization and health equity. The
authors note that the “reorganization of production and service provision across
multiple national borders by transnational corporations” is a central element of
globalization made possible by trade liberalization (p. 3). However, the mechanisms
through which trade liberalization supports these changes is not delineated.
The authors highlight three related issues identified in recent globalization
and labour market literature. First, that “ a genuinely global labour market is
gradually emerging, driven in part by the integration of India, China, and the former
transition economies into the global marketplace” (p. 3). Second, that “the need to
appear ‘business-friendly’ may limit governments’ ability to adopt and implement
labour standards, health and safety regulations, and other redistributive social policy
measures” (p. 3). Finally that “production is being fragmented and reorganized
across multiple national borders in global commodity chains or value chains, in
which each element of production is located where it contributes most to overall
returns while reducing financial risks” (p. 4).
The health consequences of these processes are seen to be related to:
“growing economic and social inequalities among workers; falling wages and
deteriorating working conditions for many or most workers; eventual loss of some
jobs to jurisdictions, notably China, which can offer even lower labour costs;
increased workplace hazards and industrial pollution exposure to which is in turn
related to labour market position” (p. 4).

77
The authors note that these consequences are not just the result of economic
integration, and that distributions of gains and losses depend on ability of workers,
firms and national economic policies to carve out niches in global value chains. It is
also noted that the winners are typically those with “access to the necessary financial
resources, skills (‘human capital’), and technology” (p. 4).
In this framework, the reorganization of labour markets is also seen to
disproportionately impact women. A study by the United Nations Research Institute
for Social Development is presented to have found that women’s work in export
industries can increase their income and entitlement to benefits; however, the authors
note that these gains are also more vulnerable to economic crises and to labour
market flexibility pressures.
Another broad framework is presented by Muntaner and colleagues (2010).
This framework draws on work done by the Employment and Working Conditions
Knowledge Network of the WHO CSDOH and outlines “the mechanisms leading
from globalization and related macro-scale social process to health inequalities by
way of employment conditions” (p. 57). Two flow charts are used to illustrate the
overarching conceptual framework, one at the macro-level and one at the micro-level
(see figures 15 and 16 respectively).

Figure 15 Macro-level framework and policy entry points (Muntaner et al. 2010)

78
Figure 16 Micro-level framework and policy entry points (Muntaner et al. 2010)
Within the micro-level, the authors provide a comprehensive analysis of the
pathways “between employment conditions and health inequalities through a number
of behavioural, psychosocial, and physiopathological pathways” (Benach et al.,
2007, p. 32). At this level, four main categories of risk exposure (physical, chemical,
ergonomic, and psychosocial) are seen to be mediated by social mechanisms and
influenced by six different types of employment conditions (full employment,
unemployment, precarious employment, informal employment, child labour, and
slavery & bonded labour).
At the macro-level, the framework seeks to contextualize employment
relations by illustrating the role of power relations in influencing both labour market
and welfare state policies. In the final report of EMCONET to the WHO CSDOH
(Benach et al., 2007), a historical account of the macro-level context is given. The
authors note that “[k]ey influences affecting changes to employment dimensions over
the past thirty years have been the growing influence of powerful corporations and
abandonment of Keynesian economic policy and social compacts in favour of neo-
liberalism” (p. 102). Further, the authors note that “[p]olicies and practices flowing
from the belief that competitive markets deliver the best outcomes include rejecting
public spending as a method of managing unemployment rates; removing barriers to
trade, commerce and competition; tax cuts; privatization; corporatism; competitive
tendering; outsourcing/off-shoring; downsizing; and (more rhetorically than in
practice) small government” (p. 102).

79
It is acknowledged that in wealthy countries these policies and practices have
resulted in “ a reduced welfare net for the unemployed and disadvantaged; job losses
in the public sector; growth in job insecurity and precarious employment; a
weakening (in practice) of regulatory protections; and the historical re-emergence of
an informal economy, including home-based work and some child labour” (p. 102). It
is noted that
“[i]n poor countries the dominance of neoliberalism has
translated into a new model of economic development oriented
toward productivity and supplying products to global markets
(including “race to the bottom” working conditions to attract
overseas capital and the use of corporate-friendly low regulatory
special export zones) irrespective of the effects on local
communities, such as decreased domestic food production, rural
dislocation, and social instability” (p. 102).
On the basis of this account, trade liberalization can be understood as part of
the macro-economic environment shaping labour markets, welfare states, and
employment relations, though specific mechanisms through which this occurs are not
explored. Trade liberalization is also not conceptualized in relation to power
relations, which occupies the crux of the macroeconomic environment. However, the
authors do characterize export processing zones (EPZs) on the basis of their
“relentlessness hostility to trade unions” which demonstrates one pathway through
which trade liberalizing strategies impact the domain of power relations (p. 50).
Other framework authors help to provide a more comprehensive
conceptualization of trade liberalization in relation to labour markets and health.
Corrigall and colleagues (2008) for example, explore the relationship between
global trade and mental health. Like Muntaner and colleagues (2010), the authors
highlight the role of work-stress in mediating a range of health outcomes including
depression, aggression, unhealthy lifestyle habits, alcohol abuse and musculoskeletal
disorders. Trade liberalization is seen as part of the Washington Consensus, and “a
key principal of WTO agreements” which seeks to “ensure that trade is not
unnecessarily restricted by tariff or non-tariff barriers” (p. 336). Whereas tariff
barriers are understood as those which restrict trade through financial methods, such
as import taxes, non-tariff barriers are seen to be those which “refer to laws and
regulations that affect trade such as those based on a threat to public health” (p. 336).
80
The authors relate the opening of national economies in general to changes in
the nature of the working environment both in industrialized countries and middle- to
low-income countries. In industrialized countries, impacts are characterized by job
losses, whereas in middle to low income countries, impacts are characterized by the
increased global supply of unskilled labour, the depression of wages, and decreases
in the quality of work. The authors note that while the specific impacts of trade
policies or agreements are hard to quantify, “evidence indicates job demands have
increased while job control has decreased and workers are exposed to a greater
number of occupational hazards” (p. 345). In middle- to low-income countries these
problems are noted to be exacerbated by weak monitoring and regulatory
enforcement capacities, as well as by “weakly organized labour movements and the
overriding need for economic growth” (p. 436). It is also noted that those most likely
to work in the informal sector, and therefore be exposed to greater occupational risks
and job insecurity, are women and children.
On the other hand, Loewenson (2001) highlights that globalization has largely
benefited industrialized countries, however, the author is more aligned with Corrigal
and colleagues (2008) in noting that the greatest burdens have fallen on those in
middle- to low-income countries. The author highlights that “[f]or the large majority
of workers in the less-industrialized countries, liberalized trade has been
accompanied by transfer of obsolete and hazardous technologies, chemicals, process
and wastes, including asbestos and pesticides no longer produced or used in
industrialized countries” (p. 864). Health is therefore seen to be impacted through a
range of physical, chemical, and psychosocial influences such as “mechanical,
electrical and physical hazards”, environmental pollution, and psychological stress
(p. 864).
Again similar to Corrigall and colleagues (2008), the author notes that trade
liberalization “has also been associated with an increase in assembly line, low-
quality jobs, with minimal options for advancement, and a growth of insecure, casual
employment in a small-scale informal sector” (p. 864). Social protection, as a system
for protecting workers’ is noted to be rarely provided by these types of employment
shifting the “liability for working conditions to the worker” (p. 864).
Trade “liberalization is also associated with deregulation of production laws,
adding to pressures on occupational health standards” (p. 866). EPZs are highlighted
as an example of how health is impacted “under liberalized tax and trade regimes”
81
(p. 864) and is the only example of how a specific trade liberalizing policy influences
health. The authors note that in some EPZs occupational health laws do not apply,
breach of laws are penalized at “absurdly low levels” and criminal sanctions rarely
evoked (p. 865).
Furthermore, both women and migrant workers are acknowledged to be
disproportionately impacted by labour market changes. While new production
patterns are seen to have increased women’s participation in the labour force, with
increased income generating opportunities, women are also noted to work more often
than men in insecure employment and have less access to important resources such
as “credit, land, services training, and other production inputs” (p. 864). They are
also noted to mostly occupy “low-skilled, low-paid jobs where rates of union
membership are low” and to carry the double burden of both employment and
household work (p. 864). These burdens come at the cost of increased psychological
stress, decreased time for rest, and increased exposure to occupational risks.
Elaborating on Loewenson’s (2001) characterization of the impacts trade
liberalization on women’s health, are the following four framework authors: Doyal
(2002), Grown (2005), Yaşar (2010) and Loewenson and colleagues (2010). Most of
these framework authors incorporate trade liberalizing pathways within broader
frameworks of globalization; Grown (2005) however, exclusively looks at the
pathways between trade liberalization and women’s reproductive health. Doyal
(2002) explores the impacts of globalization on women’s health in general, Yaşar
(2010) explores the impact of globalization on women’s exposure to HIV/AIDS in
Cambodia, and Loewenson and colleagues (2010) explore the impact of globalization
on women’s nutritional outcomes in Sub-Saharan Africa.
Except for Grown, these authors neglect to explicitly define trade
liberalization and instead appeal to ideas of ‘free trade’, ‘free markets’, ‘greater
integration’, etc. Providing a more comprehensive understanding of trade
liberalization, Grown (2005) highlights that “[t]rade liberalization can occur through
an autonomous decision of a government to remove or reduce barrier to exports,
eliminate subsidies to domestic industries and firms, and privatize goods and
services, all of which are intended to result in the freer movement of capital, goods,
and labor across borders” (p. 29). Grown also highlights that “trade liberalization
more commonly occurs as a result of multilateral trade negotiations of the World
Trade Organization, through regional or bilateral trade agreements, or via conditions
82
attached to IMF or World Bank loans” (p. 29). This latter contextualization of trade
liberalization is shared by the other authors, who also emphasize the particular roles
of neoliberalism and corporate power in shaping negotiations and agreements.
All of these framework authors acknowledge that women’s employment has
increased as a result of globalization. Grown (2005), however is the only one to note
trade liberalization’s specific role in this trend and distinguishes between the growth
in women’s share of employment in semi-industrialized countries, those which are
agriculturally oriented, and those which are oriented toward the service sector. The
author highlights that “[b]oth increased foreign investment and elimination of export
tariffs have increased the demand for female labour and provided women access to
manufacturing, services, and some types of agricultural employment in many
countries” (p. 36). In semi-industrialized countries, women’s work has increased in
export oriented industries. It is noted however, that once “these economies mature,
the process of feminization of export employment may decline or even reverse” (p.
36). The promotion of cash crops as a liberalization strategy in agriculturally oriented
economies is seen to have increased women’s work as “seasonal, contract workers or
as labourers on husbands’ or relatives’ land” (p. 36). Finally, women are seen to
constitute a large share of export workers in economies which emphasize service
exports, such as informatics and tourism.
Across these frameworks, women’s health is understood to be impacted by
factors both within and outside the workplace. Grown (2005) highlights four types of
work environments which have different impacts on women’s reproductive health:
factory employment; home-based work; sex work; and part-time and seasonal work.
These categories provide a useful way of summarizing pathways to women’s health
offered by other framework authors.

Factory Work
Unsurprisingly, trade liberalization is related to women’s health largely
through employment in EPZs. Authors note that this type of employment can be both
positive and negative for women’s health. Benefits come from women’s better access
to income, and thus “increased decision making and control over household
spending” (Loewenson et al., 2010, p. 11). Loewenson and colleagues (2010)
additionally highlight the benefits this has on children through improved nutrition,
although note that income does not always increase for workers in Sub-Saharan

83
Africa who are without education.
Disadvantages of factory work include job and income insecurity, as well as a
lack of rights, and social benefits. Factory work can also increase food insecurity
especially when work requires relocation to urban areas where women are faced with
“competing demands for spending, higher costs of food purchases and limited
possibilities …to produce food in urban environments” (p. 11). In terms of
reproductive health, factory work is seen to put women at “greater risk of early
sexual activity and sexual harassment” (Grown, 2005, p. 37).
Factory work also has important occupational and environmental health
impacts through increased exposure to both “old and new hazards” (Doyal, 2002, p.
244). New hazards are associated with industry shifts towards modern technology
and include for example, repetitive strain injury stemming from computer based
work. Traditional hazards include “strenuous, monotonous and ergonomically
unsound” work (Doyal, 2002, p. 241) as well as “high levels of machine-related
accidents, dust, noise, poor ventilation, and exposure to toxic chemicals” (Grown,
2005, p. 37). These factors are also seen to increase the stress women experience, and
can affect women’s reproductive health through miscarriage, pregnancy problems or
poor fetal health (Grown, 2005).
Furthermore, it is important to note that women’s share of employment is not
always sustained given increased capital mobility and labour market flexibility
(Doyal, 2002). Yaşar (2010) for example, notes that the garment industry is the
single largest employer in Cambodia but that it is under constant threat by
liberalizing policies promoting international competition. Because other industries do
not have the capacity to absorb any labour surplus resulting from what is described
as inevitable job loss, the author warns that unemployment is likely to drive many
female garment workers into the sex trade, especially when “they are not able to
return to rural Cambodia, primarily because of unpaid debt and stigma (p. 14).

Home-Based Work
Home-based work is also associated with trade liberalization, although the
mechanisms relating the two are never fully explored. Predominately employing
women, home-based work is characterized as insecure due to a lack of formal
contracts and also due to a lack of access to leave or health benefits (Grown, 2005).
Women who must juggle this kind of work with childcare are seen to face additional

84
burdens (Doyal, 2002). In addition, the home environment itself can pose unique
occupational hazards, stemming from for example, home storage of pesticides, or
cooking stove pollution. Moreover home-based workers comprise an ‘invisible
workforce’, which often means their needs are overlooked in the formulation of
labour market or health policies (Grown, 2005, p. 38).

Sex Work
In addition to the pathways previously presented by De Vogli and Birbeck
(2005), Yaşar (2010) argues that trade liberalization exacerbates inequalities and
creates “an environment characterized by insecure paid employment, no safety nets,
sexual harassment, landlessness, and the like” (p. 5). This environment is seen to
increase women’s likelihood of entering the commercial sex market and their
exposure to HIV/AIDS as well as other sexually transmitted diseases. Additionally,
the growth of the service sector and tourism industry are noted by Grown (2005) to
increase the demand for sex workers thereby increasing women’s risk of sexually
transmitted diseases.

Part-time and Seasonal Work


Grown (2005) explores the impact of part-time and seasonal work on
women’s health and notes that women working in this type of work “are often paid
by the piece, receive no incomes for out of season unemployment, and are not
covered by employer based health insurance plans” (p. 38). The author notes that this
type of employment is found both in factory and agricultural work.
Agricultural work is less explored by Grown (2005) but considered by
Loewenson and colleagues (2010) for its impact on women’s nutritional outcomes.
Changes in economic trends in general are associated with “changes in support for
women as farmers, in the quality and security of non-farm wage employment, and in
demands on women in their household occupational roles” (p. 11). Trade
liberalization’s specific role in these changes however, is not easily identifiable.
Together with changes in non-agricultural employment, these trends are seen
to impact the nutritional health of women in Sub-Saharan Africa through the
following pathways: “diminished access to purchased or grown food; competing time
demands between food sourcing and preparation and other household activities;
increased women’s (and often children’s) physical burdens and energy expenditures
and exposure to disease; mental and emotional stress and neglect by women of their
85
own health and wellbeing; female children dropping out of school to contribute to
household labour, undermining opportunities and capacities for health, and women’s
lack of control over finances and decision making, undermining health-promoting
food choices and timely health-seeking behaviors” (p. 12).

Other Health Impacts


As Loewenson and colleagues (2010) make clear, changes in the labour
market can also impact women’s health, outside of their workplace. An oft-cited
concern is the double-burden women face as both income earners and caretakers,
increasing their emotional and physical stress. For example, Doyal (2002) notes that
the increase of women in the workforce has rarely been matched by an increase of
men taking on domestic responsibilities. Cornia (2001) additionally warns that if
“growth in economic activity by women is not accompanied by the development of
adequate child care institutions there may be an increase in injury and malnutrition
among children despite a rise in family incomes” (p. 837). Labour market
restructuring is also noted to have environmental impacts, such as deforestation, that
are more likely to affect women due to their responsibilities in the household. Doyal
(2002) notes that “[i]n parts of India, for example, deforestation means that women
now have to walk many extra hours each day for a headload of firewood, while in
many African communities water may be many hours away” (p. 241).
Child labour is another important, though less explored, pathway through
which trade liberalization affects health in the context of labour markets. Polakoff
(2007) explores the relationship between economic globalization, world poverty,
child labor and health. Trade liberalization can be understood as central to economic
globalization which the author defines as "the removal of barriers to free trade and
the closer integration of national economies” (Siglitz, 2003 cited in Polakoff, 2007,
p. 260). Regional trade agreements, as well as the WTO are noted for facilitating the
removal of these barriers. Trade liberalization is also conceptualized as an ‘essential
ingredient’ of structural adjustment which “has consistently resulted in an increase in
the number of poor and exacerbated the desperate conditions of life for the majority
of people in the countries on which it has been imposed” (p. 261).
Structural adjustment is related to greater world hunger, unemployment,
environmental degradation, social fragmentation, and severe crises. These
environments are in turn are noted to have “increased exploitation of, and

86
dependence on, child labor” (p. 263). Economic globalization is also related to
increased poverty and child labor in the global north, but it is unclear on what basis.
The author notes that in the US, child laborers can be found in the agricultural
sector, given their lack of legal protection in labour standards. Figures from the
United Farm Workers Union are presented which in 2000 placed the total number of
child farm workers in the US at 800,000. Child farm workers’ health is impacted
through “excessive and inappropriate hours of exhausting work that is performed
under unhealthy and often dangerous conditions” (p. 267). Pesticide exposure is
highlighted as a major risk which is linked to range of neurological and endocrine
problems. Other hazards include “lack of drinking water, water for hand washing,
and toilet facilities” (p. 268). Child farm workers often are paid very little, have
trouble maintaining a normal school schedule, fall behind in their education, and are
more likely to drop out of school altogether.
In the global south, the author explores child labour within the industrial
workforce. Here trade liberalization is introduced in the context of ‘tax holidays’,
which Polakoff describes as a benefit often rewarded to corporations through which
they enjoy a 5 or 10 year reprieve from paying import or export duties. The author
blames corporations for targeting child workers by “seek[ing] out the cheapest labor
force in order to reap the greatest profit”, capital mobility for facilitating this, and
also poverty for leading children into work (p. 270). Working conditions are
characterized by low wages, long hours, environmental contamination, and exposure
to dangerous hazards, physical abuse and criminal violence, including the trafficking
of children for sex work.

Summary
In the context of labour markets, framework authors challenge the assumption
that trade liberalization automatically improves health through growth in
employment and income. They do this by highlighting for instance that employment
is not always secure and wages extremely low, particularly in middle- to low-income
countries. Additionally, framework authors identify a range of health determining
factors which are related to trade liberalization and its impact on the labour market
outside of employment rates and income, for instance through physical, chemical and
psychosocial working conditions. Therefore, even if access to employment and
income increases, benefits may be outweighed by the health costs of working

87
conditions. Framework authors also note burdens are likely to fall hardest on women,
migrant and children workers and often identify the role of (and need for) unions in
promoting healthier conditions.
Muntaner and colleagues (2010), based on the EMCONET framework
(Benach et al., 2007) provide the most comprehensive framework in terms of
describing how employment conditions influence health, and while the role of social
policies in mediating labour market conditions and health is not entirely overlooked
by other framework authors, Muntaner and colleagues are unique in their systematic
incorporation of welfare state policies as a mediating link both in the relationship
between the macro-political contexts and labour market policies, as well as between
labour market policies and health. However, the authors fail to thoroughly account
for trade liberalization’s role in mediating power relations, labour market policies,
welfare state policies, and employment conditions. For example, while Labonté and
colleagues (2007) acknowledge the role of transnational corporations in reorganizing
modes of production and service provision across national borders, trade
liberalization is identified as largely driving these changes.
Even while other framework authors are better at identifying trade
liberalization’s influence in labour markets, specific policies and mechanisms
relating the two are scantily explored. One of the most specified aspects of trade
liberalization which is explored is the working environment of EPZs.

2.4 Preliminary Summary of Literature Review


In this chapter a systematic review of the literature was undertaken as an
initial step towards answering the first research objective and question of this thesis:

Research Objective 1 (RO1): To identify how trade liberalization and social policy
interact to influence health and its social determinants.

Research Question 1 (RQ1): How do researchers theorize the pathways and


mechanisms mediating the trade liberalization and health relationship?

Studies were located for this review through a database search (ASSIA, PAIS,
Econlit, WOK) and by using Google Scholar. Studies were included if they provided
a clear analytical framework for conceptualizing pathways between trade
liberalization and health. Because of an initial low recall of studies, a refined search
strategy was adopted whereby reference lists of included studies were hand searched.

88
Google Scholar’s ‘cited by’ feature was also utilized. Frameworks identified through
these supplementary strategies were excluded if they focused exclusively on the
health impacts of trade liberalization in reference to health services, tobacco, alcohol,
medicines or food. This is because each of these health determinants already has a
rich and extensive research environment and this review was primarily interested in
the understudied impacts of trade liberalization on the fundamental determinants of
health.

In total, 41 studies were identified for inclusion in this review. In order to


answer the first research question, data was extracted from identified studies to
answer the following questions. These questions were posed since the ways in which
trade liberalization and health are understood can both conceal and illuminate
pathways and mechanisms important for health.

1. How is trade liberalization understood in analytical frameworks relating trade


to health?
2. How is health conceptualized in these frameworks?
3. How do researchers theorize the mechanisms and pathways mediating the
liberalization and health relationship?

These questions provided a useful way of organizing the findings of this


review. Overall it is found that researchers conceptualize both trade liberalization
and health in a variety of ways. Few studies included in the review explore the
exclusive impact of trade liberalization on health, most explore the health impacts of
trade liberalization within broader globalization and health frameworks. However,
this is noted to likely reflect a logical progression of research. Four main contexts are
identified through which liberalizing process may impact health: (1) increased flows
of goods and people (2) agricultural and food trade (3) structural adjustment
programs and Poverty Reduction Strategy Papers and (4) labour markets. In these
contexts, researchers identify a range of pathways between trade liberalization and
health. In the next chapter, these findings are critically discussed in greater depth for
a more comprehensive account how far they take us towards answering the first
research objective and question of this thesis.

89
CHAPTER 3 MOVING THE RESEARCH AGENDA FORWARD
This chapter will critically discuss the results of the literature review. It will
first look at the problems associated with the ways in which trade liberalization is
conceptualized and defined by framework authors. It will then discuss the ways in
which health is conceptualized and highlight the importance of differentiating
between pathways to health and pathways to health inequalities. How health is
framed and the implications this has for the acceptance of frameworks in global
health discussions will also be highlighted.

Next, the pathways presented by frameworks authors will be critically


discussed. The strengths and weaknesses of particularly influential frameworks will
be noted. The pathways presented by framework authors across the four identified
contexts of trade liberalization will also be considered. Areas of overlap between
pathways identified in the four contexts will be highlighted since they identify
aspects of trade liberalization that may be relevant in different contexts.

It will be argued that frameworks included in this review offer only partial
glimpses of a conceivably much larger and more comprehensive trade liberalization
and health framework, but that exploring these pathways exposes some general
considerations that should be taken into account if a more comprehensive and clear
picture of how trade liberalization impacts health is to be achieved.

Drawing on the points raised in this discussion as well as social


epidemiological theories, this chapter will conclude with a proposed conceptual
framework which seeks to outline broad pathways between trade liberalization and
health, as well as areas for future research.

3.1 Towards a Common Definition of Trade Liberalization


How trade liberalization is conceptualized and defined can both conceal and
illuminate pathways and mechanisms mediating its relationship to health. While
conceptualizations offer context and broad understandings behind trade liberalizing
processes, definitions should aim to provide specific accounts of what these
processes are.

Most framework authors conceptualize trade liberalization in the context of


globalization and by referring to ideas like ‘market integration’, ‘free trade‘, and

90
‘opening-up’, or by proxy of increased trade flows. Explicit definitions of trade
liberalization on the other hand, are largely lacking.

In the absence of concrete definitions, conceptualizing trade liberalization in


these ways becomes problematic. This is because drawing on terms which
themselves are vague and undefined makes it difficult to distinguish between
pathways which are believed to originate from trade liberalization specifically and
those which are believed to originate from other globalization processes in general.
Conceptualizing trade liberalization in these ways is also problematic since the
particular role of trade liberalization in these notions is unclear. For example,
openness can broadly refer to deregulation policies or it can be used to indicate a
country’s degree of integration with global economic forces. This latter notion of
openness however, may depend on factors unrelated to trade liberalization, such as
countries’ natural resource endowments (Subasat, 2008). Moreover, authors using
trade flows as a proxy for trade liberalization confuse the processes of trade
liberalization with its presumed outcomes. This is especially problematic given the
recognition that trade liberalization does not inevitably lead to increased trade flows
(Rodriguez & Rodrik, 2001).

On its face, the concept of trade liberalization may seem rather simple.
However, this review demonstrates the difficulty in formally defining its
constitution. For example, while many framework authors conceptualize financial
liberalization as a central component of trade liberalization, others seem to see it as a
distinct process.

Interestingly, currency devaluation and export support policies are


acknowledged by one author as important components of trade liberalization (Thow,
2009) and GATS and TRIPS (outside of their typical associations with health
services and the protection of drug patents) by others (Thow, 2009; Grown, 2005;
Corrigall et al., 2008).

In sum, frameworks included in the literature rely largely on vague


conceptualizations of what trade liberalization entails. When specific definitions are
offered, a range of different and sometimes contradictory processes are identified by
different authors. While the variety of identified processes begin to paint a more

91
comprehensive picture of trade liberalization as it relates to health, it also expose a
lack of consensus on what exactly trade liberalization is.

The need for consensus in this regard is articulated by Starfield (2001) who
argues that “common definitions of concepts are necessary in order for researchers to
develop common measures, whether in descriptive, analytic, pathway, or intervention
studies. The absence of a common definition will make it difficult if not impossible
to draw any conclusions about the policy implications of the findings of research” (p.
552).

In other words, while explicit definitions of trade liberalization are needed to


make clear theorized pathways to health, a common definition of trade liberalization
is especially needed if we want to comprehensively explore the health impact of
trade liberalizing processes and draw more informed conclusions about their
combined impact on health.

In a glossary of trade terms for public health, authors Labonté and Sanger
(Labonté & Sanger, 2006) define trade liberalization as “the process of reducing
tariff levels and of making other government measures less trade restrictive” (p.
656). However, there are two main weaknesses with this definition.

First, it is extremely vague in regards to exactly which processes, outside of


tariff reductions, follow from trade liberalizing policies. It thus provides little
direction in terms of both mapping out and measuring the health impacts of trade
liberalization. Second, its emphasis on the notion of reducing “trade restrictiveness”
is problematic for reasons which are similar to those outlined above in relation to
using notions of openness and increased trade flows as a proxy for trade
liberalization. For example, does a reduction in trade restrictiveness refer to policies
which increase the flow of goods into and out of a country? If so, export promotion
strategies, such as export subsidies, can be considered as processes of trade
liberalization. If however, trade restrictiveness refers only to government
intervention, then export promotion strategies like export subsidies should not be
considered an instance of trade liberalization. Moreover, in this latter case, a country
could be characterized as undertaking a liberalization process even when this may
reduce the overall amount of goods being traded (i.e when trade is further restricted).

92
For example, when export subsidies are reduced and consequently a country’s
exports fall.

In synthesising researchers’ different conceptualizations of trade


liberalization, this literature review underscores that trade liberalization is a multi-
faceted concept shaped by the ideology of neoliberalism, characterized by unequal
power relations, and promoted by powerful interests through a range of institutions,
trade agreements and policies in the domains of goods, services and investment. This
understanding of trade liberalization provides some direction for mapping out a wide
range of liberalizing processes and their combined impact on health. It also is aligned
with the majority of framework authors who see investment as a domain of trade
liberalization. Moreover, by drawing attention to the political contexts and actors
shaping the context within which liberalization policies are pursed, it also directs
attention to how well the policy implications of research will be adopted.

However, while this notion conceives of trade liberalization as being shaped


and promoted by certain ideas and actors, it does not say what it actually is. To
further consider the problem of defining trade liberalization, this thesis will now turn
to development economics, a field in which debates about the merits of trade
liberalization have been ongoing for decades.

3.1.1 Definitional Insights from Development Economics


The aim of this next section is to present a brief overview of how trade
liberalization has been defined in development economics. It is this literature base
that is turned to because this is where understandings of trade liberalization have
been most explored. As a research field, development economics has largely been
driven by an agenda which seeks to understand the broader links between trade and
economic performance. Debates about the merits of trade liberalization in terms of
economic growth have been going on for decades and continue today. This literature
base is therefore not only hugely expansive, but also involves several levels of
abstraction: theoretical, empirical, methodological, historical, political, and
ideological. It is beyond the scope of this section to provide a comprehensive account
of this literature. Instead, a general overview will be given of how mainstream
literature in development economics has defined trade liberalization or contributed to
its definitional understandings.

93
Much of the contemporary work exploring the relationship between trade and
growth has its roots in the 1970s. Little, Scitovsky and Scott (1970) and Balassa
(1971) are often described as pioneering the investigation into the links between
trade and economic performance. However, it wasn’t until the classic National
Bureau of Economic Research (NBER) study, directed by Krueger (1978) and
Bhagwati (1978), that trade liberalizing notions were used in a systematic way to
explore these relationships (Edwards, 1989, 1993).

In this latter study, the authors aim to explore how a country’s trade
orientation impacts growth. Here, trade liberalization is defined as any policy that
reduces a country’s ‘anti-export bias’. While the technicalities of this idea are not
important, what is vital is that this definition allows for a country to be characterized
as ‘liberalized’ yet still employ very high import tariffs. This definition of trade
liberalization thus does not preclude government intervention. Edwards (1989)
argues that it is this point which prompted researchers to demarcate a ‘liberal’ trade
regime, i.e. one which is characterized by a near absence of government intervention.

In the years after the NEBR study, some continued to view liberalization as
distinct from laissez-fair (Cooper, 1987), however the two are now generally
understood to be conceptually aligned (Banuri, 1991; Edwards, 1989, 1993). Less
controversial however, is that liberalization should be seen as a process (i.e a change
in country’s trade policy) (Cooper, 1987; Edwards, 1989). Moreover, important
strides have been made in terms of distinguishing between trade liberalization and
other related concepts such as openness and the orientation of a country’s trade
policy. For example Banuri (1991) distinguishes between several related concepts
found in the literature on liberalization and growth. In this work, openness is
described as an indication of a country’s degree of integration with global economic
forces. Trade orientation is taken to be a description of government policies which
create the conditions for openness. Trade liberalization by contrast, is defined as “a
strategy of policy reform intended to take the economy from a state of ‘illiberalism’
to that of liberalism’, or, in more hackneyed terms, towards laissez-faire” (Banuri,
1991, p. 10).

Other work emphasizes that a country’s openness may have less to do with
government policies than a country’s structural characteristics related for instance, to
their natural endowments, proximity to export markets, and size of domestic markets

94
(Subasat, 2008). A country’s trade flows may also depend on exogenous factors such
as transport costs or changes in world demand (Rodriguez & Rodrik, 2001). Together
these considerations indicate that that trade liberalization should be distinguished
from notions of market integration as well as from other overall measures of trade
flows since these may be related to other characteristics of a country or changes in
global market in general.

3.1.2 Defining Trade Liberalization for Public Health


With the above considerations in mind this thesis presents the following
working definition of trade liberalization for public health researchers: Trade
liberalization is the process of reducing government intervention in matters of trade.
In contrast to the definition provided by Labonté and Sanger (2006) which
focuses on reducing the restrictiveness of trade, at the heart of this definition is
making trade free from government restrictions. The distinction is subtle, but
important. Characterizing trade liberalization in this way prevents the confounding of
liberalizing policies for processes which may increase a country’s market integration,
like export promotion strategies. It also ensures that liberalizing policies are not
equated to increased trade flows or other measures which can conflate trade
liberalizing policies for their presumed outcomes.
This definition also suggests potential methods for measuring instances of
trade liberalization. For instance, trade liberalization may occur by way of reducing
government intervention in terms of imports (e.g. by reducing import tariffs), or by
way of reducing government intervention in terms of exports (e.g. by reducing export
subsidies). Importantly, this definition also points to ways trade liberalization should
not be measured, for example via changes in trade flows (e.g. by reference to the
value of imports and exports in relation to a country’s gross domestic product).

3.2 Considering Health Conceptualizations

3.2.1 Overall Conceptualizations


While both broad and specific conceptualizations of health provide insight
into the links between trade liberalization and health, framework authors in this
review largely discuss these relationships by conceptualizing health and health
inequalities/inequities in general and often undefined terms. Nutrition related
outcomes were by far the most commonly explored specific health/health inequality

95
outcome and made up the majority of frameworks categorized in the context of
agricultural and food trade. Communicable and non-communicable diseases were
more likely to be discussed in the area of hazardous flows, and occupational health
outcomes more likely in the context of labour markets.
Frameworks were almost equally split between those investigating health and
those investigating health inequalities/inequities. Framework authors investigating
health inequalities/inequities were more likely to contextualize globalization and
trade liberalization with references to power imbalances within and between
countries, and by considering the role of neoliberalism as an ideological force driving
the processes of trade liberalization. Outside of this distinction however, it wasn’t
readily apparent how pathways leading to health and pathways to health
inequalities/inequities differ, a distinction which is increasingly recognized in public
health literature (Nancy Krieger, 2011; Starfield, 2001).

3.2.2 Health Inequalities Versus Health Inequities


Frameworks which explore differences in health do so either by characterizing
these differences as health inequalities or health inequities. Acknowledging the
differences between these two concepts is important given divergent policy
implications (Braveman, 2006). Health inequalities, sometimes referred to as health
disparities in the US, typically highlight crude differences in health. Health inequities
on the other hand, are often understood as “differences in health that are not only
unnecessary and avoidable but in addition, are considered unfair and unjust”
(Whitehead, 1992, p. 430). When understood in this sense, considerations of health
equity consequently launch us into discussions of ethics, equality and justice.
However, commonly used health inequality measures are also value laden and data
are often interpreted through normative judgments of particular health distributions
(Harper et al., 2010).

3.2.3 Framing Health


Frameworks included in this review were not only differentiated on the basis
of their chosen overall health/health inequality outcome, but one third of all
frameworks used a readily identifiable ‘frame’ of health to contextualize chosen
outcomes. These frameworks characterize health in the context of development
policy, as a human right, and/or in reference to global public goods.
The manner in which authors frame health, or health inequalities, did not have

96
a distinguishable impact on pathways presented in this review but since these frames
constitute ways in which health is discussed in both global trade and global health
discourses, they have important implications for how well frameworks are received
and acted upon. They also have important implications for health equity (Labonté,
2008). These implications should be taken into consideration if a comprehensive
understanding of how trade liberalization impacts health is to be accepted by actors
in the global arena with positive impacts on health equity.

3.3 Developing our Understanding of the Links Between Trade


Liberalization and Health

3.3.1 Early and Influential Frameworks: Setting the Stage


Early frameworks establishing the link between globalization and health
identify important aspects of globalization that shape the context within which trade
liberalization is pursued, adopted and responded to. These frameworks however,
focus on a limited range of pathways and fail to name specific mechanisms
mediating the trade liberalization and health relationship.
More recent frameworks have begun to account for a wider range of pathways
and mechanisms by adopting social determinants of health perspective. The
frameworks devised by Labonté and colleagues (2007) and Blouin and colleagues
(2009) are highlighted for pioneering such work.
Blouin and colleagues (2009) offer the only framework which seeks to expose
the exclusive impact of trade liberalization on a broad range of health outcomes.
Therefore, even in the absence of an explicit liberalizing definition, there is less
ambiguity in this framework as to whether the processes described are attributed to
trade liberalization versus broader global processes. Moreover, because the
framework explores the link between trade liberalization and health in general, it is
less likely that trade liberalization is conceptualized in a way which is restricted to a
specific health outcome. However, as previously discussed in section 2.3.3 many of
the identified pathways are under conceptualized.

3.3.2 Across the Contexts: Strengths, Weaknesses, Overlap and Gaps


In the previous chapter, pathways between trade liberalization and health are
discussed across four non-mutually exclusive contexts: increased flows of goods and
people; agricultural and food trade, structural adjustment policies and poverty

97
reduction strategy papers, and labour markets.
In the domain of increased flows of goods and people, trade liberalization is
associated with increases in communicable and non-communicable diseases by
increasing people’s exposure to infectious diseases and hazardous goods respectively.
Additionally, a range of health outcomes is seen to result from an increased flow of
drugs, legal and illegal, due mostly to import liberalizing strategies.
In the domain of agricultural and food trade, framework authors offer a more
conceptually developed understanding of the pathways between trade liberalization
and health. Frameworks explored in this domain identify three main pathways
linking trade liberalization and health: food security, nutrition, and food safety. In
these pathways a range of trade liberalizing strategies are explicitly addressed, but
many associations are also assumed without theoretical support.
In the domain of structural adjustment policies and poverty reduction strategy
papers, trade liberalization is viewed as a significant factor mediating the pathways
to health. Despite this fact, a reduction in consumption subsidies is the only specific
trade liberalization policy isolated for its unique impacts on nutrition and household
income.
In the final domain of labour markets, trade liberalization is seen to impact
health through a variety of pathways including, economic and social inequalities,
wages, working conditions (both physical and pyschosocial), job loss, job insecurity,
and particularly for women, work-life (in) balances. Trade liberalization's specific
influence on each of these pathways is less conceptually developed than it is in
relation to pathways identified in the context of food trade and agricultural policy.
EPZs are the most commonly identified liberalizing strategy in this context. More
than in any other context, social policies are identified in the context of labour
markets as a mediating factor between health and trade liberalization. Power
relations are also more emphasized in this context and the strength of organized
labour movements is consistently acknowledged as an important mediating factor,
though one not comprehensively explored.
As previously discussed, literature relating trade liberalization to health was
identified through a selective search strategy which favored frameworks moving
away from the health impacts of trade in hazardous goods, foods, medicines, and the
liberalization of health services. Despite this selectiveness both conceptualizations
and definitions of trade liberalization were most comprehensive in the context of
98
food trade and agricultural policy. In this context trade liberalization is related to a
wider range of agreements and policies than in other contexts, though frameworks in
all contexts identify a similar range of institutions related to trade liberalization.
At first glance, it seems reasonable that agreements are less emphasized in
other contexts. The selective search strategy utilized means that conceptualizations of
trade liberalization in the context of ‘flows of people and goods’ are unrepresentative
of research exploring this area. Moreover, it makes sense that in the context of
structural adjustment policies, trade liberalization is more related to loan
conditionalities than agreements. In the context of labour markets, it can be argued
that agreements are less relevant since liberalizing processes are not usually tied to a
specific product whose health significance can be interpreted through specific rules
or trade measures contained for instance, in agreements such as the SPS or TBT.
Instead, analyzing the more indirect health impacts of labour markets, or
macroeconomic trends in general, tends to concern broader policy options, cut across
different policy spheres, and raise ideological considerations (Fidler, Drager, & Lee,
2009).
However, the agreements offered by framework authors in the context of food
and agricultural trade are more than just focused on the rules under which food
related goods can be traded or under which conditions specific health-related
restrictions can be taken. Authors in this context demonstrate that trade agreements
are also important drivers of larger liberalization processes. Thow (2009) for
example, relates GATS and TRIMS to trade liberalization given their respective
commitments to remove constrictions on foreign ownership and companies' source of
inputs. Both of these agreements have less to do with the direct health impact of
specific products than they do with larger liberalizing processes.
A greater range of trade liberalizing policies are also considered in the context
of food trade and agricultural policy. Again, while the selective search strategy likely
played a major role in the dearth of agreements and policies identified in the context
of flows of goods, better conceptualizations and definitions of trade liberalization in
reference to structural adjustment, PSRPs, and labour markets are especially needed.
Categorizing pathways across these four contexts provided a conceptually
useful way of discussing the vast majority of frameworks included in this review and
highlighted both the strengths of, and gaps within, particular research areas.
However, these different contexts also exhibit considerable overlap with each other.
99
This overlap helps to identify interrelated and contributing pathways across contexts.
As previously mentioned, the majority of frameworks identified tend to explore
pathways within the boundaries of a singular context. Recognizing the overlap
between different contexts is therefore especially important as it allows for a more
comprehensive appreciation of the trade liberalization-health relationship.
For example, liberalizing processes identified in the context of food trade and
agricultural policy, demonstrate areas of overlap with labour markets-- the
liberalization of export markets and its facilitation of work in cash cropping schemes,
as well as decreases in farmer subsidies are both liberalizing processes also
addressed in the context of labour markets. The health impacts of agricultural work
explored in the context of labour markets also overlaps with nutritional concerns
addressed in the context of food trade and agricultural policy. Interestingly, the
differences between characterizations of trade liberalization within each of these
contexts rest largely on the framing of impacts either in terms of ‘people as
consumers’ or ‘people as producers’. Recognizing the overlap across labour markets
and other contexts not only suggests a wider range of trade liberalization
conceptualizations and definitions to consider in relation to labour markets, but also
identifies a wider range of interrelated pathways to health as exemplified in figure 17
below.

Trade Liberalization
{Food Trade and Agricultural Policy} {Labour Markets}

Increases women's
participation in
Increases incomes agricultural work but
and food spending;
little impact on
Cash is characterized by
low quality
nutrional intake
(Labonté et al. 2007)
Cropping conditions and job
insecurity
(Lowenson et al.
2010)

Health
Figure 17 Example of usefulness in recognizing overlap between contexts

100
Categorizing pathways across these four contexts also exposes areas less
explored by framework authors. While highlighted by the early and influential
frameworks, less explored by later frameworks are issues relating to trade
liberalization's impact on health through reductions in government revenues,
reductions in government’s policy space, the physical environment and the
liberalization of trade in services. While it is likely due to the selective search
strategy that a limited number of pathways addressing the liberalization of health
services were identified, the liberalization of services in general would conceivably
have impacts on health outside that of the health sector, via for example, education.
In total, pathways identified within each of these areas offer only partial
glimpses of a conceivably much larger and more comprehensive trade liberalization
and health framework. Moreover, in addition to previously raised issues, exploring
these partial pathways exposes some general considerations that should be taken into
account if a more comprehensive and clear picture of trade liberalization and its
impacts on health is to be achieved.

3.3.3 General Considerations

The Distal-Proximal Divide


An important issue that quickly became apparent in the synthesis of identified
frameworks is that the large majority conceptualize the health impacts of trade
liberalization by contextualizing determinants of health within a causal hierarchy
whereby distal determinants of health, such as institutions, policies and
socioeconomic status, are seen to influence proximal determinants of health (i.e.
physical, behavioral, and psychosocial exposures) through various intermediary
pathways. As Krieger (2008b) suggests, this can be problematic since “events at one
level can directly and profoundly affect nonadjacent levels, instantly and persistently,
without intermediaries” (Krieger, 2008b, p. 225). Income, for instance, is sometimes
described by framework authors as a distal determinant of health influencing further
'downstream' determinants (e.g. Meier 2006). This however, obscures the very direct
and important impact income (and especially its distribution) can have on population
health.

Graphical Representations
Many authors incorporate into their work graphical representations of their
frameworks. Visual conceptual models offer the benefit of “explicating theory and

101
for organizing, comprehending and contesting scientific data and knowledge”
(Krieger, 2008a, p. 1098). They can also “simultaneously spur ideas and
observations”, and act as a “tool for integrating and evaluating rapidly emerging
findings and for guiding new research” (p. 1098). However, two main problems
characterize graphical representations included in this review. First, many visual
frameworks did not incorporate graphically all the pathways explicated in the textual
descriptions. Blouin and colleagues (2009) for example, describe the important role
social policies play in mediating health outcomes; however their graphical
framework fails to consider this pathway. Second, framework authors graphically
describing trade liberalization's impact both on health and health inequality fail to
account for the fact that pathways leading to health are not necessarily the same as
those leading to health equity (Krieger, 2008a; Starfield, 2001). An example of this is
the framework by Woodward and colleagues (2001) which shows the same pathways
leading from globalization to both health and health equity.

Global and Differential Impacts


Four additional and related considerations which were exposed in the
synthesis of frameworks have to do with the fact that 1) trade liberalization policies
undertaken in one country can have important impacts on others and 2) that
liberalizing policies have differential impacts both between and within countries.
This first issue is highlighted across a variety of frameworks which illustrate
the potential impact of reductions in industrial countries' agricultural subsidies on
market opportunities for producers in low- and middle-income countries. It is also
illustrated particularly well by Hawkes (2006) who uses case studies to explore how
agricultural liberalization policies in one country can impact health in others. While
this consideration seems obvious, it is important to highlight because some
frameworks ‘disembed’ their conceptualizations of trade liberalization from global
interactions and instead concentrate on the domestic health impact of national
liberalization processes.
Related to this issue is the consideration that the large majority of frameworks
included in this review focus on the health impacts of trade liberalization in middle-
to low- income countries. A notable exemption to this is the framework by Smith and
Signal (2009) which explores pathways to nutrition-related health outcomes via trade
liberalizing reforms in New Zealand.

102
Another consideration related to the differential impact of trade liberalization
is the fact that trade liberalization policies can impact different countries
differentially. This is exemplified by Singer (2008) who demonstrates how import
liberalization has increased the availability of legal and illegal drugs with different
impacts in “countries of production, countries of trans-shipment, and countries of
targeted consumption” (p. 469).
A final consideration related to the above three is that many framework
authors highlight the differential impact of trade liberalization policies within
countries, across different actors and subgroups. Returning to the impact of
industrialized countries' use of agricultural subsidies, some framework authors point
out that a reduction in the level of these subsidies will not benefit all producers in
low- and middle-income countries the same. Authors emphasize, for example, that
larger transnational corporations are more likely to benefit from these reductions than
smaller producers. Framework authors who explore health inequalities/inequities or
the health outcomes within specific subgroups, such as women, also demonstrate
how trade liberalization affects various populations within countries differently. The
poor, women, children, and immigrants are often highlighted as particularly
vulnerable population subgroups.
Together these final issues highlight the need for a better understanding of
interactions of trade liberalization from the global to domestic level, across low-,
middle-, and high-income countries, as well as across countries differentiated by
their position in production-consumption chains.
With the issues and considerations explored in this section in mind, the next
section of this chapter will present social epidemiological theory as the appropriate
basis for developing a more comprehensive framework for exploring the links
between trade liberalization and health.

3.4 Towards a Conceptual Framework


A benefit of synthesizing a variety of frameworks in a particular research area
is to create a delimited number from which to work with (Starfield, 2001).
Frameworks authors included in this review provide a myriad of pathways through
which trade liberalization impacts health. These pathways emphasize a range of
material and psychosocial exposures. Moreover, exploring these pathways exposed a
range of considerations that should be taken into account when conceptualizing a

103
larger trade liberalization and health framework. As in any other scientific endeavor,
in order to organize these pathways and considerations into a single framework
which elucidates a testable set of ideas, theory is essential.
Concerned with the social and biological processes behind distributions of
health, social epidemiology provides an appropriate basis for theorizing how trade
liberalization impacts health. This is supported by the fact that the various theoretical
variants of social epidemiology form the basis of most of the frameworks included in
this review, albeit in a mostly implicit way.
Krieger (2001, 2011) identifies three major theoretical directions taken in
contemporary social epidemiology: (1) sociopolitical; (2) psychosocial; and (3)
ecosocial. While not mutually exclusive, these approaches provide theories of disease
distribution that emphasize different social mechanisms to explain how populations
arrive at different levels of health with different magnitudes of health inequalities.
Each of these social epidemiologic theoretical directions are characterized by
various strands but generally speaking, it can be said that sociopolitical theories
emphasize power relations, macro-level determinants of health, and the idea that
health is a human right; psychosocial oriented theories emphasize psychological
determinants of health, or, to use an oft cited expression, are concerned with how
broader, social determinants of health 'get under the skin'; finally, ecosocial theory,
the more nascent of the three theories, advances these first two by addressing their
under-theorized areas and proposing an approach which fosters a more
comprehensive analysis of health distributions by considering how interactions of
various pathways across multi-levels and spatio-temporal scales literally become
embodied, taking into account important historical, and sociopolitical drivers of these
pathways across the lifecourse.
This section will highlight the main ideas behind each of these theories, as
well as their relevance to frameworks included in this review. The different emphases
these perspectives place on various social and biological determinants of health will
be noted. On the basis of these distinctions, and taking into account considerations
raised in the previous section, ecosocial theory will be presented as the most
appropriate basis from which trade liberalization can be theorized to impact health.
However, because trade liberalization raises issues not yet explored in ecosocial
theory, an adapted ecosocial framework will be presented as a basis from which to
further explore the links between trade liberalization and health.
104
3.4.1 Sociopolitical Theories
Krieger (2011) highlights eight self-designated variants of sociopolitical
epidemiological theories:
 The social production of disease,
 The political economy of health,
 Social determinants of health,
 Population health,
 Fundamental cause,
 Political epidemiology,
 Latin American social medicine/collective health, and
 Health and human rights

3.4.2 Social Production of Disease/Political Economy of Health


The first two of these theories are considered to encompass the same
conceptual premises and thus can be examined together. Central to both is the idea
that “any given society's patterning of health and disease—including is social
inequalities in heath—is produced by the structure, values, and priorities of its
political and economic systems, in conjunction with those of the political and
economic systems of the other societies with which it interacts” (Krieger, 2011, p.
167 emphasis in original). Tantamount to these theories is the distribution of power
and the question of “who is producing what, with what technologies, for whom, and
why” (Krieger, 2011, p. 167). The 'who' in this question is conceptualized as political
and economic systems operating across multiple levels, both within and across
various geographical areas. The 'who' also focuses on the institutions and individuals
who control these systems. The 'what' of this question refers not only to the physical
output, both in financial and material terms, of these systems; but also to the social
structures created by them, the means by which social groups within these structures
are able to sustain themselves, as well as challenge their social position; and the
justifications used to support the priorities of these systems. However, typically
unaccounted for by these theories is how these social conditions, processes, and
relationships translate biologically.
Many of the frameworks included in this review emphasize the role of
neoliberalism, the 'Washington consensus', peoples' unequal bargaining positions,
and various power relations in influencing health outcomes and thus can be

105
understood as operating within a social production of disease/political economy of
health framework. However, they too fail to explicate how these conditions translate
on the biological level.

3.4.3 Social Determinants of Health, Population Health, and Fundamental


Causes
Drawing on the core components of the social production of disease/political
economy of health theories are the subsidiary theories of 'social determinants of
health', 'population health', and 'fundamental causes'. Introduced at various times,
these theories gained popularity in the mid to late 1990s and relate to the social
production of disease and political economy of health theories by emphasizing the
health importance of power and resource distribution. All three typically recognize
that health is shaped by socioeconomic conditions. All three also typically pay heed
to the social gradient in health, something frameworks in this review fail to do.
The social determinants of health and population health approaches are
usually distinguished from the social production of disease and political economy of
health approaches via their consideration of biological mechanisms. These
approaches often consider both how early life conditions translate into both
childhood and adult health, as well how cumulative (dis)advantage impacts health
over the lifecourse. Neither of these considerations are however, addressed by
frameworks included in this review highlighting areas for further exploration.
These two theories can be distinguished from the fundamental cause
approach. As described in the introduction chapter of this work, this approach is
centered on the idea that focusing on superficial risk factors of disease, without
addressing their fundamental causes, will do little to change the link between
socioeconomic status and health, since fundamental causes are associated with
multiple risk factors and multiple health outcomes (Link & Phelan, 1995).
Fundamental causes of disease are conceptualized “broadly to include money,
knowledge, power, prestige, and the kinds of interpersonal resources embodied in the
concepts of social support and social networks” (Link & Phelan, 1995, p. 87). This
approach is also centered around the recognition that we live in a dynamic world
system where new diseases and risk factors are always emerging and those with
greater access to resources will always be better positioned to respond to them.
Moreover, fundamental causes can be distinguished from contextualized risk

106
factors as the latter concerns “how people come to be exposed to individually-based
risk factors such as poor diet, cholesterol, lack of exercise, or high blood pressure”
(Link & Phelan, 1995, p. 81).
Frameworks included in this review direct attention to both fundamental
determinants of health (e.g. labour markets and factors related to social position such
as income) and contextualized risk factors (e.g unhealthy diets, and consumption of
alcohol and tobacco).

3.4.4 Latin American Social Medicine/Collective Health


Krieger (2011) notes that in Latin America, social epidemiological theories
are growing in influence. Like the social production of disease/political economy of
health approaches, these theories are focused on political and economic determinants
of societal patterns of disease. However, it is noted that they typically pay more
attention to the health “impact of capitalist development, imperialism, and politics on
health” (Krieger, 2011, p. 189). They are also differentiated from their English-
language counterparts given the higher degree of importance placed on non-state
actors and social movements in shaping distributions of health.
None of the frameworks included in this review can be said to have drawn on
this body of work, likely because frameworks which were not written in English
were excluded. However, some frameworks included in this review do emphasize the
role of people as social actors. Furthermore, collaborations that have begun to
emerge between researchers in Latin America and those in the global North (Krieger
et al., 2010) may serve to highlight additional areas of relevance to trade
liberalization and health frameworks.

3.4.5 Health and Human Rights


The final approach within the sociopolitical strand of social epidemiology is
the health and human rights approach. The basis of this approach is significantly
different from the others discussed thus far in that it focuses on governments'
obligations to “respect, protect and fulfill human rights” (Krieger, 2011, p. 190). As
previously discussed, many frameworks included in this review frame health as a
human right.

3.4.6 Psychosocial Theory


The second major strand of social epidemiology is concerned with the

107
psychosocial determinants of health. The central idea of this trend is that peoples'
perception of social conditions, social interactions and their social status have
important psychological, biological, and behavioral effects. As with some
sociopolitical theories, health is seen to be systematically distributed across status
hierarchies/social gradients. Unlike sociopolitical theories, psychosocial approaches
identify peoples' relative rank in status hierarchies as fundamental stressors which
are incorporated biologically via stress mediated mechanisms.
Also of concern to psychosocial theories are the social phenomena which can
buffer or worsen the negative impacts of these perceptions, namely social support
networks and social capital. However, it should be noted that the health relevance of
social capital is a highly contested subject of debate (for a review of this debate see
the International Journal of Epidemiology, 33, (4)).
Psychosocial theories typically afford less attention to the social and political
systems responsible for distributions of health, although a handful of frameworks
included in this review do draw attention to the psychological and mental health
impacts of certain trade liberalizing policies (see for example, Corrigall et al. 2008).
As previously mentioned they don't however, draw attention to the social gradient in
health. They also don't account for the specific biological mechanisms resulting in
these outcomes.

3.4.7 Ecosocial Theory


Compared to the other theoretical strands which have their roots in the mid
twentieth century, ecosocial theory is a more recent addition to social
epidemiological study. First introduced by Nancy Krieger (1994), ecosocial theory
seeks to embrace both sociopolitical and psychosocial perspectives by considering
both the biological and social determinants of health. However, ecosocial theory also
strives to develop these perspectives by incorporating into them core ecological
considerations, hence the prefix 'eco'. These ecological considerations relate to the
levels of organization within which health important processes occur, the
relationships and interactions that occur between different levels of organization, and
the importance of historical conditions (Krieger, 2011).
The guiding question of ecosocial theory is “who and what drives current and
changing patterns of social inequalities in health?” (Krieger, 2001, p. 672). Ecosocial
theory seeks to answer this question via attention to four core constructs:

108
embodiment, pathways of embodiment; cumulative interplay between exposure,
susceptibility and resistance; and accountability and agency (Krieger, 2001, 2011).

Embodiment
The first core construct of ecosocial theory, embodiment, refers “to how we
literally incorporate, biologically, the material and social world in which we live,
from conception to death; a corollary is that no aspect of our biology can be
understood absent knowledge of history and individual and societal ways of living”
(Krieger, 2001, p. 672). Importantly, the construct of embodiment also “recognizes
that socially-structured causal links between exposures and outcomes can vary over
time and place, a proposition consonant with contemporary ecological theorizing”
(Krieger, 2011, p. 216). This advances other social epidemiological theories by
recognizing that “explanations of disease distribution cannot be reduced solely to
explanations of disease mechanisms or to static notions of “status” or “fundamental”
causes, as the latter do not account for why actual disease rates and patterns of health
inequities change, in complex ways, over time and place” (Krieger, 2011, p. 222).
Given the dynamic nature of trade liberalization over time and place, this idea
is also particularly relevant to the aim of this review. Few frameworks included in
this review for example, incorporate considerations of the pace and sequencing of
trade liberalization strategies over time. As previously mentioned, they also fail to
account for the cumulative impacts of exposures over the lifecourse or to the
potential time-lagged health effects of liberalizing strategies. However, this idea does
align well with Labonté and Togerson’s (2003) framework which incorporates a
consideration of the super-ordinate context that is how countries' pre-existing
endowments together with their political systems can mediate trade liberalizing
processes.

Pathways of Embodiment, & the Cumulative Interplay of Exposure, Susceptibility,


and Resistance
The second and third core constructs of ecosocial theory acknowledge that
embodiment occurs via a diverse set of concurrent and interacting pathways, across
multiple levels and lifecourses. Highlighted in this regard are “adverse exposures to,
along with differential societal and biological susceptibility and resistance to:
1. economic and social deprivation;
2. toxic substances, pathogens, and hazardous conditions;

109
3. discrimination and other forms of socially inflicted trauma (mental, physical,
and sexual, directly experienced or witnessed from verbal threats to violent
acts);
4. targeted marketing of harmful commodities (e.g. “junk” food and
psychoactive substances such as tobacco, alcohol, and other licit and illicit
drugs);
5. inadequate or degrading health care; and
6. degradation of ecosystems, including as linked to systematic alienation of
Indigenous populations from their lands and corresponding traditional
economics (Krieger, 2011, p. 223).
While these pathways were conceptualized by Krieger (1999, 2011) in an
attempt to understand the ways in which racism impacts health, they also summarize
many of the pathways addressed by framework authors included in this review.
However, absent from this account, in relation to the health impacts of trade
liberalization, are the following determinants: social and economic inequalities
(whose health impacts extend beyond that of absolute deprivation) and job insecurity,
although this may be represented either by social deprivation or hazardous
conditions.

Accountability and Agency


The final and fourth construct of ecosocial theory is that of accountability and
agency. This construct addresses both how health distributions are created, and the
ways in which they are monitored, analyzed, and addressed. Drawing largely on
social production of disease and political economy of health theories, this construct
“directs attention to issues of power at each and every level, and hence to institutions'
and individual peoples' capacity to act (“agency”) and their responsibility for both
actions taken and avoided (“accountability”)” (Krieger, 2011, p. 225).
It acknowledges that this capacity to act (and consequent accountability) –
which can be directed at either at the same or different levels—is not equivalent
across levels. That is, “in line with more recent sociological, ecological, and
biological theorizing” ecosocial theory posits that “macrolevel phenomena are more
likely to drive and constrain meso- and microlevel phenomena than vice versa—even
as, under particular circumstances the micro can powerfully affect the macro”, as has
been historically demonstrated via popular movements, which among other social

110
achievements, have abolished slavery, decriminalized homosexuality, and established
welfare states (Krieger, 2011, p. 225).
Thus, ecosocial theory recognizes that causal pathways need not be linear,
reconciling the distal/proximal divide problem characterizing the majority of
frameworks included in this review. It also accounts for many frameworks’
consideration of power relations.

3.5 Trade Liberalization and Health: An Ecosocial Framework


In summary, ecosocial theory advances sociopolitical and psychosocial
approaches by drawing attention to under-theorized and under-researched issues in
these areas. Moreover, it brings these theories together by reframing simplistic
divisions of social and biological determinants of health in a more comprehensive
manner which considers the sociopolitical, historical and spatiotemporal relevance of
multi-level, and simultaneously interacting pathways, as well as their biological
embodiment across the lifecourse. In other words, “more than simply adding
'biology' to 'social' analyses, or 'social factors' to 'biological' analyses, the ecosocial
framework begins to envision a more systematic integrated approach capable of
generating new hypotheses, rather than simply reinterpreting factors identified by
one approach (e.g. biological) in terms of another (e.g. social)” (Krieger, 2001, p.
673). A graphical representation of ecosocial theory can be seen below in figure 18.

111
Figure 18 Ecosocial Theory (Krieger 2008)
By advancing an approach which embraces both sociopolitical and
pyschosocial perspectives, ecosocial theory also provides an overarching framework
which incorporates the types of theorizations employed by the majority of framework
authors included in this review. Importantly it also provides a means of advancing
areas thus far under conceptualized by framework authors. This is because
frameworks exploring the link between trade liberalization and health have yet to
consider the role biological processes play in shaping distributions of health, how
liberalizing processes may have lagged effects on health, and also how these
processes fit into lifecourse theories of epidemiology. By incorporating more
theoretically advanced notions of how sociopolitical processes interact
simultaneously across multi-levels, ecosocial theory can also free current trade
liberalization and health theorizations from the shackles of the distal-proximal
divide.
However, while providing a platform to advance current theorizations of the
pathways between trade liberalization and health, ecosocial theory in its current state
does not account for many of the considerations raised via the narrative synthesis of
frameworks included in this review.
For this reason, an adapted ecosocial framework is presented below (see
figure 19) which accounts for these considerations by incorporating the following:

112
contextualizations of trade liberalization within larger processes of globalization; the
various aspects of trade liberalization identified by framework authors, and the range
of domains and pathways through which framework authors relate trade
liberalization to health.

113
Trade Liberalization & Health:
An Ecosocial Perspective
Globalizing Context:
Trade Flows
Institutions
Economic Policies
Inequality Agreements
TNCs

Supra-Ordinate Context

Political Economy and


Ecological Context

Pathways: Domains: Levels:


Power Relations
Income Physical Global
Economic Inequality Social National
Employment
Working Conditions Competitive Forces
Regional
Job Loss Regulatory Space
Work-life Balances
Capital Markets Area/Group
Food Safety
Food Security Household
Nutrition
Social Inequalities Individual
Economic insecurity
Hazardous Goods
Government Revenues

Historical context, trans-


generational influences

Embodiment; Pathways of embodiment; Cumulative interplay of


exposure, susceptibility, & resistance; Accountability and Agency

Lifecourse:

Figure 19 Trade Liberalization and Health: An Ecosocial Perspective

114
Contextualizations of trade liberalization within larger processes of
globalization are incorporated into this new framework by drawing on the work of
Woodward and colleagues (2001) as well as that of De Vogli and colleagues (2009).
As previously discussed, Woodward and colleagues' (2001) representation of
economic globalization as a feedback loop provides a particularly useful way of not
only understanding how trade liberalization relates to other processes of
globalization, but also of how it is propagated. Similarly but uniquely, De Vogli and
colleagues (2009) demonstrate how public policies and economic inequality exist as
part of a feedback loop, whereby trade liberalizing policies contribute to the growth
of transnational corporations and thus their ability to further advance liberalizing
policies. Their resulting accumulation of wealth and power is then seen to increase
both economic and health inequalities between and within countries. The welfare
state is also highlighted in this framework as a potential mediating factor between
economic and health inequalities falling under the ‘social’ domain. Integrating these
two frameworks provides a useful way of beginning to conceptualize the context
within which trade liberalization operates.
Furthermore, a super-ordinate level, as described by Labonté and colleagues
(2003), is added to the traditional ecosocial framework as it identifies a range of
factors that shape the acceptance of, and domestic policy responses to trade
liberalizing processes.
The various aspects of trade liberalization that are identified by framework
authors are also incorporated into this new framework. Taken into account are not
only specific trade liberalizing processes/policies, but also the institutions and
agreements through which these processes are set into motion.
The pathways through which framework authors relate trade liberalization to
health are accounted for first by incorporating Labonté and Torgerson’s (2003) five
domains: the physical environment; the social environment; competitive forces;
governments’ regulatory space; and capital markets. This helps us begin to
conceptualize the major areas of society that trade liberalization impacts. While
pathways within each of these domains have yet to be comprehensively explored,
this provides a conceptually useful way of hypothesizing what they might be.
Considering these domains also highlights the importance of trade liberalization's
impact on areas under explored by framework authors including its impact on
welfare states (e.g. via government revenues), governments' policy space, and the
115
physical environment. The main pathways identified by framework authors in each
context are also noted, but this list should not be considered exhaustive.
Moreover, Labonté and colleagues' (2008) adoption of the Diderichsen et al.
(2001) health inequity model is highlighted in this review for providing a unique
analysis of the mechanisms by which various pathways create changes in the
distribution of health across populations, while not readily apparent in the graphical
representation of this new framework such mechanisms can be considered as an
element of the ecosocial construct of 'cumulative interplay of exposure, susceptibility
and resistance’.

3.5.1 Framework Limitations


The point of this adapted ecosocial framework, along with ecosocial theory in
general, is not to lay claim to a grand theory but rather to present a tool for orienting
deeper analyses into the relationships between trade liberalization and health. The
strength of this framework is that while it synthesizes a range of frameworks
exploring the link between trade liberalization and health, it also provides a tool for
conceptualizing links across areas thus far less explored. It also addresses one of the
main problems associated with conceptual frameworks relating trade liberalization to
health thus far, the distal-proximal divide.
A main weakness is that it fails to clearly map out the processes and
mechanisms linking the various elements contained within it. However, because the
framework was developed on the basis of two research fields still very much in their
early stages, it is expected that as these fields develop so too will our ability to
capture more thoroughly the precise mechanisms and interactions at play.
The aim of synthesizing frameworks relating trade liberalization to health was
to identify the potential pathways through which trade liberalization impacts health.
More specifically, given the scant attention paid to the social determinants of health,
it was primarily concerned with influences on health outside of those related to trade
in hazardous goods, food, medicines and the liberalization of health services.
Because of a low recall of frameworks, a supplementary search strategy was adopted
which purposively excluded frameworks focused on these pathways. Therefore,
while the synthesis still included frameworks emphasizing these pathways by
including those identified through the database searches, conclusions made about the
comprehensiveness of trade liberalization conceptualizations, definitions and health

116
related pathways should be considered with this limitation in mind. That a selective
search strategy was used should also be taken into account when considering the
adaptations made to the ecosocial framework which were based on these
conclusions.
It should also be noted that while the review included frameworks
highlighting pathways from trade liberalization to both health and differences in
health, by using ecosocial theory as the basis for a comprehensive framework, it
settles on an approach largely directed at explaining inequalities in health. The
framework does however, lends itself to any of the previously discussed frames of
health.
Moreover, by drawing on Woodward and colleagues (2001) contextualization
of trade liberalization, this new framework aligns itself with framework authors who
consider economic processes as the primary drivers of globalization.

3.6 Summary of Literature Review & Progress Towards Research


Objective 1
In Chapters 2 and 3, a systematic literature review was undertaken as an
initial step towards answering the first research objective and question of this thesis:

Research Objective 1 (RO1): To identify how trade liberalization and social policy
interact to influence health and its social determinants.

Research Question 1 (RQ1): How do researchers theorize the pathways and


mechanisms mediating the trade liberalization and health relationship?

In Chapter 2 it was found that researchers conceptualize both trade


liberalization and health in a variety of ways. Four main contexts are identified
through which liberalizing processes may impact health: (1) increased flows of goods
and people (2) agricultural and food trade (3) structural adjustment programs and
Poverty Reduction Strategy Papers and (4) labour markets. In Chapter 3, these
findings were critically discussed in greater depth. An important finding in this
chapter is that authors often conceptualize and define trade liberalization in a variety
of sometimes conflicting and problematic ways. For this reason literature from
development economics was briefly reviewed. Insights from this field informed the
development of a working definition of trade liberalization for public health: trade
liberalization is the process of reducing government intervention in matters of trade.

117
Another important finding is that the links between trade liberalization and health
have been especially under-conceptualized in relation to SAPs, Poverty Reduction
Strategy Papers and labour markets.

In total, studies included in the literature review are found to offer only a
partial picture of a conceivably larger and more comprehensive trade liberalization
and health agenda. However, exploring these partial pathways exposed some general
considerations that should be taken into account when considering the trade
liberalization and health relationship. For example, the importance of overcoming the
distal-proximal divide, that trade liberalization policies undertaken in country can
have important impacts on other countries, and that liberalizing policies have
differential impacts both between and within countries. With these findings and
considerations in mind, ecosocial theory emerged as a useful basis for both
organizing pathways to health and providing a means of advancing areas thus far
under conceptualized, for example, how liberalizing processes may have lagged
effects on health. With this in mind, the chapter presented an adapted ecosocial
framework of the relationship between trade liberalization and health.

Together the results of the literature review and the adapted ecosocial
framework take us some way towards answering the first research question of this
work. Importantly, synthesizing included studies brings into greater focus the
broader globalizing contexts within which trade liberalization is pursued, adopted
and responded to. Moreover a wide range of liberalizing processes is highlighted
across the four identified contexts. However, two main weaknesses of the review,
and accompanying ecosocial framework, are that they offer little by way of empirical
evidence and little in direction towards meeting the broader research objective of
identifying how trade liberalization and social policy interact to influence health and
its social determinants. This first limitation relates to the fact that few studies
included in the review drew on empirical data. The second limitation relates to the
fact that included studies rarely acknowledge social policies as a health mediating
factor, except in the context of labour markets.

As mentioned in the introduction to this thesis, the research objectives,


questions and methods of this work were not fixed from the beginning, but rather
emerged and evolved as the work progressed. As will be made clear in the following

118
pages, the second part of this thesis was undertaken as a means to build on the
findings of the literature review through empirical case study work.

119
INTRODUCTION TO PART 2
This thesis is conceptually organized into three parts. While the first part
consisted of the literature review (including the development of the adapted ecosocial
framework), the next two parts emerge directly in response to the results of this work.

The literature review presented in the previous two chapters represents the
first attempt at delineating the health impacts of trade liberalization through a
systematic narrative synthesis. While this review is noted to take us some way
towards answering the first research question of this thesis, again, two main
weaknesses are that it offers 1) little in the way of empirical evidence and 2) little in
respect to the broader research objective of identifying how trade liberalization and
social policy interact to influence health and its social determinants (RO1). As
mentioned previously, this first limitation relates to the fact that little empirical work
has been undertaken by the studies identified in the review. The second limitation
relates to that fact that included studies rarely acknowledge social policies as a health
mediating factor, except in the context of labour markets.
With these limitations in mind it was decided to undertake an empirical
exploration of the trade liberalization and health relationship by focusing specifically
on labour markets in order to advance progress towards meeting the first research
objective. It is this empirical work which is the focus of the second and third parts of
the thesis.
Considerations which were raised in the literature review, and the adapted
ecosocial framework itself, also informed various aspects of the next two parts of this
thesis. For example, the definition of trade liberalization developed in Chapter 3 was
used to locate an appropriate instance of trade liberalization to empirically explore.
Moreover, and as will be made clearer in Chapter 5, the adapted ecosocial framework
informed methodological decisions surrounding this empirical work.
Specifically, parts two and three of this thesis are a case study of how trade
liberalization and social protection policies interact to influence distributions of
health. The ‘case’ in this study is the 2005 phase-out of the Multi-fibre Arrangement
(MFA) which after nearly 50 years of protection abruptly liberalized the textile and
clothing sector. As will be made clear in the next chapter, this event fits the
definition of trade liberalization presented in Chapter 3 since it represents a trade
related process characterized by a reduction in government intervention.

120
The units of analysis in this case work are the many low-, middle- and high-
income countries for which the clothing and textile industry is an important source of
employment, especially for women. Since systems of social protection in these
countries are highly diverse, analysing how important determinants of health and
health outcomes changed in these countries after the MFA phase-out can help to
answer the first research objective of this thesis which, again, is to identify how
trade liberalization and social policy interact to influence health and its social
determinants.

It is here that the second research objective of this thesis now assumes its
significance, it is to investigate and analyse how the phase-out of the Multi-Fibre
Arrangement impacted health in countries reliant on the textile and clothing
sector for employment (RO2). The concomitant research questions are as follows:

How did health outcomes change after the phase-out of the Multi-Fibre in
countries reliant on the textile and clothing sector? (RQ2)

What are the potential causal mechanisms responsible for these changes?
(RQ3)

The significance and reasoning behind the development of these research questions
will be discussed in Chapter 5.

Three chapters are included in the next section (i.e. part two) of this thesis.
Chapter 4 will introduce the textile and clothing (T&C) industry as a particularly
valuable sector to consider when exploring the health impacts of trade liberalization
through labour markets. It will also present the main analytical framework used to
guide this case study. Chapter 5 will then detail the methods used to answer research
question two (RQ2) and Chapter 6 will present the results and discussion of this
analysis. The third section of the thesis will move to consider the third and final
research question of this thesis.

121
CHAPTER 4 THE TEXTILE AND CLOTHING SECTOR AND THE MFA
PHASE-OUT
4.1 Introduction
The literature discussed in Chapters 2 and 3 outline a range of potential
pathways between trade liberalization and health. While in totality the evidence
demonstrates a largely nascent field; it was found that the links between trade
liberalization and health have been especially under-conceptualized in relation to
labour markets. However, this under-conceptualization relates mainly to the issue
that specific liberalizing processes have rarely been identified in this context. The
literature by contrast, offers relatively developed theorizations of the pathways to
health via various employment and working conditions.
This chapter will introduce a major liberalization episode in the textile and
clothing (T&C) sector, the phase-out of the Multi-Fibre Arrangement (MFA) in 2005,
as a particularly useful case to consider when exploring the health impacts of trade
liberalization through labour markets. It will also present the main analytical
framework used to guide this case study.
The T&C sector is one of the most globalized industries in the world with
around 130 producing and exporting countries (UNDP, 2006). The reason for its
globalized structure is multi-fold, flowing from the nature of its supply chain which
“typifies the development of global production chains in the world economy”
(Heerden, Berhouet, & Caspari, 2003, p. 1), as well as from preferential trade
schemes, regional trade agreements, historical structures of protection, and most
recently the sector’s liberalization. This chapter will examine each of these aspects of
the sector in turn. In doing so, employment changes which occurred after the MFA
phase-out will be contextualized in ways that have implications for potential
pathways to health. In considering these implications, the chapter will present the
phase-out as a valuable natural policy experiment for exploring the health impacts of
trade liberalization and discuss why the EMCONET framework (Benach et al., 2007)
provides a valuable tool for investigating these impacts.

4.2 The T&C Supply Chain


Simply speaking, the supply chain of the T&C sector begins with the sourcing
of raw materials, including those of both natural (such as cotton and wool) and man-

122
made origin (such as nylon and polyester). These materials are used to create fibres
which are then spun into yarn for the production of woven or knitted textiles. Textiles
are then cut and sewn into clothing, home furnishings, or other technical textile
products, which are then sold to retail outlets (Frederick, 2010; Nordås, 2004).
Each step of the T&C production process is associated with a different
capital/labour ratio and degree of value-added (Heerden et al., 2003). The two ends
of the chain, fibre production along with the design, marketing and retail end,
incorporate the most value-added and capital intensive aspects of production. It is
these production processes that have partly remained in developed countries,
although more recently they have begun to shift towards middle-income countries
(Lopez-Acevedo & Robertson, 2012). Between the ends of production, are the
activities of cutting, sewing and finishing garments. Because these activities are
labour-intensive, have low start-up costs, and depend on relatively simple
technology, they have absorbed a significant proportion of unskilled, mostly female,
labour in developing countries.
While the majority of production and retail takes place in the middle to low
priced segment of the T&C sector, it is worth recognizing that there also remains a
high price, high quality fashion market in developed countries which is
“characterized by modern technology, relatively well-paid workers and designers and
a high degree of flexibility” (Nordås, 2004 p. 3). In this market segment, firms
respond directly to, and often influence, customer preference. In the middle- to low-
priced goods market, consumer demand is also the ultimate determinant of what is
produced and when (Nordås, 2004). However, since this demand is both seasonal and
unpredictable, the industry employs a high proportion of part-time, piece-rate, and/or
household-based workers (UNDP, 2006).
The fragmentation of production can also create opportunities for countries to
move further up the chain into processes that involve more skill and are associated
with higher wages (Lopez-Acevedo & Robertson, 2012). Indeed, it has been argued
that the T&C sector launched most developed and newly industrialized countries into
more intensive forms of manufacturing, guiding their development trajectories
(Adhikari & Yamamoto, 2008).

4.3 Trade Preference Schemes and Regional Trade Agreements


Various trade preference schemes and regional trade agreements have

123
contributed to the globalized nature of the T&C sector (Audet, 2004; OECD, 2004).
In the two dominant import markets, the US and the EU, ‘production sharing
schemes’ and ‘outward processing programmes’ allow garments to be exported to
low-cost locations for assembly and re-importation at lower tariff rates. For low-
wage countries these schemes boost access to high-quality inputs and foreign
markets, for developed countries they can strengthen the competitive position of
domestic suppliers by decreasing labour costs (OECD, 2004). Production sharing
schemes are noted to have increased industrial activities in Mexico, as well as the
Caribbean, Central America and Asia (Seyoum, 2010). Historically, these trade
preference schemes have generated a significant amount of trade, accounting for
15% of EU trade in textiles in 1995, and 24% of total clothing imports into the US in
1999 (OECD, 2004).
Other preferential trade arrangements include the general system of
preferences (GSP) as well as GSP supplemental preference programmes like the
Everything but Arms (EBA) scheme in the EU and the Africa Growth and
Opportunity Act (AGOA) in the US. Through the GSP, 27 developed countries afford
tariff preferences to over 100 beneficiary countries (Lopez-Acevedo & Robertson,
2012). Since 2001, the EBA has granted duty free access to the EU market for all
least developed countries’ products (except arms and ammunitions), without quota
restrictions (with the exception of a few agricultural products which were eventually
phased in). Partly as a result of the EBA, along with other quota and tariff free
access arrangements1 Bangladesh, Tunisia, Morocco and Mauritius represented the
4th, 6th, 7th, and 18th largest suppliers to the EU clothing sector in 2004 (Curran,
2008).
Since 2000, the US has offered preferential market access to eligible countries
in Sub-Saharan Africa under the AGOA. This has greatly expanded exports in the
region, most of which are accounted for by the T&C sector (Keane & Willem te
Velde, 2008). However, rules of origin provisions in both the EU and US trade
preference schemes can also limit the opportunities available to lower income
countries by prohibiting the importation of textiles from third countries for in-
country assembly and exportation (Barber, Gowthaman, & Rose, 2004; Keane &
Willem te Velde, 2008).

1
i.e such as those contained within the Contonou Agreement and various Euro-Med Agreements

124
Regional trade agreements (RTAs) have also shaped production and trade
opportunities in the T&C sector. For example, in the context of NAFTA, Mexico has
been able to expand its T&C industrial activities (OECD, 2004). Indeed, it has been
reported that NAFTA “was instrumental in spearheading Mexico to its position as the
world’s fourth largest clothing exporter in 2001” (OECD, 2004, p. 60).

4.4 Protection and Liberalization of the Sector

The road leading to eventual liberalization of the T&C sector is often


described as beginning in the 1950s, when Japanese cotton textile exports were
restricted into the United States via “voluntary export restraints”. Soon after
“voluntary restrictions” were placed by the United Kingdom on cotton textile exports
from Hong Kong, India, and Pakistan (Spinanger, 1999).
As East Asian countries began developing their T&C industries however,
developed countries enlisted further restraints to regulate a growing range of T&C
imports. In 1961, a forum was established within GATT to address concern among
large importing countries (largely the US and those within the EU) that increasing
imports were causing ‘market disruption’—a concept integrated into GATT only a
year prior and defined as ‘instances of sharp import increases associated with low
import prices not attributable to dumping or foreign subsidies’ (Heron, 2006, p. 3).
This forum resulted in the Short Term Arrangement Regarding International Trade in
Cotton (STA) which unilaterally imposed quotas on cotton-based textiles and
clothing for one year. An extended version of this Agreement was instituted by the
Long Term Agreement Regarding International Trade in Cotton Textiles (LTA) in
1962 (Naumann, 2006).
Because the STA and subsequent LTA regulated only cotton exports, East
Asian countries began to successfully manufacture and export artificial and non-
cotton textile products (Heron, 2006). This prompted protracted renegotiations of the
LTA (again, initiated by Western policy makers) and in 1974, the Multi-Fibre
Agreement (MFA) was born to impose quotas on materials other than cotton (such as
wool and other man-made fibres) (Heron, 2006; Nordås, 2004).
The MFA aimed to gradually open the T&C sector in an orderly way that
would avoid market disruptions. It set targets for increased trade by increasing quotas
at an average annual growth rate of 6% (Naumann, 2006). However, subsequent
negotiating rounds (in 1977, 1981, 1986, and 1991) “served to place increasingly

125
restrictive quotas on most of the leading developing country exporters” (Heron 2006,
p. 4). By the end of the second negotiating round in 1981, 80% of T&C imports from
developing countries into the US were subject to quotas. On the other hand, T&C
trade between developed countries—“which in 1990, accounted for approximately
43 percent of total world trade in textiles and 35 percent of total world trade in
clothing” —was completely free from quantitative restrictions.
In a report published by the WTO, Nordås (2004, p.13) highlights four ways
the MFA violated the free trade principles of the multilateral trading system: (1) it
violated the most favoured nation principle; (2) it applied quantitative restrictions
rather than tariffs; (3) it discriminated against developing countries; and (4) it was
non-transparent. During Uruguay Round negotiations, between 1986 and 1994, it
was finally decided to liberalize the global T&C sector. However, this was not done
with the intention of rectifying these violations of the multilateral trading system,
rather “the most often cited explanation suggests that it was due to the calculations
on the part of developed countries—particularly the US – that the liberalization of
the T&C regime would offer significantly more room for manoeuvre with regard to
trade rights and services” (Heron, 2006, p. 4).
Following the expiration of the MFA in 1994, the Agreement on Textiles and
Clothing established that countries wishing to retain quotas would have to commit to
ten year phase-out period. This phase-out was meant to take place over four stages,
with the last quotas lifted on January 1, 2005. However, using ‘safeguard’ measures,
countries were able to maintain the majority of their quotas until the final phase-out
in 2005. The US for instance, maintained about 89% of its quotas, the EU, 70% and
Canada 79% (Kowalski & Molnár, 2009). This meant that liberalization of the sector
in 2005 was both abrupt and rapid.

4.5 Employment Impacts of the MFA Phase-Out


The significance of the T&C sector for developing countries has grown
rapidly since the 1970s (Lopez-Acevedo & Robertson, 2012). In the mid-1960s
developing countries accounted for around 25% of global apparel exports, by 2000
this figure had increased to 70% (Lopez-Acevedo & Robertson, 2012). For many
low-income countries, apparel exports account for the greatest share of total
manufacturing exports. For example, in Cambodia, Bangladesh, Honduras, and Sri
Lanka, the apparel industry accounted for 85%, 71%, 49%, and 41% of total

126
merchandise exports in 2008, respectively (Lopez-Acevedo & Robertson, 2012).
Furthermore, the sector accounts for a significant proportion of employment
in many countries with large exporting industries. Employment in T&C production
for least developed and low-income countries as a share of total employment in
manufacturing ranges from 35% in selected low income countries, to 75% in
Bangladesh and 90% in other selected LDCs (e.g. Lesotho, Cambodia) (Keane &
Willem te Velde, 2008).
In these countries, the T&C sector is often the largest employer of women
after agriculture (Lopez-Acevedo & Robertson, 2012). This has brought
empowerment for many women in the form of increased control over income and
greater say in household decisions, although these advances have not been without
social costs as working conditions in the sector can be particularly dangerous and
exploitative (UNDP, 2006).
While quotas served to restrict trade from developing countries with
competitive export markets, it also served to develop the T&C industry in countries
that might not have otherwise developed their sector (Lopez Acevedo & Robertson,
2012). This is because quotas created incentives for companies in countries meeting
their export limits to set up production facilities in other, less constrained, countries.
This process was known as ‘quota-hopping’. Korean companies for example, are
noted to have established T&C production facilities in Bangladesh, the Caribbean
and Sub-Saharan Africa; Chinese companies to have “established factories in several
Asian and African locations, Indian companies in Nepal and even relatively minor
players in the global markets such as Sri Lankan and Mauritian business persons
established factories in the Maldives and Madagascar, respectively” (Adhikari and
Yamamoto, 2008, p. 184). Because the allocation of quotas was significantly more
lenient towards countries with little or no T&C industry, less developed countries
saw a significant expansion of their industry (Heerden, Berhouet, & Caspari, 2003).
Prior to the phase out of quotas in 2005, a number of studies predicted large
growth in Chinese T&C exports and in other large T&C exporting countries2. By
contrast, countries that had benefited from quota-hopping were expected to lose
market shares. Early and tentative conclusions reached by the UNDP (2006) and ILO
(2005) largely mirrored these predictions. An analysis by Curran (2008), which

2
See Curran (2008) for a list of key studies and their forecasts.

127
compares a range of forecasted impacts, also finds that predictions were mostly
accurate, except in the case of a few notable suppliers: namely Korea and Taiwan,
who saw their exports decrease in spite of more optimistic estimations, and Pakistan,
Morocco, Tunisia and Turkey, who saw their exports increase in spite of more
pessimistic estimations.
A more recent study published by the World Bank (Lopez-Acevedo &
Robertson, 2012) examines the impact of the MFA phase-out through case studies of
nine developing countries: Bangladesh, India, Sri Lanka, Pakistan, Honduras,
Mexico, Vietnam, Cambodia and Morocco. The main focus of this study is to
examine changes in apparel exports, wages, and employment. This study is unique in
that it offers an assessment of employment impacts after the MFA phase-out for a
range of countries. Most studies examining post-MFA impacts on the sector focus on
changes in exports or market shares, rather than employment.
The study finds that between 2004 and 2008 some developing countries were
able to expand their T&C exports, like Bangladesh (by 69%), India (by 67%) and
Pakistan (by 32%); while others saw their exports severely contract for example,
Mexico by 36%. Similar patterns emerged in regards to market shares, although
some countries, for instance Pakistan, Morocco, Sri Lanka, and Honduras, all saw
market share declines despite increases in export values. In terms of employment,
Pakistan saw an increase despite losses in market share, and Morocco, Sri Lanka and
Honduras all saw decreases in employment despite increases in exports. Of the nine
countries, Mexico is characterized as faring the worst with significant losses in
export values, market share, and employment. A graphical depiction of these results
is displayed below in Figure 20.

128
Figure 20 Employment changes after the MFA phase-out (Lopez-Acevedo & Robertson, 2012)
The empirical analysis within the report suggests that while changes in
employment mostly followed changes in exports, this was not always the case. Both
Honduras and Mexico saw significant losses in employment which the authors relate
to their reliance on free trade agreements with the US. Both countries also saw
decreases in the wage premium3 of working in the T&C sector. It is noted that this
may have significant impacts for poverty in Honduras given the wage premium has
previously been identified as a critical factor in reducing the country’s poverty level.
Interestingly, while Cambodia pursued a path focused on the lower end of the
value chain, the government in Sri Lanka worked to focus on higher value-added
products. Both saw increases in exports, but while employment increased in
Cambodia it is noted to have fallen in Sri Lanka. The authors point to this
comparison to highlight the importance of industry structure and policies, noting that
“[p]olicies to move up the value chain may cause the industry to move away from
precisely those entry-level positions that give opportunities for people to exit
poverty, raising the question of what other policies might be needed to complement
policies that facilitate moving up the value chain” (p. 149).
Overall, results from this study indicate that the MFA phase-out generated
differential employment impacts based on complexities of the global T&C sector
related to countries’ ties to free trade agreements and their position within the value
chain.

3
The difference between average wages in the T&C sector and economy-wide average wages

129
4.6 A Framework for Exploring the Health Impacts of the MFA
Phase-Out
In Chapter 2, it is found that studies relate trade liberalization to health
through a variety of labour market conditions like wages, working conditions, job
loss, and economic insecurity (e.g. Corrigall et al., 2008; Doyal, 2007; Grown, 2005;
Labonté & Schrecker, 2007; Loewenson, 2001; Loewenson et al. 2010). As
previously mentioned however, specific liberalizing processes have rarely been
identified within this context. The MFA phase-out represents a specific instance of
trade liberalization which has had significant impacts on T&C employment patterns
across the world. As such, it represents a valuable natural policy experiment for
exploring the health impacts of trade liberalization in the context of labour markets.
The preceding sections indicate that the T&C sector is extremely globalized
and characterized by international production chains, preferential trading schemes,
regional trade agreements, and in less developed countries, by a largely female
workforce and precarious employment. These considerations present complex
implications for how employment changes after the MFA phase-out might have
impacted health. They indicate for example, that employment changes are likely to
have affected a diverse range of countries in response to their various positions in the
global trading regime. They also indicate a complex set of potential pathways to
health. For example, employment growth in the sector may conceivably be health
promoting if it means improved material conditions through wages. However, that
poor working conditions characterize much of the work in the sector means that there
are significant ways in which employment growth may negatively impact health
outcomes. In relation to employment loss, we can also conceive of both potentially
beneficial and detrimental pathways to health. For example, less exposure to poor
working conditions may promote health while loss of income may facilitate health
deterioration through worsening material circumstances.
Within the literature focusing on trade liberalization and labour markets, the
EMCONET framework (Benach et al., 2007) is found to provide the most
comprehensive outline of the pathways leading to health in the context of labour
markets. This framework can help to deal with the complex ways employment
changes after the MFA phase-out might have impacted health by providing a tool to
understand these different considerations in a logical and coherent fashion. The

130
adapted ecosocial framework presented in Chapter 3 is also useful for thinking about
the relationship between the MFA phase-out and health. This will be further
discussed in Chapter 5 since it relates to methodological considerations.
As discussed in Chapter 2, both a micro- and a macro-level flowchart are used
to illustrate the EMCONET framework (see Figures 21 and 22). Captured within the
micro-level are a number of behavioral, psychosocial, and physiopathological
pathways. At this level, four main categories of risk exposure (physical, chemical,
ergonomic, and psychosocial) are seen to be mediated by social mechanisms and
influenced by six different types of employment conditions (full employment,
unemployment, precarious employment, informal employment, child labour, and
slavery & bonded labour).

Figure 21 Macro-level framework and policy entry points (Benach, Muntaner, & Santana, 2007)

131
Figure 22 Micro-level framework and policy entry points (Benach, Muntaner, & Santana, 2007)
At the macro-level, the framework focuses on the health importance of
protective labour market and welfare state policies. While trade liberalization is not
explicitly depicted within the framework, the authors identify the reduction of
barriers to trade as one of the main drivers of conditions within this macro-level
context.
While the micro-level aspect of the framework encompasses many of the
potentially contradictory pathways to health noted above, the macro-level aspect
indicates that these various conditions and how they impact health will in turn
depend on broader contextual factors such as countries’ labour market and welfare
state policies. This is important with regards to the overarching perspective of this
thesis which is directed at the macro-level and social (fundamental) determinants of
health. Moreover, because countries impacted by the MFA phase-out are likely to
have different systems of social protection, analysing how important determinants of
health and health outcomes changed in these countries after the MFA phase-out can
help to answer the first research objective of this thesis which is to identify how trade
liberalization and social policy interact to influence health and its social
determinants.

132
CHAPTER 5 METHODS: AN EMPRICAL CROSS-CASE ANALYSIS
5.1 Introduction
In Chapter 4 it is argued that the health impacts of the MFA phase-out are
likely to be characterized by a complex combination of causal considerations.
Recognizing this complexity, it was judged that taking advantage of a number of
countries’ experiences of the MFA phase-out would provide greater insight into how
social policies interact with trade liberalization to influence health.

For this reason, the second objective of this thesis is to investigate and
analyse how the phase-out of the Multi-Fibre Arrangement impacted health in
countries reliant on the textile and clothing sector for employment (RO2).

The concomitant research questions are:

RQ2. How did health outcomes change after the phase-out of the Multi-
Fibre in countries reliant on the textile and clothing sector?

RQ3. What are the potential causal mechanisms responsible for these
changes?

This chapter will detail the main method used to answer the first of these two
research questions. It will begin by briefly describing some methodological
considerations for exploring the health impacts of the MFA phase-out. In doing so, it
will introduce the methodological value of fuzzy-set qualitative comparative analysis
(fsQCA). It will then move to consider broader benefits and limitations of fsQCA.
Here the significance of the second research question (RQ3) will become clear. Next,
the analytical steps of fsQCA will be discussed, including the selection of cases,
health outcomes and causal conditions. Chapter 6 will then present the results of the
fsQCA.

5.2 Methodological considerations


While there are many well-known methods for cross-case analyses, the large
majority of public health scholarship which evaluates macro-level phenomena via
cross-national comparisons has relied on conventional linear regression models.
These methods assume the existence of a single, necessary and sufficient,
explanatory model, and average evidence across cases to quantify the net effect of
each explanatory variable by keeping all others in the analysis constant (Ragin, 2006).
133
In doing so, these models often neglect to take sufficient account of the contextual
dimensions of cases and the interactive effects of causal pathways (Ragin, 2006).

Such methods therefore, do not align well with the EMCONET framework
(Benach et al., 2007) which suggests that the health impacts of the MFA phase-out
are likely to vary according to different macro-level contextual conditions. They also
do not align well with the adapted ecosocial framework of trade liberalization and
health, presented in Chapter 3, which characterizes the relationship between trade
liberalization and health via reference to a complex set of concurrent and interacting
pathways.

For this reason, fuzzy-set qualitative comparative analysis (fsQCA) emerged


as the preferred method to answer the second research question of this thesis. This is
because in contrast to linear methods, FsQCA requires that a model be specified
whereby independent variables are assumed to combine in a way that responds to the
unique contextual environments of cases. As an established methodology used in
political sciences (for a review see Rihoux et al. (2013)), fsQCA offers an innovative
and promising methodological technique towards the objectives of this thesis. It is
described in greater depth in the next section.

5.3 Fuzzy-set Qualitative Comparative Analysis

Fuzzy-set qualitative comparative analysis (fsQCA) is a configurational


approach grounded in set-theory (Ragin, 2000; Ragin, 2008). That it is
configurational means that its focus is on whether specific combinations of causal
conditions (configurations) are associated with an outcome of interest. Through the
concept of ‘multiple conjunctural causation’, fsQCA recognizes that 1) outcomes are
often produced via a combination of conditions, 2) that the same outcome may be
produced by different combinations of conditions, and 3) that the context within
which conditions combine can influence the type of impact made on the outcome
(Rihoux, 2006).

The crux of fsQCA is that it enables logical statements to be made about


causal conditions that are either necessary or sufficient for a given outcome. A
condition is necessary when it must be present for an outcome to occur. Causal
conditions are said to be sufficient if the outcome occurs whenever the causal
conditions are present.

134
Besides being especially apt at addressing causal complexity, fsQCA is also
useful for supporting exploratory analyses and theory development (Grimm, 2006;
Ragin & Schneider, 2011). This is because fsQCA is a “case-oriented” analytic
method. This means that in contrast to the “variable-oriented” method, it focuses on
how conditions influence an outcome in a context-specific manner (Ragin, 2006).
This context-specific approach allows for greater exploration into the causal
conditions that are associated (or not) with an outcome.

Moreover, in fsQCA there is constant dialogue between theory and evidence


(Ragin, 2000). This allows for the identification of inconsistencies in a way which
can suggest modifications to the analysis and its surrounding theory. This iterative
method is thus extremely useful for theory development. These characteristics of
fsQCA are especially important since our understanding of the health impacts of the
MFA phase-out is relatively undeveloped.

However, fsQCA also presents some important challenges and limitations.


For example, fsQCA requires the researcher to make many subjective decisions
which introduces the potential for substantial bias into the analysis. However, this
problem is not exclusive to fsQCA, as researcher bias also characterizes variable-
oriented techniques (Ragin, 2000). Furthermore, this challenge can be dealt with
through complete transparency and careful application of substantive and theoretical
knowledge. In the subsequent sections of this chapter, explicit details of the decisions
made within each step of the fsQCA will be made clear. Moreover, just as in a
traditional regression analysis, the results generated by an fsQCA should be viewed
as potentially falsifiable hypotheses to be further tested.

There are also limits to the explanatory power of fsQCA. Recall that fsQCA is
focused on identifying whether specific configurations are associated with a certain
outcome. To do this, fsQCA looks at all of the logically possible combinations of
causal conditions. The number of logically possible combinations of causal
conditions grows exponentially with the number of included causal conditions so that
there are 2k possible configurations (combinations of causal conditions), where k
represents the number of causal conditions. Consequently, it has been argued that
there is a ratio of cases to causal conditions, below which there is a high chance that
a fsQCA will find an association due to random variation (Marx & Dusa, 2011).
Researchers are thus advised to pay careful attention to both the number of cases and

135
causal conditions included in a fsQCA. For example, a fsQCA with five causal
conditions should ideally include at least 18 cases (Marx & Dusa, 2011). As will be
made clearer below, the fsQCA undertaken within this analysis has five causal
conditions, and contains well over 18 cases.

Perhaps the main limitation of fsQCA is that, like regression analyses, it


identifies associations, not causality. It is for this reason that the third and final
research question of this thesis asks which causal mechanisms are responsible for the
changes we see in health after the MFA phase-out. As a case-oriented approach,
fsQCA can be used to identify particular types of cases for detailed within-case
analyses that can focus on identifying causal processes. How this was undertaken to
answer the third research question (RQ3) is discussed in the third section of this
thesis, beginning with Chapter 7.

5.4 Operationalizing fsQCA


FsQCA can typically be broken down into three different stages. In the first
stage, decisions are made regarding the selection of cases, outcome indicators and
causal conditions. Included cases are then assigned fuzzy membership scores for
each of the outcome indicators and causal conditions. These scores indicate each
case’s degree of membership in the category formed by the indicator. A fuzzy-set
approach responds to the need for variables which can be finely calibrated. Fuzzy
scores thus range from 0 to 1, where the former indicates non-membership in the set
created by the indicator, and the latter indicates full-membership.

In the second stage of fsQCA, examinations of necessity and sufficiency are


undertaken. In the latter analysis a truth table is constructed which outlines all
logically possible combinations of causal conditions and their relationship to the
outcomes. The truth table also demonstrates which combinations of conditions best
describe included cases. In the final stage of fsQCA, a process of ‘logical reduction’
is undertaken through which a simplified statement is made about which
combinations of conditions are necessary and sufficient for an outcome.

This chapter will now move to discuss how these three stages were
approached as a means to answer the second research objective of this thesis: How
did health outcomes change after the phase-out of the Multi-Fibre in countries
reliant on the textile and clothing sector?

136
5.4.1 Stage 1: Outcome Indicators, Case Selection and Causal Conditions
In the first stage, decisions were made regarding the selection of outcome
indicators, cases, and causal conditions.

Outcome indicators
As discussed in Chapter 4, the MFA phase-out is likely to have had impacts
on health in a range of different ways. An ideal way to explore the health impacts of
this liberalization episode would have been to follow up with workers who were
impacted by resulting shifts in employment, as well as their families. Unsurprisingly
however, this type of data is not readily available. For this reason, it was decided to
investigate changes in health at the national, population-level. A central question of
this methodological choice is to what extent we can expect health outcomes at the
national level to respond to changes in employment in the T&C sector.

There is a relatively robust body of literature in public health which finds


evidence for changes in national-level health outcomes as a result of changing
macro-economic conditions. For example, there is a substantial amount of research
focused on the health impacts of economic crises/recessions (Burgard et al., 2013;
Catalano et al., 2011; Suhrcke et al., 2011). A main pathway through which
recessions are seen to impact national levels of health is job loss (Stuckler et al.,
2009). As described in the previous chapter, there is evidence that some countries
experienced severe losses of employment after the MFA phase-out.

Although there is a sizeable body of literature relating macro-economic


conditions to changes in health via job loss, fewer studies have measured the effect
of employment growth on health (Rueda et al., 2012). However, that employment is
considered a fundamental determinant of health means that there are many pathways
through which we can expect growth in employment to potentially impact health at
the national level (Bambra, 2011; Benach et al., 2007).

Moreover, while much of the literature which deals with the health impacts of
changing macroeconomic conditions is focused on the developed world, there is also
evidence that such changes can have important implications for national levels of
health in poorer countries (Hopkins, 2006; Suhrcke & Stuckler, 2012).

For these reasons, the answer to the question of whether we can expect
changes in population health as a result of changes in employment in the T&C sector,

137
will depend in large part on which health outcomes are explored, how reliant
countries are on the T&C sector for employment, as well as how large changes in
employment were after the MFA phase-out. For reasons that are discussed below,
adult female and infant mortality rates were chosen as the most appropriate outcome
indicators to explore the health impacts of the MFA phase-out. Moreover, in ways
that will soon be made clearer, steps have been taken to ensure that countries
included in this analysis are particularly reliant on the textile and clothing sector for
employment. It will be also demonstrated that many countries saw considerably large
shifts in their T&C employment after the MFA phase-out.

Female Adult Mortality Rates


Mortality rates were chosen to explore the health impacts of the MFA phase-
out on the basis of evidence which links unemployment to mortality (Bambra, 2011;
Bartley, Ferrie, & Montgomery, 2006; Bartley & Plewis, 2002; Roelfs, Shor,
Davidson, & Schwartz, 2011). Female mortality rates were chosen in particular since
the textile and clothing sector workforce is largely female. Another benefit of
mortality rates is that they are generally available for a large set of countries,
although this data is not without its limitations (as discussed in greater detail below).

A recent meta-analysis finds that unemployment is associated with an


increase relative risk of all-cause mortality (Roelfs et al., 2011). In this study, the risk
of death among individuals who experienced unemployment is found to be 63%
higher than those who did not, after adjustment for age and other covariates. This
risk increases to 73% when only the first 5 years of unemployment are considered.
However, the magnitudes of these associations are typically greater for men than
they are for women, and studies included in this meta-analysis focus exclusively on
developed countries.

One possible reason for this gender difference is lower labour force
participation rates among women, another is that employment status is more central
to men’s identities than to women’s (Roelfs et al., 2011). However, this first issue is
addressed in this case study by focusing on a female dominated industry. Moreover,
it would seem that employment is an extremely central aspect of women’s identities
in the lower- and middle-income countries impacted by the MFA phase-out, since
this employment represents an important channel to improved material circumstances
(Kabeer, 2000; Lynch, 2007).
138
Furthermore, Suhrcke and Stuckler (2012) suggest that income shocks, like
those resulting from unemployment, are likely to generate even worse health
outcomes in poor countries since in these countries people are less likely to have
access to social protection, and live closer to or in abject poverty. This idea is
supported by evidence from Hopkins (2006) who finds that the East Asian crisis
(which was characterized by significant job losses) was associated with (short-lived)
increases in the mortality rates of Thailand and Indonesia. While little health impact
was found in Malaysia, Hopkins attributes this to the fact that, rather than following
the World Bank prescription for adjustment, which included cuts in public spending,
the country chose its own path to adjustment that minimized health impacts for
example through a fixed exchange rate and capital controls. This finding is aligned
with the analytical frameworks used to guide this study, which emphasize the
mediating role of contextual features in shaping health outcomes.

Historically, adult mortality rates have been reported by the UN Population


Division, the WHO and the World Bank. However, the usefulness of this data has
been hindered by a range of well-known weaknesses, due partly to incomplete
registrations systems and partly to the dearth of global health research focused on all-
cause adult mortality (Hill, 2003; Koyanagi & Shibuya, 2010). These agencies for
instance, have relied on models which extrapolate adult mortality from child
mortality. This has been identified as especially problematic in the context of
HIV/AIDS. Ambiguity in both the sources of data and the methods used, have also
hindered replication of results (Koyanagi & Shibuya, 2010). Indeed it has been
acknowledged by the World Bank (Bos, Vu, & Stephens, 1992) that documenting
short-term fluctuations based on these past methods, let alone linking them to
changing socio-economic contexts, requires far greater detail than such methods can
afford.

By contrast, authors of the dataset this study draws on, Rajaratnam et al.
(2010b), specifically acknowledge that one of the main strengths of the data is that it
can be used to better investigate changes in mortality rates over time and in doing so,
linked to changes in various socio-economic contexts. Moreover, new advances in
estimation techniques are argued to have greatly ameliorated past weaknesses
(Koyanagi & Shibuya, 2010).

139
In the data used for this study, adult mortality is summarized by the
probability that an individual who is 15 years old will die before reaching age 60
(45q15 in standard life table notation). It is based on the assumption that the age-
specific mortality conditions of the year are constant throughout an individual’s life.
Mortality rates from 187 countries, from 1970 to 2010, are estimated from a variety
of sources including “(1) vital registration data, (2) sample registration systems
(when available), and nationally representative survey or census data that enable
direct estimation of age-specific adult mortality rates from questions about either (3)
deaths in the household or (4) the survival of siblings of a respondent” (Rajaratnam
et al., 2010b, p. 1705). Moreover, the methods used demonstrate a higher predictive
validity than earlier estimates and are both transparent and replicable (Koyanagi &
Shibuya, 2010; Rajaratnam et al., 2010b).

While it has been acknowledged that these estimates are the “best that can
done”, it has also been cautioned that “in the places where adult mortality is highest,
we do not have the kind of data that is required to monitor and evaluate local and
international health interventions” (Deaton, 2011, p. 19). Despite these drawbacks,
this data has been drawn upon more confidently than other estimates in recent work
(see for example, Chongsuvivatwong et al., 2011). Moreover, as will be noted in
relation to case selection, countries with high and/or fluctuating levels of uncertainty
in their data are excluded from this analysis.

Infant Mortality Rates


Infant mortality rates (IMR) were chosen to explore the health impacts of the
MFA phase-out for two main reasons. First, IMR was considered a valuable indicator
since employment conditions don’t just impact workers but also their families
(Bambra, 2011). Although it is unclear what proportion of T&C workers are having
children across different countries, the second reason IMR was considered is that it is
especially recognized for its rapid response and sensitivity to more macro-level
policy changes (Bambra, 2006; Bezruchka, 2012; Ferrarini & Norström, 2010;
Reidpath & Allotey, 2003). This means that the MFA phase-out may have impacted
IMR both directly, through the employment conditions of T&C workers having
children, and indirectly if the phase-out influenced health important conditions at the
national level.

140
Two national level conditions highlighted by the EMCONET framework for
their health importance in the context of labour markets are economic inequality and
material deprivation (Benach et al., 2007). Both of these conditions may have been
impacted through a large shift in T&C employment after the MFA phase-out,
through for example, the additional provision of wages to people who were otherwise
not earning, or through the loss of wages for those losing employment. Infant
mortality rates have been associated with both of these conditions in other studies
(Schell, Reilly, Rosling, Peterson, & Ekström, 2007; Wilkinson & Pickett, 2010)

The responsiveness of IMR to macro-level conditions, in addition to its cross-


national availability, makes it one of the most widely used comparative indicators of
population-level health; however, it is an indicator not without its disadvantages
(Mathers, Salomon, & Murray, 2003; Reidpath & Allotey, 2003). Across countries,
different classification and reporting protocols can make cross-national comparisons
of IMR difficult (Bezruchka, 2012; Howell & Blondel, 1994). Moreover, estimations
are often based on small samples of large populations (Reidpath & Allotey, 2003). It
has also been argued that IMR fails to successfully capture average population health
(Mathers et al., 2003).

For these reasons IMRs are sometimes exchanged for, or supplemented with,
other population health measures such as life expectancy, disability-adjusted life
expectancy (DALE) and health-adjusted life expectancy (HALE). However, these
indicators are also not free from weaknesses. There are concerns, for instance, about
the usefulness of life expectancy as a short-term health outcome measure given its
inability to respond within any reasonable time to policy changes (Frank & Haw,
2011). Debates also exist over the relative merit of DALE/HALE (Arnesen & Nord,
1999; Gold, Stevenson, & Fryback, 2002; Reidpath & Allotey, 2003).

Despite its drawbacks, IMR continues to be an indicator of high interest in


comparative population-level analyses (Bambra, 2006; Chuang, Sung, Chang, &
Chuang, 2013). Given its sensitivity to the social determinants of health, such as
living conditions and economic status (Reidpath & Allotey, 2003), it is a particularly
useful proxy measure of population-level health in the present study. It is also an
extremely politically relevant indicator given the Fourth Millennium Development
Goal of reducing by two-thirds, between 1990 and 2015, the under-five mortality
rate-- of which IMR is one of three identified indicators.

141
Moreover, the authors of the dataset this study draws on, Rajaratnam et al.
(2010a), have recently updated international IMR data for 187 countries from 1970
to 2009. Here IMR is summarized by the probability of death before age 1,
conditional on surviving to 1 month. This work improves on previous IMR estimates
by using more recent data and new estimation methods which reduce bias and
measurement error as well as improve predictive validity. Using this data also allows
for cross-country comparisons on infant mortality trends in a way which is both
replicable and reduces sources of non-sampling error. Rather than extrapolating from
small surveys, the authors pool data from a variety of sources including civil
registrations, summary birth histories in censuses, complete birth histories, and
survey programmes such as the WHO’s Demographic and Health Surveys.

Furthermore, countries with high and/or fluctuating levels of uncertainty in


their data are excluded from this analysis, as will be noted later in relation to case
selection. Finally, the claim by Mathers et al. (2003) that IMR is not by itself a
sufficient indicator of population health is addressed since this study also explores
changes in adult female mortality rates.

Case Selection
Countries used for the analysis of the MFA phase-out were selected on the
basis of their reliance on the textile and clothing sector for employment. Countries
were deemed reliant on the sector if in 2004 (or in the latest year for which data is
available between 2000 and 2004), employment in the T&C sector (as a proportion
of total manufacturing employment) was greater than 10%, given that more than 10%
of the working population is employed in manufacturing. Total manufacturing and
T&C sector employment figures were obtained from the United Nations Industrial
Development Organization (UNIDO) Industrial Statistics Databases (2011). Data on
the proportion of the working population employed in industry were obtained from
the 2011 edition of the World Bank Development Indicators.

Table 1 displays the 65 countries which were found to have more than 10% of
total manufacturing employment in the T&C sector. Of these 65 countries, three
were excluded for having less than 10% of the working population employed in
industry: Ethiopia; Madagascar; and the United Republic of Tanzania. Additionally,
the following 9 countries were excluded because data on the proportion of the
working population employed in industry were not available: Afghanistan; China,

142
Taiwan Province; Eritrea; Fiji; Iran; Lesotho; Palestinian Territories; Tunisia and
Yemen. Therefore, 53 countries were initially considered as reliant on the T&C
sector for employment.

143
Table 1: T&C Employment as a percentage of total manufacturing employment

T&C % of T&C % of
employment working employment working
Country Year as % of total pop. in Country Year as % of total pop. in
manufacturing Industry manufacturing Industry
employment* *** employment* ***
Afghanistan 2004 13.23 NA Latvia 2004 13.74 27.3
Albania 2004 26.76 13.6 Lesotho 2004 86.86 NA
Azerbaijan 2004 12.92 11.9 Lithuania 2004 21.97 28.2
Bangladesh** 2004 40 13.7 Madagascar 2004 56.02 6.7
Bolivia 2001 14.81 20.5 Mauritius 2004 66.83 33.5
Botswana 2004 33.41 22.6 Mexico 2003 14.42 24.8
Brazil 2004 12.76 21 Mongolia 2004 51.23 16.1
Bulgaria 2004 29.53 32.9 Morocco 2004 41.53 19.5
Cambodia 2000 73.60 10.5 Nepal 2002 28.28 13.4
Palestinian
China 2004 18.11 22.5 2004 21.04 NA
Territories
China, Hong
2004 27.90 15.6 Peru 2004 35.60 41.7
Kong SAR
China,
2004 82.70 25.2 Philippines 2003 18.54 15.8
Macao SAR
China,
Taiwan 2001 11.12 NA Poland 2004 10.24 28.8
Province
Colombia 2004 22.30 19.9 Portugal 2004 24.24 31
Costa Rica 2003 17.60 22 Puerto Rico 2000 18.42 19.4
Croatia 2004 13.78 29.8 Qatar 2004 21.51 41
Republic of
Ecuador 2004 10.87 17.5 2004 10.04 27.5
Korea
Egypt 2004 28.51 20 Romania 2004 24.29 31.2
Eritrea 2004 20.65 NA Saudi Arabia 2003 14.75 21
Serbia &
Estonia 2004 17.98 34.9 2001 17.82 26.9
Montenegro
Ethiopia 2004 25.34 5.3 Slovakia 2004 10.56 39
Fiji 2004 37.05 NA South Africa 2004 12.82 26.1
Greece 2004 13.73 22.5 Sri Lanka 2001 49.32 24.1
Syrian Arab
Guatemala 2004 40.57 19.5 2004 25.41 25.6
Republic
Hungary 2004 10.19 32.8 Thailand 2002 18.04 20.5
The former
Yugoslav
India 2004 21.03 16.1 2004 36.42 32.8
Republic of
Macedonia
Indonesia 2004 22.90 18 Tunisia 2002 50.60 NA
Iran (Islamic
2004 11.06 NA Turkey 2004 34.55 23
Republic of)
United
Italy 2004 11.14 30.8 Republic of 2004 10.89 2.6
Tanzania
Jamaica 2004 16.28 18.3 Uruguay 2004 13.14 21.4
Jordan 2004 15.49 21.8 Viet Nam 2004 23.04 17.4
Kuwait 2001 16.75 18.3 Yemen 2004 12.96 NA
Kyrgyzstan 2004 11.48 17.6
Sources: United Nations Industrial Development Organization (2011): INDSTAT2, Industrial
Statistics Database (Edition: 2011). ESDS International, University of Manchester. DOI:
10.5257/unido/indstat2/2011. Industrial Statistics Database 2011 at the 2-digit level of ISIC Code
(Revision 3). * ISIC codes 17 for textiles, 18 for wearing apparel and D for total manufacturing. **

144
Since data was not available on Bangladesh in the UNIDO database, and because the country is oft
cited as extremely reliant on the textile and clothing sector, data for Bangladesh taken from the IMF
Working Paper by Mlachila and Yang (2004): The End of Textiles Quotas: A Case Study of the
Impact on The End of Textiles Quotas on Bangladesh. *** World Bank (2011): World Development
Indicators (Edition: September 2011). ESDS International, University of Manchester. DOI:
http://dx.doi.org/10.5257/wb/wdi/2011-09

Final set of included countries


While 53 countries were initially identified as reliant on the T&C textile
sector for employment, the final set of countries used for analysis was significantly
smaller with 29 countries analyzed in respect to infant mortality rates, and 27
countries analyzed in respect to adult female mortality rates, yielding a total of 32
unique countries (see Table 2 below for a final list of included countries). This
number of cases well exceeds the minimum number of cases below which there is a
high chance that a fsQCA will find an association due to random variation (Marx,
2011). As mentioned in the beginning of this chapter, a fsQCA with five causal
conditions should ideally include at least 18 cases (Marx, 2011).

Interestingly, some more highly developed countries were identified as


reliant on the textile and clothing sector for employment, such as Italy and Portugal.

Inclusion of countries was limited by the quality of mortality data sources, as


well as by the availability of data for indicators used to operationalize the causal
conditions discussed below. Countries were excluded from the study if mortality data
was characterized by relatively high and/or erratic levels of uncertainty as
documented in country specific graphs of trend data provided by the authors of the
mortality data sets (Rajaratnam et al. 2010ab).

Table 2 Final list of included countries and the type of mortality change investigated

Mortality Mortality Mortality


Country Change Country Change Country Change
Investigated Investigated Investigated
Azerbaijan IMR Hungary Female Philippines Both
Mortality
Bangladesh Both India Both Poland Both
Brazil Both Indonesia IMR Portugal Both
Bulgaria Both Italy Both Republic of Both
Korea
China Both Kyrgyzstan Both Romania Female
Mortality
Colombia Both Latvia Both Slovakia Both
Croatia Both Lithuania Both South Africa Both
Ecuador Both Mauritius Both Sri Lanka IMR
Egypt IMR Mexico Both Thailand Both
Estonia Female Mortality Morocco IMR Turkey Both
Greece Both Peru Both

145
Causal conditions
There are a variety of approaches that can be used to select causal conditions
for inclusion in a fsQCA model (Amenta & Poulsen, 1994). Since the EMCONET
framework used to guide this research predicts multiple pathways to health (Benach,
et al. 2007) conditions can be seen to be selected via the conjunctural approach. In
this approach, conditions are expected to combine in different ways to impact an
outcome. This approach is described as best aligned with the characteristics of a
fsQCA (Amenta & Poulsen, 1994).

The following five causal conditions were selected on the basis of the
EMCONET framework and the thesis’s focus on macro-level conditions: countries’
(1) Level of development; (2) Labour market protection; (3) Welfare state protection;
(4) Employment loss after the MFA phase-out (5) Employment growth after the
MFA phase-out. Additional details on the choice of these conditions are presented in
the following sections.

Assignment of Fuzzy Membership Scores


For each of the outcome indicators and causal conditions selected, scores
were assigned which describe each case’s degree of membership in the category
formed by the indicator. Again, a fuzzy-set approach responds to the need for
variables which can be finely calibrated. Fuzzy scores thus range from 0 to 1, where
the former indicates non-membership in the set created by the indicator, and the
latter indicates full-membership (Ragin, 2008). Final fuzzy-set scores for each of the
outcomes and causal conditions were calculated as described below.

Health Outcomes
This study was interested in how adult female and infant mortality rates
changed after the MFA phase-out. There are a variety of methods available for
analysing changes in mortality rates over time. This study relies primarily on
calculations of relative and absolute changes in mortality rates before and after the
MFA phase-out. Before these calculations are discussed however, it is first worth
considering two other highly relevant methodologies for analysing changes in
mortality rates over time and why this present study decided against using them.

Interrupted time series analysis (ITS) is a regression technique frequently


used to examine trends over time. It is argued to be one of the strongest quasi-
experimental approaches to evaluating the impact of an intervention or event.

146
Comparing trends before and after an intervention/event, ITS is able both to detect
immediate impacts as well as those less sudden (Gillings, Makuc, & Siegel, 1981).
Another benefit of an ITS design is that it is able to determine whether changes in
trends are large enough to be statistically significant. However, ITS requires 50-100
dependent variable observations over time and data on infant and female mortality
rates is available only from the 1970s onwards.

Joinpoint regression is another technique which allows for the assessment of


changes in trends over time. One of the major strengths of this methodology is that it
is able to identify a point in time in which a significant change in trend has occurred.
Generally speaking, joinpoint regression analyzes trend data and breaks it up into
different line segments at points where a change in trend is statistically different
(p<0.05). Using a Monte Carlo Permutation test of significance, the model tests the
null hypothesis (H0: there are k0 joinpoints) against the alternative hypothesis (H 1:
there are k1 joinpoints) (Kim, Fay, Feuer, & Midthune, 2000). When fit on the log
scale, the slopes of these line segments are interpretable as the annual percent change
in the rate. The number of joinpoints identifiable within an analysis is limited by the
number of data points being analysed. The maximum number of joinpoints
identifiable for any analysis is 5.

While this technique is appropriately aligned with the objectives of this work,
its utility is limited for two main reasons. The first reason relates to data availability
issues. In general it is recommended that data over longer periods of time be used for
Joinpoint analyses. While the dataset used for this analysis provides infant and
female mortality rates from the late 70s, the reliability of this data changes over time
(as evidenced by varying levels of uncertainty). In this scenario, it is suggested that
the standard errors of the data be included in the analysis. However, while this data
was requested from the authors of the datasets used in this analysis, it was not made
available at the time of this work.

The second reason why the utility of Joinpoint regression was limited relates
to the fact that in many of the countries included in this analysis, mortality rates have
gone through periods of dramatic reductions since the 1970s. With this in mind,
along with the consideration that only a limited number of joinpoints are able to be
identified over the course of the data, it wasn’t clear whether the analysis would then
be less sensitive to more subtle changes occurring in more recent years.

147
Relative and absolute changes in mortality rates before and after the MFA phase-
out
Relative and absolute changes in mortality rates were calculated from
Rajaratnam et al. (2010ab) to capture both immediate and delayed changes in health
after the MFA phase-out. Immediate changes are reflected in the differences in
mortality rates between 2004 and 2005, while delayed changes were calculated based
on the five year period preceding (2000-2004) and following the MFA phase-out
(2005-2009). While absolute figures give a precise account of change across
countries, relative changes are better at capturing comparative changes over time.
These calculations are displayed below across table 3 and 4 in relation to female
mortality rates, and across tables 5 and 6 in relation to infant mortality rates.

148
Adult Female Mortality
Table 3 Absolute Changes in Adult Female Mortality Rates

Adult Female Mortality Rates Absolute Reduction in Adult Difference


(per 1000) Female Mortality Rates Between
Country Pre MFA Post MFA Pre & Post
2004- (2000- (2005- MFA
2000 2004 2005 2009 2005 2004) 2009) Periods
Bangladesh 135.6 126.1 124.8 121.4 1.3 9.5 3.4 -6.1
Brazil 122.4 119 117.9 111.4 1.1 3.4 6.5 3.1
Bulgaria 97.8 91.8 92 88.5 -0.2 6 3.5 -2.5
China 118.6 105.4 102.5 92.5 2.9 13.2 10 -3.2
Colombia 88.7 78.7 76.2 68.6 2.5 10 7.6 -2.4
Croatia 74.9 66.1 64.9 62.9 1.2 8.8 2 -6.8
Ecuador 98.6 90.7 88.7 80.6 2 7.9 8.1 0.2
Estonia 119.2 102 98.7 87.6 3.3 17.2 11.1 -6.1
Greece 48.4 46.5 45.8 42.4 0.7 1.9 3.4 1.5
Hungary 115 107.7 106.6 104 1.1 7.3 2.6 -4.7
India 188.6 166.4 161.7 147.3 4.7 22.2 14.4 -7.8
Italy 50.5 45 44.2 42 0.8 5.5 2.2 -3.3
Korea 61.7 51.2 49.1 42 2.1 10.5 7.1 -3.4
Kyrgyz Republic 154.7 146 145.6 144.4 0.4 8.7 1.2 -7.5
Latvia 120.9 116.7 117.9 120 -1.2 4.2 -2.1 -6.3
Lithuania 105.1 104.5 108.5 115.5 -4 0.6 -7 -7.6
Mauritius 110 108.5 108.1 106.6 0.4 1.5 1.5 0
Mexico 101.4 95.9 94.7 89.8 1.2 5.5 4.9 -0.6
Peru 97.6 95 94.1 82.9 0.9 2.6 11.2 8.6
Philippines 118.8 120.1 119.8 116 0.3 -1.3 3.8 5.1
Poland 86.3 78.3 78.8 78.3 -0.5 8 0.5 -7.5
Portugal 66.4 59.1 56.8 50.2 2.3 7.3 6.6 -0.7
Romania 108.3 100.4 98.5 91 1.9 7.9 7.5 -0.4
Slovak Republic 81.8 77.9 77.4 75.2 0.5 3.9 2.2 -1.7
South Africa 316.5 430.2 444.2 450 -14 -113.7 -5.8 107.9
Thailand 124.9 117.2 114.1 102.7 3.1 7.7 11.4 3.7
Turkey 102.7 92.8 91.2 86.1 1.6 9.9 5.1 -4.8

149
Table 4 Relative Changes in Adult Female Mortality Rates

Percent Change in Adult


Adult Female Mortality Rate Difference
Female Mortality
(per1000) between pre
Reduction & post MFA
Country
periods
2004- 2000- 2005-
2000 2004 2005 2009 (percentage
2005 2004 2009
points)
Bangladesh 135.6 126.1 124.8 121.4 1.30 7.01 2.72 -4.28
Brazil 122.4 119 117.9 111.4 1.10 2.78 5.51 2.74
Bulgaria 97.8 91.8 92 88.5 -0.20 6.13 3.80 -2.33
China 118.6 105.4 102.5 92.5 2.90 11.13 9.76 -1.37
Colombia 88.7 78.7 76.2 68.6 2.50 11.27 9.97 -1.30
Croatia 74.9 66.1 64.9 62.9 1.20 11.75 3.08 -8.67
Ecuador 98.6 90.7 88.7 80.6 2.00 8.01 9.13 1.12
Estonia 119.2 102 98.7 87.6 3.30 14.43 11.25 -3.18
Greece 48.4 46.5 45.8 42.4 0.70 3.93 7.42 3.50
Hungary 115 107.7 106.6 104 1.10 6.35 2.44 -3.91
India 188.6 166.4 161.7 147.3 4.70 11.77 8.91 -2.87
Italy 50.5 45 44.2 42 0.80 10.89 4.98 -5.91
Korea 61.7 51.2 49.1 42 2.10 17.02 14.46 -2.56
Kyrgyz
154.7 146 145.6 144.4 0.40 5.62 0.82 -4.80
Republic
Latvia 120.9 116.7 117.9 120 -1.20 3.47 -1.78 -5.26
Lithuania 105.1 104.5 108.5 115.5 -4.00 0.57 -6.45 -7.02
Mauritius 110 108.5 108.1 106.6 0.40 1.36 1.39 0.02
Mexico 101.4 95.9 94.7 89.8 1.20 5.42 5.17 -0.25
Peru 97.6 95 94.1 82.9 0.90 2.66 11.90 9.24
Philippines 118.8 120.1 119.8 116 0.30 -1.09 3.17 4.27
Poland 86.3 78.3 78.8 78.3 -0.50 9.27 0.63 -8.64
Portugal 66.4 59.1 56.8 50.2 2.30 10.99 11.62 0.63
Romania 108.3 100.4 98.5 91 1.90 7.29 7.61 0.32
Slovak Republic 81.8 77.9 77.4 75.2 0.50 4.77 2.84 -1.93
-
316.5 430.2 444.2 450 -14.00 -1.31 34.62
South Africa 35.92
Thailand 124.9 117.2 114.1 102.7 3.10 6.16 9.99 3.83
Turkey 102.7 92.8 91.2 86.1 1.60 9.64 5.59 -4.05

In terms of adult female mortality, reductions in rates were taking place across
all except two countries prior to the MFA phase-out: the Philippines, where mortality
rates increased by 1.1% from 2000-2004; and South Africa, where rates increased by
35.9%. In the period following the MFA phase-out, this trend had reversed in the
Philippines, with mortality rates declining by 3.2% between 2005 and 2009. In South
Africa however, increases in adult female mortality rates were greatly reduced but
not reversed; in the period following the MFA phase-out, the country saw a rise in
mortality of 1.3%, a difference of 34.6 percentage points from the previous period.

150
Besides South Africa, Lithuania and Latvia also saw their adult female
mortality rates increase following the MFA phase-out. In Lithuania, adult female
mortality rates were reduced by 0.6% in the period prior to the phase-out; however,
following the phase-out this trend was reversed and mortality rates increased by
6.5%. Similarly, in Latvia adult female mortality rates were reduced by 3.5% in the
period prior to the phase-out; however, this trend was reversed and mortality rates
increased by 1.8% following the phase-out.

In terms of more immediate changes directly following the MFA phase-out


(between the years of 2004 and 2005), only five countries saw their mortality rates
increase: South Africa (by 14%), Lithuania (by 4%), Latvia (by 1.2%), Poland (by
0.5%), and Bulgaria (by 0.2%). In the years following 2005, only South Africa,
Latvia and Lithuania continued to have increasing adult female mortality rates.

The Philippines is the only country where adult female mortality rates were
increasing prior to the MFA phase-out but reversed in 2005, though the change from
2004 to 2005 was relatively small at 0.3%. In the remaining 21 countries, mortality
rates continued to decrease immediately following the MFA phase out, however to
varying extents. Six of these 21 countries saw changes between 2004 and 2005 of
less than 1% (Kyrgyz Republic, Mauritius, Slovak Republic, Greece, Italy, and Peru);
seven had changes between 1-2% (Brazil, Hungary, Croatia, Mexico, Bangladesh,
Turkey, and Romania); five had changes between 2-3% (Ecuador, Korea, Portugal,
Colombia, and China); Thailand and Estonia saw changes between 3-4% and finally,
India saw its adult female mortality rates decrease from 2004 to 2005 by 4.7%.

For the majority of countries which saw immediate reductions in adult female
mortality rates between 2004 and 2005, decreases across the period following the
phase-out were lower than they were in the period preceding it. This is true for all of
the 22 countries which saw immediate reductions with the exception of the following
seven: the Philippines, Greece, Peru, Brazil, Romania, Ecuador, Portugal, and
Thailand.

Overall, the data indicate that relative changes improved after the MFA
phase-out in 10 of the 27 analyzed countries. Some of these improvements were
relatively minor, in four of these countries (Romania, Ecudaor, Portugal and Brazil),
percent changes in adult female mortality rates improved by 0.3 to 2.7 percentage

151
points after the MFA phase-out. In the remaining countries which saw an
improvement in adult female mortality rate reductions (Greece, Philippines, Peru,
Thailand, and South Africa), improvements ranged from 3.5 to 34.6 percentage
points.

By contrast, 17 countries saw a slowing in their adult female mortality rate


reductions. Again the extent of these changes varied, with Mexico experiencing a
change of less than 1%; three countries experiencing changes between 1-2%
(Colombia, China, and the Slovak Republic); three countries experiencing changes
between 2-3% (Bulgaria, Korea and India), Estonia and Hungary experiencing
changes between 3-4%; four countries experiencing changes between 4-5% (Turkey,
Bangladesh and the Kyrgyz Republic); and five countries experiencing changes
greater than 5% (Croatia, Poland, Lithuania, Italy and Latvia).

152
Infant Mortality Rates
Table 5 Absolute changes in infant mortality rates

Infant Mortality Rate (per Absolute Change in Infant


1000) Mortality Reduction Difference
Post Between Pre
Country
Pre MFA MFA and Post MFA
2004- (2000- (2005- Periods
2000 2004 2005 2009 2005 2004) 2009)
Azerbaijan 24.53 19.1 17.76 13.17 1.34 5.43 4.59 -0.84
Bangladesh 23.15 18.71 17.76 14.79 0.95 4.44 2.97 -1.47
Brazil 13.33 10.9 10.39 8.55 0.51 2.43 1.84 -0.59
Bulgaria 7.21 5.49 5.15 4.25 0.34 1.72 0.9 -0.82
China 9.49 6.75 6.17 4.84 0.58 2.74 1.33 -1.41
Colombia 8.92 7.28 6.86 5.33 0.42 1.64 1.53 -0.11
Croatia 2.13 1.79 1.7 1.33 0.09 0.34 0.37 0.03
Ecuador 16.83 14.31 13.71 11.47 0.6 2.52 2.24 -0.28
Egypt 15.82 11.79 10.92 8.08 0.87 4.03 2.84 -1.19
Greece 1.88 1.51 1.45 1.06 0.06 0.37 0.39 0.02
India 22.36 19.44 18.8 16.42 0.64 2.92 2.38 -0.54
Indonesia 17.16 15.12 14.68 12.89 0.44 2.04 1.79 -0.25
Italy 1.45 0.94 0.9 0.78 0.04 0.51 0.12 -0.39
Korea 2.7 2.43 2.29 1.75 0.14 0.27 0.54 0.27
Kyrgyz
Republic 21.11 19.38 18.98 17.04 0.4 1.73 1.94 0.21
Latvia 4.79 3.69 3.51 3.23 0.18 1.1 0.28 -0.82
Lithuania 3.89 3.26 3.15 2.63 0.11 0.63 0.52 -0.11
Mauritius 5.02 3.67 3.51 3.12 0.16 1.35 0.39 -0.96
Mexico 12.91 10.75 10.27 8.55 0.48 2.16 1.72 -0.44
Morocco 15.97 12.91 12.22 9.79 0.69 3.06 2.43 -0.63
Pakistan 29.44 27.07 28.15 24.78 -1.08 2.37 3.37 1
Peru 13.93 11.06 10.48 8.58 0.58 2.87 1.9 -0.97
Philippines 11.7 10.5 10.15 8.84 0.35 1.2 1.31 0.11
Poland 2.54 1.98 1.9 1.65 0.08 0.56 0.25 -0.31
Portugal 2.48 1.44 1.27 0.99 0.17 1.04 0.28 -0.76
Slovak
Republic 3.59 3.1 2.96 2.31 0.14 0.49 0.65 0.16
South Africa 16.21 23.09 25.1 24.02 -2.01 -6.88 1.08 7.96
Sri Lanka 4.29 3.03 2.83 2.29 0.2 1.26 0.54 -0.72
Thailand 3.06 2.46 2.33 1.87 0.13 0.6 0.46 -0.14
Turkey 14.88 12.71 12.33 10.27 0.38 2.17 2.06 -0.11

153
Table 6 Relative changes in infant mortality rates

Infant Mortality Rate (per Percent Change in Infant Mortality


1000) Reduction Difference
between pre
Country
Pre MFA Post MFA and post
2004- (2000- (2005- MFA periods
2000 2004 2005 2009 2005 2004) 2009)
Azerbaijan 24.5 19.1 17.8 13.2 7.02 22.14 25.84 3.71
Bangladesh 23.2 18.7 17.8 14.8 5.08 19.18 16.72 -2.46
Brazil 13.3 10.9 10.4 8.6 4.68 18.23 17.71 -0.52
Bulgaria 7.2 5.5 5.2 4.3 6.19 23.86 17.48 -6.38
China 9.5 6.8 6.2 4.8 8.59 28.87 21.56 -7.32
Colombia 8.9 7.3 6.9 5.3 5.77 18.39 22.30 3.92
Croatia 2.1 1.8 1.7 1.3 5.03 15.96 21.76 5.80
Ecuador 16.8 14.3 13.7 11.5 4.19 14.97 16.34 1.37
Egypt 15.8 11.8 10.9 8.1 7.38 25.47 26.01 0.53
Greece 1.9 1.5 1.5 1.1 3.97 19.68 26.90 7.22
India 22.4 19.4 18.8 16.4 3.29 13.06 12.66 -0.40
Indonesia 17.2 15.1 14.7 12.9 2.91 11.89 12.19 0.31
Italy 1.5 0.9 0.9 0.8 4.26 35.17 13.33 -21.84
Korea 2.7 2.4 2.3 1.8 5.76 10.00 23.58 13.58
Kyrgyz
Republic 21.1 19.4 19.0 17.0 2.06 8.20 10.22 2.03
Latvia 4.8 3.7 3.5 3.2 4.88 22.96 7.98 -14.99
Lithuania 3.9 3.3 3.2 2.6 3.37 16.20 16.51 0.31
Mauritius 5.0 3.7 3.5 3.1 4.36 26.89 11.11 -15.78
Mexico 12.9 10.8 10.3 8.6 4.47 16.73 16.75 0.02
Morocco 16.0 12.9 12.2 9.8 5.34 19.16 19.89 0.72
Peru 13.9 11.1 10.5 8.6 5.24 20.60 18.13 -2.47
Philippines 11.7 10.5 10.2 8.8 3.33 10.26 12.91 2.65
Poland 2.5 2.0 1.9 1.7 4.04 22.05 13.16 -8.89
Portugal 2.5 1.4 1.3 1.0 11.81 41.94 22.05 -19.89
Slovak
Republic 3.6 3.1 3.0 2.3 4.52 13.65 21.96 8.31
South Africa 16.2 23.1 25.1 24.0 -8.71 -42.44 4.30 46.75
Sri Lanka 4.3 3.0 2.8 2.3 6.60 29.37 19.08 -10.29
Thailand 3.1 2.5 2.3 1.9 5.28 19.61 19.74 0.13
Turkey 14.9 12.7 12.3 10.3 2.99 14.58 16.71 2.12

In terms of infant mortality, reductions in rates were taking place across all
countries prior to the MFA phase-out except in South Africa, where mortality rates
increased by 42.4% from 2000-2004. In the period following the MFA phase-out,
this trend had reversed, with mortality rates declining by 4.3% between 2005 and
2009. Reductions in infant mortality rates continued in all countries in the period
following the MFA phase-out, although to varying extents.

154
South Africa is also the only country which saw an immediate negative
change in infant mortality rates directly following the MFA phase-out (between the
years of 2004 and 2005), with its IMR increasing by 8.7%. In the remaining 28
countries, IMR continued to decrease immediately following the MFA phase out.
Between 2004 and 2005 three of these 26 countries saw changes between 2-3%
(Kyrgyz Republic, Indonesia and Turkey); five had changes between 3-4% (India,
the Philippines, Lithuania, Greece and Poland); eight had changes between 4-5%
(Ecuador, Italy, Mauritius, Mexico, Slovak Republic, Brazil, Latvia, and Croatia); six
saw changes between 5-6% (Bangladesh, Peru, Thailand, Morocco, Korea, and
Colombia); and finally six saw changes greater than 6% (Bulgaria, Sri Lanka,
Azerbaijan, Egypt, China and Portugal).

In contrast to what was found with female mortality rates, for the majority of
countries which saw immediate reductions in infant mortality rates between 2004 and
2005, decreases across the period following the phase-out were higher than they were
in the period preceding it. This is true for all of the 26 countries which saw
immediate reductions with the exception of the following 12: India, Poland, Italy,
Mauritius, Brazil, Latvia, Bangladesh, Peru, Colombia, Bulgaria, China and Portugal.

Overall, the data indicate that infant mortality rates improved after the MFA
phase-out in 17 of the 29 analyzed countries. Some of these improvements were
relatively small, in ten of these countries (Thailand, Indonesia, Lithuania, Egypt,
Morocco, Mexico, Ecuador, Kyrgyz Republic, Turkey, and the Philippines) percent
changes in IMR improved by 0.1 to 2.6 percentage points after the MFA phase-out.
In the remaining seven countries (Azerbaijan, Colombia, Croatia, Greece, Slovak
Republic, Korea, and South Africa), improvements ranged from 3.7 to 46.7
percentage points.

By contrast, 12 countries saw a slowing in their IMR reductions. Again the


extent of these changes varied, with India and Brazil experiencing a change of less
than 1%; Peru and Bangladesh experiencing changes of 2.5%; four countries
experiencing changes between 6-10.5% (Sri Lanka, Poland, China, and Bulgaria) and
four countries experiencing changes between 15-22% (Italy, Portugal, Mauritius and
Latvia).

155
Two other general points regarding mortality rate changes after the MFA
phase are worth noting. First, in general, changes in infant mortality rates were larger
than those of female mortality rates. Second, countries which saw improvements in
female mortalities rates after the MFA phase-out were not necessarily the same
countries which saw improvements in infant mortality rates; similarly countries
which saw their female mortality rates worsen were not the same countries which
saw worsening IMR.

Fuzzy-set scores for changing adult female and infant mortality rates
Fuzzy membership scores for two outcome sets: a ‘health improving’ and a
‘health worsening’ set were assigned both for adult female and infant mortality rates.
It was decided to use relative changes to assign fuzzy-set scores since they are better
at capturing comparative changes over time than absolute changes. Relative changes
in mortality rates were calculated based on the five-year period preceding (2000-
2004) and following the MFA phase-out (2005-2009). In terms of female mortality
rates, this selected time period is consistent with studies which show an association
between unemployment and adult mortality rates after a similarly short period of
time (Bartley et al., 2006; Brenner, 1995; Hopkins, 2006; Lundin, Lundberg,
Hallsten, Ottosson, & Hemmingsson, 2010; Mustard et al., 2013; Roelfs et al., 2011).
However, as previously mentioned comparable evidence could not be found on the
relationship between employment growth and adult mortality.

A ‘direct calibration method’ was used to assign fuzzy membership scores in


the outcome sets of ‘health improving’ (Ragin, 2008; Schneider & Wagemann, 2012).
In the direct calibration method three thresholds need to be specified which
correspond to the qualitative breakpoints of full membership (1), the cross-over point
(.5), and full non-membership (0). Once these breakpoints are specified, fuzzy
membership scores are assigned by the fsQCA software (version 2.5). Generally
speaking, the software calculates scores by translating variable scores into the metric
of log odds. The technical details of direct calibration are explained in Ragin (2008).

In relation to female mortality rates the qualitative breakpoints for the health
improving set were conceptualized respectively as a 3% increase in mortality rate
reduction, a 0% change in mortality rate reduction and a 3% decrease in mortality
rate reductions between the pre and post MFA periods. In relation to infant mortality
rates these breakpoints were conceptualized respectively as a 4% increase in
156
mortality rate reduction, a 0% change in mortality rate reduction and a 4% decrease
in mortality rate reductions.

Fuzzy-set scores in the membership sets of ‘health worsening’ were taken to


be the negation of health improving scores and calculated by subtracting a country’s
score in the health improving membership score from 1.

The above breakpoints were chosen with a consideration of the distribution of


changes across countries. In relation to female mortality rates only 10 countries saw
an improvement in health after the MFA phase-out. Changes of less than 1% seemed
too small to deem an improvement, and beyond these smaller improvements there is
a break in the data until about the 3% mark. Moreover, a sensitivity analysis was
carried out whereby fuzzy-set scores were calibrated and run on both lower and
higher breaks points. When thresholds were set at 1% and 2% no changes were seen
in the results of the fuzzy-set analysis in regards to health improvement. At a 4%
threshold, an additional set of countries were characterized as experiencing a health
improvement however, this is because moving to this higher threshold increased the
fuzzy-set membership score of countries which did not see an improvement in health.
Because it was countries which did not see an improvement in health which were
altering these ‘health improving’ results, it was decided against using this higher
threshold.

On the other hand, 17 countries saw their adult female mortality rates worsen
after the MFA phase-out and unlike health improving data, there is no noticeable
break in the health worsening data. Therefore, while the 3% break point seemed
appropriate for denoting meaningful changes in the health improving data, for the
health worsening data there is a less strong argument for the use of this figure. When
the sensitivity analysis was carried out in regards to the membership set of health
worsening, using a threshold lower than 3% meant that certain types of countries
which were otherwise characterized as experiencing a health worsening change in
trend, were no longer characterized this way. This is largely because of increases in
the fuzzy scores of countries which did not see a health worsening change in trend.
However, it was decided against using a lower threshold for three main reasons. First,
the differences in fuzzy membership scores across the different thresholds were
relatively small for example, in the case of Ecuador a 2% threshold would mean a
fuzzy-set score of 0.16, whereas a 3% threshold would mean a fuzzy-set score of

157
0.25. Second, many countries saw a greater than 2% decrease in their female
mortality rate reductions after the MFA phase-out, therefore while the changes in
fuzzy-set membership scores between a 2% and 3% threshold were small, decreasing
the threshold would result in less fuzzy-set score diversity. Finally, it seems strange
to conclude that while a 3% increase in mortality rate reduction represents a health
improvement, a different (and lower) figure would denote a meaningful change in the
health worsening data.

In regards to infant mortality rates, similar techniques were used to arrive at


the 4% threshold. Many more countries saw an improvement in their reductions of
infant mortality rates after the MFA phase-out than was seen in regards to female
mortality rates. While breaks in the data were less evident (except at very high
levels), in terms of health improvement a noticeable change seemed to occur around
4%. In terms of health worsening, a noticeable change seemed to occur at around 6%.
However, sensitivity analyses showed no difference in terms of the results of the
fuzzy-set analysis between using a 4% or 6% threshold both in terms of health
improving and health worsening outcomes. Moreover, no changes in the overall
fuzzy-set results were seen using a 1%, 2%, 3%, or 5% threshold. A 4% threshold
was chosen over 6% threshold, since this distributed the range of membership scores
more evenly.

Final fuzzy scores for the four health outcome sets are displayed in table 7
below.

158
Table 7 Fuzzy membership scores for the outcome sets of Health Improving and Health
Worsening adult female and infant mortality rates

Difference in
Difference in
adult female
infant Fuzzy Fuzzy
mortality rate Fuzzy Score Fuzzy Score
mortality rate Score in Score in
reductions in Health in Health
Country reductions Health Health
between pre Improving Worsening
between pre Improving Worsening
(2000-2004) and Set Set
and post MFA Set Set
post (2005-2009)
periods
MFA periods
Azerbaijan NA NA NA 3.71 0.94 0.06
Bangladesh -4.28 0.01 0.99 -2.46 0.14 0.86
Brazil 2.74 0.94 0.06 -0.52 0.4 0.6
Bulgaria -2.33 0.09 0.91 -6.38 0.01 0.99
China -1.37 0.2 0.8 -7.32 0 1
Colombia -1.30 0.21 0.79 3.92 0.95 0.05
Croatia -8.67 0 1 5.80 0.99 0.01
Ecuador 1.12 0.75 0.25 1.37 0.74 0.26
Egypt NA NA NA 0.53 0.6 0.4
Estonia -3.18 0.04 0.96 NA NA NA
Greece 3.50 0.97 0.03 7.22 1 0
Hungary -3.91 0.02 0.98 NA NA NA
India -2.87 0.05 0.95 -0.40 0.43 0.57
Indonesia NA NA NA 0.31 0.56 0.44
Italy -5.91 0 1 -21.84 0 1
Korea -2.56 0.07 0.93 13.58 1 0
Kyrgyz
Republic -4.80 0.01 0.99 2.03 0.82 0.18
Latvia -5.26 0.01 0.99 -14.99 0 1
Lithuania -7.02 0 1 0.31 0.56 0.44
Mauritius 0.024 0.51 0.49 -15.78 0 1
Mexico -0.250 0.44 0.56 0.02 0.5 0.5
Morocco NA NA NA 0.72 0.63 0.37
Peru 9.24 1 0 -2.47 0.14 0.86
Philippines 4.27 0.99 0.01 2.65 0.88 0.12
Poland -8.64 0 1 -8.89 0 1
Portugal 0.63 0.65 0.35 -19.89 0 1
Romania 0.32 0.58 0.42 NA NA NA
Slovak
Republic -1.93 0.13 0.87 8.31 1 0
South
Africa 34.62 1 0 46.75 1 0
Sri Lanka NA NA NA -10.29 0 1
Thailand 3.83 0.98 0.02 0.13 0.53 0.47
Turkey -4.05 0.02 0.98 2.12 0.83 0.17

Countries’ Level of Development


A development indicator was included in this analysis to differentiate between
countries’ level of development when evaluating how causal conditions combine to

159
influence changes in health outcomes. Membership in the development set was
calculated using data from the United Nation’s Human Development Index.

The Human Development Index (HDI) is a composite index of life


expectancy, literacy and gross national income (GNI) per capita through which
countries receive a development score between 0 and 1 (UNDP, 2006). Data from the
2006 United Nation’s Human Development Index was used to assign scores in the
membership set of Highly Developed Countries. This data reflected the conditions in
countries in 2004, the year prior to the MFA-phase-out.

A direct calibration method was used to assign fuzzy membership scores in a


way which aligned with the Index’s rating of countries into the categories of High
Human Development, Medium Human Development and Low Human Development.
In the membership set of Highly Developed, the qualitative breakpoints of full
membership (1), the cross-over point (.5), and full non-membership (0) were
conceptualized as 0.9, 0.8 and 0.5 respectively. All countries receiving a HDI score
of less than 0.5 are deemed by the Index as having low human development. The
cross-over point was chosen at 0.8 since below this point countries are deemed as
having medium human development.

While HDI has been challenged on a variety of accounts (Neumayer, 2001),


the criticism most relevant to this analysis posits that the HDI ranking is not
sufficiently different from a ranking based on income (Kelley, 1991; McGillivray,
1991). However, it was found that a fuzzy-set similarly calibrated based on countries’
GNI per capita, and using the World Bank’s categorization of high, upper-middle,
and low income countries, was substantially different than the fuzzy-set constructed
on the basis of the HDI index. This is illustrated by the scatter plots contained in
figures 23-26 below which map the relationships between the two types of fuzzy-set
development membership sets and adult female and infant mortality rates in 2004.

To maintain a similar calibration approach, the GNI-based fuzzy membership


set was directly calibrated using a full-membership breakpoint of $12,276 GNI per
capita (i.e. the threshold at which the World Bank characterizes a country as high-
income); a cross-over point of $10,000 GNI per capita and a full non-membership
breakpoint of $1,005 GNI per capita (i.e. the threshold at which the World Bank
characterizes a country as low-income). Data from the World Bank Development

160
Indicators Database (2011) was used for this analysis. The cross-over point was
chosen at $10,000 since below this figure the World Bank deems countries as
middle-income. In fact, countries with a GNI per capita of $3,976 to $12,275 are
characterized by the World Bank as upper-middle income. However, using $10,000
was chosen as a cross-over point instead of $12,275 to allow for some distance from
the threshold at which the World Bank deems countries as high-income.

HDI Fuzzy-set Scores versus Adult Female Mortality Rates


Adult Female Mortality Rates (2004)

450
400
350
300
per 1000

250
200
150
100
50
0
0 0.2 0.4 0.6 0.8 1
HDI (2004) Fuzzy-Set Scores

Figure 23 HDI fuzzy-set scores versus adult female mortality rates

GNI Per Capita Fuzzy-set Scores Versus Adult Female


Mortality Rates
Adult Female Mortality Rates (2004)

450
400
350
300
per 1000

250
200
150
100
50
0
0 0.2 0.4 0.6 0.8 1
GNI Per Capita (2004) Fuzzy-set Scores

Figure 24 GNI per capita fuzzy-set scores versus adult female mortality rates

161
HDI Fuzzy-Set Scores versus IMR
25.0

IMR per 1000 (2004)


20.0

15.0

10.0

5.0

0.0
0 0.2 0.4 0.6 0.8 1
HDI (2004) Fuzzy-set Scores

Figure 25 HDI fuzzy-set scores versus IMR

GNI Per Capita Fuzzy-Set Scores versus IMR


25.0

20.0
IMR per 1000 (2004)

15.0

10.0

5.0

0.0
0 0.2 0.4 0.6 0.8 1
GNI Per Capita (2004) Fuzzy-set Scores

Figure 26 GNI per capita fuzzy-set scores versus IMR

The scatterplots in figures 23-26 also show that the fuzzy-sets constructed on
the basis of GNI per capita are also less successful at grouping countries among
others with relatively similar health profiles in terms infant mortality rates. For this
study, it is important that an indicator of development group countries in this way,
given that it may be more difficult for substantial changes in mortality rates to occur
in countries with already low rates, and vice versa.

As previously mentioned, a development indicator was included largely to


contextualize how changes in employment after the MFA phase-out impacted the
health of countries across different levels of development. However, that

162
development might have changed in countries after the MFA phase-out must also be
taken into account. For this reason, how countries HDI ranking changed in the period
from 2004-2009 was assessed.

To do so, the HDI ranking of countries from 2005-2009 was obtained from
the UNDP International Human Development Indicators Database. However, this
data is not directly comparable to the 2004 data obtained from the 2006 Human
Development Report. This is because of data revisions and changes in methodology
in later reports. While these changes have been applied to data annually from 2005
onwards, for historical data prior to 2005, revisions have only been calculated across
five year intervals beginning in 1980.

Therefore, in order to get an idea of how development rankings might have


changed within countries between 2004 and 2009, countries’ average annual change
in HDI was calculated for the years 2005-2009. This average annual change was then
used to estimate what the revised 2004 rankings might look like by subtracting the
average annual change from the 2005 ranking figure. With this information, it was
then possible to calculate countries’ ‘estimated average HDI ranking from 2004-
2009’. An average ranking was calculated since it is more relevant for this analysis to
approximate a country’s overall development level during this time period than how
much a country’s development changed. Again this relates back to the purpose of
including a development indicator in this analysis: i.e. to contextualize how changes
in T&C employment impacted countries’ health across different levels of
development.

These calculations demonstrate relatively small differences between the 2004


rankings and the estimated averaged ranking from 2004-2009. The construction of an
additional development fuzzy-set using the ‘estimated-averaged’ HDI rankings from
2004-2009, illustrate how minor the differences are for a fuzzy-set analysis. See
Table 8 below.

163
Table 8 Fuzzy-set scores using the ‘estimated-averaged’ HDI rankings from 2004-2009

HDI (Estimated and Averaged


HDI (2004)
2004-2009)
Index Fuzzy Averaged Index Fuzzy Score
Country Fuzzy Score
Score Score Score Difference
Azerbaijan 0.736 0.35 0.736 0.35 0
Bangladesh 0.53 0.06 0.486 0.04 0.02
Brazil 0.792 0.48 0.715 0.3 0.18
Bulgaria 0.816 0.62 0.768 0.42 0.2
China 0.768 0.42 0.673 0.22 0.2
Colombia 0.79 0.48 0.706 0.28 0.2
Croatia 0.846 0.8 0.798 0.5 0.3
Ecuador 0.765 0.41 0.715 0.3 0.11
Egypt 0.702 0.27 0.638 0.17 0.1
Estonia 0.858 0.85 0.834 0.73 0.12
Greece 0.921 0.97 0.871 0.89 0.08
Hungary 0.869 0.89 0.819 0.64 0.25
India 0.611 0.13 0.535 0.07 0.06
Indonesia 0.711 0.29 0.610 0.13 0.16
Italy 0.940 0.99 0.880 0.92 0.07
Korea 0.912 0.97 0.885 0.93 0.04
Kyrgyz
Republic 0.705 0.28 0.621 0.14 0.14
Latvia 0.845 0.79 0.804 0.53 0.26
Lithuania 0.857 0.85 0.810 0.57 0.28
Mauritius 0.800 0.5 0.728 0.33 0.17
Mexico 0.821 0.65 0.765 0.41 0.24
Morocco 0.640 0.17 0.577 0.1 0.07
Peru 0.767 0.42 0.714 0.3 0.12
Philippines 0.763 0.41 0.652 0.18 0.23
Poland 0.862 0.87 0.810 0.57 0.3
Portugal 0.904 0.96 0.815 0.61 0.35
Romania 0.805 0.54 0.772 0.43 0.11
Slovak
Republic 0.856 0.84 0.828 0.7 0.14
South
Africa 0.653 0.19 0.613 0.13 0.06
Sri Lanka 0.755 0.39 0.686 0.24 0.15
Thailand 0.784 0.46 0.686 0.24 0.22
Turkey 0.757 0.39 0.696 0.26 0.13

Because these results suggest that incorporating changes in HDI rankings over
time would only marginally change fuzzy-set scores, it was then decided to use the
original 2004 HDI rankings data (from the 2006 Human Development Report) to
construct the development fuzzy-set as previously explained. This is because these
calculations are more straight forward and do not rely on estimations.

164
Final fuzzy scores for countries’ level of development are displayed in table 9
below.

Table 9 Fuzzy-set membership scores in the set of Highly Developed Countries

HDI Index
Fuzzy
Country Score
Score
(2004)
Azerbaijan 0.736 0.35
Bangladesh 0.53 0.06
Brazil 0.792 0.48
Bulgaria 0.816 0.62
China 0.768 0.42
Colombia 0.79 0.48
Croatia 0.846 0.8
Ecuador 0.765 0.41
Egypt 0.702 0.27
Estonia 0.858 0.85
Greece 0.921 0.97
Hungary 0.869 0.89
India 0.611 0.13
Indonesia 0.711 0.29
Italy 0.94 0.99
Korea 0.912 0.97
Kyrgyz 0.705 0.28
Republic
Latvia 0.845 0.79
Lithuania 0.857 0.85
Mauritius 0.8 0.5
Mexico 0.821 0.65
Morocco 0.64 0.17
Peru 0.767 0.42
Philippines 0.763 0.41
Poland 0.862 0.87
Portugal 0.904 0.96
Romania 0.805 0.54
Slovak 0.856 0.84
Republic
South 0.653 0.19
Africa
Sri Lanka 0.755 0.39
Thailand 0.784 0.46
Turkey 0.757 0.39

Protection of Labour Market and Welfare State Policies


The level of protection offered by labour markets and welfare states is an
important determinant of health in the EMCONET framework (Benach et al., 2007).
This framework identifies three health important characteristics of labour markets:

165
labour regulations, collective bargaining and the power of trade unions. Due to the
dearth of internationally available data on the latter two characteristics, labour
regulation is the main characteristic explored for the purposes of this fsQCA. Labour
regulation is defined by the authors as referring “both to the specific regulation of the
labour market (employment protection legislation) and to welfare state benefits
related to the salaried relationship, such as benefits for those involuntarily leaving the
labour market, for example, income security measures for the unemployed” (Benach,
Muntaner, & Santana, 2007 p. 30). Recognizing the significant overlap between
labour markets and welfare states, the authors draw on Esping-Andersen and
Regini’s (2001) work in highlighting that “the welfare state and the labour market
are two institutions deeply inter-connected and it is not possible to understand the
labour market without considering the welfare state institutions that surround it”
(Benach et al., 2007 p. 31).

To allow for a better understanding of how both labour markets and welfare
states combine conjunctively with other causal conditions, labour regulation is
explored in this analysis across the domains of labour market protection, and welfare
state protection independently.

Labour Market Protection


Cross-country investigations often use ratification of ILO Conventions as a
proxy indicator of labour standards (Block, 2005). While the ILO Conventions are an
internationally recognized benchmark of employee protection, ratification as an
indicator it is not without its drawbacks. It’s main drawback is acknowledged by
Block (2005) who highlights that because not all countries will have the same
capacity to implement and enforce conventions, “there is likely to be substantial
measurement error in any variable that considers labour standards to be roughly
equivalent in two or more countries that have ratified the same Conventions” (Block,
2005, p. 11). Indeed, as acknowledged by both Block (2005) and Rodrick (1996),
while many less developed countries have ratified more Conventions than the United
States, it seems unreasonable to assume that they have better labour standards.

Despite this drawback, ratification of ILO conventions has been described as


a reliable way of measuring labour market protection in less developed countries and
that “it is fair to assume … that labour market protections will be relatively low in
countries that have ratified a very low number of ILO conventions (Rudra, 2007, p.
166
388). For this reason, the level of labour market protection was measured and
calibrated based on the number of fundamental ILO Conventions ratified by a
country. The Fundamental ILO conventions are:

 Freedom of Association and Protection of the Right to Organise Convention,


1948 (No. 87)
 Right to Organise and Collective Bargaining Convention, 1949 (No. 98)
 Forced Labour Convention, 1930 (No. 29)
 Abolition of Forced Labour Convention, 1957 (No. 105)
 Minimum Age Convention, 1973 (No. 138)
 Worst Forms of Child Labour Convention, 1999 (No. 182)
 Equal Remuneration Convention, 1951 (No. 100)
 Discrimination (Employment and Occupation) Convention, 1958 (No. 111)

A six-value fuzzy-set was used to assign membership in the set of Protective


Labour Market Policies. This membership set takes into account both the number of
fundamental Conventions ratified in 2004, before the MFA phase-out, as well as
additional ratifications that were made following the phase-out. The decision was
made to incorporate changes in ratification after the MFA phase-out since the health
impact of these protections would still be important following the liberalization
episode. However, evidence was not found which suggested that labour market
protections changed in countries in specific response to the MFA phase-out.

Countries that had ratified all 8 of the fundamental Conventions by 2004 were
assigned a fuzzy membership score of 1, indicating full membership in the set of
Protective Labour Market Policies; countries that had ratified 7 of the fundamental
conventions by 2004 were assigned a membership score of 0.6, unless they had
ratified the final Convention before 2009, in which case they were assigned a
membership score of 0.8. A membership score of 0.6 indicates membership “more or
less” in the set, whereas a membership score of 0.8 indicates a country is
qualitatively “mostly in” the set. Countries which ratified 6 of the fundamental
Conventions were assigned a membership score of 0.4 indicating they are “more or
less” out of the set. However, again if these countries ratified the final 2 Conventions
before 2009 they were assigned a membership of 0.8; if they ratified 1 further
Convention before 2009 they were assigned a membership score of 0.6. Countries

167
were assigned a membership score of 0.2 if they ratified 4-5 of the Fundamental
Conventions by 2004, indicating that they are “mostly out” of the set of countries
with Protective Labour Market Policies. One country within this group, Lativa, had
ratified all 8 Conventions by 2006 and thus was assigned a membership score of 0.8.
Finally, countries were assigned a membership score of 0 if they had ratified 3 or less
of the fundamental conventions by 2004. However, one country within this group,
China, had ratified an additional Convention in 2006 and so was assigned a
membership score of 0.2.

Final fuzzy scores for countries’ labour market protection are displayed in table 10
on the next page.

168
Table 10 Fuzzy membership scores in the set of Protective Labour Market Policies

Freedom of Forced Discrimination Child labour


association labour Fuzzy
Country
Score
C087 C098 C029 C105 C100 C111 C138 C182
Azerbaijan 1992 1992 1992 2000 1992 1992 1992 2004 1
Bangladesh 1972 1972 1972 1972 1998 1972 2001 0.6
Brazil 1952 1957 1965 1957 1965 2001 2000 0.6
Bulgaria 1959 1959 1932 1999 1955 1960 1980 2000 1
China 1990 2006 1999 2002 0.2
Colombia 1976 1976 1969 1963 1963 1969 2001 2005 0.8
Croatia 1991 1991 1991 1997 1991 1991 1991 2001 1
Ecuador 1967 1959 1954 1962 1957 1962 2000 2000 1
Egypt 1957 1954 1955 1958 1960 1960 1999 2002 1
Estonia 1994 1994 1996 1996 1996 2005 2007 2001 1
Greece 1962 1962 1952 1962 1975 1984 1986 2001 1
Hungary 1957 1957 1956 1994 1956 1961 1998 2000 1
India 1954 2000 1958 1960 0.2
Indonesia 1998 1957 1950 1999 1958 1999 1999 2000 1
Italy 1958 1958 1934 1968 1956 1963 1981 2000 1
Korea 1997 1998 1999 2001 0.2
Kyrgyzstan 1992 1992 1992 1999 1992 1992 1992 2004 1
Latvia 1992 1992 2006 1992 1992 1992 2006 2006 0.8
Lithuania 1994 1994 1994 1994 1994 1994 1998 2003 1
Mauritius 2005 1969 1969 1969 2002 2002 1990 2000 0.8
Mexico 1950 1934 1959 1952 1961 2000 0.4
Morocco 1957 1957 1966 1979 1963 2000 2001 0.6
Peru 1960 1964 1960 1960 1960 1970 2002 2002 1
Philippines 1953 1953 2005 1960 1953 1960 1998 2000 0.8
Poland 1957 1957 1958 1958 1954 1961 1978 2002 1
Portugal 1977 1964 1956 1959 1967 1959 1998 2000 1
Romania 1957 1958 1957 1998 1957 1973 1975 2000 1
Slovakia 1993 1993 1993 1997 1993 1993 1997 1999 1
South Africa 1996 1996 1997 1997 2000 1997 2000 2000 1
Sri Lanka 1995 1972 1950 2003 1993 1998 2000 2001 1
Thailand 1969 1969 1999 2004 2001 0.2
Turkey 1993 1952 1998 1961 1967 1967 1998 2001 1

Welfare State Protection


While welfare states certainly encompass more than income security measures,
this analysis focused on this element since it is identified in EMCONET framework
(Benach et al., 2007) as fundamental in shaping health outcomes. Welfare state
protection was thus measured and calibrated using data from the ILO income

169
security index (ILO, 2004). This index is calculated using a range of input, process
and outcome indicators.

In terms of input indicators, a country is given a positive value if it has


ratified the following ILO conventions: No 102 on Social Security (Minimum
Standards), No 26 on Minimum Wage-Fixing Machinery, No. 131 on Minimum
Wage Fixing, No 9 on Protection of Wages. Two dummy variables are also included,
with positive values given for the existence of a minimum wage law and for the
existence of laws promoting or legitimizing collective bargaining.

The selected process indicators are social security expenditure as a share of


GDP, and two dummies where a positive value is given for the existence of an
unemployment benefits scheme and for the existence of a state pension.

The outcome indicators are the national poverty rate, GDP per capita and the
Gini coefficient, a measure of foreign indebtedness (“to reflect a country’s
vulnerability to a sudden loss of what-ever level of national income security it has
achieved” p. 135), life expectancy at birth, the wage share in total value added
(“representing the extent of a relatively secure form of income earning” p. 135), the
old-age income security index from the SES security Database (“reflecting the
security of non-work income for mainly retired workers”) and finally, the ratio of
average female to male income (“as a proxy for wage differentials, gender
discrimination and disadvantage of female workers” p. 136).

In calculating the index, indicators are normalized and then added together
with outcome indicators given twice the weight of the other two. Based on the
construction of this index countries are grouped into one of four clusters,
“Pacesetters”, “Conventionalists”, “Pragmatists” or the “Much-to-be-done”.
Pacesetting countries are characterized as scoring highly both on input and process
indicators as well as on outcome indicators. Conventional countries are characterized
as scoring highly on input and process indicators but low on outcome indicators.
Pragmatists are characterized as scoring low on input and process indications but
high on outcome indicators, and Much-to-be-done countries are characterized as
scoring low both on the process/input indicators and on outcome indicators.

170
Rather than using the actual index score for each country, it was these
categorizations which were used to assign membership scores in the fuzzy-set of
countries with Protective Welfare State Policies. This is for multiple reasons.

First, the categories clearly delineate important qualitative features of


countries. The ranking of countries however, does not strictly follow these clusters—
Madagascar for instance, is clustered as a Conventional country but on the index,
scores lower than other Much-To-Be-Done countries. Albania, a Much-To-Be-Done
country, scores higher on the index than many Conventional countries. Slovakia, a
Pacesetter country, scores lower on the index than other Pragmatist and Conventional
countries. Related to this is the fact that two countries can score very similarly on the
index but be qualitatively very different. As a Pacesetter, Slovakia for instance, has
an index score of 0.626 while Estonia, as a Pragmatist, has index score of 0.627.
Therefore calibrating membership scores on the basis of index scores would cloud
important qualitative differences among countries.

Moreover, such qualitative differences are important theoretically for how the
welfare state is understood to impact health. For instance, while outcome indicators
are meant to represent the effectiveness of input and process indicators, the authors
of the index note that for some countries, despite having formal policies and
institutions to promote income security, other economic realities can “make
outcomes less than satisfactory” (p. 136). Therefore, while a Pacesetter country
might score lower on the index than a Pragmatist country (with low scores on input
and process indicators), it is in the Pacesetter country, scoring relatively high on all
three types of indicators, where we would expect a more health conductive socio-
political context.

For these reasons, a four-value fuzzy-set was used whereby Pacesetter


countries were assigned a membership score of 1, Conventional countries were
assigned a membership score of .67, Pragmatist countries were assigned a
membership score of .33, and Much-To-Be-Done countries were assigned a
membership score of 0.

While it is possible that income security measures changed in countries from


2004-2009, such changes would have needed to be relatively drastic to change a
country’s broader categorization in reference to the breakpoints used above. As such

171
it seems reasonable to expect that such drastic changes would have been rare.
Moreover, in the literature on the MFA phase-out signals were not found that
countries were altering their social policies in an anticipatory way.

Final fuzzy scores for countries’ welfare state protection are displayed in table
11 below.

Table 11 Fuzzy-set membership scores in the set of Protective Welfare State Policies

ISI Fuzzy
Country Category
(2004) Score
Azerbaijan 0.424 Much to be done 0
Bangladesh 0.365 Much to be done 0
Brazil 0.586 Conventional 0.67
Bulgaria 0.658 Conventional 0.67
China 0.428 Much to be done 0
Colombia 0.335 Much to be done 0
Croatia 0.679 Conventional 0.67
Ecuador 0.464 Conventional 0.67
Egypt 0.505 Much to be done 0
Estonia 0.627 Pragmatist 0.33
Greece 0.594 Pragmatist 0.33
Hungary 0.672 Pragmatist 0.33
India 0.288 Much to be done 0
Indonesia 0.328 Much to be done 0
Italy 0.681 Pragmatist 0.33
Korea 0.666 Pragmatist 0.33
Kyrgyzstan 0.371 Much to be done 0
Latvia 0.694 Pacesetter 1
Lithuania 0.622 Pragmatist 0.33
Mauritius 0.654 Conventional 0.67
Mexico 0.555 Conventional 0.67
Morocco 0.331 Much to be done 0
Peru 0.356 Much to be done 0
Philippines 0.432 Conventional 0.67
Poland 0.692 Pacesetter 1
Portugal 0.738 Pacesetter 1
Romania 0.514 Conventional 0.67
Slovakia 0.626 Pacesetter 1
South
0.487 0.67
Africa Conventional
Sri Lanka 0.502 Conventional 0.67
Thailand 0.408 Much to be done 0
Turkey 0.567 Conventional 0.67

172
Changes in T&C Employment Following the MFA Phase-out
After the MFA phase-out, countries reliant on the T&C sector could either
have 1) experienced no change in their T&C related employment; 2) experienced
growth in T&C related employment; or 3) experienced loss in T&C related
employment. Because employment growth can impact health through different
pathways than employment loss, two membership sets were constructed for each
type of change. Constructing two membership sets was necessary since a single
membership set was unable to account for countries’ qualitatively different
experiences. For instance, a single membership set of Employment Growth would
theoretically assign similar membership scores to two countries with very different
employment losses. This is because any country experiencing employment loss
would largely be conceptualized outside of the membership set of Employment
Growth; and likely designated a score of (or close to) 0. However, an employment
loss of 5% in a T&C reliant country will have very different health implications than
a loss of 50%.

A direct calibration method was used to assign fuzzy membership scores for
changes in T&C employment. While it is unclear precisely how much of a change in
employment might be important for health, qualitative thresholds were chosen with a
consideration of the data, that is the variation of change across included counties.
They were also chosen at points where qualitative differences between countries
experiences seemed meaningful and with a consideration that changes would need to
be somewhat significant to influence health at the population level.

For the membership sets of Employment Growth and Employment Loss,


scores were calibrated based on percent changes in employment in the T&C sector
between 2004 and 2008, or the closest years for which data was available. Again, in
the direct calibration method three thresholds need to be specified which correspond
to the qualitative breakpoints of full membership (1), the cross-over point (.5), and
full non-membership (0). For the Employment Growth membership set, these
thresholds were conceptualized at a 15% increase, a 5% increase, and 0% increase,
respectively. For the Employment Loss membership set, these thresholds were
conceptualized at a 15% decrease, a 5% decrease, and a 0% decrease, respectively.
Employment figures were obtained from the United Nations Industrial Development
Organization (UNIDO) Industrial Statistics Databases (2011). However, figures were

173
not available from UNIDO for Bangladesh and Mexico and so were taken from
Lopez-Acevedo & Robertson (2012).

Final fuzzy scores for countries’ changes in employment are displayed in


table 12 below.

Table 12Fuzzy membership scores in the sets of Employment Growth and Employment Loss

Change in Fuzzy-set Fuzzy-set


Country Years Employment Employment Employment
(percentage) Growth Loss
Azerbaijan 2004-2008 -19.78 0 0.99
Bangladesh 2004-2008* 40.00 1 0
Brazil 2004-2007 11.86 0.89 0
Bulgaria 2004-2008 -17.88 0 0.98
China 2004-2008 18.03 0.98 0
Colombia 2004-2005 -4.07 0 0.36
Croatia 2004-2008 -18.06 0 0.98
Ecuador 2004-2008 3.04 0.24 0.01
Egypt 2004-2006 -1.53 0.02 0.11
Estonia 2004-2008 -34.05 0 1
Greece 2004-2008 -3.49 0.01 0.29
Hungary 2004-2008 -41.56 0 1
India 2004-2008 21.31 0.99 0
Indonesia 2004-2008 8.62 0.75 0
Italy 2004-2008 -13.29 0 0.92
Kyrgyzstan 2004-2008 -27.22 0 1
Latvia 2004-2008 -32.26 0 1
Lithuania 2004-2008 -39.01 0 1
Mauritius 2004-2008 -14.82 0 0.95
Mexico 2004-2008* -35.00 0 1
Morocco 2004-2008 -8.37 0 0.73
Peru 2004-2008 -16.20 0 0.97
Philippines 2003-2006 -6.75 0 0.63
Poland 2004-2008 -8.42 0 0.74
Portugal 2004-2008 -12.05 0 0.89
Korea 2004-2006 -12.17 0 0.9
Romania 2004-2008 -39.96 0 1
Slovakia 2004-2008 -21.24 0 0.99
South Africa 2004-2008 -32.16 0 1
Sri Lanka 2001-2008 80.51 1 0
Thailand 2002-2006 8.06 0.71 0
Turkey 2004-2006 -0.29 0.04 0.06

174
5.4.2 Stage 2: Examining Necessity and Sufficiency
In the second stage of fsQCA, an examination of necessity and sufficiency is
undertaken.

As noted in the beginning of this chapter, for a condition to be necessary it


must be present for an outcome to occur. A condition, or combination of conditions,
is said to be sufficient if the outcome occurs whenever the causal condition(s) are
present. Because it is rare for conditions or combinations of conditions to conform
exactly to a precise subset relation of necessity or sufficiency, fsQCA offers two
measurements of how well cases fit a subset relation: consistency and coverage
(Ragin 2006, 2008)

Consistency measures the degree to which a subset relation of necessity or


sufficiency is met. For an analysis of necessity, consistency scores indicate the
degree to which cases with an outcome Y, agree in displaying condition X. A fuzzy
subset relation will exist when membership scores in an outcome Y are consistently
lower than those for the condition X. For an analysis of sufficiency, consistency
scores indicate the degree to which cases with a specific configuration agree in
displaying the outcome under investigation (Ragin, 2008). A fuzzy subset relation
will exist when membership scores in a configuration are consistently lower than
those for the outcome. Consistency scores range from 0 to 1, where 0 indicates no
consistency and 1 indicates perfect consistency. For this analysis consistency scores
were calculated by the fsQCA software (version 2.5).

Coverage scores, by contrast, provide a measure of empirical relevance after a


subset relation of either necessity or sufficiency has been established (Ragin 2006,
2008). In other words coverage scores measure how well a consistent subset “covers”
the superset. In the case of necessary causes, coverage indicates the degree to which
the cause is relevant to the outcome. In the case of sufficiency, coverage indicates the
degree to which the cause explains all occurrences of the outcome. As with
consistency scores, coverage scores range from 0 to 1, where 0 indicates no coverage
and 1 indicates full coverage. For this analysis consistency scores were calculated by
the fsQCA software (version 2.5).

Analysis of Necessity
An analysis of necessity is typically undertaken prior to an analysis of
sufficiency. It is suggested that when testing conditions for their necessity, the
175
threshold for consistency should be greater than 0.9 (Schneider & Wagemann, 2012)
and that for coverage it should not be too low (<0.5). It is also important to recognize
that no cause should be taken as necessary, independent of a theory that recognizes it
as a relevant cause (Ragin, 2008).

Analysis of Sufficiency
Sufficiency is examined through the construction of a truth table which
outlines all logically possible combinations of causal conditions (configurations). As
previously mentioned, with fsQCA there are 2k possible configurations, where k
represents the number of selected causal conditions. The fsQCA undertaken here,
with five conditions, thus has 32 possible combinations.

The truth table also outlines the empirical instances of configurations, as well
as their relationship to the outcome indicators. In fuzzy-sets, cases have varying
degrees of membership scores in each configuration. A case’s membership score in a
configuration is calculated by taking the minimum membership score of all the
conditions within the configuration. In other words, if a case has a membership score
of 0.6 in causal condition A, and a membership score of 1 in causal condition B, its
membership score in the configuration AB is 0.6.

It is also important to note that for each membership score a case has in a
causal condition, it also has a correlated score in the negation of that condition. The
case’s membership in the negation of a condition is calculated by subtracting from 1,
the case’s membership score in the non-negated condition. For example, a case with
a membership score of 0.6 in causal condition A will have a membership score of 0.4
in the condition a (not A). Based on fuzzy-set logic, each case can only score greater
than 0.5 in one configuration. Combinations of conditions which have no empirical
instances are deemed ‘logical remainders’, a common trait of social science research
(Ragin, 2000).

The relationship a configuration has to an outcome can either be set-theoretic


in nature (i.e. sufficient for the outcome) or contradictory. Again, a configuration
which is set-theoretic in nature is designated as having a subset relation to the
outcome. This “is important because it signifies an explicit connection between a
combination of causal conditions and an outcome” (Ragin, 2008, p. 137).

176
The determination of whether a configuration is set-theoretic or contradictory
depends on the consistency score of the configuration and the threshold set by the
researcher as demonstrating a basis for a set-theoretic relation. As previously
described, consistency scores indicate the degree to which cases in a specific
configuration agree in displaying the outcome under investigation (Ragin, 2008).

This study deems a sufficient set-theoretic relation present when a


configuration has an observed consistency score of 0.75 or greater. Ragin (2008)
argues that this is the minimum basis on which a configuration can be claimed to
have a set-theoretic relationship with an outcome. Moreover, upon examination of
the data, this threshold was consistent with the empirical gaps between consistency
values and thus the 0.75 minimum is maintained as an appropriate choice (Schneider
& Wagemann, 2010).

Because configurations need not have perfect consistency scores to


demonstrate set-theoretic relations, some configurations might contain ‘contradictory
cases’ (i.e. cases that have a different outcome from the one forming the set-theoretic
relation of the configuration). Moreover, configurations with consistency scores
approaching neither 1 nor 0 are deemed contradictory and indicate that there are
additional factors which need to be considered when trying to understand the
outcomes displayed by the relevant cases (Glaesser & Cooper, 2011).

5.4.3 Stage 3: Logical Reduction


The third stage of fsQCA involves a minimization process through which
configurations passing the consistency threshold of sufficiency are reduced into a
more parsimonious statement about the combinations of conditions sufficient for an
outcome. This is achieved through the use of Boolean algebra (automated in the
fsQCA software). For example, if two configurations, ABC and AbC, were
determined as consistently sufficient for an outcome, we could reduce this to just AC
(since the outcome occurs whether B is present or absent).

Some analysts using fsQCA may use ‘simplifying assumptions’ to achieve


greater parsimony in their logical reduction. This is achieved by making assumptions
about configurations which have no empirical instances on the basis of substantive
knowledge or theory (Ragin, 2008). However, because this fsQCA is largely

177
exploratory in nature, no simplifying assumptions were used. This is taken to be the
most conservative approach to logical reduction.

The effectiveness of each of the final configurations (i.e. solution paths)


identified through logical reduction is determined by a coverage score. For each
solution path two types of coverage scores are calculated: a raw coverage and a
unique coverage score. Raw coverage assesses “the relative importance of different
combinations of causally relevant conditions” (Ragin, 2006, p. 305); that is the extent
to which a solution path accounts for all cases with the outcome of interest. On the
other hand, unique coverage assesses the weight of the solution path; that is the
extent to which the path uniquely covers the outcome. A unique coverage score is
necessary since solution paths may overlap. Moreover, how much of the outcome is
covered by all the solution paths is expressed as the solution coverage. Raw coverage
is calculated by finding the proportion of cases displaying the outcome which are
captured by the solution path. Unique coverage is calculated by subtracting from the
solution coverage, the cumulative raw coverage of the other solution paths. Coverage
scores can range from 0 to 1, where scores closer to 1 indicate greater coverage
(Ragin, 2006).

178
CHAPTER 6 RESULTS AND DISCUSSION
6.1 Introduction
FSQCA was applied to the dataset described in the previous chapter. Again,
the dataset includes 5 conditions: two post MFA phase-out conditions (employment
growth and employment loss), two macro-level social policy conditions (labour
market protection and welfare state protection) and one development indicator. A
fsQCA with five conditions has 32 (i.e. 25) possible combinations. Moreover,
changes in two health indicators were examined, adult female and infant mortality
rates. For each of these indicators two outcome sets were analyzed, one of ‘health
improving’ and one of ‘health worsening’.

As previously mentioned, 27 cases are used to explore changes in female


mortality rates and 29 cases are used to explore changes in infant mortality rates.
Overall, 32 unique countries are included in the analysis, 14 of which are highly
developed and 17 of which are not. With a human development index of 0.5,
Mauritius is neither in nor out of the set of ‘highly developed countries’. This also
means that Mauritius is not represented by a single configuration in the below
discussed results, though it should be noted that the fuzzy-set scores of the country,
across all membership sets, are configured into consistency calculations. Therefore,
while not represented by a single configuration, Mauritius is still accounted for in the
analysis.

The remaining sections of this chapter are organized and ordered as follows.
First, the results of the analyses of necessity will be presented. Following this will be
a consideration of the sufficiency tests. Here some general observations about how
the results are distributed across the truth table will be discussed. The overall
findings contained within the truth table will be then summarized. An assessment of
the truth table before the minimization process is an important step in understanding
the relationships between cases, conditions, combinations of conditions, and
outcomes (Glaesser & Cooper, 2011). The solution terms of the minimization
process will then be presented. Finally, the last section will look more critically at
what the results of the fsQCA can tell us about the relationship between labour
market and welfare state policies, changes in T&C employment after the MFA
phase-out, and health outcomes.

179
6.2 Analyses of Necessity
Table 13 below indicates the consistency and coverage scores for the analyses
of necessity. Overall, none of the causal conditions are found to be necessary causes.
Recall from the previous chapter that for a condition to be necessary it must be
present for an outcome to occur. The negation of one of the causal conditions,
employment growth, meets the lowest suggested thresholds for consistency and
coverage (of 0.9 and 0.5 respectively). It does so in relation to the improvement of
infant mortality rates, with consistency and coverage scores of 0.904 and 0.592,
respectively. However, there are two main reasons this condition was not taken to be
a necessary cause.

Table 13 Tests of necessity, consistency and coverage scores

Improving Adult Worsening


Causal Female Adult Female Improving Infant Worsening Infant
Condition Mortality Mortality Mortality Mortality
Consist. Cover. Consist. Cover. Consist. Cover. Consist. Cover.
Highly
Developed 0.714 0.763 0.655 0.390 0.620 0.589 0.670 0.624
Not Highly
Developed 0.429 0.690 0.603 0.541 0.604 0.651 0.559 0.590
Protective
Labour
Market
Policies 0.828 0.671 0.860 0.389 0.883 0.557 0.847 0.524
Not Protective
Labour
Market
Policies 0.245 0.759 0.271 0.468 0.246 0.621 0.284 0.703
Protective
Welfare State
Policies 0.501 0.685 0.576 0.439 0.427 0.520 0.511 0.610
Not Protective
Welfare State
Policies 0.578 0.653 0.587 0.397 0.680 0.587 0.598 0.505
Employment
Growth 0.237 0.700 0.221 0.365 0.162 0.357 0.364 0.787
Not
Employment
Growth 0.785 0.644 0.818 0.374 0.904 0.592 0.703 0.451
Employment
Loss 0.666 0.692 0.586 0.340 0.610 0.541 0.575 0.500
Not
Employment
Loss 0.365 0.613 0.469 0.439 0.437 0.512 0.472 0.542

To begin with, it was noted that this causal condition just barely meets the
bare minimum for consistency. Often, a higher threshold than 0.9 is advocated for
(Schneider & Wagemann, 2012). More importantly however, in fuzzy-set literature it

180
is stressed that no cause should be taken as necessary, independent of a theory that
recognizes it as a relevant cause (Ragin, 2008). Indeed, a necessary cause is taken to
be a rare empirical event. Therefore, the main reason the absence of T&C
employment growth (which is not the same as employment loss), is not taken as a
necessary cause is because it is unclear why it would be a necessary condition for the
improvement of infant mortality rates in countries reliant on the textile and clothing
sector.

This uncertainty relates to the make-up of countries falling under the


membership set of ‘not employment growth’. There are two types of countries in this
membership set. The first type includes countries which experienced no change in
T&C employment after the MFA phase-out. As will be noted later in this chapter,
only 5 countries are characterized as such. The second type includes countries which
experienced employment loss. However, employment loss has a consistency score of
0.609, much below the 0.9 minimum threshold. Moreover, the coverage rate for
employment loss is 0.541. This suggests that the coverage score of ‘not employment
growth’ (0.592) is largely made of cases which experienced employment loss. The
similar coverage scores of ‘not employment growth’ and ‘employment loss’ also
suggests that the coverage score for the condition of ‘no change in employment’
would on its own be too low to be deemed a necessary cause. Since for these reasons
neither employment loss nor ‘no change in employment’ would on its own be
characterized as a necessary condition, an absence of employment growth (which is
made up of these two conditions) was not taken as one either.

6.3 Analyses of Sufficiency


Results of the sufficiency analyses are organized across the summary table in
table 14 below and the truth table displayed in table 15. The summary table displays
information for countries which experienced either T&C employment growth or loss
after the MFA phase-out. The five countries which experienced neither employment
growth nor loss are not included in this table: Ecuador, Turkey, Colombia, Egypt and
Greece. The configurations represented by these cases are however, are displayed
across the truth table and incorporated into the Boolean minimization process.

181
Table 14 Tests of Sufficiency Summary Table

Protection Change in Employment after the MFA phase-out in Not Highly Developed Countries
Type
Employment Growth Employment Loss

Countries Adult Female Infant Countries Adult Infant


Mortality Mortality Female Mortality
Mortality
Protective Brazil, Sri Health Health Philippines,
Contradictory Contradictory
LM and WS Lanka* Improving Worsening South Africa
Row Row
Policies
Azerbaijan*,
Protective
Bangladesh, Health Health Kyrgyz Contradictory Health
LM Policies
Indonesia* Worsening Worsening Republic, Row Improving
Only
Morocco*, Peru
Protective X NA NA X NA NA
WS Policies
Only
Neither India, Contradictory Health X NA NA
Protective China, Row Worsening
LM nor WS Thailand
policies

Protection Change in Employment after the MFA phase-out in Highly Developed Countries
Type
Employment Growth Employment Loss

Countries Adult Infant Countries Adult Infant


Female Mortality Female Mortality
Mortality Mortality
Bulgaria,
Croatia , Latvia,
Protective
Poland, Health Contradictory
LM and WS X NA NA
Portugal, Worsening Row
Policies
Romania** and
Slovak Republic
Protective X NA NA Estonia**, Health Contradictory
LM Policies Hungary**, Worsening Row
Only Italy and
Lithuania
Protective X NA NA Mexico Health Contradictory
WS Policies Worsening Row
Only
Neither X NA NA Korea Health Health
Protective Worsening Improving
LM nor WS
policies
LM=Labour * Only examined in reference to infant
Market mortality rates
WS= ** Only examined in reference to adult female
Welfare mortality rates
State

The truth table displays configurations for which there are empirical instances.
Combinations of conditions which have no empirical instances are deemed ‘logical
remainders’, a common trait of social science research (Ragin, 2000). While not
displayed in the truth table, a consideration of logical remainders will inform the
discussion of the results.

182
For each of the configurations which have empirical instances, the truth table
also displays three important pieces of information. First, is the number of cases
which have at least 0.5 membership in the respective configuration. Based on fuzzy-
set logic, each case can only have at least 0.5 membership in one configuration,
therefore the second column indicates which countries are described by the
respective row. Finally, the health outcome columns ‘Health Improving and ‘Health
Worsening’ indicate for each causal combination (and each health indicator),
whether the conjunction passes the consistency threshold of 0.75. As mentioned in
the previous chapter, rows with consistency scores approaching neither 1 nor 0 are
deemed ‘contradictory configurations’ and indicate that there are additional factors
which need to be considered when trying to understand the outcomes displayed by
the relevant cases (Glaesser & Cooper, 2011). On the other hand, ‘contradictory
cases’ are cases that display a different outcome other than the one which forms the
set-theoretic relation of the configuration. Both contradictory cases and contradictory
configurations will be considered in the presentation and discussion of the results of
this analysis.

183
Table 15 FsQCA truth table results

Adult Female Mortality Infant Mortality

Health Health Health Health


N N
Configuration =
Cases improving worsening =
Cases improving worsening
consistency consistency consistency consistency
1 h*M*W* G*l 1 Brazil Brazil, Sri
0.975* 0.432 2 0.486 0.993*
Lanka
2 h*m*w*G*l 3 China, China,
India, 0.426 0.702 3 India, 0.540 0.887*
Thailand Thailand
3 h*M*w*G*l 1 Bangladesh Bangladesh
0.582 0.753* 2 0.604 0.903*
, Indonesia
4 h*M*W*g*L 2 Philippines, Philippines,
South 0.705 0.693 2 South 0.642 0.560
Africa Africa
5 h*M*w*g*L 2 Kyrgyz Azerbaijan,
Republic, Kyrgyz
Peru 0.613 0.720 4 Republic, 0.787* 0.523
Morocco,
Peru
6 h*M*W*g*l 2 Ecuador, Ecuador,
0.629 0.629 2 0.881* 0.455
Turkey Turkey
7 h*M*w*g*l 1 Colombia Colombia,
0.681 0.676 2 0.933* 0.603
Egypt
8 H*M*W*g*L 7 Bulgaria, Bulgaria,
Croatia, Croatia,
Latvia, Latvia,
Poland, Poland,
0.398 0.821* 6 0.469 0.660
Portugal, Portugal,
Romania, Slovak
Slovak Republic
Republic
9 H*m*W*g*L 1 Mexico 0.555 0.867* 1 Mexico 0.673 0.667
10 H*M*w*g*L 4 Estonia, Italy,
Hungary, Lithuania
0.416 0.816* 2 0.685 0.606
Italy,
Lithuania
11 H*m*w*g*L 1 Korea 0.578 0.894* 1 Korea 0.887* 0.492
12 H*M*w*g*l 1 Greece 0.745 0.532 1 Greece 0.954* 0.526
H = Highly Developed; M = Protective Labour Market Policies; W = Protective Welfare
State Policies; G= Employment Growth; L=Employment Loss (lower case signifies the
negation of these conditions)
*Consistency greater than 0.75

6.3.1 General Inspection of the Truth Table


Included countries can be organized into 12 out of the 32 logically possible
combinations (rows in the truth table). This implies that there are 20 logical
remainders.

Of the configurations which have empirical instances, general inspection of


the truth table points to at least three interesting findings. These relate to: 1) the
distribution of employment loss and gain across countries with different levels of
development; 2) the distribution of protective labour market and welfare state
policies across countries of different levels of development; and 3) the distribution of
employment loss and gain across countries with different levels of social protection.
184
Employment change across countries with different levels of development
Of the 32 countries included in this study, only seven experienced
employment growth after the MFA phase-out, none of which were highly developed:
Brazil, China, India, Thailand, Bangladesh, Indonesia and Sri Lanka. However, an
almost equal number of less developed countries analysed in this study experienced
employment loss: Azerbaijan, Kyrgyz Republic, Morocco, Peru, Philippines, and
South Africa. Moreover, Colombia, Ecuador, Egypt, and Turkey are all less
developed countries which saw no significant change in their T&C employment.
Therefore, while the phase-out may have favoured less developed countries in terms
of employment growth, this favouring was not shared equally among all less
developed countries.

All highly developed countries included in this analysis experienced


employment loss with the exception of Greece which saw no change in its T&C
related employment after the MFA phase-out.

Labour market and welfare state policies across countries with different levels of
development
The distribution of protective labour market and welfare state policies across
the configurations is uneven. Of all the 17 not highly developed countries, six (Brazil,
Sri Lanka, the Philippines, South Africa, Ecuador and Turkey) are characterized as
having both protective labour market and welfare state policies. Eight are
characterized as having only protective labour market policies (and not protective
welfare state policies): Azerbaijan, Bangladesh, Indonesia, Kyrgyz Republic,
Morocco, Peru, Colombia, and Egypt. Finally, China, India and Thailand are
characterized as having neither protective labour market nor welfare state policies.

Of the 14 countries which are highly developed, the majority are


characterized by both protective labour market and welfare state policies. This
includes Bulgaria, Croatia, Latvia, Poland, Portugal, Romania and the Slovak
Republic. Five highly developed countries are characterized by their protective
labour market policies (but lack of protective welfare state policies): Estonia,
Hungary, Italy, Lithuania and Greece. Mexico is the only country which is
characterized by protective welfare state policies and an absence of protective labour
market policies; and Korea is the only highly developed country characterized
neither by protective labour market nor welfare state policies.

185
The difference between highly and less highly developed countries in terms of
labour market and welfare state protection is perhaps unsurprising. However, the
differences among highly and less developed countries is noteworthy.

Employment change across countries with different levels of social protection


Of the cases analysed, it doesn’t appear that employment growth or loss
favoured countries with less or more protective labour market and welfare state
policies. Among less developed countries, those with neither labour market nor
welfare state protection policies saw employment growth (China, India, and
Thailand); however, Brazil and Sri Lanka, with the most comprehensive level of
protection, similarly saw its employment in the T&C sector grow after the MFA
phase out. On the other hand, the Philippines and South Africa, which are also both
characterized by protective labour market and welfare state policies, saw
employment loss. Among high income countries, all levels of protection (or absence
thereof) were associated with employment loss.

6.3.2 Overall Truth Table Results

T&C Employment Growth and Health Outcomes


The seven countries which experienced employment growth after the MFA
phase-out are organized across three configurations: (h*M*W*G*l), (h*m*w*G*l)
and (h*M*w*G*l). These configurations are associated with both improvements and
deteriorations in adult female mortality rates. Only the combination of conditions
represented by Brazil (h*M*W*G*l) is associated with improved female mortality
rates. Interestingly however, this configuration is also sufficient for the health
outcome set of health worsening in terms of infant mortality rates. This configuration
is represented by both Brazil and Sri Lanka in this latter case. In fact, employment
growth is associated with worsening infant mortality rates across all of the
configurations characterized by employment growth.

The configuration represented by China, India and Thailand is characterized


by neither protective labour market nor welfare state policies and in relation to
female mortality rates it is contradictory in nature. This suggests that there are
additional factors which need to be considered when trying to understand the
outcomes displayed by these cases (Glaesser & Cooper, 2011).

186
Finally, despite employment growth in the context of protective labour market
policies, the configuration represented by Bangladesh is (just barely) sufficient for
the outcome set of health worsening with a consistency score of 0.753.

T&C Employment Loss and Health Outcomes


Both highly and less developed countries experienced employment loss after
the MFA phase-out. In highly developed countries, employment loss is associated
with worsening adult female mortality rates across all configurations of which there
are empirical instances. In less developed countries, the results for adult female
mortality outcomes are contradictory for the two configurations of which there are
empirical instances.

For both highly and less developed countries, the results for infant mortality
are largely contradictory. Only two of the six configurations characterized by
employment loss demonstrate a set theoretic relation to the improvement of infant
mortality rates (H*m*w*g*L and h*M*w*g*L).

6.4 Logical Reduction


The configurations which show a set theoretic relation to a health outcome
were reduced using the previously discussed approach of logical reduction. The final
reduced configurations are displayed in Table 16 below.

187
Table 16 FsQCA Logical Reduction Results

Reduced Solution Paths


Adult Female Mortality Rates
Cases Outcome Raw Unique Solution Solution
Consistency Coverage Coverage Coverage Consistency
Health
Improving
h*M*W* G*l Brazil 0.975 0.082 0.082 0.082 0.975

Health
Worsening
h*M*w* G*l Bangladesh 0.753 0.079 0.075 0.697 0.830
H*g*L Italy, Korea, 0.841 0.621 0.618 0.697 0.830
Hungary, Estonia,
Lithuania, Slovak
Republic, Croatia,
Lativia, Poland,
Bulgaria
Infant Mortality Rates
Cases Outcome Raw Unique Solution Solution
Consistency Coverage Coverage Coverage Consistency
Health
Improving
M*w*g*l Colombia, Egypt, 0.897 0.267 0.052 0.536 0.826
Greece

h*M*g*l Ecuador, Turkey 0.904 0.251 0.040 0.536 0.826


Colombia, Egypt

h*M*w* g Kyrgyz Republic, 0.824 0.400 0.168 0.536 0.826


Colombia,
Azerbaijan,
Morocco, Egypt
H*m*w*g*L Korea 0.887 0.107 0.039 0.536 0.826
Health
Worsening
h*w*G*l Bangladesh, India, 0.842 0.270 0.089 0.303 0.856
Indonesia, China,
Thailand
h*M*G*l Indonesia, Sri 0.916 0.214 0.033 0.303 0.856
Lanka, Bangladesh,
Brazil
H = Highly Developed; M = Protective Labour Market Policies; W = Protective Welfare
State Policies; G= Employment Growth; L=Employment Loss

6.4.1 Solution Paths for Adult Female Mortality


In terms of adult female mortality rates, the Boolean minimization process
resulted in only one solution path to health improvement: h*M*W*G*l. This
solution’s raw coverage however, at 0.082, is very low indicating that there are many
cases following other pathways to health improvement

The minimization process here was straightforward, since there is only one
row with a consistency above 0.75. Although the configuration represented by

188
Greece is very close to meeting this threshold at 0.745, because this represents a
clear drop in consistency from the configuration represented by Brazil (0.975) it was
decided against including it in the minimization process. If it had been included it
would have resulted in the following configuration: H*M*w*g*l. Had this
configuration been included the coverage score of the solution path would have
increased to 0.256, still relatively low; however the consistency score would have
decreased to 0.778.

In terms of worsening adult female mortality rates, the Boolean minimization


process resulted in two solution paths: H*g*L and h*M*w*G*l. Together the two
configurations have relatively high solution coverage of 0.697 and a solution
consistency of 0.823.

6.4.2 Solution Paths for Infant Mortality Rates


In terms of infant mortality rates, the Boolean minimization process resulted

in four solution paths to health improvement: (M*w*g*l); (h*M*g*l); (h*M*w*g);

and (H*m*w*g*L). The first two of these configurations are of less interest to the

objectives of this work since they are characterized neither by employment growth

nor loss. However, their respective raw coverage (0.267 and 0.251) and consistency

scores (0.897, 0.904) illustrate their contribution to the solution’s overall coverage

and consistency scores of 0.537 and 0.826.

In terms of worsening infant mortality rates, the Boolean minimization


process resulted in two solution paths: (h*w*G*l) and (h*M*G*l). Together, these
solution paths have an overall low coverage of 0.303.

6.5 Discussion
In total, seven solution paths demonstrate a sufficient relationship between
employment changes after the MFA phase-out and changes in health outcomes. In
terms of adult female mortality, the results seem to suggest that across countries
which are not highly developed, protective labour market policies in conjunction
with protective welfare state policies are health promoting in the context of T&C
employment growth, as indicated in the first solution path of which Brazil is a

189
member. This finding is aligned with the EMOCNET framework which highlights
the health importance of protective labour market and welfare state policies (Benach
et al., 2007) however, more work is needed to understand the specific causal
processes and mechanisms behind this relationship.

The results also demonstrate that for less developed countries, employment
growth in the context of labour market protection (but not welfare state protection) is
associated with the worsening of adult female mortality rates, as indicated by the
case represented by Bangladesh. This result may signify that in the absence of
protective welfare state policies, T&C employment growth is largely detrimental to
female mortality in a way in which protective labour market policies alone are unable
to compensate for. It could be for instance, that even despite protective labour market
policies, which promote healthy employment conditions, material deprivation,
income insecurity and/or economic inequalities outweigh these benefits. Alternative
explanations for this finding would be that labour market protections themselves are
damaging to adult female mortality, or that the indicator used to operationalize
labour market protections is flawed. Theoretically, it is hard to imagine how
protective labour markets might be health damaging. Conceivably, more protective
labour market policies which for instance, provide workers with greater collective
bargaining power could perhaps result in unhealthy clashes between workers and
employers; however, this is not an explanation which is supported in the literature.
That Bangladesh is characterized in this study as having protective labour market
policies, though is often in the media for its dangerous T&C working conditions
(Harris, 2013; Khaleeli, 2013), may signify that the approach used to operationalize
labour market protection is flawed. However, ratification of fundamental ILO
conventions has been described as a reliable way of measuring labour market
protection in less developed countries (Rudra, 2007), especially given the otherwise
dearth of data on labour protections. As with the previous solution, this finding
highlights the need for more research on the causal processes behind this set-
theoretic relationship.

Overall, these first two solutions seem to highlight the health importance of
protective welfare states in the context of T&C employment growth, but they do not
tell us to what degree protective welfare states are health promoting. It is unclear for
instance, whether protective welfare state policies alone (in the absence of protective

190
labour market policies) would be health promoting, especially since this type of
configuration is a logical remainder (i.e. there are no empirical cases of this
configuration).

The remaining solution paths in regards to female mortality indicate that in


highly developed countries, employment loss is universally associated with
worsening health regardless of the presence or absence of labour market and welfare
state protections. This finding is puzzling since we might expect protective policies
to act as a buffer to the potentially negative impacts of employment loss. However,
the types of alternative employment opportunities available to workers are one factor
that may be shaping these results. This highlights the causal complexity underlying
the fsQCA results and again illustrates the need for greater work on causal processes
and mechanisms.

In terms of infant mortality, employment growth appears to be largely health


worsening. Indeed every configuration characterized by employment growth is
sufficient for the worsening of infant mortality rates. This includes the configuration,
represented by Brazil and Sri Lanka, which is characterized by both protective labour
market and welfare state policies and associated with improving female mortality
rates. That infant mortality rates worsen in the context of employment growth, even
when adult female mortality rates improve, is at first glance puzzling. We might
think for instance, that mothers with access to income security and decent working
conditions are better able to provide health promoting resources for an infant.
However, there are a variety of reasons that might explain the worsening of infant
health in this context, such as the availability of the mother to care for the infant if
they continue to work after giving birth. That Thailand and Indonesia are
contradictory cases, in that they both saw infant mortality rates improve despite
employment growth, means that comparative case studies may provide greater
insight into these issues.

The impacts of employment loss on infant mortality are less clear. Three of
the four configurations characterized by highly developed countries and employment
loss present contradictory results. Only the configuration represented by Korea is
characterized by a set-theoretic relationship whereby employment loss, in the context
of neither protective labour market nor welfare state policies, is sufficient for the
improvement of infant mortality rates.

191
In regards to countries which are not highly developed, two configurations are
characterized by employment loss: h*M*W*g*L and h*M*w*g*L. The first of these
is contradictory in nature. The second is set-theoretic in nature with improving infant
mortality rates.

Interestingly, all of the configurations characterized by employment growth


and a set-theoretic relation to infant mortality indicate a health worsening
relationship and all of the configurations characterized by employment loss and a set-
theoretic relation to infant mortality indicate a health improving relationship.
However, that there are so many contradictory rows in terms of infant mortality rates
and employment loss suggests that the chosen causal conditions are not by
themselves adequate in explaining this health outcome. This should not be taken to
mean that labour market and welfare state policies are not important determinants of
infant mortality in the context of employment loss, but that infant mortality may be
best explained when these policies are considered in conjunction with other causally
important conditions.

It is also important to consider how far the solution paths go towards


explaining the health outcomes of all the included cases. The solution paths found in
relation to the worsening of adult female mortality rates cover the majority of
countries experiencing this outcome with a solution coverage score 0.697. However,
only one solution path was found to be set-theoretic in nature with the improvement
of adult female mortality rates. The raw coverage for this path is extremely low at
0.082, and indicates that there are many cases following other pathways to the
outcome. More work is needed to explain these cases. Such work will also aid in
resolving many of the contradictory cases and rows found in relation to adult female
mortality outcomes.

In regards to infant mortality rates, the solution coverage scores are also less
than ideal at 0.537 and 0.303 for the health improving and health worsening outcome
sets respectively. More work understanding the causal conditions behind infant
mortality outcomes after the MFA phase-out would also go a long way in resolving
many of the corresponding contradictory results.

192
6.6 Summary
The solutions achieved with this fsQCA analysis demonstrate clear cross-case
regularities and are a first step in elucidating the impacts of the MFA phase-out on
health. They suggest that changes in employment in the T&C sector are sometimes
associated with health outcomes in unique ways depending on countries’ level of
labour market and welfare state protection. Interestingly, increased employment
seems to be detrimental to infant health, even while beneficial to adult health.

Overall however, the results of the truth table and the solutions of the
minimization process are limited in what they can tell us about the importance of
protective labour market and welfare state policies in the context of T&C
employment changes after the MFA phase-out. Low coverage scores across the
solution paths means that many of the countries’ health outcomes remain
unexplained. Moreover, the results achieved in this analysis are compatible with
different causal processes and thus present competing explanations for the different
health outcomes investigated.

The findings however, are a useful basis from which to explore these issues in
greater depth. One of the benefits of fsQCA is that it provides a systematic approach
for choosing cases for in-depth case-study work (Schneider & Rohlfing, 2013). On
the basis of these fsQCA results, such work can potentially provide a causal account
of the resulting cross-case patterns by identifying specific explanations or narrowing
the range of feasible theoretical accounts.

193
INTRODUCTION TO PART 3
The next three chapters comprise the final half of the second research
component of this thesis: the case study of the 2005 phase-out of the Multi-Fibre
Arrangement (MFA). In the first part of this thesis, a literature review was
undertaken to better understand how researchers theorize the pathways and
mechanisms mediating the trade liberalization and health relationship? (RQ1) An
answer was sought for this question as a first step towards meeting the first research
objective of this thesis: to identify how trade liberalization and social policy
interact to influence health and its social determinants. Among other
considerations related to the trade liberalization and health relationship, this review
found that liberalizing policies have been particularly underexplored in the context of
labour markets. Social policies, however, were very much emphasized as a mediating
pathway to health in this context.

In the second part of this thesis, the MFA phase-out was introduced as a
useful case for further exploring this research objective. Furthermore, towards this
end, a second research objective was introduced: to investigate and analyse how
the phase-out of the Multi-Fibre Arrangement impacted health in countries
reliant on the textile and clothing sector for employment. Two research questions
were posed towards meeting this objective. The first asks: How did health outcomes
change after the phase-out of the Multi-Fibre Arrangement in countries reliant on the
textile and clothing sector? (RQ2) A fuzzy-set qualitative comparative analysis
(fsQCA) was undertaken to answer this question. This analysis finds that that T&C
employment changes after the MFA phase-out are sometimes associated with
changing adult female and infant mortality rates, depending on countries’ level of
labour market and welfare state protections.

A main limitation of the fsQCA however, is that it identifies associations and


not causality. For this reason, a second research question was posed. It asks: What
are the potential causal mechanisms responsible for these changes? (RQ3) This is the
question that the following three chapters will deal with. Specifically, Chapter 7 will
briefly introduce process tracing as a methodology which can help provide an
account of the cross-case patterns achieved in the fsQCA. It will also present a
systematic procedure for selecting cases for undertaking process tracing efforts after
a fsQCA. The process tracing results will be presented in Chapter 8 and then

194
critically discussed in Chapter 9. Chapter 10 will then conclude by drawing together
the three parts of this thesis.

195
CHAPTER 7 METHODS: AN EMPIRICAL WITHIN-CASE ANALYSIS
7.1 Introduction
In Chapter 6 it is found that changes in countries’ textile and clothing (T&C)
employment are sometimes associated with changing adult female and infant
mortality rates after the phase-out of the Multi-Fibre Arrangement (MFA), depending
on countries’ level of labour market and welfare state protections. These results
however, do not tell us much about the actual causal processes and mechanisms
which tie certain combinations of causal conditions to particular outcomes. Moreover,
the results of the fsQCA analysis raised questions about the appropriateness of the
indicator used to measure countries’ labour market protections and also presented
some puzzling results, such as the finding that adult female mortality rates worsened
in all highly developed countries experiencing T&C employment loss after the MFA
phase-out, regardless of the absence or presence of protective labour market and
welfare state policies. This chapter will briefly introduce process tracing as a
methodology which can help provide an account of these cross-case patterns and
concerns.

The chapter will begin with an overview of process tracing and its relevance
to work which seeks to investigate the relationship between trade liberalization and
health as well as cross-case analyses like fsQCA. Here the specific value of theory-
building process tracing will be emphasized. The chapter will then move to formally
define process tracing and its major component, causal mechanisms. Finally, a
systematic approach will be identified for carrying out process tracing which is
aligned with the objectives of this work. In reference to this approach, twelve
countries are identified for further in-depth case study work.

7.2 Process Tracing Overview


Process tracing has three broad purposes: to test whether a theorized causal
mechanism exists in reality (theory-testing), to build a causal mechanism where
theory is less developed (theory-building) and to explain a particular outcome in a
specific case (explaining outcome) (Beach & Pedersen, 2012). The ambition of
testing whether an already theorized causal mechanism exists in reality is to move us
away from mere correlations and associations achieved in cross-case research like
fsQCA. Furthermore, when there is less developed theory about an association

196
between a cause and an outcome, we can use process tracing to help build one.
Finally, when a case has a particularly puzzling outcome, process tracing can be used
to craft an explanation. Whereas these first two purposes of process tracing are
considered theory-centric, the third is considered to be largely case-centric. This is
because, in this latter scenario, the analysis is focused on explaining an outcome in a
way in which is tied to a specific case and not to a more generalized theory.

This work will draw on theory-centric process tracing since it seeks to explain
the results of the fsQCA which are not case specific, but rather based at the cross-
case level. In doing so, it will continue to draw on the EMCONET framework, which
was presented in Chapter 5 as a valuable tool for thinking about the complex ways
T&C employment changes after the MFA phase-out might have impacted health.
However, because this framework still leaves open the possibility for many different
kinds of causal processes shaping the fsQCA results, this work will utilize a theory-
building process tracing approach rather than a theory-testing one. Again, theory-
building process tracing is useful when we know there is an association between a
cause and an outcome but existing theory offers little in terms of causal explanation.
Thus the purpose of theory-building process tracing is to identify a plausible causal
mechanism that can be tested empirically in subsequent research (Beach & Pedersen,
2012).

Process tracing has been identified as one of the best ways to both describe
and evaluate the policy-relevant relationships between global processes and SDOH
(Labonté & Schrecker, 2005). While it has been cautioned that rarely is it possible to
evaluate the health impact of global processes with the same degree of
conclusiveness that is possible in standard epidemiological study designs (Labonté &
Schrecker, 2007), it is important to note that process tracing relies on a
fundamentally different kind of inference than these studies (Beach & Pedersen,
2012). In most epidemiological studies, researchers typically use a ‘frequentist logic
of inference’. They do this by investigating the size of effect that a given causal
condition (or treatment) has on an outcome versus a control group. What enables
researchers to make an inferential claim about a causal relationship are classical
statistical laws of probability.

In contrast to these kinds of studies, process tracing relies on the collection of


evidence with the aim of assessing whether a causal process has occurred. In

197
evaluating this evidence, an assessment is made whether our confidence has changed
concerning the presence/absence of a causal mechanism. Thus the type of inference
made in process tracing studies is Bayesian in nature. This type of research should
not be seen as a weaker research design. Indeed, because process tracing efforts are
focused on making inferences at the within-case level, Bayesian logic allows for
stronger inferences than would be possible using frequentist logic (which is based at
the cross-case level) (Beach & Pedersen, 2012). Moreover, it is worth reiterating that
the purpose of theory-building process tracing is to identify a plausible causal
mechanism that can be tested empirically in subsequent research.

7.3 Defining Process Tracing and Causal Mechanisms

Because process tracing is a tool used across different research areas, there are
a variety of definitions of what exactly constitutes an instance of process tracing.
This work adopts a definition of process tracing from Bennett and Checkel
(Forthcoming) which discusses contemporary developments on the philosophical and
practical dimensions of process tracing. In this work, process tracing is defined as:

“the analysis of evidence on processes, sequences, and conjunctures of


events within a case for the purposes of either developing or testing
hypotheses about causal mechanisms that might causally explain the case. Put
another way, process tracing examines the deductive observable implications
of hypothesized causal mechanisms within a case to test whether these might
in fact explain the case. Or, it inductively uses evidence from within a case to
develop hypotheses that might explain the case; the latter hypotheses may, in
turn, generate additional testable implications in the case or in other cases” (p.
7).

This definition of process tracing is adopted since its foundations are found in
the philosophy of social science which emphasizes the role of causal processes and
mechanisms in producing outcomes. This is in contrast for example, to neo-Humean
perspectives which ground causation in terms of associations or correlations (Beach
& Pedersen, 2012). Moreover, often definitions of process tracing reference the idea
of an ‘intervening variable’ in seeking to explain a process taking place between an
independent variable (or variables) and an outcome (dependent variable). However,
framing causal processes via reference to intervening variables runs the risk of

198
oversimplification by “suggest[ing] that an intervening variable is both fully caused
by the independent variable(s) that preceded it, and that it transmits this causal force,
without adding to it, subtracting from it, or altering it” (Bennett & Checkel,
forthcoming p. 6). The above definition, in contrast, allows for the possibility that the
processes taking place between a cause and an outcome may be influenced by forces
other than the independent variable under investigation, and that “they may have
amplifying or dampening effects of their own, or interactions or feedback effects”
(Bennett & Checkel, forthcoming p. 6).

The above definition is however, dependent on the notion of causal


mechanisms which also needs clarification. A definition of causal mechanisms is
adopted by the same authors, who offer that causal mechanisms are:

“ultimately unobservable physical, social, or psychological processes through


which agents with causal capacities operate, but only in specific contexts or
conditions, to transfer energy, information, or matter to other entities” (Bennett &
Checkel, forthcoming p. 15).

The value in this definition is two-fold. To begin with, it recognizes that


causal mechanisms are ultimately unobservable, making clear that causality cannot
be observed, only inferred. While some definitions frame causal mechanisms as
observable events, to suggest otherwise is especially important in reference to this
work. This is because the outcomes under investigation relate to changes in health
which are ultimately unobservable if we consider the bio-medical pathways
producing these events. Second, this definition leaves open the possibility that
evidence may be found at all levels of an analysis, for instance and specifically in
relation to this work, evidence may proceed from the micro level of epidemiological
data to the macro level of changes in broader determinants of health and vice versa.

7.4 Steps in Theory-Building Process Tracing


Despite the fact that process tracing has become a widely used approach, little
work has engaged with explicitly addressing the methodological techniques and
standards of theory-building process tracing. An exception to this is the work of
Beach and Pedersen (2012) which this thesis draws on to formulate a methodical
approach to understanding the cross-case patterns exhibited in the results of the
fsQCA analysis. As will be discussed in greater detail below, work by Schneider and

199
Rohlfing (2013) will also be drawn upon in creating this approach, as these authors
offer an especially valuable technique for the selection of cases after a fsQCA.

Beach and Pedersen (2012) emphasize three main steps in a theory-building


process tracing exercise: (1) case selection, (2) the collection of empirical material,
and (3) the building of theoretical mechanisms. The remaining sections of this
chapter will examine each of these steps in greater detail.

7.4.1 Case Selection


There are two kinds of cases useful for theory-building process tracing:
typical cases and deviant cases. Typical cases are useful for identifying plausible
causal mechanisms between a causal condition (or combination of conditions) and an
outcome. Deviant cases by contrast, are useful for learning about the context in
which a mechanism does not work (Beach & Pedersen, 2012).

Work by Schneider and Rohlfing (2013) develops the case selection logic for
theory-building process tracing when it is undertaken after a fsQCA analysis of
sufficiency. One way they do so is by clarifying that there are different types of
deviant cases, each with its own analytic purpose. Another way they do so is by
clarifying which typical and deviant cases are the most useful to process trace.

For each solution path of a fsQCA, the authors note that cases can be plotted
on a XY plot. This plot illustrates cases’ membership in the solution path, alongside
their membership in the outcome. As can be seen in figure 27, there are six types of
cases which are discernible in a fuzzy-set analysis of sufficiency. Notice while there
is only one type of typical case (in area 1 of the plot), there are three types of deviant
cases (in areas 2, 3, and 6 of the plot). There are also two types of irrelevant cases
(in areas 4 and 5 of the plot).

200
Figure 27 Case-Selection after a fsQCA from Schneider and Rohlfing (2013)

Typical cases are cases which display high membership in both the solution
path and the outcome. Like typical cases, cases in area 5 of the plot are also
consistent with a statement of sufficiency since their membership in the solution path
is lower than their membership in the outcome. However, because they are neither
members of the solution path nor members of the outcome, the inferential benefits of
studying these cases are limited. Cases in area 4 of the plot are also consistent with a
statement of sufficiency, however Schneider and Rohlfing (2013) coin these cases as
‘deviant to coverage’ since qualitatively they are members of the outcome but not the
solution path. Studying this type of deviant case can provide insight on why coverage
rates for solution paths may be low, but are of less use when trying to understand the
causal mechanism behind a solution path. This is because cases in this area do not
provide evidence against the statement of sufficiency since we do not expect the
outcome to occur when cases are not members of the solution path.

There are three types of cases which are inconsistent with a statement of
sufficiency. Cases in area 6 are deemed irrelevant since they are neither members of
the solution path nor outcome. Therefore there is little to be gained from process
tracing these cases when trying to understand the causal mechanism behind a
sufficient solution path. In this scenario, there is also little to be gained from
201
examining cases in area 2 of the plot. These cases are deemed ‘deviant cases in
degree’. Like typical cases, they are members of both the solution term and the
outcome, however they are inconsistent with a pattern of sufficiency since their
membership in the outcome is lower than their membership in the solution path.
Thus, in seeking to understand a causal mechanism behind a statement of sufficiency
we are better off studying a typical case. Finally, cases in area 3 of the plot are
puzzling since they are members of the solution path but not the outcome. These
cases provide evidence against the statement of sufficiency, and thus represent a
potentially useful case to process trace when trying to understand the causal
mechanism behind a sufficient relationship. Schneider and Rohlfing (2013) indicate
that the most likely reason for cases ‘deviant in kind’ is the omission of a causal
condition of which the deviant case is not a member but the typical cases are. Once
this omitted condition is included in the analysis, the deviant case will no longer be a
member of the solution path and thus turned into an irrelevant case in area 5 of the
plot. Since this omitted condition will have a role in the causal mechanism, studying
these deviant cases can be useful when trying to understand causal processes behind
a solution path.

Because this work is focused on better understanding the causal mechanisms


behind the sufficient configurations identified in the fsQCA, the focus here is on
typical and deviant cases in kind (in areas 1 and 3 of the above XY plot respectively).

Based on this XY plot, Schneider and Rohlfing (2013) outline multiple


principles that should be followed when selecting cases for process tracing after a
fsQCA analysis. Which principles apply depend on what basis a case is being
selected for, as described below.

Typical Cases
In aiming to understand causal mechanisms behind fsQCA solutions, the
authors highlight the importance of the principle of diverse case selection. This
principle states that at least one typical case for each of the solution paths should be
chosen for analysis. This idea is defended on two accounts: first, to have a
comprehensive understanding of the examined outcomes, one should examine at
least as many typical cases as there are solution paths. Second and relatedly, a typical
case from each of the solution paths should be examined since the insights from

202
process tracing in one solution path, will not be generalizable to another solution
path.

In the solutions with more than one typical case, the ideal typical case will
be the one which is closest to displaying a membership of 1 both in the outcome and
in the solution pathway.

It is emphasized that in testing for a theorized causal mechanism researchers


should strive to utilize a comparative case method in which multiple typical cases are
compared to each other. This is simply because “insights gathered in one typical case
can be strengthened by [comparing it with] another typical case” (Schneider &
Rohlfing, 2013, p. 569). Per Schneider and Rohlfing (2013), cases should be chosen
for a typical case comparative design via reference to the principle of max-max
difference which states that cases should be compared which span the maximum
range of membership in the solution term and outcome. Here the idea is that we can
increase our confidence in a causal mechanism by showing that it is in place across
countries which span the range of membership in the condition and outcome.

Deviant cases
Again, cases which are deviant in kind are puzzling in that they are members
of the solution term but do not display the outcome of the subset relationship. As
described above, this is likely to relate to the omission of a causal condition. For this
reason, cases which are deviant to consistency are best investigated via comparative
process tracing alongside a typical case. This is because it would be hard to discern
what is anomalous about a case, without knowing what causal mechanism is at play
in a typical case (Shneider & Rohlfing 2013).

The ideal-typical deviant case should be chosen via reference to the principle
of maxi-min difference. This principle states that when comparing a deviant case with
a typical case, the difference in the cases’ memberships in the outcome should be
maximized, while the difference in the cases’ memberships in the solution path
should be minimized. Minimizing the cases’ membership in the solution path ensures
similarity between the cases in relation to the causal conditions being investigated.
The idea behind maximizing the difference in outcomes is that this will increase the
likelihood of finding something anomalous about the deviant case.

203
Final selection of cases:
The above demonstrates important considerations to be taken into account
when selecting case studies after a fsQCA. Ideally, all typical and deviant cases
would be studied in-depth, but this would require a prohibitively large number of
studies.

With careful consideration, 12 countries were identified for greater in-depth


analysis. These countries were chosen with two main considerations in the mind. (1)
First, it was decided to prioritize typical cases since it is these cases that move us
most towards understanding the sufficient relationships achieved with the fsQCA
analysis. (2) The distribution of cases across the different solution paths was also
used to arrive at the final selection of cases. This consideration takes advantage of
the fact that cases are characterized in different ways across different solution paths.
For instance, Korea is a typical case in both the second and sixth solution path.
Bangladesh is a typical case in both the third and seventh solution path. In short, the
idea here is that by utilizing countries’ multiple characterizations, the final selection
of cases is able to address a wider range of theory-building process tracing
considerations, while at the same time restricting the number of included countries.

In the fsQCA analysis presented in the previous chapter, a total of seven


solution paths are identified. That there are seven solutions means that adhering to
the principle of diverse case selection would require a minimum of 7 single case
studies. Of the seven solution paths, four have only one typical case: solution 1
(Brazil); solution 3 (Bangladesh); solution 5 (Korea); and solution 6 (China). In the
2nd and 7th solution paths Italy and Sri Lanka were clearly the best contender for an
ideal typical case based on the solutions XY plot. (Solution paths’ XY plots are
presented and further discussed in Chapter 8). In the case of the 4th solution path, the
best contender was less clear from eyeballing the XY plot, both Azerbaijan and the
Kyrgyz Republic seem to be ideal contenders. However, because the Kyrgyz
Republic had a higher membership in the solution term, it was chosen over
Azerbaijan to be included in the final set of countries. Therefore the first seven cases
chosen as typical cases were: Brazil, Italy, Bangladesh, the Kyrgyz Republic, Korea,
China and Sri Lanka.

Using the principle of max-max difference the Slovak Republic was chosen
for a typical case comparison in solution 2. This case was also selected with a

204
consideration of the principle of maxi-min in relation to selecting Portugal as a case
deviant in kind for this solution path (as described below). Moreover, Korea besides
being the only typical case of solution 5, is also a typical case in solution 2, rendering
another possible comparative case in this solution path. Bangladesh, in addition to
being the only typical case of solution 3 is also a typical case in the 7th solution path.
Therefore, Bangladesh was used for a typical case comparative approach in solution
path 7 alongside Sri Lanka, the ideal typical case.

In regards to deviant cases, only four solution paths have cases which are
deviant in kind: solution paths 2, 4, 6 and 7. In solution paths 4 and 7, the only
deviant cases are Peru and Indonesia, respectively. Considering the principle of
maxi-min difference, in solution 2 Portugal is the most ideal case to explore deviancy
in comparison with the solution’s ideal-typical case, Italy. In solution 6, Thailand is
the most ideal case to explore deviancy in relation to the solution’s idea-typical case,
China. Therefore, the cases best suited for exploring deviancy concerns are Peru,
Indonesia, Portugal, and Thailand.

Combining the eight typical cases identified for in-depth study (Brazil, Italy,
the Slovak Republic, Bangladesh, the Kyrgyz Republic, Korea, China and Sri Lanka)
with the following four deviant countries: Thailand, Portugal, Indonesia, and Peru;
this work is able to cover 1) a typical case for each solution path 2) comparative
typical case studies for two solutions (paths 2 and 7) and 3) deviant cases in kind for
the four solutions paths that have such cases (paths 2, 4, 6 and 7).

7.4.2 Collection of Empirical Material and Building Causal Mechanisms


In the second step of theory-building process tracing, data is collected to build
an empirical narrative of the case. In the third step, existing theory is used to create a
potential causal mechanism.

Because selection bias is particularly acute in process tracing research (Beach


and Pedersen, 2012), the gathering of evidence is approached with this vulnerability
in mind. Moreover, this work attempts to minimize this bias in part by using a
systematic process to search for evidence. To locate sources of evidence Google,
Google Scholar and cablegatesearch.net were used. This latter source is a search
engine for diplomatic cables released by Wikileaks in 2010. A preliminary search
strategy for information on the MFA phase-out returned multiple results directed at

205
this source, and for this reason it was included in the broader search strategy. A
preliminary search strategy also concluded that traditional database searches returned
a dearth of relevant material and thus this search strategy was not undertaken.

In building a narrative about the T&C sector and consequences of the MFA
phase-out in countries, topic keywords common to all Google searches included the
country name, the Multi-Fibre Arrangement, health, employment, the textile and
clothing sector, apparel, garments, and ready-made garments. Once a scaffold of this
narrative was constructed, material regarding countries’ labour market and social
policies was searched for. Here, aside from Google and Google Scholar, international
organizations with a focus on these macro-level conditions were searched such as the
International Labour Organization, the World Bank and the Asian Development
Bank.

As the narrative was constructed, the aim was to inductively work backwards
in search of a plausible causal mechanism that might help explain each solution
path’s results. In terms of deviant cases, process tracing was specifically used to
investigate what about the country is anomalous in relation to the typical cases.
Again, evidence was interpreted as causally relevant with the EMCONET framework
(Benach et al., 2007) in mind. The collection of empirical evidence ceased when the
search strategy began to reveal redundant observations and/or when the form of a
causal mechanism began to take at least preliminary form.

206
CHAPTER 8 CASE STUDIES AND THE SEARCH FOR CAUSAL
MECHANISMS
8.1 Introduction
This chapter presents the process tracing results for the twelve countries
selected for in-depth case work. Evidence was collected for each country as specified
in the previous chapter. This chapter will proceed by presenting evidence in
sequential order of the fsQCA results. Recall from Chapter 6, that the fsQCA
identified seven solution paths as sufficient for changing mortality rates. The first
three of these solution paths relate to changes in adult female mortality, the final four
relate to changes in infant mortality rates.

For each solution path, this chapter will present four pieces of information: 1)
A brief overview will outline the sufficient relationship characterizing the solution
path (i.e. the causal conditions and the associated health outcome). This overview
will also present the relevant XY plot used for case selection and specify which
country/ies will be investigated. 2) Next, the potential causal mechanism will be
summarized along with a description of the sources of evidence. 3) The evidence
found in the process tracing exercise will then be presented in greater depth and
finally, 4) a summary which restates the causal mechanism and discusses it in
relation to broader methodological considerations will be presented. The following
chapter will critically discuss these case study results.

8.2 Solution Path 1: Brazil

8.2.1 Introduction
As can be seen in the below XY plot, Brazil is the only country characterized
by the first solution path. Recall that this solution path relates increases in T&C
employment, in conjunction with protective labour market and welfare state policies,
to an improvement in female mortality rates.

207
South Africa
Peru Solution 1
1 Philippines
Thailand
Greece Brazil
0.9

Health Improvement (Adult Female Mortality)


0.8
Ecuador
0.7
Portugal
0.6
Romania

0.5 Mauritius

Mexico
0.4

0.3
Colombia
0.2 China
SlovakRep
Bulgaria
0.1 Korea
India Estonia
Hungary
KyrgyzRep0 Turkey
Bangladesh Croatia
Latvia 0 Italy 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Poland
Lithuania ~HD*PLM*PWS*EG*~EL

Figure 28 Solution Path 1 XY plot

8.2.2 Collection of Empirical Material and Causal Mechanism


The results of these efforts need to be understood in the context of very
limited sources of data. In collecting empirical evidence for this solution path, the
dearth of data on the conditions of Brazilian T&C workers was an immediate
concern. Few studies (in the English language) were found which have focused on
the social or health impacts of the T&C sector in Brazil. Exceptions include two case
studies which investigate the working conditions of T&C workers in terms of
ergonomics (Melo Junior, 2012) and life style behaviours (Sant’Ana & Kovalechen,
2012). Another exception includes work which draws attention to the working
conditions of migrant workforces in Brazil (Bastia & McGrath, 2011). However,
literature on the country’s social policies along with industrial reports, which focus
on the more economic aspects of the sector, do provide some important insights in
regards to this solution path.

A range of areas within the EMCONET framework are considered in the


context of this solution path including labour market and welfare state policies,

208
employment conditions, and working conditions. Overall however, process tracing
efforts are unable to uncover a precise mechanism linking changes in the T&C
industry to the improvement of adult female mortality rates in Brazil. Reasons for
this will be made clearer below. Nonetheless, process tracing efforts undertaken in
this solution path do bring into greater focus important considerations for future
work in this area.

8.2.3 Findings
The textile and clothing industry in Brazil is reported to be the second largest
employer in the country, after the automobile industry. Recent figures estimate the
sector employs 1.7 million workers and another 8 million indirectly (Knight, 2011).
Females are noted to account for between 70-75% of this employment (Knight, 2011;
Melo Junior, 2012) and for many, the T&C sector is their first experience in the
labour market (Melo Junior, 2012). Over 65% of Brazilian T&C workers are
reported to be employed in small and medium sized firms (between 5 and 99 workers)
and 34.5% in large factories (more than 100 workers) (Knight, 2011). Large factories
represent only 3.4% of T&C factories in the country (Knight, 2011).

Evidence suggests that a considerable number of external migrants (namely


from Bolivia) are employed in the sector (Bastia & McGrath, 2011), but precise
figures were not uncovered. For these external migrants, working conditions in the
sector are noted to be particularly poor and characterized by long working hours,
forced labour, and unsafe conditions (Barrie, 2011; Bastia & McGrath, 2011). Little
work has focused on the working conditions in T&C factories employing Brazilians.
As previously mentioned, exceptions to this include studies which have focused on
the working conditions of T&C workers in terms of ergonomics (Melo Junior, 2012)
and life style behaviours (Sant’Ana & Kovalechen, 2012). However, these studies
shed little light on the industry’s overall conditions.

A report written for the ILO’s Decent Work Agenda summarizes working
conditions in Brazil through the year 2007. This work finds that recent increases in
the country’s minimum wage (particularly in 2003) significantly improved the
purchasing power of many Brazilians. The report finds that wage laws seem to be
mostly adhered to within the country, except in rural areas where agricultural work
often goes unpaid. On the other hand, data shows that many Brazilians have
excessive working hours: 35.5 percent of Brazilian workers worked more than 44
209
hours in 2007, and 20.3 percent more than 48. However, this is a phenomenon noted
to be more common among men than among women: in 2007 only 25.2 percent of
women worked more than 44 hours and 13.7 percent worked more than 48, versus
42.2 and 25.2 percent respectively for men. In terms of labour unions, 18.1 percent of
the formal workforce is unionized in Brazil (and 17% of women).

Industry reports confirm that after 2005, employment in the T&C sector grew
(Knight, 2011). If we assume this growth took place alongside relatively decent
working conditions and in conjunction with the observation that the T&C sector is a
large first-time employer, a potential causal mechanism might take shape which
relates the improvement of adult female mortality rates after the MFA phase out to
women’s enhanced material resources. However, evidence which supports the notion
that working conditions in the sector are decent is weak. There are also two other
important considerations which question this mechanism.

The first is that growth in the Brazilian T&C sector after 2005 has been
related to increases in domestic demand (Knight, 2011). This perhaps questions the
assumed relationship between the MFA phase-out and employment change, since the
change in employment in Brazil seems to be the result of domestic conditions.
However, that increased domestic demand was met by growth in domestic T&C
producers is invariably linked to the structure of the post-MFA global T&C market.
Indeed, evidence suggests that Brazil had reason for concern in anticipation of the
MFA phase-out (ILO, 2005).

The second consideration which questions the construction of a causal


mechanism in relation to this solution path is that in 2003, the Brazilian government
introduced a conditional cash transfer programme called ‘Bolsa Familia’. This
programme provides a monthly transfer of income to poor and extremely poor
households. By 2006, the programme covered 11 million household, and has been
noted for its role in reducing poverty and inequality (Sánchez-Ancochea & Mattei,
2011). In public health literature, the programme has also been associated with
reductions in childhood mortality (Rasella, Aquino, Santos, Paes-Sousa, & Barreto,
2013).

210
8.2.4 Summary
With these complications in mind, it is difficult to construct a causal
mechanism from the MFA phase-out to the improvement of adult female mortality
rates in Brazil. Greater attention to the conditions of the Brazilian T&C sector is
much needed. This is especially true given the industry is noted to be the second
largest employer in the country, with a great many more indirectly dependent on the
sector.

8.3 Solution Path 2: Italy, the Slovak Republic, Korea and Portugal

8.3.1 Introduction
The second solution path concerns high income countries. It relates decreases
in T&C employment after the MFA phase-out to the worsening of adult female
mortality rates, regardless of the presence or absence of protective labour market and
welfare state policies. As mentioned in Chapter 6, this finding appears puzzling since
we might expect protective policies to act as a buffer to the potentially negative
impacts of employment loss.

As can be seen in the below XY plot, a range of typical cases are scattered
across the second solution path. Also present are cases which are deviant in kind.
Such a configuration of cases allows us to explore this solution path through two
kinds of comparative case designs: one which looks comparatively at typical cases
and one which compares these typical cases with a case deviant in kind.

211
Solution 2
Bangladesh CroatiaLithuania
1 Poland Italy
Turkey KyrgyzRep
Latvia Hungary
India Estonia Korea
0.9 Bulgaria
SlovakRep

Health Worsening (Adult Female Mortality Rate)


0.8 China
Colombia
0.7

0.6
Mexico

0.5 Mauritius
Romania
0.4
Portugal
0.3
Ecuador
0.2

0.1
Brazil Greece
Philippines
Thailand Peru
0 South Africa
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
HD*~EG*EL

Figure 29 Solution Path 2 XY plot

The most ideal typical case (the one which is the closest to displaying a
membership of 1 both in the outcome and solution path) is Italy. For reasons
discussed in the previous chapter, the Slovak Republic and Korea were chosen as the
best candidates for a comparative investigation with Italy. In terms of cases which
are deviant in kind, Portugal represents the best candidate for a comparative design
with these typical cases.

8.3.2 Collection of Empirical Material and Causal Mechanism


The four countries investigated in this solution path are represented by three
different truth table rows in the fsQCA analysis. While each of these countries are
high-income and characterized by employment loss after the MFA phase-out, Italy is
additionally characterized by protective labour market policies, the Slovak Republic
and Portugal by both protective labour market and welfare state policies, and Korea
by neither protective labour market nor welfare state policies.

212
The theory building process for this solution term began with the collection of
empirical material for each of the typical cases. In terms of Portugal, process tracing
was specifically used to investigate what about the country stands in contrast to the
three typical cases. Portugal is explored last since it would be hard to discern what is
anomalous about a case, without knowing what causal mechanism is at play in a
typical case (Schneider & Rohlfing, 2013).

The results of these efforts need to be understood in the context of very


limited sources of data. In collecting empirical evidence for this solution path, the
dearth of data on the conditions of workers was an immediate concern across all four
countries. Perhaps unsurprisingly, literature surrounding the T&C sector in these
countries is far more focused on matters related to economic growth and firm
efficiency than social conditions. Moreover, the type and quality of data available for
each country differed. For example, while it was possible to find precise figures on
the share of female employment in the sector for Italy, the Slovak Republic and
Portugal, precise data was unable to be uncovered for the Republic of Korea.

The areas within the EMCONET framework which were most considered
within this solution path relate to various aspects of 1) employment and working
conditions, such as the size of T&C firms, the types of labour contracts through
which workers were employed, and the types of labour overall which characterize the
sector, e.g. formal versus informal, paid versus unpaid, legal versus illegal; and 2)
workers’ access to social provisions as afforded through various channels of labour
market and welfare state protection. What is found is that often the former is a
determinant for the latter and that despite countries’ characterizations of labour
market and welfare state protections, social provisions that may be health promoting
or mediating in the face of job loss are often inaccessible to T&C workers (even
when they are present and accessible to a country’s population at large).
Empirical support which confirms one of the main contextual assumptions of
the fsQCA analysis was found for all four countries examined in relation to this
solution path. In 2001, almost 70% of the T&C workforce was female in Italy
(Rinolfi, 2013), in 2004 71% of the T&C workforce was female in Portugal
(Cristovam, 2006) and in the Slovak Republic, women accounted for about 85-86%
of total T&C employment for the years between 2000 and 2011 (Cziria, 2013).

213
Employment in the sector is also noted to be very female-dominated in the Republic
of Korea (Moon, 2003), although exact figures for recent years were not available.

8.3.3 Findings
This section will now turn to country-specific evidence which elucidates how
labour market and welfare state protections within the three typical cases were
largely inaccessible to T&C workers facing employment loss after the MFA. It will
then explore Portugal and its anomalous position within the solution path.

Italy
Much of the literature on the Italian T&C sector focuses on firm-related
characteristics, specifically those characteristics related to firm size. This is likely
because a distinguishing feature of Italian manufacturing in general is the paradigm
of “industrial districts”: regionally clustered networks of small craft industries.
Industrial districts are very much part of the communities where they are located
(McCaffrey, 2013) and are regarded among some labour market economists for their
flexibility, efficiency, and innovative capacities (Asheim, 1996).
While the literature on industrial districts and the overall size of firms in the
Italian T&C sector is overwhelmingly economically focused, it does take us some
way towards the aim of developing a causal mechanism within this solution path.
This is because the literature describes where the majority of T&C workers are
employed; some important conditions which characterize these forms of employment;
how employment conditions influence workers’ access to social provisions in the
face of job loss; and finally how the MFA phase-out may have worsened the
situation for those who remained employed in the sector.
Within industrial districts, a large majority of Italian T&C workers are
employed in small enterprises. These enterprises are typically family-run, specialize
in one or a few stages of the production process and because they are in close
proximity of one another, facilitate various subcontracting activities within the sector.
In 2000, the average Italian T&C firm employed just 10 people (Dunford, 2006). In
2001, 33% of Italian T&C firms employed five or fewer people, and 89% of T&C
firms employed 15 workers or fewer, accounting for 37% of total T&C employment.
A further 26% of Italian T&C workers in 2001 were employed in firms with fewer
than 50 employees (ISTAT, 2004 cited in Dunford, 2006).

214
In terms of labour protection, Italian small firms are less regulated than larger
firms and as such have close ties to more precarious forms of labour such as
homeworkers, paid and unpaid family members, child labour, and legal and illegal
immigrant labour (Hadjimichalis, 2006). The unregulated nature of small enterprises
is a historical institution in Italy, which is argued by some to have contributed to the
very development of industrial districts (Dunford, 2006; McCaffrey, 2013; Owen &
Cannon-Jones, 2003).
Job decline in the Italian T&C sector has been particularly prevalent in
smaller enterprises (Camuffo, Pozzana, Vinelli, & Benedetti, 2008; Dunford, 2006).
An important reason for this is the outsourcing of production processes, which are
typically undertaken in smaller firms, to countries in Eastern Europe and Asia with
cheaper labour (Hadjimichalis 2006; Cumuffo et al. 2008).
That job loss is more predominate among smaller enterprises is an important
observation for this solution path. This is because in Italy, support for those facing
unemployment is unevenly distributed across different categories of workers, with
coverage and generosity increasing with the size of the firm. The most generous
systems of income support available are the ‘mobility’ allowance and the ‘CIG
straordinaria’. However, these two schemes are only available to workers employed
in firms with more than 15 employees (Ciccarone, 2011).
With T&C job loss likely more predominate among smaller firms, and given
that much of the informal labour associated with these smaller enterprises would
have been without recourse to state provisions, it seems likely that only a small
fraction of, the largely female, Italian T&C workers who became unemployed after
the MFA phase-out would have been covered by these more generous schemes.
While there exists a more minimal unemployment benefit, to qualify for this
scheme workers must have 1) been insured for two years and 2) made at least 52
weeks’ worth of social contributions over the previous two years. Less stringent
criteria (78 days worked during the previous year, plus the same insurance
requirement) are set for the ‘reduced’ unemployment allowance (Ciccarone, 2011).
However, the length and replacement rate of these unemployment benefits are rather
low, as payments are only granted for 180 days (extended to 8 months after Jan 2008)
and at a 60% rate of replacement (Brugiavini & Peracchi, 2012).
A final consideration in relation to the potential impacts of the MFA phase-
out, is the finding that accompanying jobs losses in the Italian T&C sector, has been
215
an increase in labour market flexibility and a stagnation or decrease of wages for
those remaining employed in the sector (Dunford, 2006). Hadjimichalis (2006)
additionally reports worsening working conditions after the MFA phase-out for those
remaining employed in the sector, such as longer working hours and overtime
without compensation.

Slovak Republic
While in Italy the T&C sector began to decline in late the 1990s, at this time
in the Slovak Republic, the industry was just beginning to recover from the collapse
of communism. Accordingly, much of the literature on the T&C sector in this region
is focused on the nature of the industry in the context of post-communist labour
markets, as well as how, within this context, firms have responded to various sources
of increased competition. While it was mentioned previously that data on workers
was scarce for all countries studied within this solution path, this was especially true
in the case of the Slovak Republic.
In the Slovak case, evidence towards the construction of an overall causal
mechanism is largely based on broader labour market changes which took place after
the fall of the state controlled economy and suggestions from within the literature
which signal the following: 1) that T&C employment contracts were made more
flexible, 2) that consequently, access to unemployment protection was reduced, and 3)
that women were not only more likely to face job loss, but were also in a
significantly precarious position in terms of finding new work.
As in Italy, T&C production is extremely regionalized in the Slovak
Republic 4 . In the communist regime prior to 1989, large state-owned enterprises
characterized the sector (Smith, 2003). After the fall of communism, the sector saw
its share of manufacturing employment decrease (in the clothing sector specifically,
from 3.3% in 1980 to 3.1% in 1990) however, employment began to recover in the
mid-1990s as competition drove many western European companies to Central and
Eastern Europe for lower production costs. By 1995, 5.8% of total manufacturing
employment was accounted for by the clothing sector (ŠÚSR 1993, 1999, cited in
Smith, 2003). However, due to the underreporting of labour in unregistered smaller
workshops, this figure is likely to have been much higher (Smith, 2003).

4
Textile production facilities are centred in the areas of Levice, Liptovský Mikuláš and Ružomberok;
clothing production facilities are centred in the areas of Prešov, Trenčín, Púchov and Žilina.

216
Two main processes characterize changes in the Slovak T&C sector after the
fall of communism. First, state owned enterprises became privatized and second,
T&C firms became more fragmented. This latter process of fragmentation can be
accounted for both by the dissolution of large state-owned firms into smaller
independent firms and by the creation of new private firms by managers previously
employed in the state-owned enterprises. For example, in 1990 six independent firms
were created out of one of the large state-owned firms (Smith, 2003). While data on
the exact distribution of employment across different sized firms could not be found,
these processes suggest at least a significant downsizing of firms (Smith, Pickles,
Buček, Begg, & Roukova, 2008; Smith, 2003). Moreover, it has been suggested that
one way in which larger firms have managed the threat of competitive pressures is by
subcontracting certain production processes to smaller enterprises (Smith, Pickles,
Begg, Roukova, & Buček, 2005).
Due to the division of labour across different sized firms and to the nature of
competitive pressures (Smith, Pickles, et al., 2008), it seems likely that it was largely
these smaller firms which would have characterized employment loss after the MFA
phase-out. There are two reasons why this is an important observation in relation to
this solution path. First, as mentioned above, smaller firms are more likely to be
unregistered with the government and, as in Italy, utilize unreported (sometimes
family) labour without formal employment contracts (Smith, 2003). Therefore
workers who lost employment in these unregistered firms would have been ineligible
for the country’s unemployment insurance.
Second, running parallel to changes in the T&C sector were changes in the
country’s overall labour code and welfare state policies (Barancová, 2006; Cerami,
2010; Fisher, Gould, & Haughton, 2007; Kahhancová & Martisková, 2013; A. Smith,
Stenning, Rochovská, & Świa̧tek, 2008). A main implication of these changes is that
short-term labour contracts, prevalent in smaller T&C firms, became deregulated,
with such workers no longer able to access the protections afforded to those with
more permanent contracts (Pickles & Smith, 2010). Although T&C workers
employed on the basis of fixed-term still would have had access to unemployment
insurance, amendments to the Slovak Labour Code in 2001 created greater space for
new types of precarious employment which limited employees’ access to this type of
protection. Emerging forms of employment included those related to self-
employment and various types of work agreements which exist outside of formal
217
employment relationships (and therefore do not come under the purview of standard
labour market regulations) (Barancová & Olšovská, 2011; Kahhancová &
Martisková, 2013). Unfortunately, precise data on the type of contracts held by
Slovak T&C workers could not be found however, within the literature there are
suggestions that changes in the Labour Code indeed had negative repercussions on
smaller firms within the T&C sector (Pickles & Smith, 2010). Moreover, the T&C
sector in the Slovak Republic is largely characterized by uneven power relations
where smaller subcontracted enterprises occupy a particularly tenuous position
(Smith, 2003). This perhaps adds confidence to suggestions that changes in the
Slovak Labour Code negatively impacted workers employed in smaller T&C firms.
There are two further points to consider in the construction of a causal
mechanism when it comes to the Slovak Republic. First, even for those workers who
did have access to unemployment insurance after the MFA phase-out, benefits in the
Slovak system would have been low (Ferrarini & Sjöberg, 2010; Palme, Nelson,
Sjöberg, & Minas, 2009). To be eligible for unemployment between 2005 and 2009,
full-time employees must have made 36 months of contributions during the previous
four years of employment. Net replacement levels during this period were around
65% (Palme, Nelson, Sjöberg, & Minas, 2009): a figure which assumes further
significance when we consider the average monthly wage in the Slovak T&C sector
represented only 58% of the average wage in manufacturing industries in 2005
(Wikileaks, 2006d). Moreover, in 2006, only around one-fifth of the registered
unemployed actually received unemployment benefits (Palme, Nelson, Sjöberg, &
Minas, 2009).
Second, women in the Slovak Republic were not only the first to lose jobs,
but they also found it more difficult to find new work and experienced longer periods
of unemployment (Smith, Stenning, et al., 2008). For those not eligible for social
assistance benefits, Smith (2003) notes that both informal and illegal forms of work
offered additional income for women workers in the post-communist context.

Korea
There are two classes of literature which cover the Korean T&C sector and
shed additional insight on a causal mechanism for this solution path. The first
discusses the T&C sector in relation to gender divisions within Korean society and
the second discusses the economic potential of the industry, especially in the context

218
of the East Asian Crisis of 1997. Whereas the latter class of literature highlights
characteristics of the sector’s decline over time, the former sheds light on the type of
workers employed within the industry and their unequal position in relation to a
number of labour market considerations. As in the cases of Italy and the Slovak
Republic, evidence suggests that women workers losing their employment in the
Korean T&C industry were likely to have experienced difficulty finding alternative
employment in the formal sector. Moreover, for the type of woman employed in the
sector, access to social protection after job loss would likely have been dismal.
Despite major job losses and drastic reductions of incomes brought about by
the 1997 financial crisis, Korea remained during this time one of the top T&C
exporting nations in world. After experiencing rapid industrial development in the
T&C sector in the early 1960s, job loss in the sector began in the late 1980s (as
manufacturing industries of higher order began to take precedence) and continued
through the late 1990s as high domestic wages sent production facilities abroad in
search of lower costs (McNamara, 1999). The phase-out of the MFA presented a
further significant challenge to the industry (Truett & Truett, 2011).
As in other aspects of Korean society, employment within the T&C sector is
stratified according to gender. Whereas men have historically maintained more
managerial and skilled positions, women have held the large majority of low skilled,
low-paid positions and as such have represented the majority of the workforce
(Moon, 2003). Over time however, the type of female worker within the industry has
shifted. In the early stages of the industry, the majority of female employees were
younger than 25 and unmarried (Moon, 2003). As the Korean economy developed
and as opportunities for young women to pursue education increased, younger
females were more likely to find employment in the service sector. At the same time,
two processes made the T&C sector an attractive source of employment for married
women over the age of 25. First, a decline in fertility made it possible for older,
married women to take up employment (Berik, 2008), an expectation which was held
especially in regards to married women from low- and middle-income families
(Moon, 2003). Second, within the sector, subcontracting was becoming the dominant
form of employment (Lee, 1993). This facilitated home-based employment
opportunities for the many married women over the age of 25, who remained largely
responsible for family-related work (Lee, 1993; Moon, 2003).

219
Concrete evidence which points to the specific type of employment loss after
the MFA phase-out could not be found. However, Cooke (2010) suggests that
Korean women in general are disproportionately selected for redundancy; a point
which Berik (2008) notes is particularly true in the case of manufacturing. Moreover,
as Korean T&C exporters have moved out of T&C production and into export
management (Lee, 1993), historically job loss has been concentrated amongst the
female-dominated and lower-skilled subcontracting tasks (McNamara, 1999).

This would have two important implications for women working in the T&C
sector. First, family commitments would have made it difficult to find an alternate
job in the competitive formal employment market, rendering employment in the
informal sector more likely (Moon, 2003). This is likely to be true regardless of a
woman’s prior education and her skills from previous work (Moon, 2003).

Second, while important strides have been taken to incorporate gender


sensitive social policies into Korea’s welfare state system (J. Cho, Kwon, & Ahn,
2010), benefits for women experiencing unemployment remain largely out of reach
(Cho, Cho, Surendra, & Cho, 2013). A main reason for this is that, as many Korean
scholars have pointed out, female employees are on the whole overrepresented in
nonstandard and irregular jobs (J. Cho et al., 2010; Chun, 2006; Cooke, 2010; B.-H.
Lee, Lee, & International Labour Office, 2007); a point which is also made in direct
reference to the Korean T&C sector by the Korean Women Workers Association
(KWWA, 2008).
For example, unemployment insurance wasn’t introduced in Korea until 1995.
At this time, coverage applied to firms with more than 30 employees. As a response
to the 1997 financial crisis, this was expanded in 1998 to cover workers in all firms
with at least one employee. The eligibility criterion was also relaxed after the
financial crisis: entitlement contributions which began at 12 months of contribution
were ultimately reduced to one month and benefits increased both in generosity and
in duration (Shin, 2000). However, despite these changes, a considerable number of
low-income households remain excluded from these provisions. From 1997 to 2000,
the share of employees covered by Korea’s unemployment insurance scheme
increased rapidly from 32% to 50.5%; however until 2004 this figure remained
largely stable (Lee & Park, 2006 cited in Ku, 2007). A common explanation for such
a low coverage rate is that many workers in non-standard employment are unable or

220
unwilling to make contributions towards social insurance schemes (Cho et al., 2013;
Kim, Khang, Cho, Chun, & Muntaner, 2011; Ku, 2007). In 2007, for example, the
participation rate among non-standard workers was strikingly lower than among
permanent employees: 33% vs. 83.9% (Korean Ministry of Labour 2007, cited by
Kim, Khang, Cho, Chun, & Muntaner, 2011).

Portugal
As mentioned above, the search for evidence in the case of Portugal
proceeded with a slightly different aim than it did for the 3 typical cases of Italy, the
Slovak Republic and Korea. Across the latter cases, a causal mechanism seems to
emerge which points to a variety of conditions which both would have restricted
unemployed female T&C workers from protective social provisions, and left little
alternative for them in terms of other job opportunities. With this mechanism in
mind, the search for evidence in the case of Portugal sought to identify what about
the country is comparatively anomalous. Evidence for this case is drawn from two
main sources: first, a case study carried out by Cristovam (2006) which explores the
restructuring of the T&C sector in Portugal and second, a literature base which is
primarily concerned with broader Portuguese labour market conditions.

As in the typical cases, the T&C sector in Portugal is regionally distributed


with the majority of work carried out in the Northern Region (Norte). There is also
evidence which suggests that T&C firms are typically smaller, albeit perhaps less so
than in Italy (Cristovam, 2006). In 2002, about half of Portuguese T&C workers
were employed in enterprises with fewer than 50 employees, 22% worked in micro-
sized companies with fewer than 10 employees, and 17% worked in large companies
with more than 250 workers (Eurostat, cited by Cristovam, 2006).

As previously mentioned, about 71% of the 2004 Portuguese T&C workforce


was female. Only about 10% were under the age of 25 and perhaps most remarkably,
over 30% were above the age of 50 (Eurostat, cited in Cristovam, 2006). After the
MFA phase-out, Portuguese females losing work in the T&C sector, like their
counterparts in Italy, the Slovak Republic and Korea, would also have found it
difficult to find alternative employment opportunities (Cristovam, 2006). However,
there are suggestions within the literature that at least some of the older workers
would have been able to take early retirement and as such would have been entitled
to state pensions (Cristovam, 2006).

221
Where evidence between the three typical cases and Portugal is most starkly
less aligned, relates to the form of labour female T&C workers undertake. In the
typical cases discussed above, female T&C work is largely characterized as non-
standard (and as such, access to social provisions is restricted). In Portugal, although
it could not be established with a high degree of certainty, evidence begins to suggest
that the T&C labour force is largely characterized by a standard, full-time
employment relationship (Tavora & Rubery, 2013). Despite the fact that Portugal is
known for its high level of temporary employment, Debels (2004) notes that the
main sectors associated with these types of contracts are services, agriculture, and
mining. An important potential implication of this finding is that unlike many of the
T&C workers in Italy, the Slovak Republic and Korea, many of the Portuguese
workers losing employment after the MFA phase-out would have been entitled to the
country’s unemployment benefits.

Another way in which Portugal differs from the three typical cases is in its
particularly strict employment protection legislation. Important aspects of such
legislation are the labour laws which govern collective dismissals. According to
Portuguese law, a collective dismissal occurs when an “employer terminates, either
simultaneously or within a period of three months, the contracts of employment of at
least two employees in the case of enterprises with up to fifty employees and at least
five employees in enterprises with over fifty employees, on the grounds of permanent
closure of the enterprise, the closure of one or more departments, or the need to
reduce the workforce for structural, technological or economic reasons” (Eurofound,
2014).

While evidence could not be found on the instances of collective dismissals in


the T&C sector following the MFA phase-out. There is at least partial evidence that
such a course of action has been undertaken in the sector. In 2001, a textiles
company employing 300, mostly female, workers initiated a collective redundancy
procedure when it decided to discontinue its production in the country (Cristovam,
2001). In 2003, at least two T&C companies initiated collective redundancy
procedures (EMCC, 2003). Given the scale of employment loss and the distribution
of workers across firms of different sizes, it stands to reason that many of the
redundancies following the phase-out of the MFA would also have fallen under the
purview of these regulations.

222
Between 2005 and 2009 a company undertaking a collective dismissal would
have needed to notify a representative of the workers (e.g. a workers commission, or
trade union representative) as well as the Ministry of Employment and Social
Security, of the proposed redundancies. This regulation is intended to initiate a round
of negotiating in regards to 1) how the dismissal would be carried out, and 2) the
adoption of measures which would mediate the impact of the dismissal, for example
through re-training procedures or early retirement. Whether or not an agreement
would have been reached, employees were meant to be provided with a written
notice of dismissal at least 60 days in advance. Workers dismissed through a
collective redundancy would have been entitled to time off to find other employment
(two working days per week, without loss of pay) and financial compensation in the
form of one month's basic pay for each year of service, subject to a minimum of
three months' pay (ILO, 2014a).

For example, in one of the cases of collective redundancy mentioned above,


an agreement was reached between the employer and its employees. This agreement
stipulated that all workers would receive a 3% salary increase before the closing of
the firm to allow for higher levels of compensation after dismissals, as well as a
longer duration of compensation (EMCC, 2003). However, it must be noted that
there is also evidence of companies going bankrupt, who then close operations
without giving prior notice to their workers, and in some cases denying them
compensation for hours already worked (EMCC, 2003). In cases of the latter kind, it
is not clear what happens to redundant T&C employees since the regulation
surrounding collective dismissals stipulates that an employee is still entitled to their
severance pay, even in the event that an employer fails to comply with the necessary
regulations (ILO, 2014a). While alternative job opportunities are noted to be low for
female Portugeuse T&C workers, this legislation at least provides workers with, in
addition to standard unemployment insurance, a minimum severance pay, and
possible re-training opportunities or early retirement.

In Italy, as with other labour laws, regulations regarding collective dismissals


apply only to firms with more than 15 employees (ILO, 2014a). For reasons
discussed above, many of unemployed female T&C workers would have had little
recourse to any related social provisions.

223
In the Slovak Republic, between 2004 and 2009, regulations stipulated that a
collective dismissal took place whenever an employer terminated the employment
relationship of at least 20 employees over a period of 90 days (Munková, 2004). This
would have had little impact on smaller enterprises identified as particularly relevant
in the Slovak case, particularly for those employing less than 20 employees.
Moreover, changes in the country’s Labour Code in the early 2000s, as discussed
above for making labour contracts more flexible, also introduced flexibility into the
regulations concerning collective dismissals. Consequently, workers placed on
agreements for work performed outside a typical contractual employment
relationship would have lost any entitlements related to collective redundancy
procedures (Barancová & Olšovská, 2011). Finally, in Korea, there is no statutory
severance pay corresponding to collective dismissals (ILO, 2014a).

That T&C work in Portugal seems to be largely characterized by a standard


employment contract, and that collective dismissals procedures are fairly strict,
encompassing firms of all sizes, suggests that female Portuguese T&C workers are
less ‘hidden’ than their counterparts in Italy, the Slovak Republic and the Republic of
Korea. This notion is perhaps further supported by the government’s engagement
with Microsoft in 2006 in a public-private partnership which aimed to provide re-
training services to over 3,000 T&C employees losing employment. However, by
2008 only 680 of the 1700 employees which completed the training were re-
employed, and largely in jobs with worse pay and security than their previous
position in the T&C industry (Euractiv, 2008).

8.3.4 Summary
The aim of collecting evidence within this solution path was to build a
narrative about the characteristics of each country’s T&C sector in way which might
begin to explain the results of the fsQCA analysis. The fsQCA solution path relates
T&C employment loss in high income countries to the worsening of adult female
mortality rates, regardless of the presence or absence of protective labour market and
welfare state policies.

Working inductively backwards from available evidence, a plausible causal


mechanism has taken shape which suggests that despite countries’ characterizations
of protective labour market and/or welfare state policies, female T&C workers in the
typical cases explored are employed under conditions which render 1) alternative
224
employment opportunities either particularly scarce or precarious and 2) access to
social provisions, which may be health promoting or mediating in the face of job loss,
largely inaccessible. In Portugal, a deviant case, evidence suggests that female T&C
workers are, by contrast, employed under conditions which grant access to social
provisions in the context of job loss, despite the shortage of alternative employment
opportunities.

At first glance this causal mechanism might seem to explain the puzzling
results of this solution path quite nicely. However, there are two main reasons to be
cautious with these results.

The first reason relates to data availability issues. The evidence presented in
this solution path originates from small literature bases which are not directly
concerned with the impact of the MFA phase-out on T&C workers. For this reason,
the preliminary nature of this causal mechanism needs to be emphasized.

Second, while evidence points to the importance of labour market and welfare
state policies, it is not immediately clear how these conditions are in turn, impacting
health. The EMCONET framework (Benach et al., 2007) highlights a range of
different pathways from these macro-level conditions to health. These pathways
relate to specific working conditions, to material deprivation, inequalities and to
considerations of health related behaviours, physio-pathological changes and
psychosocial factors. However, data on these conditions were not uncovered in this
work. Therefore the precise mechanisms influencing health remain unclear.

It is also worth noting that in investigating Portugal as a case deviant in kind,


the purpose was simply to see what distinguished it from the other countries. Finding
that workers have a more ‘standard’ employment relation with greater access to
social protection is not to say that these conditions were sufficient for the
improvement of female mortality rates in Portugal.

Despite these considerations, the results of these case studies do highlight an


important problem which is the often hidden nature of T&C workers in high income
countries. These results also highlight the need to be cautious about how we measure
and implement indicators of labour market and welfare state policies in public health
work. On the basis of this case study work, it may be questioned whether countries
were characterized correctly in relation to the fsQCA. However, since the search for

225
evidence in this solution path focused primarily on uncovering data in specific
relation to T&C workers, the findings do not necessarily reflect on countries’ overall
labour market and welfare policies.

8.4 Solution Path 3: Bangladesh

8.4.1 Introduction
The third solution path concerns countries which are not high income. It
relates increases in T&C employment to the worsening of adult female mortality
rates in the presence of protective labour market but not protective welfare state
policies. As can be seen in the below XY plot, there is only one typical case in this
solution path: Bangladesh. There are no cases which are deviant in kind. For this
reason, only Bangladesh will be explored in the context of this solution path.

Croatia
Poland
Lithuania
Solution 3
Italy KyrgyzRep Latvia
1 Bangladesh
Hungary Turkey
Estonia India
Korea
0.9 Bulgaria
SlovakRep
Health Worsening (Adult Female Mortality Rate)

0.8 Colombia China

0.7

0.6
Mexico

0.5 Mauritius

Romania
0.4
Portugal
0.3
Ecuador
0.2

0.1
Greece Brazil
SouthAfr Thailand
0
Peru
Philippines 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
~HD*PLM*~PWS*EG*~EL

Figure 30 Solution Path 3 XY plot

226
In the discussion of the fsQCA results, it was suggested that this solution term
might signify that in the absence of protective welfare state policies, T&C
employment growth may be detrimental to female mortality in a way in which
protective labour market policies alone are unable to compensate for. However, it
was also noted that it seemed problematic that Bangladesh was characterized as
having protective labour market policies since it is often spotlighted in the media for
its dangerous T&C working conditions. This may signal a flaw with the indicator
more generally or perhaps something exceptional about Bangladesh.

Case study work undertaken towards understanding the previous solution path
found that despite countries being characterized as having protective labour market
or welfare state policies, T&C workers are often excluded from important protective
provisions. Consequently, it was considered whether such countries were
characterized incorrectly or instead whether T&C workers represent an exception to
these characterizations. A similar question should be posed to the case of Bangladesh.

8.4.2 Collection of Empirical Material and Causal Mechanism


The collection of empirical material in this solution path aimed to build a
narrative about the Bangladeshi T&C sector in a way which would shed light on the
characteristics of the sector’s employees, the role of the MFA phase-out in
facilitating job growth, the employment and working conditions in the sector, and
how these conditions might have interacted with the characteristics of the country’s
labour market and welfare state policies. As with the previous solution path, the aim
here was to inductively work backwards in search of a plausible causal mechanism
that might help explain the solution’s results.

In contrast to the other cases examined thus far, much more scholarly work
has been carried out relation to the Bangladeshi T&C sector, especially in the context
of its social consequences. However, it should be noted that the results presented
here are not meant to provide an exhaustive account of this literature. Rather, the
literature search focused on the above mentioned conditions and evidence was
interpreted as causally relevant with the EMCONET framework in mind.

A range of areas within the EMCONET framework are considered in the


context of this solution path: labour market regulations, welfare state policies,
employment conditions, working conditions, and material deprivation. The perhaps

227
most prominent result to emerge out of the literature is that the T&C sector in
Bangladesh can be characterized by a range of complex and often contradictory
processes in terms of how it impacts the lives of its largely female workforce. This
relates primarily to the type of firms within which women work and the different
spheres of women’s lives which are impacted, sometimes negatively, sometimes
positively, by work in the sector. Ultimately a causal mechanism emerges which
points to poor working conditions in the context of lacking labour market and
welfare state protections.

Focusing first on the importance of the industry for Bangladesh’s economy,


the type of workers employed in the T&C sector, and its overall characteristics, the
results will then move to explore these complex and contradictory processes in
greater depth. The summary will conclude with how these issues begin to form a
plausible causal mechanism for the worsening of adult female mortality in the
country following the MFA phase-out.

8.4.3 Findings
It is hard to overstate the importance of the T&C sector for the Bangladeshi
economy in general and for women’s employment in the country specifically.
Clothing exports compromise nearly 80% of the country’s total exports (Staritz, 2010)
and in 2004 the sector was estimated to account for 9.5% of Bangladesh’s GDP
(Knowles, Reyes, & Jackson, 2006). Employment in the industry makes up 40% of
total manufacturing employment and grew from 0.2 million in 1986 to 3 million in
2010 (Staritz, 2010). Moreover it is estimated that the industry generates another 10
million jobs, indirectly (Staritz, 2010).

It is widely acknowledged that the Bangladeshi T&C industry has its roots in
the MFA quota system. Newly emerging economies in Asia are noted to have opened
production facilities in Bangladesh in the late 1970s to take advantage of additional
quotas (Kabeer & Mahmud, 2004b). Whereas there were only 130 T&C firms in
Bangladesh in 1983, by 2000 this figure had grown 3,000 and by 2009 to 5,500
(Kabeer & Mahmud, 2004b; Staritz, 2010). Because the sector was so dependent on
the quota system for its development, employment loss was predicted to follow the
MFA phase-out. However, despite these pessimistic forecasts, the Bangladeshi sector
continued to grow after 2005. Some attribute this post MFA phase-out growth to
safeguard protections initiated by the US and EU which protected the country from
228
Chinese competition (Ahmed, 2009), however such measures are noted by World
Bank scholar Staritz (2010) to only partially explain the sector’s growth after 2005.

Work in the Bangladeshi T&C sector has historically been, and continues to
be, carried out by a predominately female workforce. However, there are signs that
the share of employment held by females has been decreasing since the 1990s when
women represented 90% of the workforce (Staritz, 2010). Figures place the share of
female employment in the 2000s anywhere from 90 to below 80 percent (Kabeer &
Mahmud, 2004; Ahmed, 2009; Staritz, 2010). Females working in the sector are
typically noted to be young and unmarried, although Kabeer (2004) and Khatun et al.
(2007) cite observations that dispute such characteristics. Work from Kabeer (2004)
indicates that between 40-50% of female T&C workers are married, and many with
children. Khatun et al. (2007) notes that by 2006, 59% of female T&C workers were
married and that the average age of women workers increased from 19 in 1990 to 25
in 2006. Ninety percent of Bangladeshi female T&C workers are migrants from the
rural countryside5 and with strong ties to their rural households, they are noted to
send as much as of 2/3 of their income back to their family (Ahmed, 2009). For this
reason, the industry is also noted for its broader role in the reduction of rural poverty
(Kabeer & Mahmud, 2004a). Moreover, for the majority of women T&C workers,
employment in the sector often constitutes their first job (Kabeer & Mahmud, 2004b).
Importantly, Kabeer and Mahmud (2004a) find that growth in the T&C sector has
opened up new employment opportunities for females from rural areas where
otherwise there exists little opportunity for employment.

The T&C industry in Bangladesh produces two distinct categories of products:


woven garments and knitwear. While the former employs mostly women, the latter is
dominated by males. Bahkt and colleagues (2002 in Staritz, 2010) estimate that in the
early 2000s, women made up only 33% of the knitwear industry. This is because
production processes associated with knitting require a higher technological skillset
and women are often placed in jobs requiring less skills (Khosla, 2009). Another
reason for the lower participation of women in the knitting industry is that factories
often run an overnight shift and prior to 2006, women were not allowed to work
between 10 p.m. and 6 a.m. (Starlitz, 2010).
5
Similar to the experience of other T&C-reliant countries, the Bangladeshi T&C sector is very much
regionalized with the majority of firms in the capital city, Dhaka and Chittagong5 Other important
areas for T&C firms are Gazipur, and Narayanganj (Staritz 2010).

229
Growth in the knitwear industry characterized part of the overall job growth
in the sector after the MFA phase-out and is an important factor behind declining
shares of female employment in the T&C sector overall (Starlitz, 2010). It is also an
important source of the growing gap between men and women’s wages. This is
because as jobs become more technologically advanced, men’s wages rise faster than
women’s (Khosla, 2009). However, women are often paid lower wages than their
male counterpart even when the job is the same and when education and experience
are controlled for (Khosla, 2009).

As mentioned above, the T&C sector impacts the lives of female workers and
their families in a range of complex and sometimes contradictory ways. To begin
with, the type of firm within which women work can have inconsistent consequences
for workers.

Firms in the Bangladeshi T&C sector can be distinguished on the basis of


whether they operate inside or outside an export processing zone (EPZ). Those
operating inside an EPZ are mostly large and foreign owned. While foreign
investment played an important role in establishing the T&C industry in Bangladesh,
of the 4303 firms operating in 2006, only 83 were whole or partially foreign-owned
(IMF, 2007). In 2005, the average EPZ firm had 1150 employees (IMF, 2008 in
Staritz, 2010) and the average non-EPZ firm had 500 employees (IMF, 2007).
Employment in EPZ firms has been noted to account for about 12% of total T&C
employment (Kabeer & Mahmud, 2004a). Thus the vast majority of T&C
employment is in locally owned non-EPZ firms.

It is widely acknowledged that workers’ rights, earnings, and working


conditions are better within EPZ factories (Absar, 2003; Ahmed, 2006; Berik &
Rodgers, 2010; Hossain, 2011; Kabeer & Mahmud, 2004a, 2004b; Siddiqi, 2009;
USAID, 2007). For example, in a survey of EPZ (n=125) and non-EPZ (n=737) T&C
workers, Kabeer and Mahmud (2004a) found that EPZ workers had higher levels of
earnings than non-EPZ workers. Results of this survey also demonstrated that EPZ
workers were more likely than non-EPZ workers to have some sort of employment
contract, to have access to paid leave and maternity leave, to receive overtime pay
and to receive other benefits such as transport, accommodation, medical care and
child care facilities. In a study by Sidiqqi (2009), workers in non-EPZ firms are
found to experience more frequent and severe forms of sexual harassment than those

230
in firms within an EPZ. Moreover, evidence aggregated by Absar (2003) shows that
incidences of illnesses vary along the lines of whether a worker is employed in an
EPZ or non-EPZ firm, with greater incidences of illnesses among non-EPZ workers.

Another dimension of contradiction in the lives of Bangladeshi female T&C


workers relates to how the sector has improved their social and economic freedom
yet simultaneously exposed them to physically demanding and dangerous working
conditions.

Hossain (2011) provides a recent overview of literature which supports this


point. Highlighted in this work are a range of qualitative studies which demonstrate
that female T&C workers are motivated to work in the industry not only out of
economic need, but also because the industry affords them “a meaningful expansion
of their agency” (Hossain, 2011, p. 29). Kibra (1998) for example, finds that female
T&C workers perceive their employment as providing a means for investment in
their family (e.g. by enabling the education of siblings or children) and that work in
sector allows them to relieve their family of the financial burden of supporting a
daughter, avoid unwanted marriages, and increase their personal autonomy. However,
as Hossain (2011) points out, how much control female workers retain over their
income varies with respect to their household situation. In some cases women are
required to relinquish their wages to their husbands, while in others they retain more
control. This latter point underscores a wider message found in the literature which is
that Bangladesh remains characterized by patriarchal structures and as such there are
limits to the levels of empowerment that women can achieve through improved
economic capabilities (Hossain, 2011; Sidiqqi, 2009) However, research also shows
that female employment in the T&C sector has created room for negotiation and
manoeuvring within these structures (Hossain, 2011; Sidiqqi, 2009).

With regard to the level of wages in the T&C sector, Bangladesh has had a
minimum wage since 1994. However, at this time, no mechanisms were put in place
to tie it to inflation or other macroeconomic changes (Ahmed & Peerlings, 2009). In
response to labour unrest in 2006, the government increased the monthly minimum
wage for the first time from Tk930 to Tk1662 ($24). This is noted to have fallen
short of living wage estimates (Staritz, 2010) and the international poverty line of $1
a day (Berik & Rodgers, 2010). Wages have also been noted to be particularly low in
the context of rising food prices (Staritz, 2010), and Absar (2001) notes that some

231
workers’ pay fall below this regulated threshold. However, work from Kabeer and
Mahmud (2004a) indicates that wages in the majority of the sector were almost twice
as high as the monthly per capita poverty line set by the Government of Bangladesh
in 2000 (725Tk in urban areas and 635Tk in rural areas). On this basis the authors
suggest that female T&C workers were able to support at least one other adult or two
children with their T&C wage.

Overall, wages in the T&C sector are on average higher and more regularly
paid than wages in alternative work (Kabeer & Mahmud, 2004a). Therefore, while
survival on a T&C wages is no doubt difficult, women’s economic capabilities on
average have nevertheless been enhanced (Khosla, 2009).

Standing in contrast to improved autonomy and economic capabilities


however, are the physically demanding and dangerous working conditions which
characterize production in the sector. Perhaps the greatest indicator of the physical
demand of T&C work is the finding that few workers can last more than a few years
in the sector. High turnover is partially explained by women leaving their
employment to marry or have children, but evidence also points to women leaving
the sector because of the toll work takes on their health (Kabeer & Mahmud, 2004a;
Hossain, 2011). Islam and Zahid (2012) even note the fainting and death of
employees on the T&C work floor due to exhaustion and over work.

Broadly speaking, working conditions in the T&C sector are largely


characterized by long hours, unsafe factory conditions, sexual harassment, physical
and verbal abuse, a lack of security and employment benefits; denials of freedoms to
associate and bargain collectively; and often underpaid or delayed wages (Berik &
Rodgers, 2010; Hobson, 2013; Hossain, 2011; Islam & Zahid, 2012; Kabeer &
Mahmud, 2004a; Siddiqi, 2009). In total, while Bangladesh has ratified 7 of the 8
fundamental ILO conventions, all are poorly enforced (Berik & Rodgers, 2010).

A workplace safety concern that has been particularly visible of late is the
lack of fire safety equipment and structural integrity in buildings where T&C work is
undertaken. Hobson (2013), citing evidence from the International Labor Rights
Forum, highlights that at least 1800 T&C workers have died in factory fires and
building collapses since 2005. Hossain (2011) additionally points to an estimate from

232
a Bangladeshi newspaper that some 60% of T&C firms lacked fire safety equipment
in 2010.

A World Bank study (Lopez-Acevedo & Robertson, 2012) which examined


the T&C sector in Bangladesh in the period after the MFA phase-out, found that
while working conditions improved for men, for women they worsened. In this study,
working conditions were measured by dummy variables for holding a regular paid
job and working for 40 or less hours per week. The authors speculate that “female
workers increasingly face a trade-off in which they obtain jobs and higher wages at
the expense of a nonregular paid (hence less stable) position” (p. 80).

Other evidence about the impacts of the MFA phase-out in Bangladesh also
suggests that working conditions would have worsened for women after 2005. To
begin with, it is noted that while production increased in the sector, it moved away
from foreign owned firms in EPZs (Staritz, 2010). This means a greater number of
women would have been exposed to the unsafe working conditions characterizing the
majority of the sector in non-EPZ firms. In addition, the phase-out has been noted to
increase pressures on firms to lower prices (Staritz, 2010). Work by Kabeer and
Mahmud (2004b) suggests that firms which are under greater price pressures have
worse working conditions when they do not deal directly with buyers (a
characterization which fits a large majority of Bangladeshi T&C firms). For this
reason, it seems likely than that any increase in price pressure would also place many
women T&C workers in a more precarious working situation.

Complicating matters further is the recognition that buyers are simultaneously


putting greater pressure on producers to improve working conditions however, again,
this pressure is felt greatest among firms which deal directly with buyers such as
those in EPZs (Kabeer & Mahmud, 2004a).

8.4.4 Summary
The aim of collecting evidence within this solution path was to build a
narrative about the characteristics of Bangladesh’s T&C sector in way which might
begin to explain the worsening of adult female mortality rates after the MFA phase-
out. Again, this solution path relates T&C employment growth in not high-income
countries to the worsening of adult female mortality rates in the presence of
protective labour market but not welfare state policies. Prior to collecting evidence, it

233
was questioned whether this solution suggests that T&C employment growth is
detrimental to female mortality in a way in which protective labour market policies
alone are unable to compensate for. It was also noted that it seemed problematic that
Bangladesh was characterized in the fsQCA analysis as having protective labour
market policies since it is often spotlighted in the media for its dangerous T&C
working conditions.

In pulling together the various findings examined in this solution path, what
emerges is a complex set of conditions which can impact female Bangladeshi T&C
workers in a range of contradictory ways. If we examine these conditions in relation
to the EMCONET framework, we find that while female T&C workers are on
average exposed to a range of risky and dangerous working conditions, work in the
sector can also mean improvements in material circumstances and in the perhaps
more psychosocial-related senses of greater self-reliance and autonomy. However,
the extent to which these conditions apply can vary according to the type of firm
within which women work and their household-specific characteristics.

This finding seems at least to discredit the idea that T&C employment growth
in Bangladesh is detrimental to female mortality in a way in which protective labour
market policies alone are unable to compensate for. Evidence that would have
supported this idea might have shown for example, at least a minimum standard of
decent working conditions but perhaps wages that were below subsistence levels.
Instead, working conditions were found to be extremely poor and livelihoods
enhanced by the income earning opportunities the sector generates, especially in the
context of otherwise scarce opportunities.

Evidence suggests that after the MFA phase-out, T&C job growth for females
largely took place in non-EPZ firms. These firms are associated with both lower
wages and worse working conditions when compared to EPZ firms. Evidence also
suggests that working conditions were likely to deteriorate further in these non-EPZ
firms after the MFA phase-out due to increasing price pressures. On this basis, a
plausible causal mechanism seems to emerge which relates the worsening of adult
female mortality rates in Bangladesh to females’ greater exposure to dangerous
working conditions. However, while working conditions are no doubt an important
aspect of how the Bangladeshi T&C sector impacts female workers’ health, these
conditions are inevitably linked to an absence of protections that provide an effective

234
means for workers to address workplace grievances. For this reason, the causal
mechanism which takes form within this solution path must direct attention not just
to poor working conditions, but also to an absence of labour market and welfare state
protections.

The fsQCA analysis characterized Bangladesh as having protective labour


market but not welfare state policies. This latter characterization is supported by the
evidence found within the solution path, but the former is not. In the previous
solution path it was considered how the results of case study work might reflect back
on the fsQCA. It was noted that since evidence is considered in specific relation to
T&C workers, generalizing back to countries’ overall labour market and welfare
state policies is difficult. This is true also in the case of this solution path. However,
as in the previous solution path, it clear that of central importance to future analyses
of the health impact of liberalization processes is the accessibility of labour market
and welfare state protections specifically for the workers most impacted by these
processes.

8.5 Solution path 4: The Kyrgyz Republic

8.5.1 Introduction
The fourth solution path relates non high-income countries to the
improvement of infant mortality rates in the context of protective labour market
policies (but not protective welfare state policies) and either T&C employment loss,
or no change in T&C employment. As can be seen in the below XY plot, both
Azerbaijan and the Kyrgyz Republic represent the most ideal-typical cases in this
solution path. Peru is the only case which is deviant in kind.

235
Korea Croatia
Solution 4
SlovakRep Greece
1 Colombia
South Africa Azerbaijan
0.9 Philippines
KyrgyzRep
Turkey
0.8
Health Improving (Infant Mortality Rates) Ecuador
0.7
Morocco
Egypt
0.6 Lithuania
Indonesia
Thailand Mexico
0.5
India
Brazil
0.4

0.3

0.2 Peru
Bangladesh
0.1
Latvia China Bulgaria
Poland0
Portugal Italy Mauritius
SriLanka 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
~HD*PLM*~PWS*~EG

Figure 31Solution Path 4 XY plot

As mentioned in the previous chapter, the Kyrgyz Republic was chosen to


explore this solution path since it has a higher membership in the solution path.

8.5.2 Collection of Empirical Material and the Causal Mechanism


The process tracing exercise for this solution term began with the collection
of empirical material for the Kyrgyz Republic. As with previous solution paths, the
goal was to build a narrative about the country’s T&C sector and inductively work
backwards in search of a plausible causal mechanism that might help explain the
solution’s results. In terms of Peru, process tracing was specifically used to
investigate what about the country is anomalous when compared to the Kyrgyz
Republic.

Again, the results of these efforts need to be understood in the context of data
sources and availability. In collecting empirical evidence for this solution path, a
dearth of data (in the English language) was found across the two countries, both in
terms of the sector’s more general characteristics and in relation to employment and

236
working conditions. In regards to the Kyrgyz Republic, evidence about the type of
workers employed in the sector, and their employment conditions, comes largely
from a report published by the Harvard Business School (Birkman, Kaloshkina,
Khan, Shavurov, & Smallhouse, 2012) and work undertaken by the ILO (2012b).
Evidence from these reports is then related to characteristics of the country’s labour
market and welfare state policies. Comparative data for T&C sector in Peru is drawn
mainly from a series of embassy cables which were released by Wikileaks in 2010.
These cables were sent by the Peruvian Embassy in Lima to government officials in
the U.S..

The areas within the EMCONET framework which were most considered
within this solution path relate to various aspects of 1) employment conditions,
particularly the types of labour which characterize the sector, i.e. formal versus
informal and 2) workers’ access to social provisions as afforded through various
channels of labour market and welfare state protection. As with other solution paths,
it is found that that despite countries’ characterizations of protective labour market
policies, provisions which may be health promoting or mediating in the face of job
loss are often inaccessible to T&C workers (even when they are present and
accessible to a country’s population at large). However, in contrast to previous
solution paths, this finding does not facilitate the construction of a plausible causal
mechanism. As will be discussed in greater detail below, this is because the
implications of these findings seem to be at odds with the improvement of infant
mortality rates. Moreover, T&C employment loss in the Kyrgyz Republic is found to
be accompanied by employment growth in the informal T&C sector. For these
reasons, and others which will be made clearer below, a potential causal mechanism
begins to emerge which relates the improvement of infant mortality rates in this
solution path to other changes in the country’s macroeconomic context.

The next section will turn to country-specific evidence which elucidates the
importance of the T&C sector for employment within the Kyrgyz Republic as well as
its economy as a whole. It will be shown that labour market protections within the
Kyrgyz Republic would have been largely inaccessible to T&C workers facing
employment loss after the MFA phase-out. It will also be shown that employment
loss in the formal T&C sector was accompanied by employment growth in the

237
informal sector. Peru and its anomalous position within the solution path will then be
explored.

8.5.3 Findings

Kyrgyz Republic
The Kyrgyz Republic became an independent state in 1991, after the collapse
of the USSR. As in the Slovak Republic, the T&C sector emerged when the country
was characterized by a planned economy and at this stage the sector was
characterized by a small number of large firms. After the demise of the planned
economy, the sector experienced a sharp decline, and subsequently rose again (in a
more fragmented form) in the early 2000s (Birkman et al., 2012).

As in other countries studied thus far, the T&C sector is regionally


concentrated in the Kyrgyz Republic, with 96% of firms located in the city of
Bishreck. Between 70-85% of workers in the sector are women, and the large
majority of firms are micro (employing 10 people or less) or small enterprises
(employing 10-50 people) (Birkman et al., 2012).

Informality in the sector is high (ILO, 2012b; Schwegler-Rohmeis, Mummert,


& Jarck, 2013) with estimates placing the share of informal firms between 40 percent
(ILO, 2012b; Umurzako & Burzhubaev, 2010) and 60-80 percent (Birkman et al.,
2012). In relation to these former estimates, the informal economy is defined as “all
economic activities by workers and economic units that are – in law or in practice –
not covered or insufficiently covered by formal arrangements. Their activities are not
included in the law, which means that they are operating outside the formal reach of
the law; or they are not covered in practice, which means that – although they are
operating within the formal reach of the law, the law is not applied or not enforced;
or the law discourages compliance because it is inappropriate, burdensome, or
imposes excessive costs” (ILO, 2012b, p. 4). In the latter estimates, informality is
defined as the “diversified set of economic activities, enterprises, and workers that
are not regulated or protected by the state” (Birkman et al., 2012, p. 16).

The sector is also recognized as an important contributor to overall


employment in the country. Schwegler-Rohmeis and colleagues (2013) estimate that
8% of the country’s total employment in 2010 was generated by the T&C industry.
Birkman and colleagues (2012) place this figure at 12 percent. Moreover, the ILO

238
(2012b) estimates that the sector’s contribution to the country’s GDP was 3 percent
in 2010, although Birkman et al. (2012) estimate that this figure ranges from 5-15
percent.

According to Birkman et al. (2012), official T&C employment figures began


to decline in 2007. Importantly for this solution path however, this is noted to likely
reflect a shift of employment to the informal sector, rather than a strict loss of jobs.
This suggestion is perhaps further supported by the steady increase of T&C exports
from the country (ILO 2012b) and broader characterizations of the country’s labour
market which point to a high level of informality across many sectors including the
T&C industry (Bernabè & Kolev, 2003).

This presents a scenario which is in some ways similar to the findings


revealed across high income countries losing T&C employment (i.e. in solution path
2). In these countries the post MFA phase-out context is characterized by not only
employment loss, but also by a shift to more precarious working and employment
conditions. In Italy for example, job loss in the T&C sector was noted to be
accompanied by worsening working conditions for those remaining employed in the
sector (Hadjimichalis, 2006). In the Slovak Republic, it was suggested that
liberalization of the country’s labour code, in conjunction with increased competitive
pressures, meant that many of those employed in the sector saw their contracts
effectively deregulated (with decreased access to social provisions once unemployed).
In the Kyrgyz Republic, a shift in employment from the formal to informal sector
also represents a move to more precarious employment conditions. However, in this
case a fundamental difference is that on the whole, the number of those employed in
the sector, may not have changed.

In the shift from formal to informal employment, those losing formal


employment are likely to be many of the same individuals taking up informal
positions. On the one hand, this relates to the regional concentration of the sector,
whereby companies are unlikely to be moving to locations out of reach of former
formal employees, as well as to the context of high unemployment in the country
which renders alternative employment opportunities particularly scarce, especially
for women (Bernabè & Kolev, 2003).

239
However, a perhaps more convincing reason why this is likely relates to the
structure of informality in the sector. Based on interviews with T&C firms and other
key experts in the country, an ILO (2012b) study finds that there are two broad types
of informality across the industry. The first type corresponds to legally registered
firms which underreport certain aspects of their production, such as their output, the
number of workers, or profits. The second type corresponds to enterprises which
work without formal registration. These establishments do not report anything at all
to the government. In regards to the former, the ILO study reports that “practically
all registered enterprises work half "in shade" and use various means to hide data on
actual production output” (p. 4). If we keep this mind and also remember that official
production figures were increasing while formal employment figures were
decreasing, it stands to reason that many formal employees were shifted to the
informal market, while remaining employed at the same firm. This notion is perhaps
further supported by 1) the low productivity which characterizes the country’s T&C
labour force (Birkman et al., 2012; ILO, 2012b), which demonstrates that increases
in production figures are unlikely to have resulted from increased productivity; and 2)
the fact that literature on the sector does not point to a reduction in formal firms.

For these reasons, an increase in informality in the T&C sector in the Kyrgyz
Republic likely means an increase in employment precariousness for formal
employees rendered informal. It may also mean an increase in informal employees
who may not have been previously employed by the sector, but this was neither
confirmed nor rejected by the available literature.

Workers losing employment in the formal sector were unlikely to benefit from
any job loss related social provisions. This follows from two main considerations.
First, while the Kyrgyz Republic does have collective redundancy provisions
(including severance pay) within its labour code, like Italy, workers employed in
firms with fewer than 15 employees are excluded from these provisions (ILO, 2014a).
Since so many of the country’s T&C firms are micro and small enterprises, it is
unlikely many dismissals would have fallen under the remit of these regulations.
Second, unemployment benefits in the Kyrgyz Republic are afforded on the basis of
a social insurance system. However, the ILO (2012b) reports that in analysing social
security records, the majority of T&C workers do not make social security
contributions.

240
Beyond social provisions, working conditions within T&C sector of the
Kyrgyz Republic are reported to be quite poor. According to the ILO (2012b), small
T&C firms “are usually located in basements of multi-storey buildings, private
homes or on the territory of abandoned large industrial buildings” (p. 5). Labour laws
and sanitary norms are acknowledged to be largely ignored, especially in firms which
operate informally, employees often work 14-17 hours a day, and labour accidents
are underreported or hidden. Wages in the sector are lower than both the average
wage in the country and in the manufacturing industry specifically. Moreover, any
increases in wages have been noted to be offset by simultaneous rises in consumer
prices. Interestingly, women’s wages are reported to be 2-5 percent higher than
men’s (ILO, 2012b). It is also important to note that much of the work in the sector is
seasonal, so for many, employment in the sector is insecure. For formal employees
shifting to informal employment it is not clear how much of a change they would
have seen in working conditions.

While the nature of insights considered above have facilitated plausible causal
mechanisms in relation to changes in adult female mortality rates in other countries
examined thus far, it is difficult to reconcile these considerations with the
improvement of infant mortality rates in the current solution path. A primary reason
for this is the dearth of information about the types of women working in the sector.
Evidence for instance, documenting the age and other socio-cultural characteristics
of female T&C workers in the Kyrgyz Republic could not be found. If we assume
women in the Kyrgyz Republic are having children while employed in the T&C
sector, it is unlikely that workers shifting from formal to informal employment
experienced a change in working conditions that might have facilitated the conditions
sufficient to improve infant health in their families, such as an increase in wages, or
improvement of maternity related labour provisions. However, if employment on a
whole grew, then for women newly employed in the informal labour market, their
incomes are likely to have improved, although this must be reconciled against the
typically poor working conditions under which they are employed and the resulting
impact this might have on infant health.

Aside from these direct impacts on infant health, the MFA phase-out may
have facilitated changes in the sector with indirect impacts on infant health. Two
factors highlighted by the EMCONET framework (Benach et al., 2007) which may

241
be important in this regard are economic inequality and material deprivation. In
regards to the former, data from the World Bank indicates that economic inequality,
as measured by the Gini coefficient, worsened in the Kyrgyz Republic between 2004
and 2009 (World Bank, 2013). Whether or not this worsening may have been
influenced by changes in the T&C sector, it is difficult to relate it to an improvement
in the country’s infant mortality rate since higher inequality is typically associated
with worse infant health (Wilkinson & Pickett, 2010).

In regards to material deprivation, it is worth noting that social spending as


share of total government spending has been declining since the late 1990s
(Mogilevsky & Omorova, 2011). Moreover, social assistance programs in the
Kyrgyz Republic are noted to be ineffective at reducing poverty on account of their
extremely low benefit levels (Hoelscher & Alexander, 2010). However, poverty
levels are noted to have declined in the country from 2002 -2007 (Birkman et al.,
2012), which might go some way towards explaining improved infant mortality rates.
Here poverty is measured by the share of people living on less than $2 and $1.25 a
day. A report by United Nations Department for Social and Economic Affairs
(Mogilevsky & Omorova, 2011) examines the Kyrgyz Republic’s achievement
towards the Millenium Development Goals and also notes declines in poverty
throughout the 2000s. Here poverty is measured by a general poverty line and an
extreme (food) poverty line. Whereas the later “reflects the cost of a food basket
consumption of 2100 Kcal per person per day…[the former] takes the food poverty
line and adds-up the cost of basic non-food expenditures” (p. 8). This report accounts
for these declines largely by pointing to increased economic growth. Such growth is
postulated to have been pro-poor due to the prevalence of low inflation, and to the
increasing value of remittances which are estimated to account for as much as 28%
of the GDP and largely accrue to the poor rural areas (Mogilevsky & Omorova,
2011).

Peru
As mentioned above, the search for evidence in the case of Peru proceeded
with a slightly different aim than it did for Kyrgyz Republic. Across the latter case, it
is hard to reconcile changes in the T&C sector with an improvement in the country’s
infant mortality rates. The search for evidence in the case of Peru sought to identify

242
what about the country is comparatively anomalous when considered in relation to
the Kyrgyz Republic.

As in the Kyrgyz Republic, T&C firms in Peru are found to be generally small.
In the export sector 92% of firms are reportedly small and medium-sized, although
larger companies are noted to generate 65% of the value in exports (Wikileaks,
2007b). An embassy cable sent in 2006, reveals that just 25 companies accounted for
72 percent of the country’s total apparel exports (Wikileaks, 2006b).

Overall, T&C employment is noted to account for 3% of total employment in


the country. However, the accuracy of these figures is questioned on the grounds of a
significant informal sector (Wikileaks, 2005).

The Peruvian T&C sector is noted to have expanded greatly in the early 2000s,
largely due to the implementation of a free trade agreement with the US: the Andean
Trade Promotion and Drug Eradication Act. This agreement which was implemented
in 2002 granted T&C exports duty free access to the US. Under it, Peruvian textile
exports to the United States are noted to have increased 60%: from $532 million in
2002 to $887 million in 2004 (Wikileaks, 2005). Expansion in the sector was
primarily seen in the export sector, versus the domestic market, which increased by
only 2% from 2002 to 2005 (Wikileaks, 2005).

The distinction between firms serving the export and domestic market is
important in regards to this solution path because in the otherwise absence of
evidence, it tells us something about where employment changes were likely to have
occurred in the industry. In the post MFA phase-out context, firms operating in the
domestic market were hit hard by cheap Chinese imports while sales in the export
market continued to grow (Wikileaks, 2005, 2006, 2007). For this reason, it seems
likely that some of the declines in employment figures originate from job loss in the
domestic market. However, it may also be the case that job loss occurred in smaller
T&C firms in the export market, since larger T&C firms are typically described as
better at managing competitive pressures.

Unlike evidence found in relation to the Krygyz Republic, sources of


evidence in relation to Peru do not indicate that job losses in the formal sector were
accompanied by growth in informal sector. If true, this would represent a significant

243
departure from the experience of the Kyrgyz Republic documented above however,
sources of evidence are far less diverse and informative in the Peruvian context.

8.5.4 Summary
The aim of collecting evidence within this solution path was to explain the
results of the fsQCA which relate the improvement of infant mortality rates in non-
highly developed countries to either T&C employment loss, or no change in T&C
employment, in the context of protective labour market policies but not protective
welfare state policies.

Working inductively backwards from available evidence, a causal mechanism


seems to emerge which accounts for an improvement in infant mortality rates in the
Kyrgyz Republic via wider macroeconomic changes. Interestingly, employment loss
in the country’s T&C sector seems to indicate a shift to informal T&C employment.
However, how likely it is that this experience is replicated across other countries
characterized by this solution path is unclear. Sources of evidence examined in
relation to Peru do not indicate that job losses in the formal sector were accompanied
by growth in informal sector. Future analyses examining employment changes in the
T&C sector should consider this type of interaction between the formal and informal
sector.

Evidence examined in relation to the Kyrgyz Republic does, however,


demonstrate that employment conditions within the sector became more precarious
after the MFA phase-out. This finding is aligned with evidence found in relation to
other countries, such as Italy and Bangladesh, where working conditions are
suggested to have worsened after the MFA phase-out.

8.6 Solution Path 5: Korea

8.6.1 Introduction
The fifth solution path relates high-income countries to the improvement of
infant mortality rates in the context of non-protective labour market and welfare state
policies and employment loss. As can be seen in the below XY plot, Korea is the
only typical case within this solution path and there are no cases deviant in kind. For
this reason, Korea is the only country studied in relation to this solution path.

244
South Africa Solution 5
SlovakRep
1 Croatia Korea
Greece
Azerbaijan Colombia
0.9
Philippines
Turkey
0.8 KyrgyzRep
Health Improving (Infant mortality rate)
Ecuador
0.7

Morocco
0.6 Egypt
Indonesia Lithuania
Thailand
0.5
Mexico
India
0.4 Brazil

0.3

0.2
Peru
0.1 Bangladesh
SriLanka Bulgaria
Italy 0
Portugal Mauritius
Poland 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Latvia HD*~PLM*~PWS*~EG*EL
China

Figure 32 Solution Path 5 XY plot

8.6.2 Collection of Empirical Material and the Causal Mechanism


The process tracing exercise for this solution term began with an evaluation of
the empirical material which was previously collected for Korea. Next, the aim was
to inductively work backwards in the search of a plausible causal mechanism that
might help explain the results of the present, fifth solution path.

As with previous solutions, the results of these efforts need to be understood


in the context of data sources and availability. Process tracing efforts undertaken in
regards to the second solution path found that there are two classes of literature
which cover the Korean T&C sector. The first discusses the T&C sector in relation to
gender divisions within Korean society, and the second discusses the economic
potential of the industry, especially in relation to the East Asian Crisis of 1997.
While this literature offers some general information about Korean T&C workers and

245
how the MFA phase-out might have facilitated employment loss, it is limited in what
it can tell us about changing infant mortality rates in the country.

The areas within the EMCONET framework which were most considered in
relation to this solution path concern various aspects of employment conditions, as
well as to workers’ access to social provisions as afforded through various channels
of labour market and welfare state protection.

Data collected in reference to Korea in the second solution path suggest that
after the MFA phase-out, T&C workers losing employment were likely to be women
engaged in various home-based subcontracting activities (Berik, 2008; Lee, 1993;
McNamara, 1999; Moon, 2003). Evidence also suggests that these workers were
likely to be married, over the age of 25, and have children (Berik, 2008; Lee, 1993;
Moon, 2003). However, precise data on the demographics of the industry, and on
those losing employment could not be found.

It was also noted that for workers losing T&C employment, family
commitments would have made it difficult to find alternative employment in the
competitive formal market (Moon, 2003), and that such workers would have had
little access to unemployment benefits, or severance pay (ILO, 2014b; Kim et al.,
2011; Kim, Khang, Muntaner, Chun, & Cho, 2008; Ku, 2007).

With limited data on the characteristics of Korean T&C workers it is hard to


relate changes in infant mortality directly to changes in T&C employment after the
MFA phase-out. This is because while evidence suggests that many women working
in the sector are mothers, it is not clear at what point women began their employment
in the sector, before or after having children. If women began working in the sector
before having children, then employment loss may have directly impacted infant
mortality rates. However, if women became employed in the sector after having a
child, direct impacts are likely to be less relevant. This relates to the low fertility
rates in the country. The World Bank (2013) reports that in 2005, women had on
average 1.1 children. Therefore, if women began working in the sector after having a
child, it would be unlikely that they’d be having another. Of course, it is possible that
fertility rates are higher among T&C workers, however no data could be found to
support either side of this possibility.

246
It is also hard to relate changes in T&C employment to changes in infant
mortality rates indirectly. The EMCONET framework identifies economic inequality
and material deprivation as two population-level factors which may influence health
(Benach et al., 2007). However, neither the OECD nor the World Bank has data on
income inequality in Korea prior to the MFA phase-out. Data from the OECD
(OECD, 2014) suggests that income inequality remained relatively stable between
2006 and 2009 with the following reported Gini coefficients for each respective year:
0.306, 0.312, 0.314, 0.3146. In relation to material deprivation, work by the Asian
Development Bank (Asian Development Bank, 2012b) indicates that the majority of
social spending in Korea goes to the non-poor. For this reason, even if employment
loss in the sector was accompanied by economic growth in the industry, it is unlikely
such growth would have improved the country’s social spending in a way which
would have led to decreased material deprivation among the poor.

8.6.3 Summary
With these complications in mind, it is difficult to construct a causal
mechanism from the MFA phase-out to the improvement of infant mortality rates in
Korea. Better data on the demographics of Korean T&C workers is much needed to
better understand the health impacts of changes in the sector.

8.7 Solution path 6: China and Thailand

8.7.1 Introduction
The sixth solution path relates non-high-income countries to the worsening of
infant mortality rates in the context of non-protective welfare state policies and
employment growth, regardless of protective labour market policies. As can be seen
in the below XY plot, China is the only typical case within this solution path. While
Bangladesh and India represent cases which are deviant in degree, only Thailand and
Indonesia are cases which are deviant in kind and thus suitable for comparative case
work. As mentioned in the previous chapter, Thailand is best positioned to be
examined comparatively with China since its membership in the solution path is
closer to China’s than that of Indonesia’s.

6
The OECD “Gini coefficient is based on the comparison of cumulative proportions of the population
against cumulative proportions of income they receive, and it ranges between 0 in the case of perfect
equality and 1 in the case of perfect inequality” (OECD, 2014).

247
8.7.2 Collection of Empirical Material and the Causal Mechanism
As with the previous solution paths, the process tracing exercise for this
solution term began with the collection of empirical material for China. In terms of
Thailand, process tracing was specifically used to investigate what about the country
is anomalous when compared to China.

Poland Portugal
Mauritius Solution 6
Italy Latvia
1 China
Bulgaria SriLanka
0.9 Bangladesh
Peru
0.8

0.7
Brazil
0.6 India
Health Worsening (infant mortality rates)

Mexico Thailand
0.5 Indonesia
Lithuania
0.4 Egypt
Morocco
0.3 Ecuador

KyrgyzRep
0.2
Turkey
Philippines
0.1
Azerbaijan
Colombia
Croatia
0 Greece
Korea
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
SlovakRep
SouthAfr ~HD*~PWS*EG*~EL

Figure 33 Solution Path 6 XY plot

As with previous solutions, the results of these efforts need to be understood


in the context of data sources and availability. In regards to China two main sources
are relied upon to understand the condition of T&C workers. One is a drafted report
prepared by Davin (2001) for the United Nations Research Institute for Social
Development’s Project on Globalization. This report investigates the impact of
export-oriented manufacturing on Chinese workers. Another is a peer-reviewed
journal article by Pun Ngai (2004) which seeks to understand the working conditions

248
of female migrant workers in China. In regards to Thailand, in-depth anthropological
work on Thai T&C workers by Pangsapa (2007) is relied upon to structure an overall
picture of the sector. However, this work does not account for the T&C sector
immediately prior to and following the MFA phase-out. For this reason work which
investigates the competitiveness of the sector is drawn upon to understand how these
characterizations may have changed. Overall, there is a relatively small literature
base on the Thai T&C sector.

The areas within the EMCONET framework which were most considered
within this solution path relate to various aspects of employment and working
conditions, as well as material deprivation and workers’ access to social provisions
as afforded through various channels of labour market and welfare state protection.
Similar to the case of Bangladesh, it is found that Chinese T&C workers face poor
working conditions in the context of non-protective labour market and welfare state
policies. While the demographics of the Chinese T&C workforce makes it difficult to
directly link changes in T&C employment to infant mortality rates, an indirect
pathway is identified which may have a role in explaining worsening infant mortality
rates in the context of increased T&C employment growth.

8.7.3 Findings

China
In China the T&C sector is rooted in liberalization policies which began in
1979 with the creation of special economic zones. As with other cases studied thus
far, the sector is regionalized with production concentrated in the provinces of
Zheijiang, Guangdong, Jiangsu, Shanghai, Shandong and Fujian (Wick, 2009). The
majority of Chinese T&C workers are reported to be female, although precise figures
could not be found (Barrientos & Howell, 2006; Wick, 2009).

As predicted by analysts, China has been a clear winner following the MFA-
phase out, at least economically speaking. Immediately after the phase-out, Chinese
exports to US and EU markets surged. For example, Chinese exports of cotton
trousers to the US increased by 1500% and exports of pullovers to the EU increased
by 534% (Kowalski & Molnár, 2009). Consequently, both the US and EU initiated
bilateral talks with China through which ‘memorandums of understanding’ were
reached (via the safeguard clause of China’s accession into the WTO). These
memorandums of understanding limited Chinese T&C exports across a range of

249
products categories until the end of 2007 in regards to the EU, and until the end of
2008 in regards to the US (Kowalski & Molnár, 2009). Despite these safeguard
provisions, the WTO’s International Trade Statistics reveals that China continued to
capture a growing share of the global T&C market following the MFA phase-out.
For example, China increased its exports of clothing to the global market by 20% in
2005, 29% in 2006 (WTO, 2007), 21% in 2007 and by 4% in 2008 (WTO, 2009). It
increased its exports of textiles to the global market by 23% in 2005, 19% in 2006
(WTO, 2007), by 15% in 2007 and 17% in 2008 (WTO, 2009).

As in Bangladesh, the majority of women working in the sector are internal


migrants and come from rural areas where otherwise little opportunity for paid
employment exists. There are also Chinese T&C homeworkers who are not internal
migrants; however, many of these women subcontract work from factories which
produce for the domestic market (Hong Kong Christian Industrial Committee, 2003)
and thus were presumably minimally impacted by export growth following the MFA
phase-out. Evidence suggests that non-migrant homeworkers who subcontract work
from export-oriented factories represent a relatively small proportion of the total
export-oriented workforce, which is largely composed of migrant workers (Hong
Kong Christian Industrial Committee, 2003).

Similar to the experience of female Bangladeshi T&C workers, employment


in the sector has facilitated an improvement in the autonomy of many rural born
female Chinese T&C workers (Davin, 2001). However, and also similarly to the
Bangladeshi case, these improvements have been limited by patriarchal structures
(Davin, 2001; Ngai, 2004) and exist in tandem with extremely poor working
conditions including long working hours, underpayment of and/or delayed wages,
unsafe factory conditions, a lack of security and employment benefits; and denials of
freedoms to associate and bargain collectively (Barrientos & Howell, 2006; China
Labour Watch, 2011; Davin, 2001; Ngai, 2004; Rodgers & Berik, 2006; Wick, 2009).
Such conditions persist despite legislated protections which stipulate the requirement
of employment contracts, maximum working hours and various health and safety
regulations (Barrientos & Howell, 2006; Ngai, 2004; Rodgers & Berik, 2006).

Moreover, related to their status as migrant workers, Chinese female T&C


workers are generally excluded from social welfare schemes (Davin, 2001). These
conditions arise in part from a national policy, ‘hukou’, which restricts the rights of

250
rural workers in urban areas. Broadly speaking, hukou is a system of household
registration which officially identifies a person’s place of residence. It originated in
1958 as a means to restrict migration from rural to urban areas. It also set out to
prescribe an individual’s eligibility for state provided benefits such as health care,
retirement provisions, and subsidized food and housing. Due to their agricultural ties,
those living in rural areas were cast off as self-sufficient and largely excluded from
social provisions (Chan & Buckingham, 2008).

Once China began liberalizing its economy, hukou was revised to


accommodate a growing demand for low-skilled workers in export-oriented
industrial sectors. Rural workers are now able to migrate to urban areas for work
however, the system continues to divide citizens into a rural and urban class,
whereby the former are “treated as inferior second-class citizens” (Chan &
Buckingham, 2008). Rural migrants for instance, are afforded only temporary
residence and are required to return to their rural home once an employment contract
ends (Davin, 2001). That T&C contracts are rarely put into writing (as required by
law) or are otherwise extremely short-term (Wick, 2009), means that hukou is a
significant source of labour control. Moreover, migrant workers have no claim to
urban social provisions which remain largely unmatched in rural communities where
familial support is the dominant means of social protection (Ngai, 2004).

Hukou also impacts other aspects of female T&C workers’ lives which are
important when considering pathways to health. To begin with, the temporary
residency status of migrants restricts workers’ ability to stay in the city. This
contributes to the fact that the majority of female T&C workers are housed in
dormitories, owned by factories or local authorities (Ngai, 2004). This ‘dormitory
labour regime’ (Ngai, 2004) also aligns with patriarchal structures within Chinese
society (Davin, 2001; Ngai, 2004). Conditions within the dormitories include
overcrowding, inadequate ventilation and lighting, and restriction of workers’
freedom of movement (e.g. via dormitory curfews) (Davin, 2001; Ngai, 2004).
Moreover, that accommodation is so tightly bound to employment is another
significant source of labour control. Workers living in factory dormitories are
reportedly prevented from looking for alternative work (Ngai, 2004) and from
developing relationships with men (Davin, 2001).

251
Hukou stipulations also restrict the registration of marriages and childbirth by
rural migrants in urban areas. This contributes to the type of worker employed in the
T&C sector: young (under 25), single, and childless. This also means that workers
often leave the sector after three to five years of employment in order to marry and
procreate (Ngai, 2004). In terms of this solution path, a main implication here is that
changes in T&C employment are unlikely to have a direct impact on infant mortality
rates, since the majority of those employed in the sector are not becoming pregnant.

However, as has been discussed with other cases, growth in the sector after
the MFA phase-out may have had an indirect impact on conditions which are
important for infant health at the population level. Economic inequality and material
deprivation are two areas identified by the EMCONET framework as important for
health (Benach et al., 2007). In regards to the former, data from the World Bank
indicates that economic inequality, as measured by the Gini coefficient, has remained
stable between 2004 and 2008 (World Bank, 2013). In regards to the latter, growth in
the T&C sector presumably would have increased the government’s spending
capacity however, that infant mortality rates worsened suggests that additional
factors need to be considered.

Feng and colleagues (2012) undertook an extensive analysis which looked at


the determinants of child mortality in 30 Chinese provinces from 1990-2006. Results
of the analysis point to the importance of social determinants of health in reducing
mortality rates of children under 5. This is an important consideration to take into
account since neonatal mortality rates have been shown to account for a high
proportion of the under 5 mortality rate in China. In the Gansu province for instance,
Yi and colleagues (2011) find that neonatal mortality accounts for 67.6% of under 5
mortality in urban areas and 59.0 % in rural areas. Social determinants identified as
important for under 5 mortality in China include mother’s education, household
crowding and access to clean water and sanitation (Feng et al., 2012). Of these
determinants, access to clean water presents the most obvious potential link to the
T&C sector.

Industrial growth has been associated with the widespread pollution of


Chinese water sources (WHO & UNDP, 2001; World Bank, 2001; C Wu, Maurer,
Wang, Xue, & Davis, 1999; Zhang et al., 2010). The T&C sector in particular has
been implicated as an especially significant contributor to water pollution in China

252
(Friends of Nature, the Institute of Public & Environmental Affairs, Green Beagle,
Environmental Protection Commonwealth Association and Nanjing Green Stone
Environmental Action Network, 2012; Jahiel, 2006; Jun et al., 2012; You, Cheng, &
Yan, 2009). While 96% of those living in large cities have access to safe drinking
water, this is true of less than 30% of people in rural areas (Tang et al., 2008), many
of whose water sources are downstream from T&C production and waste dumping
facilities. Contaminated drinking water has been shown to be associated with a range
of health problems (WHO & UNDP, 2001; C Wu et al., 1999; Zhang et al., 2010).
Importantly for this work, He and Perloff (2013) find that water pollution has a
significant negative effect on infant mortality, which may go some way towards
explaining the worsening of infant mortality following T&C production expansion
after the MFA phase-out.

Changes in air pollution have also been related to changes in infant mortality
rates in China (Tanaka, 2012) and as such may represent another environmental
pathway through which growth in the Chinese T&C sector contributed to the
worsening the country’s infant mortality rates.

Thailand
As mentioned above, the search for evidence in the case of Thailand
proceeded with a slightly different aim than it did for China. With regards to the
latter, a possible causal mechanism emerges which accounts for the worsening of
infant mortality rates (following employment growth in the T&C sector) via
reference to environmental pathways, particularly through the worsening of waste
water and air pollution. With this mechanism in mind, the search for evidence in the
case of Thailand sought to identify what about the country is comparatively
anomalous.

Thailand’s T&C sector shares many important features with China’s. To


begin with, females account for around 80-90% of the sector’s total labour force
(International Labor Rights Fund, 2002, The Fair Wear Foundation and Ethical
Trading Initiative-Norway, 2007); many are migrants, (although there is a significant
share of external as well as internal migrants); and working conditions within the
sector are extremely poor: many of the same deleterious conditions found in China
are also found in Thailand including long working hours, unsafe factory conditions,
delayed or under payment of wages, harassment and abuse and a lack of a written

253
contract (Pangsapa, 2007; The Fair Wear Foundation and Ethical Trading Initiative-
Norway, 2007).

Piya Pangsapa (2007) offers an extensive account of Thai T&C workers in the
time period leading up to and after the 1997 Asian financial crisis. On the basis of
this work we can distinguish Thailand’s T&C sector from China’s on at least three
related accounts which are important for this solution path. First, the majority of Thai
T&C workers do not live in dormitories. While the large majority migrate from rural
areas to Bangkok or its suburbs for employment, most live in private, albeit often
shared, accommodation. However, factory dormitory-based housing is noted to exist
to a minor extent. Second, while precise figures could not be found, evidence from
Pangsapa’s work suggests that a significant proportion of Thai T&C workers are
older, married and having children. Finally, while Chinese workers are typically
employed in the industry for a short period of time before getting married and having
children, many Thai workers have a long history of working in the sector.

An additional issue to consider is that after the 1997 financial crisis, many
production facilities relocated to border areas in search of cheaper labour (Pangsapa,
2007). As a result, migrants from other countries, and informal workers in rural
localities, began to accumulate a greater share of T&C employment (Arnold, 2007).
This is an important consideration for this solution path because if external migrants
occupy a significant share of T&C work, this will have consequences for how
changes in T&C employment can be related to the health of Thai nationals.
Moreover since Pangsapa’s (2007) analysis does not cover the time period after the
MFA phase-out, we must also consider how characteristics of the sector described
thus far, might have changed since her analysis.

Despite the relocation of production facilities to border areas, evidence


suggests that the majority of formal workers in the sector were Thai leading up to
and immediately after the MFA phase-out. This evidence comes from researchers
Goto and Endo (2014) who find that in 2006, 75.1% of formal T&C workers were
employed in areas not associated with migrant labour, namely, Bangkok and its
surrounding provinces. However, by the year 2010, this figure dropped to 56.6% due
in part to employment loss in the greater Bangkok area, as well as to employment
growth in border regions, particularly in the province of Tak (Goto & Endo, 2014).
Tak is home to Mae Sot, a city which is well-documented for its use of migrant

254
labour from Myanmar (Kusakabe & Pearson, 2010, 2014).What this signals is that
after the MFA phase-out, employment growth largely occurred in areas of Thailand
characterized by migrant labour, and that in fact many Thai T&C workers in the
greater Bangkok area experienced employment loss. However, it must be noted that
these figures are based on formal registered workers, and while some migrant labour
is accounted for in these official figures, unavailable are the numbers of both
informal Thai and external migrant workers.

These changes have multiple implications for this solution path. To begin
with, that many Thai T&C workers lost their employment represents a significant
departure from the post MFA phase-out context in China, where workers were
experiencing employment growth. For this reason we must consider whether such
employment loss was significant enough to impact national health outcomes.
Between 2006 and 2010, 12,964 or 6.1% of Thai workers in the greater Bangkok
area lost their T&C employment (Goto & Endo, 2014). While a sizeable figure, it
falls short of 15% benchmark used in the fsQCA analysis to qualitatively categorize
a country as fully in the membership set of employment loss. Because it is unlikely
this figure of employment loss can be significantly associated with changes in
national health outcomes, we are in a predicament similar to the one we encountered
in the case of China, whereby we were unable to relate changes in T&C employment
directly to infant mortality rates. However, employment loss within the sector may
have had other, wider ramifications on the sector, for instance by facilitating greater
job insecurity amongst Thai T&C workers maintaining their positions.

Complicating matters further is the fact that the sector did, overall, experience
employment growth and while we also cannot relate this growth directly to changes
in Thailand’s infant mortality rates, (since it occurred in a population of workers
whose children would not be accounted for in the examined health figures), as in the
case of China, there may be indirect pathways infant health was implicated. Again,
the EMCONET framework identifies levels of economic inequality and material
deprivation as two areas that are important for health in the context of labour markets
(Benach et al., 2007). According to World Bank data (World Bank, 2013),
inequality largely decreased in Thailand after the MFA phase-out. In 2002, the
country had a Gini coefficient of 42; in 2006 it increased slightly to 42.6 but by 2008
it had decreased to 40. This is contrasted with China, where the Gini coefficient

255
remained stable (World Bank, 2013). In 2004, the Thai T&C sector represented
about 4% of the country’s GDP (Pangsapa, 2007). While it would seem growth in the
sector would have improved the state’s spending capacity on factors that could have
reduced population level material deprivation, Goto and Engo (2014) note that in fact
the value of T&C exports peaked in 2005, and has been declining ever since.

Water and air pollution emerged as possible links between growth in the
Chinese T&C sector and the worsening infant mortality rates. In the case of Thailand
while some evidence suggests similar concerns (Greenpeace, 2011) less information
was available on these factors.

There is a final complexity that must be taken into consideration in the case of
Thailand. In late December 2004 a large Tsunami took the lives of 8,500 Thai people,
which may in part explain improving mortality rates from 2005. However, there are a
range of ways in which health may be impacted after a Tsunami, with some impacts
lasting for weeks and months beyond the disaster (Johnson & Galea, 2009). For this
reason more work is needed to consider how the Tsunami impacted mortality rates in
Thailand, both in 2004 and beyond.

8.7.4 Summary
The aim of collecting evidence within this solution path was to explain the
results of the fsQCA which relate T&C employment growth in non-highly developed
countries to the worsening of infant mortality rates in the context of non-protective
welfare state policies, regardless of protective labour market policies.

Working inductively backwards from available evidence, a potential causal


mechanism begins to take shape which accounts for the worsening of infant mortality
rates in China via an increase in T&C pollution after the MFA phase-out. However,
much more work is needed to investigate this potential pathway to health, and it is
unclear how likely pollution might have influenced infant mortality rates in other
countries characterized by this solution path. Future analyses examining changes in
the T&C sector should consider this question in greater depth.

Examining Thailand as a deviant case, adds little to the construction of a


causal mechanism in relation to this solution path. However, process tracing in this
case does highlight some interesting considerations about changes in the T&C sector
after the MFA phase-out. To begin with, while T&C employment growth is noted to

256
characterize the post MFA phase-out context, this portrayal veils a more complex
situation whereby many Thai nationals experienced employment loss and migrants
from neighbouring countries such as Myanmar, experienced employment growth.
Furthermore, while migrants are accounted for to some extent within national
employment figures, unaccounted for is the likely growth in informal T&C
employment, among both Thai nationals and external migrants. This finding is
aligned with work undertaken in previous solution paths which finds an increase in
precarious work after the MFA phase-out. It is also aligned with work undertaken in
regards to the Kyrgyz Republic, which found that official employment figures
masked interactions between the formal and informal market.

There are two main implications which follow from these considerations. First,
while it is noted that it is difficult to relate employment loss among Thai nationals to
infant mortality rates, more work is needed to understand the impacts of employment
changes on migrant and informal workers. Second, as indicated previously, future
work which investigates the health impact of T&C employment changes should
better account for interactions between the formal and informal segments of the
market.

8.8 Solution path 7: Sri Lanka, Bangladesh and Indonesia

8.8.1 Introduction
The seventh and final solution path relates less developed countries to the
worsening of infant mortality rates in the context of protective labour market policies
and employment growth, regardless of the protectiveness of welfare state policies. As
can be seen in the below XY plot, Brazil, Bangladesh and Sri Lanka are all typical
cases within this solution path. Sri Lanka and Bangladesh represents the most ideal
typical cases and as such will be examined comparatively. Indonesia represents a
case which is deviant in kind and thus suitable for comparative case work.

257
Latvia Portugal Solution 7
Mauritius Poland
1 Italy China Sri Lanka
Bulgaria
0.9
Peru Bangladesh
0.8
Health Worsening (Infant Mortality Rate)

0.7

0.6 Brazil
India
0.5 Mexico
Thailand
Lithuania Indonesia
0.4 Egypt
Morocco
0.3
Ecuador
0.2 KyrgyzRep
Turkey
Philippines
0.1
Azerbaijan
Colombia
0 Croatia
SlovakRep Greece
SouthAfr 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Korea ~HD*PLM*EG*~EL

Figure 34 Solution Path 7 XY plot

8.8.2 Collection of Empirical Material and the Causal Mechanism


The three countries investigated within this solution path are represented by
two different truth table rows in the fsQCA analysis. While Sri Lanka is
characterized by both protective welfare state and labour market policies, Bangladesh
and Indonesia are characterized by only protective labour market policies. However,
while examining Bangladesh in relation to the third solution path it was questioned
whether this was the correct characterization.

As with other solution paths, here the theory building process began with the
collection of empirical material. The goal was first to build a narrative about Sri
Lanka’s T&C sector similar to what was undertaken in relation to other solution
paths. This narrative was then compared with characterizations of the Bangladeshi
T&C sector. Here evidence from the third solution path was drawn upon; however,

258
new material was also gathered in specific relation to Bangladesh and changing
infant mortality rates. Next, the aim was to inductively work backwards in the search
of a plausible causal mechanism that might help explain the solution’s results. In
terms of Indonesia, process tracing was specifically used to investigate what about
the country is anomalous when compared to Sri Lanka and Bangladesh.

As with previous solutions, the results of these efforts need to be understood


within the context of data sources and availability. Scholarly work has focused more
on the social consequences of the Sri Lankan and Bangladeshi T&C sector than it has
in relation to other countries studied in this thesis. However, the results presented
here are not meant to provide an exhaustive account of this literature. Rather, the
focus is on conditions which were interpreted as causally relevant with the
EMCONET framework in mind. In regards to Indonesia, little scholarly work has
focused on the characteristics of T&C workers and their employment conditions.
However, literature on the country’s social policies, along with scholarly work which
focuses on the more economic aspects of the sector do provide some insights on how
the country and its T&C sector is comparatively situated in reference to Bangladesh
and Sri Lanka.

A range of areas within the EMCONET framework are considered in the


context of this solution path: labour market regulations, welfare state policies,
employment conditions, working conditions, economic inequality and material
deprivation.

Literature surrounding the T&C sector in Sri Lanka indicates a range of


similarities to the Bangladeshi sector; however, there are also important differences
that must be taken into account. Overall, focusing on evidence in relation to the
EMCONET framework fails to shed light on a precise mechanism linking changes in
the T&C industry to the worsening of infant mortality rates in Sri Lanka and
Bangladesh. Nonetheless, process tracing efforts undertaken in this solution path
bring into greater focus important considerations for future work in this area.

Presentation of the results will focus first on the importance of the industry for
the Sri Lankan economy. Similarities between the Sri Lanka’s and Bangladesh’s
T&C sector will then be explored. Finally, what about Indonesia is anomalous when
compared to these typical cases will be considered.

259
8.8.3 Findings

Sri Lanka and Bangladesh


As was the case with regards to the T&C sector in Bangladesh, it is also hard
to overstate the importance of the T&C sector for the Sri Lankan economy and for
women’s employment in the country overall. In 2004, the sector is noted to have
accounted for 48.8% of total export revenue (Kelegama, 2009), to have generated
9.5% of the country’s GDP and employed 6% of the country’s total labour force
(Kelegama, 2009). In 2006, the apparel sector alone is estimated to have employed
between 290,000 (Lopez-Acevedo & Robertson, 2012) and 383,000 (rounded,
authors calculations of COMTRADE data) workers. Moreover, over 1 million people
are estimated to be indirectly dependent on the sector (Lopez-Acevedo & Robertson,
2012).

There are other similarities between the Bangladeshi and Sri Lankan T&C
sector. First, it is widely acknowledged that in both countries the sector is rooted in
the MFA quota system (Kelegama, 2009; Lopez-Acevedo & Robertson, 2012).
Moreover, as in Bangladesh, T&C work in Sri Lanka has historically been, and
continues to be, carried out by a predominately female workforce. Figures place the
share of female employment in the range of 73% to 90% (Atkinson, 2007; Jayaweera,
2003; Savchenko & Lopez-Acevedo, 2012). In both countries, female workers are
also noted to remain in the sector for only a short period of time. This is for reasons
related both to the sector’s strenuous working conditions as well as to workers’
marriage and childbearing decisions (Prasanna & Gowthaman, 2006). Finally, as in
Bangladesh, the Sri Lankan T&C industry has brought many women the benefit of
increased autonomy and independence while at the same time, exposing them to poor
working conditions including low wages, long working hours, employment without
contracts, high work intensity, verbal and physical abuse, and a suppression of union
rights (Attanapola, 2004; Hancock, 2009; Lopez-Acevedo & Robertson, 2012; Lynch,
2007; Shaw, 2007).

The Sri Lankan T&C sector however, can be differentiated from Bangladesh’s
on at least four important accounts. First, while many of the sector’s workers are
internal migrants, leaving their rural villages for employment in the large urban-
based free trade zones (FTZs) (namely in Colombo and its outlying areas), a
significant proportion of Sri Lankan T&C workers are employed in production

260
facilities based in rural areas, in close proximity to their familial villages 7 . This
unique feature of the sector has its roots in a national strategy called the 200 Garment
Factory Programme. Created in 1992, this Programme sought to promote the industry
while simultaneously combating both rural unemployment and the stigma attached to
urban garment workers. Popular discourse in Sri Lankan society characterizes urban
garment workers as a threat to the country’s morality and national identify
(Hewamanne, 2008; Lynch, 2007). Since villages are considered sources of
‘pureness and virtuosity’, it was believed this image could be improved (and namely,
the industry promoted) by expanding the sector into rural localities.

This urban/rural dichotomy within the Sri Lanka T&C sector has important
implications for pathways to health. One implication relates to workers’ living
conditions: women migrating to urban areas for T&C employment often live in
dormitories, which, as in China, are noted for being overcrowded, unhygienic and
unsafe (Attanapola, 2004; Shaw, 2007). In contrast to China however, these
dormitories are owned and operated by individuals otherwise disassociated with the
factories. Rural-based workers by contrast, often continue living with their families
and so do not face these same accommodation problems. Research by Attanapola
(2004) demonstrates that for Sri Lankan women migrating to larger cities for work in
Free Trade Zones (FTZs), working and living conditions combine to contribute to a
worsening of their health. Another study shows that a main reason for workers’
dissatisfaction with FTZ work is the non-work environment (Shaw, 2007). Given the
poor record of working conditions within factories, this finding is particularly
revealing.

Another health implication of the sector’s urban/rural dichotomy relates to


working conditions as well as to broader community development considerations.
Goger (2013) identifies the 200 Garment Factory Programme as a catalyst for
improved working conditions in rural-based production facilities. This is because
companies with production facilities in Sri Lanka’s villages must display evidence of
moral responsibility in order to be accepted into local communities. Firms have been
highlighted for responding to this need by not only ensuring acceptable working
conditions within the factories but also by contributing to the improvement of the

7
A precise figure on the distribution of employment across these rural and urban settings could not be
found.

261
surrounding communities’ schools and housing. Village based T&C factories have
also been known to establish community centres and offer women’s empowerment
workshops in the areas where they are located (Goger, 2013). Such locally focused
initiatives are something not typically taken on by FTZ-based factories.

This is not to say that all village-based factories have better working
conditions than those based in FTZs. To be sure, many FTZ-based companies are
under pressure to ensure that they are meeting the standards of their international
buyers. Better working conditions are also a means of attracting and retaining
workers in a country where the stigmatization of garment workers means high labour
turnover and labour shortages. For reasons related to this later point, many Sri
Lankan T&C factories are noted to provide workers with free or subsidized meals,
transport, and health care services, even while other poor working conditions, as
noted earlier, prevail. Moreover, T&C workers living at home may experience
problems that those working in FTZ-based factories do not, such as domestic
violence, conflicts over changing income-earning roles in households, and a higher
degree of surveillance and patriarchal control (Goger, 2013). Another point to
consider is that not all the community-based investments that rural factory owners
pursue will be inherently good for workers or their communities. For example, as
Groger (2013) notes, such initiatives may serve to reinforce various forms of
patriarchal control.

Overall, it suffices to say that pathways to health in relation to the Sri Lankan
T&C sector will differ across the location of production facilities both in terms of
workers’ living and working conditions.

Sri Lanka is also differentiated from Bangladesh on the basis that there is at
least some evidence that its labour market and welfare state protections are
marginally better. In fact the Asian Development Bank (2012) considers labour
regulations in Sri Lanka to be relatively higher than those in most South Asian
countries. At least three examples can be provided towards evidencing this claim.

To begin with, up until 2010, Sri Lanka was the only Asian country to benefit
from the Generalized System of Preferences Plus (GSP+), a trade exception
established in 2005 which extends preferential access to European markets for
developing countries which have effectively implemented a series of 27 core

262
international conventions related to human rights, labour rights, the environment,
narcotics and corruption (Yap, 2013). However, in 2010, these benefits were revoked
from Sri Lanka for human right abuses committed by the government during the
country’s civil war (Yap, 2013).

Second, Sri Lanka’s ‘Garments Without Guilt’ campaign, while extremely


rudimentary and limited in its scope, seems to evidence a national commitment to
labour market protections that has not surfaced in relation to other Asian countries
that this thesis has investigated. This campaign, launched by the government and a
collection of industry associations in 2006, seeks to portray Sri Lanka as an ethical
source of T&C products. In doing so, it has created a certification system which is
guided by 5 principles. These principles aim for T&C working conditions to be
ethical and free of child labour, forced labour, discrimination on any grounds and
sweatshop practices. Compliance with the campaign is independently monitored by
the Swiss-based group SGS (JAAF, n.d.).

Finally, Sri Lanka has in place a system of social security protections for
T&C workers which are unmatched in countries characterized as lacking protective
welfare state policies. The Employees’ Provident Fund (EPF) and the Employees’
Trust Fund (ETF) form the main structure of social protections for Sri Lankan
workers in the private sector (ADB 2012a). The EPF requires a minimum
contribution equal to 20% of the employee’s gross earnings. The employer is
responsible for 12% of this contribution and the employee 8%. Employees are able to
withdraw their balances upon reaching retirement age (55 for males, 50 for females),
or upon retiring after marriage (for females only). Benefits can also be claimed on
medical grounds, for reasons related to migration, or upon death of the employee
(Asian Development Bank, 2012b).

The ETF requires employers to make contributions equal to 3% of an


employee’s total monthly earnings (Asian Development Bank, 2012b). These
contributions are not to be deducted from the employee’s wage. Through this Fund,
employees receive death benefits, disability benefits, financial assistance for heart
and eye surgeries, and financial scholarships for their children. Moreover, employees
can claim the balance of the fund (with interest and dividends) when they retire,
resign, or depart from their post without resignation. They can also claim these funds

263
when their employer ceases operations, or when they are dismissed from
employment (Asian Development Bank, 2012b).

Of course, none of these examples of protective labour market and welfare


policies are without their weaknesses. For instance, it is reported that many workers
were denied severance pay following T&C factory closures in the immediate
aftermath after the MFA phase-out (Atkinson, 2007). Some companies were also
found to have defaulted on their payments towards the Employees’ Provident Fund
(Sivananthiran, n.d.in Arunatilake, 2013). Moreover, the scope of the Garments
Without Guilt campaign is noted to be extremely limited. In 2010, only 20% of Sri
Lanka’s T&C production facilities were certified, a figure that appears to be
decreasing with increasing competitive pressures (Goger, 2013). The country also
has an extremely poor record with regards to allowing workers to freely organize
(Atkinson, 2007), despite having ratified ILO conventions which protect workers’
right to collective bargaining and freedom of association.

A third difference between Bangladesh and Sri Lanka relates to the type of
female undertaking T&C work. While in Bangladesh married women with children
represent a growing demographic among T&C workers (Kabeer, 2004), evidence
suggests that in Sri Lanka workers remain mostly young, unmarried, and childless
(Prasanna & Gowthaman, 2006; Shaw, 2007). This has important implications for
the ways in which we can process trace potential pathways to infant mortality in both
countries. In Sri Lanka the focus must be on indirect pathways since the large
majority of T&C workers are not having children. However, in Bangladesh, in
addition to indirect pathways, changes in infant mortality rates may be significantly
impacted via direct pathways since a larger proportion of T&C workers are noted to
be having children (between 40-50%) (Kabeer, 2004).

The final difference between the two countries relates to the employment
changes in the T&C sector following the MFA phase-out. In Bangladesh, process
tracing confirmed that employment in the sector continued to grow following the
phase-out. In Sri Lanka however, evidence is less conclusive.

World Bank authors Lopez-Acevedo and Robertson (2012) draw on figures


which demonstrate a reduction in T&C employment after the MFA phase-out.
However, there are at least three important weaknesses in these figures. The first is

264
that they are drawn from multiple sources with different estimation procedures: some
figures are drawn for example from industry estimations whereas others are drawn
from survey results. Second, the industry sources cited do not reference where their
figures are obtained from. For example, an article (Barrie, 2009) is cited from the
industry website, Just Style. The article, published in 2009, notes that 270,000 are
employed in the T&C sector; however, neither a year nor source for this figure is
provided. Finally, the figures provided in the World Bank report are noted to account
only for apparel employment, not textile. This seems strange especially since some
of the sources they cite seem to account for both, such as the aforementioned article
posted on Just Style. While textile employment is less prevalent in Sri Lanka than
that of apparel, it nonetheless represents an important source of work in the industry.
Data from INDSTAT demonstrates that in 2008, almost 13% of those employed in
the Sri Lankan T&C industry were employed in the textile sector (author’s
calculations based on INDSTAT data).

Contradicting the claim that employment decreased after the MFA phase-out
are embassy cables released by Wikileaks, as well as data from INDSTAT (which
was used to calculate T&C employment changes for the fsQCA analysis). Embassy
cables from Sri Lanka to the office of the United State Trade Representative in 2006
and 2007 report some initial T&C employment loss after the phase-out of quotas
(Wikileaks, 2006a, 2006c, 2007a). However, redundant workers are noted to have
been reabsorbed back into the industry. In the cables, evidence of this is offered via
reference to labour shortages in T&C facilities. Moreover, a released embassy cable
from 2008 which discusses the status of the industry does not point to employment
losses since the phase-out, although neither does it point to employment growth
(Wikileaks, 2008).

Data from INDSTAT, as used in the fsQCA analysis, exhibits a positive


change in employment in the pre- versus post-MFA phase-out period. This data is
collected from the Sri Lankan Department of Census and Statistics. However,
because of missing values the pre- phase-out employment figure used in the analysis
is from 2001 (instead of the ideal pre-phase-out year of 2004). In 2001, 212,668
workers are noted to have been employed in the sector. In 2006, 580,003 workers are
noted to have been employed in the sector. However, it could be the case that
employment grew substantially between 2001 and 2004, only to decrease in 2005

265
following the phase-out. In fact, between 2006 and 2008 we see a downward trend
from 580,003 workers in 2006 to 499,351 workers in 2007 and 383,882 workers in
2008 (author’s calculations from INDSTAT data). Therefore, INDSTAT data
provides little assistance in confirming the direction of employment change after the
MFA phase-out in Sri Lanka.

Despite these problematic employment figures, understanding the situation for


workers following the MFA phase-out seems at least somewhat reconcilable. A
major employment survey carried out by OXFAM immediately prior to and
following the MFA-phase out records initial employment loss after the MFA phase-
out (Atkinson, 2007). This survey forms part of the data that the World Bank authors
use to paint an overall picture of employment loss. However, employment loss is
noted to mostly have occurred in small, more rural-based factories. While it is likely
that some workers losing employment in rural-based facilities were reabsorbed into
urban-based factories, because T&C work in urban areas is highly stigmatized
(Hewamanne, 2008; Lynch, 2007) it is likely such transfers would have been limited.
In other words, we are at least able to deduce that following the MFA phase-out, Sri
Lankan workers in rural-based production facilities were more likely to face
employment loss than their urban based counter-parts. While some losing
employment may have been reemployed in urban factories, it is unlikely all those
losing their employment were reabsorbed. Therefore, without knowing whether on
the whole T&C employment increased or decreased after the phase-out, we can at
least safely presume that in general, many workers in rural based factories
experienced employment loss, while others were likely to either be re-employed or
employed for the first time in urban-based factories.

What are the implications of the above considerations for this solution path?
In the case of Bangladesh, we can envision how employment growth may have been
directly health promoting for IMR, namely through increased household resources.
However, as discussed in Solution 3, the tedious conditions under which women
work, and their own poor health status, could present problems for the prenatal
environment as well as for infants’ health after birth, if the mother must resume work
soon after giving birth. In terms of indirect pathways, again, the EMCONET
framework highlights the importance of economic inequality and material
deprivation. Of the data available from the World Bank we find that the Gini

266
coefficient in Bangladesh decreased (i.e. inequality decreased) faster in the 5 years
following the MFA phase-out than the 5 years preceding it. In 2000, the Gini
coefficient was 33.5, compared to 33.2 in 2005 and 32.1 in 2010 (World Bank, 2013).
While some of this reduction may be related to employment growth in the T&C
sector, overall the change is small and inequality remains high. Moreover, with
substantial changes in economic inequality we would expect improvement in infant
mortality rates (Wilkinson & Pickett, 2010).

In terms of social expenditure aimed at relieving material deprivation,


Bangladesh spends less than 2% of its GDP on social protection (Asian Development
Bank, 2012a). Moreover, social protection programs in Bangladesh fail to efficiently
target the poor and are instead directed towards the non-poor (Asian Development
Bank, 2012a). For these reasons, it would be difficult to relate changes in material
deprivation via social spending to changes in national infant mortality rates, whether
or not this spending capacity increased as a result of industry growth.

In the case of Sri Lanka, it is unproblematic that we do not know the overall
trend of employment changes when trying to understand pathways to IMR. This
follows from two considerations. First, since few Sri Lankan women working in the
T&C sector are having children, employment change in either direction would have
presumably had little direct impact on national infant mortality rates. Second, while
evidence doesn’t agree on the employment impact of the phase-out, it is agreed that
the industry overall grew after quotas were removed (Lopez-Acevedo & Robertson,
2012). For this reason, we can consider whether overall growth within the industry
indirectly influenced infant mortality rates. Compared with Bangladesh, economic
inequality is higher in Sri Lanka though over the same 10 year period decreased
faster. World Bank data indicates a Gini coefficient of 41.1 in Sri Lanka in 2000,
40.3 in 2007, and 36.4 in 2010 (World Bank, 2013). Without data for other years
however, it is unclear how these decreases are positioned in relation to the MFA
phase-out. Moreover, we would expect improvements in infant mortality rates with
substantial changes in economic inequality, not the worsening of rates as is seen in
the case in Sri Lanka. In terms of social expenditure, Sri Lanka spends 3.2% of its
GDP on social protection (Asian Development Bank, 2012b). While this is relatively
high in comparison to other low- and middle-income Asian countries, the social

267
protection programs in Sri Lanka, as in Bangladesh, are noted to be ineffective at
reaching the poor (Asian Development Bank, 2012b).

A final consideration to take into account is, as discussed in the case of


Thailand, on December 26, 2004 a large Tsunami struck Southeast Asia. As a result,
an estimated 31,000 -37,000 lives were lost in Sri Lanka (UNEP, 2005). The
circumstantial risk factors surrounding this disaster, which may differ across nations,
may in part explain worsening infant mortality rates from 2005 (Johnson & Galea,
2009).

In sum, a causal mechanism for this solution path appears to point to


considerations beyond those contained with the EMCONET framework. This finding
mirrors the results found in regards to other solution paths concerned with changes in
infant mortality rates. While process tracing efforts in the cases of the Kyrgyz
Republic and China revealed other potential pathways to health, (poverty reduction
in the case of the Kyrgyz Republic and industrial pollution in the case of China), this
was not the case in regards to this solution path.

However, in Bangladesh, high levels of arsenic in drinking water have been


identified as a significant source of health problems by the World Health
Organization as well as related to industrial output (WHO, 2014). More work is
needed however, to understand these potential pathways in both countries.

Indonesia
As mentioned above, the search for evidence in the case of Indonesia
proceeded with a slightly different aim than it did for Bangladesh and Sri Lanka.
Across the latter cases, it is hard to reconcile changes in the T&C sector with the
countries’ worsening infant mortality rates after the MFA phase-out. The search for
evidence in the case of Indonesia sought to identify whether something about the
country is comparatively anomalous.

As in Sri Lanka and Bangladesh, the T&C sector in Indonesia is rooted in the
MFA quota system (Hassler, 2004; Vickers, 2012) and characterized by a
predominately female workforce: estimates place the share of females working in the
sector between 50 (Robertson, Brown, Pierre, & Sanchez-Puerta, 2009) and 78
percent (ILO, 2012a). T&C companies are mostly centred in Java, particularly in
Jakarata and its outlying areas, as well as in Bandung, West Java’s capital

268
(Ferenschild & Wick, 2004; Vickers, 2012). Female workers are reported to come
from largely rural, and agricultural based livelihoods, and there is some evidence of
dormitory-based accommodation (Vickers, 2012). There is also some evidence that
Indonesian T&C workers are mothers (Ferenschild & Wick, 2004). Moreover, as in
Bangladesh, there is evidence of a large informal market within the T&C sector
(Vickers, 2012).

In comparison to Bangladesh and Sri Lanka, there is less scholarly work


focused on the working conditions of Indonesian T&C workers; however, there is
some evidence which suggests that working conditions are quite poor and
characterized by long working hours, low wages, sexual harassment, violations of
health standards and a suppression of union activities (Ferenschild & Wick, 2004).

After the MFA phase-out, Indonesian T&C export values are reported to have
increased (Thee, 2009; Vickers, 2012), however, additional information on the
employment impacts of the phase-out could not be found. There is evidence that
some employment loss (20,000 jobs) occurred immediately after the phase-out
(Chongbo Wu, 2007), although it is unclear whether workers losing their work would
have been reabsorbed back into the market. Relatedly, it is unclear what type of
employment alternatives Indonesian T&C workers have, if any.

Despite the limited data on the T&C workforce in Indonesia, there are three
main ways in which the country’s T&C sector can be distinguished from the typical
cases explored in this solution path. A main difference relates to the protectiveness of
the countries’ welfare state policies. An argument was made above that Sri Lanka’s
welfare state polices are marginally more protective than Bangladesh’s. Indonesia’s
welfare state policies can also be described as more protective than Bangladesh’s, but
they are perhaps not as protective as Sri Lanka’s. This notion is supported by work
undertaken by the Asian Development Bank (2013) which analyses data on social
protection programs in 35 countries in Asia and the Pacific. On the basis of this
analysis a social protection index (SPI) is developed which is essentially a ratio of
countries’ social protection expenditures to their intended beneficiaries. This ratio is
compared with ‘poverty-line expenditures’ so that if a country has a SPI of 0.100,
this would indicate “that total social protection expenditures (per intended

269
beneficiary) represent 10% of poverty-line expenditures” (p. xi)8. Therefore, a higher
index number indicates better performance. In this work Bangladesh is given a score
of 0.043, Indonesia a score of 0.044 and Sri Lanka a score of 0.121.

In terms of social policies which have specific relevance to T&C workers,


Indonesia also seems to rank somewhere between Bangladesh and Sri Lanka. For
instance, in terms of retirement funds, Sri Lankan employers and employees together
contribute 20% of a worker’s salary (12% is contributed by the employer and 8% by
the employee). In Indonesia, employers are required to contribute 5.7% of a worker’s
wage and employees are not required to contribute anything. While Sri Lankan
female T&C workers have access to their fund upon retiring for marriage, on medical
grounds, for reasons related to migration, or upon death of the employee (ADB,
2012a), in Indonesia, accumulated funds can be claimed before retirement in cases of
death, permanent total disability, permanent migration overseas, when becoming a
police officer, when joining the armed forces, or in the event of involuntary
unemployment. However, in the latter case the worker to is required to have been
participating in the scheme for at least 5 years and must submit to a 1 month waiting
period. There are no comparable policies in Bangladesh (ADB, 2012b).

That Sri Lankan T&C workers have access to funds after marriage seems
especially valuable since workers often leave the sector for marriage purposes. It also
seems particularly valuable that Indonesian workers are able to access benefits in the
case of involuntary unemployment. However, it is not clear whether the 5 year
working requirement is especially accessible to the demographic of T&C workers in
this country, particularly so given the large share of informal workers. Sri Lankan
employees also have access to funds in the case of unemployment, involuntary or
otherwise: recall from above that the Sri Lankan Employees’ Trust Fund requires
employers to make contributions equal to 3% of an employee’s total monthly
earnings and that the value of this fund can be claimed by employees when they
retire, resign, or depart from their post without resignation. They can also claim these
funds when their employer ceases operations, or when they are dismissed from
employment (ADB, 2012a).

8
Countries’ poverty-line expenditures represent one-quarter of their GDP per capita (Asian
Development Bank, 2013)

270
Based on the countries’ distribution across different rows of the fsQCA truth
table, the differences in welfare state policies between countries is somewhat
expected: recall that while Sri Lanka is characterized as having protective welfare
state policies in the fsQCA analysis, Indonesia and Bangladesh are not.

In terms of labour market policies, the fsQCA analysis characterizes all three
countries as protective. Process tracing efforts in the third solution path however,
revealed a dearth of labour market protections in Bangladesh, owing largely to the
failed implementation of mandated policies. In Sri Lanka, weaknesses are also noted
in the implementation of protective labour market policies, especially in the
aftermath of the MFA phase-out, but perhaps to a lesser degree than in Bangladesh.
In Indonesia, the implementation of protective labour market policies seems to be
achieved in a way that is superior to implementation efforts in both Bangladesh and
Sri Lanka. However, as will be described below, weaknesses exist in other policy
areas and better implementation hasn’t necessarily lead to an improvement in
workers’ employment conditions. This finding represents the second way Indonesia
can be distinguished from the typical cases of this solution path.

For example, all three countries have relatively similar labour market
regulations on severance pay, a key aspect of employment protection regulation (ILO,
2014a). Given the findings of solution path 3 it is doubtful whether such regulations
are adhered to in Bangladesh within any reasonable respect, outside of the small
population of foreign-owned factories within EPZs. Moreover, as noted earlier, after
the MFA phase-out many Sri Lankans were left jobless without severance pay. In
Indonesia, despite some factory closures after the MFA phase-out, employers’ failure
to abide by severance pay regulations is not something which emerged from the
literature. While this may be the result of a smaller literature on the Indonesian T&C
workforce, existing literature points to another possible reason why this outcome is
observed.

In 1998, the emergence of democratic rule is noted to have led to significant


changes in Indonesian labour market policies and their implementation (Lake, 2008).
In this transformation, an overall pro-employer labour market position is noted to
have given way to stronger worker rights. T&C companies are reported to have
responded specifically to a shift towards stronger severance pay regulations by
exchanging permanent employees for a greater number of contract workers, to whom

271
the regulations on severance pay do not apply (Lake, 2008). While this seems to
represent greater adherence to labour regulations, it also represents deterioration in
workers’ employment conditions since contract workers face greater job insecurity
and decreased access to social provisions. Moreover, in making this move, T&C
companies are noted to evade a different aspect of labour regulation; namely one
which precludes the long term employment of contract workers (Lake, 2008).

Indonesian T&C companies deploy a similar tactic in response to minimum


wage laws, which are decentralized to local authorities. Without betraying local
regulations on workers’ minimum pay, companies instead relocate to localities with
lower wages (Lake, 2008).

Scholars relate these pro-worker regulations on severance pay and minimum


wages to the strength of unions in Indonesia (James, Ray, & Minor, 2003). That
Indonesian unions are characterized as having an active role in shaping regulation is
the third way in which Indonesia may be distinguished from Bangladesh and Sri
Lanka, since labour unions in these countries are best characterized as suppressed.
However, the ways in which T&C companies have responded to protective labour
market policies suggests that unions in the Indonesian T&C sector may be
weakening. This is because workers on temporary contracts may find it harder to
organize. Moreover the threat of relocation may curb workers’ desire to collectively
organize. Rasiah (2012) notes that after the MFA phase-out, increased competitive
pressures weakened Indonesian unions. Therefore, whether active unions is an
enduring feature of T&C labour in Indonesia or one which was short-lived after the
country’s shift to democracy, is something that needs to be further investigated.

8.8.4 Summary
Recall that this solution path relates non-high-income countries to the
worsening of infant mortality rates in the context of protective labour market policies
and employment growth, regardless of the protectiveness of welfare state policies.
Working inductively backwards from available evidence however, changes in the
T&C sector do not seem related to changes in Bangladesh’s and Sri Lanka’s infant
mortality rates. Moreover, examining Indonesia as a deviant case adds little to these
findings.

272
Results from this case-study work however, do allow for some reflection on
the fsQCA analysis and its characterizations of countries’ labour market and welfare
state policies. As previously noted the fsQCA analysis characterizes Bangladesh and
Indonesia as having protective labour market but not protective welfare state policies.
It characterizes Sri Lanka as having both protective labour market and welfare state
policies. As noted in relation to other solution paths, it is difficult to relate the case
study findings back to countries’ overall labour market and welfare state
characterizations since these findings are focused on T&C workers; however, in this
case it is also difficult to answer how these considerations should be taken into
account in future cross case analyses.
It has already been noted that T&C workers in Bangladesh are characterized
by neither protective labour market nor welfare state policies. However, in terms of
Indonesia and Sri Lanka, how labour market and welfare state policies should be
characterized in relation to T&C workers is perhaps less clear than in other cases.
This is because there appears to be something qualitatively different about T&C
workers in Sri Lanka and Indonesia than in many of the other countries investigated
thus far, even in regards to more highly developed countries, with the possible
exception of Portugal. For example, in both Sri Lanka and Indonesia, T&C
employers (and T&C employees in the case of Sri Lanka) are required to contribute
to workers’ retirement funds. Moreover these funds are available to workers for
reasons related to voluntary or involuntary unemployment. These types of policies
were not uncovered in relation to other countries, except in Portugal where older
workers were noted to have the option of early retirement.

Additionally, in Sri Lanka, the government is pursuing a national marketing


strategy to portray the country as an ethical supplier of T&C goods. In Indonesia,
workers’ unions are noted for having an active role in shaping labour market and
social policies. For these reasons, T&C workers in Sri Lanka and Indonesia seem less
hidden than in other investigated countries (although a large informal market is noted
to operate in Indonesia). However, in both countries the genuine social protection
offered by these considerations is questioned. This relates both to noted weaknesses
in implementation, as well as to competitive pressures which are highlighted for
diminishing the impact of potentially health protective pathways. For example,
competitive pressures are noted to reduce the impact of Sri Lanka’s ethical garment

273
production campaign. In Indonesia, T&C companies respond to competitive
pressures by circumventing more (costly) protective policies and in doing so,
increase the precariousness of employment in the sector. Further work is needed to
investigate how these types of considerations should be taken into account in future
analyses.

274
CHAPTER 9 DISCUSSION
9.1 Introduction
This chapter will critically discuss the results of the in-depth case study work.
Recall that the aim of undertaking this case work was to provide a potential causal
account of the cross-case patterns achieved in the fsQCA analysis. The fsQCA
analysis identified seven solution paths as sufficient for changing mortality rates after
the MFA phase-out. Twelve countries were selected for in-depth case work based on
Schneider and Rohlfing’s (2013) systematic procedure for choosing cases after a
fsQCA. Theory-building process tracing was used in an attempt to construct potential
causal mechanisms for each of these seven solution paths. This was guided by the
EMCONET framework (Benach et al., 2007) which outlines pathways to health in
the context of labour markets.
This chapter will begin with a summary of the process tracing results
to indicate the type of plausible causal mechanisms which emerged from this work. It
will then move to critically consider the tentative nature of these results and other
limitations of the process tracing efforts. Next, the chapter will consider the process
tracing results in relation to broader considerations of the health impact of the MFA
phase-out and the relationship between trade liberalization and health more
generally. Finally, the chapter will conclude with some thoughts on how combining
process tracing with fsQCA leaves us better positioned to further investigate the
health impacts of the MFA-phase-out.

9.2 Summary of Causal Mechanisms


The fsQCA analysis identified seven solution paths as sufficient for changing
mortality rates after the MFA phase-out. Three of these solution paths related to
changing adult female mortality rates and four related to changing infant mortality
rates. Across these seven solution paths, specific potential causal mechanisms
relating changes in T&C employment after the MFA phase-out to health emerged for
only two.
Brazil was investigated in the first solution path, which related employment
growth in conjunction with protective labour market and welfare state policies, to an
improvement in adult female mortality rates. Unfortunately, process tracing in this
solution path did not lead to the construction of a potential causal mechanism. This is

275
because it was found that a conditional cash transfer programme (‘Bolsa Familia’)
was introduced in Brazil in 2003, which was likely to have influenced positively on
adult female mortality rates after the phase-out in 2005. By 2006, this programme
covered 11 million household, and has since been noted for its role in reducing
poverty and inequality (Sánchez-Ancochea & Mattei, 2011). In public health
literature it has also been associated with reductions in childhood mortality (Rasella
et al., 2013).
In the second solution path, four countries were investigated: Italy, the Slovak
Republic and Korea were investigated as typical cases, and Portugal was explored as
a deviant case. The fsQCA solution characterizing these countries was at first glance
puzzling. This is because it related decreases in T&C employment in highly
developed countries to a worsening of adult female mortality rates, regardless of the
presence or absence of protective labour market and welfare state policies. This was
puzzling since we might expect protective policies to act as a buffer to the potentially
negative impacts of employment loss.
Process tracing efforts however, were able to uncover a potential causal
mechanism that might explain these results. Across the typical cases investigated,
evidence was found which suggests that regardless of whether a country could be
characterized by protective welfare state or labour market policies, T&C workers
losing their employment after the MFA phase-out would have had little access to
social protection. It was also found that T&C workers in these countries would have
had few alternative employment opportunities.
As a deviant case, Portugal by contrast did not experience a worsening of
adult female mortality rates. Despite being a member of the solution term, here
evidence was found which suggests that Portuguese T&C workers had greater access
to social protection after the MFA phase-out, given that their work was likely to be
characterized by a more standard, full-time employment relationship. In addition to
unemployment insurance, the evidence suggests that Portuguese T&C workers losing
their employment would have been covered by the country’s collective dismissal
regulations. These regulations require employers to give workers advance notice of
dismissals and time off to look for alternative work. Workers are also entitled to a
minimum severance pay and possible re-training opportunities or early retirement.

276
The third solution path was characterized by only one country: Bangladesh.
This solution path described a sufficient relationship between the worsening of adult
female mortality rates and employment growth in less developed countries with
protective labour market (but not welfare state) policies. Process tracing efforts
undertaken in regards to this solution found that the T&C sector in Bangladesh is
characterized by a range of complex and often contradictory processes in terms of
how it impacts the lives of its largely female workforce. This relates primarily to the
type of firms within which women work and the different spheres of women’s lives
which are impacted, sometimes negatively, sometimes positively, by work in the
sector. Because evidence suggested that Bangladeshi T&C workers have little access
to social protection and that employment growth after the MFA phase-out took place
in firms with poor working conditions, ultimately a potential causal mechanism
emerges which directs attention to these conditions.

The final four solution paths relate to changes in infant mortality rates.
Process tracing efforts across these different solution paths however, were unable to
uncover any specific potential causal mechanisms linking changes in health to T&C
employment changes after the MFA phase-out. In some cases this related to a dearth
of evidence on whether T&C workers were having children. In other cases, such as
in regards to Thailand and Sri Lanka, there was uncertainty about the health impact
of a large Tsunami which struck some of the analysed countries in late December
2004, just prior to the MFA phase-out. Pathways related to national levels of
economic inequality and material deprivation were also explored in relation to infant
mortality rates, but tying these issues to changes in the T&C sector proved difficult.
In relation to the Kyrgyz Republic, a causal mechanism began to emerge which
accounts for an improvement in infant mortality rates via wider macroeconomic
changes, unrelated to the T&C sector. In China, a potential causal mechanism began
to take shape which attributes the worsening of infant mortality rates to an increase
in T&C pollution after the MFA phase-out. However, much more work is needed to
investigate this potential pathway to health, and it is unclear how likely pollution
may be influencing infant mortality rates in other countries characterized by the same
combinations of causal conditions.
. Overall, it is found that in highly developed countries which experienced
employment loss after the MFA phase-out, the worsening of adult female mortality

277
rates seems to be related to T&C workers’ inability to access social protections as
well as few alternative work opportunities. In less developed countries which
experienced T&C employment growth after the MFA phase-out, it is found that the
worsening of adult female mortality rates seems to be related to an absence of social
protection policies, in combination with poor working conditions which were likely
to have worsened after the removal of T&C quotas. The next section of this chapter
will look at the reasons why these results themselves are limited and in some ways
quite tentative. However, despite the provisional nature of these results, it is worth
stressing that they do point to the importance of social protection policies in
determining the health impacts of trade liberalizing processes.

9.3 Limitations of Results:

9.3.1 Data Availability


A major factor which limited process tracing efforts was data availability.
This not only hampered the construction of causal mechanisms across the majority of
solution paths as discussed above, but also renders the two potential causal
mechanisms which emerged out of this work rather tentative.
For example, the potential causal mechanism which emerged in relation to
Italy, the Slovak Republic, Korea, and Portugal rests heavily on a small literature
base which was far more focused on matters related to economic growth and firm
efficiency rather than social conditions. While care was taken in drawing inferences
from this literature, the preliminary nature of this causal mechanism needs to be
emphasized.
The second causal mechanism was constructed with evidence gathered in
relation to Bangladesh. In contrast to the previous causal mechanism, much more
scholarly work has been carried out relation to the Bangladeshi T&C sector,
especially in the context of its social consequences. However, data was still limited
in specific relation to the MFA phase-out. For this reason, the preliminary nature of
this causal mechanism must also be emphasized.

9.3.2 Precision of Causal Mechanisms


Another limitation of the potential causal mechanisms is how far they go in
specifying precise pathways to health. In relation to developed countries, the first
potential causal mechanism emphasises the health importance of social protection

278
policies related for example to collective dismissals, severance pay and
unemployment insurance. However, it is not immediately clear how these conditions,
or their absence, in turn, may be impacting health. The EMCONET framework
identifies a range of different pathways which may characterize the pathways to
health from these conditions for example material deprivation, health behaviours,
physio-pathological changes and psychosocial factors. However, data on these
conditions were not uncovered in this work. Therefore precise mechanisms to health
remain unclear.
In relation to the second causal mechanism, working conditions in addition to
social protection policies are emphasized. However again, this leaves a range of
more precise processes unclear. That the sector is noted to sometimes impact
positively on workers adds an additional level of complexity to this mechanism.

9.3.3 Generalizability of Causal Mechanisms


A final limitation of the process tracing results relates to their embeddedness
within a larger cross-case analysis. A primary concern here is that while the causal
mechanisms seem to explain (at least tentatively) some of the countries investigated
in greater depth, these results are limited in explaining the situation of other countries
characterized by similar solution paths of the fsQCA. This relates less to the inherent
nature of combining fsQCA with process tracing efforts (which will be discussed in
the next section) than to the process tracing results themselves. This is because the
causal mechanisms which emerged were not associated with the original causal
conditions of the fsQCA, but rather drew attention to the need for refinement in these
conditions.
For example, in relation to the first causal mechanism, despite countries’
characterizations of protective labour market or welfare state policies, T&C workers
were found to have little access to social protection. This means that for the causal
mechanism to be relevant to other countries characterized by the original solution
path, we must know more about T&C workers’ access to social provisions in these
other countries. If however, the original conditions of the fsQCA were related more
directly to the causal mechanism, we would have greater reason to believe they are
relevant to these other countries. Consider for example, a hypothetical solution path
which relates strong social protections to improved mortality rates. Suppose also that
this solution path is characterized by 7 typical cases, but only 3 are investigated in-

279
depth to build a potential causal mechanism. If a causal mechanism were to emerge
that showed how this social protection was health improving in these 3 cases, it
would be easier to assume that this causal mechanism might be in place in the other 4
typical cases, given that we already know that these other typical cases have
protective social policies. This stands in contrast to the first causal mechanism found
in this work, since reference is made to conditions that we don’t yet have information
on for other typical cases.
In relation to Bangladesh and the second causal mechanism, the situation is
slightly different. Here the causal mechanism refers to causal conditions that we do
have information on in relation to other countries. In this causal mechanism, an
absence of social protection policies is related to the worsening of adult female
mortality rates, in addition to poor working conditions. While Bangladesh was
characterized in the fsQCA as having protective labour market policies, this causal
mechanism shows that Bangladesh is instead best characterized by neither protective
labour market nor welfare state policies. This is a combination of conditions which
characterizes China, India and Thailand in the fsQCA. These countries however, are
not characterized by a sufficient relationship in the fsQCA, although if the analysis is
re-run with Bangladesh re-categorized, a sufficient relationship may arise. In relation
to the second causal mechanism then, the limitation is not that it is unclear whether
the causal mechanism applies to other countries, but that the other countries to which
it is likely to apply are not currently characterized by a sufficient relationship.
Despite these limitations, it is important to keep in mind that process tracing
was undertaken with the objective of identifying plausible causal mechanisms that
can be tested empirically in subsequent research. As the next section of this chapter
will show, aside from aiding in the construction of two potential causal mechanisms,
process tracing results also shed meaningful light on broader considerations related
to the health impact of the MFA phase-out, and on the relationship between trade
liberalization and health more generally. These are findings which can be taken into
account in subsequent research. Moreover, the concluding section of this chapter will
illustrate how combining fsQCA with process tracing has established a solid basis
from which many of the challenges noted above can be overcome.

280
9.4 Broader Insights
Aside from aiding in the construction of two potential causal mechanisms,
process tracing results also shed meaningful light on broader considerations related
to the health impact of the MFA phase-out, and on the relationship between trade
liberalization and health more generally.

First, aligned with previous comparative welfare state studies (Beckfield and
Krieger, 2009), the results of this work seem to indicate the health importance of
social protection. Two causal mechanisms emerged which suggest that the worsening
of AFM is related to T&C workers’ inability to access social protections. This is
found to be the case in the context of both T&C employment growth (in less
developed countries) and loss (in highly developed countries). Evidence collected in
relation to the first solution path also points to the health importance of protective
social policies despite the fact that a causal mechanism could not be constructed.
Here we encounter evidence from Brazil where reductions in poverty, inequality and
child mortality have been associated with an expansive conditional cash transfer
programme implemented just two years prior to the phase-out.

In the context of the MFA phase-out, trade liberalization was found to


increase competition within the sector via an intensification of price pressures. This
competition in turn, impacted T&C labour markets both in terms of shifting
employment and working conditions. Changes in employment were found to extend
beyond strict gains or losses. In regards to the Kyrgyz Republic for instance,
evidence suggests that T&C employment loss in the formal sector corresponded with
employment growth in the informal sector. In regards to Thailand, evidence suggests
that figures pointing to employment growth mask a greater reliance on migrant and
informal labour, which has resulted in employment losses for many Thai nationals.
Furthermore, both in countries experiencing employment growth (e.g. Bangladesh)
and employment loss (e.g Italy), evidence was found for a worsening of T&C
working conditions after the MFA phase-out.
Social protection policies were found to interact with these considerations,
and in response to the employment and working conditions of the sector in general,
in many health important ways. For example, factors such as the size of T&C firms,
the types of labour contracts through which workers were employed, and the types of
labour overall which characterize the sector (e.g. formal versus informal, legal versus

281
illegal), were found to determine T&C workers’ access to protective labour market
and welfare state policies. These employment and working conditions often in turn,
relate back to countries’ labour regulations and also respond to the competitive
pressures of the sector. In other words, social protection policies can both moderate
pathways to health in the context of labour markets and influence the type of health
related pathways resulting from trade liberalizing policies. They can moderate
pathways to health by influencing the type of social protection available to affected
workers. They can influence the type of health related pathways resulting from trade
liberalizing policies by shaping for example, the type of employment contracts
through which T&C firms can hire workers. In the Slovak Republic for example,
labour regulation policies allowed for more precarious forms of T&C employment.
These forms of employment would have precluded T&C workers losing their
employment from accessing social protection policies like unemployment insurance.

Overall, findings suggest that that social protection is often inaccessible to the
type of workers who may be the most vulnerable to processes of liberalization, and
that many workers are particularly vulnerable to processes of liberalization due to the
structure of labour market and social protection policies. Perhaps surprisingly, this
seems to be the case in both developed and less developed countries. In fact, social
protection for T&C workers seems to be greater in some less developed than more
developed countries. In Sri Lanka and Indonesia for instance, T&C employers (and
T&C employees in the case of Sri Lanka) are required to contribute to workers’
retirement funds. Moreover, these funds are available to workers for reasons related
to voluntary or involuntary unemployment. This can be contrasted with T&C
workers in Italy, who are often (when employed in firms of less than 15 employees),
exempt from protective labour regulations and social policies in the case of
employment loss.

9.5 Combining fsQCA with Process Tracing


Recall that the fsQCA was undertaken as an exploratory analysis and process
tracing as a means to build potential causal mechanisms that could be tested in
subsequent research. The chapter will now move to discuss how combining fsQCA
with process tracing leaves us better positioned to further investigate the health
impacts of the MFA phase-out. Chapter 10, the concluding chapter, will more

282
broadly consider how the results of this thesis leave us in terms of further
investigating the health impacts of trade liberalization.
To begin with, process tracing results uncovered a range of considerations
which, if accounted for, can improve the validity of future cross-case analyses
concerned with the health impacts of the MFA phase-out. For example, the fsQCA
undertaken in this work characterized countries in terms of the protectiveness of their
overall labour market and welfare state policies. However, results of the case study
work demonstrate that future investigations would be better off considering how well
such policies align with the type of workers employed in the T&C sector. For
example, while Italy was characterized as a country with protective labour market
policies, Italian T&C workers had little access to any potentially protective
provisions. This is because Italian labour regulations excluded precisely the type of
firms where a majority of T&C workers were employed prior to the phase-out (those
with fewer than 15 workers).

Process tracing results also indicate the importance of regulations related to


collective dismissals, severance pay and unemployment benefits. Considerations of
these regulations in future cross-case analyses may offer an alternative means of
measuring the protectiveness of labour market policies in the context of the T&C
sector. A valuable source of this information is the International Labour
Organization’s Employment Protection Legislation Database. This database outlines
employment termination legislation for 95 countries, and includes information such
as the type of establishments excluded from dismissal regulations and the conditions
under which severance pay is mandated. Results indicate that future analyses should
also consider that changes in employment extend beyond simple dichotomies of
losses and gains.

Limiting the efforts of future analyses will be similar data availability issues
which were encountered in the work of this thesis. However, by providing an in-
depth account of causally relevant factors associated with the T&C sector, process
tracing results do establish a basis for the MFA phase-out to be investigated in ways
that can potentially avoid some of the complications encountered in this work, such
as those noted in relation to Brazil and countries struck by the large tsunami in 2004.
For example, a finding typical across all of the cases studied in-depth is that the T&C
sector is highly regionalized within countries. Using health data disaggregated by

283
region may therefore provide further insight about the health impacts of the phase-
out in ways which limit the confounding influence of other issues.

284
CHAPTER 10 OVERALL FINDINGS, CONTRIBUTIONS AND NEXT
STEPS
10.1 Introduction
In this final chapter, the contributions this thesis makes to the literature are
reviewed. The main results, strengths, and limitations of the thesis will be considered
along with where we are left in terms of next steps. These considerations will
illustrate that while the relationship between trade liberalization and health is indeed
enormously complex, this thesis has made significant strides towards tempering this
complexity and additionally provides strong footing for future advancements in this
regard.

10.2 What are the Pathways Between Trade Liberalization and


Health: Insights from the Literature Review
The contributions this thesis makes to the literature begin with the findings
and analysis of the literature review. Over the years, several frameworks have been
developed to synthesize the complex pathways between globalization and health
(some discussed in this thesis are Woodward et al., 2001; Labonté and Torgerson,
2003; and Huynen et al., 2005). Together these frameworks identify a range of global
processes that have important health consequences.
These broader frameworks also identify important aspects of globalization
that shape the context within which trade liberalization is pursued, adopted and
responded to. Overall however, they focus on a limited range of liberalizing
pathways and fail to name specific mechanisms mediating the trade liberalization and
health relationship. However, recent work in this area has begun to account for a
wider range of liberalizing pathways, some by adopting a social determinants of
health perspective. Frameworks devised by Labonté and colleagues (2007) and
Blouin and colleagues (2009) are discussed in this thesis for pioneering work in this
latter regard.
Thus far, few researchers have focused on the exclusive health impact of trade
liberalization. Exceptions include Thow (2009), Rayner et al. (2007), Grown (2005)
and Blouin and colleagues (2009). These frameworks are important given that much
of the work which examines trade liberalization in conjunction with other global
processes fails to distinguish between pathways which originate from globalizing

285
processes more generally and trade liberalization specifically.
To date however, only Blouin and colleagues (2009) offer a framework which
seeks to expose the exclusive impact of trade liberalization on a variety of health
outcomes. Thow (2009), Rayner et al. (2007) and Grown (2005) by contrast, are
focused on specific health issues related to nutritional outcomes and reproductive
health.
Leaving aside the more widely studied impacts of trade on access to
medicines and health services, Blouin and colleagues (2009) identify four additional
pathways through which trade liberalization can impact health: income, inequality,
economic insecurity and unhealthy diets. One criticism that has been aimed at this
work however, is that in discussing these pathways the authors have not considered
the ways in which trade liberalization may impact positively on health (Paccaud,
2009). Moreover, the identified pathways seem under conceptualized. For example,
labour markets and wage differentials are of primary concern in the income
inequality pathway, although conceivably trade liberalization may impact income
inequality through other domains like food prices and government spending.
Additionally, while the important mediating role of social policies is emphasized by
the authors, a consideration of the impacts of trade liberalization on government
revenues, a pathway recognized elsewhere (Labonté & Schrecker, 2007), is also
absent.
This thesis aimed to provide a more comprehensive picture of the links
between trade liberalization and health by undertaking a narrative synthesis of
literature which explicitly identifies the ways in which trade liberalization may
impact health. With a particular focus on SDOH related pathways, this review
identifies important aspects of globalization that shape the context within which trade
liberalization is pursued and points to four main contexts through which liberalizing
processes may impact health: (1) increased flows of goods and people; (2)
agricultural and food trade (3) structural adjustment programs 9 and Poverty
Reduction Strategy Papers10 and (4) labour markets. In these contexts, researchers
identify a range of pathways between trade liberalization and health.
For example, in the context of increased flows of goods and people, trade

9
Structural Adjustment Programs are loan conditionalities imposed by either the IMF or World Bank.
10
Poverty Reduction Strategy Papers emerged in 1999 as a new requirement for countries to receive
grants or loans from the IMF or World Bank.

286
liberalization is associated with increases in communicable and non-communicable
diseases by increasing people’s exposure to infectious diseases and hazardous goods,
respectively. Additionally, a range of health outcomes is seen to result from an
increased flow of drugs, legal and illegal, due mostly to import liberalizing strategies.
In the context of agricultural and food trade, a variety of liberalizing processes link
trade liberalization to health, particularly in reference to food security, nutrition, and
food safety. In the context of SAPs, trade liberalization is viewed as a significant
factor mediating related pathways to health. For example, reductions in consumption
subsidies, a prominent component of SAPs, are noted for their negative impact on
nutrition and household income. Finally, in the context of labour markets, trade
liberalization is seen to impact health through a number of pathways related to
various employment and working conditions as well as economic and social
inequalities.
This literature review represents the first attempt at delineating the health
impacts of trade liberalization through a systematic narrative synthesis. While this
takes us some way towards answering the first research question of this thesis (i.e
what are the pathways and mechanisms mediating the trade liberalization and health
relationship?), two main weaknesses are that it offers 1) little in the way of empirical
evidence and 2) little in respect to the broader objective of identifying how trade
liberalization and social policy interact to influence health and its social determinants
(RO1). This first limitation relates to the fact that little empirical work has been
undertaken by the studies identified by the review. The second limitation relates to
the fact that included studies rarely acknowledge social policies as a health mediating
factor, except in the context of labour markets. Moreover, due to other weaknesses in
our understanding of the trade liberalization and health relationship, as will be
discussed below, even in this context it is difficult to construct a clear conception of
how social policies interact with liberalizing processes to influence health.
To begin with, this review finds that a dearth of specific liberalizing policies
has been explored across the four contexts identified above, but particularly in
relation to SAPs, PRSPs and labour markets. In these contexts, trade liberalization is
often discussed in broad, imprecise ways with the exception of studies which
consider for example, the health impact of reductions in consumption subsidies in the
context of SAPs (Cheru, 2002; Labonté & Schrecker, 2006) and export processing
zones (Corrigal et al. 2008; Muntaner et al., 2010) and reductions in export tariffs
287
(Grown, 2005) in the context of labour markets.
While one benefit of research synthesis is that it can identify such gaps in the
literature, another is that it can facilitate the creation of a delimited number of
frameworks from which to work. The value of which has been described by
population health researcher Barbara Starfield (2001) and by Labonté and Torgerson
(2003) in direct relation to trade and health work. Drawing on the various pathways
identified in the literature review, as well as social epidemiological theories in
general, this thesis presents an ‘adapted ecosocial framework’ (see section 3.5) which
outlines the major features of the trade liberalization and health relationship as
identified by studies included in the literature review. Ecosocial theory is used to
frame these features since it is found both to account for authors’ conceptualizations
of the relationship between trade liberalization and health and to provide a useful
means of advancing areas thus far under conceptualized. For example, framing
pathways to health in relation to ecosocial theory emphasizes that liberalizing
processes may have lagged effects on health and need to be considered in relation to
lifecourse theories of epidemiology. These two considerations are very much
underdeveloped in work undertaken thus far.
By incorporating more theoretically advanced notions of how sociopolitical
processes interact simultaneously across multi-levels, ecosocial theory can also free
current trade liberalization and health theorizations from the shackles of the ‘distal-
proximal divide’ (i.e. false assumptions that health is impacted in a linear fashion,
through determinants whose causal strength is determined by their ‘nearness’ to the
individuals or groups under consideration) (Krieger, 2008b). The adapted ecosocial
framework also highlights the importance of trade liberalization's impact on other
areas thus far under explored including its impact on welfare states (e.g. via
government revenues), governments' policy space, and the physical environment.
The aim of this adapted ecosocial framework is not to lay claim to a grand
theory but rather to present a tool for orienting deeper analyses into the relationships
between trade liberalization and health. A main weakness of the framework is that it
fails to clearly map out the processes and mechanisms linking the various elements
contained within it. However, because the framework was developed on the basis of
research still very much in its early stages, it is expected that as our knowledge about
trade liberalization develops so too will our ability to capture more thoroughly the
precise mechanisms and interactions at play.
288
It is also worth considering how critically engaging with the literature review
leaves us better equipped to deal with some of the complexities characterizing the
trade liberalization and health relationship and how in turn, these considerations
represent important contributions to the literature.
Aside from identifying the pathways between trade liberalization and health,
one of the main objectives of the literature review was to understand how researchers
both conceptualize and define trade liberalization. Informing this question was the
recognition that the ways in which trade liberalization are understood can both
conceal and illuminate pathways important for health. Drawing on influential work
by Lister (2004), this thesis regarded trade liberalizing concepts as the broad
meanings lying behind liberalizing processes, for example ‘market integration’ or
‘free trade’, as well as the larger contexts within which authors perceived these
processes as taking place, for example in reference to ‘globalization’ or
‘neoliberalism’. Definitions by contrast, were characterized as offering more specific
explications of trade liberalizing processes, such as a reduction in import tariffs, or a
reduction in consumption subsidies.
A major finding of the literature review was that researchers conceptualize
trade liberalization in a variety of (sometimes conflicting) ways. For example,
different authors vary in their identification of financial liberalization or foreign
investment as a subset of a larger trade liberalization agenda. While many authors
clearly position financial flows and foreign investment within conceptualizations of
trade liberalization (Borghesi & Vercelli 2003; Doyal, 2002; Grown, 2005; Labonté
& Torgerson, 2003; Muntaner et al., 2010; Polakoff, 2007; Smith & Signal, 2009;
Woodward et al., 2001), others seem to position these concepts in separate domains
(De Vogli & Birbeck, 2005; De Vogli et al., 2009).
Moreover, authors often conceptualize trade liberalization by referring to
abstract ideas like ‘openness’, ‘free trade’ or by proxy of increased trade flows.
Conceptualizing trade liberalization in these ways is problematic since the particular
role of trade liberalization in these notions is unclear. For example, openness can
broadly refer to deregulation policies or it can be used to indicate a country’s degree
of integration with global economic forces. This latter notion of openness however,
may depend on factors unrelated to trade liberalization, such as countries’ natural
resource endowments (Subasat, 2008). Moreover, authors using trade flows as a
proxy for trade liberalization confuse the processes of trade liberalization with its
289
presumed outcomes. This is especially problematic given the recognition that trade
liberalization does not inevitably lead to increased trade flows (Rodrik, 1999;
Rodriguez & Rodrik, 2000).
In synthesising researchers’ different understandings of trade liberalization,
this thesis underscores that trade liberalization is a multi-faceted concept shaped by
the ideology of neoliberalism, characterized by unequal power relations, and
promoted by powerful interests through a range of institutions, trade agreements and
policies in the domains of goods, services and investment. This understanding of
trade liberalization provides direction for mapping out a wide range of liberalizing
processes and their combined impact on health. Moreover, by drawing attention to
the political contexts and actors shaping the context within which liberalization
policies are pursued, it also directs attention to how well the policy implications of
research will be adopted.
This review also found that conceptualizations of trade liberalization are often
presented without an explicit definition of what exactly the notion of trade
liberalization entails. This is problematic for two reasons. First, explicit definitions of
trade liberalization are needed for the obvious reason of making clear researchers’
theorized pathways to health. Second, a common understanding of trade
liberalization is especially needed if we want to comprehensively explore the health
impact of trade liberalizing processes and draw more informed conclusions about
their combined impact on health. The need for common definitions in health research
is strongly advocated for by Starfield (2001).
To further consider the problem of defining trade liberalization, this thesis
turned to development economics, a field in which debates about the merits of trade
liberalization have been ongoing for decades. A brief review of this work finds that
trade liberalization should be distinguished from notions of market integration and
other overall measures of trade flows. Moreover, two considerations are found to be
important in identifying an instance of trade liberalization. The first relates to
recognizing that trade liberalization is a process. The second relates to understanding
liberalization policies as a move toward a more lassiez-faire economic environment
(i.e. one in which government intervention is reduced).
With these considerations in mind this thesis presents the following working
definition of trade liberalization for public health researchers: Trade liberalization is
the process of reducing government intervention in matters of trade.
290
Characterizing trade liberalization in this way prevents the confounding of
liberalizing policies for processes which may increase a country’s market integration;
like export promotion strategies. It also ensures that liberalizing policies are not
equated to increased trade flows or other measures which can conflate trade
liberalizing policies for their presumed outcomes.
This definition also suggests potential methods for measuring instances of
trade liberalization. For instance, trade liberalization may occur by way reducing
government intervention in terms of imports (e.g. by reducing import tariffs), or by
way of reducing government intervention in terms of exports (e.g. by reducing export
subsidies). Importantly, this definition also points to ways trade liberalization should
not be measured, for example via changes in trade flows (e.g. by reference to the
value of imports and exports in relation to a country’s GDP). Ultimately,
understanding trade liberalization in this way will allow public health researchers to
better understand and study the relevant pathways to health via more well-defined
research boundaries.

10.3 Moving from the Theoretical to the Empirical

Following the literature review, this thesis moved to empirically explore a


specific instance of trade liberalization (i.e. the phase-out of the Multi-Fibre
Arrangement in 2005) and its impact on health via changes in the labour market. This
move responded to the limitations mentioned previously of the literature review and
adapted ecosocial framework, namely lacking empirical evidence, the fact that
liberalization processes are especially under-conceptualized in relation to labour
markets and that the mediating role of social policies were largely identified in the
context of labour markets. This latter issue was important since the initial research
objective of this thesis was to identify how trade liberalization and social policy
interact to influence health and its social determinants.

Other considerations raised in the literature review also played an important


role in the empirical work of this thesis. For example, the working definition of trade
liberalization discussed above aided in the selection of a trade liberalizing episode.
Moreover, the ecosocial framework guided important methodological considerations
(as discussed in chapter 5 and mentioned briefly again in the next section).

291
With this case study in mind the second research objective of this thesis was
to investigate and analyse how the phase-out of the Multi-Fibre Arrangement
impacted health in countries reliant on the textile and clothing sector for employment
(RO2). The concomitant research questions were as follows:

How did health outcomes change after the phase-out of the Multi-Fibre in
countries reliant on the textile and clothing sector? (RQ2)

What are the potential causal mechanisms responsible for these changes?
(RQ3)

The next section of this chapter will discuss the main progress towards this research
objective and associated questions, as well as related contributions to the literature.
Following this will be a consideration of how these results then link back to the
literature review, the adapted ecosocial framework, and the first research objective of
this thesis. Finally a consideration of where we are left in terms of next steps will be
presented.

10.4 The Phase-Out of the Multi-Fibre Arrangement and its Impact


on Health Outcomes in Countries Reliant on the Textile and
Clothing Sector
This chapter will now turn to the contributions this thesis makes to the
literature via its investigation of the health impacts of the MFA phase-out. To begin
with, the thesis applied an original and innovative method in analyzing how trade
and social policies interact to influence health: fuzzy-set qualitative comparative
analysis (fsQCA) combined with process tracing methods. To date the large majority
of public health scholarship which investigates macro-level conditions has relied on
conventional multivariate regression models. These linear models average evidence
across cases and isolate each explanatory variable by keeping all others in the
analysis constant. While providing important information about what variables
maximize health outcomes, these techniques neglect to take sufficient account of the
contextual dimensions of cases and fail to address the interactive effects of causal
pathways. This issue is especially important given considerations raised during the
literature review, and captured in the adapted ecosocial framework, which emphasize
that the relationship between trade liberalization and health is characterized by a
complex set of concurrent and interacting pathways. In utilizing fsQCA, this thesis

292
has demonstrated the merits of an approach which assumes causal complexity and
requires that a model be specified whereby independent variables are assumed to
combine in a way which responds to the unique contextual environments of cases.

This case study work also offers some of the first empirical evidence of how
trade liberalization interacts with social policies in the context of labour markets to
influence health. It terms of answering the second and third research questions of this
work, it is found that after the MFA phase-out only less developed countries
experienced T&C employment growth. However, many less developed countries also
experienced employment loss. In less developed countries which experienced T&C
employment growth after the MFA phase-out, it is found that the worsening of adult
female mortality rates seems to be related to an absence of social protection policies
in combination with poor working conditions, which were likely to have worsened
after the phase-out.

In highly developed countries which experienced employment loss after the


MFA phase-out, it is found that the worsening of adult female mortality rates seems
to be related to T&C workers’ inability to access social protections, as well as few
alternative work opportunities. This can be contrasted with a country like Portugal,
which did not experience a worsening of adult female mortality rates. Evidence
suggests that in Portugal, T&C work is characterized by a more standard, full-time
employment relationship. As such, Portuguese T&C workers losing their
employment after the MFA phase-out were likely to be covered by the country’s
collective dismissal regulations. These regulations would have required employers to
give workers advance notice of dismissals and time off to look for alternative work.
Evidence suggests that dismissed Portuguese T&C workers would have also been
entitled to a minimum severance pay and possible re-training opportunities or early
retirement.

However, there are noteworthy limitations to these above results. To begin


with, the results are quite tentative in nature. As discussed in Chapter 9, this relates
to data availability issues as well as broader methodological issues.A main limitation
concerns the embeddedness of the case studies within a larger cross-case analysis.
This relates less to the inherent nature of combining fsQCA with case studies than to
the process tracing results themselves. Overall, T&C workers were found to have
little access to social protection regardless of how countries were characterized in the

293
fsQCA. This means that the mechanisms which emerged were not associated with
the original conditions of the fsQCA and that we are thus both limited in making
claims about sufficiency and in generalizing these results to the situation of other
countries characterized by similar solution paths
The precise causal mechanisms behind these results are also unclear. The
EMCONET framework identifies a range of different pathways which may be
influencing these results such as those related to material deprivation, health
behaviours, physio-pathological changes and psychosocial factors. However, data on
these more micro-level conditions were not uncovered in this work.
Moreover, while 53 countries were identified as reliant on the T&C sector,
only 32 were included in the analyses. Inclusion of countries was limited by the
quality of mortality data sources, as well as by the availability of data related to
social protections. While excluded countries were comprised of both highly
developed and less developed countries, it is unclear how their inclusion might have
impacted the results of the analyses. This work thus reiterates calls for better quality
cross-national health and social policy data.
Although there was a theoretical basis for doing so, by focusing on pathways
to health in the context of labour markets, this work also neglects other potential
health impacts of the MFA phase-out. It is likely for example, that the phase-out also
had impacts on the physical environment through changes in industrial pollution with
possible health consequences. In fact, changes in the physical environment arose as a
potential health determining mechanism in some of the cases investigated in greater
depth.
This work also focused on changes in national level health, but in some
countries many T&C workers are migrants and thus not represented in national
health figures (in Italy for example, there is a large population of Chinese T&C
workers, in Thailand there is a large population of Burmese T&C workers). For these
reasons, the health impacts of the MFA phase-out are likely underestimated.
Finally, this work is unable to offer much insight about countries which
experienced improving adult mortality rates after the MFA phase-out. It is also
unable to offer much insight about changing infant mortality rates after the MFA
phase out. This is either because evidence supporting the sufficient relationships
found in the fsQCA could not be fleshed out in the case studies or because these
outcomes were not found to be sufficient with the conditions included in the fsQCA.
294
Despite these limitations, there are two important considerations to keep in
mind. First is that the results of the analyses do point to the health importance of
social protection policies and as such are aligned with previous comparative welfare
state studies (Beckfield and Krieger, 2009). This relates not only to the causal
mechanisms that emerged from this research, but also to case study work undertaken
in relation to Brazil where evidence relates reductions in poverty, inequality and
child mortality to an expansive conditional cash transfer programme. Despite the fact
that a causal mechanism could not be constructed here, this evidence also indicates
the health importance of protective social policies.
Second is that the fsQCA was undertaken as an exploratory analysis and
process tracing was undertaken with the objective of identifying plausible causal
mechanisms that could be tested empirically in subsequent research. As discussed in
Chapter 9, these analyses do leave us better positioned to further investigate the
impacts of the MFA phase-out. Process tracing results for example, uncovered a
range of considerations, which if accounted for, can improve the validity of future
cross-case analyses.
Moreover, aside from shedding causal (albeit partial) light on the relationship
between the MFA phase-out and countries which experienced a worsening of adult
female mortality rates, combining fsQCA with process tracing also shed meaningful
light on broader considerations related to the health impact of the MFA phase-out,
and on the relationship between trade liberalization and health more generally. This
is described in Chapter 9 and reviewed again the next section.

10.5 Moving from the Empirical back to the Theoretical


In the context of the MFA phase-out, trade liberalization was found to
increase competition within the sector via an intensification of price pressures. This
competition in turn, impacted T&C labour markets both in terms of shifting
employment and worsening working conditions. Social protection policies were
found to interact with these considerations, and in response to the employment and
working conditions of the sector in general, in many health important ways. For
example, factors such as the size of T&C firms, the types of labour contracts through
which workers were employed, and the types of labour overall which characterizes
the sector (e.g. formal versus informal, legal versus illegal), were found to determine
T&C workers’ access to protective labour market and welfare state policies. These

295
employment and working conditions often in turn, relate back to countries’ labour
regulations and also respond to the competitive pressures of the sector.
In consideration of the first research objective of this thesis the following can
be offered: social protection policies were found to both moderate pathways to health
in the context of labour markets and influence the type of health related pathways
resulting from trade liberalizing policies. Overall, the empirical findings suggest that
social protection is often inaccessible to the type of workers who may be the most
vulnerable to processes of liberalization, and that many workers are particularly
vulnerable to processes of liberalization due to the structure of labour market and
social protection policies. Perhaps surprisingly, this seems to be the case in both
developed and less developed countries. In fact, social protection for T&C workers
seemed to be greater in some less developed than more developed countries.
The results of the case study work also reemphasize considerations raised in
the literature review. For example, that liberalizing processes in one part of the world
can have important health implications in another. Also emphasized is one of the
main considerations of the adapted ecosocial framework: that pathways from trade
liberalization to health consist of complex, concurrent and interacting pathways. This
was exemplified in the case work of Bangladesh where the MFA phase-out was
found to impact female T&C workers in potentially contradictory ways: by enabling
both improvements in material circumstances (and personal autonomy) as well as
greater exposure to dangerous working conditions.

Findings from the empirical work do not however, suggest any adjustments to
the adapted ecosocial framework, although they do underscore the difficulty in
empirically considering the multiple dimensions and considerations which the
framework highlights. For example, while the fsQCA analysis incorporated lagged
health impacts, how the phase-out might fit into lifecourse theories of health was
something this thesis was unable to explore. But this point perhaps drives home one
of the main intentions and strengths of the adapted ecosocial framework which is that
it can be used for initiating and thinking more in-depth about the full range of
potential pathways to health originating from liberalizing processes.

The next and final section of this chapter will further consider where the
findings of this thesis leave us in terms of next steps.

296
10.6 Next Steps
As previously discussed, the results of this thesis provide a basis for further
delving into the relationship between the MFA phase-out and health. They also have
important implications for undertaking novel work in related areas of public health
and social research.

In this thesis, trade liberalization was examined as a process of quota removal.


While the gradual liberalization of world markets means that quotas are now a
relatively rare phenomenon, trade agreements may contain other processes of
liberalization with important labour market impacts. This is important since much of
the research which examines the health impacts of trade agreements is focused on
biomedical and behavioral/lifestyle factors, related for instance to health care, diet,
smoking and alcohol consumption.
For example, the Trans-Pacific Partnership Agreement (TPP) is a proposed
trade agreement which has been under negotiation since 2010. Negotiations are
currently taking place between 12 countries: Australia, Brunei, Chile, Canada, Japan,
Malaysia, Mexico, New Zealand, Peru, Singapore, the United States and Vietnam.
Several studies have directed attention to the potential impact of the TPP on health.
Gleeson and Friel (2013) argue that the TPP will impact health largely through two
main pathways. First, it is argued that the TPP will reduce access to medicines by
increasing protections on intellectual property rights. Second, it is noted that the
Agreement’s clauses related to investor–state dispute settlement provisions ‘could
restrict the ability of governments to regulate industries that produce goods that
contribute to the growing burden of non-communicable diseases, such as tobacco,
alcohol, and highly processed foods’ (p. 1508). Other work has also pointed the
potential health impacts of the Agreement in relation to tobacco (Fooks & Gilmore,
2013; Kelsey, 2012, 2013; Pattemore, 2013), alcohol (Kelsey, 2012) and unhealthy
foods (Friel et al., 2013).
However, work has yet to explore the health impacts of the TPP via its impact
on labour markets. Indeed a particularly contentious issue of the TPP relates to the
Agreement’s impact on T&C labour markets (Bradner, 2013; Elms, 2012; Platzer,
2013). One industry source reports that the TPP could displace as many as 22,000
workers in El Salvador’s textile industry (and another 15,000 jobs indirectly)
(Fibre2fashion, 2013a). The textile and clothing sector in Vietnam, by contrast, is

297
expected to gain T&C employment as a result of the Agreement (Fibre2fashion,
2013b).
Another area of research relevant to the findings of this thesis is the body of
literature focused more broadly on corporations and health. This literature is largely
concerned with how corporate actors affect health through practices such as political
lobbying, production techniques, product marketing, retail distribution, and pricing.
Much of the work on corporations and health focuses its attention on six main
industries: those related to alcohol, automobiles, firearms, food and beverages,
pharmaceutical products, and tobacco (Freudenberg, 2014). While work in these
areas is focused on the direct health impacts of these goods, the results of this thesis
demonstrate an additional pathway through which corporate actors may influence
health: labour markets. Taking methodological cues from this literature can thus
facilitate new and important lines of inquiry that broaden our understanding of how
corporate actors in the T&C sector, and indeed other sectors vulnerable to trade
liberalization and poor labour market conditions, might impact health.
For example, this thesis was largely concerned with documenting the health
consequences of the MFA phase-out. While this required investigation of important
social and political determinants of health, such as labour market and welfare state
policies, much needs to be done in terms of critically analysing the power relations
shaping these and other relevant factors. For example, the power imbalances between
different types of firms within the sector’s value chain, as well as those between
employees and employers.
Without acknowledging these types of power relations, the findings of this
thesis, seem to direct exclusive attention to a need for greater availability of health
promoting resources like unemployment insurance, higher wages, or better working
conditions. However, investigating the sector from a more critical perspective, by
focusing on corporate actors, can additionally direct attention to interventions that
address exploitative and exclusionary processes, such as those aimed at strengthening
the basis from which T&C workers can collectively bargain for greater labour and
social protections. This critical perspective aligns with the conceptualization of trade
liberalization which arose out of the literature review with its emphasis on the role of
neoliberal ideology, unequal power relations, and the undue influence of powerful
interests.

298
BIBLIOGRAPHY
Absar, S. S. (2001). Problems surrounding wages: the ready made garments sector in

Bangladesh. Labour and Management in Development Journal, 2(7).

Absar, S. S. (2003). Health Hazards and Labour Laws in Bangladesh: A Narrative-based Study

on Women Garment Workers. Asian Journal of Social Science, 31(3), 452–477.

doi:10.1163/156853103322895342

Adhikari, R., & Yamamoto, Y. (2008). The textile and clothing Industry: Adjusting to the

post-quota world. In D. O’Connor & M. Kjollerstrom (Eds.), Industrial Development

for the 21st Century. London: Zed Books Ltd.

Ahmed, N. (2006). Bangladesh apparel industry and its workers in a changing world

economy (PhD thesis). Netherlands: Wageningen University. Retrieved from

http://edepot.wur.nl/121787

Ahmed, N. (2009). Sustaining Ready-made Garment Exports from Bangladesh. Journal of

Contemporary Asia, 39(4), 597–618. doi:10.1080/00472330903076891

Ahmed, N., & Peerlings, J. H. M. (2009). Addressing Workers’ Rights in the Textile and

Apparel Industries: Consequences for the Bangladesh Economy. World

Development, 37(3), 661–675. doi:10.1016/j.worlddev.2008.06.003

Amenta, E., & Poulsen, J. D. (1994). Where to Begin A Survey of Five Approaches to

Selecting Independent Variables for Qualitative Comparative Analysis. Sociological

Methods & Research, 23(1), 22–53. doi:10.1177/0049124194023001002

Arestis, P., & Caner, A. (2004). Financial Liberalization and Poverty: Channels of Influence

(Working Paper No. 411). NY: The Levy Economics Institute. Retrieved from

http://papers.ssrn.com/abstract=569663

Arnesen, T., & Nord, E. (1999). The value of DALY life: problems with ethics and validity of

disability adjusted life years. BMJ : British Medical Journal, 319(7222), 1423–1425.

299
Arnold, D. (2007). Capital Expansion and Migrant Workers Flexible Labor in the Thai - Burma

Border Economy (HUMAN RIGHTS IN ASIA SERIES). Bangkok, Thailand: Mahidol

University. Retrieved from http://burmalibrary.org/docs4/Arnold.pdf

Arunatilake, N. (2013). Precarious Work in Sri Lanka. American Behavioral Scientist, 57(4),

488–506. doi:10.1177/0002764212466246

Asheim, B. T. (1996). Industrial districts as “learning regions”: A condition for prosperity.

European Planning Studies, 4(4), 379–400. doi:10.1080/09654319608720354

Asian Development Bank. (2012a). People’s Republic of Bangladesh: Updating and

Improving the Social Protection Index (Technical Assistance Consultant’s Report:).

Asian Development Bank.

Asian Development Bank. (2012b). The Democratic Socialist Republic of Sri Lanka: Updating

and Improving the Social Protection Index (Technical Assistance Consultant’s

Report:). Asian Development Bank.

Asian Development Bank. (2013). The Social Protection Index Assessing Results for Asia and

the Pacific. Philippines.

Atkinson, J. (2007). International NGOs and Southern Advocacy: Case studies of two Oxfam

campaigns in South Asia. London School of Economics and Political Science.

Retrieved from

http://www.lse.ac.uk/internationalDevelopment/research/NGPA/fellowships/pract

ionerfellowships/pdfs/Oxfam_Atkinson_Report.pdf

Attanapola, C. T. (2004). Changing gender roles and health impacts among female workers

in export-processing industries in Sri Lanka. Social Science & Medicine, 58(11),

2301–2312. doi:10.1016/j.socscimed.2003.08.022

Audet, D. (2004). Structural Adjustment in Textiles and Clothing the Post-ATC Trading

Environment (OECD Trade Policy Working Paper No. 4). Organisation for Economic

Co-operation and Development.

300
Bakht, Z., Yunnus, M., & Salimullah, M. D. (2002). Machinery Industry in Bangladesh (IDEAS

Machinery Industry Study Report No. 4). Tokyo Institute of Development

Economies Advanced School.

Balassa, B. A. (1971). The Structure of Protection in Developing Countries. Published for the

International Bank for Reconstruction and Development and the Inter-American

Development Bank by the Johns Hopkins Press.

Bambra, C. (2006). Health Status and the Worlds of Welfare. Social Policy and Society, 5(01),

53–62. doi:10.1017/S1474746405002721

Bambra, C. (2011). Work, Worklessness, and the Political Economy of Health by Bambra,

Clare. OUP, USA,.

Bambra, C., & Eikemo, T. A. (2009). Welfare state regimes, unemployment and health: a

comparative study of the relationship between unemployment and self-reported

health in 23 European countries. Journal of Epidemiology and Community Health,

63(2), 92–98. doi:10.1136/jech.2008.077354

Banuri, T. (1991). Economic Liberalization: No Panacea: The Experiences of Latin America

and Asia. Clarendon Press.

Barancová, H. (2006). EU Adhesion of the Slovak Republic and the Development of

Employment Legislation. Transition Studies Review, 13(1), 9–12.

doi:10.1007/s11300-006-0078-z

Barancová, H., & Olšovská, A. (2011). Labour Law in Slovak Republic. Kluwer Law

International.

Barber, C., Gowthaman, B., & Rose, J. (2004). Stitched Up: How rich country protectionism in

textiles and clothing trade prevents poverty alleviation | Oxfam GB (Oxfam Briefing

Papers) (p. 44). Oxfam International. Retrieved from http://policy-

practice.oxfam.org.uk/publications/stitched-up-how-rich-country-protectionism-in-

textiles-and-clothing-trade-preve-114604

301
Barrie, L. (2009, December 4). Sri Lanka apparel firms bullish over GSP+. Just Style.

Retrieved February 11, 2014, from http://www.just-style.com/news/apparel-firms-

in-bullish-mood-over-GSP_Id106116.aspx

Barrie, L. (2011, September 20). BRAZIL: Zara owner ramps up supply chain controls. Just

Style. Retrieved February 17, 2014, from http://www.just-style.com/news/zara-

owner-ramps-up-supply-chain-controls_id112218.aspx

Barrientos, S., & Howell, J. (2006). The ETI code of labour practice: Do workers really benefit?

China (Report on the ETI Impact Assessment No. 2f). Institute of Development

Studies - University of Sussex.

Bartley, M., Ferrie, J., & Montgomery, S. F. (2006). Health and labour market disadvantage:

unemployment, non-employment, and job insecurity. In Social Determinants of

Health (pp. 78–96). New York: Oxford University Press. Retrieved from

http://www.oxfordscholarship.com/view/10.1093/acprof:oso/9780198565895.001.

0001/acprof-9780198565895-chapter-05

Bartley, M., & Plewis, I. (2002). Accumulated labour market disadvantage and limiting long-

term illness: data from the 1971–1991 Office for National Statistics’ Longitudinal

Study. International Journal of Epidemiology, 31(2), 336–341.

doi:10.1093/ije/31.2.336

Bastia, T., & McGrath, S. (2011). Temporality, Migration and Unfree Labour: Migrant

Garment Workers (Working Paper No. 6). The University of Manchester Centre for

the Study of Political Economy. Retrieved from

http://www.socialsciences.manchester.ac.uk/PEI/publications/wp/documents/Bast

iaamdMcGrathunfreepaper.pdf

Beach, D., & Pedersen, R. B. (2012). Process-Tracing Methods: Foundations and Guidelines.

Ann Arbor: The University of Michigan Press.

302
Beaglehole, R., & Yach, D. (2003). Globalisation and the prevention and control of non-

communicable disease: the neglected chronic diseases of adults. Lancet, 362(9387),

903–908. doi:10.1016/S0140-6736(03)14335-8

Beckfield, J., & Krieger, N. (2009). Epi + demos + cracy: Linking Political Systems and

Priorities to the Magnitude of Health Inequities—Evidence, Gaps, and a Research

Agenda. Epidemiologic Reviews, 31(1), 152–177. doi:10.1093/epirev/mxp002

Benach, J., Muntaner, C., & Santana, V. (2007). Employment conditions and health

inequalities (Final report to the WHO Commission on Social Determinants of Health

(CSDH) Employment Conditions Knowledge Network (EMCONET)).

Bennett, A., & Checkel, J. (Forthcoming). Process Tracing: From Philosophical Roots to Best

Practices. In Process Tracing: From Metaphor to Analytic Tool. Cambridge:

Cambridge University Press. Retrieved from

http://www.sfu.ca/content/dam/sfu/internationalstudies/checkel/PT-MS-

Chapter1-1213.pdf

Berg, A., & Krueger, A. (2002). Lifting All Boats: Why Openness Helps Curb Poverty. Finance

and Development, A Quarterly Magazine of the International Monetary Fund, 39(3),

16–19.

Berik, G. (2008). Growth with Gender Inequity: Another Look at East Asian Development. In

G. Berik, Y. Rodgers, & A. Zammit (Eds.), Social Justice and Gender Equality:

Rethinking Development Strategies and Macroeconomic Policies. New York and

London: Routledge. Retrieved from

http://www.unrisd.org/80256B3C005BCCF9/(httpAuxPages)/62428D42EFAD8BDEC

12578D50054B60A/$file/10Berik.pdf

Berik, G., & Rodgers, Y. V. D. M. (2010). Options for enforcing labour standards: Lessons

from Bangladesh And Cambodia. Journal of International Development, 22, 56–85.

doi:10.1002/jid.1534

303
Bernabè, S., & Kolev, A. (2003). Identifying vulnerable groups in the Kyrgyz labour market:

some implications for the national poverty reduction strategy (CASEpaper No. 71).

London: Centre for Analysis of Social Exclusion, London School of Economics.

Retrieved from http://eprints.lse.ac.uk/6334/

Bettcher, D. W., Yach, D., & Guindon, G. E. (2000). Global trade and health: key linkages and

future challenges. Bulletin of the World Health Organization, 78(4), 521–534.

Bezruchka, S. (2012). The hurrider I go the behinder I get: the deteriorating international

ranking of U.S. health status. Annual Review of Public Health, 33, 157–173.

doi:10.1146/annurev-publhealth-031811-124649

Bhagwati, J. N. (1978). Anatomy and consequences of exchange control regimes. Ballinger

Pub. Co. for NEBR.

Birkman, L., Kaloshkina, M., Khan, M., Shavurov, U., & Smallhouse, S. (2012). Textile and

Apparel Cluster in Kyrgyzstan. Harvard Kennedy School; Harvard Business School.

Birn, A.-E., Pillay, Y., & Holtz, T. H. (2009). Textbook of International Health: Global Health in

a Dynamic World (3 edition.). New York: Oxford University Press.

Block, R., N. (2005). Indicators of labour standards: An overview and comparison (Working

Paper No. 54). International Labor Organization. Retrieved from

http://www.ilo.org/integration/resources/papers/WCMS_079143/lang--

en/index.htm

Blouin, C., Chopra, M., & van der Hoeven, R. (2009). Trade and social determinants of

health. The Lancet, 373(9662), 502–507. doi:10.1016/S0140-6736(08)61777-8

Borghesi, S., & Vercelli, A. (2003). Globalisation, Inequality and Health (Department of

Economics University of Siena No. 413). Department of Economics, University of

Siena. Retrieved from http://ideas.repec.org/p/usi/wpaper/413.html

304
Bos, E., Vu, M. T., & Stephens, P. W. (1992). Sources of W\orld Bank Estimates of Current

Mortality Rates (Policy Reserach Working Papers Population, Health, and Nutition

No. 0851). Washington, D.C.: World Bank.

Bourguignon, F., & Goh, C. (2003). Trade and Labor Market Vulnerability in Indonesia, Korea,

and Thailand. In K. Krumm & H. Kharas (Eds.), East Asia integrates: a trade policy

agenda for shared growth. The World Bank. Retrieved from

http://siteresources.worldbank.org/INTEAPREGTOPINTECOTRA/Resources/chapter

%2B10.pdf

Bovet, P., & Paccaud, F. (2009). Negative and positive effects of trade on health. The Lancet,

373(9672), 1338. doi:10.1016/S0140-6736(09)60770-4

Bradner, E. (2013, September 19). Two days, 240 meetings: Apparel pushes free trade on

Hill. Politico. Retrieved February 4, 2014, from

http://www.politico.com/story/2013/09/apparel-trade-meetings-capitol-hill-

97092.html

Braveman, P. (2006). Health disparities and health equity: concepts and measurement.

Annual Review of Public Health, 27, 167–194.

doi:10.1146/annurev.publhealth.27.021405.102103

Breman, A., & Shelton, C. (2007). Structural Adjustment Programs and Health. In I. Kawachi

& S. Wamala (Eds.), Globalization and Health. Oxford University Press.

Brenner, H. (1995). Political economy and health. In B. C. Amick & A. R. Tarlov (Eds.), Society

and Health (First Edition, First Printing edition., pp. 211–246). New York: Oxford

University Press.

Brugiavini, A., & Peracchi. (2012). Health Status, Welfare Programs Participation and Labor

Force Activity in Italy. In D. A. Wise (Ed.), Social Security Programs and Retirement

Around the World: Historical Trends in Mortality and Health, Employment, and

305
Disability Insurance Participation and Reforms (pp. 175–215). University of Chicago

Press.

Burgard, S. A., Ailshire, J. A., & Kalousova, L. (2013). The Great Recession and Health |

People, Populations, and Disparities. The ANNALS of the American Academy of

Political and Social Science, 650(1), 194–213. doi:10.1177/0002716213500212

Camuffo, A., Pozzana, R., Vinelli, A., & Benedetti, L. (2008). Not Doomed to Death: A Map of

Small Firms’ Business Models in the Italian Textile Apparel Industry (Industry Studies

Working Paper Series No. 94). Retrieved from http://isapapers.pitt.edu/94/

Carpiano, R. M., & Daley, D. M. (2006). A guide and glossary on postpositivist theory

building for population health. Journal of Epidemiology and Community Health,

60(7), 564–570. doi:10.1136/jech.2004.031534

Catalano, R., Goldman-Mellor, S., Saxton, K., Margerison-Zilko, C., Subbaraman, M., LeWinn,

K., & Anderson, E. (2011). THE HEALTH EFFECTS OF ECONOMIC DECLINE. Annual

Review of Public Health, 32. doi:10.1146/annurev-publhealth-031210-101146

Cerami, A. (2010). The Politics of Social Security Reforms in the Czech Republic, Hungary,

Poland, and Slovakia. In B. Palier (Ed.), A Long Goodbye to Bismarck?: The Politics of

Welfare Reforms in Continental Europe (pp. 233–253). Amsterdam: Amsterdam

University Press.

Chan, K. W., & Buckingham, W. (2008). Is China Abolishing the Hukou System? The China

Quarterly, 195, 582–606. doi:10.1017/S0305741008000787

Cheru, F. (2002). Debt, adjustment and the politics of effective response to HIV/AIDS in

Africa. Third World Quarterly, 23(2), 299–312. doi:10.1080/01436590220126658

China Labour Watch. (2011). Investigation of Two Clothing and Apparel Factories in China:

Excessive Overtime, Student Workers, and Exploitative Wage Systems. Retrieved

from

306
http://digitalcommons.ilr.cornell.edu/cgi/viewcontent.cgi?article=2167&context=gl

obaldocs

Cho, H.-Y., Cho, H., Surendra, L., & Cho, H. (2013). Contemporary South Korean Society: A

Critical Perspective. Routledge.

Cho, J., Kwon, T., & Ahn, J. (2010). Half success, half failure in Korean Affirmative Action: An

empirical evaluation on corporate progress. Women’s Studies International Forum,

33(3), 264–273. doi:10.1016/j.wsif.2010.02.020

Chongsuvivatwong, V., Phua, K. H., Yap, M. T., Pocock, N. S., Hashim, J. H., Chhem, R., …

Lopez, A. D. (2011). Health and health-care systems in southeast Asia: diversity and

transitions. The Lancet, 377(9763), 429–437. doi:10.1016/S0140-6736(10)61507-3

Chuang, K.-Y., Sung, P.-W., Chang, C.-J., & Chuang, Y.-C. (2013). Political and economic

characteristics as moderators of the relationship between health services and

infant mortality in less-developed countries. Journal of Epidemiology and

Community Health, 67(12), 1006–1012. doi:10.1136/jech-2013-202685

Chun, J. (2006). The Contested Politics of Gender and Employment: Revitalizing the South

Korean Labor Movement. In D. Pillay, I. Lindberg, & A. Bieler (Eds.), Draft study for

‘Global Working Class Project.

Ciccarone, G. (2011). EEO Review: Adapting unemployment benefit systems to the

economic cycle, 2011 (p. 10). European Employment Observatory. Retrieved from

http://www.eu-employment-observatory.net/resources/reviews/Italy-

UBRvw2011.pdf

Cooke, F. L. (2010). Women’s participation in employment in Asia: a comparative analysis of

China, India, Japan and South Korea. The International Journal of Human Resource

Management, 21(12), 2249–2270. doi:10.1080/09585192.2010.509627

307
Cooper, R. N. (1987). Round Table Discussion. In V. Corbo, M. Goldstein, & M. S. Khan (Eds.),

Growth-oriented adjustment programs: Proceedings of a Symposium Held in

Washington, D.C. (pp. 516–523). Washington, D.C.: International Monetary Fund.

Cornia, G. A. (2001). Globalization and health: results and options. Bulletin of the World

Health Organization, 79(9), 834–841.

Corrigall, J., Plagerson, S., Lund, C., & Myers, J. (2008). Global Trade and Mental Health.

Global Social Policy, 8(3), 335–358. doi:10.1177/1468018108095632

Cristovam, M. L. (2001, April 28). Restructuring in multinationals hits employment in

Portugal. European Industrial Relations Obsevatory On-line. Retrieved June 4, 2014,

from http://www.eurofound.europa.eu/eiro/2001/04/feature/pt0104144f.htm

Cristovam, M. L. (2006). RESTRUCTURING CASE STUDY THE TEXTILE AND CLOTHING

INDUSTRY IN PORTUGAL. In Restructuring in Europe: The Anticipation of negative

labour market effects Conclusions from Case Studies on the labour market impact of

large-scale economic restructuring - Compilation of case studies in full length.

European Commission, Directorate-General for Employment, Social Affairs and

Equal Opportunities. Retrieved from

https://www.google.co.uk/url?sa=t&rct=j&q=&esrc=s&source=web&cd=1&cad=rja

&uact=8&ved=0CCgQFjAA&url=https%3A%2F%2Ffanyv88.com%3A443%2Fhttp%2Fec.europa.eu%2Fsocial%2FBlobSer

vlet%3FdocId%3D3674%26langId%3Den&ei=GEePU52VGIvo7AbgsoDgDA&usg=AFQ

jCNHn5hQpzz_5FEvXyHn-m65p7QaZbw&bvm=bv.68235269,d.ZGU

CSDOH. (2008). Closing the gap in a generation: health equity through action on the social

determinants of health. Final report of the commission on social determinants of

health. Geneva: World Health Organization.

Curran, L. (2008). Unpicking the Textiles Trade--The Impact of Liberalization on the Global

Textiles Sector. Global Trade and Customs Journal, 3(7), 261–274.

308
Cziria, L. (2013, July 25). Slovakia: The representativeness of trade unions and employer

associations in the textile and clothing sector. European Industrial Relations

Obsevatory On-line. Retrieved June 8, 2014, from

http://www.eurofound.europa.eu/eiro/studies/tn1302024s/sk1302029q.htm

Davin, D. (2001). The impact of export-oriented manufacturing on Chinese women workers.

United Nations Research Institute for Social Development (UNRISD). Retrieved from

http://www.unrisd.org/UNRISD/website/document.nsf/ab82a6805797760f80256b

4f005da1ab/23b9b6703e326969c1256bb800535363/$FILE/davin.pdf

De Vogli, R., & Birbeck, G. L. (2005). Potential impact of adjustment policies on vulnerability

of women and children to HIV/AIDS in sub-Saharan Africa. Journal of Health,

Population, and Nutrition, 23(2), 105–120.

De Vogli, R., Gimeno, D., & Mistry, R. (2009). The policies–inequality feedback and health:

the case of globalisation. Journal of Epidemiology and Community Health, 63(9),

688–691. doi:10.1136/jech.2008.081588

Deaton, A. (2011). Measuring Development: Different Data, Different Conclusions? In M.

Aglietta, S. Alkire, & F. Bourguignon (Eds.), Measure for Measure: How Well Do We

Measure Development ? : Proceedings of the 8th AFD-EUDN Conference, Paris,

December 1st, 2010. Agence française de développement. Retrieved from

http://www.princeton.edu/~deaton/downloads/deaton_different_data_different_

conclusions_paris_2010.pdf

Debels, A. (2004). Temporary jobs: Segmented security through the labour and welfare

system (Paper for TLM.Net Conference: Quality in Labour Market Transitions: A

European Challenge). Belgium: KULeuven Department of Sociology –.

Diaz-Bonilla, E., Babinard, J., & Pinstrup-Andersen, P. (2002). Opportunities and Risks for the

Poor in Developing Countries (Working Paper No. 83). New Delhi: Indian Council for

309
Research on International Economic Relations. Retrieved from

http://icrier.org/pdf/risk.pdf

Doyal, L. (2002). Putting gender into health and globalisation debates: New perspectives

and old challenges. Third World Quarterly, 23(2), 233–250.

doi:10.1080/01436590220126612

Doyal, L. (2005). Understanding Gender, Health and Globalization: Opportunities and

Challenges. In I. Kickbusch, K. Hartwig, & J. List (Eds.), Globalization, Women, and

Health in the 21st Century (pp. 9–28). New York: Palgrave Macmillan.

Dunford, M. (2006). Industrial Districts, Magic Circles, and the Restructuring of the Italian

Textiles and Clothing Chain. Economic Geography, 82(1), 27–59.

doi:10.1111/j.1944-8287.2006.tb00287.x

Edwards, S. (1989). Openness, Outward Orientation, Trade Liberalization, and Economic

Performance in Developing Countries (Policy, Planning and Research Working

Papers Trade Policy No. 191). Washington, D.C.: World Bank. Retrieved from

http://econ.worldbank.org/external/default/main?pagePK=64165259&theSitePK=4

69372&piPK=64165421&menuPK=64166322&entityID=000009265_396092800121

Edwards, S. (1993). Openness, trade liberalization, and growth in developing countries.

Journal of Economic Literature, 31(3), 1358–1393.

Elms, D. (2012). Getting from Here to There: Stitching Together Goods Agreements in the

Trans-Pacific Partnership (TPP) Agreement (Working Paper No. No. 235). Singapore:

S. Rajaratnam School of International Studies.

EMCC. (2003). (European Restructuring Monitor quarterly No. 3). European Monitoring

Centre on Change. Retrieved from

http://www.eurofound.europa.eu/emcc/erm/templates/displaydoc.php?docID=2

310
Esping-Andersen, G., & Regini, M. (2001). Why Deregulate Labour Markets? Oxford

University Press.

Euractiv. (2008, May 13). ICT comes to aid of unemployed Portuguese textile workers |

EurActiv. Euractiv.com. Retrieved October 16, 2013, from

http://www.euractiv.com/socialeurope/ict-comes-aid-unemployed-portugu-news-

219788

Eurofound. (2014). COLLECTIVE DISMISSAL: PORTUGAL. The European Foundation for the

Improvement of Living and Working Conditions (Eurofound). Retrieved June 4, 2014,

from http://www.eurofound.europa.eu/emire/PORTUGAL/COLLECTIVEDISMISSAL-

PT.htm

Feldbaum, H., Lee, K., & Michaud, J. (2010). Global Health and Foreign Policy. Epidemiologic

Reviews, 32(1), 82–92. doi:10.1093/epirev/mxq006

Feng, X. L., Theodoratou, E., Liu, L., Chan, K. Y., Hipgrave, D., Scherpbier, R., … Chopra, M.

(2012). Social, economic, political and health system and program determinants of

child mortality reduction in China between 1990 and 2006: a systematic analysis.

Journal of Global Health, 2(1). Retrieved from

http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3484751/

Ferenschild, S., & Wick, I. (2004). Global game for cuffs and collars the phase-out of the

WTO Agreement on Textiles and Clothing aggravates social divisions. Siegburg;

Neuwied: Südwind e.V. ; Ökumenisches Netz Rhein-Mosel-Saar. Retrieved from

http://catalog.hathitrust.org/api/volumes/oclc/76739910.html

Ferrarini, T., & Norström, T. (2010). Family policy, economic development and infant

mortality: a longitudinal comparative analysis. International Journal of Social

Welfare, 19, S89–S102. doi:10.1111/j.1468-2397.2010.00736.x

311
Ferrarini, T., & Sjöberg, O. (2010). Social policy and health: transition countries in a

comparative perspective. International Journal of Social Welfare, 19, S60–S88.

doi:10.1111/j.1468-2397.2010.00729.x

Fibre2fashion. (2013a, February 2). El Salvador textile sector may lose 22,000 jobs to TPP.

Fibre2fashion. Retrieved February 1, 2014, from

http://www.fibre2fashion.com/news/textile-

news/newsdetails.aspx?news_id=121475

Fibre2fashion. (2013b, November 21). TPP can strongly boost Vietnamese garment sector.

Fibre2fashion. Retrieved February 4, 2014, from

http://www.fibre2fashion.com/news/garment-apparel-

news/vietnam/newsdetails.aspx?news_id=156038

Fidler, D. P., Drager, N., & Lee, K. (2009). Managing the pursuit of health and wealth: the

key challenges. Lancet, 373(9660), 325–331. doi:10.1016/S0140-6736(08)61775-4

Fisher, S., Gould, J., & Haughton, T. (2007). Slovakia’s Neoliberal Turn. Europe-Asia Studies,

59(6), 977–998. doi:10.1080/09668130701489170

Fooks, G., & Gilmore, A. B. (2013). International trade law, plain packaging and tobacco

industry political activity: the Trans-Pacific Partnership. Tobacco Control,

tobaccocontrol–2012–050869. doi:10.1136/tobaccocontrol-2012-050869

Fox, A. M., & Meier, B. M. (2009). Health as freedom: addressing social determinants of

global health inequities through the human right to development. Bioethics, 23(2),

112–122. doi:10.1111/j.1467-8519.2008.00718.x

Frank, J., & Haw, S. (2011). Best Practice Guidelines for Monitoring Socioeconomic

Inequalities in Health Status: Lessons from Scotland. The Milbank Quarterly, 89(4),

658–693. doi:10.1111/j.1468-0009.2011.00646.x

Frederick, S. E. (2010). Development and Application of a Value Chain Research Approach to

Understand and Evaluate Internal and External Factors and Relationships Affecting

312
Economic Competitiveness in the Textile Value Chain (PhD dissertation). North

Carolina State University. Retrieved from

repository.lib.ncsu.edu/ir/bitstream/1840.16/6190/1/etd.pdf

Freudenberg, N. (2014). Lethal But Legal: Corporations, Consumption, and Protecting Public

Health. Oxford ; New York: OUP USA.

Friel, S., Gleeson, D., Thow, A.-M., Labonté, R., Stuckler, D., Kay, A., & Snowdon, W. (2013).

A new generation of trade policy: potential risks to diet-related health from the

trans pacific partnership agreement. Globalization and Health, 9(1), 46.

doi:10.1186/1744-8603-9-46

Friends of Nature, the Institute of Public & Environmental Affairs, Green Beagle,

Environmental Protection Commonwealth Association and Nanjing Green Stone

Environmental Action Network. (2012). Cleaning up the Fashion Industry (Green

Choice Apparel Supply Chain Investigation – Draft Report).

Gillings, D., Makuc, D., & Siegel, E. (1981). Analysis of interrupted time series mortality

trends: an example to evaluate regionalized perinatal care. American Journal of

Public Health, 71(1), 38–46. doi:10.2105/AJPH.71.1.38

Glaesser, J., & Cooper, B. (2011). Selecting cases for in-depth study from a survey dataset:

an application of Ragin’s configurational methods. Methodological Innovations

Online, 6(2), 52–70.

Gleeson, D., & Friel, S. (2013). Emerging threats to public health from regional trade

agreements. The Lancet, 381(9876), 1507–1509. doi:10.1016/S0140-

6736(13)60312-8

Goger, A. (2013). From disposable to empowered: rearticulating labor in Sri Lankan apparel

factories. Environment and Planning A, 45(11), 2628–2645. doi:10.1068/a45694

Gold, M. R., Stevenson, D., & Fryback, D. G. (2002). HALYS and QALYS and DALYS, Oh My:

similarities and differences in summary measures of population Health. Annual

313
Review of Public Health, 23, 115–134.

doi:10.1146/annurev.publhealth.23.100901.140513

Goto, K., & Endo, T. (2014). Upgrading, Relocating, Informalising? Local Strategies in the Era

of Globalisation: The Thai Garment Industry. Journal of Contemporary Asia, 44(1),

1–18. doi:10.1080/00472336.2013.794365

Greenpeace. (2011). Dirty Laundry: Unravelling the corporate connections to toxic water

pollution in China. Amsterdam: Greenpeace.

Grimm, H. (2006). Entrepreneurship Policy and Regional Economic Growth. Exploring the

Link and Theoretical Implications. In H. Grimm & B. Rihoux (Eds.), Innovative

Comparative Methods for Policy Analysis: Beyond the Quantitative-Qualitative

Divide (2006 edition., pp. 123–144). New York, NY: Springer.

Grown, C. (2005). Trade Liberalization and Reproductive Health: A framework for

understanding the linkages. Development, 48(4), 28–42.

doi:10.1057/palgrave.development.1100198

Hadjimichalis, C. (2006). The End of Third Italy As We Knew It? Antipode, 38(1), 82–106.

doi:10.1111/j.0066-4812.2006.00566.x

Hancock, P. (2009). Gender, Status and Empowerment A Study among Women Who Work

in a Sri Lankan Export Processing Zone (EPZ). Journal of Developing Societies, 25(4),

393–420. doi:10.1177/0169796X0902500401

Harper, S., King, N. B., Meersman, S. C., Reichman, M. E., Breen, N., & Lynch, J. (2010).

Implicit value judgments in the measurement of health inequalities. The Milbank

Quarterly, 88(1), 4–29. doi:10.1111/j.1468-0009.2010.00587.x

Harris, G. (2013, December 22). Bangladeshi Factory Owners Charged in Fire That Killed 112.

The New York Times. Retrieved from

http://www.nytimes.com/2013/12/23/world/asia/bangladeshi-factory-owners-

charged-in-fatal-fire.html

314
Hassler, M. (2004). Changes in the Indonesian clothing industry: Trade and regulation.

Singapore Journal of Tropical Geography, 25(1), 64–76.

Hawkes, C. (2006). Uneven dietary development: linking the policies and processes of

globalization with the nutrition transition, obesity and diet-related chronic diseases.

Globalization and Health, 2(1), 4. doi:10.1186/1744-8603-2-4

He, G., & Perloff, J. M. (2013). Surface Water Quality and Infant Mortality in China.

Department of Agricultural and Resource Economics, UC Berkeley. Retrieved from

http://areweb.berkeley.edu/candidate/sites/areweb.berkeley.edu/files/water_and

_infant_0.pdf

Heerden, A. van, Berhouet, M., & Caspari, C. (2003). Rags or riches?: Phasing-out the multi-

fibre arrangement (SEED Working Paper No. 40). International Labor Organization.

Retrieved from http://www.ilo.org/empent/Publications/WCMS_117697/lang--

en/index.htm

Heron, T. (2006). The Ending of the Multifibre Arrangement: A Development Boon for the

South? The European Journal of Development Research, 18(1), 1–21.

doi:10.1080/09578810600576461

Hewamanne, S. (2008). Stitching identities in a free trade zone gender and politics in Sri

Lanka. Philadelphia, Pa.: University of Pennsylvania Press.

Hill, K. (2003). ADULT MORTALITY IN THE DEVELOPING WORLD; WHAT WE KNOW AND

HOW WE KNOW IT. Presented at the TRAINING WORKSHOP ON HIV/AIDS AND

ADULT MORTALITY IN DEVELOPING COUNTRIES, New York: United Nations

Secretariat. Retrieved from

http://www.un.org/esa/population/publications/adultmort/HILL_Paper1.pdf

Hobson, J. (2013). To die for? The health and safety of fast fashion. Occupational Medicine,

63(5), 317–319. doi:10.1093/occmed/kqt079

315
Hoelscher, P., & Alexander, G. (2010). Social protection in times of crisis: experiences in

Eastern Europe and Central Asia. Journal of Poverty and Social Justice, 18(3), 255–

268.

Hong Kong Christian Industrial Committee. (2003). Garment Industry Supply Chains and

Women Workers In the Guangdong Province China. Clean Clothes Campaign.

Retrieved from https://www.cleanclothes.org/resources/national-cccs/garment-

report-www.pdf/view

Hopkins, S. (2006). Economic stability and health status: Evidence from East Asia before and

after the 1990s economic crisis. Health Policy, 75(3), 347–357.

doi:10.1016/j.healthpol.2005.04.002

Hossain, N. (2011). Exports, Equity, and Empowerment: Th effects of readymade garments

manufacturing on gender equality in Bangladesh (World Development Report 2012

Background Report). Washington, D.C.: World Bank.

Howell, E. M., & Blondel, B. (1994). International infant mortality rates: bias from reporting

differences. American Journal of Public Health, 84(5), 850–852.

Huynen, M. M., Martens, P., & Hilderink, H. B. (2005a). The health impacts of globalisation:

a conceptual framework. Globalization and Health, 1(1), 14. doi:10.1186/1744-

8603-1-14

Huynen, M. M., Martens, P., & Hilderink, H. B. (2005b). The health impacts of globalisation:

a conceptual framework (p. 14). Netherlands Environmental Assessment Agency

(MNP). Retrieved from

http://www.pbl.nl/en/publications/2005/The_health_impacts_of_globalization_a_

conceptual_framework

ILO. (2004). Economic security for a better world. International Labour Office.

ILO. (2005). Promoting fair globalization in textiles and clothing in a post-MFA environment

(Report for discussion at the Tripartite Meeting on Promoting Fair Globalization in

316
Textiles and Clothing in a Post-MFA Environment). International Labor Organization.

Retrieved from

http://www.ilo.org/sector/Resources/publications/WCMS_161673/lang--

en/index.htm

ILO. (2012a). Better work Indonesia: garment industry baseline report : worker perspectives

from the factory and beyond. Geneva.

ILO. (2012b). Skills for Trade and Economic Diversification in the Kyrgyz garment sector

(Employment Report No. 19). Geneva: International Labor Organization.

ILO. (2014a). Employment protection legislation database - EPLex. ILO. Retrieved February 8,

2014, from http://www.ilo.org/dyn/eplex/termmain.home

ILO. (2014b). Employment protection legislation database - EPLex: Korea, Republic of.

Employment protection legislation database - EPLex. Retrieved February 8, 2014,

from

http://www.ilo.org/dyn/eplex/termmain.showCountry?p_lang=en&p_country_id=5

IMF. (2007). Bangladesh: Selected Issues (IMF Country Report No. 07/230). Washington,

D.C.: International Monetary Fund.

International Labor Rights Fund. (2002). Sexual Harassment in the Workplace: A report from

Field Research in Thailand. Retrieved from

http://digitalcommons.ilr.cornell.edu/cgi/viewcontent.cgi?article=1469&context=gl

obaldocs

Islam, M. K., & Zahid, D. (2012). Socioeconomic Deprivation and Garment Worker

Movement in Bangladesh: A Sociological Analysis. American Journal of Sociological

Research, 2(4), 82–89.

317
ISTAT. (2004). Censimento generale dell’industria e dei servizi, 1971, 1981, 1991, and 2001

[General census of industry and services, 1971, 1981, 1991 and 2001]. ISTAT:Rome.

Retrieved from http://dwcis.istat.it/cis/ index.htm

JAAF. (n.d.). Ethical Manufacturing is not just a business standard, it is a way of life.

Colombo: Joint Apparel Association Forum.

Jahiel, A. R. (2006). China, the WTO, and implications for the environment. Environmental

Politics, 15(2), 310–329. doi:10.1080/09644010600562666

James, W. E., Ray, D. J., & Minor, P. J. (2003). INDONESIA’S TEXTILES AND APPAREL: THE

CHALLENGES AHEAD. Bulletin of Indonesian Economic Studies, 39(1), 93–103.

doi:10.1080/00074910302005

Jamrozik, K., & Hobbs, M. (2004). Medical care and public health. Oxford University Press.

Retrieved from http://digital.library.adelaide.edu.au/dspace/handle/2440/47454

Jayaweera, S. (2003). Continuity and Change: Women Workers in Garment and Textile

Industries in Sri Lanka. In S. Mukhopadhyay & R. M. Sudarshan (Eds.), Tracking

Gender Equity Under Economic Reforms: Continuity and Change in South Asia. IDRC.

Johnson, J., & Galea, S. (2009). Disasters and Population Health. In K. E. Cherry (Ed.), .

Springer-Verlag.

Jun, M., Jingjing, W., Collins, M., Malei, W., Orlins, S., & Jie, L. (2012). Sustainable Apparel’s

Critical Blind Spot. Friends of Nature, the Institute of Public & Environmental Affairs,

Envirofriends and Nanjing Green Stone.

Kabeer, N. (2000). The Power to Choose: Bangladeshi Women and Labour Market Decisions

in London and Dhaka. Verso.

Kabeer, N. (2004). Globalization, labor standards, and women’s rights: dilemmas of

collective (in)action in an interdependent world. Feminist Economics, 10(1), 3–35.

doi:10.1080/1354570042000198227

318
Kabeer, N., & Mahmud, S. (2004a). Globalization, gender and poverty: Bangladeshi women

workers in export and local markets. Journal of International Development, 16(1),

93–109. doi:10.1002/jid.1065

Kabeer, N., & Mahmud, S. (2004b). Rags, Riches and Women Workers: Export-oriented

Garment Manufacturing in Bangladesh. In M. Carr (Ed.), Chains of Fortune: Linking

Women Producers and Workers with Global Markets (pp. 133–162). London:

Commonwealth Secretariat.

Kahhancová, M., & Martisková, M. (2013). FROM COLLECTIVE BARGAINING TO POLITICAL

ACTION: TRADE UNION RESPONSES TO PRECARIOUS EMPLOYMENT IN THE SLOVAK

REPUBLIC (CELSI Research Report No. 2). Bratislava, Slovak Republic: Central

European Labour Studies Institute. Retrieved from

http://www.celsi.sk/media/research_reports/celsi-rr-2.pdf

Keane, J., & Willem te Velde, D. (2008). The role of textile and clothing industries in growth

and development strategies. Overseas Development Institute: Investment and

Growth Programme. Retrieved from http://www.odi.org.uk/publications/2493-

role-clothing-textile-industries-growth-development-strategies

Kelegama, S. (2009). Ready-made Garment Exports from Sri Lanka. Journal of Contemporary

Asia, 39, 579–596. doi:10.1080/00472330903076875

Kelley, A. (1991). The Human Development Index: “Handle with Care.” Population and

Development Review, 17(2), 315–324.

Kelsey, J. (2012). New-generation free trade agreements threaten progressive tobacco and

alcohol policies. Addiction, 107(10), 1719–1721. doi:10.1111/j.1360-

0443.2012.03874.x

Kelsey, J. (2013). The Trans-Pacific Partnership agreement: a gold-plated gift to the global

tobacco industry? American Journal of Law & Medicine, 39(2-3), 237–64.

319
Kennedy, G., Nantel, G., & Shetty, P. (2004). Globalization of food systems in developing

countries: a synthesis of country case studies. (FAO Food and Nutrition Paper) (pp.

1–24).

Khaleeli, H. (2013, April 29). Inside Bangladesh’s garment factories: life and work in a

dangerous industry. The Guardian. Retrieved from

http://www.theguardian.com/world/2013/apr/29/inside-bangladesh-garment-

factories

Khatun, F., Rahman, M., Bhattacharya, D., Moazzem, K. G., & Shahrin, A. (2007). Gender and

Trade Liberalization in Bangladesh: the Case of the Ready-Made Garments (Greater

Access to Trade Expansion (GATE) Project). United States Agency for International

Development (USAID).

Khosla, N. (2009). The Ready-Made Garments Industry in Bangladesh: A Means to Reducing

Gender-Based Social Exclusion of Women? Journal of International Women’s

Studies, 11(1), 289–303.

Kibria, N. (1998). Becoming a Garments Worker: The Mobilization of Women into the

Garments Factories of Bangladesh (Occasional Paper No. 9). United Nations

Development Programme. Retrieved from

http://www.unrisd.org/80256B3C005BCCF9/(httpAuxPages)/523115D41019B9D98

0256B67005B6EF8/$file/opb9.pdf

Kim, H. J., Fay, M. P., Feuer, E. J., & Midthune, D. N. (2000). Permutation tests for joinpoint

regression with applications to cancer rates. Statistics in Medicine, 19(3), 335–351.

Kim, I.-H., Khang, Y.-H., Cho, S.-I., Chun, H., & Muntaner, C. (2011). Gender, Professional

and Non-Professional Work, and the Changing Pattern of Employment-Related

Inequality in Poor Self-Rated Health, 1995-2006 in South Korea. Journal of

Preventive Medicine and Public Health, 44(1), 22. doi:10.3961/jpmph.2011.44.1.22

320
Kim, I.-H., Khang, Y.-H., Muntaner, C., Chun, H., & Cho, S.-I. (2008). Gender, precarious work,

and chronic diseases in South Korea. American Journal of Industrial Medicine,

51(10), 748–757. doi:10.1002/ajim.20626

Kindig, D., & Stoddart, G. (2003). What Is Population Health? American Journal of Public

Health, 93(3), 380–383. doi:10.2105/AJPH.93.3.380

Knight, P. (2011). Prospects for the Textile and Clothing Industry in Brazil (Textile Outlook

International No. 153). Textiles Intelligence Limited.

Knowles, A., Reyes, C., & Jackson, K. (2006). GENDER, MIGRATION AND REMITTANCES: A

BANGLADESHI EXPERIENCE. In A. Thornton, McGregor, & Knowles (Eds.), Southern

Perspectives on Development: Dialogue or Division? (pp. 229–246). The University

of Otago, Dunedin.

Koivusalo, M. (2006). The Impact of Economic Globalisation on Health. Theoretical Medicine

and Bioethics, 27(1), 13–34. doi:10.1007/s11017-005-5757-y

Kowalski, P., & Molnár, M. (2009). Economic Impacts of the Phase-Out in 2005 of

Quantitative Restrictions under the Agreement on Textiles and Clothing (OECD

Trade Policy Papers). Paris: Organisation for Economic Co-operation and

Development. Retrieved from http://www.oecd-

ilibrary.org/content/workingpaper/220185234525

Koyanagi, A., & Shibuya, K. (2010). What do we really know about adult mortality

worldwide? The Lancet, 375(9727), 1668–1670. doi:10.1016/S0140-6736(10)60629-

Krieger, N. (1994). Epidemiology and the web of causation: has anyone seen the spider?

Social Science & Medicine (1982), 39(7), 887–903.

Krieger, N. (2001). Theories for social epidemiology in the 21st century: an ecosocial

perspective. International Journal of Epidemiology, 30(4), 668–677.

321
Krieger, N. (2008a). Ladders, pyramids and champagne: the iconography of health

inequities. Journal of Epidemiology and Community Health, 62(12), 1098–1104.

doi:10.1136/jech.2008.079061

Krieger, N. (2008b). Proximal, Distal, and the Politics of Causation: What’s Level Got to Do

With It? American Journal of Public Health, 98(2), 221–230.

doi:10.2105/AJPH.2007.111278

Krieger, N. (2011). Epidemiology and the people’s health: theory and context. Oxford; New

York: Oxford University Press.

Krieger, N., Alegría, M., Almeida-Filho, N., Barbosa da Silva, J., Barreto, M. L., Beckfield, J., …

Walters, K. L. (2010). Who, and what, causes health inequities? Reflections on

emerging debates from an exploratory Latin American/North American workshop.

Journal of Epidemiology and Community Health, 64(9), 747–749.

doi:10.1136/jech.2009.106906

Ku, I. (2007). Social welfare reform since the 1997 economic crisis in Korea: achievement,

limits, and future prospects. Asian Social Work and Policy Review, 1(1), 21–35.

Kusakabe, K., & Pearson, R. (2010). Transborder Migration, Social Reproduction and

Economic Development: A Case Study of Burmese Women Workers in Thailand:

Burmese women workers in Thailand. International Migration, 48(6), 13–43.

doi:10.1111/j.1468-2435.2010.00630.x

Kusakabe, K., & Pearson, R. (2014). Burmese Female Migrant Workers in Thailand:

Managing Productive and Reproductive Responsibilities. In T.-D. Truong, D. Gasper,

J. Handmaker, & S. I. Bergh (Eds.), Migration, Gender and Social Justice (pp. 69–85).

Springer Berlin Heidelberg. Retrieved from

http://link.springer.com/chapter/10.1007/978-3-642-28012-2_4

KWWA. (2008, March 20). The current situation of women employment in garment industry

and its requested policy. Korean Women Workers Association: [vol.11]. Korean

322
Women Workers Association (KWWA). Retrieved June 8, 2014, from

http://kwwa.tistory.com/239

Labonté, R. (2008). Global health in public policy: finding the right frame? Critical Public

Health, 18(4), 467–482. doi:10.1080/09581590802443588

Labonté, R., Blouin, C., Chopra, M., Lee, K., Packer, C., Rowson, M., … Woodward, D. (2007).

Towards health-equitable globalization: rights, regulation and redistribution.

Globalization knowledge network final report to the Commission on Social

Determinants of Health. Institute of Population Health, University of Ottawa.

Retrieved from

http://www.who.int/social_determinants/resources/globlalization_kn_07_2007.pd

Labonté, R., & Sanger, M. (2006). Glossary of the World Trade Organisation and public

health: part 1. Journal of Epidemiology and Community Health, 60(8), 655–661.

doi:10.1136/jech.2005.037895

Labonté, R., & Schrecker, T. (2006). Globalization and social determinants of health:

Analytic and strategic review paper. University of Ottawa: Institute of Population

Health.

Labonté, R., & Schrecker, T. (2007). Globalization and social determinants of health:

Introduction and methodological background (part 1 of 3). Globalization and Health,

3(1), 1–10. doi:10.1186/1744-8603-3-5

Labonté, R., & Schrecker, T. (2009). Introduction: Globalization’s Challenges to People’s

Health. In R. Labonté, T. Schrecker, C. Packer, & V. Runnels (Eds.), Globalization and

Health: Pathways, Evidence and Policy (1 edition., pp. 1–33). New York; London:

Routledge.

Labonté, R., & Torgerson, R. (2003). Frameworks for analyzing the links between

globalization and health. Geneva: World Health Organization.

323
Labonté, R., & Torgerson, R. (2005). Interrogating globalization, health and development:

Towards a comprehensive framework for research, policy and political action.

Critical Public Health, 15(2), 157–179. doi:10.1080/09581590500186117

Lake, H. (2008). Analysis of human resourse management practicies: Indonesia labor

intensive light manufacturing industries. USAID.

Lee, B.-H., Lee, S., & International Labour Office. (2007). Minding the gaps: non-regular

employment and labour market segmentation in the Republic of Korea. Geneva.

Lee, K. (2000). The impact of globalization on public health: implications for the UK Faculty

of Public Health Medicine. Journal of Public Health Medicine, 22(3), 253–262.

Lee, K., & Park, S. (2006). Employment insurance system implementation and evaluation. In

Employment insurance in Korea: The first ten years (pp. 85–122). Seoul: Korea Labor

Institute.

Lee, O.-J. (1993). Gender-Differentiated Employment Practices in the South Korean Textile

Industry. Gender and Society, 7(4), 507–528.

Link, B. G., & Phelan, J. (1995). Social conditions as fundamental causes of disease. Journal

of Health and Social Behavior, Spec No, 80–94.

Little, I. M. D., Scitovsky, T., & Scott, M. F. (1970). Industry and trade in some developing

countries: a comparative study. Published for the Development Centre of the

Organization for Economic Co-operation and Development by Oxford U.P.

Liu, Z. (2005). Institution and inequality: the hukou system in China. Journal of Comparative

Economics, 33(1), 133–157. doi:10.1016/j.jce.2004.11.001

Loewenson, R. (2001). Globalization and occupational health: a perspective from southern

Africa. Bulletin of the World Health Organization, 79(9), 863–868.

Loewenson, R., Nolen, L. B., & Wamala, S. (2010). Review article: Globalisation and

women’s health in Sub-Saharan Africa: would paying attention to women’s

324
occupational roles improve nutritional outcomes? Scandinavian Journal of Public

Health, 38(4 Suppl), 6–17. doi:10.1177/1403494809358276

Lopez-Acevedo, G., & Robertson, R. (2012). Sewing Success? Employment, Wages, and

Poverty following the End of the Multi-Fibre Arrangement. Washington, DC: World

Bank. Retrieved from https://openknowledge.worldbank.org/handle/10986/13137

Lundberg, O. (2008). Commentary: Politics and public health—some conceptual

considerations concerning welfare state characteristics and public health outcomes.

International Journal of Epidemiology, 37(5), 1105–1108. doi:10.1093/ije/dyn078

Lundin, A., Lundberg, I., Hallsten, L., Ottosson, J., & Hemmingsson, T. (2010).

Unemployment and mortality—a longitudinal prospective study on selection and

causation in 49321 Swedish middle-aged men. Journal of Epidemiology and

Community Health, 64(01), 22–28. doi:10.1136/jech.2008.079269

Lynch, C. (2007). Juki girls, good girls: gender and cultural politics in Sri Lanka’s global

garment industry. Ithaca: ILR Press/Cornell University Press.

Martens, P., Akin, S.-M., Maud, H., & Mohsin, R. (2010). Is globalization healthy: a statistical

indicator analysis of the impacts of globalization on health. Globalization and

Health, 6(1), 16. doi:10.1186/1744-8603-6-16

Marx, A., & Dusa, A. (2011). Crisp-Set Qualitative Comparative Analysis (csQCA):

Contradictions and consistency benchmarks for model specification.

Methodological Innovations Online, 6(2), 103–148.

Mathers, C. D., Salomon, J. A., & Murray, C. J. L. (2003). Infant mortality is not an adequate

summary measure of population health. Journal of Epidemiology and Community

Health, 57(5), 319–319. doi:10.1136/jech.57.5.319

McCaffrey, S. J. (2013). Tacit-rich districts and globalization: changes in the Italian textile

and apparel production system. Socio-Economic Review, 11(4), 657–685.

doi:10.1093/ser/mwt005

325
McGillivray, M. (1991). The human development index: Yet another redundant composite

development indicator? World Development, 19(10), 1461–1468.

doi:10.1016/0305-750X(91)90088-Y

McNamara, D. (1999). Global Adjustment in Korean Textiles. Creativity and Innovation

Management, 8(1), 48–56.

Meier, B. M. (2006). Employing Health Rights for Global Justice: The Promise of Public

Health in Response to the Insalubrious Ramifications of Globalization. Cornell

International Law Journal, 39, 711–774.

Melo Junior, A. S. (2012). The risk of developing repetitive stress injury in seamstresses, in

the clothing industry, under the perspective of ergonomic work analysis: a case

study. Work: A Journal of Prevention, Assessment and Rehabilitation, 41, 1670–

1676.

Mendoza, R. U. (2009). Aggregate Shocks, Poor Households and Children Transmission

Channels and Policy Responses. Global Social Policy, 9(1 suppl), 55–78.

doi:10.1177/1468018109106885

Mogilevsky, R., & Omorova, A. (2011). Assessing Development Strategies to Achieve the

MDGs in The Kyrgyz Republic (Country Report). United Nations Department for

Social and Economic Affairs.

Moon, K.-H. (2003). Integration of women into development: women’s working conditions

inthe Korean textile and apparel industries. Labour and Management in

Development Journal, Asia Pacific Press, 4(1). Retrieved from

https://digitalcollections.anu.edu.au/handle/1885/41083

Munková, M. (2004). Regulation of dismissals examined. European Industrial Relations

Obsevatory On-line. Retrieved June 4, 2014, from

http://www.eurofound.europa.eu/eiro/2004/01/feature/sk0401111f.htm

326
Muntaner, C., Borrell, C., Ng, E., Chung, H., Espelt, A., Rodriguez-Sanz, M., … O’Campo, P.

(2011). Politics, welfare regimes, and population health: controversies and

evidence. Sociology of Health & Illness, 33(6), 946–964. doi:10.1111/j.1467-

9566.2011.01339.x

Muntaner, C., Chung, H., Solar, O., Santana, V., Castedo, A., Benach, J., & EMCONET

Network. (2010). A macro-level model of employment relations and health

inequalities. International Journal of Health Services: Planning, Administration,

Evaluation, 40(2), 215–221.

Mustard, C. A., Bielecky, A., Etches, J., Wilkins, R., Tjepkema, M., Amick, B. C., … Aronson, K.

J. (2013). Mortality following unemployment in Canada, 1991–2001. BMC Public

Health, 13(1), 441. doi:10.1186/1471-2458-13-441

Naumann, E. (2006). The Multifibre Agreement – WTO Agreement on Textiles and Clothing

(Trade Law Centre for Southern Africa Working Paper No. 4). Stellenbosch, South

Africa. Retrieved from http://www.fibre2fashion.com/industry-article/pdffiles/the-

multifibre-agreement.pdf

Navarro, V., & Muntaner, C. (2004). Conclusion: Political, Economic, and Cultural

Determinants of Population Health--A Research Agenda. In Political and Economic

Determinants of Population Health and Well-being: Controversies and

Developments (pp. 551–556). Amityville, N.Y: Baywood Publishing Company Inc.

Neumayer, E. (2001). The human development index and sustainability — a constructive

proposal. Ecological Economics, 39(1), 101–114. doi:10.1016/S0921-

8009(01)00201-4

Ngai, P. (2004). Women workers and precarious employment in Shenzhen special economic

zone, China. Gender & Development, 12(2), 29–36.

Nordås, H. K. (2004). The Global Textile and Clothing Industry Post the Agreement on

Textiles and Clothing (Discussion Paper No. 5). Geneva: World Trade Organization.

327
Retrieved from

http://www.wto.org/english/res_e/booksp_e/discussion_papers5_e.pdf

OECD. (2004). A New World Map in Textiles and Clothing Adjusting to Change: Adjusting to

Change. OECD. Retrieved from

http://www.oecd.org/tad/anewworldmapintextilesandclothingadjustingtochange.h

tm

OECD. (2014, February 3). OECD.Stat. Retrieved February 8, 2014, from http://www.oecd-

ilibrary.org/content/data/data-00285-en

Owen, N., & Cannon-Jones, A. (2003, April 1). DTI Economics Paper No. 2: A comparative

study of the British and Italian Textile and Clothing Industries. Monograph.

Retrieved June 4, 2014, from http://www.bis.gov.uk/files/file14772.pdf

Palme, J., Nelson, K., Sjöberg, O., & Minas, R. (2009). European social models, protection

and inclusion (Research Report). Stockholm: Institute for Futures Studies. Retrieved

from

http://people.su.se/~kennethn/European%20Social%20Models,%20Protection%20

and%20Inclusion.pdf

Pangsapa, P. (2007). Textures of struggle: the emergence of resistance among garment

workers in Thailand. Ithaca: ILR Press/Cornell University Press.

Pattemore, P. K. (2013). Tobacco or Healthy Children: The Two Cannot Co-Exist. Frontiers in

Pediatrics, 1, 20. doi:10.3389/fped.2013.00020

Pickles, J., & Smith, A. (2010). Clothing workers after worker states: the consequences for

work and labour of outsourcing, nearshoring and delocalization in postsocialist

Europe. In S. Mcgrath-champ, A. Herod, & A. Rainnie (Eds.), Handbook of

Employment and Society: Working Space (pp. 106–123). Cheltenham: Edward Elgar.

Platzer, M. D. (2013). U.S. textile manufacturing and the Trans-Pacific Partnership

negotiations. Washington D.C.: Congressional Research Service. Retrieved from

328
http://digitalcommons.ilr.cornell.edu/cgi/viewcontent.cgi?article=2197&context=k

ey_workplace

Polakoff, E. G. (2007). Globalization and Child Labor Review of the Issues. Journal of

Developing Societies, 23(1-2), 259–283. doi:10.1177/0169796X0602300215

Popkin, B. M. (1998). The nutrition transition and its health implications in lower-income

countries. Public Health Nutrition, 1(1), 5–21.

Prasanna, R., & Gowthaman, B. (2006). Sector Specific Living Wage For Sri Lankan Apparel

Industry Workers. Apparel-industry Labour Rights Movement (ALaRM), Clean

Clothes Campaign. Retrieved from https://www.modepoly.org/documents/06-

ALaRM_LIVING_WAGE_sri_lanka.pdf

Ragin, C. (2000). Fuzzy-Set Social Science. University of Chicago Press.

Ragin, C. (2006). The Limitations of Net-Effects Thinking. In H. Grimm & B. Rihoux (Eds.),

Innovative Comparative Methods for Policy Analysis: Beyond the Quantitative-

Qualitative Divide (2006 edition., pp. 13–41). New York, NY: Springer.

Ragin, C. (2008). Redesigning Social Inquiry: Fuzzy Sets and Beyond. Chicago: University of

Chicago Press.

Ragin, C., & Schneider, G. A. (2011). Case-Oriented Theory Building and Theory Testing. In

M. Williams & W. P. Vogt (Eds.), The SAGE Handbook of Innovation in Social

Research Methods (pp. 150–166). London: SAGE Publications Ltd. Retrieved from

http://srmo.sagepub.com/view/sage-hdbk-innovation-in-social-research-

methods/n11.xml

Rajaratnam, J. K., Marcus, J. R., Flaxman, A. D., Wang, H., Levin-Rector, A., Dwyer, L., …

Murray, C. J. (2010). Neonatal, postneonatal, childhood, and under-5 mortality for

187 countries, 1970–2010: a systematic analysis of progress towards Millennium

Development Goal 4. The Lancet, 375(9730), 1988–2008. doi:10.1016/S0140-

6736(10)60703-9

329
Raphael, D. (2011). A discourse analysis of the social determinants of health. Critical Public

Health, 21(2), 221–236. doi:10.1080/09581596.2010.485606

Rasella, D., Aquino, R., Santos, C. A., Paes-Sousa, R., & Barreto, M. L. (2013). Effect of a

conditional cash transfer programme on childhood mortality: a nationwide analysis

of Brazilian municipalities. The Lancet, 382(9886), 57–64. doi:10.1016/S0140-

6736(13)60715-1

Rasiah, R. (2012). Beyond the Multi-Fibre Agreement: How are Workers in East Asia Faring?

International Journal of Institutions and Economies, 4(3). Retrieved from http://e-

journal.um.edu.my/filebank/published_article/4114/Fulltext1.pdf

Rayner, G., Hawkes, C., Lang, T., & Bello, W. (2006). Trade liberalization and the diet

transition: a public health response. Health Promotion International, 21(suppl 1),

67–74. doi:10.1093/heapro/dal053

Reidpath, D. D., & Allotey, P. (2003). Infant mortality rate as an indicator of population

health. Journal of Epidemiology and Community Health, 57(5), 344–346.

doi:10.1136/jech.57.5.344

Rihoux, B. (2006). Qualitative Comparative Analysis (QCA) and Related Systematic

Comparative Methods | Recent Advances and Remaining Challenges for Social

Science Research. International Sociology, 21(5), 679–706.

doi:10.1177/0268580906067836

Rinolfi, V. (2013, July 25). Italy: The representativeness of trade unions and employer

associations in the textile and clothing sector. European Industrial Relations

Obsevatory On-line. Retrieved June 8, 2014, from

http://www.eurofound.europa.eu/eiro/studies/tn1302024s/it1302029q.htm

Robertson, R., Brown, D., Pierre, G., & Sanchez-Puerta, M. L. (2009). Globalization, wages,

and the quality of jobs: five country studies. World Bank. Retrieved from

http://books.google.com/books?hl=en&lr=&id=BU9sU7sG3DoC&oi=fnd&pg=PR5&d

330
q=%22trade%E2%80%94Social+aspects.+4.+Developing+countries%E2%80%94Com

merce.+I.+Robertson,+Raymond,+1969-

%22+%22Robertson,+Drusilla+Brown,+Ga%C3%ABlle+Pierre,+and+Mar%C3%ADa+L

aura%22+%22background:+copyright+Rob%22+%22design+by:+Serif+Design,%22+

%22Links+between+Globalization+and+Working+Conditi&ots=oD9f9r-

dZE&sig=moGdIW_MJJvOC_ihW9L1oMw1EhU

Rodgers, Y. V. D. M., & Berik, G. (2006). Asia’s race to capture post-MFA markets: A

snapshot of labor standards, compliance, and impacts on competitiveness. Asian

Development Review, 23(1), 55–86.

Rodriguez, F., & Rodrik, D. (2001). Trade Policy and Economic Growth: A Skeptic’s Guide to

the Cross-National Evidence. NBER, 261–338.

Rodrik, D. (1996). Labor Standards in International Trade: Do They Matter and what Do We

Do about Them? In R. Z. Lawrence, D. Rodrik, & J. Whalley (Eds.), Emerging agenda

for global trade: high stakes for developing countries. Washington, D.C.: Johns

Hopkins University Press.

Roelfs, D. J., Shor, E., Davidson, K. W., & Schwartz, J. E. (2011). Losing life and livelihood: A

systematic review and meta-analysis of unemployment and all-cause mortality.

Social Science & Medicine, 72(6), 840–854. doi:10.1016/j.socscimed.2011.01.005

Roos, N. P., Brownwell, M., & Menec, M. L. (2006). Universal Medical Care and Health

Inequalities: Right Objectives, Insufficient Tools. In Healthier Societies. Oxford

University Press. Retrieved from

http://www.oxfordscholarship.com/view/10.1093/acprof:oso/9780195179200.001.

0001/acprof-9780195179200

Rudra, N. (2007). Welfare States in Developing Countries: Unique or Universal? The Journal

of Politics, 69(02), 378–396. doi:10.1111/j.1468-2508.2007.00538.x

331
Rueda, S., Chambers, L., Wilson, M., Mustard, C., Rourke, S. B., Bayoumi, A., … Lavis, J.

(2012). Association of returning to work with better health in working-aged adults:

a systematic review. American Journal of Public Health, 102(3), 541–556.

doi:10.2105/AJPH.2011.300401

Sánchez-Ancochea, D., & Mattei, L. (2011). Bolsa Família, poverty and inequality: Political

and economic effects in the short and long run. Global Social Policy, 11(2-3), 299–

318. doi:10.1177/1468018111421297

Sant’Ana, M. A., & Kovalechen, F. (2012). Evaluation of the health risks to garment workers

in the city of Xambrê-PR, Brazil. Work: A Journal of Prevention, Assessment and

Rehabilitation, 41, 5647–5649.

Savchenko, Y., & Lopez-Acevedo, G. (2012). Female Wages in the Apparel Industry Post-

MFA: The Cases of Cambodia and Sri Lanka (Policy Research Working Paper No.

6061). Washington, DC: World Bank.

Schell, C. O., Reilly, M., Rosling, H., Peterson, S., & Ekström, A. M. (2007). Socioeconomic

determinants of infant mortality: a worldwide study of 152 low-, middle-, and high-

income countries. Scandinavian Journal of Public Health, 35(3), 288–297.

doi:10.1080/14034940600979171

Schneider, C. Q., & Rohlfing, I. (2013). Combining QCA and Process Tracing in Set-Theoretic

Multi-Method Research. Sociological Methods & Research, 42(4), 559–597.

doi:10.1177/0049124113481341

Schneider, C. Q., & Wagemann, C. (2010). Standards of Good Practice in Qualitative

Comparative Analysis (QCA) and Fuzzy-Sets. Comparative Sociology, 9(3), 397–418.

doi:10.1163/156913210X12493538729793

Schneider, C. Q., & Wagemann, C. (2012). Set-Theoretic Methods for the Social Sciences: A

Guide to Qualitative Comparative Analysis. Cambridge University Press.

332
Schwegler-Rohmeis, W., Mummert, A., & Jarck, K. (2013). Study “Labour Market and

Employment Policy in the Kyrgyz Republic” Identifying constraints and options for

employment development. Bishkek: Deutsche Gesellschaft für Internationale

Zusammenarbeit (GIZ) GmbH.

Seyoum, B. (2010). Trade Liberalization in Textiles and Clothing and Developing Countries:

An Analysis with Special Emphasis on the US Import Market. The International

Trade Journal, 24(2), 149–181. doi:10.1080/08853901003652351

Shaw, J. (2007). “There is No Work in My Village”: The Employment Decisions of Female

Garment Workers in Sri Lanka’s Export Processing Zones. Journal of Developing

Societies, 23(1-2), 37–58. doi:10.1177/0169796X0602300203

Shin, D.-M. (2000). Financial Crisis and Social Security: The Paradox of the Republic of Korea.

International Social Security Review, 53(3), 83–107. doi:10.1111/1468-246X.00079

Siddiqi, D. M. (2009). do Bangladeshi factory workers need saving? Sisterhood in the post-

sweatshop era1. Feminist Review, 91(1), 154–174. doi:10.1057/fr.2008.55

Singer, M. (2008). Drugs and development: The global impact of drug use and trafficking on

social and economic development. International Journal of Drug Policy, 19(6), 467–

478. doi:10.1016/j.drugpo.2006.12.007

Sivananthiran, A. (n.d.). Promoting decent work in export processing zones (EPZs) in Sri

Lanka. Retrieved from

http://oit.org/public/french/dialogue/download/epzsrilanka.pdf

Smith, A. (2003). Power Relations, Industrial Clusters, and Regional Transformations: Pan-

European Integration and Outward Processing in the Slovak Clothing Industry.

Economic Geography, 79(1), 17–40. doi:10.1111/j.1944-8287.2003.tb00200.x

Smith, A., Pickles, J., Begg, R., Roukova, P., & Buček, M. (2005). Outward Processing, EU

Enlargement and Regional Relocation in the European Textiles and Clothing

Industry: Reflections on the European Commission’s Communication on “the Future

333
of the Textiles and Clothing Sector in the Enlarged European Union.” European

Urban and Regional Studies, 12(1), 83–91. doi:10.1177/0969776405046266

Smith, A., Pickles, J., Buček, M., Begg, R., & Roukova, P. (2008). Reconfiguring “post-socialist”

regions: cross-border networks and regional competition in the Slovak and

Ukrainian clothing industry. Global Networks, 8(3), 281–307. doi:10.1111/j.1471-

0374.2008.00196.x

Smith, A., Stenning, A., Rochovská, A., & Świa̧ tek, D. (2008). The Emergence of a Working

Poor: Labour Markets, Neoliberalisation and Diverse Economies in Post-Socialist

Cities. Antipode, 40(2), 283–311. doi:10.1111/j.1467-8330.2008.00592.x

Smith, M. B., & Signal, L. (2009). Global influences on milk purchasing in New Zealand -

implications for health and inequalities. Globalization and Health, 5, 1.

doi:10.1186/1744-8603-5-1

Spinanger, D. (1999). Textiles Beyond the MFA Phase-Out. World Economy, 22(4), 455–476.

doi:10.1111/1467-9701.00213

Starfield, B. (2001). Improving equity in health: a research agenda. International Journal of

Health Services: Planning, Administration, Evaluation, 31(3), 545–566.

Staritz, C. (2010). Making the Cut? The World Bank. Retrieved from

http://elibrary.worldbank.org/doi/book/10.1596/978-0-8213-8636-1

Stuckler, D., & Basu, S. (2009). The International Monetary Fund’s effects on global health:

before and after the 2008 financial crisis. International Journal of Health Services:

Planning, Administration, Evaluation, 39(4), 771–781.

Stuckler, D., Basu, S., Suhrcke, M., Coutts, A., & McKee, M. (2009). The public health effect

of economic crises and alternative policy responses in Europe: an empirical analysis.

The Lancet, 374(9686), 315–323. doi:10.1016/S0140-6736(09)61124-7

Subasat, T. (2008). Do liberal trade policies promote trade openness? International Review

of Applied Economics, 22(1), 45–61. doi:10.1080/02692170701745887

334
Suhrcke, M., & Stuckler, D. (2012). Will the recession be bad for our health? It depends.

Social Science & Medicine (1982), 74(5), 647–653.

doi:10.1016/j.socscimed.2011.12.011

Suhrcke, M., Stuckler, D., Suk, J. E., Desai, M., Senek, M., McKee, M., … Semenza, J. C.

(2011). The Impact of Economic Crises on Communicable Disease Transmission and

Control: A Systematic Review of the Evidence. PLoS ONE, 6(6), e20724.

doi:10.1371/journal.pone.0020724

Tanaka, S. (2012). Environmental Regulations on Air Pollution in China and Their Impact on

Infant Mortality (IDEC DP2 Series No. 2-11). Hiroshima University, Graduate School

for International Development and Cooperation (IDEC). Retrieved from

http://ideas.repec.org/p/hir/idecdp/2-11.html

Tang, S., Meng, Q., Chen, L., Bekedam, H., Evans, T., & Whitehead, M. (2008). Tackling the

challenges to health equity in China. The Lancet, 372(9648), 1493–1501.

doi:10.1016/S0140-6736(08)61364-1

Tavora, I., & Rubery, J. (2013). Female employment, labour market institutions and gender

culture in Portugal. European Journal of Industrial Relations, 19(3), 221–237.

doi:10.1177/0959680113493374

The Fair Wear Foundation and Ethical Trading Initiative-Norway. (2007). Background Study

Thailand. Amsterdam.

Thow, A. M. (2009). Trade liberalisation and the nutrition transition: mapping the pathways

for public health nutritionists. Public Health Nutrition, 12(11), 2150–2158.

doi:10.1017/S1368980009005680

Torres, R. (2001).Towards a socially sustainable world economy: an analysis of the social

pillars of globalization. Geneva: International Labour Office.

335
Truett, L. J., & Truett, D. B. (2011). The Korean textile industry: still competitive, after all

these years? Applied Economics, 43(22), 2983–2992.

doi:10.1080/00036846.2010.528374

Umurzako, K., & Burzhubaev, T. (2010). KYRGYZSTAN: AID FOR TRADE NEEDS ASSESSMENT.

Bishkek: United Nations Development Programme.

UNDP. (2006). Human Development Report 2006: Beyond scarcity: Power, poverty and the

global water crisis. New York, NY: United Nations Development Programme.

Retrieved from http://hdr.undp.org/en/content/human-development-report-2006

UNEP. (2005). National Rapid Environmental Assessment- Sri Lanka. United Natition

Environment Programme (UNEP). Retrieved from

http://www.unep.org/tsunami/reports/tsunami_srilanka_layout.pdf

UNIDO. (2011). INDSTAT2, Industrial Statistics Database (Edition: 2011).

doi:10.5257/unido/indstat2/2011

United Nations Development Programme (UNDP). (2006). Asia-Pacific Human Development

Report 2006. New York: United Nations Development Programme. Retrieved from

http://www.undp.org/content/undp/en/home/librarypage/hdr/Asia-Pacific-

Human-Development-Report-2006/

United Nations Food and Agriculture Organization. (2006). Trade Reforms and Food Security:

Country Case Studies and Synthesis. Rome, Italy: Food & Agriculture Organization.

USAID. (2007). GENDER AND TRADE LIBERALIZATION IN BANGLADESH: THE CASE OF THE

READY-MADE GARMENTS (GREATER ACCESS TO TRADE EXPANSION (GATE)

PROJECT UNDER THE WOMEN IN DEVELOPMENT IQC). USAID. Retrieved from

http://www.usaidwidworkshops.org/GATE_CD/session4.html

Vågerö, D. (2006). Where does new theory come from? Journal of Epidemiology and

Community Health, 60(7), 573–574. doi:10.1136/jech.2005.038893

336
Veselý, A., & Smith, M. L. (2008). Macro-Micro Linkages and the Role of Mechanisms in

Social Stratification Research. Czech Sociological Review/Sociologický Asopis, 44(3),

491–509.

Vickers, A. (2012). Clothing Production in Indonesia: A Divided Industry. Institutions and

Economies, 4(3), 41–60.

Whitehead, M. (1992). The concepts and principles of equity and health. International

Journal of Health Services: Planning, Administration, Evaluation, 22(3), 429–445.

WHO. (2014). Arsenic. Fact sheet N°372. World Health Organization. Retrieved June 8, 2014,

from http://www.who.int/mediacentre/factsheets/fs372/en/

WHO, & UNDP. (2001). Environment and People’s Health in China. World Health

Organization and United Nations Development Programme. Retrieved from

http://www.wpro.who.int/environmental_health/documents/docs/CHNEnvironme

ntalHealth.pdf

Wick, I. (2009). The social impact of the liberalised world market for textiles and clothing:

Strategies of trade unions and women´s organisations (Workbook No. 62).

Frankfurt/Main: Otto Brenner Foundation. Retrieved from http://www.otto-

brenner-shop.de/uploads/tx_mplightshop/AH62_en_01.pdf

Wikileaks. (2005, September 27). Embassy Cable: Peru: Updated Textile And Apparel

Information. Wikileaks. Retrieved from

https://cablegatesearch.wikileaks.org/cable.php?id=05LIMA4184&q=peru%20textil

Wikileaks. (2006a, April 6). Embassy Cable: Garment Quota Expiration No Drastic Impact On

Labor. Wikileaks. Retrieved from

https://cablegatesearch.wikileaks.org/cable.php?id=06COLOMBO537

Wikileaks. (2006b, May 11). Embassy Cable: Textile Sector Growing At Diminished Rate.

Wikileaks. Retrieved from

337
https://cablegatesearch.wikileaks.org/cable.php?id=06LIMA1825&q=peru%20textil

Wikileaks. (2006c, September 28). Embassy Cable: Textiles And Apparel Sector: Updated

Statistics And Projections Of Future Competitivenss - Sri Lanka. Wikileaks. Retrieved

from http://www.cablegatesearch.net/cable.php?id=06COLOMBO1594

Wikileaks. (2006d, October 4). Embassy Cable: Textiles And Apparel Sector: Slovakia.

Wikileaks. Retrieved from

http://www.cablegatesearch.net/cable.php?id=06BRATISLAVA804&q=slovak%20te

xtiles

Wikileaks. (2007a, October 2). Embassy Cable: Sri Lanka Textiles And Apparel Sector:

Updated Stats And Projections For Future Competitiveness. Wikileaks. Retrieved

from https://cablegatesearch.wikileaks.org/cable.php?id=07COLOMBO1355

Wikileaks. (2007b, October 3). Embassy Cable: Peru Textiles And Apparel Exports Grow.

Wikileaks. Retrieved from

https://cablegatesearch.wikileaks.org/cable.php?id=07LIMA3339&q=peru%20textil

Wikileaks. (2008, January 24). Embassy Cable: Sri Lanka: Textile And Apparel Sector

Concerns For 2008: U.s. Recession And Eu Gsp+. Wikileaks. Retrieved from

http://www.cablegatesearch.net/cable.php?id=08COLOMBO92

Wilkinson, R., & Pickett, K. (2010). The Spirit Level: Why Equality is Better for Everyone.

Penguin UK.

Woodward, D. (2005). The GATS and trade in health services: implications for health care in

developing countries. Review of International Political Economy - REV INT POLIT

ECON, 12(3), 511–534. doi:10.1080/09692290500171021

338
Woodward, D., Drager, N., Beaglehole, R., & Lipson, D. (2001). Globalization and health: a

framework for analysis and action. Bulletin of the World Health Organization, 79(9),

875–881. doi:10.1590/S0042-96862001000900014

Woodward, D., Drager, N., Beaglehole, R., & Lipson, D. (2002). Globalization, global public

goods, and health. In N. Drager & C. Vieira (Eds.), Trade in Health Services: Global,

Regional, And Country Perspectives (pp. 3–11). Washington, D.C.: Pan American

Health Organization.

World Bank. (2001). China: air, land, and water : environmental priorities for a new

millennium. Washington, D.C.: World Bank.

World Bank. (2011). World Development Indicators 2011. World Bank Development

Indicators. Retrieved June 8, 2014, from http://data.worldbank.org/data-

catalog/world-development-indicators/wdi-2011

World Bank. (2013, October 16). World Development Indicators 2013. The World Bank.

Retrieved February 8, 2014, from http://data.worldbank.org/data-catalog/world-

development-indicators

WTO. (2007). International Trade Statistics 2007. Geneva: World Trade Organization.

Retrieved from http://www.wto.org/english/res_e/statis_e/statis_e.htm

WTO. (2009). International Trade Statistics 2009. Geneva: World Trade Organization.

Retrieved from http://www.wto.org/english/res_e/statis_e/statis_e.htm

Wu, C. (2007). Studies on the Indonesian textile and garment industry. Labour and

Management in Development Journal, 7(5), 5–14.

Wu, C., Maurer, C., Wang, Y., Xue, S., & Davis, D. L. (1999). Water pollution and human

health in China. Environmental Health Perspectives, 107(4), 251–256.

Yap, J. (2013). Beyond “Don”t Be Evil’: The European Union GSP+ Trade Preference Scheme

and the Incentivisation of the Sri Lankan Garment Industry to Foster Human Rights.

European Law Journal, 19(2), 283–301.

339
Yaşar, Y. (2010). Gender, Development, and Neoliberalism: HIV/ AIDS in Cambodia. Review

of Radical Political Economics, 42(4), 528–548. doi:10.1177/0486613410375062

Yi, B., Wu, L., Liu, H., Fang, W., Hu, Y., & Wang, Y. (2011). Rural-urban differences of

neonatal mortality in a poorly developed province of China. BMC Public Health,

11(1), 477.

You, S., Cheng, S., & Yan, H. (2009). The impact of textile industry on China’s environment.

International Journal of Fashion Design, Technology and Education, 2(1), 33–43.

doi:10.1080/17543260903055141

Zhang, J., Mauzerall, D. L., Zhu, T., Liang, S., Ezzati, M., & Remais, J. V. (2010).

Environmental health in China: progress towards clean air and safe water. The

Lancet, 375(9720), 1110–1119. doi:10.1016/S0140-6736(10)60062-1

340

You might also like