TECHNOLOGICAL UPGRADING AND DECENT WORK IN THE
MANUFACTURING SECTOR：EVIDENCE FROM SEVEN COASTLINE
PROVINCES, CHINA, 2002-2014
by
Jingrui Li
B.A., University of Northwest Agriculture and Forestry University, 2012

THESIS SUBMITTED IN PARTIAL FULFILLMENT OF
THE REQUIREMENTS FOR THE DEGREE OF
MASTER OF ARTS
IN
DEVELOPMENT ECONOMICS

UNIVERSITY OF NORTHERN BRITISH COLUMBIA
November 2016
© Jingrui Li, 2016

Abstract
This research examines the relationship between technological upgrading and decent work in the
manufacturing sector in China. By analyzing data from seven selected coastline provinces
(Jiangsu, Guangdong, Zhejiang, Shandong, Fujian, Guangxi, and Hainan), both technological
upgrading and an increase in decent work have been observed between 2002 and 2014. Decent
work, however, is not distributed evenly among Chinese citizens. The average wage is identified
as a key decent work indicator, and selected as the dependent variable in the econometric
regression model. The impact of technological upgrading on the average nominal wage is
estimated by a time-series-cross-section regression model using provincial level data from 20032014. Regression results indicate that technological upgrading accelerates wage growth in the
manufacturing sector. My suggestions for future government development strategies are, to
firstly, revise the Hukou system to promote equal opportunities for all citizens, and secondly,
offer preferential policies to attract more FDI and encourage domestic R&D activities.

ii

Table of Contents
Abstract ................................................................................................................................................ii
List of Tables .......................................................................................................................................vi
List of Figures ....................................................................................................................................vii
Acknowledgement ........................................................................................................................... viii
Chapter 1: Introduction ..................................................................................................................... 1
1.1 Overview ..................................................................................................................................... 1
1.2 Research goals and methodology............................................................................................... 3
1.3 Importance of this research ........................................................................................................ 4
1.4 Outline of this research ............................................................................................................... 5
1.5 Key definitions ............................................................................................................................ 5
Chapter 2: Literature review ............................................................................................................ 7
2.1 Technological upgrading ............................................................................................................ 7
2.1.2 Definition of technological upgrading................................................................................ 8
2.2 Evidence of technological upgrading in China ....................................................................... 13
2.2.1 Evidence of disembodied technological upgrading ......................................................... 13
2.2.2 Evidence of embodied and disembodied technological upgrading ................................ 15
2.2.3 Enterprise ownership and technological upgrading......................................................... 16
2.3 Definition of decent work......................................................................................................... 20
iii

2.3.1 Definition of formal and informal employment .............................................................. 20
2.3.2 Definition of decent work ................................................................................................. 22
2.4 Implication of technological upgrading for decent work ....................................................... 23
2.5 Summary ................................................................................................................................... 26
Chapter 3: Technological upgrading and decent work in selected provinces ......................... 28
3.1 Introduction ............................................................................................................................... 28
3.2 Methodology ............................................................................................................................. 28
3.3 Evidence of technological upgrading in seven selected provinces ........................................ 31
3.4 Evidence of an increase in decent work .................................................................................. 34
3.4.1 Manufacturing employment .............................................................................................. 34
3.4.2 Manufacturing wages ........................................................................................................ 38
3.4.3 Social security .................................................................................................................... 43
3.5 Correlation between technological upgrading and decent work ............................................ 48
3.6 Decent working time and employment contracts.................................................................... 49
3.6.1 Overview of decent working time in China ..................................................................... 49
3.6.2 Working time and resident registration type .................................................................... 51
Chapter 4: Econometric modeling analysis .................................................................................. 55
4.1 Methodology ............................................................................................................................. 55
4.2 Macroeconomic theories of wage determination .................................................................... 56
4.3 Review of macroeconomic wage models ................................................................................ 57
iv

4.4 The model .................................................................................................................................. 64
4.4.1 Wage model design ........................................................................................................... 64
4.4.2 Econometric issues and solutions ..................................................................................... 66
4.5 Data for the model .................................................................................................................... 71
4.6 Model results ............................................................................................................................. 72
4.7 Summary ................................................................................................................................... 82
Chapter 5: Conclusion ..................................................................................................................... 84
5.1 Research goals and achievements ............................................................................................ 84
5.2 Policy implications ................................................................................................................... 87
5.3 Limitations and recommendations for future research ........................................................... 90
Bibliography ...................................................................................................................................... 93
Appendix: Descriptive statistics for data used in econometric model ....................................103

v

List of Tables
Table 1. Technology indicators in absolute terms, selected provinces, 2002 and 2014 …………32
Table 2. Technology indicators in relative terms, selected provinces, 2002 and 2014………...33
Table 3. Employment by industrial sector (number in ‘000s), urban units, selected provinces,
2014………………………………………………………………………………………………36
Table 4. Share of manufacturing employment (%), urban units, selected provinces, 2002 and
2014………………………………………………………………………………………………38
Table 5. The number of dependents that can be supported by the average manufacturing
wage………………………………………………………………………….…………………43
Table 6. Basic data for China's pension income (in CNY per year and million people) ...……...47
Table 7. Correlation between technology indicators and decent work indicators, across seven
provinces and years 2002-2014………………………….…………………………………….48
Table 8. Average weekly working hours of employees, urban areas, 2012……………………...50
Table 9. Weekly working hours of employed persons by household registration (%), urban
areas ………………………………………………………………………………….………….52
Table 10. Migrant workers’ working hours…………………………………………………….52
Table 11. Work contracts type and daily working hours, migrant workers, urban area………….53
Table 12. Migrant workers with employment contracts (%)…………………………………...53
Table 13. The D.W. critical result………………………...……………………………………...70
Table 14. Model variable details…………………………….……………...…………………74
Table 15. a Wage equation for 7 provinces, using data for 2003-2014…………………………76
Table 15. b Wage equation for 7 provinces, using data for 2003-2014…………………………77

vi

List of Figures
Figure 1. Ownership structure in China………………………………........................................17
Figure 2. Value of high-tech exports by region in 2013…………………………………………30
Figure 3. Employment in manufacturing (in ‘000s), urban units, selected provinces, 20022014………………………………………………………………………………………………37
Figure 4. Average annual nominal wages (CNY) in manufacturing, urban units, selected
provinces,

2002-2014………………………………….……………………………………….39

Figure 5. Average annual nominal wage (manufacturing sector and all urban units) and per capita
consumption expenditure (CNY), selected provinces, 2005……………...……………………41
Figure 6. Average annual nominal wage (manufacturing sector and all urban units) and per capita
consumption expenditure (CNY), selected provinces, 2014……………………………………42
Figure 7. Employment insurance participation ratio (%), selected provinces, 2002-2014…....44
Figure 8. Pension revenue (in billion CNY) ……………...……………………………………45

vii

Acknowledgement
Firstly, I would like to express my sincere gratitude to my supervisor Dr. Fiona MacPhail
for her continuous support of my graduate thesis and related research, for her patience,
motivation, and immense knowledge. She consistently allowed this thesis to be my own work,
but steered me in the right direction whenever she thought I needed it. Her guidance helped me at
all times during of the research and writing of this thesis. I could not have imagined having a
better advisor and mentor for my Master of Arts program.
Besides my thesis supervisor, I would like to thank Dr. Paul Bowles, who provided me an
opportunity to join his research project as a Research Assistant. By participating in this project, I
had the opportunity to access useful literature and information for my research. I am sincerely
grateful to him for sharing his illuminating views on a number of issues related to the project and
my research.
My sincere thanks are also extended to my thesis supervisory committee members, Dr.
Balbinder S. Deo, and Dr. Baotai Wang, for their insightful comments and encouragement, and
also for the hard questions which led me to widen my research and understanding from various
perspectives.
I thank my classmates and friends, Ananya Bhattacharya and Michelle Metzger, for their
time helping me revising my thesis. I also would like to thank my boyfriend, Yunke Li, for his
mathematics and emotional support when I met difficulties during my graduate study. Last but
not least, I would like to thank my family: my parents for supporting me emotionally and
financially throughout writing this thesis and my life in general.

viii

Chapter 1: Introduction
1.1 Overview
After an average GDP growth rate around 10% during 1980-2008 in China, GDP growth slowed
down to 7.4% in 2014 and will likely to drop to 6.9% in 2017 (World Bank 2015a). In
contemporary China, as a high middle-income country, the export-driven and investmentoriented economic growth strategies are facing difficulty in providing sustainable economic
growth (Kai Guo and Papa N’Diaye 2009; Gangming Yuan 2013). Firstly, after the global
financial crises, the sluggish recovery in global demand makes it a challenge for China to rely on
export-driven economic growth strategies. Secondly, the increase in labour cost in China is
reducing its comparative advantage in labour-intensive industries compared to lower wage
economies. Further expansion of the manufacturing industry has become an insufficient strategy
to support a sustainable increase in global market share for China. 1 The industrial transformation
towards technology-intensive production or towards higher domestic value-added production has
become a challenge for Chinese manufacturers who engage in labour-intensive production.
More sustainable economic strategies are required to enhance economic growth along with
social harmony. In recent years, Chinese government policies have continued transitioning to an
economic system driven by domestic consumption instead of exports (National People's
Congress of the People's Republic of China 2011). One of the ways to raise domestic
consumption is to improve social welfare, which will reduce precautionary savings and support
economic growth (Guo and N'Diaye 2009). The other way to enhance domestic consumption is
to raise people’s income. Recent policy guidance on achieving higher income levels emphasize
China has the highest investment rate in the world, which is around 40% of GDP between 2004-2008, and
investment is predominantly in the manufacturing sector (30%), the infrastructure sector (18%) and the real estate
sector (23%) (Guo and N'Diaye 2009).
1

1

technological upgrading by shifting policy preference from labour-intensive industries towards
technology-intensive industries. For example, China’s five-year plan for 2011-2015 emphasizes
technological progress and innovation. One of the important goals in this plan is to climb up the
technological ladder through means such as biotechnology, new materials, new information
technology, and high-end manufacturing (National People's Congress of the People's Republic of
China 2011). Moreover, China has set the innovation target at 3.3 patents per 10,000 people
(National People's Congress of the People's Republic of China 2011). This shows the
government’s determination to maintain prosperity and to shift Chinese economy to a highincome one.
This climbing the technology ladder strategy raises my concern for the implications of the
strategy for decent work in the manufacturing sector. The manufacturing industry in China has
enjoyed a very fast development since the opening up of China in 1979. In 2011, the Chinese
manufacturing sector surpassed the manufacturing sector in the United States and has become
the world’s largest producer of manufacturing goods (Karel Eloot, Alan Huang and Martin
Lehinch 2013). In addition, as an important foundation for economic and social development, the
manufacturing sector has become the main driver of urbanization and provider of employment in
China. For years, Chinese manufacturing has relied on a low cost labour force, and export tax
rebates and other preferential policies. The label of “made in China” is mainly concentrated in
low-tech production such as footwear and clothing. Nowadays, however, this label can be easily
found in high-tech industries as well, such as aircraft components, smart phones, and medical
equipment. This phenomenon raises my curiosity about whether shifting from low-tech products
towards high-tech products makes manufacturing workers better off.

2

1.2 Research goals and methodology
The purpose of this research is to identify the impacts of technological upgrading on decent work
in the manufacturing sector in China. This research mainly focuses on how technological
upgrading affects wages in the manufacturing sector. Different technological upgrading
indicators are used in this research to examine their impacts on wages; therefore, the type of the
technological upgrading that raises workers’ wages will be identified.
This research also addresses the following questions:
1. Has there been technological upgrading in China?
2. Has there been an improvement in decent work in China?
3. Does everyone benefit from improvement in decent work equally?
4. Does technological upgrading improve decent work in the manufacturing sector?
In order to answer the research questions, the following methodologies are adopted.
Firstly, a literature review is used to identify the definition of key concepts. For example, in
Chapter 2, by using a literature review, the definitions of technological upgrading and decent
work are identified. Secondly, based upon these definitions, indicators of key concepts are
identified in empirical research publications, which facilitates building an appropriate database in
this research. Thirdly, descriptive analysis is used to answer questions 1, 2, and 3. For instance,
in Chapter 3, some technological upgrading indicators are used to examine if there has been
technological upgrading in seven selected provinces between 2002 and 2014. Lastly,
econometrics analysis is used to examine the relationship between technological upgrading and
decent work (which is captured by wage in this research).

3

1.3 Importance of this research
This research topic is important for two main reasons. Firstly, this research relates to the
national development goal as expressed in the Five Year Development Plan 2011-2015.
Technological upgrading is emphasized as a national strategy in this five-year plan and is
intended to help China achieve its high-income goal (National People's Congress of the People's
Republic of China 2011). Secondly, this research examines decent work which relates to one of
the United Nation’s Sustainable Development Goals (SDGs) (United Nation 2015). Moreover,
work is the main factor affecting people’s standard of living.
The research contributes to the literature in three ways. Firstly, the research contributes to
the analysis of decent work which remains fairly limited in China. I apply the ILO decent work
concept to analyze trends and patterns in decent work in seven provinces, although the analysis is
limited by data availability. Secondly, the research contributes to the analysis of the relationship
between technological upgrading and job quality and expands the existing scope of analysis
beyond single indicators of technological upgrading, such as single indicators such as R&D
expenditure or FDI, and single labour market indicators, such as employment and wages. This
research analyses ten indicators of technological change and multiple indicators of job quality or
decent work. More specifically, correlations between technological upgrading and decent work
indicators are examined, along with regression analysis of the impact of technological upgrading
on wages in the manufacturing sector. Thirdly, the methodological approach taken here to
investigate the relationship between technological upgrading and decent work in China may be
useful to researchers interested in these questions in other country specific contexts.

4

1.4 Outline of this research
The thesis is arranged into five chapters. The second chapter is a review of the literature about
the definition of technological upgrading, the relationship between enterprise ownership and
technological upgrading, the definition of decent work, and the implication of technological
upgrading on decent work.
The third chapter provides a descriptive analysis of technological upgrading and increase in
decent work in selected provinces. Decent working time will be analyzed in the last part in this
chapter as a supplement to the decent work discussion.
Chapter 4, the econometric modeling of wage determination, contains a review of
macroeconomic theories of wage determination and an econometrics model to estimate the
relationship between technological upgrading and wage.
Finally, Chapter 5 presents conclusions, recommendations, limitations and avenues for
future research.
1.5 Key definitions
Key definitions are provided below and they are fully discussed with references to the literature
in the following Chapter. Technological upgrading refers to the outcomes of adopting new
embodied or/and disembodied technology including process upgrading, production upgrading,
functional upgrading, chain or inter-sectoral upgrading, knowledge and knowledge-creating
activities.
Disembodied technological upgrading refers to knowledge and knowledge-creating
activities arising from research, invention and development.
Embodied technological upgrading refers to increased productivity arising from the efficient
use of new or higher quality capital.

5

Decent work refers to work that is productive and delivers a fair income, it involves security
in the workplace and social protection for families, better prospects for personal development
and social integration, freedom for people to express their concerns, organize and participate in
the decisions that affect their lives and equality of opportunity and treatment for all women and
men.
Productive employment refers to employment yielding sufficient returns to labour to permit
the worker and her/his dependents a level of consumption above the poverty line.

6

Chapter 2: Literature review
In this chapter, the literature on technological upgrading and decent work, with particular
reference to China, is selectively reviewed in order to guide the methodology and analysis in this
thesis. The chapter starts with an examination of the concept of technological upgrading and its
contribution to economic growth. The second section examines the evidence to ascertain whether
technological change has occurred in China. The meaning of decent work is discussed in the
third section and the relationship between technological change and decent work is assessed in
the fourth section. The implications of the literature for the thesis are summarized in the final
section.
2.1 Technological upgrading
2.1.1 Importance of technological upgrading
Technological upgrading is of importance to national income growth. On the way to achieving
national income growth, a developing country is likely to face the “middle-income trap”. The
middle-income trap has been defined by Breda Griffith (2011, 39) as “a situation where a
middle-income country is failing to transition to a high-income country due to rising costs and
declining competitiveness”. Many studies have highlighted that China is experiencing pressure
that could result in this problem (OECD 2013b; Wing Thye Woo 2012; Hongbin Cai 2014). For
example, Chinese GDP growth has declined from 10.4% in 2010 to 7.4% in 2014 and will likely
drop to 6.9% in 2017 (World Bank 2015a). Increasing labour productivity, promoting innovation
and accelerating industrial restructuring and upgrading are suggested as the solutions to the
“middle-income trap” (Liqun Jin 2015).
There are many reasons that the middle-income trap occurs. When a growing economy is at
the early stages, a large population with low wages makes a low-income country competitive.
7

Thus, a developing country (like China) may transition from a low-income country to a middleincome one (Hernriue Schneider 2013) through a low-wage, labour-intensive development
strategy. According to Homi Kharas and Harinder Kohli (2011), a middle-income country may
be trapped because, firstly, as income increases, so do the wage costs; low-technology and
labour-intensive manufacturing loses its comparative advantage to lower-wage economies.
Secondly, the rural labour surplus declines overtime contributing to wage increases as an
economy grows towards a high-income stage; thirdly, imported foreign technology starts to have
diminishing marginal returns. Meanwhile, the middle-income country still cannot compete with
high-technology and skill-intensive economies.
In order to escape the middle-income trap, suitable strategies are required after a nation has
achieved high middle-income status. Researchers emphasize that policy-makers need to
concentrate on total factor productivity (TFP) based development or technological upgrading
(Kharas and Kohli 2011; Barry Eichengreen, Donghyun Park and Kwanho Shin 2012; Jelle van
den Berg 2013; Eva Paus 2012).
Given the emphasis on technological upgrading as a strategy to overcome the middleincome trap, this thesis examines the relationship between technological upgrading and decent
work. Decent work is defined below and in general, the relationship between technological
change and employment or work is important given that for the majority of people, employment
is the major determinant of their economic well-being. The thesis does not examine the direct
connection between technological upgrading and economic growth.
2.1.2 Definition of technological upgrading
In this section, the definition of technological upgrading and its classification as well as its
measurement will be reviewed. Technological upgrading is an abstract concept that cannot be

8

measured directly. The ways in which technological upgrading is measured in the literature is
also reviewed in order to inform the variable selection and facilitate later analysis in this thesis.
Technological upgrading is used to describe the overall progress of invention and
innovation as well as diffusion of technology. This term generally has the same meaning as
technological development, technological achievement, technical progress and technical change.
Technological upgrading is the primary cause of economic growth, and it involves an increase in
productivity, the improvement in product quality, and/or the development of new products.
Technological change is adopted if it increases productivity.
Technological change or total factor productivity (TFP) is typically measured by a Cobb
Douglas production function presented as: Y=ALα K1-β (where Y is output, and a function of two
inputs, namely labour (L) and capital (K), and technological change or total factor productivity
(A); and α and β refer to the output elasticities of labour and capital and capture the returns to
scale. Dale Weldeau Jorgenson and Zvi Griliches (1967, 249) state that the “[i]dentification of
measured growth in total factor productivity (TFP) with embodied or disembodied technical
change provides methods for measuring technical change, but provides no genuine explanation
of the underlying changes in real output and input”. While the TFP correctly measures the shift
in production possibilities and can be traced back to an initial shift in technology, it is still
necessary to consider the induced effect as the contribution of capital growth (OECD 2010).
Therefore, this research does not aim to measure TFP growth, but it should be clear that TFP
growth needs to be considered as a proxy of technological upgrading in the literature review.
Technological progress arises from embodied and/or disembodied technological change.
Embodied technological upgrading arises from the efficient use of new or higher quality capital
such as machinery, equipment and components that incorporate new technology. Disembodied

9

technological change refers to knowledge and knowledge-creating activities arising from
research invention or Research and Development (R&D), technical expertise, or technology in a
way that does not necessarily involve the purchase of machinery and equipment (Charles
Kennedy and Anthony P. Thirlwall 1972; Renuka Mahadevan 2003; Shuzo Ueda and Kazuo
Ogawa 2012). Disembodied technological progress captures scientific and technological
innovations or R&D activities. However, Alberto Gabriele (2002, 334) goes beyond this
definition:
In fact, technical progress also encompasses the web of imitative and adaptive
changes-in the realms of production organization, product design, materials and
energy consumption, procurement, sales and distribution, management,
finance, administration, and other economically relevant activities–which result
in higher productivity and jointly foster the progressive climbing of the
technological ladder and a more favorable positioning in the international
division of labor.
Technological upgrading has a broader meaning in Gabriele’s (2002) definition above, which
goes beyond new technology captured in capital and R&D; it identifies specific types of
knowledge-creating activities affecting production (for example, product design) and reflects
upon a technological ladder in relationship to a global production system.
Technological upgrading is also the main driving force of industrial upgrading.
Technological upgrading facilitates an industry to produce better quality, higher value-added,
more environmentally friendly and more competitive products. In terms of the definition of
industrial upgrading, it also indicates the connections with technological upgrading. Carlo
Pietrobelli and Roberta Rabellotti (2006, 1) state, “[u]pgrading is usually defined as the ability to
make better products, to make products more efficiently, or to move into more skilled activities.”
Also, Pietrobelli and Rabellotti (2006, 1) refer to “innovating to increase value-added” as being
part of upgrading. Moreover, Peter Gibbon and Stefano Ponte (2005, 87-88) define industrial
upgrading as “the possibility for (developing country) producers to move up the value chain,
10

either by shifting to more rewarding functional positions or by making products that have more
value-added invested in them and that can provide better returns to producers”. 2
From these definitions and discussion, I realize that “better products”, “innovating to
increase value-added”, “making products more efficiently” and “move up the value chain” are
associated with new machinery purchasing and/or knowledge creating activities. By making
connections between the definition of technological upgrading and industrial upgrading, the
number of technological upgrading indicators are expanded.
Moving up the value chain is classified as technological progress in this research. Different
paths of value chain upgrading have been identified by John Humphrey and Hubert Schmits
(2002) and each of the paths indicates technological upgrading. Four value chain upgrading paths
are defined as follows:
1. Process upgrading: transforming inputs into outputs more efficiently by
reorganizing the production system or introducing superior technology; 2.
Product upgrading: moving into more sophisticated product lines (which can be
defined in terms of increased unit values); 3. Functional upgrading: acquiring
new functions (or abandoning existing functions) to increase the overall skill
content of activities; 4. Chain or inter-sectoral upgrading: firms of clusters
move into new productive activities (Humphrey and Schmits 2002, 1020).
Embodied technological upgrading is associated with paths 1, 2, and 4, as in these paths, the
purchasing of new and better equipment and machineries are necessary, in order to produce more
efficiently, increase unit values and inter-sectoral upgrading. Disembodied technological
upgrading is associated with paths 1, 2, 3, and 4, as all the paths require new knowledge
understanding or knowledge creating activities. Under most of the circumstances, both types of
technological upgrading take place to contribute to the upgrading. Therefore, it is normal for
indicators to be classified under both embodied and disembodied technological upgrading

2 A value chain is a chain of activities that a firm operating in a specific industry performs in order to deliver a
valuable product or service for the market (Michael E. Porter 1985).

11

categories. The strict distinction between embodied technological upgrading and disembodied
technological upgrading is unnecessary for this research. For example, product upgrading could
be associated with disembodied technological upgrading and/or embodied technological
upgrading. No matter which classification the indicator belongs to, it still captures technological
upgrading.
For this research, technological upgrading is defined as the outcome of adopting embodied
or disembodied technology including: process upgrading, production upgrading, functional
upgrading, chain or inter-sectoral upgrading, following Humphrey and Schmits (2002). Given
the definition, a connection between selected technological upgrading indicators and this
definition will be made below.
Disembodied technological upgrading refers to knowledge and knowledge-creating
activities arising from research, invention and development. It also includes production
organization, product design, management and financial activity optimization, technical
expertise, or technology in a way that does not necessarily involve the purchase of machinery
and equipment incorporating new technology. Based upon those definitions, I classify some
indicators that capture disembodied technological upgrading such as: R&D expenditure, the
number of patents, FDI technology spillover, improved management, and increased skill level. 3
Embodied technological upgrading or progress refers to increased productivity arising from
the efficient use of new or higher quality capital. According to the definition, indicators of
embodied technological upgrading are identified as new capital goods purchasing, new
3 Technology spillover has been defined as the external effects of technological change (Philippe Aghoin and Peter
Howitt 1992; Gene Grossman and Elhanan Helpman 1991; Paul Romer 1990). Theories emphasize two aspects of
technology change: 1. the marginal cost for an additional agent to use the technology are negligible; 2. the return to
technological investments is partly private and partly public. These two characteristics indicate that technological
investments offer benefits to not only investors but also other individuals, and these external effects are called
technology spillover (Keller 2004).

12

equipment purchases, and increased productivity. Indicators that involve both embodied and
disembodied technological upgrading include new products, FDI, product value-added, product
sophistication level, and high-tech products.
2.2 Evidence of technological upgrading in China
By using the definition and categories of disembodied and embodied technological upgrading,
the evidence on technological upgrading in China is evaluated in this section. In general, the
evidence supports the hypothesis, that technological upgrading has occurred in the
manufacturing sector in China.
2.2.1 Evidence of disembodied technological upgrading
Most academic research uses R&D expenditure (or ratio of R&D expenditure to sales, or ratio of
R&D expenditure to GDP) as the indicator to capture technological upgrading (Fredrik Sjöholm
and Nannan Lundin 2013; Jaejoon Woo 2012; Xiaolan Fu and Yundan Gong 2011; Xiaolan Fu
and Carlo Pietrobelli 2011; Ganeshan Wignaraja 2012; Miaojie Yu 2011; Nicholas Bloom,
Mirko Draca and John Van Reenen 2011). Innovations or patents are also frequently used as
proxies for technological upgrading (see Fu and Pietrobelli 2011; Wignaraja 2012; J. Woo 2009).
J. Woo (2012) finds that China has the second largest stock of human resources in science
and technology in the world. He also finds that in China, the R&D intensity has grown beyond
1% of GDP.4 More specifically, R&D expenditure as a share of GDP was nearly 0.5% in 1990s,
then increased to 1% in 2000 and further to 1.5% in 2007 (J. Woo 2012). OECD (2013a) reports
that the R&D/GDP ratio was almost 2% in 2012. Moreover, the Medium and Long-Term
National Plan aims to lift the R&D intensity to 2.5% of GDP (OECD 2013a). In addition, the

4 R&D intensity indicates the non-personnel R&D expenditure per scientist or engineer.

13

number of patents that belong to the Triadic Patent Families registered by Chinese researchers
has increased more than 7 times from 2000 to 2007 (Xiaolan Fu and Pietrobelli 2011).5
For a growing country like China, learning new technology from developed economies is
the quickest manner to update its own technology level. It is widely accepted that international
trade and FDI are crucial channels for technology transfer from one country to another (David T.
Coe and Elhanan Helpman 1995; Amy Jocelyn Glass and Kamal Saggi 1998, 1999; Wolfgang
Keller 1998; UNCTAD 2004; Camal Saggi 2002; Lei Zhu and Bang Nam Jeon 2007). Empirical
evidence indicates that Chinese firms are learning and benefiting from MNC and/or FDI
technology spillover (Sjöholm and Lundin 2013; Ping Lin, Zhuomin Liu and Yifan Zhang 2009;
Yiying Zhu 2010).
In general, foreign investment and foreign firms facilitate technological upgrading in China.
For instance, Sjöholm and Lundin (2013), based upon a survey of Chinese manufacturing firms
over the period 1998-2004, found that FDI has positive effects on technology spillover. 6
Moreover, Lin et al. (2009) have used a large panel data set covering a sample of manufacturing
firms from 1998-2005. Their results indicate that FDI has positive vertical technology spillover
effects on Chinese domestic firms. 7 Both Sjöholm and Lundin (2013) and Lin et al. (2009), use
their own methodology to capture the FDI spillover effect, as it cannot be obtained directly from
any database, therefore, FDI spillover effect will not be used as a technological upgrading
indicator in this research. Furthermore, Y. Zhu (2010) suggests, by 2002, nearly 90% MNC

5 A Triadic Patent Family refers to a set of patents registered in various countries (i.e. patent offices) to protect the
same invention and this indicator is measured as a number (OECD ilibrary n.d.).
6 Sjöholm and Lundin (2013) define the FDI spillover effect as technology externalities from FDI that might affect
productivity in domestic firms.
7 Lin et al. (2009) define the FDI horizontal spillover effect as the external productivity effect on domestic firms in
the same industry; the FDI vertical spillover effect refers to domestic firms’ productivity improvements because of
forward or backward linkages with foreign invested firms.

14

subsidiaries adopted core technology in local production and by 2008, there were about 1200
R&D centers set up in China by MNCs.8
2.2.2 Evidence of embodied and disembodied technological upgrading
Based upon the definition of embodied technological change, the following indicators are
identified in this research as proxies of embodied technological upgrading: new capital goods
purchases (such as capital imports and capital investment). There are many indicators which
involve both embodied and disembodied technological upgrading such as the product valueadded, FDI (when the investment goes into both R&D activities and machinery purchasing) and
the product sophistication level. It is plausible to expect that when more domestic value is added
in exports, more profit is earned. Therefore, better equipment, more skillful workers and more
sophisticated production procedures are needed to facilitate a value chain upgrading. Many other
indicators have been widely used to illustrate technological change such as: an export
sophistication index, value of capital equipment imports, and export value chain upgrading or
value-added (J. Woo 2012; Stephen Young and Ping Lan 2010; Robert Koopman, Zhi Wang and
Shang-Jin Wei 2008; Bin Xu 2010; Yu 2011; Peter K. Schott 2008).
J. Woo (2012) finds that capital equipment has been largely imported in China over the
period 1995-2008 and this is associated with a shift in the export structure towards high-tech
categories. Precise calculations about Chinese high-tech exports were conducted by Koopman et
al. (2008) for Chinese manufacturing exports; they observed that the domestic value-added as a
share of the total value of exports increased from 47.1% in 1997 to 51.3% in 2002. Evidence of
technological upgrading is also reflected in the share of high-tech exports. The R&D intensive
exports increased from 3% in 2000 to 13% in 2008 (Fu and Pietrobelli 2011). Wignaraja (2012)
8 “Core technology is a generic term used to describe the key technology or components used in a product. These
key components are vital for the product to work, and in many cases, are patent-protected” (WikiAnswers n.d.).

15

discovers that electronics have increased in the share of total exports from 5.8% to 9.7% between
2001 and 2009. He also finds that exporting firms have higher technology index scores than nonexporters in this industry. During the same period, China’s share of world automobile exports
increased from 0.6% to 2.8%, facilitated by the technology diffusion and market network of
MNCs (Wignaraja 2012).
However, the technology level in China is still low. For instance, Schott (2008) discovers
that although China’s export products have a very high overall sophistication level, the vertical
sophistication turns out to be exceptionally low. The “overall sophistication” of Chinese exports
indicates the overlap between Chinese exports categories and developed economy exports
categories; the “vertical sophistication” of Chinese exports indicates the level of value-added in
the products (Schott 2008). J. Woo (2012) indicates that the measured overall technological level
is much lower than the technological sophistication of exports. This implies that even though
much of the export merchandise belongs to high-tech categories, they are at a very shallow level
or in other words, they have very low domestic value-added. Besides, Chinese workers share a
meager portion of the value from the processing industry. Kenneth L. Kraemer, Greg Linder and
Jason Dedrick (2011) find, for example, for each iPhone 4 or iPad (16GB) that Apple sells at 549
USD and 424 USD respectively, there is only 10 USD or less paid to Chinese workers.
2.2.3 Enterprise ownership and technological upgrading
The purpose of this section is to discover whether the type of enterprise ownership plays a role in
technological upgrading. The hypothesis is that different enterprise ownerships have different
levels of contribution towards technological upgrading. More specifically, I will argue, following
Fu and Gong (2011), Yan Liang (2008) and Albert Guangzhou Hu (2014), that, foreign-invested
enterprises are more likely to be the leading power of technological upgrading, while domestic-

16

invested enterprises are less likely to lead technological upgrading in China. Enterprise
ownership types are described below in order to aid the analysis of the relationship between
enterprise ownership and technological upgrading.
There are three main enterprise ownership types: domestic-invested enterprises (DIEs),
foreign-invested enterprises (FIEs), and Hong Kong–Macao-Taiwan-invested enterprises
(HMTs) (National Bureau of Statistics of P.R.C. 2006). See Figure 1 for the enterprise ownership
structure in China. The first group includes state-owned-enterprises (SOEs), collective-owned
enterprises, shareholding cooperative enterprises, associate ownership enterprises, private
enterprises, limited liability corporations, shareholding corporations, and other types of
enterprises. Generally, the SOEs have higher growth in medium and high technology industries
while the other domestic-invested enterprises grow fastest in low and medium technology

Figure 1. Ownership Structure in China
Ownership Stucture

Domestic-invested
Enterprises (DIEs)

Foreign-invested
Enterprises (FIEs)

HongKong-Macao-Taiwaninvested Enterprises
(HMTs)

State-Owned Enterprises

Other Domestic-invested Enterprises
Source: figure compiled by author based upon information in National Bureau of Statistics
of P.R.C (2006)

17

industries (Fu and Gong 2011).9 The FIEs and HMTs are more likely to grow in high-tech
sectors because a great portion of the high-tech processing and assembling type exports are
produced by them (Liang 2008).
Technological progress and SOEs
In the past, SOEs were considered as the leaders of R&D, but the situation has changed in the
last decade. Hu (2014) has found that over the period 1995-1999, the government-funded sectors
(including SOEs) undertook more R&D than other domestic-invested enterprises, but after 1999,
other domestic-invested enterprises surpassed the government sector. Anming Zhang, Yimin
Zhang and Ronald Zhao (2003) state that SOEs have significantly lower R&D efficiency, but
second highest R&D intensity compared to non-state owned enterprises. They explain that
although SOEs spend more money on R&D activities than others do, the effect of their
expenditure on productivity is insignificant. These results are consistent with Hu (2001) who
finds the direct contribution of government R&D to firm productivity is statistically
insignificant. Therefore, the SOEs or government funded research sectors are less likely to be the
leader of technological upgrading in China.
Technological progress, FIEs and HMTs
In general, FIEs and HMTs contribute to technological upgrading in China. However, whether
they have positive effects on domestic R&D activities is still under debate as outlined below.
Some of the economics literature suggests that FIEs contribute to technological progress in
China. For example, Sjöholm and Lundin (2013) have conducted a study on a large number of
Chinese manufacturing firms over the period 1998-2004 and have found FDI is positively related
9 See Fu and Gong (2011, 1224) for the classification of high-technology industry, medium-high technology
industry, and low-technology industry. For example, high-tech industry including medical and pharmaceutical
products, electronic and telecommunications etc., medium-high technology industry including special purposes
equipment, transport equipment etc., medium-low technology industry contains ordinary machinery, artifact and
other manufacturing etc., low-technology industry contains printing and record media reproduction, textile industry
etc.

18

to technology spillover. Lin et al. (2009) used a large panel data set covering all state-owned and
non-state owned manufacturing firms from 1998-2005, and the results suggest that FDI has
positive vertical technology spillover effects on Chinese domestic firms. Moreover, J. Woo
(2009) has investigated the effects of FDI on TFP growth in a large sample of countries
(including China) over the period 1970 to 2000. He finds a significant positive effect on TFP
from FDI and proposes that if the FDI is from OECD countries the effects will be more
substantial. Furthermore, Zhu (2010) reports that by 2002, nearly 90% of MNCs subsidiaries had
adopted their core technology in local production and in 2008, and there were about 1200 R&D
centers set up in China by MNCs. Zhu (2010) also argues that the establishment of high-class
R&D will provide good opportunities for local businesses in terms of learning and facilitating
local innovation ability, thus creating a positive system for domestic R&D.
However, there are also findings indicating that FDI has a negative effect on R&D in
domestic firms. For instance, Lin et al. (2009) find negative spillovers from HMT originated
FDI, and positive from OECD countries in China; Fu and Gong (2010) find foreign invested
firms have a significant negative effect on R&D of local firms over 2001 to 2005 in China. In
other countries, similar cases have been observed, which indicates that a country should never
totally rely on imported technologies. Taking Belgium as an example, Reinhilde Veugelers and
Peter Vanden Houte (1990) find FDI has a negative effect on domestic R&D. Subash Sasidharan
and Vinish Kathuria (2011) addressed the relationship between FDI and domestic R&D in India
during 1994-2005; their findings indicate that foreign equity participation acts as a disincentive
for investment in R&D in most cases. The reason for this trend is that the use of foreign
technology makes local firms rely on imported technology leading to crowding out of local R&D
(Fu and Pietrobelli 2011). In addition, the negative effect of FIEs on domestic R&D activities

19

occurs when there is a poor technology absorptive capacity (Fu and Petrobelli 2011). The
technology absorptive capacity depends on many aspects such as human capital, financial
conditions and R&D expenditures of the receiving county. Therefore, it is also important for a
country to enhance its technology absorptive capacity in order to reduce negative effects of
introducing foreign technology.
2.3 Definition of decent work
2.3.1 Definition of formal and informal employment
The International Labour Organization (ILO) (1993) defines formal employment as work in
the formal sector, and informal employment as work in the informal sector. The informal sector
is regarded as a group of household enterprises or unincorporated enterprises owned by
households. The formal sector refers to economic activity carried out by enterprises which are
typically registered with the government, are larger in scale (compared to informal enterprises)
measured by sales revenue or employees, and are monitored and/or taxed by the government.
However, the ILO has moved away from the enterprise-based definition of formal and
informal employment to a job-based view, which allows one to identify informal jobs within the
formal sector. As an enterprise-based definition cannot capture all the dimensions of informal
employment, informal employment now refers to work which is not regulated by legal or
regulatory frameworks, such as self-employment, wage employment which escapes regulation,
and non-paid work in an income-producing sector (ILO 2002). Although the job-based
dimension addresses the shortcomings of the enterprise-based definition, my research will only
use the enterprise-based definition, since the job-based data are absent in national, provincial or
municipal government statistical yearbooks.

20

In the Chinese context, the ILO (2003) has defined the informal sector as follows: 1. small
or micro-enterprises, which are normally private enterprises; 2. household-based workshops
involved in simple production and service activities; and 3. independent service providers. Based
on the ILO (2003) definition, Jinjun Xue, Wenshu Gao and Lin Guo (2014) classified the
informal sectors as: individual businesses (no more than 7 employees) and household sector. Ralf
Hussmanns (2004) suggests that the informal employment emphasizes jobs that remain
undeclared to the government to avoid paying social insurance and taxes, and supervision by
labour law, as well as to avoid meeting health and safety standards.
The formal sector refers to governmental and institutional organizations, SOEs, collective
enterprises and private enterprises (Xue, Gao and Guo 2014). In the past, jobs in the formal
sector were considered as formal employment or standard employment, but now there are
increasing numbers of formal sectors hiring non-standard workers. 10 Compared to those engaged
in the formal employment, non-standard workers have unstable working hour arrangements,
lower wages and lower social security. The government does not disclose the number of nonstandard workers employed in formal sectors. Therefore, even though the official data of people
employed in the formal sector has increased over time, this does not necessarily mean an
increase in the number of standard jobs.
My research focuses on the manufacturing industry. Normally, enterprises in the
manufacturing industry do not fall into the informal sector, because in order to survive in the

Standard employment refers to “adequately compensated full-time jobs that are fully covered by the social
security system, have no contractual time limitation and provide unity of work and appointment” (Marcel Garz
2013, 349). Non-standard work refers to job that fall outside the realm of standard work arrangements, including
temporary or fixed-term contracts, temporary agency or dispatched work, dependent self-employment, as well as
part-time work, including marginal part-time work, which is characterized by short, variable, and often
unpredictable, hours (International Labour Organization b).
10

21

manufacturing industry, an enterprise usually needs to produce at a large scale to bring down the
cost. Enterprises in the informal sector are unlikely to reach this scale.
This research will use the government statistical yearbooks as the data source, and the
informal sector is not included. By using the enterprise-based definition of formal and informal
employment, people who work in the manufacturing industry are considered as formally
employed. This research cannot answer whether the number of standard workers in the
manufacturing industry has increased or not, since the indicator is absent from government
statistical yearbooks. Formally, employed workers have access to social security provisions;
therefore, social security indicators could capture the change in decent work. The proxy for
social security includes the number of social insurance participants, number of employment
insurance participants or employment insurance participant ratio (the number of people who have
employment insurance divided by the number of employed people over the whole economy), and
total pension revenue.
2.3.2 Definition of decent work
Decent work, a concept initially developed by the International Labour Organization, has gained
wide support as indicated by its incorporation as the eighth goal in the United Nations
Sustainable Development Goals. Decent work has been defined as below:
Decent work sums up the aspirations of people in their working lives. It
involves opportunities for work that is productive and delivers a fair income,
security in the workplace and social protection for families, better prospects for
personal development and social integration, freedom for people to express
their concerns, organize and participate in the decisions that affect their lives
and equality of opportunity and treatment for all women and men (International
Labour Organization a).
Decent work is a higher quality of employment than formal employment, because formal
employment only requires the workers to be employed in the formal sector. Decent work

22

addresses fair income, social protection for families, workplace security, personal development,
freedom to express ideas, social integration and equality of opportunity. As the definition of the
decent work combines so many aspects (such as access to full and productive employment with
rights at work, social protection, and the promotion of social dialogue), measurement of decent
work from every aspect is a difficult task. Hence, this research is going to focus on only two
aspects of decent work: productive employment and social protection for families.
Productive employment is defined as “employment yielding sufficient returns to labour to
permit the worker and her/his dependents a level of consumption above the poverty line”
(International Labour Organization 2012, 3). Based upon the definition, whether the monetary
wage could support the worker’s annual consumption expenditure and the number of dependents
a worker could support becomes a method to identify decent work.
Decent work should provide social protection to workers in terms of supporting families
cope with unexpected risks, such as unemployment, illness, maternity, retirement and old age.
Therefore, the social security indicators such as, the unemployment insurance participation ratio
becomes a quantifiable indicator for decent work.
2.4 Implication of technological upgrading for decent work
The way in which technological upgrading relates to workers’ wages and employment requires
explanation. Technological progress has the potential to enhance productivity and living
standards, reduce poverty and shift labour from inefficient sectors towards more efficient ones.
For example, with the use of machinery, an economy is able to transform inputs into outputs
more efficiently. This is also called productivity growth, which has the potential to raise peoples’
real incomes through two main mechanisms. First, higher productivity means lower costs and if
these savings are passed on to consumers, it reduces the price of goods; therefore, consumers
23

will be able to purchase more goods and services (John R. Baldwin, Wulong Gu, Ryan
Macdonald, and Beiling Yan 2014). Second, technological progress may change the demand for
employment type and the demand for labour skills. More specifically, if technological upgrading
is capital-saving it leads to more employment, but if it is labour-saving then more workers will
be replaced by capital. The technological change may also be associated with a change in the
relative demand for high or low skilled labour. In addition, if technological upgrading results in
economic growth, even if it is labour-saving, there may be an increase in demand for labour. The
impact of technological upgrading on employment also needs to be examined country by
country, since different social, political, and economic situations in different areas will affect the
result.
Technological upgrading contributes to economic growth, which positively affects the
employment rate. In general, empirical research indicates that technology contributes to job
creation. For instance, John Van Reenen (1997) and Smolny (1998) discover that technology has
a strong and positive effect on job creation in developed economies such as the United Kingdom
and Germany. Moreover, with respect to China, Braunstein and Epstein (2002) find that from
1986 to 1999, FDI raised not only employment but also wages. Since FDI is a proxy for
technological upgrading, it can be inferred that technological upgrading generates employment in
China.
However, there is also evidence to suggest that the technological progress does not affect
positively impact employment in China. For instance, Sjöholm and Lundin (2010) have analyzed
the relationship between technological progress and employment based on large and mediumsized enterprises in Chinese manufacturing industry between 1998 and 2004. They adopt “the
ratio of science and technology (S&T) expenditures to sales” as their technological upgrading

24

indicator (Sjöholm and Lundin 2010, 17). Based upon this indicator of technological change,
they conclude that there is no significant effect of technological progress on employment. As the
technological upgrading indicator used by Sjöholm and Lundin (2010) is different from the
indicator used by Braunstein and Epstein (2002), the difference of the results could be due to the
different indicators of technological upgrading used. Changwen Zhao (2010) argues that Sjöholm
and Lundin used data from 1998-2004, which is the early stage of Chinese economy. Hence, the
comparative advantage of that stage is lower labour costs rather than technology (Zhao 2010).
Zhao’s argument is persuasive and explains why Sjöholm and Lundin (2010) do not find a
significant positive employment effect from their research. Since China is facing higher labour
costs in recent years, it is reasonable to expect that now technological progress may have a
positive significant effect on employment or wages.
In general, even if technological upgrading has a positive overall impact on the labour
market in terms of increased employment and wages, it may not benefit all workers equally
(Khalid Nadvi 2004). Peter Knorringa and Lee Pegler (2006) conclude that value chain

upgrading is not only increasing flexibility in the labour market but also increasing insecurity
and working hours of workers. Besides, they suggest that core workers in the key supplier firms
may enjoy improved employment conditions while others may not. Knorringa and Pegler (2006)
argue that wokers employed in global value chain (GVC) supplier firms face physical and
emotional pressures and stress of new work systems. 11 Workers engaged in higher value-added
activities or in firms that supply higher value buyers, gain higher income and have better
working conditions comparing with workers engaged in lower value-added activities. Nadvi
(2004, 28) argues that, “workers within GVCs, at all points, are increasingly vulnerable to
global value chain refers to different stages of the production process being located across different countries
to facilitate international production, trade and investments (OECD n.d.).
11 The

25

changing employment contracts and the increasing casualization of work”. However, Nadvi’s
view on casualization and vulnerability of work is narrow as he fails to recognize that the
vulnerablity of employment depends on whether a worker engages in high value-added activities
or low value-added activities. Workers employed at the upper stream of the value chain (for
example the R&D sector or design sector) will be less likely to experience vulnerablity in their
employment. On the other hand, the situation of workers engaged in low value-added activities
will depend on their skill level and learning abilities.
Although the distribution of benefits from technological upgrading is not equal among
workers, technological upgrading potentially enhances the working conditions and wages on
average or overall. Therefore, technological change may improve decent work through increases
in income and improvements in working conditions. By looking at the number of workers with a
labour contract helps identify the change in decent work as well, since labour contract offers the
opportunity to regulate the rights and the obligations of both the employee and the employer.
2.5 Summary
Based upon this review of the literature, it is reasonable to conclude that there has been
technological upgrading in China. By reviewing the relationship between enterprise ownership
and technological upgrading, I realize that ownership type does play an important role in
technological upgrading. However, this research cannot determine how exactly enterprise
ownership would affect technological upgrading, therefore, this ownership will not be considered
as technological upgrading indicator. The literature review offers methodologies to capture
technological upgrading, appropriate proxies for technological upgrading are listed as: R&D
expenditure, the number of R&D employees, FDI inflows, the value of mechanical imports or
exports, and the value of high-tech imports or exports.
26

By reviewing the meaning of decent work and its relationship with technological upgrading
in the second part of this chapter, I find that the evidence from various countries does not suggest
a strong positive effect of technological upgrading and decent work for all workers. In fact, how
technological upgrading affects workers depends on: workers skill level, the stage of an
economy, the labour market situation, government policies and so on. But it is clear that with
low-value-added industry transferred to underdeveloped areas, the wage levels and benefits to
workers are likely to increase in both developed and underdeveloped areas. Technological
upgrading also shifts the enterprise labour reqirement from low-skilled towards highly-skill.
Comparing with better-educated urban workers, less-educated migrant workers may have
difficulties finding skill-intensive jobs in developed areas. Therefore, migrant wokers will be less
likely to share in the benefits of technological upgrading. The literature review demonstrates
that decent work is comprised of a variety of dimensions and various indicators are used to
measure these dimensions. In this thesis, I will use average income relative to the poverty line
and annual consumption expenditure to reflect productive employment.
Based upon this literature review, two main gaps in the research are identified. Firstly, the
literature focuses on the impact of technological upgrading on productivity growth or
employment growth. I have not found literature which studies the relationship between
technological upgrading and decent work or wages, and my research is intended to fill this gap.
Secondly, research on decent work in China is very limited. This research will contribute the
analysis of decent work in the Chinese labour market.

27

Chapter 3: Technological upgrading and decent work in selected
provinces
3.1 Introduction
The purpose of this chapter is to assess the extent of technological upgrading and the
improvement of decent work in the seven selected provinces: Jiangsu, Zhejiang, Fujian,
Guangdong, Guangxi, Hainan, and Shandong. As discussed in the previous chapter, while an
assessment of the impact of technological upgrading on improving decent work is the intended
goal of this thesis, unfortunately, data to measure decent work are limited. Therefore, the
quantitative analysis will focus primarily upon employment and wages in the manufacturing
sector, and will be complemented with an analysis of selected indicators of social security. The
analysis in this chapter is intended to establish whether or not technological upgrading and
decent work have occurred which provides the foundation for the regression analysis in the
subsequent chapter on the association between technological upgrading and decent work. More
specifically, this chapter provides an analysis of the:
1. Evidence of technological upgrading and technology endowment in seven selected provinces;
2. Evidence of an increase in decent work in terms of the number of manufacturing employment,
manufacturing average annual wage, and social security indicators;
3. Correlation between technological upgrading indicators and decent work indicators;
4. Decent working time.
3.2 Methodology
The main methodological challenge addressed in this chapter is to find reliable and publicly
available data to capture the concepts of technological upgrading and decent work. The
28

literature review in the previous chapter provides definitions of these key concepts and highlights
the types of indicators that have been used in the literature to measure these concepts. I tried to
find data on similar indicators.
The data on technological upgrading and decent work in this chapter are collected primarily
from provincial statistical yearbooks. Most provincial statistical yearbooks for the most recent
decade can be accessed on provincial government websites, with the exception of data from
Hainan province, which must be obtained from the National Knowledge Infrastructure (CNKI)
website. The CNKI provides the photocopy edition of statistical yearbooks for all provinces in
China. I used both provincial government websites and the CNKI web to download statistical
yearbooks, and then manually enter the data to complete data set. However, some indicators
cannot be found consistently from provincial statistical yearbooks, such as the number of people
who participate in unemployment insurance programs. Therefore, the National Bureau of
Statistics website is also used to collect data for my database. The time period selection is based
upon the availability of two of the technological upgrading indicators, high-tech products
imports and exports, which starts from 2002. Although I realized there are other technological
upgrading indicators in both the provincial statistical yearbooks and National Bureau of Statistics
websites, such as the number of R&D achievements in large and medium enterprises, they are
not consistently reported annually. The short time period of the database becomes one of the
limitations of this research.
The empirical approach adopted here is to focus upon the change in technological
upgrading and decent work over the time period 2002-2014, as well as the variation among the
selected provinces. To take account of differences in the size of the economy and population
among the provinces, where appropriate, the data are presented relative to the size of the

29

province. After providing a descriptive analysis of the changes in technological upgrading and
decent work based upon several different indicators of each concept, the correlation between the
two concepts is presented. As discussed further in the section on the correlation, I would expect
that indicators of technological upgrading and decent work to be positively correlated.
These seven coastline provinces are selected for analysis because manufacturing enterprises
are clustered in coastline provinces. As their products are primarily destined for export, being on
the coastline facilitates the transport of these goods to overseas markets. Besides, due to their
special geographical location, coastal cities are an important window for foreign trade and
communication. The convenience of maritime transport attracts export-oriented and high-tech
firms to settle in the coastal cities. 12 Therefore, technological upgrading is more likely to take

Figure 2. Value of high-tech exports by region in 2013

Source: Ministry of Science and Technology of the PRC ( 2015, 33).
12 There are also high-tech firms located in inland provinces, however, inland provinces do not disclose data on the
value of high-tech products that are produced or exported in their statistical yearbooks. Future research could extend
the data to other provinces if reliable data resources are available.

30

place in coastal provinces. 13 Information from Figure 2 indicates that coastal provinces export
more high-tech products than inland provinces. For example, five of the seven selected provinces
exported more than 10,000 million USD high-tech products in 2013.
3.3 Evidence of technological upgrading in seven selected provinces
One way to identify technological upgrading in selected provinces is by examining the changes
in the technology proxies, as illustrated in Table 1. By comparing the number of each technology
indicator in 2002 and 2014, we can conclude that there has been technological upgrading
in each of the selected provinces. For instance, the actual FDI utilized in Jiangsu is 10.4 billion
USD in 2002, and it increased to 28.2 billion USD in 2014. However, the R&D employment in
large and medium (LM) enterprises (in ‘000s people) in Guangxi and Hainan slightly decreased
from 20.4 to 20.1 and from 6.2 to 3.4, respectively. This should not change the conclusion that
technological upgrading has occurred in these two provinces, since all other five technology
indicators have shown an increase.
By comparing the numbers in Table 1 horizontally, the numbers marked in bold indicate the
province with highest level of technology for the given year. For example, in 2014, Jiangsu has
the highest actual FDI utilized of 28.2 billion USD compared to other provinces in that year.
Based upon most indicators, Guangdong province and Jiangsu province have the highest
technology endowments compared to other provinces. This conclusion is based upon the
technology indicator expressed in absolute terms. Yet, comparison of the indicators in absolute
terms ignores the impact arising from the size of each province. For instance, while Guangdong
has the highest technology endowments for many indicators, this could be due to its large
13 The original plan for this research was to obtain data for all coastline provinces and cities, however, not every
coastline province and city publish statistical yearbooks and some provinces do not report consistent data for some
vital technological upgrading indicators. Therefore, in this research I have selected the seven coastline provinces for
which consistent data on indicators of technological upgrading exist.

31

population size and area. Hence, using the indicators in absolute terms is not enough. Using
indicators in relative form as presented in Table 2, we can see that Jiangsu has the highest
technology endowments rather than Guangdong. I would argue that the results from relative term
of numbers are more convincing in terms of understanding which provinces can be considered to
have a high technology endowment.

Table 1. Technology indicators in absolute terms, selected provinces, 2002 and 2014
Year Jiang Guang
Fu
Shan
Zhe
Guang Hai
su
dong
jian
dong
jiang
xi
nan
Actual FDI utilized
2002
10.4
4.3
6.5
3.2
0.4
0.5
13.1
(in billion USD)
2014
27.3
7.7
16.0
15.8
1.0
1.9
28.2
Mechanical and
2002
21.2
2.5
5.0
29.4
0.3
0.1
72.7
electronics exports
2014
221.5
40.4
56.2
112.5
10.7
0.4
428.6
(in billion USD)
High-tech exports
2002
12.1
3.2
1.1
1.4
0.0
0.0
31.0
(in billion USD)
2014
129.4
15.0
20.6
15.5
2.9
0.3
231.0
R&D expenditure in 2002
16.2
2.3
15.9
8.8
1.3
0.1
18.3
LM enterprises (in
2014
96.9
28.0 109.2
68.8
7.5
0.8
137.5
billion CNY)
R&D employment in 2002
158.1
104.1
25.6 158.8
45.0
20.4
6.2
LM enterprises (in
2014
544.9
141.7 342.3 298.9
20.1
3.4
598.9
‘000s)
Patents granted (in
2002
7.6
4.0
7.3
10.5
1.1
0.2
22.8
‘000s)
2014
180.0
37.9
72.8
188.5
11.4
1.6
200.0
Note: LM enterprise refers to large and medium enterprises; Chinese High-tech Products
Catalog 2006 defines high-tech exports (Guangdong High-tech Enterprise Association 2015);
mechanical and electronics exports are defined by Mechanical and Electronics Product Range
(Ministry of Commerce of the P.R.C 2008).
Source: compiled by author from provincial statistical yearbooks 2003 and 2015.

32

Table 2. Technology indicators in relative terms, selected provinces, 2002 and 2014
Year
Jiang Guang Fu
Shan Zhe
Guang Hai
su
dong
jian
dong jiang xi
nan
FDI per worker (USD)
2002 232.6 317.1 248.3 118.0 110.5
16.2
146.0
2014 591.8 441.2 291.2 241.5 425.3
36.2
353.5
Mechanical and
2002
36.4
46.8
20.8
18.0
16.1
13.4
78.1
electronics exports share
2014
63.4
33.1
35.2
37.2
30.5
17.7
68.0
(%)
High-tech exports share
2002
26.1
18.4
5.4
4.7
1.5
1.1
31.5
(%)
2014
35.8
13.3
14.2
5.7
11.8
7.5
38.3
R&D expenditure share
2002
1.2
0.5
1.6
1.1
0.5
0.1
1.7
in LM enterprises (%)
2014
1.5
1.2
1.8
1.7
0.5
0.2
2.0
R&D employment share
2002
3.8
1.9
5.9
4.3
3.6
4.4
7.3
in LM enterprises (%)
2014
5.4
6.1
8.1
8.5
2.6
3.6
9.9
Patents per 1,000
2002
0.2
0.2
0.1
0.4
0.0
0.1
0.6
workers
2014
4.2
2.9
1.4
1.1
0.4
0.3
5.1
Note: LM refers to large and medium enterprises; FDI per worker is the actual FDI utilized
divided by total number of employed workers over the whole province; Mechanical and
electronics export share is the value of mechanical and electronics exports divided by the value
of total exports; High-tech exports share is the value of high-tech exports divided by the value
of total exports; R&D expenditure share in LM enterprises is the R&D expenditure in LM
enterprises as a share of GDP; R&D employment share is the R&D employment as a share of
total manufacturing employment; Patents per 1,000 workers is the total number of patents
granted divided by total number of employed workers over the whole province (in 000’s).
Source: calculated by author from provincial statistical yearbooks 2013 and 2015; and National
Bureau of Statistics of P.R.C. (n.d.)

33

3.4 Evidence of an increase in decent work
3.4.1 Manufacturing employment
Manufacturing employment in urban units will be analyzed as an indicator of decent work. 14As
discussed previously, data in the provincial statistical yearbooks capture the number of
manufacturing workers with labour contracts. Therefore, workers without a labour contract but
working in the manufacturing sector are excluded from the manufacturing employment number;
as discussed further below, it is more likely to be the case that rural-urban migrant workers,
rather than workers with an urban identity, will be the workers without contracts. By signing a
work contract, workers are officially protected by the labour law in terms of their payments and
benefits, although there are problems associated with monitoring and enforcement. Since there is
the potential for contracts to be enforced, manufacturing work is considered as decent work.
In 2014, the total population in China reached 1.37 billion, and the urban population and
rural population are 0.75 billion (54.8% of total population) and 0.61 billion (45.2% of total
population), respectively (National Bureau of Statistics of P.R.C. 2015). In the same year, the
number of urban employed persons and rural employed persons are 0.39 billion and 0.37 billion,
respectively. Although there is not a large difference between the number of rural and urban
employed persons, these numbers can only reflect employed people registered with the
government (National Bureau of Statistics of P.R.C. 2015). I would argue that urban areas offer
more opportunities for wage employment than rural areas, and wage employment is more likely
to provide decent work than non-wage employment. In urban units, the number of people
employed in the manufacturing sector accounts for 28.7% of total employment in 2014, which is

14 The urban unit refers to the independent accounting legal entity which does not include private enterprises and individual

industrial and commercial households, including state-owned units, collective units, joint-stock enterprises, joint-stock
enterprises, foreign-invested enterprises and Hong Kong, Macao and Taiwan investment enterprises (Beijing Municipal Bureau
of Statistics n.d.) .

34

the largest sector in terms of employment (National Bureau of Statistics of P.R.C. 2015). This is
one of the reasons why my research focuses on the manufacturing sector in urban units.
The manufacturing sector in six of the seven selected provinces (including Jiangsu,
Guangdong, Fujian, Shandong, Zhejiang and Guangxi) is the largest industrial sector in terms of
employment, for urban units in 2014 (See Table 3). Going back to 2002, the manufacturing
sector has provided the largest number of jobs in urban units based on data from the provincial
statistical yearbook. This evidence suggests that the economy in these provinces is dominated by
the manufacturing sector, and this is followed by construction. The total number of workers
engaged in these two sectors account for more than 60% of total employment in all selected
seven provinces. This phenomenon also indicates that the investment-driven economic growth
pattern has not changed substantially over the 2002-2014 period being considered.

35

Table 3. Employment by industrial sector (number in ‘000s), urban units, selected provinces, 2014
Jiang
Guang
Fu
Shan
Zhe
Guang
Hai
su
dong
jian
dong
jiang
xi
nan
Total
15,128.2 19,732.9 5,599.5 12,663.0 11,026.8 4,014.6 979.4
Agriculture, forestry,
59.9
56.7
27.5
17.0
5.6
82.8
99.4
animal husbandry and
fishery
Mining and quarrying
115.8
29.9
23.3
712.0
8.6
38.0
6.9
Manufacturing
6,026.7 10,151.6 2,317.6 4,258.0
3,505.5
782.0
95.7
Production and supply of
179.2
308.0
84.0
237.0
134.2
140.6
21.7
electric power, gas and
water
Construction
4,047.3
1,494.1 1,064.4 1,775.0
3,294.5
609.6
64.4
Wholesale and retail
571.3
958.4
189.8
628.0
417.3
132.9
59.0
trade
Transport, storage and
480.4
854.0
55.5
497.0
326.9
209.1
53.3
postal services
Hotels and catering
160.8
370.6
247.0
156.0
138.4
50.1
63.9
services
Information transmission,
284.4
346.3
95.8
170.0
164.4
44.6
15.6
computer services and
software
Finance
266.3
431.5
118.7
388.0
379.6
117.6
30.3
Real estate
207.5
558.4
133.3
256.0
197.5
79.5
72.6
Leasing and business
301.5
606.1
97.4
216.0
284.1
110.7
24.5
services
Scientific research and
204.5
319.3
77.8
184.0
161.0
97.7
21.4
technical services
Water conservancy,
130.4
172.8
45.8
167.0
116.8
95.9
26.3
environment and
public facilities
management
Resident services and
36.3
74.9
13.7
31.0
26.2
8.2
4.2
other services
Education
903.4
1,245.0
464.4
1,203.0
700.8
618.6
126.3
Health care and social
424.1
596.8
187.4
587.0
416.3
298.9
55.7
work
Culture, sports and
72.3
115.2
34.6
71.0
72.1
33.6
11.8
recreation
Public administration and
656.1
1,043.3
321.5
1,110.0
677.0
464.2
126.4
social organizations
Source: compiled by author from provincial statistical yearbooks 2015.

36

According to the information in Figure 3, the number of people employed in the
manufacturing sector is large and increasing, in each of the provinces with the exception of
Guangxi and Hainan provinces. Guangdong and Jiangsu had a significant increase in
manufacturing employment in 2013. The provincial statistical yearbooks for these two provinces
in 2014 and 2015 do not indicate a change in statistical standards; hence, this substantial increase
cannot be explained.
Figure 3. Employment in manufacturing (in ‘000s), urban units, selected provinces, 20022014
12,000
10,000

8,000
6,000
4,000
2,000
0

Jiangsu Guangdong

Fujian

Shandong

Zhejiang

2002

2003

2004

2005

2006

2007

2009

2010

2011

2012

2013

2014

Guangxi

Hainan

2008

Source: compiled by author from provincial statistical yearbooks 2003 and 2015.

The manufacturing employment has increased not only in absolute terms, but also in
relative terms. Comparing those numbers between 2002 and 2014, the manufacturing
employment as a share of total employment in urban units has increased in Jiangsu, Guangdong,
Fujian and Zhejiang (see Table 4). The ratio has a slight decrease in Shandong and Guangxi and
a sharp decline in Hainan. This relatively high share indicates that the manufacturing industry
37

continues to play an important role in these provinces over last decade. Moreover, with the
development of the economy, the position of manufacturing sector is still increasing in terms of
employment. In 2014, the share of manufacturing employment in Guangdong accounts for 51.45%
employment in urban units.
Table 4. Share of manufacturing employment (%), urban units, selected provinces, 2002 and
2014
Jiangsu
Guangdong Fujian Shandong Zhejiang Guangxi
Hainan
2002
35.07
34.43
41.21
35.53
28.30
20.79
19.42
2014
39.84
51.45
41.39
33.25
31.79
19.48
9.77
Note: Share of manufacturing employment is the number of people employed in the
manufacturing sector divided by the total number of people employed in urban units times
100 percent.
Source: calculated by author from provincial statistical yearbooks 2003 and 2015.
3.4.2 Manufacturing wages
Productive employment emphasizes that a job should “yield sufficient returns to labour to permit
the worker and her/his dependents a level of consumption above the poverty line” (International
Labour Organization 2012, 3). Hence, the average annual wage and the average annual
consumption expenditure are two quantifiable indicators for decent work. Figure 4 indicates that
the manufacturing average annual wage in urban units in all the selected provinces exhibits an
increasing trend between 2002 and 2014. Guangdong province has the highest average annual
wage in the first six years (between 2002 and 2007). In 2008, Jiangsu’s average annual wage
surpassed Guangdong’s and reached 58,409 CNY (approximately 9,513 USD) in 2014. 15 In
terms of the lowest average annual wage, Hainan’s urban manufacturing average annual wage is
at the bottom for most of the given years.

The conversion is based on the average exchange rate in 2014 of USD 100 = CNY 614.28 (National Bureau of
Statistics of P.R.C. n.d.)
15

38

The poverty line has been set by Chinese government at 2,300 yuan per year (1.33 USD per
day in purchasing power parity (PPP) terms) in 2011 (The Economist 2011).16 By comparing the
lowest average annual wage (9,388 CNY, Hainan in 2002) and the poverty line, we can conclude
that the manufacturing average annual wage in all the selected provinces is beyond this poverty
line. Therefore, workers employed in the manufacturing sector are likely not to trapped in
absolute poverty.
Figure 4. Average annual nominal wages (CNY) in manufacturing, urban units, selected
provinces, 2002-2014
80,000
60,000
40,000
20,000
0

2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014

Jiangsu

Guangdong

Fujian

Zhejiang

Guangxi

Hainan

Shandong

Source: compiled by author from provincial statistical yearbooks 2003-2015.

While the average manufacturing wage is higher than the absolute poverty level, it is lower
than the average annual wage in many other sectors. In 2005, the average manufacturing wage is
lower than the average annual wage across all industrial sectors (in urban units), in all selected
provinces (Figure 5).17 The manufacturing wage in Zhejiang is the worst case among all the
selected provinces in 2005, since it accounts for only 64% of the average annual wage in urban

16 The new global poverty line was updated to $1.90 per day in PPP term in October 2015 (World Bank 2015b).

17 Year 2005 is selected in this part is due to the absence of some statistical yearbook chapters in my database,
therefore, I cannot compile the figure back to year 2002.

39

units. In comparison, Guangxi’s manufacturing industry offers a relatively fair income in 2005,
since the average manufacturing wage reaches 95% of the overall average annual wage, for
urban units.
Manufacturing workers are better off in 2014 compared to 2005, since the gap between the
urban wage averaged across all sectors and the average manufacturing wage had declined (see
Figure 6). In 2014, the average manufacturing wage reached more than 80% of average annual
wage in urban units, in all the selected provinces. The average manufacturing wage in Jiangsu
performs the best, as it reaches 95% of the urban average annual wage. Zhejiang showed the
largest improvement in the manufacturing-urban average annual wage ratio, as it increased from
64% in 2005 to 83% in 2014.
In terms of the annual consumption expenditure, the average manufacturing wages in all
provinces are able to support the workers’ annual consumption expenditures, based upon data
from Figures 5 and 6.18
From Figure 5, I find that in Shandong and Zhejiang provinces, manufacturing workers
need to spend 75% of their wage to live in the urban area in 2005. In Jiangsu, Guangdong and
Fujian, the annual consumption expenditure accounts for more than 50% of manufacturing
worker’s wage. While in Guangxi and Hainan, workers only need to spend 4 4% and 46%,
respectively, to support their basic annual consumption expenditure.
Data in Figure 6 indicates that manufacturing workers in 2014 use less than 50% of their
wage to cover living expenditure, except for workers in Zhejiang. Shandong province had the
largest improvement as the expenditure-income ratio decreased from 75% to 42%. Combining
Figure 5 and Figure 6, we can conclude that manufacturing workers are able to save more money
18 The urban per capita consumption expenditure includes food, tobacco and liquor, clothing, living, daily

necessities and services, transportation and telecommunication, education, culture and entertainment, health service
and medical care, other necessities and services (National Bureau of Statistics of P.R.C. 2013).

40

in 2014 than in 2005. This means that workers’ living standard has improved given the
possibility of greater savings. However, this analysis does not take account of the number of
people that each worker is supporting and how this might have changed the workers’ living
standard.

Figure 5. Average annual nominal wage (manufacturing sector and all urban units) and per
capita consumption expenditure (CNY), selected provinces, 2005
30,000
25,000
20,000
15,000
10,000
5,000
0

Jiangsu Guangdong

Fujian

Shandong

Zhejiang

Guangxi

Annual urban per capita consumption expenditure

Manufacturing annual average nominal wage
Urban units annual average nominal wage
Source: compiled by author from provincial statistical yearbooks 2006.

41

Hainan

Figure 6. Average annual nominal wage (manufacturing sector and all urban units) and
per capita consumption expenditure (CNY), selected provinces, 2014

70,000
60,000
50,000
40,000
30,000
20,000

10,000
0

Jiangsu Guangdong

Fujian

Shandong Zhejiang

Guangxi

Hainan

Annual urban per capita consumption expenditure

Manufacturing annual average nominal wage
Urban units annual average nominal wage
Source: compiled by author from provincial statistical yearbooks 2015.

The number of dependents that a manufacturing worker can support is not an indicator that
can be found in the statistical yearbooks. Therefore, I assume that the basic annual consumption
expenditure for an urban dependent equals the urban per capita annual consumption expenditure.
This assumption helps me to measure the number of dependents that a manufacturing worker can
support and is a method to assess productive employment. The results are illustrated in Table 5.
In 2005, a manufacturing worker’s wage in Guangdong, Fujian, Shandong, and Zhejiang
provinces was insufficient to support even one dependent. However, in 2014, the manufacturing
wage in all provinces had risen sufficiently that the average worker could support one dependent.
Manufacturing workers in Jiangsu, Guangdong, Fujian, Shandong, Guangxi and Hainan
42

provinces would still have some savings after the consumption expenses associated with one
dependent person. From this perspective, one could argue that there has been movement toward
decent work in the manufacturing sector in recent years. However, it must be kept in mind that
the analysis focuses upon the average annual wage and has not taken into account the wide
distribution of wages around the average annual wage; thus, workers in the lower tail of the wage
distribution may not be able to cover annual consumption expenditures of dependents.

Table 5. The number of dependents that can be supported by the average manufacturing wage
Jiangsu
Guangdong Fujian Shandong Zhejiang Guangxi Hainan
2005
1.0
0.5
0.6
0.3
0.3
1.3
1.2
2014
1.5
1.2
1.1
1.4
0.9
1.8
1.6
Note: the number of dependents is calculated as the average manufacturing wage divided by per
capita annual consumption expenditure minus one; and is based on the assumption that a
dependent’s annual consumption expenditure equals the average urban per capita annual
consumption expenditure.
Source: compiled by author from provincial statistical yearbooks 2015.

3.4.3 Social security
Social security mechanisms are designed to reduce income losses due to social risks, such as
unemployment, sickness, accident, aging, work place injury, and so on. The availability of social
security or insurance programs is an important component of decent work. China’s social
security system is comprised of pension insurance, medical insurance, unemployment insurance,
work injury insurance, and maternity insurance and a housing fund. Two social security
indicators are selected in this research: the unemployment insurance participant ratio (total
number of unemployment insurance participants divided by total number of workers) and the
government pension revenue. Unfortunately, data pertaining to these social security provisions
are only available at the provincial level aggregated across all industrial sectors and thus, I

43

cannot analyze social security proxies for the manufacturing sector alone. Nonetheless, changes
in the indicators demonstrate the changing of social security at provincial level.
Unemployment insurance benefits help individuals and families by assisting them through
times of unemployment without a dramatic change in lifestyle. This means that workers losing
jobs will have some income for basic expenditures.
Figure 7 illustrates that the unemployment insurance participant ratio increased during the
past decade, although the ratio in all provinces is less than 50%. 19 The unemployment insurance
participant ratio is only 8-9% in Guangxi which is the lowest ratio of all selected provinces and
the ratio does not show any improvement between 2002 and 2014. Data indicate that more than
half of the total labour force in each province has no unemployment insurance protection, so they
are more vulnerable to income loss if they become unemployed. The absence of unemployment

Figure 7. Employment insurance participation ratio (%), selected provinces, 2002-2014
50
40
30
20
10
0

Jiangsu
2002
2009

Guangdong
2003
2010

Fujian
2004
2011

Shandong
2005
2012

Zhejiang
2006
2013

Guangxi
2007
2014

Hainan
2008

Note: the unemployment insurance participation rate is the total number of the
unemployment insurance participants divided by total number of employed people.
Source: compiled by author using National Bureau of Statistics of P.R.C. (n.d.) (selected
years).

Unemployment participation ratio is calculated as: the number of workers with unemployment insurance divided
by the total number of workers (%)
19

44

insurance does not only adversely affect the unemployed workers, but also negatively impacts
the dependents and community of the unemployed workers. Since without a regular income,
spending would fall sharply, resulting in hardship for the person’s dependents, as well as, local
businesses and their workers, which depend upon the worker’s expenditures.
Figure 8 illustrates the pension revenue in each province over 2002 and 2014. The growth
rate of pension revenue in Fujian, Guangxi and Hainan is much lower than that of Guangdong,
Jiangsu, Shandong and Zhejiang. The graph still indicates that pension revenue in each province
has increased over time. The growth of pension revenue is good for social stability. Pension
benefits provide basic living income to retired workers, so that they can have a sense of security.
Figure 8. Pension revenue (in billion CNY)
200
150
100

50
0

2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013
Jiangsu

Guangdong

Fujian

Zhejiang

Guangxi

Hainan

Shandong

Note: pension revenue is the total public pension revenue collected by government.
Source: compiled by author using National Bureau of Statistics of P.R.C. (n.d.) (selected
years).

The pension schemes in China are organized in five different types, and the formula
determining a person’s pension income depends on his or her pension scheme (Lijian Wang,
Daniel Béland and Sifeng Zhang 2014). The first type is called the basic old-age insurance
45

system for enterprise employees; the second is the basic old-age insurance system for public
institutions employees; the third one is old-age insurance system for urban residents; the fourth is
the new type of rural social endowment insurance; and the last one is a pension system designed
for civil servants (see Wang, Béland and Zhang (2014, 26) for a detailed discussion about these
five schemes). The pension income in each scenario in 2011 is illustrated in Table 6.
Employment in the manufacturing sector is classified as enterprise employment; therefore,
the basic old-age pension income is 18,095CNY in 2011 on average. Comparing with the
average annual consumption expenditure in urban areas at the national level, which was
15,160.89 CNY in 2011 (National Bureau of Statistics of P.R.C. 2012), the basic old age pension
income can provide a basic living income for a retired manufacturing worker. However, their
pension income only accounts for 44% of their wage, which means that their living standard
could drop sharply after retirement if there is no other source of retirement income. Since the
data in Table 6 reflects a calculation for all workers in China in 2011, the average pension
income within a particular province could be different from the data in this Table. Unfortunately,
there are no data available in the statistical yearbooks to measure the average pension income at
the provincial level; therefore, I am not able to duplicate a similar table for those selected
provinces in this research.
The pension income gap between each group of employment type is substantial. The civil
servants, who are also known as government officers, enjoy the highest pension income among
all five types of employment. The difference between their annual wage income and retirement
income is around 4,800 CNY, which means that their retirement living standard will not have a
significant difference compared with their service period. The pension income for public
institution employees reaches 26,107 CNY, which is around 57% of their average annual wage.

46

Therefore, their retirement living standard is lower than their employed living standard but is still
well above the absolute poverty line. Rural residents are entitled to only 943CNY annual
retirement income, which is lower than the poverty line (2,300 CNY per year). The retirement
living standard for rural people is therefore very low, and they have to rely on family members
and other sources of income. The old age insurance for urban residents is also lower than the
poverty line, and therefore, they also need to rely on other source of income to support retirement
life. The fairness of pension income in China is also a problem that should not be ignored
and future policy making should address it in order to enhance the social harmony.
Table 6. Basic data for China's pension income (in CNY per year and million people)
Pension scheme

Pension
income

Pension
contribution

Pension
demand

Working
generation's
average
annual wage
40,674.32

Number
of
retirees

Basic old-age insurance
18,095.82
7,753.63
6,634.10
for enterprise employees
Basic old-age insurance
26,107.21
8,839.50
6,634.10
46,092.01
for public institutions
employees
Old-age insurance for
1,347.25
251.35
6,634.10
23,979.20
urban residents
New type of rural social
943.77
192.04
3,978.17
9,833.14
endowment insurance
Pensions for civil
40,060.25
0
6,634.10
44,845.21
servants
Note: numbers in this table are average values except number of retirees.
Source: this table is reproduced from Wang, Béland and Zhang (2014, 31).

63.14
5.13
2.35
89.22
7.48

For workers in the manufacturing sector who have an employment contract, they may have
the potential to obtain access to a pension plan while they are working and receive an income
above the absolute poverty line when they are retired. However, there are still a great number of
manufacturing workers working without a contract. These workers will be unable to access to the
enterprise pension and thus, will need to rely on the pension benefits for rural and urban
47

residents described above which provide very low levels of pension income.
3.5 Correlation between technological upgrading and decent work
Based upon the graphs and the tables above, evidence suggests that there has been technological
upgrading and an improvement in decent work in all the seven selected provinces. Moreover, we
can find that when a province has higher technology endowments, it also has a higher
manufacturing wage, employment, and social security than those provinces with lower
technology endowments. For example, Jiangsu is the most technology intensive province based
upon the data in Table 2 and it offers the highest manufacturing wage since 2008 (Figure 4); the
manufacturing wage reaches 95% of the urban average annual wage and average annual
consumption expenditure is only 40% of workers’ wages. The correlation between technology
indicators and decent work indicators (both in absolute terms) further demonstrates this
relationship, as shown in Table 7. The wage, employment and social security are positively
correlated with technology indicators, and therefore, it is expected that technological upgrading
will improve decent work. The number of observations is ninety-one which covers seven
provinces over thirteen years.
Table 7. Correlation between technology indicators and decent work indicators, across seven
provinces and years 2002-2014
FDI

High-tech Mechanical and
R&D
exports electronics exports Patents employment R&D expenditure

Wage
0.48
0.42
0.47
0.70
0.58
Manufacturing
employment
0.81
0.78
0.83
0.71
0.88
Employment
insurance
0.85
0.88
0.92
0.76
0.91
Pension
revenue
0.81
0.70
0.75
0.88
0.92
Source: table compiled by author using provincial statistical yearbooks 2003-2015.

48

0.65
0.90
0.90
0.96

3.6 Decent working time and employment contracts
3.6.1 Overview of decent working time in China
The need for working time to be both healthy and safe is fundamental. Jon C. Messenger (2004)
has proposed five significant dimensions of “decent working time” as: “[w]orking time
arrangements should be healthy; ‘family-friendly’; promote gender equality; advance enterprise
productivity; and facilitate worker choice and influence over their hours of work” (Messenger
(2004), as cited by Jean-Yves Boulin, Michel Lallement, Jon C. Messenger, and François
Michon (2006, v)). For instance, regular long working hours, night work and holiday work are
neither preferred by workers nor healthy for workers.
The following discussion of decent working time will mainly address the working hours,
rather than other dimensions of decent working time, due to limited data. At the global level, the
weekly working-hour limit differs from country to country or sector to sector, and “[i]n the past,
work organization in both the manufacturing industry and the service sector was based on fulltime employment and the 8-hour day associated with it, and on the 48-hour and, later, the 40hour week” (Boulin et al. 2006, 45). In China, the standard working hour is regulated by the
labour law and “State Council Decree No. 174”. The core content has suggested that the standard
working hour should not be more than 8 hours a day and no more than 40 hours a week, and with
at least one day off per week (General Office of the State Council of the P.R.C 1995). Therefore,
the benchmark for decent working time for this research is 8 hours per day and 40 hours per
week.
Since there are no provincial level data available for working time, this part of discussion
will provide an analysis of working time at the national level. The national level analysis will
provide a general view and understanding of working time and the decent work distribution in
China.
49

Overtime working is prevalent in urban China. According to Table 8, the average weekly
working time for Chinese urban employees is 46.3 hours, which is 6.3 hours more than the
standard working time. If the standard of 8-hour working day and at least one day off per week is
applied, workers will have to work 6 days a week in most sectors. Among all sectors, only those
people who engage in Agriculture, Forestry, Animal Husbandry and Fishery do not work
overtime. Although the average weekly working hours in the manufacturing sector is not the
highest among all the sectors, it is still 8.2 hours longer than standard working time and 1.9 hours
longer than the average working hours.
Table 8. Average weekly working hours of employees, urban areas, 2012
Average
Male
Average
46.3
47.1
Hotels and Catering Services
51.4
51.9
Wholesale and Retail Trade
50.2
50.9
Construction
49.4
49.8
Resident Services and Other Services
49.1
50.0
Transport, Storage and Postal Services
48.8
49.6
Manufacturing
48.2
48.2
Information Transmission, Computer Services
47.8
47.7
and Software
Leasing and Business Services
46.2
46.6
Real Estate
45.9
46.8
Mining and Quarrying
45.7
46.2
Culture, Sports and Recreation
45.6
46.0
Health Care and Social Work
44.1
45.1
Water Conservancy, Environment and
43.8
43.7
Public Facilities Management
Scientific Research and Technical Services
43.4
43.2

Female
45.2
50.8
49.7
47.1
48.1
45.3
48.0
47.9
45.7
44.5
44.2
45.1
43.4
44.1
43.6

Production and Supply of Electric Power, Gas
43.3
43.5
42.9
and Water
Finance
43.2
43.6
42.8
Education
42.5
43.2
42.0
Public Administration and Social Organizations
41.8
42.1
41.3
Agriculture, Forestry, Animal Husbandry and
38.2
40.7
35.7
Fishery
Source: compiled by author from Desheng Lai, Dahu Meng, Changan Li and Qi Wang (2014,
433) (in Chinese).

50

3.6.2 Working time and resident registration type
The average working hours depends not only on the sector of employment, but also the type of
Hukou a worker holds. Based upon Chinese houshold registration system, also known as the
“Hukou System”, Chinese citizens are divided into two types: agriculture citizens (rural people)
and non-agriculture citizens (urban people). 20 This household registration system limits rural
people from obtaining decent work and many scholars see it as the major source of hardship for
rural-urban migrant workers living in cities (Kam Wing Chan and Li Zhang 1999; Peter
Alexander and Anita Chan 2004; Au Loong-Yu and Nan Shan 2007). The Hukou System in
China allows people to work in their hometown legally, however, in order to migrate to other
areas it is necessary to apply for different kinds of permits to work and live in that area (Chan
and Zhang 1999).
For workers with a non-agriculture Hukou, urban citizens, almost 50% of them could work
at the standard (40 hours per week) and 45.4% of them work overtime. Turning to workers with
an agricultural Hukou, rural citizens, who are working in urban areas, as shown in Table 9, 63%
work more than the 40 hours per week, while only 37% of work less than or equal to the standard
working weekly hours. Among the 37% agriculture citizens, part of them are engaged in
Agriculture, Forestry, Animal Husbandry and Fishery sector, which does not require them to
work over 40 hours per week in the first place. We can deduce that if an agriculture citizen
migrates to urban area, he or she will have a high chance of working more than the standard 40
hour level.

20

See Chan and Zhang (1999) for a more detailed disscussion of the Hukou System and rural-urban migration.

51

Table 9. Weekly working hours of employed persons by household registration (%), urban
areas
Household
Urban
1-8
9-19
20-39
40
41-48
48
Registration
Employed
Hours
Hours
Hours
Hours
Hours
Hours
Persons
Total
100
0.5
1.3
8.4
36.0
19.9
33.9
Agriculture
100
0.8
2.3
13.1
20.6
20.4
42.8
Non-Agriculture
100
0.3
0.4
4.3
49.5
19.4
26.0
Source: compiled from National Bureau of Statistics of P.R.C. (2014, Table 3-30).
Going further, as shown in Table 10, in 2014, rural-urban migrant workers work about 8.8
hours a day and more than 85% of them work overtime. 21 From Table 10, it is also noticed that
that around 40% of migrant workers work more than 8 hours a day.
Table 10. Migrant workers’ working hours
Average Daily Working Hours
Proportion of Workers Working Over 8 Hours
per Day (%)

2012
8.7

2013
8.8

2014
8.8

39.6

41.0

40.8

Proportion of Workers Working Over 44
84.4
84.7
85.4
Hours per Week（%）
Source: compiled by author from National Bureau of Statistics of P.R.C (2015c) (in Chinese)
http://www.stats.gov.cn/tjsj/zxfb/201504/t20150429797821.html.
If migrant workers have a permanent employment contract or a contract longer than 1 year,
they are more likely to work less or equal to 8 hours per day than those with a work contract less
than 1 year or temporary contract (see Table 11). To be more specific, if a worker has a formal
permanent contract, he or she will have 43.1% chance to work less or equal to 8 hours a day,
compared with a temporary term contract worker who has only 29.22% chance to do so.

Rural migrant worker refers to an agriculture citizen who engages in non-agriculture sectors in local area or
outside registration area for more than six months (National Bureau of Statistics of P.R.C. 2015a).
21

52

Table 11. Work contract type and daily working hours, migrant workers, urban areas
≤8
9~10
11~12
>12
Permanent contract (%)
43.1
38.1
15.0
3.8
Contract ≥ 1 year (%)
39.7
38.6
16.2
5.5
Contract < 1 year (%)
31.0
47.1
15.9
6.0
Temporary contract (%)
29.2
48.6
16.0
6.1
Source: compiled by author from Desheng Lai, Dahu Meng, Changan Li, and Qi Wang (2014,
205) (in Chinese)
The scale of migrant worker in 2014 reached 274 million and 31.3% of them work in
manufacturing sector (National Bureau of Statistics of P.R.C. 2015c). From Table 12, in 2014,
around 62% of rural migrant workers do not have an employment contract, and only 21.2% have
an employment contract for 1 year or more. Therefore, in 2014 almost 80% of rural migrant
workers not only engage in unstable employment (with a employment contract duration less than
one year or without a contract) but also work long hours. Comparing with the scale of migrant
workers, the work permit issue ratio is very low (Qiang Li and Zhuang Tang 2002). This
situation also blocks migrant workers from obtaining work contracts. Manufacturing workers
data that has been used in this research are for workers with work contracts. The provincial
government statistical yearbooks disclose neither the scale of migrant workers in the
manufacturing sector nor the duration of their work contracts. Lack of information disclosure
blocks further disscussion of decent working time at the provincial level.

Table 12. Migrant workers with employment contracts (%)
Non-fixed term
Less than 1-year
1 Year and over
Without a
contract
contract
contract
contract
2013
13.7
3.2
21.2
61.9
2014
13.7
3.1
21.2
62.0
Source: compiled by author from National Bureau of Statistics of P.R.C. (2015c)
http://www.stats.gov.cn/tjsj/zxfb/201504/t20150429_797821.html.

53

There are not many rural-urban migrant workers that obtain decent work or decent working
time in the manufacturing sector. Even though there has been a technological upgrading and an
increase in decent work, not everyone shares equally from these changes.

54

Chapter 4: Econometric modeling analysis
4.1 Methodology
The purpose of this chapter is to empirically assess the impact of technological upgrading on
decent work, using the average wage in the manufacturing sector as a proxy for decent work, for
seven coastline provinces in China. The methodological approach is to design a regression
model and estimate the model using data for China for 2003 to 2014. More specifically,
econometric modeling will be used to estimate the relationship between technological upgrading
and the wage in manufacturing sector. The model is developed based upon a review of the
macroeconomic theories of wage determination and the application of these theories to wage
determination in the Chinese context. Appropriate literature was found using a set of search
terms and databases including EconLit and Google Scholar. The search terms included the
following: technological upgrading and wage; technical change and wage; macro-level wage
model; macro-level and wage determinants; wage growth model; trade and wage; and
globalization and wage. A positive relationship between technological upgrading and the
manufacturing wage is expected, after controlling for other factors affecting the wage.
Before estimating the model, potential econometric issues that I could encounter in the
estimation are highlighted so that I can take account of potential biases and procedures for
addressing the problems.
The remainder of this chapter will review macroeconomic theories of wage determinants,
outline the model to be used in this thesis, identify econometrics issues that may occur, discuss
the data, and analyze the results.

55

4.2 Macroeconomic theories of wage determination
The determinants of wages in macroeconomic theory are generally well known and advanced in
standard economics textbooks. Olivier Blanchard and David Johnson (2013), for example,
identify the wage determinants as the expected price level, labour market conditions, and all
other variables that affect the outcome of wage setting, such as bargaining power, employment
insurance, and economic structural change. The following equation (1) captures the main wage
determinants:
𝑊 = 𝑃𝑒 𝐹(𝑢, 𝑧)

(1)

where W refers to the nominal wage, 𝑃𝑒 refers to the expected price level, 𝑢 represents the
unemployment rate (which captures labour market conditions) and z captures all other factors
that affect wage setting (Blanchard and Johnson 2013, 173). Labour market conditions are
expected to impact wages and for example, the lower the unemployment rate, the higher the
wages. Price expectations, likewise, would influence wages such that the higher the expected
prices, the higher the wage. This research will consider technological upgrading as one of the “z”
factors and investigate whether technological upgrading has an impact on wages at the
macroeconomic level.
The relationship between technology and wages is generally positive and researchers have
explored the mechanisms underlying the relationship. On the productivity side, because of
technology upgrading, the whole society produces more goods and services within a certain time,
which reduces people’s living cost and increases their real income (Baldwin, et al. 2014). On the
worker side, learning new technology and new skills makes them more competitive, which
increase their wages as well. For instance, Wilfred Edward Graham Salter (1960) suggests that
technological upgrading is directly linked with increasing labour productivity, and this could be
attributed to labour itself, such as greater working skills and efforts. Since industry can improve
56

its productivity by hiring higher-skilled labour compared to other industries, firms are willing to
pay the higher wages necessary to attract these wages (Derek Leslie 1985). Moreover, Jeremy
Greenwood and Mehmet Yorukoglu (1997) state that rapid technological advancement should be
associated with a rise in wages. This occurrence is due to the adoption of new technologies
involving a learning process, which offers workers opportunities to gain working skills and
eventually contribute to a rise in wages. Therefore, it can be concluded that technological
upgrading has a positive impact on labour wage growth.
When it comes to the relationship between wages and technological upgrading, it should be
noted that the majority of research is based upon microeconomic level firm data drawn from
sample surveys. 22 Using microeconomic level data helps identify the character of technical
upgrading and predetermines the applicable wage models. However, the micro level model
cannot explore the effects of technological upgrading on overall wage growth. More importantly,
the results may be biased by the sample selection.
The macroeconomic theory of wage determination is appropriately used when the focus is
on analyzing the impacts across the entire economy. Therefore, the following section will review
some macroeconomic wage models.
4.3 Review of macroeconomic wage models
The models reviewed below are selected because firstly, they all use macroeconomic level data,
such as national level data or provincial level data, to estimate their model. Secondly, they have
some aspect related to this research: for example, Elissa Braunstein and Gerald Epstein (2002)

Derek Leslie (1985) uses two-digit industry level data; Ann P. Bartel and Nachum Sicherman (1997) use National
Longitudinal Surveys of firms; Sanna-Mari Hynninen, Jari Ojala, Jaakko Pehkonen (2013) adopt individual level
data of workers.
22

57

focus on China; António Afonso and Pedro Gomes (2014) use technology as one of the
independent variables.
Braunstein and Epstein (2002) construct a macroeconomic-level wage model to determine
the impact of FDI on wages in China. The main hypothesis behind the model is that FDI
increases labour demand, and thereby increases wages. Besides, foreign enterprises tend to have
higher productivity and pay higher wages, and FDI may raise wages through a spillover effect on
domestic firms. The FDI spillover effect in their research refers to effects beyond what FDI
directly contributes to domestic investment such as managerial know-how, international
connections, and advanced production processes that enhance domestic overall productivity
(Braunstein and Epstein 2002). They used province-level panel data for China’s 29 provinces
between 1986 and 1999. Their data are drawn from the statistical yearbooks of individual
provinces and regions in China, supplemented by data taken from the China Statistical Yearbook.
Their wage data set focuses on the average annual wage across all sectors. Based on their
econometric model, they find a statistically significant and positive effect of FDI on wages. Their
regression equation is:
lnwit = αi + β1 lnIit + β2 lnFDIit + β3 lnTit + β4 lnLFit + β5 lnq it + β6 L2it + timetrend + εit (2)
where 𝑤𝑖𝑡 is the annual wage for a particular province i, in year t; α = provincial fixed effect; I
=value of total investment; FDI = value of foreign direct investment; T = total foreign trade
(imports + exports); LF = total available labour force, defined as the population fifteen and over;
q = productivity, which is captured by GDP/employment; L= liberalization variable, the ratio of
state sector output to all industrial output; timetrend = a time trend has been added to control for
uniform shocks; and ε = random error.

58

Comparing with the theoretical model (1), we can find that in equation (2) the labour market
condition is captured by the total available labour force (LF); but the price expectation has not
been considered as an independent variable, which is a weakness of their model. Additionally,
they considered investment, FDI, trade, productivity, market liberalization and time trend as
variables as “z” factors. The limitation of their model is that they do not have a theoretical
framework section for their modeling, which makes their model less convincing. The advantage
of their research is that their model is easy to follow and relevant for this thesis, since they use
macroeconomic data.
The second example of a wage determination model is a combined micro and
macroeconomic one, proposed by Elissa Braunstein and Mark Brenner (2007). Their research
focuses on the impacts of FDI on male and female wages in urban China for 1995 and 2002.
Their model is a cross-section model shown below:
lnwij = α + X ij β + Zjδ + εij

(3)

where: 𝑤𝑖𝑗 indicates wage of person i in province j; 𝑋𝑖𝑗 =a vector of individual-level variables for
males and females; 𝑍𝑗 = a vector of province-level variables; and 𝜀𝑖𝑗 = error term.
The individual-level (microeconomic level) variables include education, age, ethnic
minority, communist party member, ownership of firm, and so on. At the microeconomic level,
the wage is determined by worker’s characteristics, while at the macroeconomic level it is
affected by labour market conditions and economy-wide conditions. At the provincial level
(macroeconomic level), independent variables include: total investment as a share of GDP
(affects the demand side of the labour market); SOE output as a share of gross industrial output;
cumulative GDP growth; FDI as a share of total investment; and trade (exports plus imports) as a
share of GDP.

59

Braunstein and Brenner (2007) combine macroeconomic-level data and microeconomiclevel data, which is the key innovation of their study. They acknowledge that their independent
variables are at different levels of aggregation, which causes a downward bias when estimating
regression coefficient standard errors. The issue has been addressed by estimating standard errors
clustered on provinces (Braunstein and Brenner 2007). However, like Braunstein and Epstein
(2002), this model does not contain price expectation as one of the independent variables as well,
which is a weakness of this model. As I am interested in a macroeconomics model of wage
determination, reviewing this model is helpful. However, I will not be able to adopt the
microeconomic variables incorporated by Braunstein and Brenner (2007), since I do not have
access to the microeconomic data. Therefore, I will try the macroeconomics level variables and
abandon microeconomics level ones. As a result, my research will not be able to capture
individual characteristics of workers and its implications on wages.
The third model, proposed by Afonso and Gomes (2014), examines the interactions between
private and public sector wages per employee in OECD countries. The model is:
p

p

p

p

ωit = αi + δp ωit−1 + θpX it + πp Zit + kp Eit−1 + μit

(4)

where the i (i=1…, N) denotes the country; t (t=1, …, T) denotes the period; 𝛼𝑖 stands for the
individual effects to be estimated for each country i; 𝜇𝑖𝑡 indicates the disturbances which are
𝑝

independent across countries; and 𝜔𝑖𝑡 is the growth rate of real compensation per employee in
𝑝

the private sector. 𝑋𝑖𝑡 is a vector of macroeconomic variables that might be endogenous to
private sector wage growth. This vector includes: public sector employees’ real compensation;
the growth rate of the consumer price index (CPI); growth rate of total factor prductivity (TFP);
change in the unemployment rate; change in urbanization rate; growth rate of per worker average
hours worked; growth rate of trade; change in the tax wedge and growth rate of public sector

60

𝑝

employment. 𝑍𝑖𝑡 is a vector of institutional exogenous variables such as union density changes
and central bank independence. 𝐸𝑖𝑡−1 is defined as the percentage difference between public and
private sector wages.
This model incorporates the lagged wage as an independent variable to capture price
expectations. As price expectation is an important variable in the theoretical framework, it will
be adopted in my model. In addition, the TFP is an independent variable in this model, which
indicates that technological upgrading has been regarded as an important factor for wage growth.
However, because of the data consistency and availability problem of my data source, there are
some variables in their model (for instance, urbanization rate and growth rate of average hours
worked per worker), which will not be used in my modeling.
The following macroeconomic level wage models will be discussed in a comparatively
general manner, since they do not closely align with my research purpose (general wage
determination). They mainly focus on specific wage determination modeling, for example,
Phillips curve wage determinations, or wage flexibility. Torbjorn Eika (1993) has built a stylized
wage equation to estimate wage determination. The model is based on the Phillips curve, where
wage growth depends on the tightness of the labour market and the growth in import prices and
GDP.23 Independent variables in Eika’s model are the private consumption deflator, output per
head, import price, indirect tax rate, direct tax rate, employment tax rate and unemployment rate.
Petri Böckerman, Seppo Laaksonen and Jari Vainiomäki (2010) advance a wage model to
examine the micro- and macro-flexibility of wages in Finland. In terms of the macroeconomics
side, they have regressed the average change in nominal wages on unemployment, productivity

The Phillips curve is a historical inverse relationship between rates of unemployment and corresponding rates of
inflation generated in an economy (Irving Fisher1973). However, the Phillips' original curve describes the behavior
of money wages. William Phillips (1958) observed an inverse relationship between money wage changes and
unemployment in the British economy over 1861-1957.
23

61

growth, expected inflation and industry-level bargains (the number of employer and union
bargaining rounds). Mark E. Schweitzer (2007) examines wage flexibility in macro level data in
UK. The independent variables for his model are the lagged wage, the unemployment rate, time
dummy variables, regional dummy variables, productivity growth and inflation.
There are two theoretical frameworks to capture price expectations, namely, adaptive and
rational expectations, although adaptive expectations is the most common form in empirical
wage determination research. Before moving to the difference between adaptive and rational
price expectations, it is noted that in general price expectation can be measured by using inflation
expectation (Ignazio Visco 1984).
Adaptive price expectations captures the idea that people’s expectation of inflation is based
upon the latest known inflation rate (Blanchard and Johnson 2013). “By the 1970s, evidence is
that people formed expectations by expecting this year’s inflation rate to be the same as last
year’s” (Blanchard and Johnson 2013, 234). In the empirical literature, adaptive price
expectations are typically proxied using some form of the lagged inflation rate in wage
determination models (James E. Pesando 1975).
In contrast, rational price expectations considers how price expectations are formed based
upon expectations about current or future events. John F. Muth (1961) proposed a rational price
expectations framework to capture price expectations based upon the assumption that economic
agents use all available information and expectations are affected by the structure of the entire
economy system market. Clearly, expectation would be affected by so many factors, rather than
the latest known price or inflation rate only.
In the empirical work, I have chosen to use the adaptive price expectation framework.
Firstly, empirically estimating how price expectations are formed is beyond the scope of this

62

research. Secondly, unlike other market economies, where inflation is usually caused by private
and public consumption, the national economy including inflation is controlled by government in
China (Yasheng Huang 1999). In other words, the Chinese government can surprise economic
agents through unexpected monetary policy changes. Thirdly, adopting adaptive price
expectations and modelling it through the use of lagged inflation rates is consistent with other
empirical wage models in the literature. For example, Böckerman, Laaksonen and Vainiomäki
(2010) use expected inflation as their price expectation variable, which is measured by using a
three year moving average of realized inflation rates. Because of limited data, the three-year
moving average of realized inflation rates method is not suitable for my research and therefore, I
choose the lagged inflation as my price expectation indicator.
In general, Braunstein and Epstein (2002), Braunstein and Brenner (2007), and Afonso and
Gomes (2014) all use the unemployment rate and productivity as independent variables. This
indicates that unemployment and productivity are important variables in macro wage models.
Including productivity as an independent variable in my modeling may cause a multicollinearity
problem. Since technological upgrading contributes to productivity, this makes technology and
productivity indicators highly correlated, and result in a multicollinearity problem.
Summary
Reviewing macroeconomic wage models helps me identify independent variables that need to be
included in my wage model. I will include:
1. A price expectation variable, such as the lagged inflation rate;
2. A labour market condition variable, such as GDP growth rate;
3. A technology variable, such as high-tech imports, patents granted and so on;

63

4. A labour market institutions, or institutional power variable, such as the ratio of state sector
output to all industrial output.
While it is common to include an indicator of labour market institutions such as union
density in the wage models, effective unions do not exist in China. The All-China Federation of
Trade Unions is not independent from government, and trade unions with the full right to
organize a strike or bargain for better wages and working conditions for workers do not exist.
Since employment in the SOE sector is more likely to offer decent work conditions than in the
private sector, an indicator for SOE is included as a proxy for labour market institutions.
4.4 The model
4.4.1 Wage model design
In this part, an econometric model is built to investigate how technological upgrading affects
wages. The rationale behind this model is that technological upgrading may enhance workers’
skill level, which increases workers’ wages. For example, when a factory introduces new
production technologies, it needs to offer necessary training to workers before they can use the
technology. When workers gain more skills, the employer will have to pay higher wages to retain
them. Therefore, high-tech firms pay more to keep workers compared to other firms where
workers are doing routine activity (Blanchard and Johnson 2013).
In terms of the dependent variable, some research uses wage data in natural logarithm form
(Braunstein and Brenner 2007; Braunstein and Epstein 2002; and Luca Nunziata 2005); other
research adopts the wage growth rate (Afonso and Gomes 2014; and Ana Lamo, Javier J. Pérez
and Ludger Schuknecht 2012). Since wages in the manufacturing sector have increased steadily
during the selected years (see Figure 4. in Chapter 3), the growth rate form of the wage data
allows some fluctuation in wages and thus, is preferred to the natural logarithm form of wage. In
64

fact, the first difference of wage growth rate will be used as the dependent variable since it
allows this model to examine which technological upgrading variable could accelerate wage
growth.
As discussed in the theoretical model, the variables that will be included are: price
expectations, labour market conditions, labour market institutions, and technology upgrading.
The econometric model to estimate how technological upgrading affects the wage is:
Wspit = α + β1 priceexpit + β2 labourit , + β3 institution2it + σ Techupit + uit

(5)

where i represents province and t represents time:
Wsp = wage growth speed;
a = the constant
priceexp = price expectations
labour = labour market conditions
institution = labour market institution power
Techup = a group of technological upgrading variables.
Based upon the theoretical discussion of wage determination in 4.2, price expectation is one
of the factors which is expected to have a positive effect on wages, which means the higher the
price expectations, the higher the wages will be.
Labour market condition is expected to have a positive relationship with wages as well, and
more specifically, when there are more people employed, wage increases. The relationship
between wage and institutional power is uncertain. Lastly, technological upgrading indicators are
expected to have positive effects on wage.
The limitation of this model is that it cannot capture how microeconomic level
characteristics affect wages. In addition, some other factors that affect wages may not be

65

captured in this model, such as bargaining power. The benefits of this model are that we can see
how technological upgrading affects wages in the manufacturing sector and what form of
technological upgrading is leading the wage change. Moreover, the macroeconomic level model
will not be biased by the sample selection.
4.4.2 Econometric issues and solutions
This research will use Eviews 8 to run all the models. Ordinary least squares (OLS) method is
adopted to estimate all the unknown parameters in this linear regression model. The data are
constructed as a time-series-cross-section data set rather than simple time-series data set.
Econometrics issues, such as non-stationarity, multicollinearity, heteroscedasticity, and
autocorrelation may occur, as discussed below.
Non-stationarity
In non-stationary time series, shocks will be permanent, and the mean and the variance will
depend on time (Dimitrios Asteriou and Stephen G. Hall 2011). The unit root test is used to
determine whether error terms are white noise or not. If errors are not white noise, which means
that error terms do not have a mean of zero and they are correlated between values at different
times, the time series is non-stationary. The problem with non-stationary data is that “the
standard OLS regression procedures can easily lead to incorrect conclusions” (Asteriou and Hall
2011, 338). In other words, the regression results from a non-stationary time series will be
spurious.
The augmented Dickey-Fuller (ADF) test for unit roots will be adopted in this research for
each variable.24 Eviews offers the ADF test in its program; the test result indicates whether the
variable has a unit root or not, and the number of the unit root. If it is known that a series has a
The mathematical framework of ADF test is beyond the scope of this research; please refer to Asteriou and Hall
(2011, 342-344) for an mathematical explanation.
24

66

unit root, the series can be differenced to render it stationary. For example, the process is to test a
series Yt by using Eviews ADF test. Eviews will produce a ADF test report, and if the ADF test
statistics of Yt is bigger than the critical t-statistics (for this research I choose 10% level) listed in
the same test result report, the Yt is recognized as a stationary series (noted as I (0)), Otherwise,
it is non-stationary, and the first difference needs to be taken: Δ Yt =Yt-Yt-1. It is necessary to
take the ADF test again, and if Δ Yt is stationary then Yt is stationary at first order differencing
(noted as I (1)). If a series is not stationary after taking the first order differencing, the second
differences is taken and tested again; the procedure can be repeated, until the test results indicate
the series is stationary.
Multicollinearity
Multicollinearity refers to a phenomenon in which the explanatory variables in a linear
regression model are moderately or highly correlated (PennState Eberly College of Science n.d.).
In this situation, the estimated model parameters remain unbiased, consistent and effective, but
the estimated parameters (including the parameter signs) are very sensitive to small changes to
the model or data. Consequences of multicollinearity are summarized as:
1. Estimates of the OLS coefficients may be imprecise in the sense that large
standard errors lead to wide confidence intervals; 2. Affected coefficients may
fail to attain statistical significance because of low t-statistics; 3. The signs of
estimated coefficients can be the opposite of those expected; 4. The addition of
deletion of a few observations may result in substantial changes in the
estimated coefficients (Asteriou and Hall 2011, 101).
To detect multicollinearity, a correlation table can be built for all the variables to disclose
the relationships between them. Alternatively, multicollinearity can be identified from the
regression results. For instance, there is multicollinearity when the regression result illustrates a
high R2 but low t-statistics for most variables. In order to resolve or avoid multicollinearity,
variables should be added into the model one by one. Thus, when there are correlated variables,
67

variables can be identified and replaced immediately. For this research, the model will be run
several times with only one technology variable each time to avoid the problem of
multicollinearity.
Heteroscedasticity
The classic linear regression model requires that error terms are homoscedastic, which means the
errors have the same scatter regardless of the value of independent variable. When the scatter of
the errors is different depending on the value of one or more of the independent variables, the
error terms are heteroscedastic (Asteriou and Hall 2011).
Heteroscedasticity may arise from the following issues. Firstly, heteroscedasticity arises if
the model excludes some of the explanatory variables (Jingshui Sun and Shuqin Ma 1991). If the
model contains only a few major factors, the impact of other factors is included in the random
error term, and therefore, the random error term could produce heteroscedasticity. Secondly, it
may arise if the form of the regression equation is mis-specified (Sun and Ma 1991). In general,
the relationship between the explanatory variables and the dependent variable can be complex.
Constructing the linear regression model to estimate the non-linear relationship between
explanatory variables and the dependent variables increases the errors and leads to
heteroscedasticity. Thirdly, heteroscedasticity occurs if the variables contain measurement errors
(Sun and Ma 1991). On one hand, measurement error in the sample data often accumulate over
time, which increases the variance of the random error term over time. On the other hand,
improvements in sampling techniques and other data collection methods reduce random errors.
Thus, in the time series data, items do not have a random homoscedasticity. Lastly, random
factors also contribute to heteroscedasticity (Sun and Ma 1991). Economic variables themselves

68

are affected by a number of random factors (such as policy changes, natural disasters or financial
crises, etc.), therefore, variables do not have certainty and repeatability.
The consequences of heteroscedasticity are recognized as below:
1. Heteroscedasticity does not result in biased parameter estimates; 2. among
all the unbiased estimators, OLS does not provide the estimate with the
smallest variance. Depending on the nature of the heteroscedasticity,
significance tests can be too high or too low; 3. the standard errors are biased
when heteroscedasticity is present. This in turn leads to bias in test statistics
and confidence intervals; 4. significance tests are virtually unaffected, and thus
OLS estimation can be used without concern of serious distortion (Richard
Williams 2015, 2-3).
To identify heteroscedasticity, the White test will be used. The White test, proposed by
Halbert White (1980), is a statistical test through the establishment of an auxiliary regression
method to determine heteroscedasticity. Since Eviews includes a routine for executing the
White’s test, I am going use the White test (cross-product) function to detect the residual. If the
White’s test has a Lagrange Multiplier statistic (LM-statistic) bigger than the critical value or if
the p-value is less than the level of significance α (usually α=0.05), then there is significant
evidence of heteroscedasticity.25
If the evidence indicates heteroscedasticity, then autocorrelation consistent (HAC) standard
error estimators will be used to resolve it. 26 Eviews can compute White’s heteroscedasticitycorrected variances and standard errors automatically, therefore, the manual calculation is
unnecessary. By using this method, standard errors of all variables will be corrected and
estimators will be more accurate.
A critical value is used in statistical hypothesis testing, the value of it corresponding to a given significance level.
“The critical value approach involves determining "likely" or "unlikely" by determining whether or not the observed
test statistic is more extreme than would be expected if the null hypothesis were true. That is, it entails comparing
the observed test statistic to some cutoff value, called the "critical value." If the test statistic is more extreme than the
critical value, then the null hypothesis is rejected in favor of the alternative hypothesis. If the test statistic is not as
extreme as the critical value, then the null hypothesis is not rejected” (PennState Eberly College of Science n.d.)
26
The mathematical details of this method are beyond the scope of this research, please refer to White (1980) for
more details.
25

69

Autocorrelation
An assumption of the linear regression is that error terms are independently distributed. If this
assumption does not hold, the disturbances are autocorrelated. Autocorrelation is most likely to
occur in a time series framework because the error in a certain period may be affected by errors
in other periods (Asteriou and Hall 2011, 149). Autocorrelation can occur due to reasons such as
omitted variables, misspecification, and systematic errors in measurement (Asteriou and Hall
2011, 149).
Autocorrelation of the errors violates the OLS assumption that the error terms are
uncorrelated. The consequences of autocorrelation are summarized as:
1. The OLS estimators are still unbiased and consistent; 2. The estimators will
be inefficient and no longer the Best Linear Unbiased Estimators (BLUE); 3.
The estimated variances of the regression coefficients will be biased and
inconsistent, and therefore hypothesis testing is no longer valid (Asteriou and
Hall 2011, 152).
To detect autocorrelation, the most frequently used method is the Durbin-Watson (D.W.)
test. It is named after James Durbin and Geoffrey Watson (1950) (see Asteriou and Hall (2011,
156-158) for mathematical details about D.W. test). Although Eviews reports the D.W. test result
(d) of every regression output, the D.W. critical values still need to be calculated. Since there are
4 variables (excluding the constant) in this model, k’= 4. Based upon the D.W. significance table
(Asteriou and Hall 2011, 485) and 84 observations in my data set, the dL and the dU are 1.41 and
1.60 respectively. See Table 13 below for the D.W. critical.
Table 13. The D.W. critical result
0≤d≤1.41
Positive serial
correlation

1.41<d<1.60
Test
inconclusive

1.60≤d≤2.40
No
autocorrelation

2.40<d<2.59
Test
inconclusive

2.59≤d≤4.00
Negative serial
correlation

Source: table compiled by author based upon the D.W. test table in Asterious and Hall (2011,
157)
70

If the test indicates an autocorrelation of the model, the solution is to add autoregressive
errors of order 1 (AR (1)) to estimate the model again. If the new D.W. statistics indicates that
autocorrelation still exists, the autoregressive errors of order 2 or/and 3…n will be needed to
estimate the model until D.W. test indicates there is no autocorrelation.
4.5 Data for the model
The accessibility of appropriate data is an important part of any empirical research. The
quantitative analysis uses data from China at the provincial level. All the data come from
government official databases such as the:
1. China statistical yearbook (2002-2015)
2. Jiangsu statistical yearbook (2002-2015)
Guangdong statistical yearbook (2002-2015)
Zhejiang statistical yearbook (2002-2015)
Fujian statistical yearbook (2002-2015)
Guangxi statistical yearbook (2002-2015)
Hainan statistical yearbook (2002-2015)
Shandong statistical yearbook (2002-2015)
4. National Bureau of Statistics of P.R.C. of the P.R.C. website:
http://www.stats.gov.cn/english/
5. State Intellectual Property Office of the P.R.C. website: http://english.sipo.gov.cn/
6. General Administration of Customs of the P.R.C website: http://english.customs.gov.cn/
My database on most indicators covers the time period from 2002 to 2014. However, I extend
the wage data back to 2001, as I need to calculate wage growth rate and take first difference of it.
The whole modeling data set cannot be extended to 2001 because two of my key high-tech
71

indicators (high-tech exports and high-tech imports) are only available in the statistical
yearbooks since the year 2002. Since calculating the change in the wage growth rate (wage
growth speed) for a given year requires data for the current year and preceding two years, the
data used in the regression analysis covers the time period 2003 to 2014. The data are collected
for seven coastline provinces, the total number of observations is 91 (7 provinces and 13 years,
from 2002 to 2013) and after the data processing the number of observations that I use in my
model is 84 (7 provinces and 12 years, from 2003 to 2014). Only seven coastline provinces are
included in this data set due to the data source limitation. The original plan for this research is to
obtain data for all coastline provinces and cities, however, this plan cannot be fulfilled because
not every coastline province and city publish statistical yearbook and some provinces do not
report consistent data for some vital technological upgrading indicators.
The indicator I use for price expectations is the lagged inflation rate as the inflation rate
indicates the goods and service price change each year, and therefore I assume that people expect
the price change in the next year would be as same as the current year. Labour market conditions
is captured by GDP growth rate, as Okun’s law indicates that when GDP increases, the labour
market tightens. Lastly, ten technological upgrading indicators are used, as listed in Table 14; the
indicator selection is based upon literature review of technological upgrading and data
availability.
4.6 Model results
Based upon econometric issues addressed in section 4.3.2, necessary procedures are taken to
identify and resolve the issues to yield efficient and accurate results. Firstly, an ADF unit root
test is used to determine the stationarity of each variable; the results indicate that the following
variables have unit roots: FDI per worker, R&D expenditure as a share of GDP, value of high72

tech export, value of high-tech import, and the value of high-tech import divided by total number
of manufacturing workers. After taking the first difference, all these five variables become
stationary. Therefore, they are stationary at first order differencing.
Secondly, the technological upgrading variables are added one by one into the model, and
therefore, multicollinearity among the variables is avoided in the ten models. Thirdly,
heteroscedasticity is identified in all ten models, and therefore, the HAC standard error
estimators is used to resolve it. Lastly, since autocorrelation has been found in all ten models,
AR (1) added in to all models to resolve the autocorrelation. The first difference of wage growth
rate offers variation in the dependent variable, and therefore wage growth speed is adopted as the
dependent variable.
Fourthly, the descriptive statistics in Table 14 offers some information about my data set.
The Obs. indicates the total number of my observations, and for this research the number of
observations is 84. The Mean indicates the average of the data, which is the sum of all the
observations divided by the number of observations. For example, the wage growth speed is -0.2,
which means that on average, wage is increasing at a decreasing speed of -0.2% annually. The
Max is the largest data value in the data set and the Min is the smallest data value in the data set.
These two descriptive statistics are used to identify a possible outlier or a data-error in the data
set. For instance, the wage growth speed has a Max of 15.8 and Min of -18.4; and the big
difference between these two data observations led me to verify the data entry and calculation.
The Standard deviation (Std. Dev.) is used to measure how the data vary around the mean. The
higher the standard deviation, the greater the spread of the data. For instance, the standard
deviation of annual GDP growth is 5, which means that on average, the GDP growth rate for a
province deviates from the mean (which is 15.5%) by about 5%.

73

Table 14. Model variable details
Variables
Definition
Obs. Mean
Wage growth Wage growth speed is the first 84
-0.2
speed
difference of the wage growth
rate; it is also known as the
change in wage inflation27; The
original wage data is the annual
nominal average annual wage
of staff and workers in the
manufacturing industry in urban
units
Price
Lagged inflation rate
84
2.8
expectations

Max
15.8

Min
-18.4

Std. Dev.
5.8

7.8

-2.3

2.0

Labour
market
conditions
Labour
market
institutions

Nominal GDP annual growth 84
rate (%)

15.5

24.8

6.3

5.0

Value of total SOE output 84
divided by the value of total
industry output (%)

12.0

65.0

0.7

11.6

Technological 1. Value of mechanical and 84
upgrading
electronics exports as a share of
variables
the value of total exports (%)

42.9

69.9

10.0

18.4

2. Value of mechanical and 84
electronics imports as a share of
the value of total imports (%)

40.0

73.8

1.4

19.8

3. First difference of actual FDI 84
utilized divided by total number
of employed workers for the
entire province (USD/worker)

14.2

122

-213.5

49.4

4. R&D employees in LM 84
enterprises divided by the total
number of staff and workers in
the manufacturing sector 28(%)

5.7

16.0

1.1

2.8

Staff and Workers are people who signed a labour contract with the employer and are paid by the employer. The
employer has to pay salary and offer social security package and housing fund. Staff and Worker includes people
who are on injury leave, maternity leave, learning leave, sick leave, but still getting paid by their employer (National
Bureau of Statistics of P.R.C. 2015a)
28
The classification of large- and medium-sized (LM) enterprise is based on a combined firm-size indicator, where
operating income and numbers of employees, total assets are taking into account. For the manufacturing industry,
large-enterprise requires more than 1,000 employees and over 0.4 billion CNY operating income; while mediumenterprise requires more than 300 employees and over 20 million CNY operating income (National Bureau of
Statistics of P.R.C. 2013).
27

74

5. First difference of R&D 84
internal expenditure in LM
enterprises as a share of GDP
(%) 29

0.0

0.4

-0.8

0.2

6. Number of patents granted 84
divided by total number of
employed worker (in 000’s)
over the whole province (case)

1.1

5.7

0.0

1.4

7. First difference of the value 84
of High-tech exports (billion
USD)
8. First difference of the value 84
of High-tech imports (billion
USD)
9. Value of high-tech exports 84
divided by the total number of
staff and workers in the
manufacturing
sector
(USD/worker)

4.4

36.0

-25.4

9.8

3.3

33.9

-25.4

8.3

10,179.1 41,251.1 56.8

12,320.9

10. First difference of the value 84
457.4
10,124.0 -13,032.3 3,191.4
of high-tech imports divided by
the total number of staff and
workers in the manufacturing
sector(USD/worker)
Note: the descriptive statistics for the raw data are presented in the Appendix. The first difference of
variable is used to address the non-stationary issue as discussed in the section above. 30

R&D internal expenditure refers to the actual expenditure on basic research, applied research, experimental
development, direct expenditure on research project, management and service, that spent by the R&D unit. It does
not include expenditure on production activities, repay loans and cooperation or commission expenditure transferred
to external R&D units (National Bureau of Statistics of P.R.C. 2015a).
30
An Augmented Dickey-Fuller (ADF) unit root test (including both intercept, and trend and intercept in the
equation) is used to determine non- stationarity for each variable.
29

75

Table 15. a Wage equation for 7 provinces, using data for 2003-2014
(Estimation with HAC standard errors and covariance; dependent variable: wage growth speed)
variable
model I
model II
model III
model IV
model V
c
-5.43***
-5.92***
-4.95***
-5.84***
-5.29***
(1.13)
(1.41)
(1.02)
(1.39)
(1.07)
Lag inflation rate

0.72***
(0.23)

0.73***
(0.23)

0.75***
(0.23)

0.74***
(0.23)

0.68***
(0.25)

GDP growth rate

0.18**
(0.08)

0.19**
(0.08)

0.14*
(0.08)

0.19**
(0.08)

0.19**
(0.08)

SOE share^2

0.00
(0.00)
0.00
(0.02)

0.00
(0.00)

0.00
(0.00)

0.00
(0.00)

0.00
(0.00)

Mechanical

&

electronics

import

share
Mechanical

&

electronics

export

0.01
(0.02)

share
∆ FDI per worker
R&D

0.02**
(0.01)

employment

share
∆R&D

0.08
(0.12)

expenditure/

3.26**
(1.56)

GDP
R square
N

0.32
84

0.32
84

0.36
84

0.32
84

Note: ∆ means first difference, and it is used to address non-stationary in the variables;
Standard errors in parentheses.
* Denotes significant at the 10 percent level.
** Denotes significant at the 5 percent level.
*** Denotes significant at the 1 percent level.

76

0.33
84

Table 15. b Wage equation for 7 provinces, using data for 2003-2014
(Estimation with HAC standard errors and covariance; dependent variable: wage growth speed)
variable
model VI
model VII model VIII
model IX
model X
c
-7.27***
-5.22***
-5.69***
-5.23***
-4.88***
(1.58)
(1.13)
(1.08)
(1.12)
(1.25)
Lag inflation rate
GDP growth rate
SOE share^2
Patent

per

(1000

0.75***
(0.23)
0.27***
(0.10)

0.72***
(0.23)
0.18**
(0.08)

0.73***
(0.23)
0.19**
(0.08)

0.72***
(0.23)
0.18**
(0.08)

0.73***
(0.23)
0.16*
(0.09)

0.00
(0.00)

0.00
(0.00)

0.00
(0.00)

0.00
(0.00)

0.00
(0.00)

0.51**
(0.24)

worker)
∆ High-tech export
High-tech export per
manufacture worker
∆ High-tech import

0.01
(0.04)

0.00
(0.00)

0.02
(0.05)

∆High-tech import per
0.00
manufacturing worker
(0.00)
R square
0.34
0.32
0.32
0.32
0.32
N
84
84
84
84
84
Note: ∆ means first difference, and it is used to address non-stationary in the variables;
Standard errors in parentheses.
* Denotes significant at the 10 percent level.
** Denotes significant at the 5 percent level.
*** Denotes significant at the 1 percent level.

77

Turning to my regression results, Table 15.a and Table 15.b present estimates for each
model. “In traditional statistics the significance level is usually set at 1%, 5%, and/or 10%”
(Asteriou and Hall 2011, 9), and therefore, this research follows the traditional way of
confidence level setting. Corresponding to the significance levels, the t-critical values are 2.57,
1.96, and 1.64. 31 The significance level is used as a way to identify whether an independent
variable has a statistically significant impact on the dependent variable or not. More specifically,
if the t-statistic of an independent variable is greater than the 1% significance level of 2.57, then
with 99% of confidence, I can conclude that this independent variable has a statistically
significant effect on the dependent variable. The following discussion of regression results is
based upon the statistical significance test results in Table 15.a and Table 15.b
For the macroeconomic variables, the results are as expected. Starting with the price
expectations variable, which is proxied by the lagged inflation rate, and looking horizontally
through Table 15.a and Table 15.b, I find that it has positive and significant effects on the
manufacturing wage growth speed in all ten models, which means the higher the lagged inflation
rate, the faster the wage growth speed.
The Labour market conditions variable, which is proxied by the nominal GDP growth rate,
plays a positive significant role in all wage models. The results indicate a higher GDP growth
rate will speed up the wage. This is consistent with the theoretical hypothesis of Okun’s law that
when GDP increases, the labour market tightens, which further raises wages.
Labour market institutions, which is proxied by squared SOE production share, does not
play a role in manufacturing wage growth speed in all models. The SOEs’ production share is
also regarded as a market liberalization variable since SOEs are owned by the state. From this
perspective, the modeling results suggest that whether the economy dominated by the state or not
31

This research uses the t-table with right tail probabilities in Asteriou and Hall (2011, 479).

78

does not affect the wage growth in the manufacturing sector. In other words, for workers in the
manufacturing sector, the ownership type of the enterprise does not make a significant difference
in terms of their wage growth.
In terms of technological upgrading indicators, three of them pass the critical level of tstatistics: the first difference of FDI per worker; the first difference of R&D expenditure as a
share of GDP; and Patents per 1,000 workers. The FDI indicator suggests that when there is an
increase in FDI per worker, the average annual wage in manufacturing sector grows faster. Given
that the FDI variable is measured as the first difference of FDI per worker, the magnitude of the
regression coefficient can be interpreted as follows. If FDI per worker increases by 1 dollar from
one year to the next, the wage growth rate will increase by 0.02 percent in the manufacturing
sector. Therefore, introducing more FDI speeds up the wage growth and it can be a good strategy
for development. This result is consistent with Braunstein and Epstein (2002) and Braunstein and
Brenner (2007) results, in which FDI has a statistically significant and positive impact on wages
in China.
The R&D expenditure variable suggests that an increase in R&D expenditure in large and
medium enterprises as a share of GDP accelerates the wage growth significantly in the
manufacturing sector. This result is reasonable since the Chinese government at different levels
(national, provincial and municipal) is promoting domestic technological upgrading in recent
years. Economic and technology development zones are being built to facilitate high-tech, highvalue-added firms to operate and expand production. To attract high-tech enterprises, Chinese
governments at different levels are providing tax benefits and subsidies on equipment. To
promote domestic technological upgrading, local governments encourage the resource based and
labour intensive industries to relocate from developed areas to underdeveloped areas within the

79

province, so that they can provide enough land for high-tech firms. For instance, in 2002, the
Jiangsu Municipal Commission of Development & Reform started organizing the industrial
transfer from south Jiangsu (a more developed area) to north Jiangsu (an underdeveloped area).
As of 2012, there are 17,439 projects that have been moved to north Jiangsu (including Xuzhou,
Yancheng, Huaian, Lianuangang, Suqian). Each project is worth more than five million CNY
(0.8 million USD), and total investment reached about 1,015.1 billion CNY (160.9 billion USD)
(Nanjing Municipal Commission of Development 2012).32 The strategy has two main
implications for workers: 1. the shift in industries to underdeveloped areas will increase the
numbers of jobs in underdeveloped areas; 2. for workers who stay in the developed areas, they
are likely to get more training and higher wages from those high-tech firms. The first implication
will reduce the number of migrant worker in developed areas. It will also push industries’
technological upgrading towards high-value-added production or towards labour-saving
production in developed areas. The second implication can be explained as that high-tech
enterprises located in developed areas (coastline areas) need to offer training to workers, which
can push the wage increases further. High-tech enterprises do not normally concentrate on
labour-intensive products and they normally engage in skill-intensive products that requires a
labour force with higher education levels. The education requirement of labour will increase over
time, and workers who stay in developed areas will earn higher wages. For instance, in the past,
most recruitment criteria used to be “no academic requirements” in the Guangzhou labour
market, but in recent years there is a significant increase of “college and higher education” job
requirement (Ruiqiu Yang 2015).
Turning to the last statistically significant indicator, Patents per 1,000 workers, the results
suggest that when patents increase, wages grow faster. This indicator can also be seen as a
32

The exchange rate is based on 100 USD = 631.0 CNY (National Bureau of Statistics of P.R.C. n.d.).

80

domestic R&D indicator and consistent with the previously reported findings that R&D
expenditure as a share of GDP has a positive and statistically significant impact on wage growth
speed. Therefore, encouraging domestic technological upgrading is another good strategy to
increase workers’ wages.
Surprisingly, the other seven technological upgrading proxies do not have a statistically
significant effect on the manufacturing wages. Neither high-tech exports nor high-tech imports
indicators are statistically significant which is surprising because these are two very important
technological upgrading indicators reflecting the capability of producing high-tech products.
Other studies of Chinese high-tech exports may help explain why these two indicators do not
affect wages. For instance, Schott (2008) discovers that although China’s export products have a
very high overall sophistication level, the vertical sophistication turns out to be exceptionally
low.33 This result is supported by Koopman, Wang and Wei (2008) who find that electronic
devices, which are likely to be labelled as sophisticated products, have particularly low domestic
content (only about 30% or less). Yuqing Xing (2014, 122) finds that the trade statistics reported
by Chinese government “mistakenly credit entire values of assembled high-tech products to
China”. Xing (2014, 122) claims that “China’s real contribution to 82% of reported high-tech
exports is labor and not technology”. These findings imply that even though huge amounts of
high-tech products are exported by China, not many are produced in China from the beginning to
the end. A lot of so called “high-tech products” are imported as semi-finished products and
assembled by the manufacturing sector and exported as finished high-tech products. If the
domestic value-added in the high-tech exports is at a low level (less than 30% as Koopman,
Wang and Wei (2008) suggest), then this could be an explanation of why high-tech imports or
The “overall sophistication” of Chinese exports indicates the overlaps between Chinese exports categories and
developed economy exports categories; the “vertical sophistication” of Chinese exports indicates the level of valueadded in the products (Schott 2008).
33

81

exports indicators do not affect the manufacturing sector wage. The value-added in Chinese
high-tech products is mainly labour rather than technology. Besides, mechanical and electronics
imports and exports are also insignificant, which could further verify that the processing
activities do not contribute to the wage growth. 34 Therefore, the result from my modeling
suggests that low domestic value-added activities cannot be a solution to increasing wages in
manufacturing sector.
4.7 Summary
In sum, the empirical evidence indicates that three of technological upgrading indicators have
positive and statistically significant effects on wage growth in the manufacturing sector. More
specifically, facilitating the increase of FDI and/or R&D activities in large and medium
enterprises as well as domestic innovations will accelerate wage growth in the manufacturing
sector. Although the correlations between the wage and various indicators of technological
upgrading (including high-tech exports, high-tech imports, mechanical and electronic exports,
mechanical and electronic imports) are positive, those indicators do not affect the wage growth
during the period 2003 to 2014. This further suggests that processing semi-products does not
help the wage growth, and domestic value-added needs to be increased in those exported hightech categories.
The drawback of the model is that the R square of each model is a low level around 0.32 to
0.34. This means that there is only around 32% to 34% of the variation in the dependent variable
being explained by the regressors. There could be some other variables that affect wage growth
that are not included in my database; therefore, further research needs to examine some new

I find that there are many overlapping parts between high-tech products and mechanicals & electronics products
by checking the products’ codes in both High-tech Products Catalog 2006 and China Custom Statistical Yearbook
2006 (Guangdong High-tech Enterprise Association 2015 )
34

82

variables or to extend the time horizon for the data set to enhance the explanatory power of the
model. For example, the amount of FDI spent on R&D, the amount of domestic value-added in
high-tech exports, and the amount of domestic value-added in mechanicals and electronics
exports would more accurately capture technological upgrading compared to indicators used in
this study and having such data would help in estimating the relationship between technological
upgrading and wages.

83

Chapter 5: Conclusion
5.1 Research goals and achievements
The goal of this research is to identify the impacts of technological upgrading on decent work in
the manufacturing sector in China by answering questions:
1. Has there been technological upgrading in China?
2. Has there been an improvement in decent work in China?
3. Does everyone benefit from improvement in decent work equally?
4. Does technological upgrading improve decent work in the manufacturing sector?
The descriptive analysis of trends in technological upgrading and decent work is based
upon provincial level data, for 2002 to 2014, for seven coastline provinces: Jiangsu, Guangdong,
Fujian, Zhejiang, Shandong, Guangxi, and Hainan. By using descriptive analysis, firstly, I find
that during the selected period, technological upgrading has taken place in all provinces; Jiangsu
and Guangdong have a higher technology endowment than the other five provinces, in absolute
and relative terms.
Secondly, there has been an improvement in decent work in the manufacturing sector,
based upon indicators such as the average annual nominal wage, the number of dependents that a
worker can support with this annual wage, and social security provisions.
Thirdly, while an improvement in decent work has been observed, the distribution of these
benefits is uneven between workers with an agricultural hukou (rural workers) and those with a
non-agricultural hukou (urban workers). Rural workers who migrate to cities have fewer chances
to obtain decent work or decent working time compared to workers with a non-agricultural
hukou. This is demonstrated by data reporting that over 60% of migrant workers have no work
contract, and because of this, we could not expect them to have any social security coverage.
84

Access to decent work differs among workers not only in terms of their residential status
but also by industrial sector of employment. Some sectors provide workers with regular working
hours and good pension plans such as Public Administration sector. Some sectors in urban area
(such as Hotels and Catering Services sector or Wholesale and Retail Trade sector) require
regular over-time work, which is neither healthy for workers nor good for employment over the
whole economy. It is disappointing that even though the standard working hour in China is
regulated, overtime working is still common in 90% of all the sectors. Absence of government
supervision blocks the effectiveness of the labour law.
Fourthly, there is some evidence that technological upgrading improves decent work.
Provinces with higher technology endowments tend to show better performance in those decent
work indicators, such as the monetary wage in the manufacturing sector and employment
insurance participation rates. This is demonstrated by the positive correlation between
technological upgrading indicators and decent work indicators. For example, Jiangsu and
Guangdong, provinces with higher technological endowments, also have higher average annual
nominal wages and employment insurance participation rates.
While the fourth question to be addressed in this thesis is whether technological upgrading
improves decent work, given that few indicators of decent work exist, the econometric analysis
uses the average annual nominal wage as the main indicator of decent work. The regression
analysis is based upon data for the time period 2003 to 2014, with the wage data needed to
construct the wage growth speed variable being extended back to 2001. Various proxies for
technological upgrading are used in the regression analysis. Only three of technological
upgrading proxies pass the t-test, and these are: the first difference of FDI per worker; the first
difference of R&D expenditure as a share of GDP; and Patents per 1,000 workers. Thus, the

85

regression model results provide some support for the hypothesis that technological upgrading
will improve wages. The impacts on wage growth caused by inflation rate and nominal GDP
growth rates are significant in all modeling results.
The main contribution of this research is that it focuses on the technological upgrading in
manufacturing sector, and the impact of it on decent work. The conclusion of this research is that
technological upgrading enhances decent work, as measured by the wage, in China. Moreover, I
have not found literature with similar topic. However, not all technological upgrading indicators
have had a statistically significant impact on decent work, such as those so called “high-tech”
exports.35 This research finds that although there has been technological upgrading during the
transformation from labour-intensive manufacturing to the low domestic value-added but
technology-intensive manufacturing, this change has not had a statistically significant impact on
workers’ wages. This research also confirms that the impact of FDI on wages is significant and
positive in China, which aligns with the results from Braunstein and Epstein (2002). Khalid
Nadvi (2004) recognized that even if technological upgrading has a positive overall impact on
the labour market in terms of increased employment and wages, it may not benefit all workers
equally. This research verified this point and find out that workers could be blocked by their
Hukou type, and working sectors in China.
This research adds to the understanding of the relationship between technological upgrading
and decent work and finds that they are positively related. Moreover, this research also fills some
gaps in the literature on decent work study in terms of the Chinese labour market. For example,
the decent work working time distribution between citizens with different Hukou types.

As Xing (2014, 122) claims, “China’s real contribution to 82% of reported high-tech exports is labor and not
technology”.
35

86

5.2 Policy implications
Policy implications of this research are organized as follows. Firstly, based upon my
estimation results, future policy-making should focus on introducing more FDI, increasing the
R&D expenditure in large and medium enterprises and encouraging domestic innovation to
accelerate the growth of wage. In other countries, there has been concern that increased FDI is
associated with increased income inequality because it raises the demand for, and wages of,
skilled workers, relative to unskilled workers, as shown for example, in the case of Mexico
(Robert C. Feenstra and Gordon H. Hanson 1997). However, FDI in different countries may have
different impacts on the labour market depending upon the nature of the production and demand
for skills. For China, Xiaofei Tian, Baotai Wang and Ajit Dayanandan (2008) find that instead of
worsening income inequality, the FDI inflows during 1978 to 2006, consistently improved the
income distribution in China. They suggest that this finding supports the hypothesis that trade
liberalization will increase the demand for the abundant labour force.
However, as firms further engage in technological upgrading this may increase the demand
for higher level skilled workers, relative to lower skilled workers, and thus, may contribute to
increased income inequality in the future. This is an area for further research and also points to
the importance of continuing to increase education and skills of the labour force.
Secondly, my estimation results suggest that low domestic value-added activities (for
example, processing or assembling activities) do not affect the wage growth. Therefore, future
policy should facilitate the expansion of industries which produce high domestic value-added
products (high-tech producing industries). The distinction between high-tech processing and
high-tech producing industries should be considered by policy makers. More specifically, hightech producing industries (high domestic value-added products) emphasize that domestic
producers should utilize the technology (in both embodied and disembodied forms) to produce
87

the goods from the beginning to the end; while high-tech processing industries are, those
industries are mainly adding labour into semi-finished high-tech products without owning any
core technology. The overestimation of Chinese high-tech exports is an issue which needs to be
investigated, as it appears that the measurement may not sufficiently distinguish between hightech processing exports and high-tech producing exports. The inappropriate measurement may
limit policy makers’ abilities to assess and address the real changes necessary to develop a
technological upgrading strategy and move the economy to a higher-income level.
Thirdly, the econometrics regression results also indicate that the inflation rate and GDP
growth rate have positive and significant effects on wage growth. Therefore, a set of
expansionary monetary policies, such as permitting higher inflation rates or lowering the deposit
reserve ratio, may benefit nominal wage growth. However, simply raising the inflation rate does
not really bring up the living standard for citizens, therefore, raising inflation rate should not be a
long-term policy recommendation. Increasing GDP growth is another method to bring up the
income, but what is the right way? As a high middle-income country, the export-driven and
investment oriented economic growth strategies are facing difficulties to provide sustainable
economic growth for China (Kai Guo and Papa N’Diaye 2009, Gangming Yuan 2013). As I
discussed before, China is at an economic stage that it is losing its comparative advantage (low
labour-cost) to lower-wage economies, and at the same time, given its technology level it still
cannot compete with high-income economies. By depressing the wage to keep its comparative
advantages is an unrealistic option. Therefore, future policy-making should find ways to more
introduce cutting-edge technology and to facilitate domestic R&D activities.
Fourthly, the Hukou system in China is viewed as the major source of hardship for migrant
workers living in cities (Kam Wing Chan and Li Zhang 1999, Peter Alexander and Anita Chan

88

2004, Au Loong-Yu and Nan Shan 2007). This research also find that Hukou system makes it
unequal for workers to benefit from technological upgrading. Therefore, changes in legislation
regarding Hukou system are required. Legislation should not be used to make residents unequal
in terms of finding a job or accessing public goods and services (for example, medical care,
education and so on).
Lastly, as this research recognizes, overtime work is prevalent in most sectors in China.
Daniel Fu Keung Wong, Chang Ying Li and He Xue Song (2007) discovered that overtime work
and inadequate rest is one of the most stressful work-related issues for Chinese rural-urban
migrants. To protect labour from over-time work and to increase more employment
opportunities, government monitoring of labour standards and labour protection is more
necessary than ever. On the one hand, long daily working time reduces workers’ happiness,
increase their stress and the chance of getting occupational injuries (A. E. Dembe, J. B. Erickson,
R. G. Delbos and S. M. Banks 2005). On the other hand, the marginal productivity of each
worker reduces after certain hours of work, adding more working hours does not bring sufficient
output, which is not a wise choice for the employer as well. Cutting down the average working
hours to the standard working hours will provide more job vacancies. For example, ceteris
paribus, cutting down the working hours from 50 hours per week to 40 hours per week for 4
workers on the same position, will create new work positions with 40 hours per week. Besides,
when there is more leisure time, workers consume more products and services, which benefits
economy growth as well. Technological upgrading will be required to increase wages and decent
work, with the reduction in working hours, to maintain and increase individual worker incomes.

89

5.3 Limitations and recommendations for future research
Several limitations were encountered in undertaking this research. Firstly, the number of
observations is small primarily because consistent data could only be obtained for seven
provinces. Not all coastline provinces or cities publish statistical yearbooks such as Shanghai and
Shenzhen, which makes the data collection impossible for them. The consistency of indicators in
statistical yearbooks is another problem; for example, some good technological upgrading
indicators exist for years prior to 2003 and then are no longer collected or publicly available
(such as the number of R&D achievements in large and medium enterprise). Moreover, inland
provinces do not publish data of their provincial exports, and therefore, indicators such as hightech exports or mechanical exports cannot be found for inland provinces. These problems
resulted in a small number of observations being used in the econometric analysis. The
regression results may also be weak because the models are estimated on data for coastline
provinces which tend to be the most developed provinces and thus, there is relatively little
variation in the data. If the data set included more provinces, then greater variation among the
provinces might contribute to stronger impacts of technological upgrading on wages, my
measure of decent work. Moreover, extending the sample size, for example to a time horizon of
more than 30 years or the geographical scope of the sample, may enhance the performance of
technological upgrading proxies in the wage modeling.
Secondly, to assess technological upgrading a useful indicator would be the amount of
domestic value-added in Chinese high-tech exports. Unfortunately, data on such an indicator is
not available in the statistical yearbooks. This indicator should be a better indicator than hightech exports/imports, since the classification of high-tech products and the statistical standard
could offer an overestimation of the Chinese technology level. For example, in 2015, Guangdong
province was still using “Chinese high-tech products catalog 2006” (Guangdong High-tech
90

Enterprise Association 2015 ). This catalogue is published almost 10 years ago and it is still the
most recent edition. There are so many products in this catalog (for example, computer keyboard,
mouse, and black and white TV) which should no longer be counted as high-tech products,
nonetheless, such products are still classified as high tech in the provincial statistical yearbook.
Finally, there are some other factors affecting wages which have not been captured by my
model which makes the R square low in all my models. Better wage determination variables are
needed to revise the model. For example, data on the number of strikes in each province, the
flow of migrant worker in each province, domestic value added in provincial exports would help
enhance the wage modeling.
Future research could focus on obtaining data for more provinces and extending the time
horizon so that the data set provides more observations and greater variation. Secondly, obtaining
additional indicators for labour market conditions and bargaining power should be considered in
future studies, as the SOE share of output is a limited proxy for workers’ bargaining power and
voice. Moreover, future research may also focus on discovering better technological upgrading
indicators to enhance the estimation result.
Technological upgrading, especially the domestic R&D in China, should be the engine of
enhancing the national income in the future development strategy. The Chinese manufacturing
sector should set its goal towards the global high-end manufacturing center and focus on
innovation and creation. Although the research results indicate that technological upgrading is a
way to improve decent work in terms of wages, simply relying on technological upgrading is not
sufficient. A successful economic development strategy must cooperate with good government
regulation and supervision which promotes decent work. Generating more decent work
opportunities is a crucial way to drive sustainable social and economic development

91

(International Labour Organization a). Future policy-making should enhance social equity and
justice, so that more people are able to access decent work.

92

Bibliography
Afonso, António, and Pedro Gomes. 2014. "Interactions between Private and Public Sector
Wages." Journal of Macroeconomics 39 (Part A): 97-112.
Alexander, Peter, and Anita Chan. 2004. "Does China Have an Apartheid Pass System?" Journal
of Ethnic and Migration Studies 30 (4): 609-629.
Asteriou, Dimitrios, and Stephen G. Hall. 2011. Applied Econometrics. London: Palgrave
Macmillan.
Baldwin, John R., Wulong Gu, Ryan Macdonald, and Beiling Yan. 2014. Productivity: What Is
It? How Is It Measured? What Has Canada's Performance Been Over the Period 1961 to
2012? The Canadian Productivity Review, Ottawa: Statistics Canada.
Banister, Judith. 2005. Manufacturing Employment and Compensation in China. Federal
Publications, New York: Cornell University.
Banister, Judith, and George Cook. 2011. "China's Employment and Compensation Costs in
Manufacturing through 2008." Monthly Labor Review 134 (3): 39-52.
Bartel, Ann P., and Nachum Sicherman. 1997. Technological Change and Wages: An InterIndustry Analysis. Working Paper No. 5941, NBER.
Beijing

Municipal Bureau of Statistics. n.d.
www.bjstats.gov.cn/sjfb/bssj/ndsj/pjgz.doc.

Accessed

November

3,

2016.

Berg, Jelle van den. 2013. Escaping the Middle Income Trap by Moving Up the Technology
Ladder? An Empirical Investigation. Master Thesis, Rotterdam: Erasmus University.
Accessed April 4, 2014. http://thesis.eur.nl/pub/14092/.
Blanchard, Olivier, and David Johnson. 2013. Macro Economics. New Jersey: Pearson
Education, Inc.
Bloom, Nicholas, Mirko Draca, and John Van Reenen. 2011. Trade Induced Technical Change?
The Impact of Chinese Imports on Innovation, IT and Productivity. Working Paper 16717,
Cambridge, MA 02138: NBER, 63. Accessed November 29, 2014.
http://www.nber.org/papers/w16717.
Böckerman, Petri, Seppo Laaksonen, and Jari Vainiomäki. 2010. "Micro and Macro Level Wage
Rigidity: Lessons from Finland." Finnish Economic Papers 23 (1): 27.
Boulin, Jean-Yves, Michel Lallement, Jon C. Messenger, and François Michon. 2006. Decent
Working Time: New Trends, New Issues. Geneva: International Labour Organization.
Braunstein, Elissa, and Gerald Epstein. 2002. Bargaining Power and Foreign Direct Investment
in China: Can 1.3 Billion Consumers Tame the Multinationals? Working Paper,
University of Massachusetts Amhest, Amherst: Political Economy Research Institute, 42.
93

Braunstein, Elissa, and Mark Brenner. 2007. "Foreign Direct Investment and Gendered Wages in
Urban China." Feminist Economics 13 (3-4): 213-237.
Cai, Fang, and Meiyan Wang. 2010. "Growth and Structural Changes in Employment in
Transition China." Journal of Comparative Economics 38 (1): 71-81.
Cai, Hongbin. 2014. "Overcoming the Middle Income Trap in China." In The Oxford Companion
to the Economics of China, by Shenggeng Fan, Ravi Kanbur, Shang-Jin Wei and Xiaobo
Zhang, 428-432. Oxford: Oxford University Press.
Chan, Kam Wing, and Li Zhang. 1999. "The Hukou System and Rural-urban Migration in China:
Processes and Changes." The China Quarterly No. 160: 818-855.
Coe, David T., and Elhanan Helpman. 1995. "International R&D Spillover." European Economic
Review 39 (5): 859-887.
Dembe, A. E., J. B. Erickson, R. G. Delbos, and S. M. Banks. 2005. "The Impact of Overtime
and Long Work Hours on Occupational Injuries and Illnesses: New Evidence from the
United States." Occup Environ Med 62 (9): 588-597.
Durbin, James, and Geoffrey Watson. 1950. "Testing for Serial Correlation in Least Squares
Regression, I." Biometrika 38 (3-4): 409-428.
Eichengreen, Barry, Donghyun Park, and Kwanho Shin. 2012. "When Fast-growing Economies
Slow Down: International Evidence and Implications for China." Asian Economic Papers
11
(1):
42-87.
Accessed
April
2,
2014.
http://www.mitpressjournals.org/doi/pdf/10.1162/ASEP_a_00118.
Eika, Torbjorn. 1993. Wage Equations in Macro Models. Discussion Paper, Research
Department, Central Bureau of Statistics, Norway.
Eloot, Karel, Alan Huang, and Martin Lehnich. 2013. A New Era for Manufacturing in China.
June.
Accessed
July
4,
2016.
http://www.mckinsey.com/businessfunctions/operations/our-insights/a-new-era-for-manufacturing-in-china.
Feenstra, Robert C., and Gordon H. Hanson. 1997. "Foreign Direct Investment and Relative
Wages: Evidence from Mexico's Maquiladoras." Journal of International Economics 42
(3-4): 371-393.
Fisher, Irving. 1973. "I Discovered the Phillips Curve: "A Statistical Relation between
Unemployment and Price Changes." The Journal of Political Economy 81 (2): 496-502.
Fu, Xiaolan, and Carlo Pietrobelli. 2011. "The Role of Foreign Technology and Indigenous
Innovation in the Emerging Economies: Technological Change and Catching-up." World
Development
39
(7):
1204-1212.
Accessed
July
13,
2014.
doi:10.1016/j.worlddev.2010.05.009.
Fu, Xiaolan, and Yundan Gong. 2011. "Indigenous and Foreign Innovation Efforts and Drivers of
94

Technological Upgrading: Evidence from China." World Development 39 (7): 1213-1225.
Accessed November 13, 2014. doi:10.1016/j.worlddev.2010.05.010.
Fujian Statistics Bureau. 2002-2015. Fujian Statistical Yearbook. Fuzhou: China Statistic Press.
Gabriele, Alberto. 2002. "S&T Policies and Technical Progress in China's Industry." Review of
International Political Economy 9 (2): 333-373.
Garz, Marcel. 2013. "Labour Market Segmentation: Standard and Non-standard Employment in
Germany." German Economic Review 14 (3): 349-371.
General Office of the State Council of the P.R.C. 1995. Communique of the State Council
People's Republic of China. Beijing: The Printing House of the Secretarial Bureau of the
General Office of the State Council.
Gibbon, Peter, and Stefano Ponte. 2005. Trading Down: Africa, Value Chains, and the Global
Economy. Philadelphia: Temple University Press.
Glass, Amy Jocelyn, and Kamal Saggi. 1999. "Foreign Direct Investment and the Nature of
R&D." Canadian Journal of Economics 32 (1): 92-117.
Glass, Amy Jocelyn, and Kamal Saggi. 1998. "International Technology Transfer and the
Technology Gap." Journal of Development Economics 55 (2): 369-398.
Greenwood, Jeremy, and Mehmet Yorukoglu. 1997. "1974." Carnegie-Rochester Conference
Series on Public Policy 46: 49-95.
Griffith, Breda. 2011. "Middle Income Trap." In Frontiers in Development Policy: A Primer on
Emerging Issues, by Raj Nallari, Shahid Yusuf and Breda Griffith, 39-43. Geneva: World
Bank
Group.
Accessed
January
6,
2015.
http://documents.worldbank.org/curated/en/2013/09/18220959/middle-income-trapsconceptual-empirical-survey.
Grossman, Gene, and Elhanan Helpman. 1991. Innovation and Growth in the World Economy.
Cambridge, MA: MIT Press.
Guangdong High-tech Enterprise Association. 2015 . Notice on the Declareation of High-tech
Products in Guangdong 2015. September 24. Accessed April 18, 2016.
http://www.gdhte.cn/front/frontManageAction.do?action_method=dispNewsByNewsIdT
oFront&newsId=12939.
Guangdong Statistics Bureau. 2002-2015. Guangdong Statistical Yearbook. Guangzhou: China
Statistic Press.
Guangxi Statistics Bureau. 2002-2015. Guangxi Statistical Yearbook. Nanning: China Statistic
Press.
Guo, Kai, and Papa N'Diaye. 2009. Is China's Export-Oriented Growth Sustainable? IMF
95

Working Paper, IMF.
Hainan Statistics Bureau. 2002-2015. Hainan Statistical Yearbook. Haikou: China Statistic Press.
Hu, Albert Guangzhou. 2014. "China's R&D Intensification and the Role of the Sate." In China's
Evolving Industrial Policies and Economic Restructuring, by Yongnian Zheng and Sarah
Y. Tong, 74-86. New York: Routledge Press.
Huang, Yasheng. 1999. Inflation and Investment Controls in China: The Political Economy of
Central-Local Relations during the Reform Era. Cambridge University Press.
Humphrey, John, and Hubert Schmits. 2002. "How Does Insertion in Global Value Chains Affect
Upgrading in Industrial Clusters?" Reginal Studies 36 (9): 1017-1027.
Hynninen, Sanna-Mari, Jari Ojala, and Jaakko Pehkonen. 2013. "Technological Change and
Wage Premiums: Historical Evidence from Linked Employer–Employee Data." Labour
Economics 24: 1-11.
International Labour Organization. 2012. Understanding Deficits of Productive Employment and
Setting Target. Geneva: International Labour Organization.
—. 2003. Guidelines Concerning a Statistcal Definition of Informal Sector. Conference Report,
Geneva: ILO Publications.
—. 2002. Women and Men in the Informal Economy: A Statistical Picture. Geneva: International
Labor Organization.
—. a. Decent Work. Accessed October 21, 2015. http://www.ilo.org/global/topics/decentwork/lang--en/index.htm.
—.

b.
Non-standard
Forms
of
Employment.
Accessed
March 7,
2015.
http://www.ilo.org/global/topics/employment-security/non-standard-employment/lang-en/index.htm.

—. 1993. Resolution Concerning Statistics of Employment in the Informal Sector. Geneva:
International Labor Organization.
Jiangsu Statistics Bureau. 2002-2015. Jiangsu Statiscial Yearbook. Nanjing: China Statistics
Press.
Jin, Liqun. 2015. How China Across the Middle-Income Trap. Commentator Article, Beijing:
People's Daily.
Jorgenson, Dale Weldeau, and Zvi Griliches. 1967. "The Explanation of Productivity Change."
The Review of Economic Studies 34 (3): 249-283.
Keller, Wolfgang. 1998. "Are International R&D Spillover Trade Related? Analyzing Spillovers
among Randomly Matched Trade Partners." European Economic Review 42 (8): 146996

1481.
Keller, Wolfgang. 2004. "International Technology Diffusion." Journal of economic Literature
42 (3): 752-782.
Kennedy, Charles, and Anthony P. Thirlwall. 1972. "Surveys in Applied Economics: Technical
Progress." The Economic Journal 82 (325): 11-72.
Kharas, Homi, and Harinder Kohli. 2011. "What Is the Middle Income Trap, Why Do Countries
Fall into It, and How Can It Be Avoided?" Global Journal of Emerging Market
Economies 3 (3): 281-189.
Knorringa, Peter, and Lee Pegler. 2006. "Globalisation, Firm Upgrading and Impacts on Labour."
Tijdschrift voor economische en sociale geografie 97 (5): 470-479.
Koopman, Robert, Zhi Wang, and Shang-Jin Wei. 2008. How Much of Chinese Exports Is Really
Made in China?-Assessing Domestic Value-added When Processing Trade Is Pervasive.
NBER Working Paper, 14109, Cambridge MA 02138: National Bureau of Economic
Research. Accessed November 16, 2014. http://www.nber.org/papers/w14109.
Kraemer, Kenneth L., Greg Linder, and Jason Dedrick. 2011. Capturing Value in Global
Networks: Apple's iPad and iPhone. Working Paper, Personal Computing Industry
Center, California: UC Irvine. 3, 11.
Lai, Desheng, Dahu Meng, Changan Li, and Qi Wang. 2014. 2014 China Labour Market
Development Report. Beijing: Beijing Capital Printing Co., Ltd.
Lamo, Ana, Javier J. Pérez, and Ludger Schuknecht. 2012. "Public or Private Sector Wage
Leadership? An International Perspective." The Scandinavian Journal of Economics 114
(1): 228-244.
Leslie, Derek. 1985. "Real Wage Growth, Technical Change and Competition in the Labor
Market ." The Review of Economics and Statistics 67 (4): 640-647.
Li, Qiang, and Zhuang Tang. 2002. "Migrant Workers and Informal Employment in Urban Area
in China." Social Sience (in Chinese) 13-25.
Liang, Yan. 2008. "Why Are China's Exports Special?" The Chinese Economy 41 (6): 99-118.
Lin, Ping, Zhuomin Liu, and Yifan Zhang. 2009. "Do Chinese Domestic Firms Benefit from FDI
Inflow? Evidence of Horizontal and Vertical Spillovers." China Economic Review 20 (4):
677-691. Accessed July 23, 2014. doi:10.1016/j.chieco.2009.05.010.
Loong-Yu, Au, and Nan Shan. 2007. "Chinese Women Migrants and the Social Apartheid."
Development 50: 76-82.
Mahadevan, Renuka. 2003. "To Measure or not to Measure Total Factor Productivity Growth?"
Oxford Development Studies 31 (3): 365-378.
97

Messenger, Jon C. 2004. Working Time and Workers’ Preferences in Industrialized Countries.
Finding the Balance. London and New York: Routledge.
Ministry of Commerce of the P.R.C. 2008. Ministry of Commerce, General Administration of
Customs, AQSIQ Decree No. 7 of 2008, "Mechanical and Electrical Products Import
Management
Approach".
April
30.
Accessed
May
14,
2016.
http://www.mofcom.gov.cn/article/b/e/200804/20080405506217.shtml.
Ministry of Science and Technology of the PRC. 2015. "China High-tech Industry Statistics
2014."
China
Science
Statistics.
Accessed
May
25,
2015.
http://www.sts.org.cn/sjkl/gjscy/index.htm.
Muth, John F. 1961. "Rational Expectations and The Theory of Price Movements." Econometrica
29 (3): 315-335.
Nadvi, Khalid. 2004. "Globalization and Poverty: How Can Global Value Chain Research Inform
the Policy Debate?" IDS Bulletin 35 (1): 20-30.
Nanjing Municipal Commission of Development. 2012. Jiangsu South and North Industry
Transfer Meeting Held in Nanjing (in Chinese). September 24. Accessed April 26, 2015.
http://fgw.nanjing.gov.cn/zwxx/dkzy/201209/t20120924_167421.html.
National Bureau of Statistics of P.R.C. 2015a. FAQs XIV, Household Surveys (8). May 15.
Accessed January 18, 2016. http://www.stats.gov.cn/tjzs/cjwtjd/201308/t20130829_74325.html.
—.

2015b. Household Surveys FAQs. May 15. Accessed August
http://www.stats.gov.cn/tjzs/cjwtjd/201308/t20130829_74325.html.

13,

2015.

—. 2015c. 2014 National Monitoring Report of Migrant Workers (in Chinese). April 29.
Accessed
December
22,
2015.
http://www.stats.gov.cn/tjsj/zxfb/201504/t20150429_797821.html.
—. 2014. China Populatio & Employment Statistic Yearbook 2014. Beijing: China Statistics
Press.
—. 2013. China Statistics Yearbook on High Technology Industry. Beijing: China Statisitc Press.
—.

2013.
Indicator
Explanation.
http://www.stats.gov.cn/tjsj/zbjs/.

October

29.

Accessed

June

3,

2015.

—. 2012. China Statistical Yearbook. Government Report, Beijing: China Statistical Publishing
House.
—. 2011. Statistics Classification of Micro-, Small-, Medium- and Large-enterprise (in Chinese).
September
2.
Accessed
December
15,
2014.
http://www.stats.gov.cn/statsinfo/auto2073/201310/t20131031_450691.html.
—. 2006. Provisions of the Enterprises Ownership Registered. December 5. Accessed March 12,
98

2015. http://www.stats.gov.cn/statsinfo/auto2073/201310/t20131031_450535.html.
—. 2002-2015. China Statistical Yearbook. Beijing: China Statistics Press.
—.

n.d. International Trade (in Chinese).
http://data.stats.gov.cn/easyquery.htm?cn=C01.

—.

n.d.
Annual
Data.
Accessed
http://data.stats.gov.cn/easyquery.htm?cn=C01.

Accessed

Feburary

December

12,
1,

2016.
2015.

National People's Congress of the People's Republic of China. 2011. China's 12th Five-Year
Plan. Government Report, Beijing: NPC. Accessed December 23, 2014.
http://www.npc.gov.cn/npc/xinwen/2011-03/17/content_1647851.htm.
Nunziata, Luca. 2005. "Institutions and Wage Determination: a Multi-country Approach." Oxford
Bulletin of Economics and Statistics 67 (4): 435-466.
OECD ilibrary. n.d. Research and Development. Accessed October 23, 2015b. http://www.oecdilibrary.org/content/indicator/6a8d10f4-en.
OECD. 2013a. OECD Economic Surveys China. Annual Report, Paris: OECD.
—. 2013 b. The People's Republic of China--- Avoiding the Middle-Income Trap: Policies for
Sustained and Inclusive Growth. Research Report, Paris: OECD.
—. 2010. Measuring Productivity-Measurement of Aggregate and Industry-Level Productivity
Growth. Research Report, Paris: OECD.
—. n.d. Global Value Chain. Accessed March 22, 2015. http://www.oecd.org/industry/ind/globalvalue-chains.htm.
Paus, Eva. 2012. "Confronting the Middle Income Trap: Insights from Small Latecomers."
Studies in Comparative International Development 47 (2): 115-138. Accessed April 16,
2014. doi:10.1007/s12116-012-9110-y.
PennState Eberly College of Science. n.d. Statistic Online. Accessed July 6, 2016.
https://onlinecourses.science.psu.edu/statprogram/node/137.
Pesando, James E. 1975. "A Note on the Rationality of the Livingston Price Expectations."
Journal of Political Economy 83 (4): 849-858.
Phillips, William. 1958. "The Relation between Unemployment and the Rate of Change of
Money Wage Rates in the United Kingdom, 1861-1957." Economica 25 (100): 283-299.
Pietrobelli, Carlo, and Roberta Rabelloti. 2006. Clusters and Value Chains in Latin America: In
Search of an Integrated Approach. Research Report, Washington, DC: Inter-American
Development Bank.
Porter, Michael E. 1985. Competitive Advantage: Creating and Sustaining Superior
99

Performance. New York: Simon and Schuster.
Romer, Paul. 1990. "Endogenous Technological Change." J. Polit. Econ. 98 (5): S71-S102.
Saggi, Kamal. 2002. "Trade, Foreign Direct Investment, and International Technology Transfer:
A Survey." The World Bank Research Observer 17 (2): 191-235.
Salter, Wilfred Edward Graham. 1960. Productivity and Technical Change. Cambridge:
Cambridge University Press.
Sasidharan, Subash, and Vinish Kathuria. 2011. "Foreign Direct Investment and R&D:
Substitutes or Complements-A Case of Indian Manufacturing After 1991 Reforms."
World Development 39 (7): 1226-1239.
Schneider, Hernriue. 2013. "An Austrian View on the Middle Income Trap." Journal of Chinese
Economic and Business Studies 11 (3): 169-178.
Schott, Peter K. 2008. "The Relative Sophistication of Chinese Exports." Economic Policy 32
(53): 5-49. Accessed November 11, 2014. doi:10.1111/j.1468-0327.2007.00195.x.
Schweitzer, Mark E. 2007. Wage Flexibility in Britain: Some Micro and Macro Evidence.
Working Paper, London: Bank of England.
Shandong Statistics Bureau. 2002-2015. Shandong Statistical Yearbook. Jinan: China Statistic
Press.
Sjöholm, Fredrik, and Nannan Lundin. 2013. "Foreign Firms and Indigenous Technology
Development in the People's Republic of China." Asian Development Review 30 (2): 4975. Accessed November 11, 2014. doi:10.1162/ADEV_a_00015.
Sjöholm, Fredrik, and Nannan Lundin. 2010. "Will Science and Technology Solve China's
Unemployment Problem?" Asian Economic Papers 9 (2): 1-28.
Smolny, Werner. 1998. "Innovations, Prices and Employment: A Theoretical Model and an
Empirical Application for West German Manufacturing Firms." The Journal of Industrial
Economics 46 (3): 359-381.
Sun, Jingshui, and Shuqin Ma. 1991. Econometrics. Beijing: Tsinghua University Press.
The Economist. 2011. Life at the Bottom of the Middle Kingdom. December 2. Accessed January
11, 2016. http://www.economist.com/blogs/freeexchange/2011/12/chinas-poverty-line.
Tian, Xiaofei, Baotai Wang, and Ajit Dayanandan. 2008. "The Impact of Economic Globalization
on Income Distribution: Empirical Evidence in China." Economics Bulletin 4 (35): 1-8.
Ueda, Shuzo, and Kazuo Ogawa. 2012. "On the Cost-reducing Effects of Embodied Technical
Progress: A Panel Study of the Steel Industry in Japan." Journal of Productivity Analysis
37 (2): 141-153.
100

UNCTAD. 2004. World Investment Report. New York and Geneva: United Nations Publication.
United Nation. 2015. Sustainable Development Goals. September 25. Accessed April 4, 2016.
http://www.un.org/sustainabledevelopment/sustainable-development-goals/.
Van Reenen, John. 1997. "Employment and Technology Innovation: Evidence from UK
Manufacturing Firms." Journal of Labor Economic 15 (2): 255-284.
Veugelers, Reinhilde, and Peter Vanden Houte. 1990. "Domestic R&D in the Presence of
Multinational Enterprises." International Journal of Industrial Organization 8 (1): 1-15.
Visco, Ignazio. 1984. Price Expectations in Rising Inflation. North Holland.
Wang, Lijian, Daniel Béland, and Sifeng Zhang. 2014. "Pension Fairness in China." China
Economic Review 28: 25-36.
White, Halbert. 1980. "A Heteroskedasticity-Consistent Covariance Matrix Estimator and a
Direct Test for Heteroskedasticity." Econometrics 48 (4): 817-838.
Wignaraja, Ganeshan. 2012. "Innovation, Learning, and Exporting in China: Does R&D or A
Technology Index Matter?" Journal of Asian Economics 23 (3): 224-233. Accessed
November 11, 2014. doi:10.1016/j.asieco.2011.02.001.
WikiAnswers.
n.d.
What
is
core
technology?
Accessed
http://www.answers.com/Q/What_is_core_technology?#slide=1.

April

2,

2016.

Williams, Richard. 2015. "Heteroskedasticity." University of Notre Dame . January 30. Accessed
July 5, 2016. https://www3.nd.edu/~rwilliam/stats2/l25.pdf.
Wong, Daniel Fu Keung, Chang Ying Li, and He Xue Song. 2007. "Rural Migrant Workers in
Urban China: Living A Marginalised Life." International Journal of Social Welfare 16 (1):
32-40.
Woo, Jaejoon. 2009. "Productivity Growth and Technological Diffusion through Foreign Direct
Investment." Economic Inquiry 47 (2): 226-248.
Woo, Jaejoon. 2012. Technological Upgrading in China and India: What Do We Know? Working
Papers No.308, OECD, Paris: OECD Publishing. Accessed October 3, 2014.
doi:10.1787/18151949.
Woo, Wing Thye. 2012. "China Meets the Middle-income Trap: the Large Potholes in the Road
to Catching-up." Journal of Chinese Economic and Business Studies 10 (4): 313-336.
World Bank. 2015a. Global Economic Prospects-Forecasts. Accessed Oct 10, 2015.
http://data.worldbank.org/country/china.
—. 2015b. FAQs: Global Poverty Line Update. September 30. Accessed Feburary 26, 2016.
http://www.worldbank.org/en/topic/poverty/brief/global-poverty-line-faq.
101

Xing, Yuqing. 2014. "China’s High-tech Exports: Myth and Reality." Asian Economic Papers 13
(1): 109-123.
Xu, Bin. 2010. "The Sophistication of Exports: Is China Special?" China Economic Review 21:
482-493. Accessed April 12, 2014. doi:10.1016/j.chieco.2010.04.005.
Xue, Jinjun, Wenshu Gao, and Lin Guo. 2014. "Informal Employment and Its Effect on the
Income Distribution in Urban China." China Economic Review 31: 84-93.
Yang, Ruiqiu. 2015. Industrial Upgrading and Economic Restructuring: Labor Shortage in
Guangdong Meets Turning Point Now (in Chinese). April 7. Accessed May 3, 2015.
http://www.gdeconomy.com/shendu/2015-04-07/6742.html.
Young, Stephen, and Ping Lan. 2010. "Technology Transfer to China through Foreign Direct
Investment." Regional Studies 31 (7): 669-679. Accessed November 14, 2014.
doi:10.1080/00343409750130759.
Yu, Miaojie. 2011. Moving Up the Value Chain in Manufacturing for China. Working Paper
2011/009, Beijing: Peking University.
Yuan, Gangming. 2013. GDP Heavily Rely on Investment Causes Pressures on Current Economy
(in
Chinese).
October
18.
Accessed
January
6,
2016.
http://www.chinanews.com/gn/2013/10-18/5398994.shtml.
Zhang, Anming, Yimin Zhang, and Ronald Zhao. 2003. "A Study of the R&D Efficiency and
Productivity of Chinese Firms." Journal of Comparative Economics 31 (3): 444-464.
Zhao, Changwen. 2010. "No Simple Relationship between Technological Improvement and
Employment Generation in a Large Developing Country like China: Thought Stimulated
by Sjöholm and Lundin." Asian Economic Papers 9 (2): 35-43.
Zhejiang Statistics Bureau. 2002-2015. Zhejiang Statistical Yearbook. Hangzhou: China Statistic
Press.
Zhu, Lei, and Bang Nam Jeon. 2007. "International R&D Spillovers: Trade, FDI, and
Information Technology as Spillover Channels." Review of International Economics 15
(5): 955-976.
Zhu, Yiying. 2010. "An Analysis on Technology Spillover Effect of Foreign Direct Investment
and Its Countermeasures." International Journal of Business and Management 5 (4): 178182.

102

Appendix: Descriptive statistics for data used in econometric model
Descriptive statistics for raw data (7 provinces, 2002-2014)
Inflation rate (%)

GDP (in billion
CNY)

Value of total
SOE production
(in billion CNY)

Value of total
industry output (in
billion CNY)

Mean
Median
Maximum
Minimum
Std. Dev.
Skewness
Kurtosis

2.5
2.3
7.8
-2.3
2.2
-0.1
2.6

2,118.5
1,567.3
6,779.2
64.3
1,802.8
0.9
2.8

314.7
197.5
1,822.6
12.3
385.3
2.2
7.8

3,907.5
2,483.6
14,301.7
25.0
3,880.0
1.1
3.2

Jarque-Bera
Probability

0.8
0.7

12.7
0.0

159.2
0.0

18.0
0.0

Sum
Sum Sq.
Dev.

228.9

192,786.6

28,640.1

355,585.2

427.0

292,000,000.0

13,400,000.0

1,350,000,000.0

Observations

91

91

91

91

103

Value of total
imports (in
billion USD)

Total actual
utilized FDI (in
billion USD)

Mean
Median
Maximum
Minimum
Std. Dev.
Skewness
Kurtosis

84.8
38.3
455.2
0.9
105.6
1.8
5.7

10.5
10.1
35.8
0.3
8.8
0.8
3.1

Value of total
mechanical &
electronics
exports (in
billion USD)
741.9
302.4
4,395.7
1.2
1,034.8
1.9
5.9

Jarque-Bera
Probability

73.5
0.0

10.0
0.0

85.7
0.0

83.9
0.0

Sum
Sum Sq. Dev.

7,716.6
1,003,517.0

955.0
6,938.2

67,514.4
96,378,655.0

4,015.6
393,017.7

Observations

91

91

91

91

104

Value of total
mechanical &
electronics imports
(in billion USD)
44.1
14.5
283.7
0.2
66.1
1.9
5.8

Mean
Median
Maximum
Minimum
Std. Dev.
Skewness
Kurtosis

Total number of
employed workers
over whole
province (in
1,000s)
35,539.9
34,865.3
66,065.0
3,498.9
18,797.8
-0.1
2.0

Jarque-Bera
Probability

4.0
0.1

15.9
0.0

25.2
0.0

79.8
0.0

3,234,128.0

15,030.6

2,864.6

224,381.9

31,800,000,000.0

2,248,025.0

97,292.6

327,175,400.0

91

91

Sum
Sum Sq.
Dev.
Observations

165.2
139.6
598.9
0.8
158.0
1.0
3.4

R&D internal
expenditure in
large and medium
enterprises (in
billion CNY)
31.5
20.2
137.5
0.1
32.9
1.2
3.8

Total number
of
manufacturing
workers (in
1,000s)
2,465.7
2,451.7
10,151.6
67.4
1,906.6
1.3
6.8

R&D employees in
large and medium
enterprises (in
1,000s)

105

91

91

Mean
Median
Maximum
Minimum
Std. Dev.
Skewness
Kurtosis

Value of total
high-tech
exports (in
billion USD)
37.0
10.5
256.4
0.0
59.7
1.9
6.0

Value of total
high-tech
imports (in
billion USD)
29.5
7.7
218.7
0.1
48.3
2.1
7.1

Jarque-Bera
Probability

89.9
0.0

Sum
Sum Sq.
Dev.
Observations

126.7
75.1
646.1
0.8
150.7
1.6
5.4

Manufacturing
sector average
annual wage
(CNY)
25,361.4
21,895.0
58,409.0
8,762.0
12,330.3
0.7
2.5

131.7
0.0

62.6
0.0

8.6
0.0

60.7
0.0

3,364.3

2,680.9

11,531.1

2,307,891.0

4,212,876.0

321,255.9

210,325.7

2,042,677.0

91

91

91

Value of total
exports (in
billion USD)

106

Number of patents
granted annually
46,295.3
15,937.0
269,944.0
199.0
63,905.1
1.7
5.0

13,700,000,000.0 368,000,000,000.0
91

91

Descriptive statistics for modeling data (7 provinces, 2003-2014)

Wage growth
speed (%)

Lagged
inflation rate

GDP growth
rate (%)

SOE share
(%)

Mechanical
export share (%)

Median
Maximum
Minimum
Std. Dev.
Skewness
Kurtosis

-0.2
-0.5

2.8
2.5

15.5
15.9

12.0
6.9

42.9
41.2

15.8
-18.4
5.8
-0.1
4.1

7.8
-2.3
2.0
-0.2
3.0

24.8
6.3
5.0
-0.1
1.9

65.0
0.7
11.6
2.5
9.8

69.9
10.0
18.4
0.1
1.9

Jarque-Bera
Probability

4.6
0.1

0.3
0.8

4.7
0.1

247.2
0.0

4.0
0.1

Sum
Sum Sq. Dev.

-15.8

234.6

1,298.9

1,008.6

3602.5

2,761.3

342.4

2,042.4

11,141.6

27980.6

Observations

84

84

84

84

84

Mean

107

R&D
employme
nt share
(%)

First
difference of
R&D
expenditure/G
DP (%)

Patents per
1,000
workers

Mechanical
import
share (%)

High-tech export
per manufacturing
worker (CNY per
worker)

Mean
Median
Maximum
Minimum
Std. Dev.
Skewness
Kurtosis

5.7
5.4
16.0
1.1
2.8
1.0
4.3

0.0
0.0
0.4
-0.8
0.2
-2.3
10.4

1.1
0.5
5.7
0.0
1.4
1.8
5.4

40.0
34.6
73.8
1.4
19.8
0.0
1.7

10,179.1
4,134.1
41,251.1
56.8
12,320.9
1.3
3.1

Jarque-Bera
Probability

19.1
0.0

265.1
0.0

67.1
0.0

5.9
0.1

22.8
0.0

Sum
Sum Sq.
Dev.

475.7
666.9

2.3
3.1

95.1
167.4

3,359.5
32,424.9

855,048.1
12,600,000,000.0

Observations

84

84

84

84

84

108

First difference
of high-tech
exports (in
billion USD)
4.4
0.9

First difference
of high-tech
imports (in
billion USD)
3.3
0.6

First difference of
high-tech imports per
manufacturing worker
(USD per worker)
457.4
177.5

First difference
of FDI per
worker (USD
per worker)
14.2
14.3

36.0
-25.4
9.8
1.2
5.6

33.9
-25.4
8.3
1.2
7.5

10,124.0
-13,032.3
3,191.4
-0.5
7.4

122.0
-213.5
49.4
-1.3
8.1

Jarque-Bera
Probability

42.8
0.0

90.6
0.0

72.5
0.0

115.8
0.0

Sum
Sum Sq.
Dev.

365.9

276.0

38,422.7

1,191.8

7,973.7

5,720.2

845,000,000.0

202,379.9

84

84

84

84

Mean
Median
Maximum
Minimum
Std. Dev.
Skewness
Kurtosis

Observations

109