China's Path To Innovation

Chinas Path to Innovation
Over the past three decades, China has experienced rapid economic growth
and a fascinating transformation of its industry. However, much of this
success is the result of industrial imitation and Chinas continuing success
now relies heavily on its ability to strengthen its indigenous innovation
capability. In this book, Xiaolan Fu investigates how China can develop a
strategy of compressed development to emerge as a leading innovative
nation. The book draws on quantitative and qualitative research that
includes cross-country, cross-province and cross-rm analysis. Large
multi-level panel datasets, unique survey databases, and in-depth industry
case studies are explored. Different theoretical approaches are also used to
examine the motivations, obstacles and consequences of Chinas innovation
with a wider discussion around what other countries can learn from Chinas
experience. This book will appeal to scholars and policy-makers working in
elds such as innovation policy, technology management, development and
international economics and China studies.
xiaolan fu is Professor of Technology and International Development
and Founding Director of the Technology and Management Centre for
Development at the University of Oxford. Her research interests include
innovation, technology and industrialisation; trade, foreign direct investment and economic development; emerging Asian economies; and innovation and productivity in the UK and US. She also has rst-hand
experience working in the business and academic sectors in China before
coming to the UK.
Professor Fu has integrated a variety of scholarly articles from the rm level to

the industry level to the national policy level, to produce the rst comprehensive
treatment of Chinese innovation activities from an open innovation perspective.
Her masterful book points the way towards open innovation with Chinese
characteristics.
Henry Chesbrough, Professor at UC Berkeleys Haas School of Business and
author of Open Innovation
That China is the workshop of the world is now a given. Professor Fu has invested
years in the search for clues to how the goods in the China workshop have moved
from being assembled in China through invented in China to invented and
commercialised in a complex open engagement with the worlds capital and skilled
labour. This book is the authoritative result and is essential reading.
Barbara Harriss-White, Emeritus Professor of Development Studies, Oxford
University and co-editor of China-India: Pathways of Economic and Social
Development
This fascinating book by a leading Chinese scholar is hugely informative of the
challenges China faces in its quest to become a major global innovative economy. It
populates a knowledge gap, challenges our conventional wisdom and provides
important insights for corporate and government policy makers alike.
Raphael Kaplinsky, Professor of International Development, The Open University
China achieved an average annual growth rate of 9.8% for 35 years, made
possible only by continuous technological innovations, after the transition from a
planning economy to a market economy in 1979. Such a long period of extraordinary growth was unprecedented in human history. This book carefully studies
Chinas open national innovation system at national, regional and rm levels. It
deciphers how China was able to achieve such a remarkable success in the past,
examines how China may sustain dynamic growth in the future and suggests what
other countries can learn from Chinas success. The book is a must-read for anyone
who wants to understand Chinese economic development.
Justin Yifu Lin, Professor, Peking University and Former Chief Economist, The
World Bank
Finally, we have an analytical volume that combines economic theory, international experience, and Chinas socio-economic conditions to formulate a most
credible strategy to greatly strengthen Chinas capacity to innovate. Xiaolan Fus
Open National Innovation System approach deserves careful study by other
developing countries because it is also applicable to them.
Wing Woo, Professor, University of California at Davis and President, Jeffrey
Cheah Institute on Southeast Asia, Malaysia
Chinas Path
to Innovation
xiaolan fu
University of Oxford
University Printing House, Cambridge CB2 8BS, United Kingdom

Cambridge University Press is part of the University of Cambridge.
It furthers the Universitys mission by disseminating knowledge in the pursuit of
education, learning and research at the highest international levels of excellence.
www.cambridge.org
Information on this title: www.cambridge.org/9781107046993
Xiaolan Fu 2015
This publication is in copyright. Subject to statutory exception
and to the provisions of relevant collective licensing agreements,
no reproduction of any part may take place without the written
permission of Cambridge University Press.
First published 2015
Printed in the United Kingdom by Clays, St Ives plc
A catalogue record for this publication is available from the British Library.
Library of Congress Cataloguing in Publication data
Fu, Xiaolan, 1967
Chinas path to innovation / Xiaolan Fu.
pages cm
ISBN 978-1-107-04699-3 (hardback)
1. Economic development China. 2. Industrial policy China.
3. Information technology Management China. I. Title.
HC427.95.F798 2015
3380 .0640951dc23
2014048687
ISBN 978-1-107-04699-3 Hardback
Cambridge University Press has no responsibility for the persistence or accuracy of
URLs for external or third-party internet websites referred to in this publication,
and does not guarantee that any content on such websites is, or will remain,
accurate or appropriate.
Contents
List of gures
List of tables
page vii
x
Preface
xiii
List of abbrevations
xvii
Introduction
Introduction
Innovation in China since the reforms: An overview
15
Part I International knowledge transfer and technological

takeoff
45
Foreign direct investment, absorptive capacity and regional

innovative capabilities: Evidence from China
47
Processing trade, FDI and international competitiveness of

the Chinese high-technology industries
74
3
4
5
Indigenous and foreign innovation efforts and

technological upgrading in China
Part II
Development of indigenous innovation capacity

and catch-up
108
139
The role of state policy in shaping innovation practices: The

case of open innovation in China
141
Open innovation as a response to constraints and risks
170
The dual role of universities in industrial innovation:

comparing China and the UK
201
vi
10
11
Contents
Technological learning, tacit knowledge acquisition and

industrial upgrading: The Chinese optical bre and cable
industry
236
Leapfrogging in green technology: The solar-PV industry

in China and India
256
Part III
277
Towards a global innovation leader
Internationalisation, reverse learning and capabilities

upgrading: The case of Huawei and ZTE
279
12
International collaboration and radical innovation
314
13
Innovation efciency and the cross-country gap in

innovation
325
Capabilities, incentives, institutions and national

innovation performance
358
14
Conclusions
15
379
Conclusions: Open national innovation system and Chinas

path to innovation
381
References
396
Index
430
Figures
2.1 Chinas R&D expenditure, 19952012

page 17
2.2 R&D expenditure in China and other economies,
19952012
18
2.3 Composition of R&D by types of activities: basic
research, applied research and experimental
development
19
2.4 R&D composition by funding sources, 20032011
20
2.5a Total number of R&D personnel
21
2.5b R&D personnel by executive entity, 2009
21
2.6 Comparison of R&D personnel per thousand employees
in China and OECD economies, 20002011
22
2.7a Expenditure for acquisition of foreign technology in
China, 20002012
23
2.7b Foreign technology imports in China, percentages, 2012
24
2.8 Number of patent applications vs number of granted
patents, 19952012
25
2.9 Triadic patent families across BRICS, 19992011
26
2.10 Number of triadic patent families in China and other
OECD economies, 19992011
27
2.11 Output of R&D: index articles published in international
journals
28
2.12 Citations of scientic papers published, 20002010
28
2.13 Innovation performance: international comparison,
2012
29
2.14 R&D expenditures and intensity of enterprises in the
above-scale manufacturing industries, 2012
30
2.15 Number of R&D personnel across sectors, 2009
31
2.16 Innovation output across industries: number of patent
applications, 2010
32
2.17 Innovation outputs across industries: new product sales
and new product export values, 2012
33
vii
viii
List of gures
2.18 Export and import volume of high-tech products,

19952012
2.19 Innovation outputs by ownership of rms: new product
sales and new product export values, 2012
2.20 R&D expenditures across different ownership structures
in industrial enterprises above designated size,
20062012
2.21 R&D expenditures across different ownership structures
in industrial enterprises above designated size, 2006 and
2012
2.22 R&D personnel across different ownership structures in
industrial enterprises above designated size, 20062012
2.23 Patent applications across different ownership structures
in industrial enterprises above designated size,
20092012
2.24 New product sales and new product exports across
different ownership structures in industrial enterprises
above designated size, 20092012
2.25 R&D expenditures and R&D/GDP ratios across China,
2012
2.26 R&D expenditures across China, 2000 and 2012
2.27 Ratio of R&D expenditures in GDP across China, 2000
and 2012
2.28 Mean patent application numbers across China, 2000
and 2012
2.29 R&D funding sources across China, 2009
3.1 Trade and FDI in China, 19852004
3.2 Regional distribution of FDI stock in China, 2005
3.3 Regional distribution of number of invention
applications, 2004
3.4 Regional distribution of sales of new products, 2004
3.5 Regional distribution of industrial R&D, 2004
3.6 Innovation activities: foreign and indigenous enterprises
in China, 2004
3.7 Regional innovation efciency, China
4.1 Share of processing exports in Chinas total
high-technology exports
4.2 Percentage of processing exports in total exports,
electronic industry of China, 2006
34
35
36
36
37
38
39
40
40
41
42
42
53
53
56
56
58
60
67
81
92
List of gures
5.1 Technical efciency by ownership and technology

category
5.2 TFP growth in Chinese manufacturing rms, 20012005
5.3 Distribution of rms on the frontier
5.4 Technical change and efciency improvement in Chinese
manufacturing rms, 20012005
8.1 Percentage of innovative Chinese rms reporting R&D
collaboration with various types of partners
8.2 Percentage of Chinese rms reporting R&D
collaboration with universities in various countries
8.3 Percentage of UK rms reporting R&D collaboration
with various types of partners
8.4 Percentage of UK rms reporting R&D collaboration
with universities in various countries
9.1 The production chain in the optical bre and cable
industry
10.1 Technology transfer and indigenous knowledge-creation
mechanisms adopted by solar-PV industry in China and
India
10.2 Mixing and sequencing of technology creation and
acquisition mechanisms in solar-PV industry in China
and India
11.1 Stock of Chinas outward direct investment, mil$
11.2 Manufacturing rms with overseas investment
11.3 Main incentives for investing in developed countries
11.4 Main incentives for investing in developing countries
11.5 The reverse learning chain from customers
13.1 National innovation system: a system theoretic
framework
13.2 Total national R&D as percentage of GDP for major
industrial countries
13.3 Patents granted by USPTO, 19822006
13.4 Patenting capacity of selected countries
13.5 Patenting efciency of selected countries
13.6 Expected innovation capacity: emerging economies
13.7 Actual performance: patents granted by USPTO
13.8 Innovation efciency of emerging economies
14.1 National innovation performance
ix
122
123
124
125
216
216
226
227
242
269
270
284
286
286
287
298
330
333
334
347
347
349
350
351
360
Tables
3.1 FDI intensity: share of foreign assets in total industrial

assets
page 55
3.2 Expenditure on technology acquisition, 2004
57
3.3 Growth rate of R&D expenditure
59
3.4 Impact of FDI on regional innovation capacity
64
3.5 Impact of FDI on regional innovation efciency
68
3.6 The FDI-innovation-growth linkage in the coastal and
inland regions
70
4.1 Summary statistics of the variables
85
4.2 Exports and innovation indicators of domestic and
foreign rms, 2007
91
4.3 Export performance of top 10 exporting hightechnology industries: SIC four-digit level, 2007
93
4.4 Probit model estimates of export decisions in the hightechnology industry
95
4.5 Spillovers from PT-FDI and export value of domestic
rms
98
4.6 Determinants of export performance in industries with
fast indigenous export growth: IV estimates
101
4.7 Spillovers from PT-FDI and export performance of
indigenous rms: IV estimates of alternative spillover
measurements
103
5.1 Summary statistics of the variables
115
5.2 Determinants of TFP growth and technical change:
GMM estimates of full sample of domestic rms
126
5.3 Determinants of technical change: GMM estimates by
industry group
129
5.4 Determinants of efciency improvement: GMM
estimates by industry group
130
6.1 Sources of product innovation
143
6.2 Sources of process innovation
144
x
List of tables
6.3 The importance of information sources in innovation

activities
6.4 The main public policies for open innovation in China
6.5 Selected cases of open innovation in Chinese rms
6.6 Types of innovation mode and frequencies
7.1 Descriptive statistics and correlations
7.2 Determinants of breadth of openness: ordered-Logit
model estimates
7.3 Determinants of depth of openness: ordered-Logit model
estimates
7.4 Robustness check using OLS
7.5 Firms responses to constraints and risks by ownership
7.6 Effects of constraints and risks by size and industry
8.1 Denition of variables
8.2 Universities and rm innovation in China: Tobit model
estimates
8.3 Robustness check: Generalised Tobit model estimates
8.4 Robustness check: instrumental variable model estimates
8.5 Universities and rm innovation in selected universityconcentrated cities in China
8.6 Universities and rm innovation in the UK: standard
Tobit model estimates
9.1 Perception of the technological characteristics of the
optical bre and cable industry
9.2 Denition of variables
9.3 Factor analysis
9.4 Knowledge sources and technological learning
9.5 Does the embeddedness of tacit knowledge in the
industry matter?
9.6 Does the embeddedness of tacit knowledge in the
industry matter?
10.1 Renewable technology targets for 2020 in China
10.2 Renewable technology targets for 2022 (end of
13th Plan) in India (in MW)
10.3 Details of leading solar-PV companies in China and
India
10.4 Sustainability-oriented innovation systems of China and
India
xi
145
156
160
164
183
184
186
188
191
194
212
218
221
224
225
228
244
245
247
249
251
252
263
265
266
272
xii
List of tables
11.1 Industry distribution of Chinas OFDI: by FDI stock and

number of subsidiaries, 2010
11.2 Innovativeness of rms: comparing rms with and
without OFDI
11.3 Huaweis history of alliances
12.1 Collaborative innovation activities
12.2 Types of collaborators and regional distribution
12.3 The impact of collaboration on innovation performance
in Chinese rms: Tobit model estimation results
12.4 The impact of international collaboration on novel
innovation in China: Tobit model estimation results
13.1 Denition and sources of variables
13.2 Stochastic frontier analysis of basic patenting capacity
13.3 Determinants of patenting efciency
285
288
301
317
318
319
321
341
342
344
Preface
This book is an outcome of my research on innovations in China over

the past 10 years. I started my research on innovation as a project team
member of a Cambridge-MIT joint project on International Innovation
Benchmarking, which compared innovation activities in Europe, the
UK in particular, and the United States. Innovations were a key area of
concern for policy makers and business managers, as well as the academic scholars. Although Chinas economic reforms started in 1978, up
to the early 2000s the Chinese economy had been driven by system
reforms, investment, exports and foreign direct investment (FDI) at
different periods. Innovation was still a concept that had never really
come to the centre of Chinas development strategy, nor had it arrived at
the debate on the drivers of its long-term economic growth. Having
witnessed the competition among industrialised countries for leadership
in innovation and the priority that has clearly been given to skills and
innovation in policy making in these countries, I started my pursuit to
understand how to build innovation capabilities in a developing country such as China. I strongly believe that this will be a crucial area for
China, not only for academic research but also for policy making and
business management in the real-world context. In fact, without much
waiting, the Chinese government began to change its development
strategy in 2006. It placed development of indigenous innovation as
the top priority in its ofcial national development plan.
As a researcher who has a strong interest in Chinas industrial competitiveness and its external trade and investment, I have worked extensively in these areas; hence, I started my research quest from my
understanding of the sources of innovative knowledge and technology
in China and the processes of innovation diffusion. I gradually moved
on to researching the conditions and processes of innovation creation in
developing countries. I also examined the relationship between different
paths at different stages of development of a country, their advantages
and disadvantages and the conditions for effectiveness of each path. My
xiii
xiv
Preface
past work experiences and interest in studying innovations in the developed and other developing countries allowed me to examine Chinas
path from an international comparative perspective. This has formed
the basic structure and approach of research that I present in this book.
Chinas Path to Innovation is a combination of a selected number of
my published journal papers and several new studies on some of the most
recent topics regarding Chinas ongoing transformation from imitation to
innovation. It is a serious academic book based on 10 years of research
and reection. All published and unpublished new papers are selected and
organised to provide a systematic, comprehensive and coherent study of
Chinas path to innovation, although each chapter is fairly self-contained.
Journal publication offers a great advantage that ones research comes
under close scrutiny through the peer review process and can benet
greatly from it. My aim is to publish the more original parts of the book
in this way with updated data and information. The hope is that the
whole will add up to more than the sum of its parts and that we can
identify and develop a general model of the technology development
strategy of the developing countries based on a series of peer-reviewed,
in-depth studies of individual factors, mechanisms and cases.
Many acknowledgements and thanks are due. Since 2005, my research
on innovation and on China has been supported by substantial grants
from the Economic and Social Sciences Research Council (ESRC), the
Engineering and Physics Sciences Research Council (EPSRC), the British
Academy, the Cairncross Foundation and the State Administration of
Foreign Experts Affairs (SAFEA) of China. I thank these bodies for their
nancial support and for the condence that they have shown in my
research.
I am grateful to the copyright holders of the following journals for
permitting me, with acknowledgement, to present full or part of the
papers that have appeared in their journals in suitably revised form:
Xiaolan Fu, Carlo Pietrobelli and Luc Soete, The role of foreign
technology and indigenous innovation in emerging economies: technological change and catch-up, World Development, 39 (2011) (part of
the Introduction); Xiaolan Fu, Foreign direct investment, absorptive
capacity and regional innovation capabilities in China, Oxford
Development Studies, 36 (2008) (Chapter 3); Xiaolan Fu, Processingtrade, FDI and exports of indigenous rms: rm-level evidence from hightechnology industries in China, Oxford Bulletin of Economics and
Statistics, 73 (2011) (Chapter 4); Xiaolan Fu and Yundan Gong,
Preface
xv
Indigenous and foreign innovation efforts and drivers of technological

upgrading, World Development, 39 (2011) (Chapter 5); Xiaolan Fu
and Hongru Xiong, Open innovation in China: policies and practices,
Journal of Science & Technology Policy in China, 2 (2011) (Chapter 6);
Xiaolan Fu, Hongru Xiong and Jizhen Li, Open innovation as a response
to constraints and risks and the moderating role of governance, Asian
Economic Papers, 13 (2014) (Chapter 7); Xiaolan Fu, Jizhen Li and
Martin Johnson, Internal and external sources of tacit knowledge: evidence from the optical bre and cable industry in China, Journal of
Chinese Economic and Business Studies, 9 (2011) (Chapter 9); Xiaolan
Fu and Jing Zhang, Technology transfer, indigenous innovation and
leapfrogging in green technology: solar-PV panel industries in China
and India, Journal of Chinese Economic and Business Studies, 9
(2011) (Chapter 10); Xiaolan Fu and Qing Gong Yang, Exploring the
cross-country gap in patenting: a stochastic frontier approach, Research
Policy, 38 (2009) (part of Chapter 13); Xiaolan Fu and Rongping Mu,
Enhancing Chinas innovation performance: the policy choice, China &
World Economy, 22 (2014) (part of Chapter 14).
Chapter 8 is based on a paper written jointly with Jizhen Li published
by the Oxford University Technology and Management Centre for
Development Working Paper series. Chapter 11 is based on a new
unpublished paper written jointly with Zhongjuan Sun. I would like to
thank my co-authors for allowing results of our joint research to be
included in this volume. Prototypes of various chapters have been presented, in the past years, at seminars at many universities and various
conferences, which cannot be acknowledged here in detail. I am grateful
to the organisers and participants for providing me with these opportunities to obtain helpful feedback on the research from experts.
Among the numerous colleagues and friends whom I wish to thank for
helpful and constructive comments and discussions are Luc Soete, Henry
Chesbrough, Alan Hughes, Adrian Wood, Jinglian Wu, Justin Yifu Lin,
Barbara Harriss-White, Edmund Valpy FitzGerald, Wing Thye Woo,
Anne Miroux, Carlo Pietrobelli, Angang Hu, Rongping Mu, Erniu
Xuan, Rajneesh Narula, Pierre Mohnen, Raphie Kaplinsky, Mammo
Muchie, Xin Fang, Wei Lv, Jiangang Victor Zhang, Mingjie Ma, Jian
Gao, Poh Kam Wong, Jin Chen, Guisheng Wu, Jizhen Li, V. N.
Balasubramanyam, Cyril Lin, John Knight, Andy Cosh, Michael
Kitson, Carl Dahlman, David Kaplan, Jose Katz, Frances Stewart, Steve
Rayner, Christine Greenhalgh, Yongding Yu, Linda Yueh, Yang Yao,
xvi
Preface
Jing Zhang, Yundan Gong, Hongru Xiong, Qing Gong Yang, Marc
Ventresca, Diego Sanchez-Ancocher, Eric Thun, Bin Hao, Chunyan
Zhang and Xiaming Liu.
I am grateful to Jun Hou, Martin Johnson, Peter Luke and Zifu Wang
for excellent research assistance. The support of Paula Parish and Claire
Wood of Cambridge University Press was essential for the publication
of the book.
I thank the Department of International Development of Oxford
University for hosting the research and providing nancial support
and the Centre for Business Research of Cambridge University for
support during the early stage of my research on innovation and
China. I also thank colleagues and students of the Technology and
Management Centre for Development, Oxford University, for stimulating discussions and help. The collegiality of the Fellowship of Green
Templeton College, Oxford, has also been an inspirational support.
Finally, I can hardly express the debt of gratitude I owe to my family,
especially my husband, Shaohui, and my son, Yujie, for their great love
and support. Without their support, the book would not have come to
fruition.
Xiaolan Fu
Abbreviations
ABC
CDB
CDMA
CILG
CIS
COEs
CPO
DEA
EIS
EP
EPZs
EU
FDI
FIEs
GDP
GLLAMMs
GMM
GVC
HKTM rms
HMT
ICT
IMD
IPR
ITU
JV
MLEs
MNEs
MOST
MSTI
NIEs
Absorptive capacity
China Development Bank
Code division multiple access
Central innovation leading group
Community Innovation Survey
Collective-owned enterprises
Chinese State Intellectual Property Ofce
Data envelopment analysis
European Innovation Scoreboard
Export processing
Export processing zones
European Union
Foreign direct investment
Foreign invested enterprises
Gross domestic products
Generalized linear latent and mixed models
Generalised method of moments
Global value chains
Hong Kong, Taiwan and Macao rms
Hong Kong, Macau and Taiwan
Information and communication technology
Institute for Management Development
Intellectual property rights
International Telecommunication Union
Joint venture
Medium- and large-sized enterprises
Multinational enterprises
Ministry of Science and Technology of China
Main science and technology indicators
Newly industrialised economics
xvii
xviii
NIP
NIS
OBM
ODM
OECD
OEM
OFDI
OI
OLS
ONIS
OT
POEs
PRIs
PT-FDI
R&D
REF
S&T
SEZs
SFA
SHCs
SMEs
SOEs
SSB
STP
TFP
UNCTAD
USPTO
WBDI
WCY
WFO
WTO
ZTE
List of abbrevations
National innovation performance

National innovation system
Original brand manufacturers
Original design manufacturers
Organisation for Economic Co-operation and
Development
Original equipment manufacturers
Outward foreign direct investment
Open innovation
Ordinary least squares
Open national innovation system
Ordinary trade
Privately owned enterprises
Public research institutions
Processing tradeFDI
Research and development
Research excellence framework
Science and technology
Special economic zones
Stochastic frontier analysis
Shareholding companies
Small and medium enterprises
State-owned enterprises
State Statistical Bureau
Strategic technology partnering
Total factor productivity
United Nations Conference on Trade and Development
United States Patent and Trademark Ofce
World Bank development indicators
World Competitiveness Yearbook
Wholly foreign owned
World Trade Organization
Zhongxing Telecommunications Equipment
Corporation
Introduction
Introduction
The past three decades have witnessed rapid economic growth and a
fascinating transformation of Chinas economy and industry, from
an economy dominated by agriculture to one that is referred to as a
world manufacturing plant, from a small exporter of resource- and
unskilled-labourintensive products to a major producer of manufactured exports. The total industrial output of China increased from
US$91 million in 1980 to US$3,728 million in 2013, and the share of
industrial products in total exports has increased from 50 per cent in
1980 to more than 95 per cent in 2012 (NBS, 2013).
Increasing industrial competitiveness as revealed through surging
exports and upgraded export composition has also astonished the
rest of world. Chinas total exports and imports increased from
US$38 billion in 1980 to US$4,265 billion in 2012. Chinas share in
the world markets for exports of goods rose from 0.9 per cent in 1980 to
11 per cent in 2012. More signicant is the export of manufactured
products from China, which increased from US$9 billion in 1980 to
US$1,948 billion in 2012, 38 per cent of which constituted hightechnology products, accounting for 16.5 per cent of the worlds total
high-technology exports (UNCTAD, 2014). China is now the worlds
largest economy in terms of trade. The country has also maintained a fast
growth rate despite the recent global economic crisis that severely
affected the industrialised economies.
However, the country also faces signicant criticisms of its growth
model because of its heavy dependency on foreign technology transfer
and imitation and its lack of creativity and indigenous capabilities in
core technology. Moreover, with the amount of surplus unskilled
labour in China falling, and the resource and environmental constraints
for sustainable growth becoming increasingly signicant, China is being
forced towards a more skill-intensive and technology-intensive growth
path as its own Lewis Turning Point approaches, that is, when the
surplus labour in the subsistence sector is fully absorbed into the
3
Introduction
modern sector. China now faces signicant challenges in moving from

imitation to innovation. The success of this transformation will be of
crucial importance for China to avoid the middle-income trap and
sustain its long-term economic growth (Wu, 2013).
Therefore, given Chinas remarkable achievement in industrialisation
and modernisation in the past three decades as well as the challenges of
sustainable development and structural change ahead, it is pertinent to
ask the following questions: What was Chinas path to innovation in the
past, and whither the future? How has China managed to develop and
upgrade its technological capabilities at such a remarkable speed? In the
twenty-rst century, how can China signicantly enhance its indigenous
innovation capability and accomplish the transition from imitation to
innovation, thereby becoming an innovative nation? As one of the
major economies in the world, how can China develop a path of compressed development and leapfrog the conventional latecomer path of
imitative industrialisation, progressing up the value chain, taking a lead
in the low-carbon industrial revolution and reemerge as the worlds
leading innovation power as documented for an earlier era in Joseph
Needhams (1954) seminal work? Is there a China model of innovation?
What are the lessons that other countries can learn from Chinas experience? These are all important questions of great interest not only to
academic researchers but also to policy makers and practitioners.
As the worlds second largest economy and one that is still rmly on a
path of stable and promising economic growth, any fundamental
changes in China will have signicant impact on global business and
global economies. Moreover, China has increasingly been seen as an
exemplary model for other emerging economies. Its successes and challenges will thus be closely watched by policy makers in both developed
and emerging economies. Therefore, ndings from this book will have
signicant policy and practical implications for both developed and
other developing countries.
Innovation and its sources

Innovation is a process of creative destruction, taking place as a process
of industrial mutation that incessantly revolutionises the economic
structure from within, incessantly destroying the old one, incessantly
creating a new one (Schumpeter, 1942: 83). It is widely recognised
as a major driver of long-term growth and a key element of
Introduction
industrialisation and catch-up in developing countries (Romer, 1990). In

the present context, innovation concerns not only novel innovations but
also innovation via diffusion of existing ideas and techniques. It includes
not only technological innovations but also non-technological innovations, such as new management practices and new institutional structures. In other words, innovation refers to the introduction or adoption of
new products, new production processes, new ways of organisation and
management, new methods of marketing and new business models. A
complete innovation chain includes both the creation and commercialisation of new knowledge.
Innovation can occur as a result of a concerted focus by a range of
different agents, by chance or as a result of changes in industry structure, in market structure; in local and global demographics; in human
perception, mood and meaning; in the amount of already available
scientic knowledge, and so on (Drucker, 1985). At the micro level,
the sources of innovation may come from internal focus efforts, for
example, R&D activities or other organised innovation efforts, or
externally from the acquisition of useful technology or knowledge
created by other organisations or by users of the technology, that is,
the so-called end-user innovation identied by von Hippel (1988).
Currently, with the innovation paradigm shifting from closed to open,
rms may also open up their innovation process and create new products and processes by tapping into external resources and collaborating
with other partners (Chesbrough, 2003). At the macro level, innovations may be created by focused efforts or by chance from a range of
different agents in the country, such as rms, universities and research
institutions. They may also emerge as a result of acquisition of innovations created in foreign countries through several channels.
Innovation can be diffused between rms and across regions and
countries through various transmission mechanisms. These include
(1) licensing; (2) movement of goods through international trade, especially imports; (3) movement of capital through inward and outward
foreign direct investment (FDI and OFDI); (4) movement of people
through migration, travel, and foreign education of students and workers; (5) international research collaboration; (6) diffusion through
media and the Internet of disembodied knowledge; and (7) integration
into global value chains to benet from the foreign technology transferred within the supply chain. Some knowledge is transferred intentionally from the knowledge owner to the recipient and this may spur
Introduction
a learning process but a large proportion of knowledge spillovers take

place as unintended knowledge leakage. In recent years, the mode of
innovation is becoming more and more open and good use is made of
external resources. International knowledge diffusion can therefore
benet rms innovation at every stage of the innovation process. The
growing technological diversication of companies makes successful
integration of new external knowledge into the innovation process
increasingly important. Such successful integration further fosters innovation performance. The factors that explain the accelerating trend of
utilising external sources of knowledge include, among others, technological convergence, declining transaction costs of acquiring external
R&D inputs and shortening product cycle times (Narula, 2003).
The development strategies for industrialisation and catch-up in latecomer developing countries, the relative role of international technology transfer and indigenous innovation and the role of industry policy
in the process have been the most important but also controversial
issues in development studies and science and technology studies. One
of the controversies is whether the sources of technological change are
indigenous or rather based on foreign innovation efforts or a combination of the two, and which combination of different foreign sources of
innovation with different degrees of emphasis. On the one hand, innovation is costly, risky and path dependent. Hence, it is more efcient for
developing countries simply to acquire foreign technology created in
developed countries. In principle, if innovations are easy to diffuse and
adopt, a technologically backwards country can catch up rapidly, even
leapfrog through the acquisition and more rapid deployment of the
most advanced technologies (Soete, 1985; Grossman and Helpman,
1991, 1994; Romer, 1994; Eaton and Kortum, 1995).
On the other hand, there is the view that technology diffusion and
adoption are neither costless nor unconditional. They rely on substantial and well-directed technological efforts (Lall, 2001) and on absorptive capacity (Cohen and Levinthal, 1989). An additional related
difculty in the debate on indigenous versus foreign technology upgrading is that technical change is often biased in a particular direction so that
foreign technologies developed in industrialised countries may not be
appropriate to the economic and social conditions of developing
countries (Atkinson and Stiglitz, 1969; Basu and Weil, 1998;
Acemoglu, 2002; Fu and Gong, 2010). In addition, we cannot simplistically assume that the private interests of multinationals coincide with the
social interests of the host countries (Lall and Urata, 2003). The available
Introduction
empirical evidence on the effects of the sources of indigenous or foreign

innovation is mixed. Studies largely fail to provide convincing evidence
indicating signicant positive technological transfer and spillover
effects of FDI on local rms (Gorg and Strobl, 2001).
Accompanying this ongoing and inclusive debate on the role of technology transfer and indigenous innovation, the role of state and industry
policy in the process of industrialisation and economic development is also
subject to a wide, ongoing debate. While some argue that industry policy is
crucial for the success of the newly industrialised economics (NIEs) such as
Japan, South Korea and Signapore (Amsden, 2001; Chang, 2003; Pack
and Saggi, 2006), there are also strong arguments for the role of the market
and free competition in allocating resources efciency and enhancing the
productive efciency of the enterprises in an economy based on the recent
success of the East Asian Tigers such as Malaysia, Thailand and the
Philippines (Kruger, 1974; Bhagwati, 1982; World Bank, 1996, 2005).
Some argue for a third way for structural change to occur, suggesting
that sustained economic development is driven by changes in factor
endowments and continuous technological innovation; therefore, industry policy should encourage the development of sectors that comply with a
countrys comparative advantage while the private sector and the market
should be the major players in the process (Lin, 2011). Market forces and
private entrepreneurship would be in the driving seat of this agenda, but
governments would also perform a strategic and coordinating role in the
productive sphere beyond simply ensuring property rights, contract
enforcement, and macroeconomic stability (Rodrik, 2004). This debate
on the role of the state and policy is relevant for our analysis of national
innovation capabilities and performance because of the nature of innovation as a public product, the signicant positive externalities that knowledge and ideas may generate and the presence of market failures resulting
from the great uncertainty related to the innovation process.
The literature
Chinas experience with innovation and technological upgrading is also
the subject of wide-ranging interests amongst a variety of stakeholders
in economics and politics. The literature in this area can be broadly
classied into several categories. The rst category relates to studies of
the impact of Chinas rising innovation and technological capabilities
on the rest of the world, for example, MacDonald et al. (2008), Barlow
(2013) and Someren and Someren-Wang (2013). These studies argue
Introduction
that the United States, the EU and China have reached a crossroads, and
whether China will be a threat or an opportunity depends on the main
players in government and public and private organisations rethinking
their innovation policies and business development paths (Someren and
Someren-Wang, 2013). They also contend that the rules for survival in
R&D and education are changing in favour of China, in terms of basic
R&D parameters such as research expenditure, scientists trained,
papers published and patents awarded (MacDonald et al., 2008).
The second category in the literature on Chinas innovation capabilities concerns one or several individual factors in the national
innovation system or one type of innovation in China, for example,
university-industry linkages, state-rm coordination, high-end talents,
disruptive innovation in China and cost innovations (e.g., Zeng and
Williamson, 2007; Feng, 2009; Simon and Cao, 2009; Tan, 2011). The
third stream of literature relates to industry case studies, most of which
focus on the high-technology industries, the information and communication technology (ICT) sector and green technologies (e.g., Lu, 2000;
Jakobson, 2007; Wang, 2012; and Liu et al., 2012).
All these studies have provided useful insights about the development
of innovation and technological capabilities in China. However, they
are based on studies of a particular industry, a particular type of
innovation, or one specic driver of innovation. What is Chinas
national strategy and path to innovation? Comprehensive and systematic analysis of Chinas overall innovation strategy, driver and outcome
is rare with very few exceptions (e.g., Varum et al., 2007; OECD, 2008).
Varum et al. (2007) present a comprehensive description of the transformation of innovation policies and the reform of science and technology systems in China from 1978 to 2004. OECD (2008) provides a
comprehensive and systematic review of Chinas national innovation
system. Features and performances of each of the major players, that is,
government, industry and universities, and the role of policy and governance are examined. Both of these studies set up their analysis under the
national innovation system framework. They provide a valuable description of the relevant policies and the status and performance of the important agencies in Chinas national innovation system. However, how China
achieved its current success and how china can achieve its new objective to
transform itself into an innovation-driven economy is still underresearched. Our understanding is limited with regard to the evolution of
Chinas path to innovation, in particular the evolution of strategies,
Introduction
processes and drivers of innovation at different stages of development and

their impact on Chinas innovation capabilities and technological
upgrading.
The objective and structure of the book

The objective of this book is to provide a systematic, comprehensive and
rigorous study of Chinas drive towards innovation in the past and for
the future. It draws on my research of more than a decade to understand, analyse and evaluate this process. The research employs the
rigorous analysis and empirical methodology of modern economics as
well as in-depth case studies of representative industries and leading
Chinese companies. Much of the evidence is based on either survey data
or longitudinal data at rm-, industry-, regional- or country-level. But a
systematic approach is adopted: economic and management theory,
development and evolutionary theory, institutional analysis and political economy are used to explain the motivation, sources, obstacles,
policy measures, rm responses and consequences of Chinas drive
towards being an innovative nation, and the roles played by the state,
the market, the private sector and the non-market, non-state institutions
such as universities and public research institutions.
In addition to the analysis of Chinas experience in the past three
decades, the book also investigates some of Chinas most recent efforts
in innovation, for example, internationalisation of Chinese MNEs and
outward direct investment for technology acquisition and upgrading,
international innovation collaboration, reforms of incentive structure at
multiple levels, and the development of green technologies. Moreover,
the research places China in a global context, and an international
comparative perspective is taken comparing China with other emerging
economies such as India and more advanced countries such as the
UK. The book also critically reviews Chinas experience, provides an
in-depth discussion of the likely way forward, and what other countries
can learn from Chinas experience.
In light of the economic and management theories on the sources of
innovation, taking on board the innovation systems framework and the
capabilities approach, the book is organised into three parts focusing on
the drivers of innovation at different stages of development, in addition
to the Introduction and Conclusions chapters. Part I examines the role
of international knowledge transfer and technological takeoff in China
10
Introduction
at the early stage of the reforms. Part II analyses the development of

indigenous innovation capability in the catch-up stage of industrial
development in China. Part III focuses on Chinas current efforts to
leapfrog the country into the role of global innovation leader and
assesses the role of incentive structure, institutional arrangement and
unconventional knowledge sourcing and co-creation measures in the
process. Before embarking on these analyses, an overview of Chinas
innovation efforts and performance in the past three decades since the
reforms is presented in Chapter 2.
Part I on the role of international knowledge transfer and technological
takeoff in China includes four chapters. Chapter 3 investigates the impact
of foreign direct investment on the development of regional innovation
capabilities using a panel data set of Chinese regions. It nds that FDI has
a signicant positive impact on the overall regional innovation capacity.
FDI intensity is also positively associated with innovation efciency in the
host region. The strength of this positive effect depends, however, on the
availability of the absorptive capacity and the presence of innovationcomplementary assets in the host region. This increased regional innovation and technological capability has contributed further to regional
economic growth in Chinas coastal regions but not in the inland regions.
It concludes that the type and quality of FDI inows and the strength of
local absorptive capacity and complementary assets in the host regions
are crucial for FDI to serve as a driver of knowledge-based development.
Policy implications are discussed.
Chapter 4 examines the impact of processing trade-oriented FDI on
the export competitiveness of indigenous rms using disaggregated
rm-level production data and product-level trade data from China
covering 2000 to 2007. The estimation results show that processing
trade-FDI has generated signicant positive information spillover
effects on the export performance of indigenous rms. However, the
effect of technology spillovers on the development of international
competitiveness in indigenous rms is limited and in fact exerts a
signicant depressive effect on the propensity to export in these rms.
Indigenous innovation, economies of scale and productivity are found
to be the main drivers of export performance in indigenous rms in the
high-technology industries.
Chapter 5 explores the role of indigenous and foreign innovation
efforts in technological upgrading in developing countries, taking into
account sectoral specicities in technical change. Using a Chinese
Introduction
11
rm-level panel data set covering 2001 to 2005, the chapter decomposes productivity growth into technical change and efciency
improvement and examines the impact of indigenous and foreign
innovation efforts on these changes. Indigenous rms are found to
be the leading force on the technological frontier in the low- and
medium-technology industries, whereas foreign-invested rms enjoy
a clear lead in the high-technology sector. Collective indigenous R&D
activities at the industry level are found to be the major driver of
technology upgrading of indigenous rms that pushes out the technology frontier. While foreign investment appears to contribute to
static industry capabilities, R&D activities of foreign-invested rms
have exerted a signicant negative effect on the technical change of
local rms over the sample period.
Part II on the development of indigenous innovation capability in the
catch-up stage includes ve chapters. Chapter 6 attempts to review the
evolution of policies and practices of open innovation in China using
historical archives and case study approaches, covering policies and
practices at both the macro and micro levels. It nds that Chinese
rms have in practice employed a variety of open innovation models
since the reforms of science and technology systems in the mid-1980s.
Policies introduced by the Chinese government with respect to inbound
and outbound open innovation as well as policies encouraging open
innovation networks have encouraged Chinese rms to adopt various
open innovation modes and practices. With the increasing internationalisation of R&D and globalisation of production, open innovation is
diffusing rapidly in China. Challenges to adoption of open innovation
for latecomer rms and the implications for latecomer rms in building
indigenous innovation capability are also discussed.
Chapter 7 examines how Chinese rms use open innovation as a
response to the constraints and risks of innovation that they face.
A national rm-level survey of 1,400 rms in the manufacturing sector
is used as the basis of the analysis. It found that institutional-, nancialand knowledge/skills-related risks and constraints are all signicantly
associated with rms depth and breadth of openness in innovation.
The responses, however, vary across rms of different ownership types.
Foreign invested rms appear to be most responsive and take action
to widen and deepen their openness in innovation. Privately owned
rms have made signicant responses to market/institution-related and
nance/risk-related impediments but not to knowledge/skills-related
12
Introduction
ones. State-owned rms appear to be least responsive in the use of open

innovation. Firm size and industry-specic effects also appear to have
signicant moderating effects on rms responses to the various constraints. These ndings are supported by an in-depth study of the
Chinese semiconductor industry.
Chapter 8 attempts to investigate the role of universities in industrial
innovation in emerging economies using a rm-level survey database
from China. It also benchmarks the Chinese pattern against that of the
UK. It nds that domestic universities have played a signicant role in
the promotion of the diffusion of frontier technology and the creation of
new country- or rm-level innovation outcomes in China. In contrast to
the traditional view that collaboration with universities will lead to
greater novel innovation (an outcome that is supported by evidence
from the UK), the contribution of domestic universities to the creation
of groundbreaking innovations is limited in China. International innovation collaboration with foreign universities, especially those in the
NIEs and the emerging South, appears to be fruitful in enhancing the
creation of groundbreaking innovations in Chinese rms.
Chapter 9 provides an analysis of the relative signicance of various methods of acquiring tacit knowledge within the Chinese optical
bre and cable industry. The chapter contributes to the denition,
understanding and investigation of tacit knowledge using rm-level
data in a developing country context, helping complete a gap in the
existing broader literature on technological learning. The research
suggests that in industries where tacit knowledge is a more important
component of technological learning than codied knowledge,
internal R&D activities and domestic peers are important knowledge
sources. Additionally, universities are shown to be an important asset
in creating learning organisations, and they provide effective knowledge sources of both tacit and codied knowledge. However, imports
of equipment and licensing are a less effective learning channel in the
acquisition of tacit foreign technology.
In recent years, China and India have achieved tremendous technological progress and development in the solar photovoltaic (PV) industry.
Using case studies, Chapter 10 analyses and compares the technology
progress processes in the solar PV industry in China and India, and
it discusses the role national innovation systems played in sustaining
technology acquisition, adaptation and development. It illustrates that
both countries adopted a strategy of mixing and sequencing different
Introduction
13
technology transfer and indigenous innovation mechanisms. The experiences of both countries also suggest that a functional national environmental innovation system is important in sustaining and advancing
technology acquisition, adaptation and development. This chapter provides an alternative pathway for developing countries to follow in catching up with developed countries in the emergent green industries and
leapfrogging towards an internationally competitive green economy.
Part III on Chinas current efforts to leapfrog the country into a global
innovation leader includes four chapters. Chapter 11 analyses the patterns of reverse learning and sequential capabilities development in
Huawei and ZTE, Chinas two successful ICT multinational enterprises. The chapter analyses the internationalisation process of
Huawei and ZTE, their learning activities in the host countries, channels for reverse knowledge transfer from one subsidiary located in
developed countries to headquarters and other subsidiaries worldwide.
Findings from this chapter reveal three processes of reverse learning and
capabilities upgrading: learning from customers, collaborators and
other subsidiaries of the company group. Findings from this case
study have valuable implications for organisational learning in MNEs
from developing countries.
Chapter 12 examines the role of international innovation collaboration in the process of radical innovation in China. Radical innovations
represent major departures from existing practices and involve the
disruptive creation of new insights. Accordingly, the launch of radical
innovation requires an extension of both the depth and breadth of
knowledge. This chapter investigates the patterns of international and
domestic innovation collaboration in 819 Chinese rms from 2006 to
2008. It nds that collaborations with foreign partners have made a
signicant positive impact of the creation of novel innovation in
Chinese rms. The type of foreign partners that Chinese rms may
benet by collaborating with covers a wide range, including foreign
customers, suppliers, universities, private research institutions and rms
in the same industry. Collaboration with foreign customers generates
the largest benets in the creation of novel innovation. Collaboration
with foreign universities also proved to be fruitful for the generation of
novel innovations that are new to the world, which is consistent with the
ndings of Chapter 8.
Chapter 13 benchmarks the patent activities of countries against the
world frontier and explores the sources of the cross-country differences
14
Introduction
in innovation (proxied by patenting). A patent production frontier is rst

estimated for a panel of 21 OECD countries from the 1990 to 2002
period using stochastic frontier analysis. Patenting performance for each
country is decomposed into basic patenting capacity and patenting efciency. The gap between Europe and the world leaders in terms of basic
patenting capacity remains substantial with little sign of convergence
over the sample period. In terms of patenting efciency, Japan,
Germany and Italy have improved their relative positions in recent
years. Institutional factors are found to be signicantly associated with
the patenting efciency of an economy. Then, China and other emerging
economies are brought into this benchmarking exercise using crosscountry panel data covering 2005 to 2011. The gap in patenting performance between China and the world frontier results from Chinas relative
underperformance mainly in the efciency of innovation production.
Chapter 14 examines the policy choices for Chinas drive to transforming the country into an innovation-driven economy. Innovation
capabilities, incentives and institutional frameworks are examined. The
chapter argues that China should continue to increase its investment in
R&D and in education, and that there should also be an attempt to
strengthen the incentive system at the macro, meso and micro levels.
This strengthening may include the following reforms: release the
power of competition and guide resources towards innovative sectors;
adopt appropriate human resource management policies, such as
appraisal and remuneration systems; create effective policies for
research funding management; and evaluate the efciency of research
to encourage the creativity of researchers, managers and employees.
The chapter also discusses the space for industrial policy in the twentyrst century.
Chapter 15 summarises the main ndings of the book, discusses the
implications for other countries and identies issues for further research.
Based on the comprehensive analysis of Chinas path to innovation that is
presented in the book, a model of an open national innovation system
(ONIS) is developed and the stage-specic mixing and sequencing characteristics of the model and implications for other developing countries
are discussed. Instead of being the often presented state-led model of
innovation in China or a pure market-driven model of innovation, the
ONIS model in China is a multi-driver model led by the state, the private
sector and the MNEs, with each of them playing a leading role in different
segments of the economy and the innovation system.
Innovation in China since the reforms:

An overview
2.1 Introduction
The intensive investment and rapid development in science and technology (S&T) in China has attracted growing attention. The unprecedented increase in expenditure on R&D activities and human capital
for innovation together with industrial structure upgrading and
knowledge-intensive trade are strengthening Chinas transformation
from a labour-intensive economic growth model into a knowledgebased one.
As outlined in Chinas 12th Five-Year Plan, innovation is a key
element to promote a more balanced pattern of development. The
15-year science and technology development blueprint National
Innovation Strategy launched in 2006 calls for R&D spending to
rise from 1.2 per cent of GDP to more than 2.5 per cent, which is
slightly above the average for OECD nations, by 2020. Demonstrating
its commitment to boosting investment in science and technology to
build a high-performing innovation-oriented economy, the government allocated more than US$ 77.82 billion to invest in science and
technology in 2011 with nearly a 20 per cent annual growth rate since
2005. As a result, China has moved faster than most of its peers in the
developing world in establishing the foundations of a world-class
innovation system.
This chapter provides a detailed description of the state of Chinas
innovation and the rapid ongoing transformation of the country into an
innovation-driven economy. It rst presents an overall assessment of
Chinas progress in building innovation capabilities, based on aggregate
national statistics on innovation inputs and outputs. It then gives a
concise picture of the innovation behaviours across industries, ownership structures, and geographic regions.
15
16
Introduction
2.2 Innovation inputs: R&D and research personnel

Starting in the 1980s, China began to reform its science and technology
system through a series of policies such as Key Technologies R&D in
1982, High-Technology Research and Development in 1986 and the
Torch Plan in 1988. After 1990, the focus shifted to innovation. In early
2006, China convened a National Science & Technology Summit that
issued the outline of the National Medium- and Long-Term Science and
Technology Development Plan 20062020. A distinct feature of the
plan is the adoption of innovation as a new national strategy and goal to
propel China into the ranks of innovation-oriented countries by 2020.
Since then, innovation has been recognised as the engine to shift Chinas
long-standing export-led, low-cost manufacturing growth model to a
knowledge-based sustainable growth model. The traditional focus on
development relying on exports and heavy xed-asset investment must
give way to more knowledge-intensive, higher-quality growth driven by
innovation.
Financial and human resource inputs, in particular into R&D,
directly contribute to the development of science and technology. We
start with illustrating nancial inputs and the stock of research human
capital, followed by indicators of measuring innovation performance.
R&D investment and composition of R&D

As a key indicator of its innovative efforts, R&D1 expenditure in China
has steadily increased at an exceptional rate from RMB 34.9 billion in
1995 to RMB 10,298 billion in 2012 a growth factor of almost 30.
With close to 20 per cent annual real growth in R&D spending, China
has become the third largest global investor in R&D, just behind
the United States and Japan but ahead of individual economies of the
EU.2 Figure 2.1 shows the huge increase in investment in R&D and
R&D spending as a percentage of GDP from 1995 to 2012. The R&D
intensity of Chinas economy, measured as R&D expenditure as a
1
Research and development involves the creative work undertaken on a systematic

basis to increase the stock of knowledge (including knowledge of humanity,
culture and society) and the use of this knowledge to devise new applications.
This is the authors calculation based on data collected from the OECD Structural
Analysis (STAN) database published by the OECD and the World Development
Indicators published by the World Bank.
Innovation in China since the reforms
17
100 million RMB
Percentage
12000
2.50
1.98
10000
1.76
1.84
2.00
1.70
8000
6000
0.90
4000
0.95
1.07
1.13
1.23
1.32
1.39 1.40
1.47
1.00
0.76
0.57 0.57
1.50
0.64 0.65
0.50
2000
0.00
19
95
19
96
19
97
19
98
19
99
20
00
20
01
20
02
20
03
20
04
20
05
20
06
20
07
20
08
20
09
20
10
20
11
20
12
R&D expenditures
R&D spending as percentage of GDP
Figure 2.1 Chinas R&D expenditure, 19952012

Source: China Statistical Yearbook (19962013).
percentage of GDP, has also greatly increased since 1995. As a proportion of GDP, R&D was at a low of 0.57 per cent of GDP in 1995, rising
to 1.32 per cent in 2005 and 1.98 per cent in 2012.
Among OECD countries, the United States, Japan and Germany are
the main performers. Their ratios of R&D spending to GDP maintain
stable levels between 2.5 and 3.5 per cent, while the ratio has been
growing the fastest in South Korea, with average annual growth
rates around 10 per cent as depicted in Figure 2.2. Although it has
experienced a remarkable increase, the R&D intensity in China is still
low in comparison with OECD countries. The highest ratio of R&D to
GDP in China was 1.98 per cent in 2012, which is still far behind the
OECD average at 2.39 per cent. The National Guidelines for S&T
Development outlines an ambitious target for R&D investment: it
should reach 2 per cent of GDP by 2010 and 2.5 per cent or more by
2020. As the goal of reaching 2 per cent by 2010 was not met (1.75 per
cent in 2010 as shown in Figure 2.1), increasing the R&D intensity to
2.5 per cent by 2020 is a serious challenge taking into account the fast
growth rate of GDP. This indicates that R&D expenditure needs to
increase at least 25 per cent annually starting from 2012.
According to the classication given by the OECD, R&D covers
three activities: basic research, applied research and experimental
18
Introduction
R&D spending in GDP

5
4.5
Korea
4
3.5
3
2.5
2
Japan
Germany
US
OECD
France
China
UK
1.5
1
0.5
0
2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012
Figure 2.2 R&D expenditure in China and other economies, 19952012

development.3 The breakdown of R&D expenditure by type of activity reveals important structural features of the innovation system in
China. A notable feature of Chinas R&D is that the share of basic
research is considerably lower, whereas spending on experimental
development accounts for a relatively higher share, as shown in
Figure 2.3. Expenditure on experimental development has grown the
fastest, accounting for 84 per cent of total R&D in 2006 (72 per cent in
1998). In contrast to experimental development, the expenditure on
applied research has declined from 5.3 per cent in 1998 to 4.8 per cent
in 2012. Spending on basic research has remained at a stable level of
around 5 per cent. The composition of R&D activities implies that a
Basic research is experimental or theoretical work undertaken primarily to acquire

new knowledge of the underlying foundation of phenomena and observable facts,
without any particular application or use in view. Applied research is also original
investigation undertaken to acquire new knowledge. It is, however, directed
primarily towards a specic practical aim or objective; experimental development
is systematic work, drawing on existing knowledge gained from research and/or
practical experience, which is directed to producing new materials, products or
devices; to installing new processes, systems and services; or to improving
substantially those already produced or installed. More information can be
found at www.oecd-ilibrary.org/sites/factbook-2013-en/08/01/01/index.html?
itemId=/content/chapter/factbook-2013-60-en.
19
Percentage
100
90
80
70
60
50
40
30
20
10
0
1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012
Basic research
Applied research
Experimental development
Figure 2.3 Composition of R&D by types of activities: basic research, applied

research and experimental development
big proportion of the increase in R&D expenditure in the past decades

can be attributed to the investment in experimental development.
To sustain the rapid growth of R&D expenditure, strong nancial
supports from various channels are essential driving forces. R&D
expenditure can be classied according to the funding origins: government funding, enterprise funding, foreign funding and other funding.4
In the comparison between 2003 and 2011, Figure 2.4 shows that the
growth of R&D expenditure is mainly supported by enterprises. It
reects, to some extent, that enterprises in the business sector have
become an important player in the level of innovative sophistication
and the structural change of the innovation system in China. Enterprises
and government funding (noted as public institutions in the gure) are
the two largest nancial resources; together they have accounted for up
to 90 per cent of R&D spending in China since 2004, in particular
enterprises stood at about 74 per cent of total R&D funding in 2011. In
contrast, foreign-funded R&D in China has been very limited.
4
Other funding may include self-raised funding, in particular for independent

research institutions and the higher education sector, and leftover government
money from previous years/grants.
20
Introduction
Percentages
100
90
80
70
0.60
0.66
0.67
0.69
0.70
0.72
0.72
0.72
0.74
60
50
40
30
20
0.30
0.27
0.26
0.25
0.25
0.24
0.23
0.24
0.22
2003
2004
2005
2006
2007
2008
2009
2010
2011
10
0
Public institutions
Enterprises
Foreign
Other funding
Figure 2.4 R&D composition by funding sources, 20032011

R&D personnel
Human resources are equally important as R&D inputs and are a
crucial building block in forming an innovation pool. R&D personnel
include all persons employed directly on research and development
activities, as well as those providing direct services such as R&D managers, administrators and clerical staffs (OECD, 2002). The success of
transforming to a knowledge-based economy requires a large number
of R&D personnel to carry out the corresponding large-scale increase in
R&D activities. Since 2006, China has had the second highest number
of researchers in the world, just behind the United States. As shown
in Figure 2.5a, the number of R&D personnel in China, calculated in
full-time equivalents (FTE), has increased steadily since 1995. Breaking
down R&D personnel by type of activity shows that the notable
increase in the number of researchers is a response to the tremendous
growth of experimental development activities. There is also a moderate
rise in the number of researchers for basic research and applied research
during the period shown.
The number of R&D personnel is further broken down by R&D
executive entities. Enterprises have the largest share of R&D personnel,
21
Researchers per 10,000 people

350
300
250
200
150
100
50
19
95
19
96
19
97
19
98
19
99
20
00
20
01
20
02
20
03
20
04
20
05
20
06
20
07
20
08
20
09
20
10
20
11
20
12
Basic research
Applied research
Experimental development
Figure 2.5a Total number of R&D personnel

University
6%
Other
8%
Enterprise
46%
Research
institute
40%
Figure 2.5b R&D personnel by executive entity, 2009

while research institutes take almost an equally important share,

46 per cent and 40 per cent respectively in 2009 (Figure 2.5b). The
distribution of R&D personnel across different performing entities is
in line with the type of R&D activities carried out in China. For
22
Introduction
R&D personnel per

1000 employees
16
14
12
Korea
France
Germany
Japan
UK
10
8
6
4
China
2
0
2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011
Figure 2.6 Comparison of R&D personnel per thousand employees in China

and OECD economies, 20002011
Source: OECD science and technology database available at http://stats.
oecd.org.
example, universities absorb only 6 per cent of R&D researchers,

similar to the amount of money spent in basic research activities, and
the majority of R&D personnel are working for enterprises in which
the main R&D activity is experimental development.
The large R&D workforce is one of the most important strengths for
China on its path to become an innovative economy. The reason that
China is able to compete with other advanced countries in absolute
numbers is its large population. Yet, accounting for the quality of
innovation by way of an international comparison is still a challenge.
Although China has more personnel employed in R&D in absolute
numbers than most of the OECD countries, the ratio of R&D personnel
weighted by per thousand employees is at a much lower level and the
growth is modest (Figure 2.6).
Apart from relying on in-house R&D activities, China has tried to
acquire advanced technologies via various sources. Importing technologies from foreign countries, complementing indigenous efforts,
has been commonly acknowledged as a substantial driver to technological upgrading in China. One of the most straightforward ways to
obtain foreign technology is through global market transaction such
as purchasing licenses, patents and copyrights. Figure 2.7a exhibits
23
100 million RMB

500.0
466.9
450.0
400.0
350.0
422.2
397.4
449.0
393.9
304.9
300.0
250.0
200.0
150.0
100.0
50.0
0.0
2000
2004
2008
2009
2011
2012
Figure 2.7a Expenditure for acquisition of foreign technology in China,

20002012
Source: China Statistical Yearbook (20012013). The 2013 Statistical Yearbook
does not cover the expenditure for acquisition of technology import in 2002,
2003, 2005, 2006, 2007 or 2010.
the trend of expenditure for purchasing foreign technology in China

from 2000 to 2012. Since joining the World Trade Organisation
(WTO) in 2001, the scale of foreign technology acquisition in
China had a 30 per cent annual increase and reached a peak in
2008 with RMB 46.69 billion. After the nancial crisis in 2008,
the annual expenditure for foreign technology import has gone
through some uctuations but the volume has always been above
39.39 billion RMB.
Among different types of foreign technology imports in 2012, the
majority of spending was for technology licensing (37 per cent) and
foreign technology consultation (32 per cent) as the left pie graph in
Figure 2.7b shows. In 2012, China spent about 15 per cent of its
budget for technology import for patent technology licensing and
transfer, while the remaining 16 per cent went for importing computer
software (6 per cent), trade mark license (1 per cent) and other technology imports (9 per cent). Noticeably, a regional disparity has also
been shown for the technology imports. The right pie graph in
Figure 2.7b shows the total expenditure for technology import
across different regions of China. The eastern region as the most
economically developed area absorbed nearly 68 per cent of imported
24
Introduction
Types of imports
Trade mark
license
1%
Import of
computer
software
6%
Technology
consultation
and service
32%
Other
9%
Regions of technology imports
Patent technology
and license
transfer
15%
Western
region
23%
Northeast
region
4%
Eastern
region
68%
Middle
region
5%
Technology
license and
transfer
37%
Figure 2.7b Foreign technology imports in China, percentages, 2012

Source: China Statistical Yearbook on Science and Technology (2013).
technologies. The western region accounted for 23 per cent, while the
middle and northeast region in total received 9 per cent of the
imported technology in 2012.
2.3 Innovation outputs: patents and published journal articles

Patents
Beyond the increasing tendency for the use of innovation inputs, equal
attention is being paid to the performance of innovation, which is of
interest in assessing the quality of the innovation. One of the commonly used indicators at the aggregated level to measure innovation is
the total number of patent applications. Domestic patent applications
to the Chinese State Intellectual Property Ofce (SIPO) increased
nearly sixfold between 1995 and 2005, while the number of granted
patents increased slightly less, about vefold during the same period of
time as illustrated in Figure 2.8. Foreign applications, though fewer in
absolute numbers, rose at an even higher rate by more than sixfold and
tenfold for applications and patents granted, respectively. Such a
surge in patent applications can be explained by the persistent efforts
devoted to R&D, as well as the increasing awareness of the value of
IPR protection. The growth of patent applications increased even
more after the launch of National Innovation Strategy in 2006.
25
Number of patent applications, 1000 units

Outside of China
2012
2011
2010
2009
2008
2007
2006
2005
2004
2003
2002
2001
2000
1999
1998
1997
1996
1995
138
129
113
99
111
107
103
93
75
57
47
38
30
24
26
24
21
14
500
In China
1912.151
1504.67
1109.428
877.611
717.144
586.498
470.342
383.157
278.943
251.238
205.544
165.773
140.339
109.958
96.233
90.071
82.207
68.88
0
500
1000
1500
2000
Number of granted patents, 1000 units

Outside of China
2012
2011
2010
2009
2008
2007
2006
2005
2004
2003
2002
2001
2000
1999
1998
1997
1996
1995
92
77
74
80
60
50
44
42
39
33
20
15
10
8
7
5
4
4
250
In China
1163.226
883.861
740.62
501.786
352.406
301.632
223.86
171.619
151.328
149.588
112.103
99.278
95.236
92.101
61.378
46.389
39.725
41.248
250
750
1250
Figure 2.8 Number of patent applications vs number of granted patents,

19952012
Patent applications to the SIPO reached 1.912 million during all of

2012, whereas the number of granted patents reached 1,163,000,
nearly ve and seven times, respectively, higher compared to the levels
in 2005.
26
1000
900
800
700
600
500
400
300
200
100
0
Introduction
China
India
Russian Federation
South Africa
1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011
Brazil
China
India
Russian Federation
South Africa
Figure 2.9 Triadic patent families across BRICS, 19992011

Source: OECD patent database available at http://stats.oecd.org.
As in the United States, the European Union and Japan, the number of
Chinese triadic patents,5 which are acknowledged as a better measure of
the value of innovation, has been increasing, as shown in Figure 2.9. The
number of triadic patents granted to China applicants in 2011 was 909,
which was the highest number among the emerging BRIC countries
(Brazil, Russia, India and China) and ve times that of India.6 Brazil,
South Africa and the Russian Federation signicantly lagged behind.
Nevertheless, the total number of triadic patents led by Chinese residents is still less than one-tenth of that of the United States and Japan
(Figure 2.10). The United States and Japan are still the leading performers
in triadic patent families, followed by Germany, France, South Korea and
the UK. Application numbers from China accounted for one-fth of the
number from Germany, and less than one-tenth of those from Japan and
the United States. This level is still low for China and the gap from OECD
countries remains extremely large.
Journal articles
The higher education sector not only contributes to innovation development through its direct participation in various innovation activities but
5
Triadic patents are a series of corresponding patents led at the European Patent
Ofce, the United States Patent and Trademark Ofce and the Japan Patent Ofce,
for the same invention, by the same applicant or inventor.
The data come from www.wipo.int/ipstats/en/statistics/patents.
27
Number of triadic
patent families
16000
14000
Japan
12000
US
10000
8000
6000
Germany
4000
France
Korea
UK
China
Brazil
2000
19
99
20
00
20
01
20
02
20
03
20
04
20
05
20
06
20
07
20
08
20
09
20
10
20
11
France
Germany
Japan
Korea
UK
US
Brazil
China
Figure 2.10 Number of triadic patent families in China and other OECD
economies, 19992011
Source: OECD science and technology database available at http://stats.oecd.org
also through its education mission, which ensures a future supply of

human resources for innovation. One of the indicators to measure
innovation performance in the higher education sector is the volume of
articles published worldwide. Beyond the substantial increase of domestic publications, the articles published in international peer-reviewed
journals by Chinese authors or Chinese co-authors collaborating with
foreign researchers have also experienced a noticeable increase. As
shown in Figure 2.11, the number of articles from Chinese universities
published in international journals has been increasing gradually since
2005. In 2010, the number reached 320,345, double the number in 2005.
Yet bare numbers of publications can hardly indicate the quality by
themselves, and there is continuous debate about the productivity of
Chinese innovation personnel. The number of citations assesses the
productivity and inuence of scientic literature and also serves as a
measurement of the visibility of scientic research. The study conducted
by Hu (2011) shows that Chinana citation ranking shifted from
19th place to 10th during the 19922008 period. The 12th Five-year
Science and Technology Development Plan launched in 2011 aimed to
further raise Chinas citation ranking from 8th in 2010 to 5th by the end
of 2015 (World Bank, 2013). Figure 2.12 calculates the number of
accumulated publications and frequency of citations between 2000
28
Introduction
1000 articles
350
320.354
300
240.186
250
253.982
196.629
200
152.825
171.748
150
100
50
0
2005
2006
2007
2008
2009
2010
Figure 2.11 Output of R&D: index articles published in international journals

10,000 Citations
Citations
350
300
18
US
16
Netherlands
UK
GermanyCanada
250
14
France
12
Australia Italy
200
Spain Japan
10
8
150
Korea
China
100
India
Russia
er
U
K
m
an
C
an y
ad
Fr a
an
c
Au e
st
ra
lia
Ita
ly
Sp
ai
n
Ja
pa
n
Ko
re
a
C
hi
na
In
di
a
R
us
si
a
N
et
he US
rla
nd
s
50
Number of publications
Citations per paper
Figure 2.12 Citations of scientic papers published, 20002010

and 2010. China ranks fourth in the output of publications during the
period under review, after the United States, Japan and Germany. Not
surprisingly, and consistent with the ndings from other studies (Simon
and Cao 2009; Royal Society 2011), the citation frequency ranking of
0.73
0.72
0.69
Country
South Korea
United States
Japan
EU27
Canada
Australia
China
Russian Federation
India
Brazil
South Africa
29
0.60
0.50
0.33
0.26
0.21
0.16
0.15
0.08
0
0.1
0.2
0.3
0.4
0.5
0.6
Innovation Scoreboard
0.7
0.8
Figure 2.13 Innovation performance: international comparison, 2012

Source: EC Innovation Scoreboard (2013).
publications from China is in general lower compared to OECD economies. Given the immense investment of R&D and the large number of
R&D personnel, the lower level of productive research clearly shows
that the quality of innovation in China is severely limited. Further
efforts should be exerted to improve quality rather than quantity.
2.4 Overall innovation performance: an international

comparison
Consistent ndings are obtained when evaluating Chinas national innovation capabilities using another comprehensive index, the Innovation
Scoreboard created by the European Commission.7 Chinas score is only
one-third that of the United States, South Korea and Japan, although it
the best performer among BRIC countries (Figure 2.13). Despite these
remarkable achievements, China faces signicant challenges in enhancing
its innovation capability.
2.5 Industry distribution

The extent to which enterprises generate and exploit knowledge
varies across industries. To gauge the nature and underlying role of
7
This index is based on 25 indicators covering a wide range of innovation enablers,

rm activities and outputs.
30
Introduction
100 million RMB

1200
R&D intensity
2.5
1000
800
1.5
600
1
400
0.5
200
0
A A
C Ch uto gro
he em m -f
C mi ic otiv ood
El ult ca al e p
ec ur l m fib m ro
tri e a re an ce
ca an te m u s
l m d ria a fac sin
ac ent ls nuf tur g
C hin erta and act ing
lo e
u
Fa thin ry inm pro ring
br g, an en du
d
t
i
ca ap e g cts
G
t p q o
en
El ed m are uip ods
er
al F ec e l i me
eq err tro tal nd nt
ui ou nic pr us
pm s e o try
e m q du
In F nt eta uip cts
s
N tru ood ma l sm me
on m
n
n
-fe en ma ufa elti t
rro t m nu ctu ng
f
r
us a ac in
N
m nuf tur g
on
et ac in
-m Le
al tu g
a
e
sm rin
Ph Pa tal the
g
l
r
ar pe ic a Fu elti
m r a mi nd rn ng
Sp
ac n ne f it
o
u
eu d ra o re
ec
ia
t p l tw
l e Pe ica ape pro ea
R qu tro l ma r p du r
ub ip le n ro ct
be me um uf du s
W
r a nt p ac cts
in
nd m ro tur
e,
be
pl anu ces ing
as fa si
v
Tr era
tic ct ng
an g
pr urin
sp e a
o
or nd P du g
W
ta r r cts
oo
tio ef in
n ine tin
d
eq d g
pr
oc
ui te
pm a
es
si
Te en
ng
t
T
an o xtile
d ba s
pr cc
od o
uc
ts
R&D expenditures
R&D intensity (R&D / Industry output)
Figure 2.14 R&D expenditures and intensity of enterprises in the above-scale

manufacturing industries, 2012
technology in economic growth, it is essential to assess the innovation

activities and performance on the basis of industrial specicities.
R&D expenditure by industry provides valuable information on the
technological strengths of industries. In particular, the association of
R&D with industries permits the link between technology and different dimensions of industries economic performance. Figure 2.14
shows that the technology-intensive industries such as electronic
equipment and electrical machinery attract the most R&D investment, while, among low-tech industries, the level of R&D spending
remains substantially low.8 The decomposition of the Chinese R&D
portfolio by industry shows the emergence of Chinese high-tech
sectors, which reported a higher share of technological efforts.
8
On the basis of industry classication by technology intensity introduced by the

OECD, manufacturing industries are classied in four exclusive categories:
high technology, medium-high technology, medium-low technology and low
technology. The classication is based on indicators such as R&D expenditures
divided by value added and R&D expenditures divided by production,
reecting to different degrees technology producer and technology adopter
aspects.
Health and social

welfare
5%
Water, environment
industry
0%
Culture, sports and

entertainment
0%
31
Agriculture
0%
Mining
4%
Education
17%
Manufacturing
50%
Scientific research and

technical services
17%
Leasing and
business
services
1%
Financial sector
0%
ICT
industry
2%
Building
industry
3%
Electricity and water

industry
1%
Transportation
and postal
services
0%
Figure 2.15 Number of R&D personnel across sectors, 2009

Figure 2.15 breaks down the total number of the R&D workforce
across sectors. The pie graph highlights the dominance of the manufacturing sector and how it absorbs most of the R&D personnel. In 2009,
the education and technical service sectors shared an equal amount of
R&D personnel in China (17 per cent), followed by health and social
welfare (5 per cent) and the mining sector (4 per cent).
Turning to the innovation output across industries, a signicant part
of patent applications is attributed to high-tech industries. Calculated
at the aggregate level, the patent application share of high-tech industries accounted for 75 per cent, whereas that of low-tech industries
equalled 25 per cent in 2010.9 The sum of the electronic equipment and
The Technology Intensity Denition is in accordance with OECD classication.

For more details, please see www.oecd.org/sti/ind/48350231.pdf
32
Introduction
46209
Electronic equipment
Electrical machinery and equipment
Transportation equipment
General equipment manufacturing
Special equipment manufacturing
Textiles
Ferrous metal smelting
Pharmaceutical manufacturing
Chemical materials and products
Fabricated metal products
Instrument manufacturing
Non-ferrous metal smelting
Electricity, heat production and supply
Plastic products
Culture and entertainment goods
Wine, beverage and refined tea
Agro-food processing
Clothing, apparel industry
Food manufacturing
Other manufacturing
Oil and gas exploration
Chemical fiber manufacturing
Furniture
Rubber products
Coal mining and washing
Paper and paper products
Leather and footwear
Tobacco
Wood processing and products
Petroleum processing
Printing
Non-ferrous metal
Ferrous metal mining
Nonmetal mining
Water production and supply
Gas production and supply
28978
23700
13922
13467
Industry
6388
5813
5767
5743
5355
5131
3335
2891
2868
1995
1961
1908
1907
1875
1825
1654
1609
1581
1443
1314
1247
1003
766
594
558
521
135
90
59
51
22
0
5000 10000 15000 20000 25000 30000 35000 40000 45000 50000
Patent applications: total number in 2010
Figure 2.16 Innovation output across industries: number of patent applications,

2010
electrical machinery industries was responsible for 38.9 per cent of

total patent applications to Chinas State Intellectual Property Ofce
(Figure 2.16), followed by medium-high-technology industries such as
transportation equipment and equipment manufacturing. Mediumlow and low-tech industries share only fractions in the total number
of patent applications in China.
A similar pattern was observed for new product sales and new
product exports across industries, as shown in Figure 2.17. In response
to the large R&D expenditures in high-tech industries, both sales
and exports due to new products are remarkably higher in 2012
because of new products, notably for industries such as electronic
equipment, electrical machinery and chemical material. Medium-low
33
Electronic equipment
Electrical machinery and equipment
Chemical materials and products
Ferrous metal smelting
General equipment manufacturing
Special equipment manufacturing
Non-ferrous metal smelting
Textiles
Pharmaceutical manufacturing
Fabricated metal products
Agro-food processing
Non-metallic mineral products
Petroleum processing
Chemical fiber manufacturing
Tobacco
Paper and paper products
Coal mining and washing
Food manufacturing
Printing
Non-ferrous metal
Wood processing and products
Furniture
Electricity, heat production and supply
Gas production and supply
Nonmetal mining
Ferrous metal mining
Oil and gas exploration
Water production and supply
11000
New product export value
1000
9000
19000
New product sales
Figure 2.17 Innovation outputs across industries: new product sales and new
product export values, 2012
and low-tech industries are in general less productive in sales and

exports from new products.
The high-tech product export and import volumes are presented in
Figure 2.18. From 1995 to 2012, exports of high-technology products
from China increased from RMB 101 billion to RMB 6,012 billion.
This rapid increase in high-tech product trade reects the increasing
importance of this trade for the Chinese economy. China has gradually become a major exporter of high-technology nal products in the
world. Interestingly, between 1995 and 2012, imports and exports of
high-tech goods have increased almost at the same pace. With reference to the volumes, the exports of products from high-tech industries
were not substantially higher than those of imports before 2005. The
gap between exports and imports in high-tech industries gradually
widened after 2005. The highest surplus in high-technology product
34
Introduction
Billion USD
7000
6011.7
6000
5488
5068.6
4923.79
5000
4156.06
3769.3
4000
3478
2814.5
3000
2182.48
3418.2
3098.53
High tech product export
12
11
20
10
20
09
20
08
20
07
20
06
20
05
20
04
20
03
20
02
20
01
20
00
20
99
20
98
19
97
19
19
19
19
96
1977.08
1103
1613
464.52 679
202.51
370.43
1193
247.04
828
100.91126.63 163.1
525.07 641.07
224.69 238.93 292.01 375.98
218.27
95
2870
2473
1654
2000
1000
4632
4126.55
High tech product import
Figure 2.18 Export and import volume of high-tech products, 19952012

trade occurred in 2012. The total of exports is US$943 billion

(18.6 per cent) higher than total imports in the high-technology sector.
This, to a certain extent, reects that Chinas export-oriented manufacturing industry is still engaged in processing and assembly operations. Export competitiveness is predominantly based on low factor
costs (Fu, 2003).
The extent of innovation outputs in China that accrues to foreign
invested rms should not be neglected. As shown in Figure 2.19,
foreign-owned rms produce almost a quarter of new product sales
(24 per cent), and they had the highest share of new product exports in
2012, accounting for 44 per cent of total new product exports value.
Domestic rms occupy the leading role in new product sales and
account for nearly 66 per cent of total sales due to new products in
2012. The new-product exports from domestic Chinese rms take up
to 41 per cent, which is slightly less than for foreign owned rms.
Hong Kong, Macau and Taiwan invested rms account for one-tenth
of total new product sales, and their share of total new-product export
value is about 15 per cent.
New product sales
35
New product exports value

Foreign
invested firms
44%
Foreign
invested firms
24%
Domestic
firms
41%
Domestic
firms
66%
HK, Macau and

Taiwan invested firms
10%
HK, Macau and

Taiwan invested firms
15%
Figure 2.19 Innovation outputs by ownership of rms: new product sales and
new product export values, 2012
2.6 Innovativeness across ownership structures

The paradigm of the National Innovation System suggests that heterogeneous national innovation output may be explained by institutional
divergences (Lundvall, 1992; Freeman, 1995). Chinese rms have distinctive characteristics of corporate governance, including concentrated
ownership structures: state ownership (SOEs), other domestic ownership and the emergence of foreign investors. Differences in ownership
structures explain R&D activity and innovation performance across
countries (Hoskisson et al., 2002). As shown in Figure 2.20, although
the total amount of R&D investment increased signicantly over the
2006 to 2012 period, the ownership composition of R&D investment
has remained relative stable. Figure 2.21 compares the ownership
structures of R&D spending in 2006 and 2012. Domestic rms (excluding SOEs) hold the dominant role, while the share in the total R&D
expenditure by SOEs and foreign owned enterprises declined 2 per cent
and 3 per cent, respectively. The proportions shared by Hong Kong,
Macau and Taiwan (HMT) owned enterprises did not differ from 2006
to 2012.
A similar pattern is displayed when we look at the ownership structure of R&D personnel in China (Figure 2.22). Domestic enterprises
absorbed the majority of the R&D workforce, 66 per cent, for both
2006 and 2012. There has been a slight increase in R&D personnel
36
Introduction
RMB billion
800
700
600
500
400
300
200
100
0
2006
2007
2008
2009
2010
2011
2012
Foreign ownership
Other domestic ownership
Hong Kong, Macau and Taiwan ownership
State ownership
Figure 2.20 R&D expenditures across different ownership structures in industrial enterprises above designated size, 20062012
R&D spending, 2006

Foreign
ownership
18%
Hong Kong,
Macau and
Taiwan
ownership
9%
State
ownership
10%
Other
domestic
ownership
63%
R&D spending, 2012

Foreign
ownership
15%
Hong Kong,
Macau and
Taiwan
ownership
9%
State
ownership
8%
Other
domestic
ownership
68%
Figure 2.21 R&D expenditures across different ownership structures in industrial

enterprises above designated size, 2006 and 2012
Source: China Statistical Yearbook (2007 and 2013).
37
1000 people, per year

2500
2000
1500
1000
500
0
2006
2007
2008
2009
2011
2012
Foreign ownership
State ownership
Figure 2.22 R&D personnel across different ownership structures in industrial

enterprises above designated size, 20062012
Data for 2010 were not published in the Yearbook.
shares for both foreign owned (from 7 per cent to 12 per cent) and
HMT owned enterprises (from 13 per cent to 15 per cent), whereas the
share for SOEs has dropped to half, from 14 per cent in 2006 to only
7 per cent in 2012, although the number of R&D personnel in absolute
terms in this sector increased.
With respect to innovation outputs, ownership composition of patent
applications is well matched with the corresponding structures of R&D
expenditure and R&D personnel. With relatively higher scales of input
of R&D and human capital, domestic rms take up the greatest part of
patent applications in China, followed by enterprises with foreign and
HMT ownership (Figure 2.23). In terms of the absolute number, patent
applications led by domestic rms in 2012 reached 348,421, nearly
seven times higher compared to 2006. In parallel with the small shares
in R&D and human capital investment, SOEs take the lowest proportion with respect to the number of patent applications in China.
Interesting ndings are uncovered if innovation output is measured
by the sales and exports due to new products. Accompanied by the
38
Introduction
Pieces of patent in percentages

100
9183
13460
23403
38264
52711
59300
9551
9927
19108
29741
41595
51434
45887
66773
121635
184717
271023
348421
4388
5745
9427
13086
20746
30790
2006
2007
2008
2009
2011
2012
90
80
70
60
50
40
30
20
10
0
Foreign ownership
State ownership
Figure 2.23 Patent applications across different ownership structures in industrial enterprises above designated size, 20092012
Data for 2010 were not published in the Yearbook.
intensive investment in innovation inputs, the value of new product

sales generated by domestic rms has experienced remarkable growth
and surged to RMB 6,532 billion in 2012 as shown in Figure 2.24.
Although the absolute value of new product sales generated by SOEs
increased, the corresponding share has declined gradually from 10 per
cent in 2009 to 7 per cent in 2012. The only observed increasing trend in
the share of new product sales during the period under review is for
domestic enterprises, from 54 per cent to 59 per cent. Such moderate
change may be considered evidence of improvement in innovation
efciency among domestic enterprises in China. Having reached RMB
957 billion in 2012, foreign owned enterprises account for 44 per cent
of total new product exports. Corresponding values reported for
domestic enterprise is RMB 869 billion, nearly double the value in
2009. SOEs have had a less important performance in exporting innovative products with a share of 2 per cent in 2012.
2012
New product exports
2011
New product exports
2010
New product exports
2009
New product exports
39
New product sales
New product sales
New product sales
New product sales

0
2000
4000
6000
8000
10000
12000
RMB billion
State ownership
Foreign ownership
Figure 2.24 New product sales and new product exports across different
ownership structures in industrial enterprises above designated size,
20092012
Source: China Statistical Yearbook on Science and Technology (20102013)
2.7 Regional disparity

Given Chinas vast size and diversity, its municipalities and provinces
have distinct characteristics and substantial populations. Statistics of
national averages under these circumstances can be particularly misleading. One of the notable features of the Chinese innovation system is
the large regional disparity of R&D, as illustrated by Figure 2.25. The
investment of R&D concentrated in several provinces and cities, such as
Jiangsu, Beijing, Guangdong and Shandong, amounted to 45 per cent of
total R&D expenditure in 2012.
In 2000, the highest R&D expenditures were found in Beijing, followed by Gansu, Shanghai and Jiangxi as shown in Figure 2.26.
Southern and western China were evidently less innovative and R&D
expenditures were scarce. Beneting from the reform and opening policy
and export-oriented industrialisation, the East and coastal regions were
catching up at a remarkable pace, especially in Guangdong and
Zhejiang provinces. With R&D expenditure below RMB 80 thousand
in 2000, Guangdong has successfully become one of the most active
provinces in R&D investment, which reached more than RMB 123
40
Introduction
100 million RMB
Percentage
7
1200
1000
800
600
400
200
Be
i
Ti jing
an
H jin
In
ne S eb
r M h ei
on anx
Li go i
ao lia
ni
H
n
ei
lo Jil g
ng in
Sh jia
an ng
Ji gha
a
Zh ng i
ej su
ia
An ng
Fu hui
ji
Sh Jian an
an gx
do i
H ng
en
H an
u
G H be
ua u i
ng na
n
G don
ua g
ng
C Ha xi
ho in
ng an
Si qin
c g
G hua
ui n
z
Yu ho
nn u
an
Sh Tib
aa et
G nxi
a
Q ns
in u
N gha
in i
Xi gx
nj ia
ia
ng
1400
R&D expenditure
R&D/GDP ratio
Figure 2.25 R&D expenditures and R&D/GDP ratios across China, 2012
R&D expenditure, 2012
R&D expenditure, 2000
Figure 2.26 R&D expenditures across China, 2000 and 2012

million in 2012. Similar exceptional increases have also been observed

for Zhejiang, Jiangsu and Shandong provinces, while inland China was
also catching up but at a moderate speed. It is clearly shown by
Figure 2.26 that profound disparities of R&D investments are still
prevalent across China, where coastal areas absorbed a large proportion
of R&D spending and inland China, especially the western part,
remains less innovative.
R&D expenditure in GDP, 2012
41
R&D expenditure in GDP, 2000
Figure 2.27 Ratio of R&D expenditures in GDP across China, 2000 and 2012
Figure 2.27 presents the ratios of R&D spending in regional GDP

across provinces in China. Except for Beijing and Gansu, innovation
intensities rose from 2000 to 2012 across China. The highest R&D
intensity was found in Beijing, at 6.28 per cent of GDP in 2000 and
5.95 per cent in 2012. The ratio for Beijing in 2012 was nearly three
times that of the national average and the lowest for that year was in
Tibet, at only 0.25 per cent of GDP, or just above one-eighth of the
national average.
Being aware of the divergence among regions and the risk that the gap
will further enlarge, the central government launched the Go West
strategy in 2000 to accelerate convergence. R&D expenditure and
intensity in the middle and the western regions have been increasing,
but the levels of knowledge stocks still remain at a lower level. It is
worth noting that the large R&D spending and R&D intensity in the
Gansu, Shaaxi and Shaanxi regions may be explained by the inherited
R&D facilities, including military research bases, government research
institutions and universities, which were located there for strategic
reason during the cold war.
In addition to R&D expenditure, the large regional disparities are
also observed in relation to innovation output. Figure 2.28 displays the
mean value of patent applications across China in 2000 and 2012. The
mean value of patent application, indicated by the horizontal bar, has
risen signicantly from 2,757 pieces in 2000 to 60,824 pieces in 2012.
The statistics on the map clearly show that the pattern of disparity is
similar to the innovation input indicators: the municipalities and
42
Introduction
Patent applications, 2012
Patent applications, 2000
Figure 2.28 Mean patent application numbers across China, 2000 and 2012
Note: Grey area indicates the ratio of patent applications weighted by local
population (10,000 people) in each regions; the bar denotes the mean of patent
application numbers at the national level.
Funding composition, 2012
Government
Enterprise
Foreign
Other
Funding composition, 2007
Government
Enterprise
Other
Figure 2.29 R&D funding sources across China, 2009

provinces with the highest R&D intensity are in the lead, with other
parts of the country following at a staggered pace.
Accordingly, adopting effective innovation policies as a tool together
with other government incentives to narrow the existing gap is imperative. The breaking down of R&D funding source across regions is
shown in Figure 2.29. Whereas enterprises funding plays a dominant
role in sustaining R&D activities in the East and coastal regions,
43
including Beijing, Shanghai, Guangdong, Jiangsu and Zhejiang, the

government funds are mainly assimilated by the central and west provinces. Comparing the composition from 2012 to that of 2000 shows a
tendency of increasing funding from enterprises, whereas the government funding becomes less prevalent. With respect to the foreign funding for innovation, only limited proportions have emerged among the
giant cities such as Beijing and Shanghai.
part i
International knowledge transfer

and technological takeoff
Foreign direct investment, absorptive

capacity and regional innovative
capabilities: Evidence from China
3.1 Introduction
Technological capabilities are a key component of competitiveness at
national, regional and rm levels. The development of regional innovation capabilities has been of crucial importance for competitiveness
building in both developed and developing countries. The past decade
has seen increasing research on the regional innovation system. It is
argued that there is something distinctive and systemic about innovation
as a localised phenomenon that cannot be predicted by the more familiar
national innovation systems frameworks (Metcalfe, 1997; Braczyk,
Cooke and Heidenreich, 1998). Although knowledge is a non-rival
public production asset, which can generate positive externalities or
spillovers to others (Nelson, 1959; Arrow, 1962; Griliches, 1979),
knowledge spillovers are geographically localised (Jaffe, Trajtenberg
and Henderson, 1993; Audretsch and Feldman, 1996; Almeida and
Kogut, 1997, Anselin, Varga and Acs, 1997). Knowledge and information may ow more easily among agents located within the same area
because of social bonds that foster reciprocal trust and frequent face-toface contacts (Breschi and Lissoni, 2001). Therefore, there may be geographic boundaries to information ows or knowledge spillovers among
the rms in an industry (Krugman, 1991). These spatially bounded
knowledge spillovers allow companies operating nearby important
knowledge sources to introduce innovations at a faster rate than rival
rms located elsewhere. This is particularly the case for tacit knowledge,
that is, highly contextual and difcult to codify information, which is
therefore more easily transmitted through face-to-face contacts and personal relationships. Therefore, regional systems can much better capture
the local nature of innovation than other wider eco-geographical systems.
47
48
International knowledge transfer and technological takeoff
Among the important players in the regional innovation system are

the multinational enterprises (MNEs) and their afliates. MNEs are a
major force conducting cutting-edge research and innovation.
Multinational companies are often regarded as possessing certain
advantages that enable them to succeed in the global market. Their
advantages, their rm-specic assets, are often of a technological
nature. From 1970 to 1985, more than 80 per cent of royalty payments
for international technology transfers were made by afliates to their
parent companies (UNCTAD, 2005a). Therefore, openness of the local
economy to multinational enterprises located in the region may have a
signicant impact on regional innovation performance. The strength of
the effect of FDI on regional innovation capacity may, however, depend
on the absorptive capacity of the host regions. Innovation is an evolutionary and accumulative process. Only with the necessary capability to
identify, assimilate and develop useful external knowledge can the host
regions effectively benet from the advanced technology embedded in
FDI. Although there has been some research on the role of absorptive
capacity at the rm level, empirical studies of the role of absorptive
capacity in the evolving regional innovation system in the developing
countries have been rare.
In this chapter, I investigate the impact of FDI on regional innovation
capabilities in China with special emphasis on the role of absorptive
capacity and complementary assets in determining the strength of the
assimilation and modication. This chapter is organised as follows:
Section 3.2 presents the theoretical framework for understanding relations between FDI, absorptive capacity and regional innovation performance. Section 3.3 provides an overview of FDI and innovation in
China. Section 3.4 discusses the impact of FDI on regional innovation
capacity. Section 3.5 analyses FDI and regional innovation efciency.
Section 3.6 examines the linkage between FDI, innovation, and regional
economic growth in China. Section 3.7 concludes with some policy
implications.
3.2 Foreign direct investment, absorptive capacity

and regional innovation capabilities in developing countries:
a theoretical framework
The development of technological capabilities is the outcome of a
complex interaction of incentive structure with human resources,
Foreign direct investment and regional innovative capabilities
49
technology efforts and institutional factors. The incentive structure

includes macroeconomic incentives incentives from competition and
factor markets, and the institutions include market and non-market
institutions such as the legal framework, industrial institutions and
training and technology institutions. It is the interplay of all these
factors in particular country settings that determines at the regional
level how well the regions employ the resources and develop their
technological capabilities (Lall, 1992). In other words, it is the efforts
of the agents in a region (i.e., business, government and the universities)
and the strength of the linkages between these agents that determine the
performance of a regional innovation system (Braczyk, Cooke and
Heidenreich, 1998; Fu et al., 2006a).
Foreign direct investment contributes to regional innovation in four
ways. First, R&D and other forms of innovation generated by foreign
rms and R&D labs of MNEs increase the innovation outputs in the
region directly. Increasing FDI contributes to the emergence of newer
economies with more sophisticated technology generation (Athreye and
Cantwell, 2007). Globalisation of R&D activities by MNEs has been a
major shift in international business in recent decades. MNEs have three
types of global R&D programmes: support laboratories, locally integrated laboratories and internationally interdependent laboratories
(Pearce, 1999). When the operations of MNEs in host countries move
from being cost based towards the supply of higher-value parts, their
R&D activities may eventually accede to locally integrated laboratory
status. Increases in the international spread of subsidiary research
efforts in MNEs have tended on average to reinforce the position of
established centres of higher-grade technological activity (Athreye and
Cantwell, 2007). Depending on the availability of low-cost but highend skilled human capital, MNEs may even establish international
interdependent laboratories that are oriented to basic research in one
or more scientic disciplines. These independent laboratories have
the potential to reinforce a countrys developing strength in basic
research that may provide technological inputs to radical innovation
(Pearce, 2005).
Second, spillovers emanating from foreign innovation activities may
affect the innovation performance in the region in which they are
located. Several channels exist for knowledge to spill over from foreign
to local rms. These include knowledge transfers through the supply
chain, skilled labour turnovers and demonstration effects. Knowledge
50
transfer via supply chain requires effective quality linkages between

foreign rms and local suppliers and customers. When cross-regional
labour mobility is low, any benets from MNEs are likely to mainly go
to local rms (Greenaway, Upward and Wright, 2002). Demonstration
effects may also be local if rms only closely observe and imitate those in
the same region (Blomstrom and Kokko, 1998). Convincing evidence
suggests a productivity advantage of MNEs over domestic rms in
developing countries and some industrialised countries (e.g., Girma,
Greenaway and Wakelin, 2001). Empirical evidence on the actual
extent of spillovers from MNEs to domestic rms is, however, mixed
(Blomstrom and Kokko, 1998; Grg and Greenaway, 2004).
Third, FDI may affect regional innovation capacity through competition effects. Market competition may also be a two-edged sword in
terms of its effect on innovation. Geroski (1990) argues that lack of
competition in a market will give rise to inefciency and result in
sluggish innovative activity. On the other hand, the traditional
Schumpeterians claim that monopoly power makes it easier for rms
to appropriate the returns from innovation and thereby provides the
incentive to invest in innovation (Cohen and Levin, 1989; Symeonidis,
2001). Foreign R&D activities may also crowd out domestic innovation activities as they attract the most talented researchers and compete in the markets of innovation products that threaten local rms,
SMEs in particular (Fu, 2004a, 2007; Aghion et al., 2005; UNCTAD,
2005b). In the Chinese electronics industry, Hu and Jefferson (2002)
nd signicant productivity depression rather than positive spillover
effects of FDI on domestic rms.
Finally, in addition to greater R&D investments by MNEs and their
afliates, FDI may contribute to regional innovation capabilities by
advanced practices and experiences in innovation management and
thereby greater efciency in innovation. Innovation is not a simple
linear transformation with basic science and other inputs at one end of
a chain and commercialisation at the other (Hughes, 2003). Successful
innovation requires more than brilliant scientists. It involves everyone
from top management to employees in its R&D, nance, production
and marketing divisions. It requires high-quality decision-making;
long-range planning; motivation and management techniques; coordination; and efcient R&D, production and marketing. Therefore,
the innovation performance of a rm is determined not only by hard
factors such as R&D personnel and R&D investment but also by
51
certain soft factors such as management practices and governance

structures (Aghion and Tirole, 1994; Bessant, Caffyn and Gilbert,
1996; Cosh, Fu and Hughes, 2004). MNEs are major participants in
innovation and are more experienced in innovation management.
They may contribute to the local innovation system by transferring
managerial know-how to local rms through spillover effects.
However, two main conditions are needed for signicant spillovers
from FDI: one is the absorptive capacity of the local rms and organisations (Cohen and Levinthal, 1989; Girma, 2005), and the second is
sufcient effective linkages generated between the foreign and domestic
economic activities (Balasubramanayam, Salisu and Sapsford, 1996; Fu,
2004b). Absorptive capacity refers to the ability of an organisation or
region to identify, assimilate and exploit knowledge from the environment (Cohen and Levinthal, 1989). Absorptive capacity is usually
proxied by the technology gap between the foreign and the domestic
rms, R&D intensities of the local rms or human capital embodied in
local rms. Studies using the rst approach nd that spillovers are
present when the technology gaps are moderate (Kokko, Tansini and
Zejan, 1996); smaller plants or plants with a low share of skilled workers
in the workforce lack the necessary absorptive capacity to benet from
FDI. R&D activities of organisations have two faces (Aghion and
Howitt, 1992, 1998; Grifth, Redding and Van Reenen, 2003). One is
the widely acknowledged knowledge creation function; the other is their
role in learning and promoting absorptive capacity, given the fact that
innovation is cumulative and path dependent. It is argued that a certain
level of R&D intensity is needed before rms benet from FDI-generated
externalities. The absorptive capacities of the local rms and organisations in turn determine the overall absorptive capacities of the host
region, as they are the basic elements in a regional innovation system.
In addition to absorptive capacity, complementary assets (ingredients)
may play an important role in forming the overall dynamic capabilities
required to convert technological opportunities into innovative sales and
competitive market advantage (Teece, 1986; Teece, Pisano and Shuen,
1997; Hughes and Scott Morton, 2006; Cosh, Fu and Hughes, 2012; Fu,
2012). The introduction of substantially new or improved products
requires parallel shifts in the strategy of organisations, the structure of
organisations and practices in management. MNEs are also more likely
to locate in regions with sound scientic and educational infrastructure,
the potential for intra- and interindustry spillovers or locally embedded
52
specialisation because they are seeking complementary assets to most

effectively transfer and commercialise their technological advantage
(Cantwell and Santangelo, 1999; Cantwell and Piscitello, 2002).
Therefore, a regions receptiveness to new ideas, the strength of its
entrepreneurship, information and communication infrastructure as
well as the presence of clusters of high-technology industries may all
enhance its capability in the assimilation of new ideas and technology
generated internally or transferred from external sources.
In summary, FDI impacts on regional innovation capabilities in
several ways. First, FDI may contribute to the overall regional innovation performance directly as a result of the greater innovation intensity
in MNEs. Second, FDI may contribute to regional innovation performance through greater innovation efciency in MNEs. Third, FDI may
affect innovation capability of local indigenous rms through technological and managerial knowledge spillovers. The strength of the FDI
effect on regional innovation capabilities depends on the absorptive
capacity of and the complementary assets in the domestic sector, the
linkages between the foreign and the domestic sectors and the technology content of the FDI.
3.3 FDI and innovation in China

FDI in China
Since it launched economic reforms and called for foreign capital participation in its economy in 1979, China has received a large part of international direct investment ows. China has become the second largest FDI
recipient in the world, after the United States, and the largest host country
among developing countries. The fastest growth of FDI inows into China
started in the spring of 1992 after Deng Xiaoping circuited Chinas southern coastal areas and Special Economic Zones (SEZs). Since then, China
has adopted a new approach, which turned away from special favourable
regimes to FDI in the SEZs and selected coastal regions towards more
nationwide implementation of open policies for FDI. Inows of FDI into
China reached a peak level of US$45.463 million in 1998. After a drop in
1999, mainly because of the impact of the Asian nancial crisis and the rise
of acquisition transactions in both OECD and non-OECD countries, FDI
inows into China resumed their fast-growing trend and reached a historical peak of US$60.63 billion in 2004 (Figure 3.1).
53
700
12000
600
10000
500
8000
400
6000
300
4000
200
2000
100
FDI
Exports
Trade and FDI in China, 19852004 (US$100 mil)

14000
19
8
19 5
8
19 6
8
19 7
8
19 8
8
19 9
9
19 0
9
19 1
9
19 2
9
19 3
9
19 4
9
19 5
9
19 6
9
19 7
9
19 8
9
20 9
0
20 0
0
20 1
0
20 2
0
20 3
04
Exports and Imports
FDI
Figure 3.1 Trade and FDI in China, 19852004

Regional distribution of FDI stock in China 2005

Inland region
13%
Coastal region
87%
Coastal region
Inland region
Figure 3.2 Regional distribution of FDI stock in China, 2005

Source: www.fdi.gov.cn/.
This huge inow of FDI is, however, highly concentrated geographically in the coastal region. By 2005, total FDI stock in the coastal provinces accounted for 87 per cent of total FDI stock in China (Figure 3.2).
This uneven distribution of FDI in China has economic, policy and
geographic attributes and generates an uneven impact on regional competitive performance in China. As a result, the average FDI intensity
54
measured by foreign assets in total industrial assets was 28 per cent in the
coastal region, which was three times higher than that in the inland region
at only 7 per cent. In Guangdong, Fujian and Shanghai, nearly half of the
net xed industrial assets are foreign invested (Table 3.1).
The sources and entry modes of inward FDI in China have also
evolved over time. While investment from overseas Chinese in Hong
Kong, Macao and Taiwan was the major source of inward FDI in the
1980s, the 1990s has seen increasing inward FDI from the major
industrialised countries and other OECD countries. The entry modes
of foreign rms have also evolved from joint ventures in the 1980s to
wholly foreign-owned enterprises in the 1990s, which accounted for
more than 70 per cent of all FDI projects in the late 1990s. Some large
MNEs have started to set up their global R&D centres in China, mostly
in its large coastal cities such as Beijing and Shanghai. With the launch
of the Western Region Development Strategy by the Chinese government, the relocation of some foreign-invested enterprises into the inland
regions has been increasing. They are, however, mostly in labour- or
land-intensive industries.
Innovation
As one of the important drivers of competitiveness, innovation effort and
performance in China are not even across the regions. Again, a signicant
gap in innovation exists between the coastal and inland regions. In 2004,
the coastal provinces accounted for 82 per cent of Chinas total invention
patent applications, 79 per cent of the total sales of new products and
73 per cent of total industrial R&D expenditure. The innovation activity
in the coastal provinces is further concentrated in several provinces
including Guangdong, Shanghai, Jiangsu, Shangdong and Zhejiang
(Figures 3.3, 3.4, 3.5). Expenditure on acquisition of technology from
abroad by the coastal region accounted for 67 per cent of Chinas total
expenditure on foreign technology acquisition in 2004 and was about
twice that of the inland region (Table 3.2). Jiangsu, Shanghai, Beijing and
Guangdong are the top four regions that spend most on foreign technology acquisition. While Shanghai and Beijing acquire mainly foreign
technology, Jiangsu and Guangdong rely on both foreign and domestic
technology. The inland regions, with the exception of Chongqin, spend
much less than most of the coastal regions, with one-third relying more on
foreign technology and two-thirds acquiring more domestic technology.
55
Table 3.1 FDI intensity: share of foreign assets in total industrial assets
1999
2004
19992004 average
Guangdong
Fujian
Shanghai
Jiangsu
Tianjin
Beijing
Zhejiang
Liaoning
Hainan
Shandong
Guangxi
Hebei
Coastal average
50%
48%
40%
28%
28%
19%
19%
16%
19%
13%
9%
12%
25%
51%
50%
49%
37%
28%
22%
22%
16%
14%
16%
14%
12%
28%
51%
50%
44%
32%
27%
20%
19%
15%
15%
14%
13%
12%
26%
Chongqing
Anhui
Jilin
Hubei
Jiangxi
Henan
Hunan
Shaanxi
Yunnan
Sichuan
Shanxi
Ningxia
Heilongjiang
Inner Mongolia
Gansu
Guizhou
Xinjiang
Qinghai
Tibet
Inland average
13%
9%
11%
11%
6%
8%
4%
6%
6%
4%
2%
4%
6%
3%
3%
2%
2%
1%
1%
5%
16%
12%
12%
9%
9%
7%
8%
4%
5%
6%
7%
8%
5%
4%
2%
3%
1%
1%
NA
7%
16%
12%
11%
10%
7%
7%
6%
6%
6%
5%
5%
5%
5%
4%
2%
2%
1%
1%
1%
6%
Source: Statistical Yearbook of China.
56
Regional distribution of number of invention applications, 2004

Guangdong
24%
Inland regions
18%
Other coastal
regions
14%
Jiangsu
7%
Zhejiang
7%
Tianjin
18%
Shanghai
12%
Figure 3.3 Regional distribution of number of invention applications, 2004

Source: First Economic Census of China (2004).
Regional distribution of sales of new products, 2004
Inland regions
21%
Guangdong
15%
Shanghai
13%
Other coastal
regions
21%
Zhejiang
7%
Shandong
11%
Jiangsu
12%
Figure 3.4 Regional distribution of sales of new products, 2004

Innovation by foreign rms

Since the late 1990s, with the increasing globalisation in innovation, the
R&D activities of foreign rms in China have been increasing at a faster
pace than those of the domestic rms. The average annual growth of
R&D expenditure from 1998 to 2004 was 38 and 33 per cent in foreign
invested enterprises and ethnic Chinese invested rms, respectively.1
1
Ethnic Chinese invested rms refer to foreign rms that have owners from Hong
Kong, Macao and Taiwan.
57
Table 3.2 Expenditure on technology acquisition, 2004 (10th)
Provinces
Expenditure
on acquisition
of technology
from abroad
Percentage
Expenditure on
acquisition of
technology
from domestic
sources
Percentage
699192
90289
33743
2865
61465
14075
6578
40682
80177
25842
11368
1108
100.0%
14.9%
14.7%
8.5%
7.0%
5.7%
4.4%
4.1%
3.8%
1.7%
0.9%
0.0%
65.9%
100.0%
12.9%
4.8%
0.4%
8.8%
2.0%
0.9%
5.8%
11.5%
3.7%
1.6%
0.2%
52.7%
36568
38464
29658
27226
10095
2695
29977
7769
27010
18144
24384
41492
6157
6.9%
4.2%
3.7%
3.5%
2.3%
2.3%
2.0%
1.8%
1.6%
1.5%
1.3%
1.3%
0.6%
5.2%
5.5%
4.2%
3.9%
1.4%
0.4%
4.3%
1.1%
3.9%
2.6%
3.5%
5.9%
0.9%
5579
15479
1850
4525
3925
0.4%
0.3%
0.2%
0.1%
0.0%
0.8%
2.2%
0.3%
0.6%
0.6%
Expenditure on
Expenditure
acquisition of
on acquisition technology
of technology from domestic
from abroad
sources
Whole
3679496
Jiangsu
549744
Shanghai
540190
Beijing
314287
Guangdong 258259
Liaoning
210968
Tianjin
163259
Zhejiang
151700
Shandong
138257
Fujian
64123
Guangxi
33388
Hainan
486
Coastal
region
total
Chongqin
253162
Hubei
154054
Hebei
137388
Henan
127227
Anhui
85003
Jilin
84493
Heilongjian
74177
Hunan
66696
Shaanxi
60553
Shanxi
54788
Jiangxi
48822
Sichuan
47248
Inner
20988
Mongolia
Gansu
16013
Yunnan
9627
Ningxia
9012
Guizhou
3966
Xinjiang
1529
58
Table 3.2 (cont.)
Provinces
Qinghai
Tibet
Inland
region
total
Expenditure on
Expenditure
acquisition of
on acquisition technology
of technology from domestic
from abroad
sources
90
na
Expenditure
on acquisition
of technology
from abroad
Percentage
Expenditure on
acquisition of
technology
from domestic
sources
Percentage
0.0%
0.0%
34.1%
47.3%
na
Source: First Economic Census of China, 2004.

Note: Expenditure on technology acquisition refers to expenditure on purchasing
technology, including for product design, process design, blueprints, prescription, patents
and relevant key equipment, instruments and samples from foreign or domestic sources.
Regional distribution of industrial R&D, 2004

Guangdong
15%
Inland regions
27%
Jiangsu
13%
Other coastal regions

17%
Liaoning
7%
Shanghai
9%
Shandong
12%
Figure 3.5 Regional distribution of industrial R&D, 2004

This is much higher than that in indigenous rms at 25 per cent over the
same time period (Table 3.3).
In 2004, the number of foreign rms was about one-third of Chinas
total number of enterprises. Although the R&D expenditure and R&D
staff of the foreign rms accounted for only 27 per cent and 18 per cent
59
Table 3.3 Growth rate of R&D expenditure

Average annual
change
1998 1999 2000 2001 2002 2003 (19982003)
Indigenous
enterprises
FIEs by HKTM
FIEs
18% 44% 19% 18% 22% 32% 25%

50% 23% 12% 25% 27% 59% 33%
7% 73% 19% 25% 53% 49% 38%
Source: First Economic Census of China, 2004.
of Chinas total industrial R&D expenditure and R&D staff, innovation outputs by foreign rms in terms of percentage of total sales of new
products and the percentage of invention patent applications by foreign
rms were both more than 40 per cent. This suggests higher productivity in innovation and better innovation management in foreign rms
than in indigenous rms. The expenditure on acquisition of foreign
technology by foreign rms accounted for nearly half of Chinas total
expenditure on foreign technology acquisition, suggesting greater effort
by foreign rms in acquiring advanced technology (Figure 3.6).2
3.4 FDI and innovation capacity: empirical evidence

The econometric analysis of the impact of FDI on regional innovation
capacity starts from a basic regions innovation production function as
follows:
Y i;t RDSi;t1 RDPi;t1 HCi;t1 FDIi;t1
3:1
where Y is innovation output, RDS is R&D expenditure, RDP is the

number of people involved in research and development activity, HC is
human capital, FDI is FDI intensity measured by the ratio of net xed
assets of FIEs to total industrial net xed assets in the region and is the
error term that has the normal property. All the variables are in
2
Expenditure on absorption of foreign technology refers to expenditure on

assimilation of foreign technology; this includes expenses for training,
assimilation, imitation, and development of foreign technology and relevant staff
costs and equipment costs.
60
Innovation activities: foreign and indigenous enterprises in China, 2004

31%
Expenditure on absorption
69%
Expenditure on acquisition of
foreign technology
46%
54%
Percentage of total invention

patent applications
40%
60%
42%
Percentage of total new sales
58%
27%
R&D expenditure
73%
18%
R&D staff
82%
32%
Number of firms
68%
0%
20%
FIEs
40%
60%
80%
Percentage of total
100%
Indigenous entreprises
Figure 3.6 Innovation activities: foreign and indigenous enterprises in China,

2004
logarithm. We use one-year lagged values for all the explanatory variables. In other words, we assume that innovation production in a given
year is reected in the patents that are granted one year in the future.
Another advantage in using one-year lag for the independent variables
is that it removes the possible endogeneity between FDI and the dependent variable, number of patents, as FDI may choose to locate in regions
that have high innovation capacity.
Here, following Jaffe (1989) and Acs, Luc and Varga (2002), we
measure innovation output by number of patents granted to domestic
applicants per 10,000 population. Although patent number has its
advantages, it also suffers from the low face validity problem that
patents do not fully reect the commercial success and the value of
new and renewed products (Acs and Audretsch, 1990; Kleinknecht,
1996). Investment in research and development is often found to be a
signicant determinant of innovation performance. Regions that invest
more in R&D are more likely to innovate because R&D directly creates
new products and processes. Labour force skills and the availability of
61
qualied scientists and engineers, particularly the qualied research

staff directly involved in R&D activities, are other widely recognised
critical factors that contribute to rm innovation performance
(Hoffman et al., 1998; Porter and Stern, 1999).
To capture the effects of absorptive capacity and complementary
assets, we expend the model by including the interaction terms of FDI
and absorptive capacity (ABC) and FDI and complementary assets (CA)
alternatively. According to the literature discussed earlier, absorptive
capacity is measured in two ways: regional R&D intensity (RDS) measured by the R&D spending to GDP ratio and labour force quality (HC)
measured by the percentage of the population with 15 years of schooling. Absorptive capacities serve to enhance a regions capacity to recognise and absorb relevant external resources for innovation (Cohen
and Levinthal, 1990). Complementary assets (CA) are also measured in
three ways: (1) number of computers per thousand households
(COMP), which may capture many innovative characteristics of a
region, for example, information infrastructure and receptiveness to
new ideas; (2) the share of value added from high-technology industry
in regional total value added (HITECS), which indicates the industry
and technology structure in a region and the industry infrastructure to
assimilate and develop the R&D externalities from FDI; and (3) transaction value in technological markets (TECHMKT), which measures
the extensiveness and depth of technology linkages, ows and transactions in the regions and reects the development level and activeness
of the institutions that facilitate the transfer of technology within a
region.3 Equation (3.1) is thus extended to the following:
Y i;t RDSi;t1 RDPi;t1 HCi;t1 FDIi;t1
ABCi;t1 FDIi;t1 CAi;t1 FDIi;t1 i;t
3:2
All the variables are in logarithm.

The three types of patents are invention, utility model and external
design. Invention is a new technical solution related to a product,
process or improvement thereof. Utility model refers to practical and
new technological proposals on the shape and structure of a product or
the combination of shape and structure. External design involves a new
3
Technological transactions include technology transfer, technology consultation,

technical service, technical training, technology-equity share exchange,
technology intermediation and various research-production co-operations.
62
design of shape, pattern or combination of colour or aesthetic properties. Invention patents are regarded as major innovations. To obtain a
patent for invention, an application must meet the requirements of
novelty, inventiveness and practical applicability according to patent
law. It usually takes about one to one and a half years to process an
invention patent application in China, but six months or even less for
utility model and design patents (Cheung and Lin, 2004). Given the
different novelty and importance of the three types of patents, we run
regression against Equation (3.2) using these three types of patent as the
dependent variable alternatively to examine the impact of FDI on
innovations of different degrees of novelty.
The data used relate to a provincial-level panel data set for the 31
provinces and municipality cities in China from 1998 to 2004. The data
are collected from the China Statistical Yearbook and the Ministry of
Science and Technology of China (MOST) online database. This study
differs from some existing studies in that it uses the data from MOST for
R&D expenditure and R&D staff rather than using the data of investment in innovation published in the China Statistical Yearbook. The
MOST data are more precise on R&D expenditure because investment
in innovation includes, in addition to spending on research and development, expenditure on renewal of xed assets, capital construction,
new site construction and expenditure on corresponding supplementary
projects for production or welfare facilities and related activities.4 The
MOST measurement of R&D is also better than using expenditure and
staff on science and technology because, according to the denition
4
It includes (1) projects listed in the innovation plan of the current year of the
central government and the local governments at various levels as well as projects
that, although not listed in the innovation plan of the current year, are to continue
being constructed in this year, using the investment listed in the plan of innovation
of previous years and carried forward to this year; (2) projects of technological
innovation or renewal of the original facilities, arranged both in the plan of
innovation and in the plan of capital construction; extension projects (main
workshops or a branch of the factory) with newly increased production capacity
(or project efciency) not up to the standard of a large and medium-sized project;
and projects moving the whole factory to a new site so as to meet the requirements
of urban environmental protection or safe production; (3) projects of
reconstruction or technological innovation with a total investment of 500,000
RMB yuan and more by state-owned units, though listed neither in the plan of
capital construction nor in the plan of innovation; the projects in the state-owned
units of moving the whole factory to a new site so as to meet the requirements of
urban environmental protection or safe production.
63
given by MOST and the State Statistical Bureau (SSB), science and
technology include the broad social science disciplines in addition to
the natural science and engineering disciplines.
Table 3.4 reports the estimation results of Equation (3.2) on the
impact of FDI on regional innovation capacity. The estimated coefcient of the FDI variable is positive and statistically signicant at the
99 per cent signicance level. This fact suggests a signicant contribution of FDI to regional innovation capacity. The magnitude of the
estimated FDI coefcient in the base equation is 0.356, suggesting a
1 per cent increase in FDI intensity increases regional patent output by
0.35 per cent. These magnitudes are of a similar level to the R&D
expenditure variable and sometimes even higher than that of the
human capital variable.
The estimated coefcients of the two proxies of absorptive capacity
RD intensity and labour force skills bear the expected positive sign
and are statistically signicant at the 99 per cent signicance level. The
estimated coefcients of the interaction terms between the absorptive
variables and the FDI variable are positive and statistically signicant as
well. This evidence provides strong support to the proposition of the
important role of absorptive capacity in assimilation of the knowledge
spillovers from FDI.
We have three proxies to capture the complementary assets for
innovation in the regions: number of computers per thousand households, the share of the high-technology sector in regional total value
added and the transaction value in the technological market. The interaction terms between FDI and the rst two complementary assets
variables are positive and statistically signicant.5 This evidence suggests that better information and communication infrastructure, greater
receptiveness to new ideas and a better developed cluster of hightechnology sectors in the local economy will complement the advanced
technology embodied in FDI, facilitate the innovation process and
therefore lead to greater innovation capacity of the region. The
5
The correlation coefcients between FDI intensity on one hand and computer
usage, share of high-technology industry and technological market transaction
values are as high as 0.57, 0.54 and 0.40, respectively, and statistically signicant
at the 1 per cent signicance level. Inclusion of FDI and these complementary
assets variables in the regression simultaneously raises signicant multicolinearity
problems. We, therefore, include FDI and the interaction terms only in the
regression in addition to other variables of interest.
Table 3.4 Impact of FDI on regional innovation capacity

Dependent variable: log (patents per thousand population)
C
LOG(RDGDP?(1))
LOG(RDSTAF?(1))
LOG(HC?(1))
LOG(FDIS?(1))
LOG(FDIS?(1))*
LOG(RDGDP?(1))
LOG(FDIS?(1))*
LOG(HC?(1))
LOG(FDIS?(1))*
LOG(COMP?(1))
LOG(FDIS?(1))*
LOG(HITECS?(1))
LOG(FDIS?(1))*
LOG(TECHMKT?(1))
LM
Hausman
Model
Coef
p-value Coef
p-value Coef
p-value Coef
p-value Coef
p-value Coef
p-value
1.309***
0.354***
0.149*
0.316***
0.357***
0.000
0.002
0.058
0.000
0.000
0.000
0.000
0.041
0.001
0.000
0.008
2.591***
0.243**
0.173**
0.724***
0.816***
0.000
0.043
0.025
0.000
0.000
1.383***
0.392***
0.157**
0.337***
0.310***
0.000
0.001
0.047
0.000
0.000
0.000
0.001
0.061
0.001
0.000
1.321***
0.411***
0.118
0.334***
0.302***
0.000
0.000
0.132
0.000
0.000
0.134**
0.018
0.020*
0.086
0.010
0.291
1.506***
0.690***
0.147**
0.304***
0.474***
0.138***
1.265***
0.403***
0.148*
0.287***
0.299***
0.041*** 0.007
258.21***
0.07
RE
175.27***
2.18
RE
247.74***
7.74
RE
212.76**
6.53
RE
Note: ***Signicance at 99 per cent level; **signicance at 95 per cent level; *signicance at 90 per cent level.
250.77**
7.85
RE
268.82
32.18***
FE
65
estimated coefcient of the interaction variable between FDI and the

technological market transaction value bears the expected positive sign
but is not statistically signicant. This may be explained by the lack of
foreign participation in the technological market as evidenced in the
survey that we discussed in an earlier section (Zhou, 2006).
3.5 FDI and regional innovation efciency

Statistical analysis to investigate the impact of FDI on regional innovation efciency involves two steps. We rst evaluate the innovation
efciency using data envelopment analysis (DEA), which allows for
multi-output and multi-input in the model. Second, we explore the
role of FDI in the determination of regional innovation efciency.
The three main approaches to the measurement of efciency are
ratio analysis, such as labour productivity and capital productivity;
the normal econometric approach, such as the Solow-type total factor
productivity (TFP) index; and the frontier approach, such as data
envelopment analysis. TFP provides a more comprehensive guide to
efciency than does partial productivity. It takes into account the
contribution of factors other than raw labour and capital, such as
managerial skills and technical know-how. The conventional Solowtype total factor productivity denes TFP growth as the residual of
output growth after the contribution of labour and capital inputs has
been subtracted from total output growth. This method, however,
attributes all the deviations from the expected output to TFP, without
taking into account measurement error. It is subject to several wellknown assumptions: (1) the form of production function is known,
(2) there are constant returns to scale, (3) rms exhibit optimising
behaviour, and (4) technical change is neutral. If these assumptions do
not hold, TFP measurements will be biased (Coelli et al., 1998; Arcelus
and Arocena, 2000).
The frontier approach evaluates a rms efciency against a measure
of best practice. Two main methods are used for the estimation. One is a
non-parametric programming approach, data envelopment analysis;
another is a parametric production function approach, stochastic frontier analysis (SFA). In the DEA approach, a best-practice function is built
empirically from observed inputs and outputs. The efciency measure of
a rms innovation activity is dened by its position relative to the frontier
of best performance established mathematically by the ratio of the
66
weighted sum of outputs to the weighted sum of inputs (Charnes, Cooper

and Rhodes, 1978). The strength of the programming approach lies not
only in its lack of parameterisation but also in that no assumptions are
made about the form of the production function. In addition, the programming approach allows us to estimate efciency with multi-output
and multi-input. This technique has a major shortcoming in that it has no
provision for statistical noise, or measurement error, in the model
(Norman and Stoker, 1991; Greene, 1997).
Given the advantages and disadvantages of the different efciency
estimation approaches, we use the DEA approach in the estimation of
innovative efciency because this method allows us to evaluate a rms
efciency in innovation against best practice, and it allows us to estimate efciency with multi-output and multi-input.
In the DEA approach, for a sample of n rms, if X and Y are the
observations on innovation inputs and outputs, assuming variable
returns to scale, the rms innovative efciency score, , is the solution
to the linear program problem,
Max;
st: yi Y 0
xi X 0
i 0
i 1
i 1; ; n:
3:3
where is a scalar and is an nx1 vector of constants. The efciency

score ranges from 0 to 1. If k = 1 and all slacks are zero, the kth rm is
deemed to be technically efcient (Cooper, Seiford and Tone, 2000).
In the analysis, as our major objective is to maximise innovation
output, we concentrate on output-oriented efciency, which reects a
rms efciency in producing maximum innovation output with given
inputs, under constant returns to scale. Innovation output is measured
by the number of patents granted to each region. Given the differences
in degrees of novelty across the three types of patents invention, utility
model and design we include these three measures of innovation as
outputs in our multi-output DEA model. Inputs in our models include
the R&D expenditure to GDP ratio and the total number of R&D staff
as a share of the total population. As invention patents are of higher
degrees of novelty, we assume their importance is twice that of the
67
utility models and design. Therefore, the weights restriction we use in

the three-output DEA model is as follows:
Qinvention 2 qutilitymodel 2 qdesign
3:4
The estimated regional innovation efciency is then regressed on FDI

intensity in each region. In the estimation of rm innovative efciency,
the efciency scores have an upper bound of 1.0 and a lower bound of
0.0; the ordinary least squares estimates would be inconsistent.
Therefore, the regression model for technical efciency is specied in
the form of the Tobit model as follows (Tobin, 1958; Fu and
Balasubramanyam, 2003):
(
Xi
if Xi 5 1
IE
3:5
1
otherwise
where IE = innovative efciency, and Xi is a vector of explanatory
and control variables including FDI intensity, labour force skills and
industry structure in the region.
Figure 3.7 shows the distribution of the estimated regional innovation efciency. Patterns revealed from this gure suggest that the innovation efciency of Chinese regions has been increasing over time from
Regional innovation efficiency, China
1.2
SH03
SH04
Efficiency score
BJ04
0.8
TJ04
BJ03
ZJ03
ZJ04
ZJ00
0.6
ZJ02
ZJ01
TJ03
GD04
TJ00
TJ01
LN00
GD03
SH02
QH01
HLJ04
NMG04
TJ02
SH00 BJ01 NMG01
SH01 FJ02
NMG00
FJ03
LN03
BJ00
QH00
GD02
LN04
FJ04 JL04
NMG03
HEB04
HLJ03
SD04
SD00
SD01
NMG02
FJ00
FJ01 GD01LN01
SD03
BJ02
HLJ02
JL00
QH04
CQ03
JL03 JS03
GD00
LN02
HLJ01
HUB04
JS04
JS00 YN00
QH02
YN04
HEB01
YN01HEB02
JL01
QH03 SX03 CQ04
HEB03
JS02
SD02
HUN04 SX04
HEB00
HUN00
HUN03
HEN03
SX00 HUN01
JS01
GX04
GX00
CQ02 HEN02 JL02 SX02 GX03
HEN04
HEN00
GX01
HUB03YN03
CQ00
GZ00 JX00
HUN02
SC04
GS04
JX01SX01
CQ01 HEN01
YN02
GX02
HUB01
JX02
SC00
GZ03
GZ04
GS02
HUB00
JX03
GZ01
HUB02 SC02
JX04 SHX04
GS00
GZ02
GS03
GS01
AH04
SC03
AH00
SC01
AH03
AH01
AH02
SHX03
SHX00
SHX02
SHX01
HLJ00
0.4
0.2
20
40
60
80
Figure 3.7 Regional innovation efciency, China

Source: Authors estimation.
100
120
140
68
Table 3.5 Impact of FDI on regional innovation efciency

Dependent variable: LOG(IE)
Variable
Coefcient
Prob.
Coefcient
Prob.
C
LOG(FDIS?(1))
LOG(HC?(1))
LOG(HITECS?(1))
Random Effects
Hausman statistics
Adjusted R-squared
0.963***
0.157**
0.000
0.044
1.481***
0.192**
0.596***
0.263**
0.001
0.021
0.000
0.012
0.16
0.14
0.05
0.50
Note: ***Signicance at 99 per cent level; **signicance at 95 per cent level;

*signicance at 90 per cent level.
2000 to 2005. Two observations form the regional innovation frontier:

Shanghai in 2003 and 2004. Other regions close to the frontier are the
several coastal provinces and municipalities including Tianjin, Beijing,
Zhejiang and Guangdong.
Table 3.5 reports the estimated results on the impact of FDI on
regional innovation efciency. The estimated coefcients of the FDI
variable are positive and statistically signicant at the 5 per cent signicance level. The results are robust across different specications,
suggesting a signicant positive impact of FDI on regional innovation
efciency. Regions with higher FDI intensity are more efcient in innovation. Human capital, measured by the percentage of the population
with a university degree, has the largest impact on regional innovation
efciency. The estimated coefcient of the high-technology share variable bears an unexpected negative sign. This may be explained by heavy
investment in R&D and large R&D personnel numbers in the domestic
high-technology sector that have not been used efciently.
3.6 Innovation and regional economic growth

Finally, we include the innovation output variable in a growth function
in the light of the new growth theory and estimate the impact of
innovation on regional economic growth:
yit ?tit lit kit xit it
3:6
69
where i and t denote regions and time, respectively. The disturbance

term, vit, varies across regions and time and has the usual properties.
The dependent variable, yit, is the real growth rate of GDP; the
explanatory variables are the number of patents granted, tit; the
growth rate of labour, lit; the growth rate of the capital stock, kit;
and real growth rate of exports, xit, and yit-1, which is the one-year lag
of yit. Lowercase denotes logarithm. We introduce exports into the
production function explicitly for three reasons. First, the incentives
associated with export orientation are likely to lead to higher total
factor productivity because of economies of scale and competition
effects. Second, exports are likely to alleviate serious foreign exchange
constraints and thereby enable the country to import more advanced
machinery and materials. Third, exports are likely to result in a higher
rate of technological innovation and dynamic learning from abroad,
although the direction of causality is debatable (Balasubramanyam,
Salisu and Sapsford, 1996).
We estimate Equation (3.6) for all the regions in the country and for
the coastal and inland region subgroups separately. To examine the
FDI-innovation-growth link in the two regional groups, we also estimate Equation (3.2) for the coastal and inland region subsample
separately. The estimated results are reported in Table 3.6. In the
coastal region, FDI has been a signicant contributor to regional
innovation performance, and innovation again contributes signicantly to output growth. The story in the inland region is, however,
not the same. In the inland region, the estimated coefcient of the FDI
variable, though bearing the expected positive sign in the innovation
function, is not statistically signicant. Again, the estimated coefcient
of the innovation variable is not statistically signicant although the
sign is positive as expected. These facts suggest that FDI is not a
signicant driver of innovation capability in the inland region, nor
does innovation lead to regional economic growth. The estimated
coefcients of the exports and innovation variables show the expected
positive sign and are statistically signicant when they enter the
regressions alternatively. However, the estimated coefcient of innovation marginally loses its signicance when it enters the regression
together with exports. This is likely due to the fact that the two
variables are correlated because greater innovation enhances competitiveness and hence more exports, and greater openness by exporting
may contribute to more innovation.
Table 3.6 The FDI-innovation-growth linkage in the coastal and inland regions
Log(rgdp)
Log(patent)
Coastal
Coef
p-value
Coef
p-value
Coef
p-value
LGL
LKY
LGREX
LPATENT
Constant
LM
Hausman
0.085
0.072***
0.097***
0.481
0.004
0.003
0.075
0.095***
0.537
0.000
0.166***
0.02
5.9
C
0.000
0.010***
0.128***
0.48
1.49
C
0.008
0.002
0.087
0.077***
0.072**
0.006
0.125***
0
5.01
C
0.467
0.002
0.045
0.119
0.002
Coef
p-value
FDIS
RDGDP
HC
2.834**
0.513**
6.176**
0.018
0.023
0.010
Constant
LM
Hausman
6.648***
85.93
6.54
C
0.000
Log(rgdp)
Log(patent)
Inland
Coef
p-value
Coef
p-value
Coef
p-value
LGL
LKY
LGREX
LPATENT
Constant
LM
Hausman
0.465*
0.055***
0.006
0.050
0.001
0.714
0.468**
0.077***
0.036
0.000
0.138***
4.55
1.86
C
0.000
0.006
0.116***
5.71**
3.08
RE
0.154
0.000
0.469**
0.078***
0.001
0.006
0.116***
5.71**
3.08
RE
0.037
0.000
0.949
0.155
0.000
Coef
p-value
FDIS
RDGDP
HC
2.694
0.199
4.231*
0.143
0.129
0.069
Constant
LM
Hausman
6.268***
133.36
6.3
C
0.000
Note: ***Signicance at 99 per cent level; **signicance at 95 per cent level; *signicance at 90 per cent level.
71
3.7 Conclusions
This chapter has investigated the impact of FDI on regional innovation
capabilities and efciency in a fast-growing developing country that has
received phenomenal amounts of foreign direct investment. We nd
that FDI can contribute signicantly to overall regional innovation
capacity. The strength of this positive effect depends, however, on the
availability of the absorptive capacity and the presence of innovationcomplementary assets in the host region. In the coastal region of China,
which possesses a pool of intelligent educated R&D staff and skilled
labour and hosts most of Chinas R&D activities and top universities
and research institutes and where inward FDI has evolved from labourintensive processing activities to more strategic asset-seeking types of
FDI by major MNEs, FDI has played a signicant role in promoting
regional innovation capacity as well as regional innovation efciency
from 2000 to 2004. This increased regional innovation and technological capability contribute further to the fast regional economic growth
in this region. However, the inland provinces have not experienced a
similar innovation-growth-promotion effect of FDI. The type and quality of FDI they have attracted and the lack of absorptive capacity and
complementary assets may all be blamed for this failure in the inland
region. FDI into the inland regions of China is mainly in labour-, landor resource-intensive production activities. The newly increased relocated FDI may also be motivated by tax holidays provided by local
government and low labour and land costs in the inland region. The
technology content of this FDI is low and spillovers are limited. The lack
of absorptive capacity and complementary assets may become another
bottleneck that hinders the upgrading of the local economy towards the
higher end of the value chain and the transition of the local economy to
a knowledge-based economy.
Our attention to innovation has long been mistakenly placed on
R&D expenditure. R&D spending is, however, only one of the important inputs of innovation. In many cases, huge R&D spending on
innovation activities has not generated sufcient innovation output,
as we had expected. The productivity of innovation and the management of innovation are of crucial importance, especially for developing countries whose resources for innovation are limited. This
research indicates that FDI contributes not only to the outputs of the
72
regional innovation system but also to the productivity of innovation

in developing countries.
The sample period under study is anchored in an era with increasing
globalisation of R&D activities. Findings from this study have important policy implications. First, globalisation of R&D may provide an
opportunity for developing countries to catch up on the technology
frontier. MNEs can become an embedded driver of knowledge-based
development. However, the quality and type of FDI are important for a
signicant innovation promotion effect of FDI. Technology-intensive,
R&D-related FDI may play an important role in this process. FDI into
the coastal regions of China in the twenty-rst century with greater
participation of global leading MNEs has been much different from the
labour-intensity, export-oriented FDI in the early stage. Second,
enhancement of local absorptive capacity has been crucial to the effective assimilation of the knowledge and technology spillovers from FDI.
Finally, the development of complementary assets in the local innovation system has also been crucial for greater success in assimilation and
development of external knowledge. Therefore, a role remains for
government policies in promoting these necessary conditions for a
successful FDI-assisted technological advancement.
Evidence from the current study, however, provides little information
on whether FDI has promoted the indigenous innovation capability of
the developing countries. According to a report by Zhou (2006), most
of the foreign R&D labs in China are independent and wholly foreign
owned for better protection of intellectual property rights. More than
90 per cent of these MNEs labs do not apply for patents to avoid
disclosure of technology know-how. About 80 per cent of the foreign
rms do not have any plan to collaborate with indigenous labs or rms
or universities. Moreover, these foreign-owned R&D labs are again
unevenly distributed geographically, with Beijing, Shanghai and
Shenzhen accounting for nearly 84 per cent of total foreign R&D labs
in China. A study by Fu et al. (2006b) based on 12 British enterprises
operating in East Asia reveals that rms intend to use technology that is
superior to that at the local level rather than world-class technology in
their subsidiary or joint ventures in China. One of the main reasons is
their concern about the poor intellectual property rights protection in
China. All this raised concerns that these active foreign R&D rms may
remain as isolated innovation poles in the developing economy. Though
they have increased the overall innovation inputs and outputs in the
73
developing economy, their benets to the technological capability building of the indigenous sector remain to be seen. This may also give rise to
another form of brain drain, in which intelligent research staff members
have been lost from the domestic sector to the foreign sector, although
they stay physically in the country. Evidence from this study also
suggests that the lack of effective protection of intellectual property
rights has been a signicant barrier preventing MNEs from introducing
more advanced technology into China and investing more in innovation. Another important step for the governments of developing countries is to strengthen the protection of intellectual property rights to
attract more technology-intensive FDI and to encourage innovation.
Processing trade, FDI and international

competitiveness of the Chinese
high-technology industries
4.1 Introduction
The fragmentation and segmentation of production at the global level
has been one of the major trends driven by multinational enterprises
(MNEs). Encouragement of the processing trade, especially foreign
direct investment (FDI) in export-oriented processing activities in
export processing zones (EPZs), has become a policy tool widely utilised
by developing countries to enter the global production network and
build up international competitiveness. The number of EPZs increased
from 79 in 1975 to more than 3,500 in 2006 (ILO, 2003).
The literature on the impact of FDI on local rms has concerned itself
mainly with ordinary FDI, which engages with substantial production
activities using materials from domestic or international markets. The
evidence is mixed, in general. Some nd positive export spillovers from
FDI; some nd no or negative impact (e.g., Aitken, Hanson and
Harrison, 1997; Kokko, Transini and Zejan, 2001; Greenaway,
Upward and Wright, 2004; Buck et al., 2007; Kneller and Pisu, 2007).
In reality, different factors are at work and the relative magnitudes of
these channels depend on host country conditions and the type of FDI
inows (Javorcik, 2008). So far, little is known about the effects of
processing tradeFDI (PT-FDI), a type of FDI engaged in exportoriented processing trade activities aiming to take advantage of the
low-cost unskilled or semi-skilled labour in developing countries and
the preferential policies provided in the EPZs. This type of FDI involves
mainly importing duty-free materials and spare parts, assembling or
processing them into nal products and exporting them to the international markets. The literature on EPZs has mostly focused on employment gains. The study of the impact of EPZs on the development of
international competitiveness of local rms is limited, with the
74
Processing trade, FDI and Chinese high-technology industries
75
exception of a few case studies. An in-depth understanding and a

rigorous evaluation of the impact of the PT-FDI on local rms are
crucial for policy making. Is PT-FDI an effective channel for the development of the global competitiveness of indigenous rms? Has such
FDI helped indigenous rms gain international competitiveness in
technology-intensive industries in developing countries? What can the
indigenous rms learn from PT-FDI? In industries where the processing
trade is pervasive, what is the impact of PT-FDI on local rms export
market participation? Research on these questions is limited.
Using rm-level panel data from China, this chapter examines
these research questions in Chinas high-technology industries. The
technology-intensive industries in China have signicant involvement
with PT-FDI. Exports resulting from the processing trade accounted for
about 85 to 90 per cent of Chinas total exports of high-technology
products (Ma and Assche, 2010).1 This chapter contributes to the literature in several ways. First, it seeks to ll the gap in the literature on the
impact of processing tradeFDI on the international competitiveness of
local rms through a systematic theoretical and empirical analysis.
Second, it is one of the rst major empirical examinations to have a
special focus on processing trade. Recent research by Yu (2010) examined the effect of engagement with the processing trade on productivity
growth in a sample of 31,000 foreign invested and local rms in China in
all manufacturing industries and found that rms (both foreign and local)
engaged in the processing trade experienced higher productivity growth.
By contrast, this research focuses on the spillover effects of PT-FDI on the
export performance of local rms in high-technology industries. Third,
unlike most of the existing literature, this chapter constructs the FDI
knowledge pools at a more disaggregated level, in order to distinguish
the different channels at work. It also takes into account the time lag that
is needed for knowledge spillover to take effect.
The chapter is organised as follows: Section 4.2 reviews the literature
and presents a theoretical framework for the research. Section 4.3
briey reviews the background of the processing trade and exports of
high-technology products in China. Sections 4.4 and 4.5 discuss the
methodology and data. Sections 4.6 and 4.7 present the results of the
estimations together with robustness checks. Finally, Section 4.8 presents
conclusions.
1
Dened as products in industries with a high ratio of R&D expenditure to sales.
76
4.2 The theory and literature

Knowledge spillovers from FDI occur when multinationals nd it difcult to protect the leakage of rm-specic assets, such as superior
production technology, management practices or marketing information to local rms (Caves, 1996). As a result, FDI may affect the export
competitiveness of domestic rms in two ways. One of the channels is
technology spillovers, which could strengthen the core competitiveness
of the local rms through demonstration effects and labour movement.
Spillovers of advanced technology may contribute to the international
competitiveness of local rms that have strong absorptive capacity. Of
course, an important precondition for effective spillovers is that the
production activities in the FIEs should be technology or skill intensive
(Sjoholm, 1999). There may also be vertical spillovers within the supply
chain when MNEs provide better intermediate materials for their producer customers or when they transfer knowledge to their suppliers for
better quality nal products. Moreover, FDI is expected to help a host
country to create assets (comparative advantage) in certain areas and
facilitate the transition of indigenous rms from FDI recipient to exporters (Dunning et al., 2001).
The other channel is export-related information spillovers. This refers
to the leakage or spillover of information relating to export market
intelligence, international marketing know-how and export operations
from foreign to domestic rms. Such information spillovers may also
take place through demonstration effects and labour movement (Aitken
et al. 1997). Exporting involves sunk costs (Melitz, 2003; Bernard and
Jensen, 2004). This includes the establishment of distribution and logistics channels, product compliance with regulations, market research to
acquire information about consumer preferences and market structure
in foreign countries. Moreover, because of information asymmetries,
the costs of export market entry are perceived by some domestic rms to
be too high. This would discourage export market participation
(Greenaway and Kneller, 2004). Such xed costs associated with the
entry into export markets are signicant for rms in differentiated
product industries (Roberts and Tybout, 1997). Export-related information spillovers can effectively lower the sunk costs associated with
export market entry (Aitken et al., 1997) so that the marginal rm nds
it protable to start exporting. It may also allow existing exporters to
have a better understanding of their existing export market as well as
77
potential new markets. Therefore, the possible spillovers of technological knowledge and export-related information may enhance export
competitiveness and export market participation of local rms.
Moreover, FDI may also generate a competition effect on the exports
of local rms. Such competition effects can serve as a double-edged
sword. On the one hand, strong foreign competition may compel local
rms to strengthen and improve their competitiveness for survival (e.g.,
Liu, 2010). On the other hand, there may be a signicant depressive
effect on the productivity of local rms when foreign invested rms
crowd domestic rms from the product market and compete with them
in the labour and resources markets (Aitken and Harrison, 1999; Hu
and Jefferson, 2002; Hu, Jefferson and Qian 2005; Fu and Gong, 2011).
The empirical literature on export spillovers from foreign rms is
relatively smaller. Using rm-level data from Mexico, Aitken et al.
(1997) nd the probability of exporting by domestic rms increases
with the concentration of exports of foreign rms operating in the same
industry and region. Kokko et al. (2001) also nd that foreign rms
positively affect the probability of domestic rms exporting using rmlevel data from Uruguay. However, Barrrios, Grg and Strobl (2003)
nd no evidence of an effect from MNEs on the export share in Spain.
Similarly, Ruane and Sutherland (2005) nd negative export spillover
effects in Ireland probably because of the use of Ireland as an export
platform to the rest of the EU. They argue that export spillovers are
unlikely in this case because competition with domestic rms in local
product markets is limited. In the case of China, Buckley et al. (2002)
suggest that FDI from Hong Kong, Macao and Taiwan is more likely to
transfer marketing skills to the local rms based on evidence from
industry-level data. Buck et al. (2007) test the mechanisms underpinning the Trade Development Path hypothesis using rm-level panel data
of Chinese rms in nine Standard Industrial Classication (SIC) twodigit industries for the period 19982001. They use the share of foreign
workers in total employment, output, exports and R&D investment as
the measure of FDI spillovers and nd evidence in support of such
spillovers. These studies provide useful insights, but none of them
explores whether there is any difference between ordinary FDI and
processing trade-FDI. The industry and regional pool of FDI presence
is normally constructed at aggregate SIC two-digit and province levels.
The selection bias arising from rms export decisions has not been
appropriately corrected either.
78
Spillovers from processing trade-FDI and export

performance of domestic rms
Processing export activities include processing or assembly with
imported materials and processing or assembly with supplied materials.
The processing rms process duty-free materials and components supplied by foreign rms and export nished products. They are paid a fee
for the assembly or processing activity. The foreign rms control both
the supply of the materials as well as the entire international marketing
of the processed or assembled products. In the case of processing or
assembly with imported materials, the processing rms import, free of
customs duty, parts and components that are used to produce nished
goods and export them to international markets. The production activity related to processing trade is often semi-skilled or unskilled-labour
intensive with low value added. This is true even in technology-intensive
industries such as electronics. For example, most of the latest iPhones
sold globally are assembled in China. The materials and components
are produced in South Korea, the United States, Germany, Italy, France,
Japan and other countries. For each iPhone that is sold at the retail price
of $600, the assembly fee earned by the China-based processing MNE is
only US$6.54 per iPhone.2
As discussed earlier, knowledge spillovers from FDI may take place
through different channels. The strength of the spillovers is determined
by the type of FDI inows as well as host country conditions (Javorcik,
2008). As for PT-FDI, although some of the transmission channels
through which spillovers take effect remain available, the strength of
the spillovers is likely to be low because of several serious constraints.
First, there is little scope for technology transfer because the activity that
is undertaken in processing trade in most developing countries is in fact
low-skilled labour intensive. The core technologies are controlled by the
MNEs and are often embedded in the imported components or the
imported production lines. This constraint will be more signicant
when technology becomes more precise and when core technologies
are increasingly embedded in imported components. As a result,
although the demonstration effect and personnel movement mechanisms are still available, little advanced technology is available to be
2
Source: Supply chain for iPhone highlights costs in China, New York Times, 5 July
2010. Available at www.nytimes.com/2010/07/06/technology/06iphone.html.
79
transferred via demonstration. The level of technical skills of assemblyline workers is low and provides limited scope of technology spillover
through labour movement.
Second, technology spillovers taking place in the supply chain via
backward linkages are found to be the strongest channel (Javorcik,
2004). However, given the two-end outside nature of processing trade
production activity, that is, both materials and markets of the products
are outside the country, the linkages between PT-FDI and domestic rms
are low in general. A range of 3 per cent to 9 per cent of inputs purchased
domestically is often observed in developing countries (Milberg, 2007).
Koopman et al. (2008) nd that processing exports from China contain
only 18 per cent Chinese content, whereas those for ordinary trade were
as high as 89 per cent. Moreover, products under processing trade cannot
be sold to other domestic rms. Therefore, vertical technology spillovers
from PT-FDI will be limited too.
In respect to export information spillovers, FDI may generate positive
export-related information spillovers to domestic rms through demonstration effects, movement of labour from foreign to local rms and
other social learning channels. Domestic rms will thus obtain information about international demand, foreign regulations of their products
and sophisticated export operations and logistics. Such export-related
information spillovers may help domestic rms reduce sunk costs and
become capable of entering export markets. In the case of PT-FDI, these
transmission mechanisms are still at work. Moreover, in industries
where processing trade is prevailing, sunk costs for export market
entry are low for domestic rms, as foreign customers will provide
materials, components and product design and take care of marketing
and sales. As a result, the export-related information spillovers will
remain to have a signicant effect on export participation of domestic
rms, especially in industries where processing trade prevails.
Finally, like ordinary FDI, PT-FDI will also generate a competition
effect on domestic rms. However, in contrast to domestic rms in
developing countries, PT-FDI has an advantage created through integration of advanced technology, low-cost labour and established international marketing channels. MNEs also have access to higher-quality
components, materials and other intermediate inputs than the domestic
rms do. Therefore, MNEs engaged in processing-trade activities will
enjoy a clear competitive advantage over the indigenous rms in international markets. As a result, PT-FDI is likely to exert a strong negative
80
competition effect on indigenous rms and crowd them out of the

export market.
It should be noted that rms engaging in processing trade are different
from original equipment manufacturers (OEMs). An OEMs products
are marketed under the purchasing companys brand name. Normally,
an OEM has strong technological and production capacities, although it
lacks compatible capabilities in design, marketing and branding. Some
OEMs have successfully upgraded to original design manufacturers
(ODMs) and eventually to original brand manufacturers (OBMs).
However, OEM in high-technology industries such as consumer electronics is conned to a few new industrialised economies (NIEs) with
strong local capabilities, mainly in Korea and Taiwan (Hobday, 1995;
Lall, 2001). In contrast, the development of the electronics industries in
the Little Asian Tigers such as Indonesia, Malaysia, the Philippines and
Thailand was mainly driven by PT-FDI following the relocation of the
labour-intensive assembly activities from the old NIEs. The upgrading of
technological capabilities in indigenous rms in these Little Tigers was
far less successful than that in Korea and Taiwan (Lall, 1996).
In sum, the knowledge that spills over from PT-MNEs to indigenous
rms is mainly in the form of marketing intelligence and management
techniques. Although there will be some spillovers of basic technology
through demonstration or the movement of assembly-line workers or
technicians, the scope of technology spillover in the processing activity is
limited within the development of export competitiveness in the indigenous sector. In contrast, the crowding-out effect of MNEs on the domestic rms in the export market is strong. Therefore, we have the following
hypotheses:
Hypothesis 41: Processing tradeFDI is likely to have a positive
export information spillover effect on the export
performance of indigenous rms.
Hypothesis 42: Processing tradeFDI is likely to have an insignicant
or even negative technology spillover effect on the
export performance of indigenous rms.
4.3 FDI and exports in technology-intensive industries

in China
Since China embarked on economic reforms in 1978, its economy has
been gradually opened up to foreign trade and investment. As part of its
81
Share of processing exports in China's total high-technology exports

95
90
Percentage
85
80
75
70
65
60
55
50
1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006
Figure 4.1 Share of processing exports in Chinas total high-technology exports

Source: Data from 2000 to 2006 based on authors own estimation; data from
1992 to 1999 are reproduced from Ma and Van Assche (2010) and Koopman,
et al. (2008).
policy to promote integration into the global production network, the

government encouraged export-oriented FDI, which mainly engaged in
processing and assembling spare parts and components for the export
markets. Export-oriented FDI was encouraged by scal and nancial
incentives, for example, tax holidays and tax rebates for exports. As a
result, exports under the export-processing (EP) regime grew much
faster than that under the ordinary-trade (OT) regime as a result of
the involvement of FDI, the preferential government policy and the
lower transaction costs of the EP regime. The average annual growth
rate of processing exports was as high as 18 per cent in the 1990s, while
that for ordinary exports was only 10 per cent (Fu, 2004). Since 1996,
exports resulting from the processing trade have accounted for more
than 50 per cent of Chinas total exports (MOFTEC, 2000).
This gure is much higher in the high-technology sector. Since 1996, the
share of high-technology exports accounted for by processing trade
reached 85 per cent. In 2004, the share increased to as high as 90 per
cent and shows little evidence of decline over time (Figure 4.1). In the entire
computer and other computer periphery equipment industries, the shares
of processing exports have been as high as 99 per cent over the 2000 to
2006 period. Most foreign-invested enterprises (FIEs) in the hightechnology industries in China have engaged in processing trade. As a
82
result, exports of high-technology products from China have increased

rapidly. In 2008, high-technology exports from China accounted for
13.1 per cent of the worlds total exports of high-technology products,
ranking China as the second largest high-technology exporter in the world.
4.4 Methodology
Two problems arise with the modelling of export performance. First,
because of the sunk costs of exports, some rms decide not to enter the
export market. So a number of rms have no export sales. There is,
therefore, a selection effect based on the decision to export or not. This
problem involves estimating a Heckman selection model for panel data
controlling for sample selection bias (Hsiao, 2003; Kneller and Pisu,
2007). Second, the value of the dependent variable is censored at zero.
Normal ordinary least squares estimator may produce biased estimates
because the dependent variable is not normally distributed, hence
imposing inappropriate restrictions on the residuals. Instead, a Tobit
model should be employed for the estimation.
Therefore, the Type II Tobit model (selection in censored data) is
used. It is a two-stage model incorporating two equations. The rst
equation of the model explains the propensity to export. Those rms
that reported having positive export sales are dened as exporters. The
second equation explains the export sales of a rm (if it exports). The
empirical model is as follows:
(1) Decision equation (Probit model)
zit wit vit
and
zit 1
4:1
if zit > 0;
zit 0 if zit 0:
(2) Export performance equation (Tobit model)
eit xit i it
and
eit eit
if zit 1
eit 0 if zit 0
it i it
vit i0 it0
4:2
83
t 1; ; t and i 1; ; N; and
it 1 it1 it
0 it 2 0 it1 0 it
it ~ N0; 2 ; 0 it ~ N0; 0
2
where e is the logarithm of total exports,3 z is a dummy variable for

exporting or not, w and x are vectors of explanatory variables for
export decision and export performance equations, is the standard
deviation and is the inverse Mills ratio. The components i and i0 are
unobserved individual specic random disturbances which are constant
over time, and it are idiosyncratic errors that vary across time and
individuals. Hence, the model allows unobserved heterogeneity, rst
order state dependence and serial correlation in the error components.
Admittedly, this approach has a drawback in terms of the difculty of
establishing a distribution of individual specic effects. When a rm
decides not to export (z = 0), the observed exports value e is zero. The
signicance of the presence of the selection effects is indicated by the
Rho statistics, which reects the correlation between the error terms of
the two equations ( and ). If there are signicant selection effects, the
Heckman selection model is preferred. Otherwise, we utilise the standard Tobit model.
Following the literature (e.g., Aitken et al., 1997; Buck et al., 2007;
Kneller and Pisu, 2007), the explanatory variables in w and x include a
vector of FDI spillovers variables and a vector of control variables. The
FDI spillover variables consist of technology spillovers (fditech) and
export information spillovers (fdiinf). In contrast to the existing literature, we use lagged values of the FDI spillover variables in the model to
capture the fact that it takes time for knowledge and information spillovers to take effect on receiving rms export performance. In line with
the literature, the control variables consist of rm size (fs), capital
intensity measured by the capital-labour ratio (kl), labour productivity
(yl), innovation measured by percentage of sales on account of new
products (rd) and the wage rate (wage) that reects labour skills on one
Firms whose exports values are zero are given a zero value after taking logarithms.
This does not affect the data distribution because the minimum positive value of
exports is greater than 1.
84
hand and labour costs on the other. Moreover, we also control for some
industry-level characteristics that may inuence a rms export market
entry. These include competition (comp) in the industry, innovation
intensity in the industry (rdi) and the export orientation of the industry
(expi). Including these variables in the model also allows us to control
for the possibility that foreign rms choose to locate in more technologically advanced or export-intensive industries.
As the vectors of covariates of xi and wi may be the same, a possible
problem of identication exists if this is the case. Following Kneller and
Pisu (2007), I include a lagged export variable in both export equations.
This is theoretically consistent with the models of exports that suggest
export performance is path dependent (Bernard et al., 2003; Melitz,
2003) and helps reduce the identication problem. Finally, to control
for the possible unobserved spatial covariance, a vector of regional
dummy variables (reg) is included in the model representing rms
located in the Yangtze River Delta, the Pearl River Delta, Beijing and
Tianjin municipalities, other coastal regions, Chongqing municipality
and the remainder of the inland regions. Therefore, w and x for the
export decision and export performance equations are as follows:
n
o
inf
wit fditech
it1 ; fdiit1 ; zit1 ; f sit ; kl it ; yl it ; rd it ; wageit ; compjt ; expjt ; rd jt ; regi
4:3
n
o
inf
xit fditech
it1 ; fdiit1 ; eit1 ; f sit ; klit ; ylit ; rd it ; wageit ; compjt ; expjt ; rd jt ; regi
4:4
i2j
where j is the industry that rm i belongs to. Details of the denition of
the variables are described in Table 4.1. The estimation of the export
performance equation involves an estimation of a Heckman sample
selection model with panel data. The generalised linear latent and
mixed models (GLLAMMs) developed by Skrondal and RabeHesketh (2004), which enable the correction of selection bias in panel
data, are employed for the estimation.
Moreover, there is possible endogeneity between exports and productivity. Whereas more efcient rms choose to enter the export
market, exports may also lead to higher productivity through learning
by exporting (Bernard and Jensen, 2004). Therefore, an instrumental
Table 4.1 Summary statistics of the variables

Variable Denition
Obs
z
e
pt-foex
pt-foei
npt-foex
npt-foei
yl
size
kl
rd
wage
comp
ind_exp
ind_rd
pts
108805
0.388
0.487
108805 2.027
4.865
108805
2.076
4.542
108805 0.725
4.051
108805 0.313
3.008
108805 1.246
2.706
108805
4.975
1.087
108215 83.31 157.60
108805 287.26 295.01
91965
0.090
0.257
108606
2.598
0.687
108805
0.501
0.294
108805
0.319
0.252
108805
7.454
9.321
108807
0.597
0.333
Export dummy, equals 1 for exporters

Ln (export value)
Export-related information spillovers of PT-FDI
Technology spillovers of PT-FDI
Export-related information spillovers of non-PT-FDI
Technology spillovers of non-PT-FDI
Labour productivity
Ln (no of employees)
Capital to labour ratio
Innovation measured by per cent of sales of new products in total sales
Ln (wage rate)
Hirschman concentration index for the industry
Export-sales ratio of the industry
Percentage of new sales of the industry
Share of processing exports in total exports of the industry
Mean
Std.
Min
Max
0.000
1.000
5.869
12.295
5.869
13.057
5.869
10.563
5.769
8.922
5.769
8.041
0.000
11.183
0.000 4981.60
15.57 1806.86
0.000
23.347
5.030
10.177
0.000
1.000
0.000
0.990
0.000
41.636
0.002
1.000
86
variables approach should be used for the estimation. Productivity of

rms is hence instrumented using a vector of instrumental variables.
The instruments used include (1) the lagged dependent (export) and
endogenous variables (productivity); (2) variables indicating the ownership structure of the rms, in particular the share of state and private
ownership in a rm; and (3) all the remaining exogenous variables in the
model. Given the drawbacks in governance of state ownership in the
Chinese rms, it can reasonably be argued that a rm will be less
productive if it has a higher proportion of assets owned by the state.
The appropriateness of the instrumental variable candidates is also
carefully tested using the Sargan test for over-identifying restrictions.
Reassuringly, it was found that the instruments are valid and appropriate. We formally test whether the assumption of endogeneity is borne
out by the data at hand. If the regressors are not endogenous after all,
the standard Tobit model would be the most efcient estimation
method to use. The Davidson-Mackinnon test of exogeneity suggests
that the endogeneity problem is not statistically signicant at the 1 per
cent level (p-value = 0.035). Therefore, the standard Tobit model estimates are preferred. Nevertheless, we report both the standard Tobit
and the IV estimates as a robustness check.
4.5 Data and measurement

This chapter uses a highly disaggregated large panel data set compiled
by linking a large rm-level production and export data set with data
derived from a large product-level trade data set.
Firm-level production and export data

The rm-level panel data set of Chinese technology-intensive industries
from 2000 to2007 was drawn from the Annual Report of Industrial
Enterprise Statistics compiled by the National Bureau of Statistics of
China, covering all state-owned rms and other types of rms with
annual turnover of more than RMB 5 million (US$0.7 million) in the
technology-intensive industries. The data include rms of both domestic
and foreign ownership. Foreign rms are dened as those with 25 per
cent or more foreign equity investment according to the Joint Venture
Law of the Peoples Republic of China. Following Eurostats denition
of high-technology industries (Eurostat, 2004), rms in industries
87
involved in the manufacture of medical and pharmaceutical products

(SIC 27), electronic products and communication equipment (SIC 40)
and instrument, stationery and ofce equipment (SIC 41) are included
in the sample. These include 52 industries at the SIC four-digit level.4
An unbalanced panel is used to allow for rm entry and exit to
capture the industrial dynamics over the sample period. Data cleaning
has been carried out to exclude the outliers and unreasonable data
entries. The nal data set consists of 108,930 observations from
53,981 rms, including 70,587 observations of domestic rms and
38,343 observations of foreign rms. Overall, 63 per cent of the rms
are domestic, 21 per cent are wholly foreign owned (WFO) subsidiaries
and 15 per cent are joint ventures (JVs). In the electronics industry,
foreign rms accounted for almost 50 per cent of the total number of
rms, and most of them (32 per cent) are WFO subsidiaries. The
share of foreign rms is the lowest in the pharmaceutical industry.
Only 18 per cent of the rms are foreign owned, and most of them
(13 per cent) are joint ventures. In the instruments industries, the
proportion of foreign invested rms is 32 per cent.
Product-level trade data

The product-level trade data were drawn from the shipment records
compiled by the General Administration of Customs of China. The data
records detailed trading information at Harmonised System (HS) eightdigit level for each shipment for the years 2000, 2002, 2004 and 2006.
In total, 68.52 million shipments were made in the sample period. We
rst aggregate the export value by trade mode for each product in each
year and then estimate the proportion of processing exports in total
exports in each product sector in each year. The share of processing
exports in the years 2001, 2003 and 2005 is estimated by interpolation,
and that for 2007 is calculated using extrapolation. The HS eight-digit
product-level trade data are then converted into SIC four-digit industrylevel export data. Finally, these industry-level processing trade data are
linked to the rm-level data according to each rms SIC four-digit
industry classication.
4
The industry classication changed from GB1994 to GB2002 in 2003. So for

years before 2003, all the SIC codes are recoded (00, 01, 02) at the SIC four-digit
level.
88
Measurement
Export performance of the rms can be measured by the value of
exports or the export-sales ratio. There are some differences between
the two measurements. The total value of exports reects a rms
strength of export competitiveness; export to sales ratio reects the
export market orientation of a rm. As the main concern of this chapter
is the international competitiveness of domestic rms, the value of
exports is used to measure the export competitiveness of rms.
Spillovers from PT-FDI are proxied using two variables that represent FDI spillovers in two different forms, that is, spillovers of technology and spillovers of export information. We rst construct the overall
FDI knowledge or information pool following the normal practice and
then estimate the PT-FDI spillovers variable by adjusting the overall FDI
spillovers variable with the processing-trade intensity in each industry
for each year. We also estimate the non-processing, ordinary FDI spillovers variable and include this in the regression model to control for
possible spillovers from other FDI sources.
Spillover effects from FDI are examined empirically by testing
whether the performance of domestic rms changes as a result of the
greater foreign presence in the same industry or region. Foreign presence is often measured in terms of the output, employment and assets of
foreign rms in an industry or region. Given the fact that foreign rms
may have different strengths, some literature goes further to distinguish
FDI spillovers in specic areas such as technology or export information. The recent literature on foreign knowledge spillovers constructs
the foreign knowledge pool using the total R&D investment or total
number of patents of foreign rms (Kneller and Stevens, 2006; Singh,
2007). Given the focus on technological knowledge and export-related
information spillovers from FDI in this chapter, the latter approach is
more appropriate for the measurement of the FDI knowledge and
information pool.
Technology spillovers from FDI are often measured by the total value
of R&D capital in FIEs or the share of FIE R&D investment in total
R&D investment in the same industry or region (e.g., Buck et al., 2007;
Fu and Gong, 2011). Admittedly, R&D investment is only one of the
major inputs in innovation and knowledge creation. Moreover, MNEs
in developing countries, especially those engaged in processing trade,
often carry out R&D at their headquarters and invest little in R&D in
89
their afliates that specialise in assembly. Therefore, sales from new

products offer a better measure of the overall technology advancement
in the MNEs. The total value of sales from new products by FIEs in the
same industry and the same region is used to proxy the technological
knowledge pool provided by FDI.
Similarly, in the estimation of export-related information spillovers,
the pool of export information in FIEs is proxied by the total value of
FIE exports in the same industry or region (e.g., Aitken et al., 1997;
Kneller and Pisu, 2007; Iwasaki et al., 2010). It is expected that the
larger the export-related information pool, the greater the corresponding information spillovers from foreign to domestic rms. Admittedly,
export volumes are not only a result of marketing information but also
reect the technological advancement of a rm. The correlation coefcient of the technology and the export-related information spillovers
variables is 0.416, indicating a certain level of overlap. By controlling
for the possible technology spillovers effect in the regression, the estimated coefcient of the export information spillover variable should, to
a considerable extent, reect the net export information spillovers effect
from FDI. Nevertheless, we should bear in mind the limitations in
making precise measurement of spillovers and observe caution when
drawing conclusions.
Based on these estimates of the export information pool and technological knowledge pool, PT-FDI spillover variables are estimated by
adjusting these knowledge pool estimates with the FIE processing
exports intensity in each SIC four-digit industry. Spillovers of nonprocessing FDI are estimated by adjusting the FDI knowledge pool
with non-processing trade share in the same industry. Most of the
existing literature constructs FDI knowledge pool at SIC two- or
three-digit level for industry classication and provincial level for geographic division. Unlike these methods, in this chapter the FDI knowledge pool is constructed at a more disaggregated level: the SIC four-digit
level for industry classication and broad-city level, which includes a
city and its surrounding rural area, for regional classication.5 The
advantage of such disaggregation is, rst, that technologies, especially
in the technology-intensive industries, are specialised and industry specic. Second, knowledge spillovers are geographically localised (Jaffe,
Trajtenberg and Henderson, 1993; Austreche and Feldman, 1996).
5
A city and its surrounding area are identied using telephone area codes.
90
Knowledge and information may ow more easily among agents

located within the same area because of social bonds that foster reciprocal trust and frequent face-to-face contacts (Breschi and Lissoni,
2001). Evidence for the United States (Acs, Luc and Varga, 2002) and
Europe (Funk and Niebuhr, 2000) nds a signicant decline in spillovers beyond a range of 5075 miles and 120 kilometres, respectively.
Many provinces in China are larger than most of the countries in
Europe. Aggregation at the provincial level is hence too wide for effective knowledge spillovers in the standard sense. Of course, the construction of knowledge and the information pool at this disaggregated
level may, on the other hand, restrict the scope of spillovers when rms
observe new product ideas in markets that are not their local markets.
We therefore estimate the models using alternative spillovers measurements at provincial level as a robustness check. Table 4.1 reports the
summary statistics of the variables used in the test.
4.6 Results
Table 4.2 reports the main export and innovation performance indicators of domestic and foreign rms in technology-intensive industries
over the sample period. In all industries, the proportion of rms participating in exporting is higher amongst foreign than domestic rms.
Foreign rms also have higher export-to-sales ratios as well as higher
value of exports per rm. On average, the electronics industry has the
highest share of exporters among the foreign rms. About 75 per cent of
the foreign rms in this industry engaged in export activity, which is
three times as high as that in the indigenous rms. The average exportsales ratio in this industry is 54 per cent, which is about ve times as
high as that in the indigenous rms. Looking at changes over time, the
propensity of export participation of foreign and indigenous rms did
not change much from 2000 to 2007. Not surprisingly, the difference
between foreign and indigenous rms is much smaller in respect to
technology and innovation measured by the percentage of sales of
new products. In the electronics and instruments industries, this ratio
is even higher in indigenous rms. Most of these high-technology
exports result from processing trade. The share of processing exports
in total exports ranged between 85 to 90 per cent from 2000 to 2006. In
the electronics and communications equipment sector, this proportion
had consistently been around 89 per cent except in the communication
Table 4.2 Exports and innovation indicators of domestic and foreign rms, 2007
Domestic
Foreign
Industry Percentage
Export-sales
Percentage Percentage Export-sales
Percentage Percentage
SIC
of total exports ratio
Ln(exports) of exporters of new sales ratio
Ln(exports) of exporters of new sales
27
40
41
Total
4
87
9
100
5.41
11.62
12.74
9.42
1.47
2.46
2.26
2.03
0.17
0.27
0.26
0.23
0.08
0.12
0.12
0.10
16.41
53.53
48.61
47.21
3.30
7.73
6.70
6.91
0.36
0.75
0.69
0.68
0.09
0.07
0.06
0.07
92
Percentage of processing exports in total exports, electronics industry of China, 2006

Electronic components and their assembly
Computer peripheral equip.
Household video equip.
Integrated circuit devices
Printed circuit board
Optical devices and other electronic devices
Mobile communication and terminal equip.
Entire computer
Household audio equip.
Semiconductor discrete devices
Other electronic equip.
Vacuum electronic devices
Communication terminal equip.
Communication transmitting equip.
App. of television equipment & other auxiliary television equip.
Equipment. for receiving TV programmes
Other communication equip.
Communication exchanging equip.
Radar and auxiliary equip.
Equipment. for producing and transmitting TV programmes
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Figure 4.2 Percentage of processing exports in total exports, electronic industry

of China, 2006
Note: The sectors are placed according to the volume of exports, from largest on
the top to smallest at the bottom.
Source: Authors estimation according to customs data.
transmitting equipment industry. In sectors such as computer, computer

peripheral equipment, computer network and printed circuit board
industries, the share of processing exports in total exports was still as
high as 99 per cent in 2006 (Figure 4.2). On average, the electronics
industry accounted for 87 per cent of Chinas total high-technology
exports from 2000 to 2007.
On average, foreign invested rms produce 89 per cent of Chinas total
exports of high-technology products. This indigenous foreign division
has been relatively stable over the sample period. Disaggregating the
industries at the four-digit level, Table 4.3 shows that the manufacture
of electronic components and their assembly (SIC 4061) has the largest
export value, followed by computer peripheral equipment manufacturing
(SIC 4043). In SIC 4061, around 80 per cent of the WFO rms engaged
in exporting. Foreign rms contributed to 88 per cent of Chinas total
exports in this industry, and 84 per cent of these exports are processing
exports. In the computer industry (SIC 4041, SIC 4042 and SIC 4043),
almost all of these exports (99 per cent) were on account of processing
trade in 2006. Moreover, again almost all of these exports were produced
in FIEs. The share of FIEs exports in total exports was as high as 97 per
cent, 98 per cent and 94 per cent in these computer industries (SIC 4041,
Table 4.3 Export performance of top 10 exporting high-technology industries: SIC four-digit level, 2007
Percent-age
of processing trade
Percentage of
total exports
Export-sales ratio
(percentage)
Percentage of
exporters
SIC
Industry
2006
WFO JV Local WFO JV Local WFO JV Local
4061
4043
4071
4053
4062
4059
4014
4041
4072
4052
Total
Electronic components and their assembly

Computer peripheral equip.
Household video equip.
Integrated circuit devices
Printed circuit board
Optical devices and other electronic devices
Mobile communication and terminal equip.
Entire computer
Household audio equip.
Semiconductor discrete devices
All high-technology industries
84
99
98
85
93
82
84
99
99
87
88
63
80
41
69
78
68
71
90
61
73
63
25
14
42
26
18
22
23
7
28
19
23
12
6
17
5
4
10
6
3
11
8
14
58
66
70
65
64
62
46
59
81
67
59
42
39
47
48
41
35
30
19
67
37
31
13
12
23
8
7
12
8
3
27
10
9
78
79
85
82
81
82
72
72
90
87
77
73
66
71
74
73
68
56
43
84
73
57
32
27
38
23
22
31
22
13
40
33
22
94
4042 and 4043), respectively. All these provide stylised examples of PTFDI and suggest that most of the high-technology exports from China are
in fact exports of PT-cum-MNEs. High-technology exports from indigenous Chinese rms are very limited.
On the other hand, there are substantial variations in the speed of
catch-up across industries. Indigenous rms in a few sectors have shown
rapid export growth, including manufacturers of integrated circuit
devices, vacuum electronic devices, industrial automatic control system,
optical devices and mobile communication and terminal equipment. In
these sectors, the indigenous rms have shown a faster export growth
than FIEs and a catch-up trajectory. However, despite the high growth
rate in the indigenous rms, partly because of a low starting level and
strong government-funded R&D programs, strong foreign dominance
in the exports of these industries is clear.
Table 4.4 reports the Probit model estimates of the spillover effects of
PT-FDI on the propensity to export in the indigenous rms. Both the
lagged and contemporaneous effects of PT-FDI are reported for comparison. Both the standard Probit and IV estimates are reported as a
robustness check. As columns (1) to (4) show, the estimated coefcient
of the lagged information spillover variable of PT-FDI is positive and
signicant at the 1 per cent level, suggesting positive export-related
information externalities emanating from PT-FDI to the indigenous
rms. The estimated results are consistent in both the baseline model
and the full model when non-processing trade or ordinary FDI is controlled for. The results are also consistent in the standard Probit and the
IV model indicating the robustness of the results. However, the estimated coefcients of technology spillovers of PT-FDI are negative and
statistically signicant at the 5 per cent level.
In other words, while the limited innovation activities in the foreign
afliates in China offer limited knowledge spillovers, the overall
technological and innovation capabilities of the MNEs have posed a
barrier to the indigenous rms in their export propensity. The results
are also robust across different model specications and different
estimation methods. All this evidence supports hypotheses 41 and
42 suggesting that, on the one hand, PT-FDI has offered positive
marketing and export-related information spillovers to local rms;
on the other hand, its overall advantage in technology, mostly
embodied in imported components and materials and coupled with
low labour costs gained through processing trade, has formed a
Table 4.4 Probit model estimates of export decisions in the high-technology industry
Lagged effect
Variables
pt_forx
pt_fori
Normal
(1)
IV
(2)
Normal
(3)
IV
(4)
Normal
(5)
IV
(6)
Normal
(7)
IV
(8)
0.009***
(0.003)
0.008**
(0.004)
0.009**
(0.003)
0.008**
(0.004)
0.0145***
(0.003)
0.010***
(0.003)
2.472***
(0.023)
0.0001
(0.0002)
0.241***
(0.020)
0.0001
(0.0001)
0.228***
(0.031)
0.012
(0.028)
0.008**
(0.003)
0.007**
(0.004)
0.013***
(0.005)
0.006
(0.005)
2.468***
(0.023)
0.0001
(0.0002)
0.241***
(0.020)
0.0001
(0.0001)
0.229***
(0.031)
0.008
(0.028)
0.0151***
(0.003)
0.010***
(0.003)
2.474***
(0.023)
0.0002***
(4.3E05)
0.230***
(0.012)
0.0001
(0.0001)
0.229***
(0.031)
0.020
(0.020)
0.008**
(0.003)
0.008**
(0.004)
0.013***
(0.005)
0.005
(0.005)
2.471***
(0.023)
0.0002***
(4.3E05)
0.229***
(0.012)
0.0001
(0.0001)
0.230***
(0.031)
0.012
(0.020)
2.437***
(0.0)20
0.0001***
(3.7E)05
0.234***
(0.01)0
0.0001
(0.000)1
0.259***
(0.029)
0.024
(0.017)
2.436***
(0.020)
0.0001
(0.0001)
0.242***
(0.018)
0.0001
(0.0001)
0.259***
(0.029)
0.018
(0.024)
0.0106***
(0.003)
0.007**
(0.003)
0.021***
(0.004)
0.003
(0.004)
2.433***
(0.020)
0.0001***
(3.7E05)
0.235***
(0.010)
0.0001
(0.0001)
0.260***
(0.029)
0.013
(0.017)
0.0100***
(0.004)
0.007**
(0.003)
0.021***
(0.004)
0.004
(0.004)
2.433***
(0.020)
0.0001
(0.0001)
0.241***
(0.018)
0.0001
(0.0001)
0.260***
(0.029)
0.009
(0.024)
npt_forx
npt_fori
z_lag
yl
size
kl
rd
wage
Contemporaneous effect
Table 4.4 (cont.)

Lagged effect
Contemporaneous effect
Normal
(1)
IV
(2)
Normal
(3)
IV
(4)
Normal
(5)
IV
(6)
Normal
(7)
IV
(8)
Region dummies
Year dummies
0.019
(0.037)
0.578***
(0.054)
0.002
(0.001)
3.143***
(0.082)
Y
Y
0.018
(0.037)
0.564***
(0.054)
0.002*
(0.001)
3.187***
(0.111)
Y
Y
0.018
(0.037)
0.551***
(0.056)
0.002
(0.001)
3.090***
(0.085)
Y
Y
0.018
(0.037)
0.543***
(0.056)
0.002*
(0.001)
3.153***
(0.114)
Y
Y
0.003
(0.032)
0.563***
(0.0470
0.003***
(0.001)
3.184***
(0.071)
Y
Y
0.003
(0.032)
0.551***
(0.047)
0.004***
(0.001)
3.206***
(0.097
Y
Y
0.004
(0.032)
0.527***
(0.048)
0.004***
(0.001)
3.110***
(0.073)
Y
Y
0.004
(0.032)
0.515***
(0.048)
0.004***
(0.001)
3.127***
(0.099)
Y
Y
N
Log-likelihood
Chi-squared
32,115
7977
14162
31,948
7946
14132
31,948
7936
14112
31,948
7942
14126
42,449
10632
18264
42,017
10554
18156
42,449
10617
18237
42,017
10540
18130
Variables
comp
Ind_exp
Ind_rd
Constant
Note: Standard errors in parentheses; *** p<0.01, ** p<0.05, * p<0.1.
97
strong competitive barrier that prevents the local rms from participating in international markets.
As expected, the export-related information spillovers of non-PTFDI have shown a positive and signicant effect on the export participation of indigenous rms. The magnitude of the estimated coefcient
is double the size of that of PT-FDI. This is consistent with our
proposition that non-PT-FDI has more linkages and interactions
with the domestic economy. As regards the technology spillovers
effect, non-PT-FDI does not appear to exert signicant technology
spillovers effects on the export market participation of indigenous
rms, which is different from the estimated effects of PT-FDI. In
other words, for non-PT-FDI, its possible positive technological spillovers facilitated via linkages with domestic rms in the domestic
market and the competition pressure in the international market offset
each other; hence, its net effect on local rms becomes insignicant on
average. Again, the estimated results are robust for both the standard
Probit and IV estimation methods.
The estimated coefcient of the past export status variable is positive
and statistically signicant at the 1 per cent level in all specications,
suggesting the importance of past export experience in rms current
export decisions. The estimated coefcient of the labour productivity
variable bears a positive sign, but it is insignicant in the IV estimation
model, although it is statistically signicant in the standard Probit
regression. As regards the remaining control variables, larger rms
appear to be more likely to enter the export market; rms with greater
innovations are more likely to export; and the wage rate does appear to
have a signicant effect on rms export propensity, which can be
explained by the fact that it reects labour skills on the one hand and
production costs on the other. The estimated coefcient of the capitallabour ratio is not statistically signicant either, probably because of the
labour-intensive nature of processing trade. Looking at the three
industry-level variables, rms in industries with greater export orientation are more likely to export. Firms in industries with more innovation
are also more likely to export. In other words, there are signicant intraindustry spillovers of indigenous innovative technology within the sectoral innovation system. Finally, domestic market competition does not
appear to have signicant impact on indigenous rms export market
participation in these industries, likely because the quality standard in
the domestic market is substantially different from that in the export
98
market, and for most rms, the export market is not an easy alternative
in the face of erce domestic competition.
The contemporaneous effect of PT-FDI on the propensity to export in
indigenous rms reported in columns 4 to 8 presents a similar picture.
The signs of the estimated coefcients are the same as those of the
lagged effects, whereas there are some slight differences in the magnitude of the estimated coefcients and the level of statistical signicance.
In general, the size and signicance of the contemporaneous information spillovers effect are greater than those of the lagged effect, whereas
those of the technological spillovers remain similar with lagged effects,
likely because of the difference in the speed of diffusion between export
information and technological knowledge. The estimated coefcients of
the control variables remain broadly consistent with the lagged effects
estimates.
Turning to the estimates of export performance measured by the
value of exports, Table 4.5 reports the Heckman selection estimates
obtained using GLLAMM that have corrected the selection bias. The
Table 4.5 Spillovers from PT-FDI and export value of domestic rms
GLLAMM
(selection in Tobit model)
Variables
pt_forx_lag
pt_fori_lag
npt_forx_lag
npt_fori_lag
e_lag
size
kl
OLSa
(random effects)
Normal
(1)
IV
(2)
Normal
(3)
IV
(4)
0.0227**
(0.010)
0.011
(0.010)
0.028**
(0.013)
0.025*
(0.015)
0.637***
(0.011)
0.487***
(0.035)
0.0001
(0.0001)
0.020**
(0.010)
0.009
(0.010)
0.027**
(0.013)
0.026*
(0.015)
0.638***
(0.011)
0.617***
(0.060)
0.0001
(0.0001)
0.012**
(0.005)
0.006
(0.005)
0.010*
(0.006)
0.008
(0.007)
0.809***
(0.007)
0.278***
(0.028)
0.0001
(0.0001)
0.011**
(0.005)
0.005
(0.005)
0.009
(0.006)
0.009
(0.007)
0.810***
(0.007)
0.299***
(0.028)
0.0001
(0.0001)
99
Table 4.5 (cont.)

GLLAMM
(selection in Tobit model)
Variables
yl
rd
wage
comp
Ind_exp
Ind_rd
Constant
Region dummies
Year dummies
N
Log-likelihood
Chi-squared
R2_between
R2_overall
OLSa
(random effects)
Normal
(1)
IV
(2)
Normal
(3)
IV
(4)
0.001***
(0.001)
0.374***
(0.097)
0.003
(0.062)
0.001
(0.104)
1.415***
(0.193)
0.009*
(0.005)
3.140***
(0.288)
Y
Y
0.002***
(0.001)
0.369***
(0.098)
0.088
(0.083)
0.006
(0.105)
1.336***
(0.194)
0.011**
(0.005)
3.430***
(0.351)
Y
Y
0.004***
(0.001)
0.351***
(0.106)
0.036
(0.028)
0.000
(0.041)
0.818***
(0.144)
0.005**
(0.002)
2.416***
(0.137)
Y
Y
0.004**
(0.001)
0.356***
(0.110)
0.032
(0.036)
0.001
(0.040)
0.795***
(0.145)
0.006**
(0.002)
2.528***
(0.132)
Y
Y
31,948
7809
5740
31,948
7831
5648
31,948
31,948
88313
0.962
0.712
93688
0.963
0.712
Note: Standard errors in parentheses. ***p<0.01, **p<0.05, *p<0.1.

a
Cluster-robust standard errors are reported for OLS estimates.
Rho-statistics indicate signicant selection effects in the full sample;

therefore, the GMMALL estimates are preferred. Consistent with our
expectations, evidence from this table shows that export-related information spillovers from FDI have a positive impact on the export performance of the local rms. A 1 per cent increase in the FIE processing
exports increases the exports of domestic rms in the same industry by
0.02 per cent. The estimated coefcients are statistically signicant at
100
the 5 per cent level. On the other hand, the export performance effect of
technological spillovers from PT-FDI bears a negative sign but is statistically insignicant. Comparing this result with the signicant negative
effect of technology spillovers of PT-FDI on domestic rms propensity
to export, the difference between the two proves a qualitative difference
between the indigenous exporters and non-exporters. Firms that have
made the breakthrough and entered the export market normally have
already reached the threshold level of competitiveness. They possess
greater capabilities than non-exporters to compete with FIEs and to
learn from FIEs through reverse engineering. However, it is difcult for
those rms that have not reached the export productivity threshold level
to imitate the technological advancement embedded in the imported
components and materials used in the FIEs.
Non-PT-FDI shows a similar signicant positive information spillover effect on the export performance of indigenous rms, similar to
that of PT-FDI. The technology spillover effect of non-PT-FDI on
indigenous rms is negative and marginally signicant at the 10 per
cent level in the Tobit model. In other words, exports from nonprocessing foreign rms that utilise locally produced material and
components impose stronger competitive pressure on the exports of
indigenous exporters partly because they have similar strength and
hence compete in narrow segments of the market. However, the statistical signicance of the estimated coefcients is not robust. Whereas
the Tobit estimates produce some signicant coefcients at the 5 and
10 per cent levels, the coefcients estimated by standard panel model are
not statistically signicant.
Factors that signicantly affect how much a domestic rm exports are
in general similar to those in the export decision regression with one
notable difference. A rms past performance in exporting still plays an
important role in shaping its current export performance. Larger rms
export more; more innovative rms export more. Again, in industries
with higher export-sales ratios and greater innovation, the export volume of individual rms is greater. The capital-labour ratio and wage
rate do not appear to have a signicant effect on the export performance
of domestic rms, probably because of similar reasons to those discussed earlier. Unlike with the results of the export decision regression, labour productivity appears to have a signicant and positive
effect on export performance. The estimated coefcient is positive and
signicant in both the selection biascorrected standard Tobit and
101
IV-instrumented Tobit estimates, suggesting a role for productivity in

the determination of export performance in rms that have already
entered the export markets.
Columns (3) and (4) of Table 4.5 report the ordinary least squares
(OLS) estimates as a robustness check, although the selection bias has
not been corrected and the left-end censored distribution of the dependent variable has not been taken into account. As the estimation is
performed at the rm level but includes industry-level variables, the
standard errors are biased downwards. Following Moulton (1990),
standard errors are clustered at the industry level to make correct
inferences. The estimated results are broadly consistent with those of
the selection biascorrected Tobit model estimates. The information
spillover effect remains positive and signicant for the PT-FDI. This
effect of non-PT-FDI is also positive but is only marginally signicant.
The estimated coefcient of the technology spillovers variable remains
negative but is not statistically signicant. The effects of past export
performance, rm size, labour productivity, in-house R&D and
industry-level export orientation and innovativeness all remain consistent with the earlier estimates.
4.7 Robustness check

Table 4.6 reports the results of the robustness check using the high
indigenous export industries subsample to determine whether PT-FDI
Table 4.6 Determinants of export performance in industries with fast
indigenous export growth: IV estimates
Export decision
Variables
pt_forx_lag
pt_fori_lag
npt_forx_lag
Export value
Probit
(1)
GLLAMM
(selection in Tobit)
(2)
0.001
(0.012)
0.0001
(0.010)
0.0577*
(0.032)
0.012
(0.034)
0.009
(0.030)
0.104
(0.101)
102
Table 4.6 (cont.)

Export decision
Variables
npt_fori_lag
z_lag
Probit
(1)
0.006
(0.032)
2.413***
(0.077)
e_lag
Export value
GLLAMM
(2)
0.066
(0.100)
Region dummies
0.0001
(0.0001)
0.346***
(0.060)
0.0001
(0.0001)
0.547***
(0.1107
0.143
(0.090)
0.025
(0.119)
0.298
(0.189)
0.003
(0.005)
3.354***
(0.357)
Y
0.613***
(0.038)
0.002
(0.001)
0.890***
(0.188)
0.0001
(0.0001)
0.809**
(0.366)
0.275
(0.277)
0.143
(0.357)
1.524***
(0.584)
0.005
(0.013)
4.970***
(1.197)
Y
N
Log(likelihood
Chi-squared
3,244
742.6
1266
3,244
696.5
506.4
yl
size
kl
rd
wage
comp
Ind_exp
Ind_rd
Constant
Note: Standard errors in parentheses. ***p<0.01, **p<0.05, *p<0.1. IV method is

used to control for the endogeneity between labour productivity and export
performance. Fast export-growth industries are those top 20 export sectors in which
the difference in export growth from 2000 to 2007 between the domestic and foreign
rms is greater than 10. These include SIC 4053, 4059, 4014, 4051, 4111 and 4119.
103
has greater and positive inuence in these industries. The subsample

includes industries that have achieved fast export growth in the domestic sector. Industries whose indigenous export growth rate over the
20002006 period is 10 times that of the FIEs are included in this
subsample. It includes more than 1,000 rms with 3,244 observations
in the integrated circuit devices, optical devices and other electronic
devices, mobile communication and terminal equipment, vacuum electronic devices, the industrial automatic control system and auxiliary
instrument and other general instrument industries. The estimated
results suggest that indigenous innovation, export information spillovers and economies of scale are the important factors that contributed
to the growth of exports in the indigenous rms. Only export information spillovers from non-PT-FDI appear to contribute signicantly to
the fast export growth in these sectors. All the remaining FDI spillover
variables show no signicant impact on the export performance of
indigenous rms. In contrast, indigenous innovation through in-house
R&D shows a positive and signicant effect on both export decision
and performance. The size of the estimated coefcients is much larger
than those for the full sample. Moreover, larger rms appear to be more
likely to export and export more.
Table 4.7 reports the robustness check results using alternative measurement of FDI spillovers. FDI information and the knowledge pool are
Table 4.7 Spillovers from PT-FDI and export performance of indigenous

rms: IV estimates of alternative spillover measurements
Variables
pt_forx_lag
pt_fori_lag
npt_forx_lag
npt_fori_lag
Export decision
Export value
Probit
(1)
GLLAMM
(2)
0.010**
(0.005)
0.002
(0.003)
0.019***
(0.005)
0.006
(0.004)
0.022
(0.015)
0.005
(0.010)
0.040**
(0.016)
0.019
(0.014)
104
Table 4.7 (cont.)
Variables
z_lag
Export decision
Export value
Probit
(1)
GLLAMM
(2)
2.428***
(0.025)
e_lag
Industry dummies
Region dummies
Year dummies
0.0003*
(0.0002)
0.269***
(0.022)
0.0001
(0.0001)
0.428***
(0.045)
0.052*
(0.029)
2.656***
(0.823)
Y
Y
Y
0.617***
(0.012)
0.0018***
(0.001)
0.668***
(0.069)
0.0001
(0.0001)
0.515***
(0.139)
0.038
(0.094)
4.281***
(1.427)
Y
Y
Y
N
Log-likelihood
Chi-squared
31868
7753
10587
24757
6085
4784
yl
size
kl
rd
wage
Constant
Note: Standard errors in parentheses. ***p<0.01, **p<0.05, *p<0.1.

FDI knowledge and information pool variables are constructed at the provincial
level.
IV method is used to control for the endogeneity between labour productivity and
export performance.
constructed at the provincial level to allow for wider export information

and technological knowledge spillovers. The general picture is broadly
consistent with the estimated results using FDI spillover variables constructed at a more disaggregated broad-city level. Consistent with
105
earlier results, FDI has generated signicant export information spillovers on the propensity of export market participation in indigenous
rms, and the strength of that effect of non-PT-FDI is stronger than that
of PT-FDI. However, unlike the estimates of city-level spillover variables, the information spillover effect of PT-FDI is insignicant at the
provincial level, conrming the argument that knowledge spillovers are
a localised phenomenon, and rms in the same cluster benet the most
from knowledge spillovers between rms. As regards the technology
spillovers from FDI, although the estimated coefcients also bear a
negative sign, which is consistent with the earlier results, they are not
statistically signicant in either model. There are two possible explanations for this change. First, technology spillovers may be transmitted to
other cities through peoples movement. Second, competition may be
stronger between rms in the same cluster because they are more likely
to compete in the same or similar segment product and factor markets.
In-house R&D, rm size and labour productivity remain the signicant
determinants of export performance of indigenous rms.
4.8 Conclusions
This chapter examines how and to what extent processing trade-FDI
inuences the international competitiveness of indigenous rms using
linked and highly disaggregated rm-level and product-level data sets
relating to production, trade mode and exports from China. Findings
show that although there had been some fast export growth in a few of
the high-technology industries, by 2007 about 90 per cent of the
technology-intensive exports from China were still the processing
exports of MNEs. Most of the indigenous rms have not built up
their international competitiveness and rely on the domestic market.
The estimated results suggest that FDI engaged in processing trade
activities offers signicant positive information spillovers to local rms
with regards to their export activities. Such spillovers have provided
domestic rms not only with foreign market intelligence but also export
and marketing techniques. However, although there will still be some
technology spillovers through demonstration and the movement of
workers, the scope for technology spillovers in such low-skilled labourintensive assembly activities is limited. Benets from PT-FDIfacilitated
technology spillovers are too small to build up export competitiveness
in the international markets, where rms have to compete with MNEs
106
that actively engage in R&D and innovation. Moreover, the competitive advantage enjoyed by the MNEs through the combination of
technological, nancial and marketing capital with low-cost labour
leads to competition with indigenous rms and crowds them out of
export markets. Such negative technology spillover effects pose a signicant barrier for non-exporters regarding export market participation. As a result, the effect of technology spillovers of PT-FDI on the
propensity of export market entry in indigenous rms is in fact negative
and signicant. However, exporters who already possess international
competitiveness are able to compete and to learn from foreign competition. The net effect of such spillovers therefore becomes insignicant.
Non-PT-FDI, on the one hand, provides slightly stronger positive
information spillovers on domestic rms; on the other hand, it still
shows a negative technology spillover effect on the export value of
indigenous exporters, although this is only marginally signicant at
the 10 per cent signicance level. The estimated results suggest that
indigenous innovation, economies of scale and export information spillovers from non-PT-FDI are the main drivers of export performance in
the indigenous rms. Labour productivity does not appear to have a
signicant role in rms export decisions, suggesting that the productivity threshold level for exporting is low in industries where processing
trade is prevalent. Labour productivity serves as a signicant factor
determining the scale of a rms exports. FDI spillover variables constructed at a more disaggregated broad-city level appear to better reect
the stronger information/knowledge spillovers that take place between
rms in the same cluster and the stronger competition effect that occurs
between rms in the same cluster compared to those taking place
between rms in different cities.
Evidence from this research has important policy implications. The
establishment of export processing zones and the encouragement of
process-tradecumFDI have been used by many developing countries
as a major policy tool to facilitate technological learning and upgrading
in indigenous rms as well as to promote job creation. Evidence from
this research suggests that although there are signicant export information spillovers, the direct technological learning effect is limited and
the benets are not provided quickly. The technology advantage of the
MNEs even poses a signicant barrier deterring the export market
participation of the indigenous rms. In the case of the technologyintensive industries in China, although some catch-up is taking place in
107
a few industries, the overall FDI spillover effects are not strong enough
to increase the international competitiveness of the indigenous rms. In
the high-technology industries, where technology becomes increasingly
sophisticated and precise, the opportunities for learning and catch-up
through labour-intensive processing-tradebased production activities
have become smaller than those experienced by Japanese and Korean
rms in the 1960s and 1970s. Complementary policy tools are needed
to provide real commitments to technology transfer and opportunities
for indigenous technological learning and local capabilities building.
Admittedly, processing-trade FDI will contribute to employment and
income growth in the host country. Income from processing trades may
be reinvested in education, which eventually will raise the skills level of
the next generation in the host country. In the long run, the host country
may gradually move up the technology ladder with improved labour
skills. Even such an indirect slow upgrading process will not come about
automatically without redistribution of the income into education and
indigenous innovation. Therefore, effective government policies and a
favourable national culture to ensure that the income from processing
trade is reinvested in skills building and indigenous innovation are
crucial in the development of technological capabilities and the international competitiveness of indigenous rms.
Indigenous and foreign innovation

efforts and technological upgrading
in China
5.1 Introduction
Technology upgrading is a key element of industrialisation in developing countries. International technology transfer through foreign
direct investment (FDI) has long been regarded as a major engine of
technology upgrading in developing countries. Many developing
countries combined competition for FDI with the expectation that
advanced technological knowledge embedded in FDI can drive technological upgrading in their countries. On the other hand, in recent
years more and more developing countries have started to question
the effectiveness of such a FDI-led technology upgrading strategy and
have called for greater emphasis on indigenous innovation as a driver
of the development of indigenous technological capabilities. It is
therefore timely to assess the following two questions: (1) what are
the major drivers of technological upgrading in developing countries,
and (2) can developing countries rely on foreign technology to catch
up with industrialised countries? Furthermore, empirical evidence on
the productivity gains from trade and FDI is mixed, and the debate on
the importance of foreign versus indigenous innovation efforts is
inconclusive.1
This chapter attempts to explore the drivers of technology upgrading
in middle-income developing countries, which often have sizeable
domestic markets, considerable human capital and a strong desire for
economic independence. It assumes that developing countries, especially middle-income countries, are not only users but also creators of
new technology in certain industrial sectors. It also takes into account
Grg and Strobl (2001) and Meyer (2004) provide excellent surveys of the relevant
literature.
108
Indigenous and foreign innovation efforts
109
the fact that the industry structure of these countries often consists of
industries of a variety of technology intensities.2
The empirical analysis is carried out using a rm-level panel data set
of 56,125 Chinese rms over the 20012005 period. China provides a
good case for this study given its huge FDI inows and its emphasis on
indigenous innovation and industry upgrading. A non-parametric frontier technique is used to decompose the total factor productivity (TFP)
growth of rms into technical change and efciency improvement. The
drivers of these changes are examined with special emphasis on the
impact of indigenous and foreign research and development (R&D)
efforts. Three types of R&D effort are considered: R&D at the rm
level, R&D in all foreign invested rms within the same industry and
region in China, and international R&D spillovers facilitated by FDI.
To test the effect of the third type of foreign R&D effort, the international industry-specic R&D stock is linked to the Chinese rm-level
data in the corresponding industry and adjusted by industry- and rmlevel FDI intensity.
This chapter is organised as follows: Section 5.2 presents a theoretical
framework for the understanding of the drivers of technological
upgrading in a middle-income developing country. Section 5.3 provides
a brief overview of FDI and innovation in China. Section 5.4 discusses
data, model and methodologies. Section 5.5 presents the empirical
results. Section 5.6 concludes with a discussion of policy implications.
5.2 Theoretical framework

Innovation is costly, risky and path dependent. Innovation activities
have therefore been largely concentrated amongst a few developed
countries. If technologies are costless to diffuse and if the effectiveness
of a technology is the same in different local contexts, developing
countries can rely on foreign technology transfer and easily catch up
with the world technology frontier without indigenous innovation.
Foreign direct investment as a bundle of technological and managerial
knowledge as well as nancial capital has long been regarded as a major
vehicle in the transfer of advanced foreign technology to developing
Technology intensity is often proxied by the ratio of R&D expenditure to value

added.
110
countries (Dunning, 1993; Lall, 2003). A further reason to expect that

FDI will lead to technology transfer derives from the fact that most of
the worlds R&D investment is concentrated amongst a few large multinational enterprises (MNEs).
FDI may contribute to technological upgrading in the host economy
in several ways. Technology spillovers from foreign invested rms may
contribute to technical change in indigenous rms. Knowledge spillovers may take place from foreign to local rms in the same industry
and region through movement of trained labour, demonstration effects
and competition effects when the competitive pressure caused by the
foreign presence forces local rms to improve their production technology and management (Caves, 1974; Fosfuri, Motta and Ronde, 2001;
Buckley, Clegg and Wang, 2002). There may also be signicant knowledge transfer within the supply chain via forward and backward linkages (Javorcik, 2004). Moreover, advanced technologies embedded in
imported machinery and equipment in foreign invested enterprises
(FIEs) may raise the average technology level of the host economy.
Multinational enterprises may also bring in advanced innovation management practices and thus improve the innovation efciency of the
local innovation system (Fu, 2008). On the other hand, the introduction
of FDI may make competing domestic rms worse off (Aitken and
Harrison, 1999) and reduce the R&D efforts of local rms (OECD,
2002). This could occur if foreign rms exploit their superior technology and marketing power to force local competitors to reduce their
outputs or if they attract the most talented researchers, which in particular might threaten local small and medium enterprises (SMEs) (Aitken
and Harrison, 1999; Hu and Jefferson, 2002; UNCTAD, 2005).
Moreover, several preconditions are needed for local rms to benet
in an effective manner from FDI spillovers. First, technology transfer via
the supply chain requires effective linkages between foreign rms and
their local suppliers and customers (Fu, 2004; Javorcik, 2004). Second,
signicant spillovers from foreign to local rms require sufcient
absorptive capacity in the local rms (Cohen and Levinthal, 1989; Fu,
2008). A threshold level of human capital is also necessary (Eaton and
Kortum, 1996; Xu, 2000). R&D activities are important as a means of
learning and accumulating absorptive capacity (Aghion and Howitt,
1992; Grifth, Redding and Van Reenen, 2003). Third, different types
of FDI have markedly different productivity spillover effects (Drifeld
and Love, 2003). Given these preconditions, it is not surprising that
111
despite the possible benets from international technology transfer,

empirical evidence in this eld is mixed.
Moreover, the need for foreign technology to be appropriate to the
specic socioeconomic and technical context of a developing country
implies that developing countries cannot rely on foreign technology for
technological upgrading and that indigenous innovation is of crucial
importance. Different technologies are specic to particular combinations
of inputs (Basu and Weil, 1998). For a particular country, an appropriate
technology is a technology tailored to t the psychosocial and biophysical
context prevailing in a particular location and period (Stewart, 1983; see
also Willoughby, 1990). Therefore, technological progress can be seen as
localised learning by doing (Atkinson and Stiglitz, 1969).
Acemoglu (2002) suggests that technologies are designed to make
optimal use of the conditions and factor supplies in the country where
the technology is developed. Most new technologies are invented and
developed in industrial countries (e.g., OECD countries), which are
abundant in skilled labour. Therefore, these new technologies often
make intensive use of skilled labour, for example, engineers, managers
and other professionals, and are usually capital augmenting or skilledlabour augmenting. Such advanced technologies might be inappropriate for conditions in developing countries and hence less productive
given the different factor endowments that the developing countries
have (Acemoglu and Zilibotti, 2001; Acemoglu, 2002). The extent of
directed technical change and the difference in factor endowment
between creator and user economies will determine how appropriate
or inappropriate a technology is with respect to the needs of the importing country.
As endowments in developing countries differ between countries, and
the demand for skilled labour varies across industries, the degree of
appropriateness of foreign technology for productivity growth in a developing country depends on the characteristics of the country and the
industry under study. As demand for unskilled and semi-skilled labour
is higher in labour-intensive industries, adoption of unskilled-labour augmenting technology will generate greater returns than the use of skilledlabour augmenting technology. In other words, rms using unskilledlabour augmenting technology will be more efcient than rms using
skilled-labour augmenting technology in low-technology industries. For
similar reasons, skilled-labour augmenting technology will be more efcient in high-technology industries. Indigenous technology created in a
112
labour-abundant developing country will be unskilled-labour augmenting, as suggested by the Directed Technical Change theory. Therefore, in
such populous developing countries, indigenous technology might be
more efcient than foreign technology in labour-intensive industries. By
contrast, foreign technology created in developed economies will be more
efcient than indigenous technology in technology-intensive industries. In
sum, technical change and the appropriateness of a technology are sector
specic. This sectoral extension of the analysis is important because in
reality countries often produce a diverse mixture of goods rather than
simply specialising in either labour- or capital-intensive production.
Middle-income countries have accumulated a pool of knowledge and
skills that distinguishes their factor endowments from those of the least
developed countries as well as those of the industrialised countries.
Therefore, middle-income economies are more likely to generate intermediate innovations with medium-level technology intensity. These
middle-income countries can reap the gains from investment in such
technologies through the sale of patents, payment of royalties or SouthSouth foreign direct investment between developing countries.
Moreover, for the same relative factor prices, the gain from introducing
new techniques is higher the larger the volume of demand. This also
implies that countries such as China, Brazil and India are more likely to
generate intermediate technology than smaller economies with the same
degree of capital scarcity (Findlay, 1978).
In sum, while there are potential gains from international technology
transfer, the extent of benets might be limited given the inappropriateness of foreign technology to local conditions and the preconditions for
effective FDI-assisted technology transfer. The relative importance of
indigenous and foreign innovation varies with the technology intensity
of the industries, the development level of the host country and the
difference in factor endowments between foreign and host countries.
5.3 FDI and innovation in China

Since the start of its economic reforms in 1978, China has received a large
volume of international direct investment and is now the second largest
FDI recipient in the world. In 2008, FDI inows into China reached a
peak of US$92.40 billion (MOFCOM, 2010). Whereas investments from
overseas Chinese in Hong Kong, Macao and Taiwan were the major
sources of inward FDI in the 1980s, the 1990s saw increasing inward FDI
113
from the major industrialised countries and other OECD countries. A

major objective of Chinas opening up to FDI is to exchange markets for
technology in the 7th Five-Year-Plan (19861990). Moreover, the science and technology system reform that began in 1985 has also specied
the acquisition of foreign technology as one of the main strategies for
technology development. FDI and imports are employed as major channels for foreign technology transfer. In 1995, the Chinese government
decided to accelerate science and technology development and pursue a
technology- and education-led development strategy. This signalled the
transition of the focus of Chinas technology development from technology transfer to indigenous innovation. The shift was further strengthened
in 2006 when the promotion of indigenous innovation was formally
listed as one of the countrys top priorities. In this process, the total
R&D expenditure in China grew from RMB 7.4 billion in 1987 to
RMB 35 billion in 1995 and to RMB 300.3 billion in 2006, at an average
annual growth rate of 21 per cent (MOST, 2010a).
5.4 Data and methodology

Data
We draw on the Annual Report of Industrial Enterprise Statistics compiled by the National Bureau of Statistics of China, covering all stateowned rms and other types of rms with annual turnover of more than
RMB 5 million (US$0.7 million). The data set includes variables such as
rm ownership structure, industry afliation, geographic location, year
of establishment, number of employment, gross output, value added,
xed assets, exports, R&D and employee-training expenditures.3 The
data currently available cover 2001 to 2005 and are broadly classied
into ve ownership categories: (1) state-owned, (2) collectively owned,
(3) privately owned, (4) foreign owned and (5) others. Foreign-owned
rms are further divided into rms with investments from Hong Kong,
Taiwan and Macao (HKTM rms) and rms with investments from
other foreign sources (FIEs). Other rms are mainly shareholding
enterprises (SHCs). Following Eurostat (2004), we used a fourfold
division of technological capabilities: high technology, medium-high
3
Nominal values are deated using industry-specic ex-factory price indexes

obtained from the China Statistical Yearbook 2006.
114
technology, medium-low technology and low-technology.4 Details of

industries in each category can be found as part of Appendix 5.1.
The sectoral pattern of their R&D activities is also reported in
Appendix 5.1.
We are interested in the drivers of technological upgrading in indigenous rms. The econometric work is hence conned to domestically
owned enterprises only. We nevertheless use the full sample of rms to
construct several variables of interest, for example, the share of foreign
rms in an industry in each region or the Herndahl index of market
concentration. We include only those rms for which there is a full set of
observations during the sample period, as the estimation of TFP growth
and its components using data envelopment analysis (DEA) requires
balanced data sets. The nal data set consists of 269,905 observations
from 53,981 rms.
The international industry-specic R&D stock is linked to the
Chinese rm-level data. The estimates of international R&D stock are
based on R&D expenditure data from the OECDs Main Science and
Technology Indicators. Following Coe and Helpman (1995), R&D
capital stock (S), which is dened here as the stock at the beginning of
a period, is calculated from R&D expenditure (R) based on the perpetual inventory model St = (1 ) St1 + Rt, where is the depreciation rate,
which is assumed to be 5, 10 and 15 per cent, alternatively. The starting
value for the stock, S0, is calculated following the procedure suggested
by Griliches (1979) as S0 = R0 / (g + ), where g is the average annual
logarithmic growth of R&D expenditures over the period for which
published R&D data were available, and R0 is the rst year for which
the data were available. Domestic R&D capital stocks are converted
into euros at 2000 constant prices.5 The R&D stocks of the 22 OECD
countries are then summed to proxy the world R&D stock. Table 5.1
Technological groups are divided according to the technology intensity (R&D

expenditure to value-added ratio) of the industries. We follow the standard used in
the Eurostat data set, which is based on the NACE Rev 1.1 Classication at the
three-digit level, which is converted into the SIC two-digit level classication.
Real R&D expenditures are nominal expenditures deated by an R&D price
index (PR), which is dened as PR = 0.5P + 0.5W, where P is the deator for
industrial output and W is an index of average industry wages. This denition of
PR implies that half of R&D expenditures is composed of labour costs, which is
broadly consistent with available data on the composition of R&D expenditures.
Table 5.1 Summary statistics of the variables

Variable
Denition
Obs.
Technical efciency
TFP growth
TE based on variable returns to scale

Geometric mean of two indexes estimated as the ratios of
distance functions of a rm from the frontier
Change in efciency between two periods
Shift in technology between the two periods
The ratio of R&D expenditure to total sales
Log of total R&D expenditure of state-owned rms in the
same industry and region
Log of total R&D expenditure of private rms in the same
industry and region
Log of total R&D expenditure of collectively owned rms in
the same industry and region
Log of total R&D expenditure of shareholding rms in the
same industry and region
Log of total R&D expenditure of Hong Kong, Taiwan and
Macauinvested rms in the same industry and region
Log of total R&D expenditure of foreign invested rms in
the same industry and region
International R&D capital stocks*foreign capital share at
rm level
International R&D capital stocks*foreign capital share at
industry level
Initial technical efciency level
269905 0.3836 0.2106

269905 0.8811 0.5825
0
0
1
11.391
269905
269905
269905
269905
0.9231
0.9639
0.0118
3.2957
0
0
0
0
16.25
7.905
1
14.7872
269905 8.3919 1.8876
12.6489
269905 6.2982 2.7883
11.3304
269905 9.8352 2.2576
15.4538
269905 7.7628 2.6586
13.8253
269905 8.1425 2.7045
14.6711
268991 1.3270 3.5505
13.7443
263331 9.2225 1.7379
13.0004
269905 0.3965 0.2043
0.013
Log of age
269905 2.6242 0.6989
0.6931
5.1874
Efciency change
Technical change
Firm R&D intensity
SOE R&D
POE R&D
COE R&D
SHC R&D
HKTM R&D
FIE R&D
International
R&D*rm
International
R&D*industry
Initial technical
efciency
Age
Mean
0.9924
1.0258
0.0017
8.1727
Std. Dev. Min
Max
Table 5.1 (cont.)

Variable
Denition
Obs.
Firm size
Market
concentration
Intangible assets per
person
Training expenditure
per person
Exports
Foreign capital share
HKTM capital share
Ownership
classications
Log of employment
Herndahl index (three-digit industry)
269905 5.2039 1.1096

269905 0.0196 0.0266
0.6931 11.9031
0.0017 1
Log of intangible assets per person
269900 0.9052 1.3799
9.0027
Log of training expenditure per person
269891 0.0834 0.1734
4.0108
Log of export sales

Share of foreign capital in a rm
Share of Hong Kong, Taiwan and Macau capital in a rm
1 State-owned enterprises (SOEs)
2 Collective-owned rms (COEs)
3 Private-owned rms (POEs)
4 HKTM rms (investors are from Hong Kong, Taiwan,
Macau)
5 Foreign invested enterprises (FIEs)
6 Others, mainly shareholding companies (SHCs)
Dummy equalling 1 for SIC 27 40 41
Dummy equalling 1 for SIC 26 28 36 37 39
Dummy equalling 1 for SIC 24 25 29 30 31 32 33 34 35 42
Dummy equalling 1 for SIC 13 14 15 16 17 18 19 20 21
22 23
269905
269905
269905
269905
269905
269905
269905
3.3919
0.0905
0.1053
0.1115
0.1093
0.2638
0.1449
4.7273
0.2610
0.2841
0.3147
0.3120
0.4407
0.3520
0
0
0
0
0
0
0
18.0558
1
1
1
1
1
1
269905
269905
269905
269905
269905
269905
0.1219
0.2488
0.0774
0.2343
0.3404
0.3479
0.3271
0.4323
0.2672
0.4236
0.4739
0.4763
0
0
0
0
0
0
High tech
Medium-high tech
Medium-low tech
Low tech
Mean
Std. Dev. Min
Max
1
1
1
1
1
1
117
gives the denitions of the variables used in the analysis along with some
summary statistics.
Methodology
The empirical study is carried out using the following steps. First, we
estimate TFP growth using the Malmquist productivity index and
decompose it into technical progress and efciency change. TFP is
estimated for each industry separately allowing for different technology
and production functions for different industries. Second, we identify
the rms that are located on the technology frontier, as measured by
rms technical efciency and compare foreign with domestic rms.
Third, we use econometric techniques to estimate the drivers of TFP
growth, technical change and efciency improvement for rms in each
technology category.
Estimation of TFP growth

The conventional technique for estimating TFP is the Solow residual
method. This method, however, suffers from several well-known limitations. For example, the production function is assumed to be known
and to be homogenous across industries and rms, and technical change
is assumed to be neutral. If these assumptions do not hold, TFP measurements will be biased (Arcelus and Arocena, 2000). Therefore, this
chapter estimates TFP growth by using a non-parametric programming
method developed by Fare et al. (1994). Following Fares approach, a
production frontier is constructed based on all existing observations.
Technical efciency is dened as the distance of each observation from
the frontier. TFP growth is dened as a geometric mean of two
Malmquist productivity indexes and is estimated on the basis of the
ratios of distance functions of observations at times t and t + 1. This
approach is capable of measuring productivity in a multi-input, multioutput setting and does not require the assumptions of the Solow
method. It also has another advantage in that it allows for the decomposition of productivity growth into two mutually exclusive and
exhaustive components: (1) technical change measured by shifts in the
technological frontier and (2) efciency change measured by movements towards (or away from) the frontier. The latter is a measurement
of catching up (Fare et al., 1994).
118
This decomposition of TFP growth enables us to investigate the

different effects of foreign and indigenous innovation on technical
progress and efciency improvement. The main shortcoming of this
technique is the lack of a provision for statistical noise or measurement
errors in the model (Greene, 1997). Any deviation from the frontier is
attributed to inefciency. The frontier is hence sensitive to outliers in
the data. To reduce the possible bias caused by this shortcoming,
careful data cleansing has been carried out on an industry-by-industry
basis.
The Malmquist TFP growth index of the production point (xt+1, yt+1)
relative to the production point (xt, yt) is given as follows:
"
M0 x
t1
;y
t1
;x;y
t
!#12

t1
Dt0 xt1 ; yt1
Dt1
; yt1
0 x
t
t
Dt0 xt ; yt
Dt1
0 x ; y
5:1
A value greater than 1 indicates positive TFP growth in period t + 1.

When performance deteriorates over time, the Malmquist index will be
less than 1.
Equation (5.1) can be rewritten as
M0 xt1 ; yt1 ; xt ; yt
t1
Dt1
; yt1
0 x
t
D xt ; yt
" 0
!
!#12
Dt0 xt1 ; yt1
Dt0 xt ; yt

t1 ; yt1
t
t
Dt1
Dt1
0 x
0 x ; y
5:2
where efficiency change
t1
Dt1
; yt1
0 x
t
t
D0 x ; yt
"
and technical change
Dt0 xt1 ; yt1

t1 ; yt1
Dt1
0 x
5:3
!
!#12
Dt0 xt ; yt
t
t
Dt1
0 x ; y
5:4
Thus, TFP change is decomposed into two components: efciency

change and technical change. Efciency change measures whether production is getting closer to or further away from the frontier. Technical
119
change captures the shift in technology between the two periods. It

indicates whether or not technical progress occurred at the input-output
combination of a particular industry. A value greater than 1 indicates
technical progress or efciency improvement correspondingly. A value
of less than 1 indicates deterioration in performance. The Malmquist
productivity index is estimated by using non-parametric linear programming techniques. Details of the estimation method are given in
Appendix 5.2. This chapter follows the basic DEA model, which
includes three inputs: labour (number of employees), capital (net xed
assets) and variable costs (intermediate costs); and one output: total
rm output. Returns to scale can be variable.
Foreign and indigenous R&D efforts and TFP growth

Having decomposed productivity growth, the next step is to estimate
the drivers of TFP growth, technical change and efciency improvement
using regression analysis. Following Findlay (1978), foreign capital and
domestic capital are regarded as distinct factors of production in this
model, reecting the fact that capital, management and technology are
inextricably combined.
The empirical analysis of the impact of indigenous and foreign innovation on indigenous technology upgrading is based on the following
model:
Y it RDit FDIit Xit Dit it
5:5
where the dependent variable Y represents TFP growth, technical

change and efciency improvement, respectively. RD is a vector of
R&D variables. FDI is FDI intensity measured by the ratio of foreign
investment to total assets. X is a vector of control variables. D is the full
set of time and sector dummies, and is a random error term; i and t
denote rm and time, respectively.
We are most interested in the set of R&D variables. Our specication
contains three types of innovation efforts: innovation at the rm, industry and international levels. We construct the variables as follows:
(1) the intensity of a rms own R&D activities; (2) the industry-level
innovation effort in each of the 171 three-digit industries and 31 provinces, which is constructed as the total R&D expenditure of different
ownership types in the same industry and region; (3) international R&D
spillover, which is constructed as the interaction term of international
120
industry-specic R&D stock and FDI share at both rm and industry

levels. The full empirical model is therefore as follows:
Y it 1 RDF it 2 RDIit 3 RDW it FDIFit
4 RDW it FDIDit FDIit Xit Dit it
5:6
where RDF is rm R&D intensity, RDI is a vector of industry-level

R&D spillover variables, RDW is world R&D stock constructed from
the OECD STAN (Structural Analysis Database) database as discussed
earlier, and FDIF and FDID are FDI intensity at rm and industry
levels, respectively.
Control variables consist of the initial level of technology efciency,
age, rm size, exports, intangible assets, labour training and market
concentration. These variables are chosen on the basis of the existing
empirical literature (e.g., Bernard and Jensen, 1999; Aw, Chung and
Roberts, 2000; Fu, 2005; Girma, Gong and Grg, 2009). Smaller rms,
rms with exports and rms with greater amounts of labour training are
more likely to have faster TFP growth, technical change and efciency
improvements. Firms already on the frontier are unlikely to grow as fast
as other rms. There is no agreed relationship between rm age, intangible assets and market concentration.6 For example, the effect of market
competition on innovation may be a double-edged sword. Geroski (1990)
argues that lack of competition in a market gives rise to inefciency and
sluggish innovative activity. On the other hand, Schumpeterians claim
that monopoly power makes it easier for rms to appropriate the returns
from innovation and thereby provides the incentive to invest in innovation
(Cohen and Levinthal, 1989; Symeonidis, 2001).
In the earlier specication, FDI is represented by two variables, the
share of foreign capital and HKTM capital at the rm level. We conjecture that rms might grow faster if they have easier access to
advanced foreign technical and managerial knowledge through their
foreign or HKTM Chinese partners. We distinguish between foreign
and HKTM capital to account for any differences in motivation
between foreign and HKTM capital. Foreign investors are more likely
to be market oriented, whereas HKTM investors might look for less
expensive labour and rent to reduce costs and export their outputs.
6
According to Accounting System for Business Enterprises, the costs to develop

intangible assets are regarded as the R&D costs of self-created products that are
registered for a legal right to the asset, such as a patent (Pacter and Yuen, 2001).
121
Some independent variables in the earlier specication are arguably

determined simultaneously with the dependent variable, TFP growth
and its components. In other words, there might be a potential endogeneity problem even after controlling for xed effects. For example,
rms with relatively large R&D activities are more likely to have
higher TFP growth and faster technical change. However, it is also
possible that rms with higher growth rates might invest more in
R&D activities to keep their technology advantages. Another example
is the foreign share of a rm. Firms with a higher foreign capital share
could have better access to foreign technology and therefore have
higher growth rates, but there also might be a cherry-picking effect
(Huang, 2003) whereby foreigners choose to invest in faster growing
rms. Similar arguments can be made in the case of export and HKTM
capital participation.
To deal with the problem of endogeneity, we employ the xed effects
generalised method of moments (GMM) regression technique (see, inter
alia, Hansen, 1982 and Arellano and Bond, 1991). The use of industry
and region dummies in the regressions is designed to mitigate part of
this potential endogeneity problem. Lagged values of the potentially
endogenous variables (exports, R&D intensity, foreign capital share
and HKTM capital share) are used as instruments. In addition, the
shares of foreign and HKTM rms in the industry and region are used
as extra instruments. Given the low level of competition from stateowned rms (Girma et al., 2009), we assume that a sector will be more
efcient if more foreign or HKTM rms are participating in it. We also
formally test whether the assumption of endogeneity is borne out by the
data at hand and whether our instruments are relevant in the sense of
exhibiting sufciently strong correlation with the potentially endogenous variables. We also test for the appropriateness of the instrumental
variable candidates using Hansen Js test for overidentifying restrictions
and the validity of the instruments using the Sargan test. Reassuringly,
we nd that our instruments are appropriate on all counts.
5.5 Results
Technical efciency and TFP growth
Figure 5.1 reports the level of technical efciency of rms by ownership.
FIEs have the highest level of average technical efciency, whereas that
122
Technical efficiency of firms

0.5
0.4
0.3
0.2
0.1
0
Low tech
SOE
Medium-low tech Medium-high tech

Collective
Private
SHC
High tech
HKTM
FIE
Figure 5.1 Technical efciency by ownership and technology category
of SOEs is the lowest. The average technical efciency of HKTM rms

ranks second in the medium-technology sectors but is lower than that of
SHCs in the high-technology sectors. Across technology groups, FIEs
enjoy an advantage in technical efciency over indigenous rms. The
gap is small in low-technology industries but becomes larger in other
categories that have higher technology intensity.7 In the hightechnology sector, the estimated technical efciency of FIEs is about
30 per cent higher than that of SOEs.
Figure 5.2 reports estimated TFP growth by rm ownership. Looking
at growth dynamics, it is clear that in all industries Chinese rms have
experienced considerable TFP growth during the 20012005 period
with growth running at an average annual rate of 4.8 per cent. The
growth is mainly due to technical change at an average annual growth
rate of 5.1 per cent rather than to efciency improvements: the average
The t-test results suggest that the size of the gap in TE between SOEs and FIEs in
the low-technology sector is signicantly smaller than that in the other three
technology categories. Although the size of this gap also differs between the other
three sectors, the magnitude of these differences is negligible.
123
TFP growth
1.07
1.06
1.05
1.04
1.03
1.02
1.01
1
Low tech
SOE
Medium-low high Medium-high tech

Collective
Private
SHC
High tech
HKTM
FIE
Figure 5.2 TFP growth in Chinese manufacturing rms, 20012005
annual growth rate of efciency improvement was only 0.7 per cent
over the sample period. The growth is widely spread across different
sectors, and indigenous rms have taken the lead in this growth process
in comparison to foreign invested rms.
Drivers of technological upgrading

Despite the lower average technical efciency of indigenous rms than
of foreign-invested rms, a large number of indigenous rms stand on
the frontier in the low- to medium-hightechnology sectors (Figure 5.3).
On the contrary, in the high-technology sector, a larger number of
foreign rms stand on the technology frontier.8 A considerable number
of SOEs are also found on the technology frontier. This might be
The industries in which foreign rms have obvious dominance include electronics
and telecommunications, instruments and measuring equipment, culture,
educational and sports goods, as well as garments and leather products for which
HKTM rms have a clear lead. Indigenous rms have dominant presence in the
low- and low-mediumtechnology industries such as food processing,
papermaking, smelting and processing of ferrous and nonferrous metals. There are
also a number of industries in which foreign and domestic rms share the lead and
push the frontier upwards together.
124
Distribution of firms on the frontier
35%
30%
25%
20%
15%
10%
5%
0%
Low tech
SOE
Medium-low
tech
Collective
Medium-high
tech
Private
SHC
High tech
HKTM
FIE
Figure 5.3 Distribution of rms on the frontier

Note: All rms in each category (e.g., low-tech) add up to 100 per cent.
explained by the fact that historically the Chinese science and technology system was inuenced by the former Soviet Union model in which
the state takes the leading role in R&D. Overall, technological upgrading
in the Chinese economy is not dominated by any single sector. Foreign
rms have considerable advantages in the high-technology sector and
indigenous rms have taken a lead in the low- and medium-technology
sectors. Shareholding companies, which are relatively rich in capital and
skilled labour, enjoy a lead in the medium-hightechnology sector.
Privately owned enterprises take a lead in the low- and medium-low
technology sectors.
The greatest amount of technical change in China has taken place in
the medium-hightechnology industries (Figure 5.4). This change is
driven by both indigenous and foreign rms, although indigenous
rms have a higher rate of change. In the low-technology industries,
again, indigenous rms, especially private rms, are the main drivers of
technical change. In contrast, in the high-technology industries, foreign
rms are the leading force, with HKTM Chinese rms and FIEs enjoying
a growth rate of 7.7 per cent and 7.3 per cent, respectively. However,
125
Technical change
1.2
1.15
1.1
1.05
1
Low tech
Medium-low tech
Medium-high tech
High tech
0.95
0.9
SOE
Collective
Private
SHC
HKTM
FIE
Efficiency change
1.2
1.15
1.1
1.05
1
0.95
Low tech
Medium-low tech
High tech
Medium-high tech
0.9
SOE
Collective
Private
SHC
HKTM
FIE
Figure 5.4 Technical change and efciency improvement in Chinese manufacturing rms, 20012005
the average efciency change indexes in these industries are less than 1,
suggesting that most followers have not been able to catch up with the
innovation leaders. On the contrary, in the medium-lowtechnology
sector, most progress is made by followers through efciency changes,
while there is a lack of signicant technical progress in the sample
period.
126
Indigenous and foreign innovation efforts and indigenous

technical change
Table 5.2 reports the GMM estimates of the drivers of TFP growth,
technical change and efciency improvements in domestic rms. Results
from the Wu-Hausman specication test suggest signicant endogeneity between R&D, exports and FDI on the one hand and the dependent
Table 5.2 Determinants of TFP growth and technical change: GMM

estimates of full sample of domestic rms
TFP growth
Firm R&D
Intensity
SOE R&D
POE R&D
COE R&D
SHC R&D
HKTM R&D
Foreign R&D
International
R&D*Firm
0.941***
(0.29)
0.00110*
(0.0006)
0.0045***
(0.001)
0.0004
(0.0006)
0.0031***
(0.0009)
0.0004
(0.0006)
0.0015**
(0.0007)
0.0160**
(0.0066)
International
0.0053***
R&D*Industry
(0.0016)
Initial Technical
0.194***
Efciency
(0.0071)
Age
0.0008
(0.0018)
Technical change
(shift of frontier)
0.219
Efciency
improvement
(catch-up with frontier)
1.388***
(0.17)
0.0016**
(0.0007)
0.0125***
(0.0013)
0.0039***
(0.0007)
0.0009
(0.0012)
0.0007
(0.0008)
0.0082***
(0.0009)
0.0133**
(0.33)
0.0006
(0.001)
0.0105***
(0.0016)
0.0024***
(0.0009)
0.0066***
(0.0016)
0.0005
(0.0010)
0.0023**
(0.0011)
0.0160*
(0.0067)
0.0034*
(0.0094)
0.0083***
(0.0019)
0.186***
(0.0027)
0.483***
(0.0077)
0.0036*
(0.002)
(0.011)
0.0035
(0.0027)
127
Table 5.2 (cont.)
TFP growth
Firm Size
Market
Concentration
Intangible Assets
Training
Expenditure
Exports
Foreign Capital
Share
HKTM Capital
Share
Constant
Observations
Exogenous Test
Hansen J Test
Technical change
(shift of frontier)
Efciency
improvement
(catch-up with frontier)
0.0223***
(0.0013)
0.0851
0.0164***
(0.0015)
0.0342
0.0006
(0.0019)
0.0574
(0.058)
0.0050***
(0.0009)
0.0722***
(0.066)
0.0045***
(0.001)
0.0419***
(0.097)
0.0001
(0.0014)
0.0531***
(0.0084)
0.0013***
(0.0004)
0.559***
(0.0081)
0.0002
(0.0004)
0.361*
(0.012)
0.0011**
(0.0006)
0.447
(0.200)
0.183**
(0.200)
0.127*
(0.290)
0.0214
(0.075)
1.326***
(0.019)
155885
0.0000
0.7605
(0.071)
0.522***
(0.022)
155,885
0.0000
0.5560
(0.091)
2.294***
(0.032)
155,885
0.0000
0.4743
Note:
1. Robust standard errors in parentheses;
2. *signicant at 10 per cent; **signicant at 5 per cent;
***signicant at 1 per cent.
3. All specications include the full set of time and two-digit industry dummies.
variable on the other hand. The GMM estimation results are therefore
preferred to the OLS estimates. As a robustness check, regressions of the
basic model with industrial and international R&D spillovers at three
alternative depreciation rates (5 per cent, 10 per cent and 15 per cent)
were carried out. The estimated coefcients from different model
128
specications are consistent, underlining the robustness of the estimated

results.9
Evidence from Table 5.2 indicates that rms own R&D efforts have
a signicant and positive impact on their TFP growth. This comes
mainly from their substantial positive impact on the catch-up process:
the estimated coefcient of 1.388 is much higher than those of all the
other explanatory variables. This fact highlights the crucial role of rmlevel R&D in the assimilation and adaptation of advanced technology
diffused from the frontier. Industry-level R&D spillovers from other
SOEs had a signicant positive effect on technical change, suggesting
positive social returns to the R&D investment made by the state sector.
Likewise, the industry-level R&D spillovers from other privately
owned rms have a signicant positive effect on the technical change
of other rms in the same industry. The magnitude and signicance
level of the estimated coefcient of R&D spillovers for POEs is greater
than that of the R&D variable for SOEs, suggesting that R&D investments made by POEs have exerted greater externalities on indigenous
technical change than those of SOEs. The overall spillover effect of
R&D investment in shareholding companies does not appear to contribute signicantly to technical change in the industry; it does, however, appear to contribute signicantly to the catch-up process of
follower rms in the industry. Unexpectedly, the overall spillover effect
of the R&D activities of collectively owned rms appears to be negative.
We will explore this later (Table 5.3 and Table 5.4) by looking into its
effect in different industries.
Turning to the spillover effects of foreign R&D, spillovers from the
R&D activities of FIEs and HKTM Chinese invested rms both appear
to have a negative effect on indigenous technical change, and this effect
is statistically signicant at the 1 per cent level for FIEs. However, R&D
efforts in the FIEs have a positive effect on the catch-up process, suggesting that R&D activities in these rms do not inuence innovations
on the frontier but contribute to the diffusion of existing relatively
superior technology. Finally, international R&D spillover through
FDI has mixed effects. Greater FDI intensity at the rm level appears
9
Both OLS and GMM estimations were carried out but only results from GMM
regressions are reported here. We have only reported the GMM estimates with a
10 per cent R&D depreciation rate because of space limitations. Results of all the
estimations are available from the authors on request.
129
Table 5.3 Determinants of technical change: GMM estimates by industry

group
High tech
Firm R&D
Intensity
SOE R&D
POE R&D
COE R&D
SHC R&D
HKTM R&D
Foreign R&D
International
R&D*Firm
International
R&D*Industry
Initial Technical
Efciency
Age
Firm Size
Market
Concentration
Intangible Asset
Training
Expenditure
0.0128
Medium-high Medium-low
tech
tech
Low tech
0.166
0.479
0.457
(0.17)
0.0049**
(0.0023)
0.0071**
(0.0031)
0.0098***
(0.0019)
0.0016
(0.0044)
0.0067**
(0.0035)
0.0056
(0.004)
0.0086
(0.28)
0.0142***
(0.0018)
0.0172***
(0.0029)
0.0108***
(0.0013)
0.0252
(0.0025)
0.0096***
(0.0016)
0.0232***
(0.0021)
0.0047
(0.34)
0.0020*
(0.0012)
0.0045**
(0.002)
0.0111***
(0.0012)
0.0084***
(0.0018)
0.0028***
(0.001)
0.0012
(0.0013)
0.0267*
(0.92)
0.0012
(0.001)
0.0137***
(0.0022)
0.0114***
(0.0011)
0.0032
(0.002)
0.0031*
(0.0017)
0.0152***
(0.0017)
0.0040
(0.0074)
0.0043
(0.011)
0.017***
(0.015)
0.0200***
(0.016)
0.0016
(0.0056)
0.00813
(0.0032)
0.222***
(0.0029)
0.222***
(0.0065)
0.176***
(0.02)
0.0008
(0.0052)
0.0021
(0.0041)
0.303**
(0.015)
0.0057
(0.004)
0.0185***
(0.0028)
0.0947
(0.013)
0.0090***
(0.0031)
0.0176***
(0.0023)
0.294***
(0.013)
0.0048
(0.0034)
0.0151***
(0.0025)
0.796***
(0.14)
0.0024
(0.0023)
0.0156
(0.098)
0.0067***
(0.0018)
0.0473***
(0.1)
0.0064***
(0.0017)
0.0470***
(0.21)
0.0001
(0.0019)
0.0486***
(0.014)
(0.015)
(0.013)
(0.016)
130
Table 5.3 (cont.)
Export
Foreign Capital
Share
HKTM Capital
Share
Constant
Observations
Exogenous Test
Hansen J Test
High tech
tech
tech
Low tech
0.0007
(0.0011)
0.326
0.0011
(0.0009)
0.162
0.0006
(0.0007)
0.737*
0.0009
(0.0007)
0.0693
(0.28)
0.174*
(0.4)
0.153
(0.44)
0.109
(0.42)
0.0817
(0.1)
0.703***
(0.077)
9,610
0.0000
0.0511
(0.26)
1.503***
(0.036)
37,377
0.0000
0.0708
(0.15)
0.799***
(0.036)
56,227
0.0000
0.4589
(0.12)
0.517***
(0.064)
51,997
0.0000
0.5955
Note:
2. *signicant at 10 per cent; **signicant at 5 per cent; ***signicant at 1 per cent.
Table 5.4 Determinants of efciency improvement: GMM estimates by

industry group
High tech
Firm R&D
Intensity
SOE R&D
POE R&D
COE R&D
SHC R&D
HKTM R&D
0.666*
(0.38)
0.0055
(0.0042)
0.0159**
(0.0064)
0.0092***
(0.0035)
0.0076
(0.0073)
0.0002
(0.0054)
tech
tech
Low tech
1.229**
2.694***
(0.61)
0.0001
(0.0025)
0.0122***
(0.0033)
0.0044**
(0.0021)
0.0080**
(0.0034)
0.0082***
(0.0018)
(0.79)
0.0036**
(0.0016)
0.0098***
(0.0029)
0.0031*
(0.0018)
0.0062***
(0.003)
0.0032***
(0.0015)
3.404*
(1.81)
0.0025
(0.0016)
0.0073***
(0.0027)
0.0109***
(0.0014)
0.0123***
(0.0026)
0.0033
(0.0022)
Table 5.4 (cont.)
High tech
Foreign R&D
International
R&D*Firm
International
R&D*Industry
Initial Technical
Efciency
Age
Firm Size
Market
Concentration
Intangible Assets
Training
Expenditure
Exports
Foreign Capital
Share
HKTM Capital
Share
Constant
Observations
Exogenous Test
Hansen J Test
tech
tech
Low tech
0.0171*** 0.0030
(0.0056)
(0.0026)
0.0002
0.0287
0.0096***
(0.0018)
0.0468*
0.0136***
(0.002)
0.0103
(0.012)
0.0005
(0.022)
0.0453***
(0.028)
0.0236***
(0.019)
0.0585***
(0.0075)
0.373***
(0.0042)
0.243***
(0.0051)
0.652***
(0.0077)
0.501***
(0.035)
0.0004
(0.0087)
0.0148**
(0.007)
0.528
(0.02)
0.0048
(0.0047)
0.0081***
(0.0034)
0.262*
(0.019)
0.0053
(0.0049)
0.0151***
(0.0036)
0.452***
(0.018)
0.0128***
(0.0045)
0.0031
(0.0033)
0.774***
(0.47)
(0.15)
0.0095** 0.0038*
(0.0039)
(0.0023)
0.0949*** 0.0737***
(0.17)
0.0021
(0.0025)
0.0336
(0.23)
0.0018
(0.0025)
0.0317
(0.028)
0.0001
(0.0017)
0.0234
(0.021)
0.0027***
(0.001)
1.143
(0.021)
0.0014
(0.0011)
1.273
(0.022)
0.0007
(0.0009)
0.275
(0.44)
0.3
(0.88)
0.674
(0.84)
0.186
(0.5)
0.0775
(0.2)
1.737***
(0.12)
9,610
0.0000
0.1312
(0.71)
0.622***
(0.051)
37,377
0.0000
0.2395
(0.18)
1.959***
(0.058)
56,227
0.0000
0.4646
(0.14)
1.837***
(0.074)
51,997
0.0000
0.8785
Note:
132
to encourage the diffusion of frontier technologies. Greater FDI intensity in the industry facilitates the diffusion of superior, rather than the
most advanced, technology. This suggests that with greater control of
rms, MNEs are more likely to bring in more advanced technology. On
the other hand, greater openness to FDI in the industry will increase the
knowledge ow into the host economy, which will help the majority of
followers move closer to the frontier.
As regards the coefcients of the control variables, rms with better
initial technical efciency tend to grow more slowly. Larger and older
rms appear to have experienced greater technical change. Firms with
high export intensity, high FDI intensity, more training and greater
intangible assets have higher TFP growth than those that lack these
characteristics. Their effects on the two components (technical change
and efciency change) are, however, different, which we will discuss in
greater detail later (Tables 5.25.4).
Taking advantage of this rich data set, we examine further the
impact of innovation efforts on technical change and efciency
improvements in different technology categories. The results reported
in Tables 5.3 and 5.4 indicate different effects of indigenous and
foreign innovation efforts in different industry groups. The indigenous
R&D of individual rms has no signicant impact on technical change.
It has, however, contributed signicantly to efciency improvements,
which reects catch-up towards the frontier. The lower the technology
intensity of the sector, the larger is the impact of indigenous rm-level
innovation. This is not surprising given that the First National
Economic Census in 2004 found that about 95 per cent of total business R&D expenditure was spent on development and only 5 per cent
was spent on basic scientic research. Interestingly, spillovers from
R&D at the industry level appear to be the main drivers of indigenous
technological upgrading. In general, industry-level R&D activities in
indigenous rms, especially POEs, SOEs and COEs, have shown signicant positive spillovers on the technical progress of indigenous
rms in most technology groups. This evidence suggests that it is
collective indigenous R&D activities, that is, R&D at industry level,
that push out the technology frontier and drive the technological
upgrading of indigenous rms.
On the other hand, the R&D activities of FIEs and HKTM rms at
the industry level show a negative spillover effect on technical change
in indigenous rms in most industry groups, with the exception of
133
HKTM rms in the medium-hightechnology sector. This can be

explained by several factors. First, foreign technology may not be
appropriate in these industries. Local rms are likely to nd it difcult
to adapt foreign technology to the local context. Second, foreign R&D
activities may well intensify competition for the limited domestic
talent pool (Chang et al., 2006) and crowd out indigenous rms
from local labour, resource and product markets. Moreover, foreign
R&D centres may have limited interest in sharing knowledge with
domestic rms and R&D labs (Chang et al., 2006; Zhou, 2006. This
may also be explained by the strict intellectual property rights protection of these high-end FIEs that aim to protect their intellectual property against indigenous rms. Finally, the ndings of recent studies
(Narula and Zanfei, 2004) suggest that the core technology development of MNEs is still carried out within the headquarters of a rm,
whereas the application and adaptation of research are principally the
tasks of its afliates in foreign countries. Therefore, these R&D activities may not contribute to technical change but might have a positive
impact on catching up. This also explains the result in Table 5.4 that
industry-level R&D of FIEs has a signicant positive effect on the
catching-up process of indigenous rms in the high and medium-low
technology industries.
As regards international R&D spillovers facilitated by FDI, the spillover effect on indigenous technical change is mostly insignicant or
negative except in the medium-lowtechnology sector, where technical
change is negligible over the sample period as Figure 5.4 illustrates. On
the other hand, the spillover effect of FDI on catch-up seems to be
mixed. In the low- and medium-hightechnology sectors where indigenous rms have led technical change in China, industry-wide openness
to FDI appears to have facilitated spillovers of second-tier technology
that helped follower rms catch up with the frontier. However, in the
medium-lowtechnology sector, since the effect of international R&D
spillovers has been signicant in pushing out the frontier, albeit to a
small extent, the effect of such spillovers on catch-up will be negative
given the way in which technical change and efciency improvement are
dened and decomposed.
The differing spillover effects of indigenous as compared with foreign
innovation efforts on the technical change of domestic rms suggests
that indigenous technologies t the local technical context better and
are therefore more likely than foreign technology to be adopted by
134
domestic rms. Innovation activities in foreign invested rms not only

fail to offer sufcient positive externalities for the domestic rms but
even exert signicant negative pressure on the technical change of local
rms. Therefore, although foreign technology transfer may facilitate
technology development at an early development stage and may assist
the diffusion of second-tier technology, developing countries have to
rely on collective indigenous innovation at the industry level in their
catch-up with the world technology frontier.
With respect to the control variables, the results show that rms with
higher initial technical efciency grew more slowly in terms of both
technical change and efciency improvement. However, the estimated
coefcient in the high-technology sector is positive, suggesting that in
this sector rms standing on the frontier tend to move faster than
followers. Moreover, larger rms have faster rates of technical change,
except in high-technology industries. However, the effect of rm size on
efciency improvements is mixed. Exports do not have a signicant
effect on technical change, a nding that is consistent with that of Fu
(2005), who uses Chinese industry-level panel data. This result suggests
that a focus on cost competitiveness based on low-cost unskilled labour
and an over-reliance on processing trade provide little effective incentive
for rms to innovate.
As expected, training exerts a signicant positive impact on efciency improvement but, surprisingly, has a signicant and negative
impact on technical change. Training in China is likely to be concerned
with the teaching of new or advanced practices but not with the
creation of frontier technologies. The estimated coefcients of the
intangible assets variable are either negative or statistically insignificant in the technical change equation, possibly because in Chinese
accounting practices, intangible assets include R&D investment at
the development stage but not at the research stage. For rms in
technology-intensive industries, novel research activities may play a
more important role in keeping them on the frontier. Moreover,
intangible assets are correlated with the xed assets that are the
measure of capital in the TFP estimation, which for accounting
reasons generate a negative association between TFP and intangible
assets. Finally, market concentration appears to promote technical
change and innovation, which is consistent with the Schumpeterian
hypothesis. Lack of competition, however, deters efciency improvements in most sectors.
135
5.6 Conclusions
This chapter explores the drivers of TFP growth and technological
upgrading in a middle-income developing country. In particular, the
role of indigenous innovation and foreign technology transfer is analysed. The main ndings can be summarised briey. First, neither foreign nor indigenous rms dominate the technology frontier in China. In
low- and medium-technology sectors, more indigenous rms are
located on the frontier. Privately owned rms have taken a lead in the
low- and medium-lowtechnology industries, and shareholding companies are the leaders in medium-hightechnology industries. In the hightechnology sector, foreign rms dominate the frontier. In other words,
foreign technology is not always superior to indigenous technology,
especially in a middle-income country. In the low-technology sector,
indigenous technology is more appropriate and hence more efcient
than foreign technology. China has also accumulated the capability to
create technology appropriate to the medium-technology sector.
Second, indigenous Chinese rms have experienced considerable TFP
growth, at an average annual growth rate of 4.8 per cent from 2001 to
2005. This growth was driven mainly by technical change in all sectors
except the medium-lowtechnology industries. In the low- and mediumhightechnology sectors, indigenous rm technical growth is considerably higher than in foreign invested rms. Collective indigenous R&D
activities at the industry level are the major driver of the technological
upgrading process. R&D investment in POEs, SOEs and COEs has had
signicant positive externality effects on technical change in indigenous
rms.
The effects of R&D activities in FIEs are, however, controversial.
Such activities have exerted a signicant negative effect on indigenous
technical change, whereas their impact on catch-up in follower rms is
positive in some industries. International R&D spillovers facilitated by
FDI are insignicant for indigenous technical change. In sum, although
foreign technology transfer may facilitate technology development at an
early development stage and assist in the diffusion of second-tier technology, developing countries have to rely on collective indigenous
innovation efforts to catch up with the world technology frontier.
These ndings have important implications for technology policy in
developing countries. First, the net technological benets from FDI are
often limited. It is collective indigenous innovation efforts that drive
136
indigenous technological upgrading in a developing country. Indigenous

technology will be more efcient than foreign technology in sectors that
intensively use factors that developing countries have in abundance.
Therefore, there is an important role for technology and industrial policies in developing countries to encourage indigenous innovation for
technological upgrading and catch-up.
Second, developing countries may not only be users of new technology but also creators of new technologies in some sectors, especially in
middle-income countries. Such technologies may be more appropriate
for other developing countries with similar factor endowment and at a
similar development stage. For countries in the developing world, foreign technology is multi-tiered, rather than solely originating in developed countries. This should be taken into account in strategies for
technology choice. International technology transfer through SouthSouth trade and FDI should be the major transmission channels.
Finally, the technology gap between developed and developing countries in certain industries remains remarkable. The Chinese experience
of the two-leg forward strategy suggests that in order to maximise the
benets from existing knowledge and accelerate the catch-up process,
both indigenous innovation and acquisition of foreign knowledge are
needed, but with the optimal mixture differing among sectors and stages
of development. An effective technology policy package may thus be
country, region and industry specic, but well-focused policies to foster
the absorptive capacity and innovation capabilities of indigenous rms
are always crucial for success.
Appendix 5.1 Sectoral pattern of R&D activities (all rms in our sample)
R&D
Fraction of
expenditure/sales
R&D investors (rm with R&D)
SIC2
High technology
27-Medical and Pharmaceutical
Products
40-Electronic and Telecommunications
41-Instruments and Meters
2001
2005
2001
2005
0.4451 0.5016 0.0172
0.0818
0.3114 0.3266 0.0183

0.3159 0.3133 0.0182
0.0284
0.0248
137
Appendix 5.1 (cont.)

R&D
Fraction of
expenditure/sales
R&D investors (rm with R&D)
Medium-high technology
36-Special Purposes Equipment
39-Electric Equipment and Machinery
37-Transport Equipment
26-Raw Chemical Materials and
Chemical Products
28-Chemical Fiber
Medium-low technology
35-Ordinary Machinery
42-Artefact and Other Manufacturing
34-Metal Products
29-Rubber Products
31-Nonmetal Mineral Products
24-Cultural, Educational and Sports
Goods
30-Plastic Products
33-Smelting and Pressing of Nonferrous
Metals
25-Petroleum Rening and Coking
32-Smelting and Pressing of Ferrous
Metals
Low-technology
23-Printing and Record Media
Reproduction
17-Textile Industry
19-Leather, Furs, Down and Related
Products
18-Garments and Other Fibre Products
21-Furniture Manufacturing
13-Food Processing
15-Beverage Industry
14-Food Production
22-Papermaking and Paper Products
20-Timber Processing
16-Tobacco Processing
0.3237
0.2476
0.2738
0.2126
0.2856
0.2218
0.2646
0.1848
0.0123
0.0138
0.0079
0.0079
0.0188
0.0128
0.0123
0.0113
0.2189 0.1791 0.0049
0.0057
0.2168
0.0926
0.1092
0.1921
0.1131
0.1143
0.0088
0.0078
0.0071
0.0060
0.0073
0.0067
0.0135
0.0092
0.0090
0.0088
0.0085
0.0074
0.0991 0.0832 0.0047

0.1683 0.1599 0.0077
0.0073
0.0055
0.1818 0.1875 0.0070

0.1388 0.0976 0.0031
0.0053
0.0051
0.0654 0.0675 0.0093
0.0100
0.0857 0.0769 0.0050

0.1049 0.0841 0.0061
0.0084
0.0078
0.0494
0.0930
0.1022
0.1692
0.1711
0.0748
0.0756
0.4271
0.0071
0.0071
0.0067
0.0066
0.0064
0.0062
0.0057
0.0031
0.1950
0.0775
0.0892
0.1746
0.0822
0.1006
0.0536
0.1008
0.0786
0.1645
0.1680
0.0601
0.0628
0.4479
0.0055
0.0048
0.0047
0.0056
0.0065
0.0046
0.0062
0.0019
138
APPENDIX 5.2
Assume a production technology St that produces a vector of outputs,
N
t
yt 2 RM
, by using a vector of inputs, x 2 R , for each time period
t=1, . . ., T. The output-based distance function at t is dened as the
reciprocal of the maximum proportional expansion of the output vector
yt, given inputs xt.
Dt0 xt ; yt inf f : xt ; yt = 2 St g
sup f : xt ; yt 2 St g
Dt0 xt ; yt 1 if and only if (xt, yt) St. Dt0 xt ; yt 1 if and only if
(xt, yt) is on the frontier. The distance to the frontier is estimated by
using non-parametric linear programming techniques. Assume
k = 1, . . . , K rms using n = 1, . . . , N inputs xk;t
n at each time period
t = 1, . . .T to produce m = 1, . . . . . .,M outputs yk;t
m . To estimate the
productivity change of each rm between t and t+1, we need to solve
t
t
four linear-programming problems for Dt0 xt ; yt , Dt1
0 x ; y ,
t1
t
D0 xt1 ; yt1 and D0 xt1 ; yt1 (Fare et al., 1994). The outputoriented LP problem for estimation of Dt0 xt ; yt under variable returns
to scale is as follows:
1
d0 xt ; yt
max?; ;
st yit Y i 0;
xit Xi 0;
i 0;
X
i 1; i 1; : : : : :; n:
where is a scalar and is a nx1 vector of constants. The lineart

t
programming problems for the estimation of Dt1
0 x ; y ,
t1
t
t1
t1
t1
t1
D0 x ; y and D0 x ; y are similar to the earlier formulation with corresponding adjustments. For details, see Fare et al. (1994).
part ii
Development of indigenous
innovation capacity and catch-up
The role of state policy in shaping

innovation practices: The case of open
innovation in China
6.1 Introduction
Open innovation (OI) has been an emergent concept in innovation
studies and a major trend in practice attracting wider attention from
academics, policy makers and practitioners. It refers to the use of
purposive inows and outows of knowledge to accelerate internal
innovation and to expand the markets for external use of innovation,
respectively (Chesbrough, Vanhaverbeke and West, 2006). Many rms
started to implement open innovation as a necessary organisational
adaption to changes in the environment (Chesbrough, 2003). Firms
may open up their innovation processes in two dimensions
(Lichtenthaler and Ernst, 2007). Inbound open innovation refers to
inward knowledge acquisition and to leveraging the discoveries of
others because rms need not rely exclusively on their own R&D
(Chesbrough and Crowther, 2006). Outbound open innovation refers
to outward technology transfer, and it suggests that rms can look
for external organisations with business models that are suited to
commercialising a technology exclusively or in addition to its internal
application (Chesbrough and Crowther, 2006). Generally, rms can
combine both technology exploitation (inside-out OI) and technology
exploration (outside-in OI) to create maximum value from their
technological capabilities or other competencies (Lichtenthaler, 2008).
Is open innovation a new phenomenon in China, or it is old in
practice and new in concept? Much of the extant literature indicates
that since the reforms, the Chinese government has encouraged rms to
source external knowledge by acquisition of foreign technology. More
recently, since 2000 it has encouraged Chinese rms to go global,
to actively search for and acquire relevant foreign technology. Some
pioneering Chinese rms have adopted various modes of open-oriented
141
142
Development of indigenous innovation capacity and catch-up
strategy to manage innovation activities (Chen and Chen, 2008). What,

therefore, is the current status of open innovation in China? What is the
role of open innovation in spurring innovation upgrading in China?
What are the possible trends and challenges of Chinese rms while
adopting open innovation? Although the era of open innovation has
become a reality for most Chinese rms, we still lack a systematic
understanding of the context, environment and mechanisms, inside
and outside of the organisation, and when and how to fully prot
from the new model for latecomer rms in emerging economies. This
chapter aims to ll the gap in the literature by providing a systematic
review of the evolution of the policies and practices of open innovation
in China. It also explores the implications of open innovation strategy
for latecomer rms in building indigenous innovation capability and
competing in todays global innovation networks. As far as we are
aware, this is one of the rst systematic review studies on open innovation policies and practices with a focus on the context of emerging
economies.
The rest of the chapter is organised as follows: Section 6.2 provides an
overview of the current situation in openness in innovation in China.
Section 6.3 discusses the theoretical framework, especially the role of
government policies in shaping innovation mode and practices in rms.
Section 6.4 examines the policies introduced by the Chinese government
in the past three decades that affected the development of open innovation in China. Section 6.5 reviews the rm-level practices of open
innovation in China. Section 6.6 discusses the trends and challenges of
open innovation in China. Section 6.7 presents conclusions.
6.2 Openness in innovation in China: an overview

China has experienced a substantial transformation in the innovation
landscape since the economic reforms and opening up in 1978. In the
early stage of the reforms, Chinese rms, motivated and facilitated by
various policies, relied on foreign technology transfer through imports
and foreign direct investment (FDI) for technology upgrading. This
contributed to the new to the rm or country imitation-type diffusionary innovation. Since the late 1990s, Chinese rms have started to
transit from a technology-transferfocused development strategy to an
indigenous innovation-oriented growth strategy (Fu and Gong, 2011).
Especially since Chinas entry into the WTO in 2001, a rapid
The role of state policy in shaping innovation practices
143
internationalisation of innovation activities in China by both foreign

and indigenous rms has occurred. The latecomer rms in China began
to learn how to utilise external R&D resources or technological knowledge via acquisition of foreign technology or equipment; joint venturing; and collaborating with users, suppliers and public or private
research institutions (Chen, 2009). In recent years, some domestic
rms also set up local and overseas R&D centres and have acquired
foreign technology-intensive rms through mergers and acquisitions
(M&A) (Bai, 2009). On the other hand, some Chinese rms have also
started to use licensing, intellectual property selling, spin-offs as well as
corporate venturing to exploit external opportunities in commercialising the inventions they possess.
Although open innovation is a new concept, a rm-level national
innovation survey carried out by the National Bureau of Statistics of
China and the Research Centre for Technological Innovation at
Tsinghua University in 2008 suggests a relatively high degree of adoption of open innovation in Chinese manufacturing rms.1 Tables 6.1
Table 6.1 Sources of product innovation
Source
Internal
Afliated group (domestic)
Afliated group (overseas)
Collaboration with other rms
or research institutes
Other rms or research institutes
Number of
rms
Proportion
(percentage)
732
130
101
357
75.9
13.5
10.5
37.0
40
4.1
Note: Firms are allowed to make multiple choices.
This survey covers all industrial sectors in the Chinese economy. The valid sample
includes 1,408 rms in 42 cities across the country. The survey asks questions
regarding rms general innovation input and performance in the three years from
2005 to 2007, which contains sufcient items to measure various indexes of
innovation activities for those indigenous rms in China. The method and types of
questions used in this innovation survey are primarily based on the Oslo Manual
(OECD, 1997, 2005) and also the core Eurostat Community Innovation Survey
(CIS) (Stockdale, 2002; DTI, 2003), although with some adaptive adjustments
because of the different survey context.
144
Table 6.2 Sources of process innovation
Source
Internal
Afliated group (domestic)
Afliated group (overseas)
Collaboration with other rms
or research institutes
Other rms or research institutes
Number of
rms
Proportion
(percentage)
820
123
81
351
86.8
13.0
8.6
37.1
46
4.9
and 6.2 report the sources of product and process innovation in the
sample. Intra-rm innovation and collaboration with other rms or
research institutes appear to be the major sources of product and
process innovation. Taking product innovation as an example, on the
one hand, most Chinese rms develop a new product or new process
heavily based on their own resources and capability (75.9 per cent of
rms innovate by themselves). On the other hand, more than a third
of the rms acquire new innovations by co-operation. Although the
phenomenon of so-called closed innovation shows a relatively higher
proportion, we still cannot ignore some surprising differences. More
than one-third of product innovations stem from industrial collaboration or research alliance (projects), which shows collaborative innovations should also be an important part of all types of innovation
activities in China. In addition, some rms innovations are still dependent on their afliated domestic or overseas groups. This mainly reects
the situation of those joint venture (JV) rms in China. An interesting
nding is that a relatively higher proportion of domestic groups resources have been used than those of overseas groups. This refers to those
joint ventures that are beginning to incline to use domestic resources to
develop new innovations in the Chinese market. Whats more, the
condition of process innovation is quite similar to that of product
innovation, although the latter relies less on closed innovation.
Table 6.3 reports the importance of different types of information in
the sampled rms. Not surprisingly, about 42 per cent of rms regard
internal sources (itself or its afliated groups) as having a relatively
higher importance compared to most other sources, which complies
Table 6.3 The importance of information sources in innovation activities
Information sources
Internal
Market information
sources
Institution information
sources
Other sources
Firm itself or its afliated group

Suppliers of equipment, raw material,
component or software
Users or consumers
Competitors or other rms within
the same industry
Consultants or private R&D institutes
Universities
Government or public R&D institutes
Government S&T plan
Trade or sales program, academic
conference
S&T journal or literature, trade
or patent literature
Professional industry association
Proportion of rms that regard

the source as being of high
importance (percentage)
Mean of
importance
41.7
17.9
3.09
2.63
52.5
34.3
3.25
2.96
5.7
8.3
10.0
14.4
22.5
1.96
2.06
2.12
2.21
2.61
9.7
2.30
18.6
2.52
146
with the heavy dependence on internal resources for developing new

products or processes as mentioned earlier. Regarding various external
sources, the differences among them are very signicant. Undoubtedly,
users (clients or consumers) are ranked rst in importance for innovation research, which accounts for 52.5 per cent in all, together with the
highest degree of importance (3.25). In particular, the index of degree of
importance suggests more about the depth of one specic information
source, whereas the proportion of importance only measures the
breadth or popularity of one source adopted by rms. Therefore, we
can nd that both users and competitors (including other rms within
the same industry) can be viewed as the most important sources for
research innovation. In addition, another important source can be
traced to those trade (or sales) programmes and academic conferences
(22.5 per cent), and even industrial associations (18.6 per cent).
Furthermore, we nd most important external sources are market
channels or information (including users, competitors, suppliers etc.),
whereas institutional information is regarded as less important. To sum
up, external sources play a crucial role in the innovation process and
present a relatively high scale of importance for most domestic rms,
whereas a large proportion of domestic rms still employ internal
channels for sourcing innovation.
6.3 The role of government policy in shaping

innovation practices of rms
A government can play an important role in framing and promoting a
credible innovation policy. In theory, three major roles for government can be identied: a role as a broker, a role in demand articulation
and a role in stimulating innovations (Faber, Kemp and Van der Veen,
2008). In this section, we discuss these three roles in relation to open
innovation practice.
First, as a broker, government commits to bringing together different
players in the innovation system. Open modes of innovation activities
need various players or platforms co-operating and complementing in
acquiring mutual benets. In particular, national framework conditions
and the public infrastructure for research cultivated by government
policy play a major role. To some extent, this corresponds to the
environmental side of public policy, which can greatly inuence rmlevel behaviours. For example, one particular policy issue, related to
147
intellectual property rights (IPR), may receive much attention. Till now,
the immature intellectual property protection regime in China has been
subject to long-running criticism by many foreign companies that enter
the Chinese market. However, the globalisation of R&D and the emergence of open innovation strategies in rms clearly raise intellectual
property issues. The shift towards IPR sharing in open innovation
strategies may require different kinds of management tools in universities and public research organisations. While strong IP protection can
attract R&D-related FDI, excessively strong protection can act as a
barrier towards open innovation strategies that rely on knowledge
sharing and access (OECD, 2008). Some scholars suggest that the platforms and repositories for the intellectual commons are greatly
required and can be facilitated by government regulations and investment in a strong Information and communication technology (ICT)
infrastructure. Another example that needs to be highlighted is the
building of efcient technology transaction markets. Most rms source
knowledge in various ways, but licensing and purchasing technology
and knowledge embodied in patents or other forms of IP are important.
Creating rules and conditions that facilitate the development and transfer of technology can facilitate open innovation (OECD, 2008).
Second, government can play a role in demand articulation. This
amounts to the creation of markets for open innovations by setting
relevant standards. For instance, markets can be created through the
use of economic incentives and special agreements with industry. In
general, it is possible to pull open innovation in a way that benets
economic and political circumstances; the theory behind this is usually
related to the work of Michael Porter, who argued that countries may
increase their competitive advantage by setting strict entry or environment standards (Porter, 1990). This idea is of great importance for the
market to be competitive and open to innovative newcomers, which
actually implies a competition policy towards open innovation.
Designing a competition policy that does not preclude cooperation
is an important challenge, especially in industries where excessive
competition can produce innovation (OECD, 2008). Besides, when
government builds more commercialisation channels (such as science
parks as incubators and industrial technological strategic alliance),
most rms, particularly new entrants in high-tech or emerging industries, would be more inclined to adopt open modes of technological
exploitation activities.
148
The third role of government takes effect on the supply side of the
national innovation system. This is embodied in various structural
public policies (science and technology [S&T] policy, scal policy,
capital markets etc.). Most OECD countries S&T policies are predominantly national in scope, but it is becoming clear that policies designed
for geographically circumscribed knowledge-based activities or for vertically integrated value chains of rms need to be reviewed (OECD,
2008). For example, policies to promote networking and clusters can
have a great inuence on promoting rms to adopt an open mode of
technological exploration in their R&D processes. Another prominent
example can be seen in governments scal policies, which mainly
supply a large number of tax incentives, scal subsidies, preferential
credits and direct capital input to stimulate innovative collaboration
activities. Besides, policies related to dynamic capital markets supported
by government do matter because a free and ourishing nancial environment can greatly stimulate open innovations. Insofar as corporate
venturing is one channel through which rms add value internally and
externally, capital markets that allow for corporate venturing and exit
to secondary markets are important for open strategies (OECD, 2008).
Finally, government policies favouring open innovation can also help
many latecomer rms catch up on technological innovation. Whatever
the normative or regulatory styles, policies have had a strong inuence
on the catching-up process (Li and Kozhikode, 2009). Hobday (1995)
observed that governments in emerging economies often protected local
rms from initial failures and encouraged them to learn from MNEs.
Kim (1997) observed that the intellectual property regime in South
Korea supported local rms by allowing them to imitate the technologies of global players during the early stages of catching up. However,
this support was only available until the Korean rms began to develop
their own innovative offerings. Although the initial innovations of the
latecomers were not ground breaking, over time and with increased
investment in R&D, both internally and in collaboration, these rms
gradually were able to develop more innovative offerings (Hobday,
1995). In addition, open innovation practice can sometimes inuence
the changes in the governance of innovation policies (OECD, 2008). On
the one hand, the challenge for government is to help rms adjust their
innovation strategies to a changing environment; on the other, many
rms, in particular latecomer rms, actively respond and react to the
changing of government policies. Hence, understanding the typical
149
patterns of open innovation practices in China would have signicant

implications for government policies.
In sum, government may have a signicant impact on the innovation
mode, practices and performances at the rm level through the design
and implementation of relevant innovation and industry policies. A
systematic examination of such policies and the corresponding innovation practices adopted by rms as a result of such policy inuence will
enable us to have an in-depth understanding of the evolution of innovation behaviour of Chinese rms, especially the push and pull factors
that motivate rms and how they shape the innovation practices in
Chinese rms.
6.4 Policies affecting open innovation in China

The governments of a growing number of emerging economies have
played an active role in making their respective economies attractive to
global R&D investments (UNCTAD, 2005). In the past few decades,
Chinese policy makers have attempted to lead the entire innovation
environment in being more open and co-operative although open innovation is not regarded as a clear national strategy. It is expected that
effective policy support could help most Chinese rms overcome many
pressing obstacles in order to cultivate a highly open condition in the era
of indigenous innovation in China. For instance, open innovation can
be considered a private-collective innovation model. Instead of the
private investment model of innovation with Schumpeters temporary
monopolistic prots, the free revealing of inventions, ndings, discoveries and knowledge is a dening characteristic of the open innovation
model (von Hippel and von Krogh, 2003). This implies the corresponding role of government in stimulating the free and efcient ow of
knowledge and technology in the domestic market.
The core development strategy of the Chinese government is to
encourage indigenous innovation and to build an innovation-oriented
economy in the new century. Many S&T, industrial and macroeconomic policies are given much attention to stimulate technological
innovation and improve competitiveness. As we know, the open business model emphasises absorbing external knowledge and exploring
external paths to commercialisation (Chesbrough, 2006). Accordingly,
Chinese policy makers have continued to make the domestic market
and institutional environment more open and favourable to innovation
150
ow. The most recent indigenous innovation policy stemmed from

two revolutionary documents: the Decision on Implementing the
Outline of the Scientic and Technological Plan and Enhancing the
Independent Innovation Capacity and the National Guideline for
Medium and Long-term Plan for Science and Technology
Development (20062020) (Fang, 2007; Wang and Liu, 2007). The
open-oriented innovation ideology has been revealed clearly in the
initiating of the national strategy of indigenous innovation within
these two policy documents, which regard integrated innovation and
innovations on the basis of introduction, digestion and absorption as
two basic forms of indigenous innovation. In fact, both these innovation
forms emphasise the utilisation and integration of external R&D
resources and commercial paths. Regarding the development of
relevant public policies for open innovation in China, many signicant
innovation policies are indeed involved, although they still lack specifics. We classify them into three main policy clusters.
Policies towards inbound open innovation

Inbound open innovation refers to the acquisition of external technology
in open exploration processes. In accessing and sourcing external technologies and knowledge, EIRMA (2004) distinguished some modes:
purchase of technology, joint venturing and alliances, contract R&D,
collaborations with universities, equity in university spin-offs and equity
in venture capital investment funds. In fact, some of them, such as
international technology transfer and assimilation, are much more common in China because of the many relevant policies stimulating inbound
open innovation for indigenous rms. Early in the 1980s, most Chinese
rms introduced a large number of foreign technologies and equipment
to attempt shortening the huge gap in R&D and also in manufacturing
capacity. This open strategy of technology introduction, digestion and
absorption was directed and facilitated by trade policies relating to
imports of equipment and machinery and the encouragement of inward
foreign direct investment. The Regulations on the Administration of
Technology Acquisition Contracts (1985), Regulations on Promoting
Technology Introduction, Digestion and Absorption (1986) and
Regulations on Administration of Import and Export of Technologies
(2001) all encouraged the import of advanced and useful technologies.
Correspondingly, the Notice of Taxation on Issuing State Industrial
151
Technology Policies (2009) clearly emphasised making extensive efforts

in international co-operation and exchange and reinforcing technological
introduction, digestion and re-innovation. For instance, introducing key
components that make it difcult to achieve localisation could take
50 per cent to 70 per cent of tax incentives when their proportion of
introduction is less than 40 per cent. Moreover, enterprises that undertake key scientic and technological projects for the state should be
exempted from import duties and import value-added taxes for the
domestically unavailable key scientic research apparatus, facilities,
raw materials and components imported by them. Those policies helped
domestic rms effectively break technology barriers in the early stages
and improved manufacturing and technology application capability in a
relatively short time.
In addition, to encourage the introduction of advanced technologies
and promote industrial restructuring and technological progress, a
large number of stimulating policies were issued to attract foreign
companies and protect common interests of co-operation between
foreign and domestic rms. Early in the 1980s, not only supplying
taxation incentives, many relevant policies also strengthened the
nancial support for foreign-funded enterprises and encouraged
foreign investment in the midwestern regions, together with guiding
them to invest in key industries and important elds. Admittedly,
foreign technology transfer is only one of the measures relating to
open innovation, in particular, the inbound aspect of open innovation,
which refers to inward technology transfer (Lichtenthaler, 2009).
Although solely relying on foreign technology transfer may weaken
the incentives and capabilities of indigenous innovation, it has played
an indispensable role in helping Chinese rms source external knowledge and produce new to the country or rm imitation-type diffusionary innovation in the past three decades. For the vast developing
countries that are still at the early stage of takeoff, this channel of open
innovation is still an important policy choice.
Whats more, policies supporting indigenous rms going global are
also on the increase. The Chinese government supports going global not
only by promoting technology and product export but also by encouraging overseas investment and acquisition. For instance, the special
loans for Chinese overseas investments since 2004 are mainly used for
supporting such enterprises as overseas resource development projects
(which can make up for the relative insufciency of domestic resources),
152
overseas productive projects and infrastructure projects (which can give

impetus to the export of domestic technologies, products, equipment,
and labour services etc.), overseas R&D centres (which may utilise
internationally advanced technologies, management experiences and
professional talents) and overseas enterprise M&A projects (which
can improve the international competitiveness of enterprises and
accelerate exploration of international markets). In sum, a long history
of favouring inbound open innovation for acquiring knowledge and
resources has been the most crucial reality in the evolution of Chinese
policies promoting innovation in an open way.
Policies towards outbound open innovation

Another primary innovation policy cluster in China is designed to
accelerate outbound open innovation activities. Outbound open innovation describes the outward transfer of technology in open exploitation processes. It also refers to earning prots by bringing ideas to
market, selling IP, and multiplying technology by transferring ideas to
the outside environment (Enkel, Gassmann and Chesbrough, 2009).
Proting from the high speed of economic development, especially
together with domestic market expansion, the Chinese government
has enacted various measures to broaden the channels of outbound
innovation. After institutional transformation towards building a
market economy, the government clearly pointed out the relevant
policies for releasing the paths of commercialisation and industrialisation of S&T achievements, which means diversifying the innovation
channels. An initiating policy document, the Decision on Strengthening
Technical Innovation, Development of High-tech and Realisation of
Its Industrialisation (1999), constructed the foundation for enriching
various business models for developing high-tech products. The Law
of Promoting the Transformation of Scientic and Technological
Achievements (1996) clearly and creatively demonstrated the strategic
principles and plentiful modes of commercial and industrial innovation
knowledge. It suggested that the holders of S&T achievements may
have their knowledge commercialised via investing in it themselves,
transferring it to another party with their achievements as the conditions for co-operation or investing their achievements as trade-in or
converted shares or as proportions of the contribution to the investment. Besides, to adopt new inventions, an enterprise may publish
153
information on its own or entrust an intermediate institution engaged in

the trade of technology to solicit the S&T achievements, and it has the
right to conduct the transformation of S&T achievements independently or jointly with domestic or foreign enterprises or institutions or
other collaborators.
Moreover, the government also issued a series of policies in support
of taxation, nance and infrastructures as well as talent incentive
mechanisms to promote processing of those commercial and industrial
modes, such as specic or risk funds or preferential tax policy. The
primary reason is that such S&T achievements or inventions need
substantial investment, involving considerable risks and promising
high yields. Furthermore, the state is increasingly exploring diversied
investing channels and building multiform business models for proting
from innovation. For instance, the government has greatly developed
venture capital markets to nance innovation. It has widely enabled
many indigenous rms in recent years to expand nancing channels,
absorb the social capital, establish and develop the social industrial
investment fund aimed at reforming the traditional industries by high
technology. In particular, building a venture capital mechanism and
developing public venture capital institutions are both emphasised by
the existing policy goals. The state also attaches importance to the
fostering of talents for management and operation of venture capital,
to the establishment of the venture capital system and venture capital
funds composed mainly of social investment and for the multi-input
structure of venture capital. In sum, the institutional environment
supporting external commercialisation of new technologies has been
greatly improved in recent years. Both indigenous and foreign rms
could benet from innovation by the utilisation of various channels
to capture useful opportunities to commercialise new products or industrialise new inventions. Admittedly, the actual business environment
advocating open innovation is still at a relatively low level compared to
other developed economies. A more open and more reliable business
environment for commercialising innovation would be a long-term
focus for policy makers in China.
Policies towards open innovation networks

The global innovation network that links people, institutions (universities, government agencies, etc.) and other companies within and across
154
different countries has developed rapidly (OECD, 2008). In response,

the innovation policies in China have tended to be more aggressive and
more open to building open innovation networks in China.
The rst policy has been to establish production-study-research combination networks that centre on the enterprises and jointly bear the risks
and to establish the open technological innovation service systems relying
on cities, suggested by the State Industrial Technology Policies of 2002.
In this document, the government greatly advocates establishing the
production-study-research associations of different enterprises, domestic
universities and public or private research institutes to form a marketoriented R&D system. And the open production-study-research mechanism based on the comparative advantages and strategic needs of China
could select the areas of independent development and advances, explore
new technology routes through system integration and mutual absorption and develop new technology with independent intellectual property
rights. In addition, regarding global R&D co-operation projects or international strategic alliances starting to emerge in China, many relevant
institutional supports have been encouraged to come forth. As the focus
shifted from purely internal R&D activities to outside innovation, rms
that do not co-operate or do not exchange knowledge would reduce their
knowledge base on a long-term basis and lose the ability to enter into
exchange relations with other rms (Koschatzky, 2001). This group of
policies has been advocated and implemented by the Ministry of Science
and Technology more and more often in recent years.
Moreover, the development of the technology transaction market is
another important policy arena favouring and reinforcing open innovation networks. This transaction market can greatly facilitate the ow
of technological knowledge across rms and other organisations.
Generally, it includes technology transfer, technology consultation, technical services, technical training, technology-equity share exchange, technology intermediation and various research-production co-operations.
At all times, active technology equity exchange and transaction markets
also play a signicant role in creating national networks of open innovation in China. China set up its rst technology stock exchange market in
Shanghai in 1999. Since then, large-scale technology equity platforms
have emerged in Beijing, Chengdu, Shenzhen, Wuhan, Tianjin and
Shenyang and accelerated the outbound trade of technological innovation as well as corporate venturing in commercialising internal new
technology, products and large numbers of patents. These transaction
155
markets were used for immediate innovation service institutions as well

as for science parks and incubators and helped indigenous rms broaden
their commercialisation channels and external technology resources.
However, the role that the technology transaction market has played
in Chinas regional innovation system was not signicant over the 1998
2004 period (Fu, 2008).
The lack of protection of intellectual property rights is seen as the
major obstacle to innovation networks building (OECD, 2008). The
protection of intellectual property has been a part of the reform and
openness policy in China and an important institutional factor for
promoting scientic, technological and cultural development. To
adapt to necessary changes in global innovation networks, China
has in recent years speeded up its IP legislation and successively
promulgated such laws as the Patent Law of the Peoples Republic
of China, the Law of the Peoples Republic of China on Technology
Contracts, the Copyright Law of the Peoples Republic of China and
the Law of the Peoples Republic of China for Countering Unfair
Competition, which have begun to link up with international standards and have played a positive role in promoting reform and openness in China. As it is only recently that China has established its IP
system, a sufcient understanding of the importance of protecting IP
is still lacking in certain regions and departments. Specically, the
Decision on Further Strengthening the Intellectual Property
Protection (1994) has introduced some important regulations on
managing IPR but still lacks all-around considerations and effective
implementations. Till now, it has been widely realised by policy
makers that the IPR regime plays a major role in getting the national
framework and business environment right, which not only helps
standardise the innovation among local organisations but also
increases the possibility of sharing knowledge from foreign entrants.
In addition, the IPR regime will be managed in a more open way
(OECD, 2008) because many studies have reported that the use of
patenting has evolved from a focus on defensive applications as part
of business and management strategy (e.g., licensing, building a
patent portfolio) to exploitation of a nancial asset (i.e., attracting
external sources of nancing).
In sum, policies affecting open innovation in China have been
introduced for a long time. They are, however, not comprehensive.
There is a need to build a more comprehensive policy system in China.
156
Based on a report on the evolution of Chinese S&T, economic and

innovation policy from 1979 to 2008 (Wu et al., 2009), Table 6.4
shows selected policies illustrating the evolution of public policy for
open innovation.
Table 6.4 The main public policies for open innovation in China
Year
Policy document
1986
Regulations on Promoting
Technology Introduction,
Digestion and Absorption
1986
1993, 2007
1996
1999
Policy focus and main measures
Technology exploration; tax

incentives to import
advanced technology and
technology learning,
localisation.
Provisions on the
Technology exploration; tax
Encouragement of Foreign
incentives for foreign
Investment
investment rms in the
technology-intensive sector.
Law on Science and
Coupled process; overall level
Technology Progress
of nancial input; R&D
expenditure as cost, credits
and loans; open to domestic
foreign organisations or
individuals to fund S&T,
build information exchange
and networks and security
systems to protect IP.
Law on Promoting the
Technology exploitation;
Transformation of Scientic
support measures including
and Technological
initiation funds, discount
Achievements
loans, subsidy funds, venture
capital and other special
establishment funds; a
preferential tax policy, loans,
venture funds; build data
bank of S&T achievements.
Plan for Vitalising Trade by
Technology exploitation; tax
Science and Technology
and export preferential
support for high-tech
products.
157
Table 6.4 (cont.)

Year
Policy document
1999
Decision on Strengthening
Technical Innovation,
Development of High-tech
and Realisation of Its
Industrialisation
2002
2002, 2009
2006
2006
Technology exploitation; R&T

input, tax preference, loans
and credits and other direct
nancial input for
commercialisation of S&T;
management reform of
domestic S&T institutions.
Measures for the
Coupled process; developing
Administration of
industrial technologies and
Technological Innovation
experiments, the popularised
Plans of the State
application of new
technologies, joint
development, technological
centres and service system,
and the trial production of
new products, etc.
State Industrial Technology
Technology exploitation; scal,
Policies
tax, investment; nancial
and government
procurement policies to
provide support for
enterprises to increase
investment in innovation.
Decision on Implementing the Technology exploitation;
Outline of the Scientic and
encouraging selling IP and
Technological Plan and
multiplying technology by
Enhancing the Independent
transferring ideas to outside,
Innovation Capacity
independent innovation and
innovative talent strategy;
cultivating a batch of largesize enterprises, enterprise
groups and technical
alliances.
National Guideline for
Coupled process; tax, loans
Medium and Long-term Plan
and credit measures to
for Science and Technology
support introducing external
Development (20062020)
R&D and innovation,
158
Table 6.4 (cont.)

Year
Policy document
together with government

procurement, especially for
indigenous innovation
products.
2008
Notice on Printing and
Distributing the Outline of
improving the intellectual
the National Intellectual
property law enforcement
Property Strategy
and administration systems
through nance, investment,
government procurement,
industry, energy and
environmental protection
policies.
2009
Opinions on Encouraging
Technology Export
encouraging IP exchange
(out-licensing and
multiplying technology) and
high-tech export.
2008, 2009, Notice of Promoting on
Coupled process; encouraging
2010
Industry Technological
domestic enterprises to form
Innovation Strategic Alliance
technical R&D institutions
outside China by way of joint
venture, co-operation,
merger, etc.; set up R&D
alliances and utilise overseas
superior forces to develop
industrial technologies with
independent IPR.
6.5 Practices of open innovation in China

The primary modes of open innovation in China represent some basic
characteristics of those indigenous rms in opening their innovation
processes. The closed and open models of innovation are typically
presented as two extremes of a spectrum ranging from doing everything
in house (vertical integration) to outsourcing everything to external
159
partners (OECD, 2008). Following Enkel, Gassmann and Chesbrough

(2009), we also divide open innovation practices into three categories:
outside-in, inside-out and doubled process. According to our review of
relevant rm-based cases adopting open innovation, Table 6.5 shows
some typical modes of open innovation and their characteristics in
practice.
Evidence from these case studies suggests three characteristics of open
innovation in China. First, the proportion of rms adopting outside-in
process, or technology exploration modes, is high. This fact suggests
that most Chinese domestic rms are inclined to acquire advanced
technology or knowledge through external sourcing, networking, cooperation or acquisition. Not surprisingly, most Chinese rms, as latecomers in the global innovation network, greatly expect to acquire
complementary assets and enhance innovation capability with methods
of technology exploration.
Second, the proportion of inside-out process, or technology exploitation, still stays at a lower level. As it broadly weakens technological
accumulation and innovation capability, it is difcult for many domestic rms to adopt licensing or corporate venturing such as spin-off
whereby a company splits off sections as separate businesses or spins
out when members of a company or university set up a new business
often based on a new technology. Limited knowledge resources make
business models gradually outdated, and the immature venture capital
market in China does not facilitate the growth of those new business
models.
Third, an interesting pattern relates to differences between sectors
and rm size. Generally, companies enjoying fast and medium-level
growth use the inside-out process actively, although it is still less than
the intensity of outside-in process usage (Gassmann et al., 2010).
However, many rms in traditional or low-growth sectors also use
inside-out modes of innovation in China. In particular, this phenomenon occurs more often when production-study-research combination
programmes are well established or involved. Moreover, it is expected
that only large multinationals have an active out-licensing strategy to
which they allocate substantial resources (Lichtenthaler and Ernst,
2007; OECD, 2008). In China, however, some start-ups in high-tech
or emerging industries would be apt to out-license their new inventions
with forms of contractual collaboration to speed up the commercialisation of new technology.
Table 6.5 Selected cases of open innovation in Chinese rms

Modes
Main Forms
Company
Industry
Source
Coupled process
Knowledge sourcing from

domestic universities, private
research institutions and
consultancy
Co-operation in R&D with
overseas partners
Absorption of foreign technology
Setting up overseas R&D
centres
Selling IP and spin-off internal
know-how
Joint-venture with MNEs
Production-study-research
combination, in particular
with universities
External networking with
suppliers
Outsourcing R&D
combination, in particular with
universities and private research
institutions
ChunLan (in Taizhou, Jiangsu

Province)
Electric equipment
Wang (2006)
FeiYue (in Taizhou, Zhejiang

Province)
Electric equipment
(sewing
machine)
Zhu &Chen
(2007)
Little Swan Co. Ltd (in Wuxi &

Hangzhou, Zhejiang Province)
Electric equipment
(washing
machine)
Li (2007)
JiangHuai Auto (JAC) (in AnHui

province)
Automobile
Tang & Zhao

(2007)
Coupled process
Outside-in
process
Outside-in
process
Coupled process
Outside-in
process
Coupled process
Coupled process
Coupled process
Coupled process
Absorption, digestion and

innovation
Building globally distributed
R&D centres
Co-operation with leading
giants
Outsourcing R&D
Knowledge sourcing from
other industries and leading
universities
Co-operation through
production-study-research
combination
Building overseas R&D centre
Participating strategic
technology alliance
combination
Transnational mergers and
acquisitions (M&A)
External networking
Outbound
commercialization
with restructuring industrial
value chain
Joint venture to market globally
International Marine Containers

(Group) Ltd. (CIMC) (in
Shenzhen)
Equipment
manufacturing
(Machinery)
Zheng, He
and Chen
(2008)
Neusoft (in Shenyang, Liaoning

Province)
Software (medical,
service)
Liu (2008)
Shenyang Machine Tool (Group)

Co. Ltd
Equipment
manufacturing
(machinery)
He (2008)
Dongbei Special Steel Group Co.

Ltd (in Liaoning Province)
Steel
Liu & Hou

(2008)
Nanjing Automobile
Automobile
Guizhou Special agriculture, i.e.,

Yanhuang Shiye Co. Ltd
Agriculture
Yu & Wang
(2008)
Zhang & Liu
(2008)
Table 6.5 (cont.)

Modes
Main Forms
Company
Industry
Source
Outside-in
process
Supplier and competitor

integration and also user
innovation
combination
Distribute
R&D
centres
globally
Absorption, digestion and
innovation
Transnational M&A
Strategic alliance
R&D outsourcing
International co-operation for
synergic R&D
Building
overseas
R&D
centres
Corporate venturing
Spin-off
BOE Technology Group Co. Ltd

(in Beijing)
Supplier of display
products
Jiang (2009)
Shenhua Group Corporation

Limited (Shendong Mine
Group, in Neimenggu
Province)
Coal
Sun & Xu
(2009)
Weichai Power Co. Ltd. (in

Shandong Province)
Automobile
(supplier)
Hou (2009)
Lenovo Venturing (Beijing),

Tsinghua Ziguang (UNIS,
Beijing), B&H Investment
(Shenzhen)
Venture capital
Xu et al.
(2009)
Coupled process
Coupled process
Inside-out
process
Source: Classied by the authors and referenced to relevant literatures presented earlier.
163
Collaborative innovation appears to be the most popular innovation mode. Firms may engage in collaboration to acquire missing
knowledge, complementary resources or nance; to spread risks; to
enlarge their social networks; or to reduce costs (Hoffman and
Schlosser, 2001). In addition, a rms innovation strategies can combine
characteristics of both innovation models, and the degree of openness
depends on factors such as the importance of the technology, the rms
business strategy and the industrys characteristics. Companies traditionally seek to retain their core capabilities and decide what to outsource or with whom to collaborate on innovation on the basis of the
previously discussed benets and the conditional factors (OECD,
2008). Here, we still use evidence from our survey data set to empirically explain the content of innovation collaboration in China.
Table 6.6 gives us the frequency results of different innovation
types. Most innovations, together with their funding and information
sources, are developed and sourced internally to a large extent as the
previous overview shows; thus, the proportion of internal R&D
activities accounts for the majority (those rms that have internal
R&D take up nearly half of our sample, whereas fewer than 10 per
cent of rms have no internal R&D). Purchasing equipment, machines
or software also occupies quite a large proportion (32.7 per cent of
rms). Whats more, it is necessary to highlight that external R&D
activities are not as popular as expected: more than 50 per cent of
rms admit that they never conduct external R&D activities.
Similarly, effective acquisition of external technological knowledge
is quite difcult for many sampled rms, with nearly half expressing
that they never conduct innovation by using external technological
knowledge. These results can reect, to some extent, that the innovation types of many domestic rms are still simply independent, and
most others are introduction-oriented or learning-oriented innovation. The frequency of using external technological sources generally
stays at a low level.
6.6 Trends and challenges of open innovation in China

Trends of open innovation in China
The organisation of innovative activities across rm boundaries is clearly
on the increase, with more balance between internal and external sources
Table 6.6 Types of innovation mode and frequencies

Never
Types of innovation mode
Internal R&D
External R&D
Purchasing machine, equipment
and software
Acquiring external technological
knowledge
Training and marketing
activities
Sometimes
Very often
No. of
rms
Proportion
(percentage)
No.
of rms
Proportion
(percentage)
No.
of rms
Proportion
(percentage)
Sum total
90
465
165
9.1
52.2
17.2
418
356
482
42.4
40.0
50.2
477
70
314
48.4
7.9
32.7
985
891
961
436
48.6
374
41.6
88
9.8
898
302
33.3
454
50.1
150
16.6
906
165
of innovation (OECD, 2008). To build an inimitable competitive advantage, many multinational companies globally integrate complementary
technology, capital, branding, channels and low-cost labour resources.
Transnational innovation becomes more popular (Fagerberg, Mowery
and Nelson, 2005). The latecomer rms in emerging economies face
a rather different situation while adopting open strategies because
of their unique policy institutions and innovation practices discussed
earlier.
The World Investment Report 2005 (UNCTAD, 2005a) illustrates
that some developing countries have an enhanced position in attracting
transnational R&D investment. In such a highly open environment,
latecomer rms could integrate globally all the potential resources of
technology, manufacturing, raw materials, channels and brand, which
means that Chinese rms should acquire the knowledge resources
distributed globally (Jiang, 2004; Xiong and Li, 2008). In fact, most
Chinese enterprises started to take advantage of R&D resources
overseas much earlier. In the 1990s, some Chinese companies in
Shanghai were the forerunners of Chinese overseas R&D investment,
and since 2000 this phenomenon has become generally more popular,
especially in the ICT industry. Nowadays, companies are much more
dependent on outside resources. In the Chinese high-tech industry, it is
increasingly difcult for rms to acquire core technology from MNEs in
China. This forces Chinese rms to go out to look for suitable R&D
resources by themselves. At the same time, it will be a new opportunity
for Chinese enterprises to benet from globalisation of innovation and
make great use of available international resources. The factors that
Gassmann (2006) recognised as driving higher performance of open
innovation models are in fact equally prominent in todays China.
Hence, the era of open innovation is ourishing in China.
Many Chinese enterprises have not developed their technology in a
closed model. On the contrary, they have learned to take advantage of
global R&D resources, for example, by introducing and absorbing
advanced foreign technology and then further studying it. In addition,
they can actively participate in international scientic and technological exchanges and co-operation for the realisation of open innovation. For example, a Chinese leading IT company, ZTE, set up an
R&D organisation overseas, does its architectural R&D overseas and
its nal product innovation at home. This has produced outstanding
results.
166
Challenges of open innovation in China

Latecomer rms that want to benet from openness face a number of
challenges. First, as Chinas economy becomes highly open and
the degree of domestic competition becomes increasingly high, most
indigenous rms need to learn how to build up the crucial competence
to make good use of open innovation in a highly open environment.
Admittedly, companies investing in open innovation activities face risks
and barriers that hinder their proting from their initiatives (Enkel,
Gassmann and Chesbrough, 2009). Some Chinese scholars have
already criticised the possibly fatal disadvantage of blindly sourcing
external knowledge (Chen, 1994; Zhang, 2003). Considering future
competition and challenges from those latecomer players, most foreign
companies prefer not to spill over their core technologies. Many local
rms often complain that those technologies by the time of their
introduction were typically out of date. Without absorbing key and
advanced technologies, many indigenous rms may always lag behind
in technical upgrading.
Moreover, as a result of the weakness of absorptive capacity, many
Chinese rms cannot improve their internal R&D and long-term innovation capacity. Latecomer rms are unable to cultivate indigenous
innovation capability by only using simple inbound open innovation.
For example, international R&D co-operation requires the reorganisation of R&D personnel, R&D expenditure, an R&D facility and R&D
information distributed across different countries or regions. Because of
the limitations of their low absorptive and integrative capacity, it is
difcult for latecomers to share the R&D outcome. Till now, many
leading Chinese enterprises chose to use M&A. However, M&A may
not be suitable to achieve technology absorption if the acquiring rm
lacks technological capability. It is hard to make foreign R&D personnel and cross-national managers stay.
Finally, we need to be cautious about conducting open innovation in
different contexts, in particular regarding catching-up economies. Open
innovation is not an imperative for every company and every innovator
(Gassmann, 2006). Openness is not a choice for the rm but an outcome
of capabilities, industrial organisation and wider innovation systems.
Existing literature also argues that latecomer enterprises cannot rely too
heavily on global R&D resources (Nolan, 2001). The development of
the Chinese automobile industry proves this point. Most joint ventures
167
are unwilling to share their core advanced technology know-how and

invest too much in developing key technology in China because of their
worries about losing a leading advantage. For example, in the past few
decades, the vehicle models introduced by joint ventures in their early
days of entering the Chinese market have always been outdated models
from foreign collaborators. Except for the fact that the overall market
demand is relatively low end, the more intrinsic reason is attributed to
the fear and unwillingness of those foreign partners. Most Chinese
collaborators only make small incremental improvements and then
have to follow the old technological path. Therefore, latecomer rms
should have clear and rational expectations when choosing appropriate
partners.
6.7 Conclusions
Open innovation has been a new innovation paradigm and a future
trend widely accepted and adopted by scholars and practitioners in the
era of increasing globalisation. This chapter provides the rst systematic
review of the policies and practices of open innovation in China. In
general, open innovation is not a new phenomenon in China. In fact, it
has been a long-evolving and ongoing innovation mode adopted by
Chinese rms in the past three decades. In other words, open innovation
is a new concept towards innovation strategy based on a mix of old
and new practices in China. Although OI is a recent concept, both
science and technology and economic policies in China have to a great
degree incorporated the creation and facilitation of an open market
environment and innovation atmosphere and the encouragement of
external knowledge sourcing and acquisition as well as the commercialisation of scientic research discoveries and inventions. Although the
implementation and execution of policies at national, regional and
especially rm levels vary greatly across regions and rms depending
on the interpretation of the policies and the capabilities of local government and rms, Chinese rms have no doubt moved in the direction of
making good use of external knowledge via inbound or outbound open
innovation to build indigenous technological capabilities. As a strong
developmental state that is widely accepted in the literature, the Chinese
government has played a crucial role in the national innovation system
in initiating and coordinating a wide set of institutions and incentives.
All the innovation-related nancial, tax, industry, trade and S&T
168
policies have served as effective linkages that connect all the relevant
players at various levels of the national innovation system and ensure
that the national innovation strategy is well communicated from policy
makers at the top to rms at the lower levels.
The evolution of the orientation of policies and hence the practices
adopted by Chinese rms echo the different stages of economic and
technological development in China. From encouraging all-around
external sourcing since the 1980s, policy makers have diversied channels of commercialising innovation in the mid-1990s. Since the 2000s,
with the increase of globalisation and the growth of technological
capabilities of Chinese rms, policy orientation has become more
open and aggressive to incentivise indigenous companies to acquire
advanced external knowledge through going global. All this reects
the Chinese governments objective of promotion of indigenous
innovation in a highly open era. Despite the strong government policy
support to open innovation, however, some critical institutional
challenges, such as the lack of a strong intellectual property rights
regime to protect and facilitate knowledge sharing, still need urgent
attention and effective effort to reinforce them. Moreover, innovation
policies can no longer be designed solely in a national context. Because
of a countrys attractiveness as a place for innovation activity becoming
a priority, framework conditions that affect the location of production
and costs become critical (OECD, 2008). So, structural policies, which
include competition and regulations, public infrastructure, S&T base
and talent education policies, are all of great importance.
Increasing globalisation and changes in global innovation systems
have driven the shift to a more open innovation mode. Our review
suggests that selected Chinese domestic rms have adopted a variety
of OI practices; for example, Chinese rms in different industries have
often used the outside-in mode, the inside-out mode and the coupled
mode at different development stages. Therefore, future research should
explore systematically why and how latecomer rms can efciently use
open innovation for catching up. In particular, what is the role of OI in
emerging economies? How can the emerging economies make the best
use of open innovation for indigenous capabilities building? These are
important questions for an in-depth understanding of open innovation
for latecomer rms in emerging economies.
Technological learning is regarded as a necessary process for latecomers to improve their technological capabilities, skipping the
169
repeated technological manufacturing cycle and thus avoiding the huge

investments in technological systems in the initial stages in order to
catch up with the developed countries (Hobday, 1995). The ability to
use inside-out and outside-in strategies is facilitated by frameworks that
allow for the purchase or sale of intellectual assets that can create value
and opportunities for rms inside or outside their core businesses.
Therefore, both technology transaction markets and IPR protection
processes do matter in fostering open innovation. Moreover, rms
may differ in their modes of learning. Some learning modes are superior
to others in terms of the learning outcome in a particular situation
(Li and Kozhikode, 2009). While rms might have a variety of sources
to choose from, the path dependency of learning and the learning
outcome makes it difcult to gauge the value of a given learning source
a priori. Hence, another avenue for future research should be concerned
with establishing the best learning strategies for latecomers to deal
with various technology or organisation challenges in opening their
innovation processes. More in-depth industrial or rm-level case studies
or relevant large-scale empirical surveys should be carried out in the
future.
Open innovation as a response

to constraints and risks
7.1 Introduction
In recent years, open innovation has become a new imperative in
innovation practice and research. It refers to the use of purposive
inows and outows of knowledge to accelerate internal innovation
and to expand the markets for external use of innovation using outsidein, inside-out or coupled modes (Chesbrough, Vanhaverbeke and West,
2006). However, to date it has mainly been analysed in the context of
the industrialised economies, particular with respect to large, hightechnology multinational enterprises (MNEs) (Chesbrough, 2003;
Chesbrough and Crowther, 2006; Keupp and Gassmann, 2009). The
research on open innovation in smaller organisations or developing
countries is scarce, with some isolated exceptions (e.g., Vrande et al.,
2009; Lee et al., 2010; Fu, 2012a). Given the rising importance of the
emerging economies, understanding the determinants of open innovation in latecomer emerging economies is of crucial importance.
Perhaps the two most important questions to address are why and
when rms in emerging economies should adopt open innovation.
Firms may choose open innovation for different reasons, from tapping
into the best talents to keeping a rms leadership in innovation
(Chesbrough, 2003) to overcoming the various impediments that a
rm faces (Keupp and Gassmann, 2009). Firms in emerging economies
face substantial institutional, resource and capability constraints for
innovation. How do rms respond to these constraints and the risks
of innovation given their limited resources and capabilities? What role
can open innovation play in rms strategic responses to constraints and
risks? How relevant or useful are existing open innovation theories and
practices, primarily developed in Western countries, when they are
applied to the emerging economies such as China?
This chapter attempts to ll the gap in the literature by providing the
rst analysis of the drivers of open innovation in emerging economies.
170
171
The chapter has a special emphasis on rms adoption decisions in

response to the institutional, nancial and capabilities constraints and
risks. A national rm-level innovation survey data set of 1,408 rms in
the Chinese manufacturing sector is employed in the research. As one
of the major emerging economies, China opened up its economy after
1978 and has actively engaged in the acquisition of external knowledge and encouraged university-industry linkages ever since. Chinese
rms in most traditional or emerging sectors are still catching up with
their foreign rivals, especially in terms of technological innovation
capability.
Technological learning will be a long-term task for most latecomer
rms (Lee and Lim, 2001; Xie and Wu, 2003). Owing to a lack of
adequate resources and capabilities to deal with technological and the
market uncertainty of innovation as well as the rising costs of internal
R&D risks, most Chinese rms are not able to meet all the challenges of
innovation on their own and choose to source external knowledge and
resources to be able to accomplish the innovation process in full. Until
the late 2000s, rms collaborated with other partners and usage of
external information sources for innovation was widely observed in
the economy (Fu and Xiong, 2011). Therefore, China provides a good
case study.
Findings from this chapter contribute not only to our understanding
of open innovation in emerging economies but also contribute to the
empirical evidence on the linkage between impediments to innovation
and rms choice of innovation mode. Evidence on this question is not
currently available (with the exception of Keupp and Gassmann [2009],
based on data from Switzerland).
The rest of this chapter is organised as follows: Section 7.2 discusses
the theoretical framework. Section 7.3 describes data and methodology.
Section 7.4 presents the empirical results. Section 7.5 discusses the
ndings in detail and then presents conclusions.
7.2 Theory and hypotheses

Determinants of open innovation
The literature has pointed out various determinants of open innovation.
It is well recognised that companies should not innovate in isolation
but instead should co-operate with external partners throughout
172
the innovation process (OECD, 2008). Instead of using the usual

exploration-exploitation dichotomy to explain rms decision to
adopt open innovation, we deploy the push-pull factor framework.
Most prior contributions have argued that the motivation of open
innovation is a result of pull factors external to rms, such as environmental change and pressures; availability of skilled workers, knowledge
or venture capital; more intense competition from rivals, suppliers or
new entrants (Chesbrough, 2003); technology intensity and fusion
(Gassmann, 2006); knowledge transfer and leveraging of spillovers
(De Bondt, 1997; Chesbrough et al., 2006); and partner advantages
(Hagedoorn, 2002) that drive the externalisation of R&D activities
beyond the boundary of the rm.
Many rms have started to implement open innovation as a necessary
organisational adaption to changes in the environment (Chesbrough,
2003). As the boundaries of rms become porous (Lichtenthaler and
Ernst, 2007), companies interact intensively with their environment;
this leads to large volumes of external technology acquisition and
external technology exploitation (Chesbrough, 2003). Firms combine
both technological exploitation and exploration to create maximum
value from their technological capabilities or other competencies
(Lichtenthaler, 2008).
On the other hand, a series of internal constraints that rms face may
also push them to open up the innovation process. Keupp and
Gassmann (2009) explain the externalisation of R&D as a result of
weaknesses internal to the rm. They examine empirically how impediments to innovation spur openness at the rm-level based on the Swiss
Innovation Survey and show that information-/capability-related
impediments and risk-related impediments will increase both the
breadth and depth of open innovation.
Moreover, much of the literature emphasises the direct effect of
internal factors on open innovation activities. EIRMA (2003) shows
that R&D managers of large corporations engage in technological
venturing not just for market-related motives such as meeting customer
demands but also to acquire new knowledge. Vrande et al. (2009) in
particular claim that small and medium enterprises (SMEs) pursue open
innovation primarily for sustainable growth or to keep up with competitors. In addition, innovation collaboration studies suggest that
enterprises may engage in collaboration to acquire missing knowledge,
complementary resources or nance; achieve abundant revenue (i.e.,
173
sale of intellectual property [IP] and leveraging); spread risks; enlarge

social networks; or reduce costs (Mohr and Spekman, 1994; Hoffmann
and Schlosser, 2001; Gassmann, 2006). Koruna (2004) identied some
further motivations, including specic industry standards, attempts to
prot from infringements, realisation of learning effects and guarantees
of freedom to operate by establishing cross-licensing agreements with
other organisations.
In sum, there has been an important shift from a focus purely on
internal R&D towards emphasising external resources and partnership for innovation (e.g., Rigby and Zook, 2002; Christensen, Olesen
and Kjar, 2005). With globally dispersed specialised knowledge,
highly mobile skilled workers, abundant venture capital, greatly
reduced product life cycles, external options for ideas currently sitting
on the shelf as well as an ever increasing number of emerging and
effective competitors, most enterprises can no longer afford to innovate on their own. Firms have therefore begun to think of new ways to
harness external ideas while leveraging their in-house R&D outside
their current operations (Chesbrough, 2003; Gassmann, 2006;
Vrande, et al., 2009). Our analysis of the determinants of open innovation in the context of an emerging economy, China, focuses on the
push factors while recognising the effects of the pull factors arising in
the dynamic environment.
Innovation and open innovation: constraints and risks

Generally, obstacles to innovation include economic risk, costs, nancing constraints, rigidities of an organisation, lack of skilled personnel,
lack of information on technology and markets, lack of customer
responsiveness to innovation and institutions (Galia and Legros,
2004). In fact, these obstacles can vary among different types of rm.
A number of prior studies have reconstructed these obstacles into
specic sets or groups, such as nance-/risk-related obstacles, market-/
institution-related obstacles and information-/skill-related obstacles
(Baldwin and Lin, 2002; Keupp and Gassmann, 2009). Some of these
erosion factors, such as the increasing costs and risks of innovation,
greater competition or the increased complexity of new technologies,
may stimulate open innovation (Chesbrough, 2003).
If rms experience these constraints and risks to innovation, they are
often more likely to open up the innovation process to overcome
174
internal rigidities caused by constraints and risks. For example, many

rms have to look for partners with complementary expertise to secure
rapid access to new leading knowledge or to learn by networking.
In particular, for indigenous rms in the emerging economies, the
nance-/risk-related obstacles, market-/institution-related obstacles or
information-/skill-related obstacles of this group of rms could be quite
distinct because their context is that of catch-up in global innovation
networks (OECD, 2008). Firms are prone to miss many opportunities
because a signicant number of these will fall outside the organisations
current business or need to be combined with external technologies to
unlock their potential (Chesbrough, 2003).
Hence, we argue that open innovation is an active response for latecomer rms in emerging economies as a means to overcome internal
rigidities and reinforce their innovation resources and capabilities. This
perspective has already been supported by some case-based illustrations
(Rigby and Zook, 2002; Xie and Wu, 2003; Chen and Chen, 2008;
OECD, 2008). To our knowledge, this hypothesised relationship has
not yet been systematically tested in the context of developing countries
or specically in the case of China.
Our rst proposition is that weaknesses external to the rm, particularly environmental innovation pressure, spur open innovation in
China. Here, environmental innovation pressure refers to intense and
low-end market competition; uncertainty of customer needs or lack of
responsiveness to innovation; and institutional challenges such as a
weak IP protection regime, immature industrial standards, norms,
legislation or regulations. We term these issues market- and
institution-related constraints to innovation. The intensity of market
competition or monopolisation in some sectors may affect the outcome of innovation strategies (Fosfuri, 2006), promoting open innovation activities in particular. Many Chinese rms are still stuck in the
low value-added element of the global value chain, and some have to
face increasingly intense competition in local markets. Accordingly,
these rms are extremely keen to nd partners who possess advanced
R&D capacity or strong branding to help them get away from the lowend value chain and extremely intense competition. In addition, in
light of the large but layered and dynamic market in China, the high
level of uncertainty with respect to customer needs leads most local
rms to be hesitant in decision making with respect to innovation, in
both the short and long term. Having realised that dependence on
175
quick imitation or introduction of simple products cannot provide a

sustainable benet, many local rms are more likely to adopt networking with suppliers, users or even potential rivals or to build or participate in strategic alliances to reduce the uncertainty of market needs
and increase customer responsiveness to new products or process
innovation.
Furthermore, the existence of relatively weak institutions forces some
local rms to try their best to build linkages with government or
universities to be rst to learn of market opportunities and knowledge.
For instance, the quality of any relationship with central or local governments is widely regarded to be one of the most important forms of
social capital in China. Networking with regional governments to
access inside information may have a great inuence on a rms future
business strategy or its acquisition of government funding and/or preferential tax rates. Thus, it is proposed that indigenous rms that suffer
from more market- and institution-related constraints to innovation
will be inclined to increase their openness to innovation via relationships with various external organisations or public institutions. We
therefore posit the following hypothesis:
H71: Firms that face greater market- and institution-related constraints to innovation are more likely to adopt greater openness
in innovation.
Our second proposition is that nancial constraints and risk-related
impediments to innovation could also stimulate a model based on an
open style of innovation. In this case, such an impediment is related to
certain input or cost considerations, including a lack of input capital or
appropriate sources of nance, excessively high innovation costs and
excessive perceived economic risk. In fact, investing and nancing
capabilities do matter in industrial development for latecomers
(Malerba, 2002). For example, notwithstanding the lack of venture
capital to commercialise new inventions or sustain investment in industrialisation, many Chinese biotech rms remain stuck at the laboratory
stage or fail because of a long R&D cycle and the fact that no further
capital is then available at the necessary time. Thus, some of these rms
try to source external strategic partners, whether leading rms or venture capitalists, to achieve technological exploitation.
In addition, technological innovation always requires signicant
capital input, which places innovators under high cost pressure,
176
especially in the case of those emerging latecomers whose prot margins

or revenue-making competences are quite unstable. Hence, they have
to co-innovate with external sources (such as lead users, suppliers or
universities) to share R&D costs and reduce failure risks. In some cases,
once the notion of inter-organisational innovation collaboration has
entered an industry, any rm that does not participate will experience
serious competitive disadvantages (Enkel, Gassmann and Chesbrough,
2009).
In addition, excessive economic risks in innovation activities have
been widely recognised and conrmed (OECD, 2005). Studies on
domestic R&D co-operation show that risks associated with innovation should promote R&D collaboration or inter-rm partnerships
(Bayona, Garca-Marco and Huerta, 2001; Hagedoorn, 2002;
Tether, 2002). Evidence based on the Swiss Community Innovation
Survey also nds that the existence of risk-related impediments leads
to greater openness amongst Swedish rms (Keupp and Gassmann,
2009). All these arguments suggest that rms that suffer from greater
nancial constraints and risk-related impediments to innovation are
more likely to make use of open innovation to obtain access to
external nance or share R&D risks. This leads to the following
hypothesis:
H72: Firms that experience nancial constraints and risk-related
impediments to innovation are likely to have greater openness
in innovation.
In our third proposition, we tentatively argue that knowledge-/skillrelated constraints can promote Chinese rms willingness to deploy
open innovation. In the Chinese case, a greater number of internal
constraints exist, such as a lack of skilled personnel, a lack of information on technologies, a lack of information on markets, a lack of
innovation collaborators, a lack of a technological basis, too large a
technological gap or even resistance towards change in the rm.
Knowledge-based assets are the primary resources latecomers lack as
compared to leading rms in the global market (Amsden, 1992).
However, knowledge and learning are key strategic levers because
knowledge development is cumulative (Tushman and Anderson, 1986).
In terms of their weak accumulation of knowledge, most Chinese
rms tend to partner with foreign entrants in order to learn. Indeed,
177
exploratory tasks such as external sourcing are typically associated

with the need for skills that transcend the current capabilities of the
organisation (Katila and Ahuja, 2002). Inter-organisational collaboration has proven to be an effective strategy to learn and manage intellectual capital (Powell, Koput and Smith-Doerr, 1996). Moreover, global
modularisation has provided a valuable opportunity for many Chinese
rms, especially those rms that produce complex products, to enter
into the global production chain. The decomposability of technology
and reduced external transaction costs as a result of the increased
availability of international R&D resources have made open innovation appealing and useful within Chinese manufacturing innovation.
Hence, Chinese rms that face more knowledge-/skill-related constraints to innovation may tend to be more open in sourcing external
knowledge and skills for innovation. Therefore, we have the following
hypothesis:
H3: Firms that experience knowledge and skill constraints to innovation are likely to have greater openness in innovation.
Most rms in emerging economies are still at a transitional phase of
catch-up and innovation capability building. It is often argued that it is
difcult for latecomers to skip stages in the evolutionary process of
learning. Hence, tapping into external resources and knowledge
forms the foundation of cultivating indigenous innovation capability
(Kim, 1997a, 1997b). Opening up the innovation process is likely to be
a natural choice for latecomers in emerging economies to mitigate R&D
pressure, overcome various constraints, diversify risks and share
uncertainties.
7.3 Data and Methodology

Data
The research principally uses the 2008 Chinese national innovation
survey of 1,408 manufacturing rms, which includes information
regarding rms innovation activities from 2005 to 2007. The
National Statistical Bureau carried out the survey. It covers 42 cities
in China in both the coastal and inland regions. A total of 1,408 valid
responses were received, with a response rate of 83.6 per cent. The
178
questionnaire is designed to have high consistency and comparability

with the European Community Innovation Survey (CIS). The sample
includes rms of all major ownership types that exist in China, and
the sample is composed of 9 per cent state or collectively owned
enterprises, 7 per cent privately owned rms, 53 per cent publicly
owned and limited liability companies, and 30 per cent foreign
invested companies that are wholly owned subsidiaries of MNEs or
joint ventures in which foreign investment accounted for more than
25 per cent of total investment. Larger innovative rms are overrepresented in the sample. About 50 per cent and 17.5 per cent of
the rms in the sample represent medium- and large-sized rms,
respectively.1 The total reported R&D investment in the sampled
rms was RMB 53,350 million, accounting for 18.2 per cent of
Chinas total R&D investment in the manufacturing sector in 2007.
After careful data cleansing to exclude observations with missing
values for the necessary variables, the nal data set used in the estimation contains 626 rms, of which 95 per cent have innovated
in their products or processes. Therefore, the results of this study
reect the decision to engage in open innovation made by innovative
Chinese rms rather than by Chinese rms generally. This is a limitation of the research that we should bear in mind when drawing
conclusions.
Measurement
Dependent variables
The operational practice of openness is nuanced; therefore, different
measures are deployed to measure a rms openness in innovation.
Laursen and Salter (2006) develop two measurements of openness
involving the breadth (range of external sources) and depth (importance
of sources) of the innovation process:
The breadth of openness refers to the number of search channels a

company draws on in its innovative activities.
1
The term medium- and large-sized enterprises (MLEs) refers to rms with between
300 and 2,000 employees and more than 2,000 employees, respectively. The total
number of MLEs in the sample is 950, accounting for 2.4 per cent of Chinas total
MLEs in the manufacturing sector in 2007, according to the 2008 National
Economic Census Yearbook.
179
The depth of innovation refers to the extent to which companies draw

intensively on different search channels or sources of innovative
ideas.
Following this perspective, in this chapter we measure the breadth of
open innovation by calculating how many of the 10 major external
knowledge sources are integrated into a rms innovation processes; we
measure the depth of open innovation by the number of these sources
being deeply integrated into a rms innovation processes. The 10
external sources include those from within the rm or the company
group: supplier; customer; competitors; consultancy companies and
private R&D institutions; universities; public research institutions; government science and technology programs; trade fair and professional
conferences; journals, magazines and patent documents; and industrial
associations.
Specically, our questionnaire asks the interviewed rms to conrm
the degree of importance of each source on a scale from 1 to 4, based on
the extent of usage by them (1 means that this source has not been used
by the rm; 4 means the highest level of usage). Breadth is calculated as
the number of external sources and is rated from 1 to 4. Depth is
calculated as the number of external sources that are rated as important
with a score of 4.
Independent variables
This survey questionnaire contains 21 Likert-scale items used to question rms regarding the extent to which a specic impediment had
signicant negative consequences for innovation activities. More specically, the 21 items involve economic risks, cost, nancing channels as
well as items on talents, knowledge accumulation, technical information, market information and collaborators, monopoly, competition,
protability, innovation of other rms in same sector and uncertainty of
market demand. The items are rated from 1 (no negative consequence)
to 4 (very strong negative consequence). We use these items to construct
scales intended to measure different kinds of constraints and risks to
innovation. We classify the 21 items into three kinds of constraints to
innovation: nance- and risk-related impediments, market- and
institution-related constraints and knowledge- and skills-related constraints. The market- and institution-related constraints to innovation
section includes nine questions, the nancial constraints and risk-
180
related impediments to innovation ve questions and the knowledgeand skill-related constraints to innovation seven questions. Each constraint/risk variable is estimated as the average of the scores of all the
relevant questions. Details of the constraints and risks are given in
Appendix 7.1.
Some of the obstacles are likely to have greater impact than others.
Hence, ideally we would know the different weights of each of the
obstacles and estimate a weighted index instead of giving equal weight
to the individual indicators. However, each rm may perceive a different constraining effect as a result of its capabilities and other
rm-specic characteristics. It would be difcult to create a reasonable
weighting pattern that suits all rms. Moreover, we have experimented with using factor analysis to generate the constraints and
found that the outcome in fact is highly consistent with the three
impediment indexes that we generated according to theory.
Therefore, the indexes that we use here are broadly consistent with
theory and the actual variation in the data. In light of the high correlation between the knowledge- and skills-related constraints and the
other two constraints and risks, we enter the three impediment variables into the regression alternatively to avoid any multi-collinearity
problem.
Control variables
Following the literature on determinants of open innovation, we include
rm size, age, ownership, industry effects, regional effects, R&D intensity and collaboration experiences as control variables.
Firm size and rm age may affect open innovation strategies and
performance because larger or older rms usually have larger technology portfolios (Lichtenthaler and Lichtenthaler, 2009). We measure
rm size based on the number of employees. Therefore, a discrete
variable that equals 1 for rms of fewer than 300 employees, 2 for
rms that have 300 to 2,000 employees, and 3 for large rms with
more than 2,000 employees is included as an explanatory variable.2
Ownership may also inuence a rms openness in innovation. We
include two dummies to represent the primary types of rm ownership
Here we use the rm size classication dened by the National Bureau of Statistics
of China.
181
in China: a state-owned dummy that equals 1 for state-owned enterprises and 0 otherwise, and a private-ownership dummy that equals 1
for privately owned enterprises and zero otherwise.3 The foreign-owned
enterprises are used as the base group.
As there are potentially obvious differences in motives and frequency
of open innovation between medium- and high-tech industries and
other low-tech industries (Chesbrough and Crowther, 2006; OECD,
2008), we control for the industry specic effects at the SIC two-digit
level by using a vector of industry dummies. In total, 29 industry
dummies are included in the regression. In a large country with signicant regional disparities, regional-specic effects may also signicantly
inuence a rms decision to adopt open innovation. For example, rms
in eastern regions are more inclined to have an open attitude towards
their innovation processes. We therefore likewise control for region by
dummy variables that are coded 1 if the rm is based in a city in the
eastern region (including four municipalities, Yangtze River Delta and
Pearl River Delta) and 0 otherwise.
Absorptive capacity is an important factor when rms choose to learn
from outside. To measure this, we use a question that directly asks rms
to evaluate the extent of both internal and external R&D input (or
participation), scaled from 1 (low input) to 2 (medium input) and 3
(extensive input) to proxy a rms R&D intensity. Finally, we also
control for a rms collaboration experience: this can greatly inuence
a rms capacity and experience in open innovation. A dummy variable
that equals 1 if a rm reports that collaboration is prevalent in its
industry and 0 otherwise is included in the regression.
To ensure that all the observations in all items drawn from the
questionnaire are statistically reliable, we use reliability analysis to test
for the reliability of the data (with Alpha = 0.8975 and 1,408 cases).
Given that the dependent variables are count values that range between
0 and 10, standard OLS estimates will be biased. We therefore use
ordered Logit for the estimation. Nevertheless, we report both the
ordered Logit and OLS estimates as a robustness check.
Companies with shareholders are classied as POEs (privately owned enterprises)

in the research. Admittedly, some state-controlled shareholding companies are
included in this group as a result. This may mask the true behaviour and
performance of the POEs, but we believe the proportion and inuence are limited.
182
7.4 Results
Table 7.1 reports the descriptive statistics of the major variables and the
correlation coefcients amongst them. The three impediments to innovation the market- and institution-related constraints, the nancial
constraints and risk-related impediments and the knowledge- and skillrelated constraints to innovation all appear to be positive and signicantly correlated to the degree of openness in terms of both breadth and
depth. There are, however, also signicant correlations amongst the
three impediments variables: this may cause a potential multicollinearity in the regression models. Therefore, we enter the three
impediments alternatively into the regression to avoid this problem.
Table 7.2 reports the ordered logistic model estimates of the determinants of the breadth of openness in Chinese manufacturing rms. All
three constraints and risks appear to have a signicant positive association with rms breadth of openness in innovation when we enter the
constraints into the regression alternatively. When the three constraints
are entered into the regression at the same time, institutional- and
market-related constraints and nancial constraints both show a robust
and consistent positive and signicant association with the breadth of
openness. For a one-unit increase in the nancial constraints and riskrelated impediments, we expect to see an approximately 35 per cent
increase in the odds of having one level higher breadth of openness.4
Similarly, the odds ratio for institution- and market-related constraints
is about 92 per cent, which is the highest impact among the three
impediments. In other words, rms are most likely to open up their
innovation process when there are institutional and market constraints.
However, the estimated coefcient of the knowledge- and skills-related
constraints variable becomes insignicant when all three constraints
variables are included in the regression at the same time.
Intramural R&D and extramural R&D appear to be positively associated with a rms breadth of openness and are statistically signicant.
For each category increase of intramural R&D, the odds of one level of
greater breadth of openness is about 74 per cent. Similarly, for each
category increase of extramural R&D, we expect to see about a 50 per
cent increase in the odds of having one level higher breadth of openness.
The odds ratio is calculated as exp(beta).
Table 7.1 Descriptive statistics and correlations

Variable Obs Mean
breadth
depth
size
age
soe
poe
region
inrd
exrd
collab2
risk2
skill2
makt2
Std. Dev. breadth depth size
626 8.527 2.230

626 2.404 2.102
626 2.377 0.595
626 17.142 19.224
626 0.096 0.295
626 0.658 0.475
626 0.610 0.488
626 2.486 0.615
626 1.649 0.641
626 2.965 1.014
626 2.606 0.761
626 2.509 0.703
626 2.366 0.624
1.00
0.34
0.20
0.10
0.02
0.07
0.00
0.29
0.22
0.05
0.16
0.11
0.19
1.00
0.07
0.10
0.00
0.10
0.03
0.21
0.15
0.10
0.11
0.13
0.14
age
soe
poe
region inrd
exrd collab2 risk2 skill2 makt2
1.00
0.12 1.00
0.03 0.34 1.00
0.08 0.05 0.45 1.00
0.04 0.12 0.18 0.12 1.00
0.21 0.09 0.03 0.08 0.05 1.00
0.16 0.05 0.02 0.07 0.04 0.35 1.00
0.02 0.00 0.02 0.02 0.05 0.02 0.02 1.00
0.02 0.09 0.04 0.13 0.09 0.06 0.09 0.10
0.00 0.04 0.02 0.08 0.03 0.02 0.12 0.08
0.03 0.02 0.06 0.10 0.05 0.07 0.12 0.09
1.00
0.49 1.00
0.51 0.69
1.00
184
Table 7.2 Determinants of breadth of openness: ordered-Logit model

estimates
Firm size
Firm age
SOE
POE
Coastal region
Intramural R&D
Extramural R&D
Collaboration
experience
Finance/risk
constraints
0.531***
(0.170)
0.004
(0.005)
0.136
(0.373)
0.278
(0.214)
0.347*
(0.201)
0.551***
(0.160)
0.399***
(0.151)
0.173**
0.489***
(0.170)
0.005
(0.005)
0.203
(0.362)
0.341
(0.213)
0.272
(0.197)
0.523***
(0.161)
0.419***
(0.152)
0.193**
0.528***
(0.168)
0.005
(0.005)
0.197
(0.370)
0.296
(0.211)
0.230
(0.198)
0.492***
(0.161)
0.409***
(0.151)
0.167*
0.559***
(0.168)
0.005
(0.005)
0.162
(0.378)
0.258
(0.212)
0.290
(0.202)
0.520***
(0.161)
0.397***
(0.150)
0.153*
(0.089)
(0.087)
0.298**
1.347**
(0.137)
0.238
0.788
(0.086)
(0.088)
0.459***
(0.115)
Knowledge/skills
constraints
0.286**
(0.123)
Institute/market
constraints
Industry
Observations
Pseudo-R2
Odds
ratio (4)
Yes
627
0.076
Yes
628
0.069
1.748***
1.005
1.176
1.295
1.336
1.682***
1.487***
1.166*
(0.173)
0.639*** 0.652*** 1.920***
(0.143)
Yes
628
0.078
(0.199)
Yes
626
0.082
Yes
***signicant at 1 per cent level; **signicant at 5 per cent level; *signicant at 10 per
cent level. Robust standard errors in parentheses; BREADTH, breadth of openness.
Note: Foreign-invested rms are the base ownership group.
185
All this suggests that rms with greater absorptive capacity are more
open in innovation and engage a greater range of external sources.
Firm size also seems to have a positive impact on a rms breadth of
openness. Larger rms tend to have a greater degree of openness in
terms of breadth of tapping into external knowledge sources. Firms
with previous collaboration experiences are also more likely to increase
their openness in innovation. However, we do not see signicant differences across ownership types and geographical location with regards to
breadth of openness when other rm-specic characteristics are controlled for. In other words, ownership does not seem to have an impact
on whether rms have used/integrated external information sources for
innovation, but rms are more likely to differ in the extent to which they
do so.
Table 7.3 reports the Ordered-Logistic model estimates of the determinants of the depth of openness in Chinese manufacturing rms. Again,
all three constraints and risks including institutional- and market-related
constraints, knowledge- and skills-related constraints and nancial constraints and risks appear to have a signicant positive association with
rms breadth of openness in innovation. The estimated results are
robust to different model specications. For a one-unit increase in the
nancial constraints and risk-related impediments, we expect to see
about a 9 per cent increase in the odds of having one level higher breadth
of openness. Similarly, the odds ratio for knowledge- and skills-related
constraints is 14 per cent and that for institution- and market-related
constraints is about 30 per cent, which is again the highest impact
amongst the three impediments. In other words, rms are most likely
to use deep openness in innovation when there are institutional and
market constraints. However, unlike in the case of breadth of openness,
the constraints in knowledge and skills for innovation seem to have
greater impact on the depth of openness than those involving nancial
constraints and risk-related impediments. The association between
impediments and depth of innovation openness is strong; even in column (4), when all the three impediments are entered into the regression
at the same time, the estimated coefcients of the impediments variables
are statistically signicant at the 1 per cent level.
Intramural and extramural R&D as well as rm size have a signicant and positive impact on a rms depth in openness innovation. In
other words, rms that have high levels of in-house R&D, higher levels
of extramural R&D and large size are more likely to have greater depth
186
Table 7.3 Determinants of depth of openness: ordered-Logit model

estimates
Firm size
Firm age
SOE
POE
Coastal region
Intramural R&D
Extramural R&D
Collaboration
experience
Finance/risk
constraints
Odds
ratio (4)
0.082***
(0.021)
0.009***
(0.001)
0.018
(0.042)
0.238***
(0.044)
0.390***
(0.042)
0.529***
(0.020)
0.166***
(0.027)
0.129***
0.057***
(0.021)
0.009***
(0.001)
0.063
(0.042)
0.244***
(0.044)
0.363***
(0.042)
0.545***
(0.020)
0.149***
(0.027)
0.121***
0.077***
(0.021)
0.010***
(0.001)
0.065
(0.041)
0.218***
(0.044)
0.338***
(0.042)
0.518***
(0.020)
0.154***
(0.027)
0.115***
0.080***
(0.022)
0.009***
(0.001)
0.052
(0.043)
0.220***
(0.047)
0.347***
(0.044)
0.522***
(0.021)
0.136***
(0.028)
0.111***
(0.015)
(0.016)
0.084*** 1.088***
(0.015)
(0.015)
0.236***
(0.018)
Knowledge/skills
constraints
(0.019)
1.009***
1.054
1.246***
1.415***
1.685***
1.146***
1.118***
(0.020)
0.130*** 1.138***
0.328***
Institute/market
constraints
1.084***
(0.021)
0.418*** 0.265*** 1.303***
Industry
Yes
Yes
(0.020)
Yes
(0.022)
Yes
Observations
Pseudo-R2
627
0.048
628
0.049
628
0.050
626
0.051
Yes
cent level. Robust standard errors in parentheses; BREADTH, breadth of openness.
Note: Foreign-invested rms are the base ownership group.
187
in open innovation. Interestingly (and unlike the case of breadth), rm

age, ownership, geographical location and previous collaboration experience all have a positive impact on rms depth of open innovation, and
these variables are statistically signicant. For every one-year increase in
rm age, the odds ratio that a rm will increase its depth of innovation
increases by 1 per cent.
Furthermore, while SOEs do not appear to be signicantly different
from foreign invested rms (FIEs) in terms of depth of open innovation (OI), privately owned rms (POEs) have a signicantly higher
propensity to engage with deeper open innovation. Finally, rms in
coastal regions also have a signicantly higher propensity to engage in
deeper open innovation than rms in the inland region. Given that
their other characteristics are the same, the likelihood of engaging in a
higher level of depth of innovation for a POE is 24 per cent greater
than for an FIE and 41 per cent for a rm in coastal as opposed to
inland regions. The lower openness in innovation in the FIEs than the
POEs is likely the result of concerns regarding weak IPR protection
in China and the potential to lose technology know-how in the open
innovation process of FIEs. FIEs may also face fewer constraints (push
factors). The greater depth of open innovation of rms in coastal
regions is likely the result of the greater overall economic openness
in the coastal region and hence the easier access to external knowledge
sources and greater openness in management practice, business models and strategic thinking amongst the entrepreneurs and management
in the coastal region. Moreover, the greater technological capacity
and hence greater absorptive capacity of rms in the coastal regions
may also be an attribute of the signicantly greater openness in
innovation amongst rms in the coastal region.
Table 7.4 reports robustness check estimates using OLS. The size,
sign and signicance of the estimated coefcients of the three impediment variables are broadly consistent with those of the OrderedLogistic model estimates, although the statistical signicance of the
OLS estimates of the impediments variables is a level lower than that
of the Ordered-Logistic estimates in the depth table. Regarding the
control variables, in respect to the breadth of openness, the estimated
results are robust with regard to the impact of rm size and intramural
and extramural R&D intensity. However, the statistical signicance of
rm size, ownership and collaboration experience changes slightly. In
respect to the depth of openness, the estimated results are robust for
Table 7.4 Robustness check using OLS

Breadth
Firm size
Firm age
SOE
POE
Coastal region
Intramural R&D
Extramural R&D
Collaboration experience
Depth
0.531***
(0.170)
0.004
(0.005)
0.136
(0.373)
0.278
(0.214)
0.347*
(0.201)
0.551***
(0.160)
0.399***
(0.151)
0.173**
(0.086)
0.489***
(0.170)
0.005
(0.005)
0.203
(0.362)
0.341
(0.213)
0.272
(0.197)
0.523***
(0.161)
0.419***
(0.152)
0.193**
(0.088)
0.528***
(0.168)
0.005
(0.005)
0.197
(0.370)
0.296
(0.211)
0.230
(0.198)
0.492***
(0.161)
0.409***
(0.151)
0.167*
(0.089)
0.061
(0.150)
0.010*
(0.005)
0.124
(0.340)
0.426**
(0.176)
0.369**
(0.185)
0.507***
(0.142)
0.211
(0.143)
0.172**
(0.085)
0.037
(0.149)
0.009
(0.006)
0.164
(0.334)
0.427**
(0.172)
0.316*
(0.181)
0.515***
(0.142)
0.195
(0.145)
0.168**
(0.086)
0.054
(0.149)
0.010*
(0.005)
0.190
(0.337)
0.420**
(0.174)
0.303*
(0.182)
0.488***
(0.142)
0.202
(0.142)
0.165*
(0.086)
Finance/risk
constraints
0.459***
0.253**
(0.115)
Knowledge/skills
constraints
(0.119)
0.286**
0.323**
(0.123)
Institute/market
constraints
(0.136)
0.639***
0.364**
Industry
Yes
Yes
(0.143
Yes
Yes
Yes
(0.148)
Yes
Observations
Pseudo-R2
627
0.076
628
0.069
628
0.078
627
0.157
628
0.160
628
0.160
***signicant at 1 per cent level; **signicant at 5 per cent level; *signicant at 10 per cent level. Robust standard errors in parentheses; Breadth,
breadth of openness; Depth, depth of openness
Note: Foreign invested rms are the base ownership group.
190
ownership, regional, intramural R&D and collaboration experience.

However, the OLS estimates lose statistical signicance with regard to
rm size, age and extramural R&D intensity probably because the real
distribution of the population violates the assumption of normal distribution involved in the OLS model. In general, the OLS estimates have
also conrmed the signicant link between the constraints and risks that
rms face and the degree of openness in innovation that rms may
choose.
The ownership effect

Table 7.5 reports the estimated results of rms responses to constraints
and risks in subsamples of rms with different types of ownership.
Firms in the sample are classied into three subsamples based on their
ownership: state-owned rms (SOEs), privately owned rms (POEs)
and foreign invested rms (FIEs).5 The evidence shown in the table
suggests that rms responses to constraints and risks vary across ownership types. The breadth of innovation openness in the SOEs, the
estimated coefcients of the nance and risks-related impediments and
the knowledge- and skills-related constraints have positive and statistically signicant relationships at the 1 per cent level, indicating that
SOEs will utilise more partnerships and a wider information search
when they have constraints in their knowledge for innovation. They
will also use more partnerships to share the risks when they face excess
risks or nance constraints in innovation. However, institution- and
market-related constraints do not appear to have a signicant impact on
the degree of openness in SOEs. It may be that the ownership and
political ties that the state-owned rms have had in a half-marketised
Another similar approach is to use an interaction term for ownership and

constraints to test the statistical signicance of the difference in the coefcients
(slopes) of the constraint variables as between rms with different types of
ownership. However, as the focus of this chapter is the behaviour of rms with
different ownership types in the face of the concerned constraints, the statistical
signicance of the estimated constraints variables is the primary concern, and this
can only be revealed by estimating the model in each subsample of rms with
different ownership types. Nevertheless, we have also estimated models with
interaction terms that conrm that the estimated coefcients in the depth models
are signicantly different from each other. However, this is not the case for the
breadth models.
Table 7.5 Firms responses to constraints and risks by ownership

SOE
Breadth
1
1.11***
(0.08)
Age
0.01**
(0.00)
Region
1.62***
(0.15)
Intramural R&D 1.36***
(0.08)
Extramural R&D 0.73***
(0.10)
Collaboration
0.58***
(0.06)
Finance/risks
0.51***
(0.07)
Knowledge/skills
Size
Depth
2
1.29***
(0.08)
0.01**
(0.00)
1.46***
(0.16)
1.32***
(0.08)
0.76***
(0.11)
0.49***
(0.07)
Yes
60
0.183
3
1.17***
(0.08)
0.01***
(0.00)
1.50***
(0.16)
1.38***
(0.08)
0.83***
(0.11)
0.51***
(0.07)
4
1.32*
(0.73)
0.01
(0.01)
0.05
(0.87)
0.69
(0.76)
0.87
(0.96)
0.18
(0.44)
0.44
(0.42)
0.35***
(0.08)
Institute/Market
Industry
Observations
r2_p
POE
Yes
60
0.177
Breadth
5
1.52**
(0.76)
0.01
(0.01)
0.16
(0.90)
0.53
(0.83)
0.85
(0.95)
0.14
(0.48)
6
1.25
(0.82)
0.01
(0.01)
0.05
(0.84)
0.81
(0.88)
1.01
(0.87)
0.13
(0.43)
7
0.40*
(0.21)
0.00
(0.01)
0.40*
(0.23)
0.63***
(0.21)
0.46**
(0.19)
0.20
(0.11)
0.39***
(0.14)
0.51
(0.45)
(0.13
(0.08)
Yes
60
0.175
Yes
60
0.167
Yes
60
0.167
Depth
8
0.38*
(0.21)
0.00
(0.01)
0.33
(0.23)
0.56***
(0.21)
0.50***
(0.19)
0.22**
(0.11)
9
0.42***
(0.03)
0.00
(0.00)
0.26***
(0.06)
0.52***
(0.03)
0.49***
(0.04)
0.17***
(0.02)
10
0.06
(0.17)
0.01
(0.01)
0.49**
(0.20)
0.58***
(0.18)
0.15
(0.16)
0.11
(0.09)
0.29**
(0.14)
0.17
(0.15)
(0.27
(0.56)
Yes
Yes
60
413
0.162 0.080
FIE
Yes
414
0.075
Breadth
11
0.08
(0.17)
0.01
(0.01)
0.4**
(0.20)
0.58***
(0.18)
0.14
(0.17)
0.11
(0.09)
12
0.05
(0.17)
0.01
(0.01)
0.38*
(0.20)
0.55***
(0.18)
0.16
(0.16)
0.08
(0.10)
13
1.25***
(0.44)
0.01
(0.02)
0.07
(0.86)
0.27
(0.36)
0.44
(0.37)
(0.26
(0.23)
1.07***
(0.36)
0.27
(0.16)
0.58***
(0.03)
Yes
Yes
414
413
0.087
0.048
Yes
414
0.047
Depth
14
1.11**
(0.47)
0.01
(0.02)
0.38
(0.80)
0.43
(0.40)
0.41
(0.37)
(0.18
(0.25)
15
1.00**
(0.46)
0.01
(0.02)
0.35
(0.85)
0.45
(0.39)
0.38
(0.37)
(0.11
(0.23)
16
0.49***
(0.04)
0.00
(0.00)
0.72***
(0.10)
0.32***
(0.04)
0.07
(0.05)
0.25***
(0.03)
0.35***
(0.04)
0.579*
(0.34)
0.49***
(0.18)
Yes
Yes
414
154
0.051
0.154
Yes
154
0.132
17
0.45***
(0.04)
0.01
(0.00)
0.61***
(0.10)
0.41***
(0.04)
0.03
(0.05)
0.22***
(0.03)
18
0.39***
(0.04)
0.00
(0.00)
0.62***
(0.10)
0.37***
(0.04)
0.07
(0.05)
0.26***
(0.03)
0.51***
(0.04)
0.82*
(0.43)
Yes
154
0.136
Yes
154
0.106
Yes
154
0.11
0.35***
(0.04)
Yes
154
0.104
***signicant at 1 per cent level; **signicant at 5 per cent level; *signicant at 10 per cent level. Robust standard errors in parentheses; Breadth: breadth of openness; Depth: depth of openness.
Note: Foreign invested rms are the base ownership group. Estimated using Ordered-Logit model.
192
transition economy such as China face fewer institution-related constraints or have other alternatives to overcome such constraints in
innovation.
With respect to the depth of openness, none of the estimated coefcients of the impediments variables are statistically signicant, which
suggests that the depth of openness in innovation in the SOEs is not
determined by the constraints or risks that these companies face. Our
results also do not support arguments based on the major determinants
traditionally suggested by the literature. In fact, other estimation results
(not reported in the table because of space limitations) suggest that
industry-specic effects have played a signicant role in explaining the
variations in depth of openness in the state-owned sector.
Interestingly, rms that are not SOEs appear to be more responsive to
open innovation as a means to overcome the constraints and risks that
they face in innovation. In privately owned rms, the estimated coefcients of the institution- and market-related constraints and the
nance- and risks-related constraints are both positive and statistically
signicant with regard to their impact on the breadth and depth of
openness. For a one-unit increase in the institution- and market-related
constraints, we expect to see about a 78 per cent increase in the odds of
having a one-level higher breadth of openness. The impact on the odds
ratio of the depth of openness is 63 per cent for a one-unit increase in
institution- and market-related constraints.
The POEs also appear to choose to open up their innovation process
to different degrees of depth relative to the nancial and risk-related
constraints. Unexpectedly, they seem not to widen or deepen openness
in innovation to respond to the constraints in knowledge- and skillsrelated issues that they face. This may be because they can directly hire
the talented staff they need and so do not need to rely on any information search or collaboration. This may also be because the POEs do not
have the capacity to use open innovation effectively to overcome the
skills and knowledge constraints.
FIEs appear to be the most responsive to widening and deepening
their openness to overcoming the various constraints they have encountered in the innovation process. In particular, FIEs have used open
innovation to share the risks they have in the host country. The estimated coefcients of the nancial- and risks-related constraints are
positive and both statistically signicant at the 1 per cent level.
Moreover, they have also deepened their openness of innovation in
193
response to all three types of constraints they encounter. The estimated

coefcients of the three impediments variables are positive and statistically signicant at the 1 per cent level. The signicant and all-around
responsiveness in the FIEs that is revealed in the table is consistent with
the pattern observed in industrialised countries (e.g., Keupp and
Gassmann, 2009). This indicates that the subsidiaries of the multinationals not only have to be responsive to internal constraints and the
external environment when innovating but also that they have the
knowledge and capability to use this new mode of innovation effectively
to overcome obstacles.
The moderating effect of rm size and industry

As the results reported in Tables 7.2 and 7.3 indicate that rm size and
industry-specic effects have signicant effects on the depth and breadth
of openness in innovation, it is also interesting to understand the types
of rms that are mostly affected by constraints: for example, rms of
different size and in different industries. Table 7.6 reports the estimated
results concerning the moderating effects of rm size and the technology
intensity of the industry.
With respect to the breadth of the openness in innovation, only the
estimated coefcient of the interaction term of nancial constraints and
high-technology industry dummy is positive and statistically signicant
at the 5 per cent level. This suggests that rms in high-technology
industries are more likely to adopt open innovation when they face
nancial constraints.
With respect to the depth of openness in innovation, the size and
technology intensity of the industry both appear to have signicant
moderating effects on the association between the various constraints
and the depth of openness. Smaller rms are more likely to undertake
deep engagement in open innovation to overcome the nancial constraints and risks in innovation, whereas larger rms are more likely
to do so to overcome constraints in knowledge and skills. However, in
the face of institutional constraints, there is no signicant difference in
rms response to open innovation. Firms in a high-technology industry are more affected by constraints in nance and skills and are more
likely to respond to these constraints by engaging in deeper open
innovation. However, in the face of institutional constraints, rms
in medium- and low-technology industries are more likely to use open
194
Table 7.6 Effects of constraints and risks by size and industry
Finance C * Size
Skill C * Size
Institution C * Size
Finance C * High-tech
Skill C * High-tech
Institution C * High-tech
Finance/risks constraints
Knowledge/skills constraints
Market/institute constraints
Other control variables
Observations
Pseudo-R2
Breadth
0.181
(0.222)
0.364
(0.283)
0.339
(0.351)
0.794**
(0.331)
0.416
(0.391)
0.529
(0.473)
0.537
(0.509)
0.966
(0.673)
1.545*
(0.838)
Yes
626
0.087
Depth
0.157***
(0.008)
0.299***
(0.009)
0.010
(0.009)
0.388***
(0.037)
0.193***
(0.039)
0.420***
(0.041)
0.360***
(0.021)
0.589***
(0.022)
0.364***
(0.024)
Yes
626
0.053
cent level. Robust standard errors in parentheses; Breadth: breadth of openness; Depth:
depth of openness.
Note: Other control variables are the same as those included in Tables 7.2
and 7.3.
innovation to overcome this constraint than rms in a hightechnology industry.
7.5 Discussion and conclusions

This chapter examines rms responses to the external and internal
constraints and risks to innovation, with special emphasis on rms
adjustment in their openness in the innovation process. It provides the
195
rst empirical evidence concerning drivers of open innovation in emerging economies based on large-scale survey data. Evidence from the
research suggests that Chinese rms that suffer from greater marketand institution-related, capability- and skills-related or nance- and
risks-related constraints are more likely to engage with open innovation
in greater depth and breadth to overcome these impediments. The
strength of such responses, however, varies across rms of different
ownership types, rm size and technology intensity. Foreign invested
rms appear to respond the most by widening and deepening their
openness in innovation. Privately owned rms have made signicant
responses to market- and institution- and nance- and risk-related
impediments but not to knowledge- and skills-related impediments,
whereas state-owned rms appear to be least responsive in using open
innovation to overcome the constraints and risks they face. These
insights complement our existing understanding of the determinants
of open innovation that are derived mostly from experiences in developed countries.
The rst major nding of the research is the direct link between the
constraints and risks to innovation and openness in innovation.
External impediments such as market- and institution-related constraints to innovation have affected Chinese rms in the past several
decades. The market and industry environment in China has been
evolving rapidly. A large number of Chinese rms attempt to upgrade
their technological capabilities to move away from low value-added
manufacturing and price competition in local markets by sourcing
external innovation ideas or networking with external partners.
Moreover, uncertainty in market demand and the lack of customer
responsiveness are widely existing phenomena in emerging economies
such as China. Although the Chinese market is large, its structure
appears highly segmented and unpredictable (Chen, Liu and Chumlin,
2007). The creation of alliances with complementary partners has been
shown to be effective in reducing market uncertainty, in particular
during the development of emerging technologies (Li and Xiong,
2009). Moreover, the lagged development of a favourable institutional
environment for innovation in China has also signicantly hindered the
growth of innovation activity. Setting up strategic innovation alliances
with other rms or engaging with government departments or rms
with political ties may help overcome the negative effect of institutional
constraints.
196
Regarding nance- and risk-related impediments to innovation,

technological innovation is an exploratory task where the returns, if
any, are distant and uncertain (Levinthal and March, 1993). Existing
nancial markets or channels in China are not sufcient for most
domestic rms to utilise. This pressures Chinese rms to engage with
partners in sharing the high costs of innovation. Many prior ndings
derived from innovation collaboration studies have clearly indicated
that the primary motives of pursuing cooperation in innovation are to
acquire complementary nance or resources and to spread risks or
reduce costs (Mohr and Spekman, 1994; Hoffmann and Schlosser,
2001). Local rms in China, especially SMEs, are lacking in adequate
and sustainable capital to carry out internal innovation. They are
incapable of acquiring enough funds to support innovation.
Therefore, they have to perform innovation searches as widely and
deeply as possible to make up for any funding shortage and to share
high R&D costs. Such situations exist widely in China, especially in
science-based sectors (Li et al., 2010). Increasing openness of innovation could also help latecomers save a large amount of early investment on basic research and better focus on the commercialisation of
development (Kim, 1997).
Firms that suffer from more knowledge- and skill-related impediments to innovation also increase the degree of openness in their
innovation process. The lack of adequate knowledge and resources
has been widely regarded as one of the primary obstacles to Chinese
indigenous innovation. As an example, the brain drain and
crowding-out effect phenomena caused by the entry of multinational
companies (Gao et al., 2007) in the past decades in China has been
argued to create obstacles. Hence, it is assumed that learning from
external partners or sources appears to be an inevitable choice for
Chinese rms wishing to acquire knowledge and build their own
capabilities. Despite this logic, more knowledge-/skill-related impediments to innovation do not automatically facilitate or even spur open
innovation because of the many specic challenges in open innovation.
Our nding is consistent with those in the existing literature that argue
that most challenges to open innovation stem from internal rm (or
organisational) weakness, in particular knowledge- and skill-related
obstacles (Chesbrough and Crowther, 2006; Vrande et al., 2009),
together with the potential costs and risks of open innovation activities
(Keupp and Gassmann, 2009).
197
The second major nding of this study is that there is considerable

variation in rms responses to constraints and risks across different
types of ownership. Foreign invested rms appear to be the most
responsive to impediments by widening and deepening their openness
in innovation. Privately owned rms have made signicant efforts to
respond to market- and institution- and nance- and risk-related constraints but not to knowledge- and skills-related constraints, whereas
state-owned rms appear to be least responsive in terms of using open
innovation to overcome the constraints and risks they face. This is
consistent with the feature of a half-marketised transitional economy
such as China and the characteristics of a developmental state. SOEs
have privileges in access to nancial resources and access to the market.
They can obtain loans more easily from the large banks that are state
owned. As a result of restrictions on market entry in some key industries, they can also enjoy monopoly power. This nancial and status
advantage also enables SOEs to attract and accumulate substantial
human capital. As a result, this type of rm may face fewer constraints
and may have other alternative methods to overcome constraints and
risks.
On the other hand, this may also reect the fact that SOEs are more
rigid and closed to embracing external linkages and new modes of
innovation management. Moreover, SOEs may be less likely to be
chosen as a partner in open innovation because of the bureaucratic
image of the SOEs and the state-owned or controlled nature of their
ownership. The foreign invested rms are responsive to all the constraints by deepening and widening the openness of innovation. This
suggests that they are driven by market forces and make necessary
responses to maximise prots; however, they are also condent and
capable of employing open innovation to overcome the impediments.
Of course, we need to note that the average degree of openness of the
FIEs is not signicantly different from that of the SOEs, suggesting that
the FIEs have not become fully embedded in the local innovation
system. This can be explained by the fact that most of the R&D function
of MNEs is located at the company headquarters rather than at the
subsidiaries and that many FIEs in China are not interacting with
potential local research partners because of their concerns regarding
weak intellectual property protection in China.
Privately owned rms are more responsive than SOEs. They use open
innovation to overcome the constraints to innovation, especially
198
institution- and market-related constraints and nance- and risksrelated impediments. This is consistent with the current weak political
status of POEs and the lack of nancial support to them in China.
Interestingly, POEs do not appear to be responsive to knowledge- and
skills-related constraints through adoption of open innovation. This
may be explained by the fact that, despite requiring external innovation
sources, it is difcult to broaden or deepen openness in a genuine way if
the rm does not possess adequate absorptive capacity and previous
collaboration experiences. In other words, when absorptive capacity is
not enough to make innovators condent of going outside, some rms
may not choose an open strategy and become more conservative
instead. Our statistical results also show that absorptive capacity
(proxied using R&D intensity) has signicant effects on openness,
which suggests that absorptive capacity is essential to launching openness in innovation. This may explain why privately owned rms do not
appear to be responsive to knowledge- and skills-related constraints
using open innovation.
Third, rms responses to the various constraints appear to vary by
rm size and industry, especially in terms of the depth of openness.
Smaller rms are more likely to undertake deep engagement in open
innovation to overcome nancial constraints and share risks in innovation, whereas larger rms are more likely to do so to overcome constraints in knowledge and skills. Firms in high-technology industries are
more likely to respond to constraints in nance and skills by engaging in
deeper open innovation. However, in the light of institutional constraints, rms in medium- and low-technology industries are more likely
to use open innovation to overcome this constraint than rms in a hightechnology industry.
Finally, our ndings reveal some essential differences with the existing conclusions in the literature developed in the context of Western
countries. Innovation leaders, mainly from developed countries, adopt
open innovation as a new innovation model as a result of the strong
capacity and resources they enjoy. In contrast, latecomers, especially
in developing countries, do their best to increase openness in innovation, but they have to overcome additional constraints in their capability and resources if they are to learn and catch up.
Admittedly, this research has some limitations. First, the determinants of openness in emerging economies are diverse and variable
because of different types and processes of open innovation (outside-
199
in, inside-out, coupled, etc.), which should be examined more deeply

and comprehensively. Future research will benet from more comprehensive analyses based on a purposefully designed survey to understand
the drivers of open innovation. Moreover, future studies should employ
panel data sets to capture the dynamics in rms responses to the various
constraints and risks and the evolution over time. Finally, opening up
within and outside the country will present different challenges and also
different effects on a rms innovation performance. Future research
should explore these differences in the determinants and impact of
openness along these two dimensions.
APPENDIX 7.1
Detailed questions used as basis of the estimation of the constraints and
risks scores
1. Market- and institution-related constraints:
(1) Market is dominated or controlled by major competitors (M1)
(2) Over-competition in the industry (M2)
(3) Prot rate in the industry is too low (M3)
(4) Peer rms in the same industry rarely engage in innovation (M4)
(5) Technological gap between the rm and main competitors is too
large (M5)
(6) There is uncertainty about the demand of the new product (M6)
(7) No innovation due to: the market lacks demand for technology
innovation
(8) No innovation due to: lack of IPR protection to innovation
outcome
(9) No innovation due to: uncertainty or restriction of the standards
in the industry
2. Financial constraints and risk-related impediments
(1) Economic risks are too large in relation to technological innovation (C1)
(2) Innovation requires too high an investment (C2)
(3) Financing costs for innovation are too high (C3)
(4) Existing nancing channels cannot support large innovation
(C4)
(5) No innovation due to: lack of the necessary funds for innovation
200
3. Knowledge- and skill-related impediments:

(1) Lack of science and technology personnel (K1)
(2) Lack of accumulated technological capacity (K2)
(3) Lack of relevant technology information (K3)
(4) Lack of relevant market information (K4)
(5) Lack of partners for innovation (K5)
(6) No innovation due to: lack of necessary human resources for
innovation
(7) No innovation due to: existing technology available
The dual role of universities

in industrial innovation
Comparing China and the UK
8.1 Introduction
As an important player in national and regional innovation systems,
universities have received increasing attention with respect to their role
in innovation, competitiveness and wider social and economic development. Universities are widely regarded as a major contributor to advances in basic scientic research and the creation of innovation of great
novelty. Moreover, recent research also suggests that the role of universities is multi-faceted, covering educating, knowledge creation in the
form of scholarly publications and patents, problem-solving activities
and public space provision (Hughes, 2010). However, most of the
received wisdom on the role of universities is based on experiences
and evidences from the developed countries. The role of universities in
innovation in developing countries, especially the middle-income
emerging economies, has received much less attention.
This chapter attempts to examine the role of universities in the
emerging developing economies using comparative rm-level analysis
of the innovation effects of universities in China and the UK. It argues
that universities have two roles in industrial innovation in emerging
countries in addition to their educational contribution. One is that they
serve the traditional role of knowledge creation in the elds of basic and
applied scientic research and hence the generation of innovation of
great novelty. The other is that they decipher and adapt advanced
foreign technology that is transferred to the developing countries.
Their collaboration with industry facilitates the birth of many forms
of innovation that are new to the country and/or particular rms in the
emerging economies. The importance of the latter role is likely to be
greater than the former in the emerging economies. China provides a
201
202
good case for research given its quickly improving performance in

innovation in recent years and the great effort made by the government
in encouraging university-based research and technology transfer
(Wu, 2007; Fu and Soete, 2010). The UK provides a classical benchmark for the analysis as one of the major developed economies in the
world and a country with an excellent university sector.
The remainder of the chapter is organised as follows: Section 8.2
discusses the literature and the theoretical framework. Section 8.3
presents a brief overview of the university sector and its role in the
national innovation system in China. Section 8.4 discusses the methodology and data. Section 8.5 presents the results. Section 8.6 presents
conclusions.
8.2 The literature and theoretical framework

The literature on national and regional innovation systems has highlighted the role of universities in the innovation systems, not only in
training and education but also as active players in knowledge creation
and transfer (Nelson, 1986; Porter and Stern, 1999; Fu and Yang,
2009). Universities may contribute to an economy in a multi-faceted
manner through education, knowledge creation in the form of scholarly
publications and patents, problem-solving activities and public space
provision (Cosh, Hughes and Lester, 2006; Kitson et al., 2009; Hughes,
2010). They disseminate knowledge to the real economy by producing
qualied students and by interacting with rms through a number of
channels such as consulting, licensing, and co-operative research programmes (Eom and Lee, 2010). In the era of the knowledge economy,
the importance of universities in contributing to economic growth has
become an increasing focus of research (Etzkowitz and Leydesdorff,
2000; Sainsbury, 2007). Fast-paced global competition and technological change also link rms to universities not only through the discovery
of knowledge but also by aiding industrialisation (Etzkowitz and
Leydesdorff, 1997; Hwang et al., 2003).
Two contrasting views exist regarding the role of a university. The
Triple Helix thesis argues that universities should form direct links with
industry to maximise capitalisation of knowledge and hence serve the
need for economic development (Etzkowitz and Leydesdorff, 1997,
2000). In contrast, the so-called New Economics of Science states that
too close a relationship between science and industry and short-run
The dual role of universities in industrial innovation
203
policies that move resources into commercial applications of scientic

knowledge would jeopardise scientic advancement over the long run
(e.g., Dasgupta and David, 1994). Criticising any application of both
theories to developing countries, Eun, Lee and Wu (2006) suggest a
contingent or context-specic perspective on industry-university relationships and that the question as to whether a university should take
part in the functioning of an industry should be answered by considering the internal resources of the university, the absorptive capacity of
the industries and the existence of intermediate institutions.
The emergence of open innovation as a new mode of innovation
suggests that universities may play an increasingly signicant role in
industrial innovation (Chesbrough, 2003). Technological convergence,
declining transaction costs of acquiring external R&D inputs and the
shortening of product development cycle times have accelerated the trend
of utilising external sources of knowledge (Grandstrand et al., 1992).
Collaborations with various partners, both public and private, are important sources of knowledge that directly strengthen rms technological
competences and may thus increase their capacity to innovate (Freeman
and Soete, 1997; Kitson et al., 2009. Collaborations among organisations facilitate the attainment of complementary assets related to innovative labour and allow rms to achieve the goals that they cannot pursue
alone (Mowery, Oxley and Silverman, 1996; Powell and Grodal, 2005).
Moreover, collaborations in innovation are found to be complementary
to in-house R&D and facilitate inter-organisational, or international
knowledge transfer. Collaborations with customers, suppliers, higher
education institutions, and even competitors allow rms to expand
their range of expertise, develop specialised products and achieve various
other corporate objectives (Porter and Stern, 1999). With sector and size
variations, networking is found to be positively associated with innovation (Goes and Park, 1997; Tsai, 2001). Firms embedded in benet-rich
networks are likely to have greater innovative performance (Powell,
Koput and Smith-Doerr, 1996). By sharing complementary knowledge
and skills, rms can break through the bottleneck that constrains their
innovation activities and enable the innovation creation process to be
more efcient (Fu et al., 2011). UK Community Innovation Survey data
suggest that global engagement in UK rms innovation is also found to
be associated with a higher propensity to innovate (Criscuolo, Haskel
and Slaughter, 2005). Yet, as argued by Laursen and Salter (2006), the
benets from openness to external knowledge and resources are subject
204
to decreasing returns as over-searching and working with too many

partners will have negative consequences for innovation efciency.
Given the widely recognised role of universities in the national
innovation system, university-rm collaboration is argued to be crucial for the promotion of technological change (Manseld and Lee,
1996). A growing number of developed and developing countries are
seeking to use universities as important drivers of knowledge-based
economic development and change (Mowery and Sampat, 2005).
Through interaction with the science base, rms are able to access a
diversied range of knowledge sources in comparison to intra-rm
collaboration (Kaufmann and Todtling, 2001). University participation in research programmes is also found to have a positive impact on
rm patenting (Darby, Zucker and Wang, 2003). Public research is
also found to have not only initiated new R&D projects but also to
have contributed to the completion of existing projects in the U.S.
manufacturing sector (Cohen, Nelson and Walsh, 2002). In sum,
contracting out research, entering into university-industry alliances
and collaborating with university researchers formally or informally
can confer substantial advantages.
However, the role of universities in innovation is multi-faceted
(Hughes, 2010). Universities may struggle in terms of reconciling the
creation of new knowledge regardless of its commercial or innovative
value and their role as promoters of technological development
(Mowery and Sampat, 2005). The extent of development of an economy will inuence the role of its universities in several ways. First, the
extent of development of an economy will determine the level of its
indigenous technological and innovation capabilities and the major
source of technology upgrading. Owing to the lack of sufcient absorptive capacity in the local industry, there is a need to tap into the expertise
of science and engineering experts in the universities. Second, the extent
of development of an economy is often in line with the level of research
capabilities of its universities and hence the type of innovation they are
to create. Finally, the extent of development of an economy determines
the type of industrial and technology policies the government will adopt
to promote economic growth, which in turn affects the type of demand
for university research.
In the middle-income emerging economies, given the level of development and the research capabilities of the economy, universities are
likely to have a dual role in the innovation of rms in addition to
205
their educational contribution. First, innovation is costly, risky and

path dependent. This provides a rationale for developing countries to
use foreign technology acquisition as a major source of technological
development. Foreign sources of technology account for a large part of
productivity growth in most countries. In fact, most innovation activities are largely concentrated in a few developed countries: the United
States, Japan and a number of European countries. International technology diffusion will therefore be an important driver in economic
growth. Hence, in the middle-income emerging economies, which are
at an intermediate stage of development, the assimilation of foreign
technology has been a major source of technology upgrading, and
innovation. The innovation that is groundbreaking at the national or
rm level will be the major type of innovation in these economies.
Moreover, most of the rms in these economies are lacking an
absorptive capacity (Eun et al., 2006). Therefore, there is a real
demand from the industrial sector for external technological experts
to help with the assimilation and adaptation of foreign technology.
Linkages with universities will help accelerate the adoption of foreign
technology.
Furthermore, the middle-income emerging economies have a higher
education sector that is relatively stronger than that of the rest of the
developing countries. Therefore, universities and especially elite universities in the middle-income emerging economies are capable of
collaborating with the industrial sector to assimilate the transferred
foreign technology and make the adaptations necessary for the foreign
technology to t within the local technical, economic and social context. Some of the industry-university alliances are not only capable of
shallow assimilation (which facilitates the normal operation of the
imported equipment in recipient rms) but also deep assimilation of
foreign technology through reverse engineering and R&D to make
modications to the transferred foreign technology. This may lead to
the transition from imitation to innovation and the creation of innovations that are groundbreaking at the world level. Moreover, knowledge transfer from universities, which is often embodied in codied
forms (e.g., publications, patents, contract R&D projects) and which
also often contains tacit knowledge (e.g., collaborative research, informal consultation), becomes in this way an important asset in creating
learning organisations. Of course, we should not be overly optimistic
about the role of universities in this respect in the middle-income
206
emerging economies. Admittedly, there is a gap between the universities in these economies and those of the major developed economies,
especially in terms of research quality and impact. Even in one of
the elds of emergent technologies, where the emerging economies
might claim a lead based on the number of scientic publications,
further analysis of the citation share as well as surrogate indicators
show the need for China to improve its research impact (Guan and
Ma, 2007).
Finally, the level of development of an economy determines the type
of industrial and technology policies the government will adopt to
promote economic growth, which in turn affects the type of demand
for university research. For example, in the United States and the
United Kingdom (UK), research universities are simultaneously
centres of learning, the foci of basic and applied research and the
source of entrepreneurship. In Asian countries such as China, Korea
and Singapore, universities have been geared towards training and
only recently have begun to pay more attention to research. Elite
universities, prompted by government incentives, have started to contribute to technology deepening (Hershberg, Nabeshima and Yusuf,
2007). Unlike the many Western countries that have experienced a
transition in science policy from one driven by curiosity to one driven
by practical needs and uses, the Chinese government has been advocating from the start a use-driven science policy, requiring universities
to serve the national economy by solving practical problems for
industry (Hong, 2006). Universities were encouraged to collaborate
closely with industry, for example, in solving production problems for
factories (Ministry of Education, 1999; Yuan, 2002). Reections lead
to the following hypothesis:
H8-1: Universities have two roles in industrial innovation in emerging
countries in addition to their educational role.
They serve the traditional role of knowledge creation in the eld of basic
and applied sciences and hence the generation of novel innovations.
They also decipher and adapt advanced foreign technology that is
transferred to the developing countries. Their collaboration with industries facilitated the diffusion of innovation that can be considered new at
the country or rm level in the emerging economies. The importance of
the latter role is likely to be greater than the former in the emerging
economies.
207
8.3 Universities and industrial innovation in China

The swift rise of the Chinese economy has been accompanied by the
rapid development of its technological capabilities. While acquisition of
foreign technology has been a major strategy in the rst two decades
since the reforms of 1978, the decade since then has witnessed rapid
growth in domestic R&D investment. The expenditure on R&D has
increased exponentially. In 2010, the total R&D expenditure in China
was greater than that of Germany, the UK and France and ranked third
in the world, after the United States and Japan. In terms of its gross
R&D expenditure to GDP ratio, China is now moving close to the EU
average and aims to reach 2.5 per cent by 2020. Moreover, since 2000,
China has experienced a rapid surge in patent applications. The number
of patent applications from Chinese researchers to the authorities of the
so-called Triadic Patent Families has increased more than seven times
from 2000 to 2007.1 Chinas share in science and engineering articles
has risen sixfold since the mid-1990s, from 9,000 to nearly 57,000 each
year, accounting for 7 per cent of the global research output in 2007
(Fu, Pietrobelli and Soete, 2010; Gilman, 2010).
Transiting from a centrally planned to a market economy, universities in China have historically played an important role in its national
innovation system, similar to that of the science and technology system
in the former Soviet Union (Liu and White, 2001). In terms of R&D
expenditure and patents, universities and research institutes played a
leading role in China (Li, 2009). Reforms started in 1985 to render the
science and innovation system more relevant to the market and signalled a departure from the Soviet model whereby scientic research at
public research institutions and production at state-owned enterprises
were completely separated (Xue, 1997).
The mid-1980s witnessed several reforms in science policy in China.
The most signicant change was the cutting of government research
funding in order to push research organisations into the market (Hong,
2008). From 1986 to 1993, government research funding decreased at
an annual rate of 5 per cent (Zhou et al., 2003). Hence, universities
began to establish their own enterprises at that time, a practice ofcially
approved by the central government in 1991. Another wave of reform
of Chinese universities began in December 1994 when a national forum
1
These are patents applied for/granted in the United States, Europe and Japan.
208
encouraged institutional mergers and decentralisation in jurisdiction for

efciency purposes. This decentralisation reform has had the effect of
promoting collaborations between universities and local industries
(Hong, 2008). In addition, the introduction of a Chinese version of
the Bayh-Dole Act allowed universities to retain titles to inventions
derived from government funding. Due to this act as well as an act of
promoting commercialisation of innovation and development of hightech industry in 1999, Chinese universities have become even more
enthusiastic about transferring knowledge to industry (Hong, 2008).
The Chinese government has been advocating a use-driven science
policy since its establishment, encouraging universities to serve the
national economy by solving practical problems for industry (Hong,
2006). On the one hand, university-industry linkages in China are built
through licensing, consulting, joint or contract R&D and technology
services, closely resembling how universities in the West interact with
industry. On the other hand, a second form of use-driven innovation
occurs as a result of university-afliated or university-run enterprises
(Zhang, 2003; Ma, 2004), a system that is uniquely Chinese. Chinese
universities since the market-oriented reforms have had strong incentives to pursue economic gains and strong internal (R&D and other)
resources to launch start-ups and thus established their own rms, given
the low absorptive capacity of industrial rms and underdeveloped
intermediary institutions (Eun et al., 2006). Government-driven spinoff formation has proved an appropriate solution for technology transfer at Chinese universities (Kroll and Liefner, 2008). Economy-wise, the
economic reforms have led to a gradual evolution of major players in
the national and regional innovation system. The importance of the
industrial sector in the innovation system has been increasing over the
years, which, combined with a varied performance in the reform of
enterprises and opening up to foreign direct investment, has led to
increased disparities in innovation across Chinese regions (Fu, 2008;
Li, 2009).
8.4 Methodology and data

The empirical test of Hypothesis 81 is carried out in two stages.
First, a statistical test is used to examine the impact of the university
linkage on rms innovation with differing degrees of novelty, that is,
innovations that are groundbreaking at world, country or rm levels.
209
Second, we compare the pattern observed from Chinese data with that
contained in UK data. To explore the effect of the technological and
cultural gap on the knowledge transfer through innovation collaboration, we distinguish universities located in a rms own country, the
newly industrialised economies, the EU, the United States, Japan and
other countries.
To assess the impact of university collaboration on the innovation
performance of industrial rms, we regress a rms innovation output
on its collaboration with universities while controlling for a vector of
rm- and industry-specic characteristics.
Measurement of the dependent variable

Here, we measure innovation output by the percentage of innovative sales
in total turnover. Innovation could be measured in different ways. One
way is to use a dummy variable that equals 1 for innovation and 0 for no
innovation. This method, however, would omit detailed information with
regard to the extent of innovation. A second widely used measure is R&D
expenditure. R&D expenditure itself is, however, in fact one of the inputs
to innovation. A third widely used indicator of innovation output is the
number of patents (Jaffe, 1989; Acs, Luc and Varga, 2002). Although an
indicator based on the number of patents has its advantages, it also suffers
from the validity problem that patents might not adequately reect the
commercial success and value of new and adapted products (Acs et al.,
2002). Some studies use innovation counts (Acs, Luc and Varga, 2002;
Feldman, 1999). These also have limitations, however, in terms of reecting the depth and breadth of innovation success.
For these reasons, we use the sales of new or improved products as a
measure of innovation output, as this information is available in the
survey data set. In the survey, rms are asked whether, besides being
new to their rm, the innovation was also new to the market. This
allows a distinction between innovations of the latter kind which may
be termed novel and innovations of the former kind which may be
considered as diffusionary innovations. As we are interested in the
different roles of universities in the creation of groundbreaking novel
innovation and in translating, deciphering and adapting transferred
foreign technology, we use two dependent variables: the proportion of
sales accounted for by products that were groundbreaking at the world
level and, second, were new to the country or rm.
210
Measurement of direct university contribution

We consider a number of variables to capture the direct university
contribution to rm innovation through innovation collaboration. We
include an indicator of the extent to which the rm co-operates with
other organisations and institutions in the course of its innovation
activity. This is proxied by a co-operation dummy that equals 1 if
rms co-operated in any innovation activities with other enterprises or
institutions, such as universities and public research institutions, suppliers of equipment/services/software, clients, competitors, consultants
and commercial labs, and 0 otherwise. This allows for a direct test of the
question as to whether collaboration signicantly inuences rm innovation performance. To highlight the role of domestic universities in
emerging economies, we distinguish universities located in a rms own
country, the newly industrialised economies, the EU, the United States,
Japan and other countries. This also allows us to examine the effects of
the technological and cultural gap and hence the appropriateness of
foreign knowledge on the strength of benets from collaborations with
universities internationally.
The control variables

The control variables include a group of variables that focus on the
extent to which the output of product innovations by a rm is a function
of the resources committed to innovative activity. These resources are,
rst, intramural R&D expenditure and extramural R&D expenditure
of the rm. Investment in research and development is often found to be
a signicant determinant of innovation. Firms engaged in R&D are
more likely to innovate because R&D directly creates new products
and processes and also because these rms are more receptive to new
external ideas. However, some economists (e.g., Baldwin 1997) argue
that R&D is neither a necessary nor a sufcient condition for innovation. Moreover, control of the size of extramural R&D is also important
as a control over the effects of other types of collaborations, for example, collaboration with suppliers, customers, other rms in the same
industry and other rms within the company group. Labour force skills,
particularly qualied scientists and engineers, is another widely recognised critical factor that contributes to rm innovation performance
(Hoffman et al., 1998; Porter and Stern, 1999). To capture this important element and the extent to which the lack of qualied R&D
211
personnel can constrain innovative activity, a dummy variable that

equals 1 for rms reporting a lack of qualied personnel as being of
medium and high importance and 0 for others is also included as a
control variable. These inputs not only directly contribute to innovation
but also enhance the rms capacity to recognise and absorb relevant
external resources for innovation (Cohen and Levinthal, 1990).
We also include other variables to capture size and age effects as well
as industry sector-specic effects. The extent to which a rm may be
able to exploit its innovative activity may depend on its size per se and
on the degree of competition in its nal product markets. Larger rms
have a greater range of market opportunities through which to exploit
innovative opportunities. The size of the rm can therefore act as a
proxy for this enhanced incentive to innovate. From the point of view of
smaller rms, the existence of a dominant market position by large rms
may inhibit their access to the market and hence their ability to translate
innovative activity into a signicant proportion of new products in their
nal sales. However, large rms face conicting possibilities that may
arise from the presence of dominant positions. Moreover, rm age is
likely to be associated with rms innovation activity: older rms may
have accumulated more experience and knowledge and be more capable in innovation. Alternatively, older rms may be constrained by
organisational rigidity and hence are less active in innovation. Finally,
because technological and innovation opportunities may occur
unevenly across sectors, we include industry dummy variables to
proxy for these effects. The full list of variables is summarised in
Table 8.1. The correlation coefcients of the variables in the China
and UK samples are shown in Appendixes 8.1 and 8.2.
Two estimation problems arise in this model. The rst is that the
dependent variable, the percentage of innovative sales, is constrained to
a value between 0 and 100 and takes a value of zero in a large proportion of sample. The ordinary least squares (OLS) estimates would thus
be biased. Therefore, a Tobit model should be introduced to reduce the
problem (Tobin, 1958). Second, a number of rms have not undertaken
any R&D activity at all and therefore have no sales of new or signicantly improved products. So, there is a selection effect based on the
decision to innovate or not. A Hurdle model that was originally suggested by Cragg (1971) as a generalised form of the Tobit model needs
to be employed to allow for the fact that rms decide whether or not to
innovate and, with respect to those that are innovative, the extent to
212
Table 8.1 Denition of variables

Variable
Denition
newsal
newsaln
Percentage of new sales

46.796
Percentage of sales of products that are groundbreaking
6.798
in world terms (novel)
Percentage of sales of products that are new to China or 40.719
the rm or are signicantly improved (diffusionary)
Ln(intramural R&D expenditure)
5.105
Ln(extramural R&D expenditure)
0.906
Firm age
16.890
Firm size dummy equalling 1 for large rm and 0 for
0.714
small rm
Human capital constraints dummy variable, 1 = the
0.797
importance of lack of qualied personnel to innovation
is medium and high; and 0 = low or unimportant
Innovation co-operation dummy variable, 1 = yes, 0 = no 0.634
Dummy variable, 1 = co-operate with other rms within
0.392
an enterprise group; 0 = no
Dummy variable, 1 = co-operate with suppliers; 0 = no
0.407
Dummy variable, 1 = co-operate with customers; 0 = no
0.396
Dummy variable, 1 = co-operate with competitors or
0.291
other rms in the same industry; 0 = no
Dummy variable, 1 = co-operate with private R&D
0.243
institutions; 0 = no
Dummy variable, 1 = co-operate with universities and
0.482
public research institutions (PRIs); 0 = no
0.476
PRIs in China; 0 = no. Dummy variable, 1 = co-operate
with universities and PRIs in the same country; 0 = no
0.012
PRIs in newly industrialised countries in East Asia
(Hong Kong, Taiwan, Singapore, Korea); 0 = no
0.019
PRIs in Europe, the United States and Japan; 0 = no
0.006
PRIs in other countries not listed above; 0 = no
Dummy variable (UK), 1 = co-operate with universities
0.228
and public research institutions (PRIs); 0 = no
0.219
and PRIs in the UK; 0 = no
newsald
lrdin
lrdex
age
Size4
lack_hc1
co
cogd
cosd
cocd
copd
coprid
counid
couni1
couni2
couni3
couni4
counid_2
couni1_2
Mean
213
Table 8.1 (cont.)

Variable
Denition
couni2_2

and PRIs in Europe; 0 = no
and PRIs in other countries not listed above; 0 = no
couni3_2
Mean
0.029
0.016
which they will do so (Mairesse and Monhen, 2002). The signicance of

the presence of the selection effects is indicated by Rho statistics, which
reect the correlation between the error terms of the two equations. If
there are signicant selection effects, the Generalised Tobit model (selection in censored data) is preferred. Otherwise, we utilise the standard
Tobit model. However, in this study, as innovative rms dominate the
China data set (about 95 per cent of rms reported having innovated in
terms of their products) and the UK data set has similar characteristics,
selection bias is not a signicant problem.2 Nevertheless, we also report
the results from the Generalised Tobit model as a robustness check.
Data
The research principally uses the 2008 Chinese national innovation
survey of 1,408 manufacturing rms in China: this contains data on
rms innovation activities from 2005 to 2007. The survey was carried
out by the National Statistical Bureau. It covers 42 cities in China in
both the coastal and inland regions. A total of 1,408 valid responses
were received, with a response rate of 83.6 per cent. The questionnaire
in the Chinese innovation survey was designed by Tsinghua University
and demonstrates high consistency and comparability with the design of
the European Community Innovation Survey (CIS). The large and
innovative rms that are responsible for most of the R&D activities
that take place in China dominate the survey. After careful data cleansing to exclude observations with missing values of the necessary variables, the nal data set used in the estimation contains 802 rms, of
which 95 per cent have innovated in their products. Therefore, the
2
This is also attested to by the estimated rho statistics of the selection model. Results
are available from the authors subject to request.
214
results of this study reect the role of universities in innovative Chinese

rms rather than in Chinese rms generally. This is a limitation of the
research that we bear in mind when drawing conclusions.
We then benchmark the innovation effects of universities in China
against those in the UK. Our sample of UK rms is drawn from the
2005 Fourth UK Community Innovation Survey carried out by the
Ofce for National Statistics on behalf of the Department for
Innovation, Universities and Skills.3 The UK survey covered all enterprises with 10 or more employees in almost all industries. The survey
received valid responses from 13,986 UK enterprises, including 4,296
rms in the manufacturing sector. The response rate of the survey is
58 per cent. It covers nine regions in England plus Scotland, Wales and
Northern Ireland. A postal questionnaire was used and responses were
voluntary. The survey provides information on UK innovation during
2002, 2003 and 2004. Admittedly, there is a three-year gap between
the UK and the Chinese innovation surveys. However, it is reasonable
to assume that the pattern of innovation and the role of universities are
unlikely to have experienced big changes in a mature developed economy such as the UK over a short period of three years. Hence, the
pattern of the role of universities in rms innovation from the survey
is likely to be equivalent to that experienced from 2005 to 2007: this
can in fact be proven by detailed comparison of the persistence and
changes in UK innovation within UK CIS4 and CIS5 from 2002 to
2006 (DTI, 2008).
These two national surveys cover rms in the whole economy. Such
data cannot be used to make straightforward comparisons because
countries differ in the distribution of activity across industries and the
distribution of innovative rms by size and sector. These factors have
an impact on the requirements for, and likelihood of, access to the
universities. Therefore, we match the cleansed smaller Chinese survey
data set to the UK survey data set by size of rm, by industry and by
product innovation in the prior three years. The last criterion is
important in the current study because innovative rms dominate
3
A copy of the questionnaire can be found at http://www.berr.gov.uk/dius/innova

tion/innovation-statistics/cis/cis4-qst/page11578.html. Department for
Innovation, Universities and Skills was formerly the Department of Trade and
Industry or DTI. In 2007, its functions were transferred to the Department for
Business, Enterprise and Regulatory Reform (BERR) and the Department for
Innovation, Universities and Skills (DIUS).
215
the Chinese data and it is important to make the data set comparable in
this aspect. By this mechanism, we are able to compare the role of
universities in the intensity of producing innovation of different
degrees of novelty amongst innovative rms of the same sizes and
industry technologies in the UK and China. Because of structural
differences between the two surveys, we achieved a comparable UK
sample of 793 rms.
8.5 Results
Figures 8.1 and 8.2 report the extent of utilisation of external resources
in innovative Chinese rms in terms of the percentage of rms that
engaged in innovation collaboration with various types of partners.4
On average, nearly half of the surveyed Chinese rms report that they
collaborated with listed external organisations. Interestingly, universities are the most popular collaborator for Chinese rms, which is not
surprising given that, as discussed earlier, historically universities and
public research institutions (PRIs) dominated the innovation system in
China and a strong government policy pushed the development of
university-industry linkages. Most of the universities collaborate with
Chinese universities and around 10 per cent of the rms that collaborate
with universities collaborate with foreign universities. In comparison,
about 40 per cent of the British rms in the matched sample reported
cooperation with other organisations, at least in some of their innovation activities. Customers and suppliers appeared to be the most frequently used collaborators of British rms: around 30 per cent of rms
have collaborated with customers or suppliers in innovation activities.
Universities and PRIs ranked as the fourth most frequently used collaborators, ahead of competitors and private R&D institutions. Domestic
universities remain the main university collaborator, but a higher proportion (18 per cent) of collaborative British rms collaborated with
4
The wording in the UK survey on this question is co-operation. Although there is

subtle difference in English between collaboration and co-operation, the Chinese
wording used in the survey (he zuo) does not imply this subtle difference. Firms
may regard both arms-length close co-operation and collaboration as being he
zuo. We therefore translate the wording into collaboration, which may include
recursive and sustained interactions in addition to arm-length co-operation. This
may explain the difference in the China and UK samples in which more Chinese
rms report having collaborated with external organisations for innovation.
216
Percentage of innovative Chinese firms reporting R&D

collaboration with various types of partners
Co with universities and PRIs
35
Co with private R&D institute
18
Co with competitor
21
Co with customer
30
Co with supplier
31
Co with others within group
29
Co with any external organization
48
10
20
30
40
Percentage
50
60
Figure 8.1 Percentage of innovative Chinese rms reporting R&D collaboration with various types of partners
Percentage of Chinese firms reporting collaboration in
R&D with universities in various countries
Others
0.5
EU/US/Japan
1.6
HK, TW, Singapore, Korea
1.0
China
34
10
15
20
25
Percentage
30
35
40
Figure 8.2 Percentage of Chinese rms reporting R&D collaboration with

universities in various countries
foreign universities. In view of the language and cultural differences

between the two economies, direct comparisons of absolute percentages
may not produce rigorous results. However, the pattern of collaboration and the relative importance of various collaborators in the two
economies still provide plausible insights.
217
Universities and industrial innovation in China

The estimated results of the role of universities in rm innovation in
China using the standard Tobit model are reported in Table 8.2.
Columns (1) to (3) report the regression results using the percentage
of sales of products that are groundbreaking innovations in world terms
as the dependent variable; columns (4) to (6) report the results of
regressions using innovations that are new to the country or rm and
signicantly improved products as the dependent variable. The results
in columns (1) and (4) suggest that collaborations with other rms or
institutions have a positive impact and are signicantly associated with
the creation of innovations that are new to the world and new to the
country/rm. The magnitude of the estimated coefcients is of similar
size, but those in the diffusionary innovation regression are of higher
signicance. However, as shown in columns (2) and (5), collaboration
with universities does not appear to have contributed signicantly to the
process in the country on average. This suggests that although there are
individual cases of successful university-industry collaboration induced
innovation in China, on average, the contribution of universities to rm
innovation is not signicant nationwide. Although universities have
formed an effective linkage and become a critical source for the industrial innovations in some regions in China, such as Beijing, the fast
growth of high-technology industries in many other regions is driven
mainly by other sources (Chen and Kenney, 2006).
Breaking down universities according to their country of origin, the
estimated results exhibit some interesting ndings in columns (3) and
(6). Collaboration with domestic universities exhibits a positive but
insignicant effect on novel new sales, which can probably be explained
by the level of quality and impact of domestic universities in comparison
to the world innovation frontier during the sample period (Guan and
Ma, 2007). However, the effect regarding diffusion of sales of innovations that are new to the country or rm or signicantly improved
products is positive and statistically signicant. Interestingly, international innovation collaboration between Chinese rms and universities
in the newly industrialised economies, namely Hong Kong, Taiwan,
Singapore and Korea, appears to have a signicant and positive effect
on the generation of innovations by Chinese manufacturing rms.
Moreover, rms that have collaboration in innovation with universities
in countries other than the NIEs and Europe, the United States and
Table 8.2 Universities and rm innovation in China: Tobit model estimates

Novel innovation
VARIABLES
co with other organisations
1
model
2
model
4
model
5.344
(5.205)
co with universities in NIEs

co with universities in US/EU/Japan
co with universities in other countries
2.106**
(0.865)
lrdex
5.967
(5.553)
5
model
6
model
9.828***
(3.321)
co with domestic universities
size4
3
model
10.40**
(5.111)
co with universities
lrdin
Diffusion innovation
1.299
(0.937)
1.623**
(0.649)
4.837
(5.673)
4.276
(3.513)
1.345
(5.094)
35.67***
(11.25)
17.29
(10.72)
35.32**
(15.97)
1.203
(0.921)
1.690***
(0.647)
3.428
(5.642)
2.949***
(0.544)
3.446
(3.728)
3.007***
(0.589)
0.0666
(0.46)
5.144
(3.932)
5.954*
(3.532)
5.64
(17.1)
13.58
(11.04)
8.254
(16.5)
2.929***
(0.595)
0.0009
(0.461)
5.629
(3.912)
Industry dummies
Constant
0.257
(0.159)
1.933
(5.134)
yes
yes
0.244
(0.168)
2.811
(5.171)
yes
yes
0.202
(0.161)
2.645
(5.067)
yes
yes
0.0266
(0.0575)
2.605
(3.63)
yes
yes
0.024
(0.0603)
6.097
(3.874)
yes
yes
0.0467
(0.0622)
6.154
(3.869)
yes
yes
Observations
F statistics
Log Likelihood
928
6.293
1454
817
5.218
1298
817
5.73
1291
910
7.746
3741
802
5.116
3320
802
3.445
3314
age
lack_hc1
Note: Robust standard errors in parentheses; ***p<0.01, **p<0.05, *p<0.1.
220
Japan, such as Russia, Israel, India and Brazil, have a signicantly

higher proportion of sales on accounted for by products that are new
to the world. Surprisingly, linkages with universities in the major industrialised economies (i.e., the United States, Japan and Europe), although
showing a positive effect, involve an estimated coefcient that is not
statistically signicant. This may be explained by the technology and
culture gap between China and these industrialised economies and the
appropriateness of the foreign knowledge of the receiving economy.
Further research is needed to investigate why collaboration with this
group of highly regarded universities is less fruitful for Chinese industrial enterprises.
Firms intramural R&D appears to be insignicantly associated with
their sales intensity of novel products after controlling for extramural
R&D spending but is positive and signicantly associated with their
diffusion of new sales. This result is consistent with the work of Fu and
Gong (2010), which examines the effect of indigenous R&D activities
on technology upgrading in China using a large rm-level panel data set
produced by the National Statistical Bureau. In contrast, rms spending on extramural R&D activities exerts a positive and signicant effect
on rms novelty of new sales but not on the diffusion of innovation.
This highlights the importance of the utilisation of external innovation
resources and extramural R&D activities for the creation of innovation
involving products that are groundbreaking in world terms by middleincome emerging economies. Firm size and age do not appear to affect
the percentage of innovative sales of rms signicantly. Although the
estimated coefcient of the constraints in the R&D personnel dummy
bears the expected negative sign, it is not statistically signicant either.
Table 8.3 reports the results of a robustness check using the
Generalised Tobit model to correct for potential selection bias. The
estimated results are broadly consistent with the standard Tobit
model estimates, especially in respect of the effect of university-industry
collaboration. The estimated coefcient of the university collaboration
variable remains positive and statistically insignicant, whereas the
pattern of the inuence of universities by country of origin remains
highly similar to that in Table 8.2. The effect of collaboration with
domestic universities remains signicant for the diffusion of new sales
but not of novel new sales. International innovation collaboration with
foreign universities in the NIEs, major industrialised economies and
other countries all demonstrate a positive and signicant effect. The
Table 8.3 Robustness check: Generalised Tobit model estimates

Novel innovation
1
2.111
(1.414)

co with universities in US/EU/Japan
co with universities in other countries
0.409*
(0.213)
lrdex
age
0.729
(1.502)
0.0534*
(0.030)
6.250**
(2.824)
size4
2.3
(1.442)
lrdin
0.357
(0.224)
0.105
(0.185)
0.856
(1.498)
0.0563*
(0.030)
3.967
(2.786)
1.045
(1.402)
12.09**
(5.914)
8.429*
(4.696)
17.08**
(8.088)
0.267
(0.227)
0.119
(0.184)
0.397
(1.501)
0.0479
(0.030)
2.308***
(0.427)
5.874**
(2.962)
0.0213
(0.058)
2.331***
(0.453)
0.197
(0.363)
5.220*
(2.968)
0.0237
(0.059)
5.239*
(2.779)
0.693
(11.99)
8.014
(9.338)
5.956
(15.87)
2.329***
(0.45)
0.108
(0.363)
5.732*
(2.979)
0.0403
(0.059)
Table 8.3 (cont.)

Novel innovation
lack_hc1
Industry dummies
Constant
Observations
Log Likelihood
0.205
(1.613)
yes
yes
817
3366
0.468
(1.573)
yes
yes
817
3367
0.484
(1.557)
yes
yes
817
3354
4.728
(3.171)
yes
yes
802
3815
5.600*
(3.104)
yes
yes
802
3819
6.201**
(3.112)
yes
yes
802
3813
223
level of signicance of the estimated coefcients of the NIE and other

university collaboration dummies is greater than that of the US/Japan/
EU university collaboration dummy, indicating the effectiveness of the
former two types of international university linkage. This is, to a certain
extent, consistent with the ndings from Table 8.2. As regards the
control variables, smaller rms appear to have greater diffusion of
new sales, and younger rms appear to create more novel innovations.
In sum, the results on the role of universities in rm innovation are
robust in the main, having allowed for any possible selection bias.
The collaboration variable is arguably determined simultaneously
with the dependent variable of innovation. In other words, there
might be a potential endogeneity problem. For example, rms that
collaborate with other rms and universities are more likely to have
more innovative sales. However, it is also possible that more innovative
rms might collaborate to a greater extent with other rms and universities. Moreover, they are also more likely to be invited into any
innovation collaboration by other organisations. To deal with the
potential problem of endogeneity, we employ an instrumental variable
regression technique. The instrumental variables used are all the exogenous variables in the model with the addition of four extra exogenous
variables, including a rm location dummy that indicates whether a
rm is located in the six university-concentrated cities, a group dummy
that equals 1 for rms belonging to a corporation group, the importance
of information from universities for rm innovation and competition in
the industry. Moreover, the use of industry dummies in the regressions
is also designed to mitigate part of this potential endogeneity problem.
We test whether the assumption of endogeneity is borne out by the data
at hand. The Wald tests of exogeneity of the collaboration variables
suggest no signicant endogeneity problem. Therefore, the standard
Tobit model estimates are preferred to the instrumental variable
model estimates. Nevertheless, we report the estimated results in
Table 8.4 as a robustness check. Consistent with the picture revealed
in Tables 8.2 and 8.3, the effect of university collaboration remains
insignicant for rm innovation, of both novel and diffusionary types.
However, the impact of the general collaboration variable also becomes
insignicant using the instrumental variable estimates. Note, however,
that because the Wald test of exogeneity indicates no signicant endogeneity problem, the standard Tobit model estimates should be used as
the valid empirical results for the research.
224
Table 8.4 Robustness check: instrumental variable model estimates
Novel innovation
lrdex
age
size4
lack_hc1
Industry dummies
Constant
Observations
Log Likelihood
Wald test of exogeneity (p-value)
0.665
(1.085)
1.308*
(0.763)
0.24
(0.164)
3.417
(5.762)
1.606
(6.547)
Yes
Yes
816
1691
0.424
13.28
(12.53)
18.68
(16.23)
lrdin
7.868
(11.86)
0.998
(1.012)
1.523**
(0.751)
0.24
(0.169)
4.733
(5.675)
2.302
(5.323)
yes
yes
816
1679
0.768
3.637***
(0.744)
0.466
(0.545)
0.00426
(0.0657)
4.457
(4.129)
10.05*
(5.407)
yes
yes
801
3693
0.077
0.775
(8.565)
3.172***
(0.66)
0.164
(0.518)
0.0181
(0.0661)
5.526
(3.923)
6.408
(4.136)
yes
yes
801
3680
0.473

Instrumental variables used are all the exogenous variables in the model and four extra
exogenous variables including a rm location dummy indicating whether a rm is
located in the six university-concentrated cities, a group dummy that equals 1 for rms
belonging to a corporation group, the importance of information from universities for
rm innovation and competition in the industry.
Universities and industrial innovation in university-concentrated

Chinese cities
Chen and Kenney (2007) found that each Chinese region reacted differently to the government policy of promotion of university-industry linkage.
In Beijing, universities and research institutions are a critical source of
knowledge. However, in Shenzhen, the rapid growth of high-technology
rms did not rely on direct linkages with universities. Moreover, geographical proximity to universities facilitates greater industry-university
225
collaboration. As the geographical distribution of research universities in

China is uneven, the impact of universities may be greater in these cities but
weak in the remaining cities and regions. Table 8.5 reports estimated results
of the contribution of universities to rm innovation in selected universityconcentrated cities, namely Beijing, Shanghai, Nanjing, Xian, Wuhan and
Chongqing. Consistent with the pattern shown in Tables 8.2 and 8.3,
Table 8.5 Universities and rm innovation in selected universityconcentrated cities in China
Novel innovation Diffusion innovation
1
1.297
(14.14)
5.047
(28.14)
co with universities in US/EU/Japan 26.98
(25.52)
co with universities in other
countries
50.65***
(12.52)
lrdin
0.429
(2.307)
lrdex
0.372
(1.603)
age
0.673*
(0.385)
size4
23.57
(20.84)
lack_hc1
13.99
(16.87)
Industry dummies
Yes
Constant
Yes
30.68***
(6.164)
2.076
(1.375)
1.184
(1.078)
0.121
(0.157)
38.92***
(11.02)
6.06
(11.10)
yes
yes
Observations
Log Likelihood
117
498.8
123
175.6
15.31*
(8.044)
3.239
(11.00)
4.418
(8.481)

The six selected university-concentrated cities are Beijing, Shanghai, Nanjin, Wuhan,
Xian and Chongqing.
226
domestic Chinese universities again appear to have a signicant effect

on the creation of diffusionary innovations in Chinese rms, whereas
their effect on novel innovation is insignicant. However, the size of
the estimated coefcient of the domestic universities variable is almost
three times that in Table 8.2, suggesting a greater innovation effect by
universities in these major cities than in the economy as a whole. The
effect of international collaboration with universities in other countries also appears to be much larger in these major cities than in the
whole economy. However, the impact of collaboration with universities in NIEs loses its statistical signicance in the six-city small sample. This is probably because universities in the NIEs collaborate more
with rms in some cities in the coastal medium- or small-sized cities,
such as Shenzhen, Guangzhou, Zhejiang and Fujian, rather than those
six domestic university-concentrated major cities.
Universities and industrial innovation in the UK

Figures 8.3 and 8.4 report the proportions of rms that collaborate in
R&D with external partners and with universities in particular from
2002 to 2004. On average, 39 per cent of British rms have engaged
Percentage of UK firms reporting R&D collaboration with
various types of partners
Co with universities and public
R&D institutes
23
20
Co with private R&D institutes

Co with competitor
16
Co with customer
31
Co with supplier
30
Co with others within group
25
39
Co with any external organisations

0
10
15
20
25
30
35
40
45
Percentage
Figure 8.3 Percentage of UK rms reporting R&D collaboration with various

types of partners
Note: For matched UK sample.
227
Percentage of UK firms reporting in R&D collaboration

with universities in various countries
Universities and public R&D

institutions in other countries
Universities and public research

institutions in Europe
Universities and public research

institutions in UK
22
10
15
Percentage
20
25
Figure 8.4 Percentage of UK rms reporting R&D collaboration with universities in various countries
Note: For matched UK sample.
with external organisations for collaborative innovation. The most

popular partners for collaborative innovation in the UK are customers
and suppliers. On average, about 31 per cent of rms report that they
have collaborated with customers for innovation, and 30 per cent of
rms report collaboration with suppliers. The proportion of rms that
have collaborated with universities is 23 per cent, which is 12 per cent
lower than that in the China sample. This may be partly the result of the
strong government encouragement of university-industry linkage in
China. But it may also be because the China sample is biased towards
innovative large rms and hence not a representative sample of all
the rms in China, whereas that of the UK survey is. Most of the
British rms collaborated with domestic universities. About 3 per cent
of the rms collaborated with universities and research institutions in
Europe, and only about 2 per cent collaborated with such academic
partners in other countries.
Table 8.6 reports the Tobit model estimates of the UK sample match
by size, sector and product innovation status. We also report the estimates of the full UK manufacturing sample to examine any differences
between the matched sample and the UK rm population that may be
caused by structural differences in industry, size and innovativeness
Table 8.6 Universities and rm innovation in the UK: standard Tobit model estimates
Matched sample
Novel innovation
1
Manufacturing full sample
4
Novel innovation
7
10
11
12
co with other
organisations
4.514**
1.687
12.92***
(1.820)
(1.831)
(1.351)
13.37***
(1.367)
1.641
0.650
9.834***
7.353***
(2.128)
(2.392)
(1.685)
(1.803)
co with domestic
universities
4.754**
1.326
9.613***
6.727***
(2.222)
(2.337)
(1.733)
(1.850)
co with universities in
Europe
4.584
6.345
0.993
4.638
(4.061)
(6.681)
(5.001)
(5.559)
co with universities in
other countries
lrdin
lrdex
4.08
3.866
6.479
2.601
(5.586)
(8.057)
(5.432)
(6.643)
1.238*** 1.060*** 1.036*** 0.138
0.116
0.13
3.130*** 3.121***
3.152***
2.691***
2.814***
(0.275)
(0.275)
(0.2770
(0.214)
(0.222)
(0.222)
(0.202)
(0.207)
(0.207)
0.864*** 0.873***
0.179
0.13
0.880***
0.904***
0.711***
0.709***
(0.32)
(0.298)
(0.293)
(0.256)
(0.258)
(0.268)
(0.268)
(0.284)
(0.281)
(0.318)
(0.261)
2.827***
0.511
0.838
1.346
4.283**
4.272**
5.184*** 3.525** 3.807*** 3.804*** 2.550*
2.419
2.411
(1.859)
(1.884)
(1.917)
(1.76)
(1.821)
(1.849)
(1.411)
(1.457)
(1.46)
(1.484)
(1.539)
(1.539)
age
3.535
3.057
2.924
13.21*** 13.29*** 13.01*** 1.078
0.893
1.005
6.524*** 6.290*** 6.391***
(3.136)
(3.172)
(3.105)
(4.193)
(4.186)
(4.003)
(1.919)
(1.92)
lack_hc1
0.0986
0.219
0.19
1.295
1.249
2.222
4.551*** 4.298***
(1.737)
(1.734)
(1.725)
(1.685)
(1.689)
(1.676)
(1.139)
(1.14)
(1.143)
(1.175)
(1.18)
(1.185)
Industry dummies
yes
yes
Yes
yes
yes
yes
yes
yes
yes
yes
yes
yes
Constant
yes
yes
Yes
yes
yes
yes
yes
yes
yes
yes
yes
yes
Observations
793
793
793
793
793
793
4,296
4,297
4,297
4,296
4,297
4,297
Size4
(1.916)
(2.247)
(2.248)
(2.246)
4.383***
5.216***
5.143***
5.145***
4.665
4.38
2.157
2.242
1.816
1.699
17.35
16.35
15.23
19.03
16.45
15.21
Log Likelihood
(2146
(2144
2132
2741
2741
2725
5886
5907
5908
7619
7652
7652
230
distribution. Co-operation with other rms and institutions in general

has a positive and signicant effect on novel innovation in the matched
sample. Its impact is positive but insignicant for diffusionary innovation. This pattern is somewhat different from that in the matched
Chinese sample where the effect of collaboration is stronger for diffusionary innovation. In line with the results in China, the estimated
coefcients of the university variable are positive and insignicant.
Breaking down the universities by their country of origin, collaboration
with domestic British universities shows a signicant and positive effect
on novel innovations reported in British enterprises, whereas its effect is
insignicant for diffusionary innovations. This is the opposite of the
pattern we observed from the matched Chinese sample. In contrast to
the results from China, international collaboration with universities,
both in Europe and in other countries, does not appear to have any
signicant effect on either novel or diffusionary innovation in the British
rms. Consistent with the case in China, extramural R&D contributes
signicantly to novel innovation only. The effects of intramural R&D
expenditure also contribute signicantly to novel innovation, which is
in contrast to the nding that rm-level R&D in Chinese rms is used
mainly for assimilation and development and hence contributes less
signicantly to diffusion innovation. Larger and older rms also appear
to have produced less diffusionary innovation. Constraints in R&D
personnel do not appear to have any signicant effect.
However, in the case of the UK manufacturing sector as a whole, the
role of collaboration, especially collaboration with universities generally and with domestic British universities in particular, demonstrates a
positive and signicant effect on both novel and diffusionary innovations. The contribution of universities is stronger regarding novel innovations than for diffusionary innovations created in British rms.
Consistent with the results obtained from the matched UK sample,
international innovation collaboration with foreign universities does
not show any signicant effect for either novel or diffusionary innovations, which raises an interesting question for future research as to why
such international collaborations are set up and why they are not
effective.
In summary, the results from Table 8.6 suggest that the role of
universities in rm innovation in the UK is consistent with the current
received wisdom that they contribute to both novel and diffusionary
innovation, but their contribution to the creation of innovations that
231
are groundbreaking at the world level is greater than that in relation to

diffusionary innovation. However, the role of universities in the emerging economies is different. Collaboration between domestic universities
and industries contributes signicantly to diffusionary activities in
China but less so to innovation that is groundbreaking in world
terms. On the other hand, international innovation collaboration with
foreign universities, especially those in the NIEs, contributes signicantly to the creation of innovation that is novel at the world level in
China.
8.6 Conclusions
This chapter attempts to investigate the role of universities in industrial
innovation in emerging economies using a rm-level innovation survey
database from China. It then benchmarks the Chinese pattern with that
from the UK, a classical industrialised economy from which a signicant
amount of the received wisdom on the role of universities and their role
in innovation has developed. One of the key ndings of this chapter is
that collaboration with domestic universities has played a signicant
role in the promotion of the diffusion of frontier technology and the
creation of innovation outcomes that are groundbreaking at country or
rm levels. In contrast to the traditional view that collaboration with
universities will lead to greater innovation, a view that is supported by
our evidence from the UK, the contribution of domestic universities to
the creation of innovation that is groundbreaking in world terms is
limited in China.
International innovation collaboration with foreign universities
appears to be fruitful in enhancing Chinese rms capabilities in the
creation of innovation outcomes that are groundbreaking in world
terms. In particular, innovation collaboration with universities in the
NIEs in East Asia, such as Hong Kong, Taiwan, Singapore and Korea,
and in countries other than the Western industrialised economies, for
example, Russia, Brazil, India and Australia, has proven to be benecial
and effective in promoting novel innovation in Chinese rms. In comparison, collaboration with universities in the Western industrialised
economies (the most frequently used innovation partners among foreign universities) does not appear to be as effective and fruitful as
expected. The overall pattern of the effectiveness of international university innovation collaboration attests to the argument regarding the
232
importance of technological and cultural gaps and hence the importance of the appropriateness of technology for effective international
technology transfer. This is consistent with the theory of directly technical change (Acemoglu, 2002) and the ndings from Fu and Gong
(2010) that the transfer of Western technology is not effective in promoting indigenous technological capabilities in many Chinese industrial sectors. In contrast, knowledge from universities in the NIEs and
the emerging South appears to be more compatible to the Chinese rms.
In summary, international universities from compatible economic and
technology backgrounds have played a role as a global source of knowledge, contributing to the creation of innovation that is groundbreaking
at the world level in the emerging economies. However, this is not the
case in the UK, where collaborative linkage with domestic universities is
the only form that contributes signicantly to industrial innovations in
British rms. Nevertheless, future research should investigate in depth,
in both China and the UK, the reasons why international innovation
collaboration with universities in the most advanced economies functions ineffectively in nurturing novel industrial innovations.
Findings from this research also indicate that when we focus on
innovative rms and control for size and industry, universities appear
to be the most popular innovation partner for Chinese rms. There is a
possible language issue in the denition and interpretation of he zuo
(collaboration) and the concept in English of co-operation that may
prevent us from a direct comparison of the levels of collaboration/cooperation between countries. Language issues notwithstanding, high
percentage of Chinese rms that report having collaborated with universities in innovation activities suggests that the strong government policy
push and the marketisation reform of the science and technology system
have effectively promoted a strong university-industry linkage in China.
Findings from the current research have important policy and practical implications for rms in both emerging and developing countries
with regard to the processes involved in tapping into knowledge and
resources from universities to promote innovation. Domestic universities are best positioned to help rms in developing countries assimilate, grasp, adapt and decipher transferred foreign technology. First,
given the importance of technology transfer in developing countries,
especially in the early stage of industrialisation, policies in the developing countries should greatly promote university-industry collaboration
as a means of enhancing the absorptive capacity of the indigenous
233
economy. Second, geographical, technological and cultural proximity

have led to a closer relationship between Chinese rms and the NIEs. In
these cases, the synergy, compatibility, level of advanced technology
and technological gap between the two partners form a creative and
knowledge-enriching basis to create innovative ideas, products and
processes. Moreover, collaboration with universities from the emerging
South also appears to make a robust positive contribution to the production of novel innovations. In recent years, Chinese rms have
increasingly established processes of international innovation collaboration with foreign universities following the increasing internationalisation of Chinese rms. For example, Huawei extended investment on
technological alliances to a number of foreign universities such as
Stanford (US), University of California, Berkeley (US), Imperial College
London (UK), University of Surrey (UK), and Poznan University of
Technology (Poland) (Fu, 2014). Such efforts also proved to be
rewarding. Moreover, Chinas technology leadership in the solar PV
industry is also obtained through collaboration with an Australian
university. Therefore, rms in the developing countries seeking international collaboration should not constrain themselves by considering
only universities in the Western countries such as the United States,
Europe and Japan: universities in the NIEs and the emerging South will
provide a more compatible and effective innovation partner in the
creation of novel innovation outcomes.
Appendix 8.1 Correlation coefcient in the Chinese sample

China
newsal newsaln newsald lrdin
newsal
lrdex
co
cogd
cosd
cocd
copd
coprid counid couni1 couni2 couni3 couni4
newsaln 0.2644 1
newsald 0.877 0.2169
lrdin
lrdex
0.2833
0.1475
0.1081
0.0611
0.2292 1
0.1115 0.4035
co
cogd
0.1839
0.1504
0.0558
0.0199
0.1618 0.3681
0.1461 0.2651
0.3194 1
0.2141 0.5965 1
cosd
0.1207
0.0409
0.1099 0.2364
0.222
cocd
copd
coprid
0.1444
0.1089
0.1145
0.027
0.035
0.0572
0.1409 0.2093
0.1041 0.2228
0.0834 0.2041
0.1984 0.6033 0.5267 0.6705 1

0.2055 0.4852 0.5151 0.635 0.6112 1
0.2645 0.433 0.4592 0.5493 0.5227 0.557
counid
0.1616
0.0407
0.1352 0.4018
0.3737 0.7315 0.4151 0.4584 0.4669 0.4329 0.4476 1
couni1
couni2
0.1643
0.0388
0.0325 0.1421 0.397

0.0763 0.0005 0.0386
couni3
couni4
0.0389
0.0655
0.0895 0.0074 0.0922 0.1019 0.1037 0.1549 0.1292 0.1313 0.1329 0.0904 0.1417 0.0882 0.2474 1
0.1267 0.0011 0.0725 0.0419 0.0595 0.0671 0.0633 0.0645 0.053 0.0641 0.0813 0.0506 0.1419 0.2229 1
0.6167 0.5988 1
0.3645 0.7224 0.4047 0.4537 0.462 0.4302 0.4319 0.9876 1

0.0817 0.08
0.0855 0.0804 0.0819 0.1389 0.13
0.1094 0.0633 1
Appendix 8.2 Correlation coefcient in the matched UK sample

UK
newsal newsaln newsald lrdin
newsal
newsaln
newsald
0.4687 1
0.5989 0.0564 1
lrdin
lrdex
0.1323
0.1024
0.1699 0.0552 1
0.146 0.0347 0.3939 1
co
0.0492
0.075
cogd
cosd
cocd
0.0573
0.0719
0.054
0.0433 0.0407 0.2423 0.2978 0.7147 1

0.0772 0.0384 0.2217 0.2766 0.8147 0.6723 1
0.0632 0.0259 0.2066 0.2658 0.8344 0.7233 0.7433 1
copd
0.0428
coprid
0.0673
counid_2 0.0414
0.014
0.0277 0.033
0.252
Lrdex co
cogd
cosd
cocd
copd
coprid counid_2 couni1_2 couni2_2 couni3_2
0.3145 1
0.1275 0.2364 0.5465 0.5731 0.5505 0.5805 1
0.0155 0.0675 0.2518 0.3782 0.6169 0.5604 0.5973 0.5947 0.5391 1

0.0682 0.009 0.2656 0.3031 0.6806 0.5198 0.5923 0.5879 0.4589 0.6336 1
couni1_2 0.0518 0.0774 0.0156 0.2564 0.295 0.6635 0.4968 0.5879 0.5708 0.4581 0.6223 0.9749
couni2_2 0.0109 0.0183 0.0226 0.1026 0.1812 0.2163 0.2503 0.2326 0.2267 0.1909 0.2938 0.3178
1
0.2352
couni3_2 0.0128
0.01
0.1955
0.4511
counid_2 0.1982
couni1_2 0.1918
UK manufacturing full sample

0.1631 0.1585 0.3204 0.2874 0.675 0.5251 0.5969 0.6049 0.5036 0.634 1
0.1564 0.1541 0.3119 0.2862 0.6609 0.5079 0.5857 0.5891 0.4943 0.6177 0.9792
couni2_2 0.0755
0.0575 0.0626 0.1417 0.149
0.2367
couni3_2 0.0708
0.0539 0.0588 0.1436 0.1438 0.1485 0.1599 0.158
0.1922
0.3489
0.0089 0.158
0.1658 0.1616 0.18
0.1766 0.1721 0.1602 0.2369 0.2374
0.2196 0.2215 0.2237 0.2203 0.2142 0.2611 0.3254

0.16
0.153
0.2052 0.22
Technological learning, tacit

knowledge acquisition and industrial
upgrading: The Chinese optical bre
and cable industry
9.1 Introduction
Although denitions of technological progress and technological learning may differ, it is clear that technological learning broadly, the process
of accumulation of technological capability to improve competitiveness
can have signicant effects on prots and growth (Lee and Lim, 2001;
Geroski and Mazzucato, 2002), productivity (Keesing and Lall, 1992),
exports (Zahra, Ireland and Hitt, 2000; Ernst, 2004) and innovation
(Leonard-Barton, 1995, Zahra, 1996, Padilla and Tunzelman, 2008).
The aim of this chapter is to contribute to our understanding of how
tacit knowledge the knowledge embodied in the experience and knowhow of workers and researchers can be acquired through various
internal and external knowledge sources as part of the technological
learning process.
Technological learning plays a particularly important role in industrialisation: it is vital to understand the processes by which less developed countries can acquire the necessary technologies from developed
countries, manage the learning process, move up the value chain and
ultimately narrow the currently wide gaps in technology and living
standards. Empirically, the experience of the newly industrialised
economies (NIEs) of South Korea, Taiwan, Singapore and Hong Kong
provides the best examples of the impact of technological learning in
practice. The latecomer studies of Pack and Westphal (1986), Lall
(1987) and Amsden (1989), as well as the industry-specic studies of
the NIEs of Hobday (1995), Kim (1997) and Lee and Lim (2001),
clearly demonstrate the impact of technological improvements that
transformed poor and technologically backwards countries into modern, afuent and world-leading economies within the space of a
236
Technological learning and industrial upgrading
237
generation. As Freeman (1994) states, the study of technological learning is an important building block in economic catch-up, which brings
unique transformative value both to the economy and human society.
The literature has thus identied an indisputable and essential role for
technological learning in development. However, the existing literature
on technological learning so far has tended to treat knowledge as
homogenous and has not distinguished knowledge of different natures.
The denition of knowledge should include tacit knowledge and
explicit, codied knowledge. Expanding the denition in this way raises
interesting research questions. For example, are the sources of tacit
knowledge the same as those of codied knowledge? Which sources
are more effective for tacit knowledge learning than for codied
knowledge?
As there has been no systematic research on these questions to date,
this chapter is an attempt to study the effective sources of tacit knowledge to understand how such knowledge can be acquired and promoted. To date, this concept has been recognised to be an important
element of the internal processes involved in the acquisition of technological knowledge, capabilities and competitive advantage, but there is
currently a lack of empirical data and reliable methodologies relating to
the investigation of this elusive but essential component of technical
learning.1 This chapter is an attempt to improve our understanding in
these areas.
The aim of this chapter is to examine the sources of the acquisition of
the tacit form of knowledge through a case study of the Chinese optical
bre and cable industry. The chapter uses an econometric estimation
process to determine the impact, if any, of tacit forms of knowledge in
the technological learning process via a rm-level survey of 95 optical
bre and cable manufacturers located in China. This Chinese optical
bre and cable industry provides a useful case study for our purposes
because the batch process structure of the supply chain separates the
industry as a whole into three sub-industries that involve different
balances between tacit and explicit knowledge. This allows us to test
which of the potential forms of knowledge is more signicant in the case
of tacit knowledgeintensive sub-industries, and vice versa.
The chapter is organised as follows: Section 9.2 discusses the literature and the theoretical framework. Section 9.3 provides a brief
1
See Kessler, Bierly and Gopalakrishnan (2000) and discussion in Section 9.2.
238
overview of the Chinese optical bre and cable industry. Section 9.4
discusses data and methodology. Section 9.5 presents the results, and
Section 9.6 provides conclusions.
9.2 External and internal sources of tacit knowledge

The precise meaning of technological learning, like many denitions, is
subject to dispute and complications. We can distinguish between at
least two senses of the term. In the rst, technological learning is a
description of the trajectory or path that rms or an economy takes to
accumulate technological capability. These paths can change over time,
becoming shorter or longer as technological capability is accumulated
at different rates. This macroeconomic denition adopts a high-level
perspective of the learning process. Alternatively, technological learning
can be studied in a much more microeconomic sense whereby attention
is paid to the processes themselves though which individuals acquire
technical knowledge that rms convert into technological capability. In
this sense, technological learning is a process itself. This chapter contributes to our understanding of the ways in which tacit knowledge can
be acquired. We examine various possible channels of tacit knowledge
and, based on the characteristics of tacit knowledge and the relatively
sparse literature on this concept, we formulate four hypotheses regarding the relative importance of the different channels to the specic case
of the optical bre and cable industry. The hypotheses are presented at
the end of this section and are based on the denitional considerations
that we next review.
Within the process-related meaning of the term, certain distinctions
can be made. Some authors dene technological learning in broad
terms,2 whereas others suggest that it involves more specic processes.3
A rich literature has developed regarding the internal processes and
linkages involved in the technological learning process. Kessler, Bierly
and Shanthi (2000) examine the various merits of internal versus
2
Bell and Pavitt (1992) dene technological learning (or technological

accumulation) as any process by which the resources for generating and managing
technical changes (technological capabilities) are increased or strengthened.
A common theme within the works of Amsden (1989), Hobday (1995) and Xie
(2004) is that technological learning requires the acquisition (copying or
borrowing) of foreign technologies, possibly followed by adaptation and
improvement.
239
external knowledge acquisition and development, nding that internal

forms offer advantages in terms of speed of innovation and lasting
competitive advantage. Figueiredo (2003) shows how the key features
underlying learning processes can inuence inter-rm differences in their
accumulation of technological capabilities and improve operational performance in a latecomer context.4 Lokshin, Belderbos and Carree (2007)
show that internal and external knowledge forms must be combined in an
optimum manner that does not involve the maximisation of either type of
approach on its own: in other words, there must be sufcient internal
absorptive capacity and in-house R&D must be of the correct standard
and type if external technical knowledge is to be used optimally.5
Similarly, Cassiman and Veugelers (2006) call for further research on
complementarities and internal research processes after nding that highquality internal R&D processes are required to complement external
knowledge.6 A heavy emphasis on examination of internal processes
brings the investigation of effective management techniques (e.g., Chen
and Qu, 2003) within the scope of research on technological learning.
The extensive literature on technological learning reviewed here has
treated tacit and codied knowledge in a bundle. In fact, they are
different in the nature of knowledge each of them relates, the way that
such knowledge is generated and stored and the transferability of each.
Codied knowledge is knowledge that is articulated into formal language, including grammatical statements, mathematical expressions,
specications, manuals and so on. This articulation means that explicit
knowledge can be readily transmitted and shared with others. Codied
knowledge can also be given a physical form via publications, patents
and textbooks. In principle, and allowing for sufcient ability in the
learner and sufciently clear instruction, codied knowledge can be
transmitted at a distance in an impersonal manner. The speed with
which the recipient assimilates codied knowledge is more a function
Figueiredo concludes that rates of technological capability accumulation and

improvement in operational performance can be accelerated through the
knowledge processes within a company, through devotion of attention to the
variety, intensity, functioning and interaction of four identied processes: external
knowledge acquisition, internal knowledge acquisition, knowledge socialisation
and knowledge codication.
Based on a six-year panel data set covering 304 Dutch manufacturing rms from
1996 to 2001.
Based on a data set covering 1,335 Belgian manufacturing rms in 1993.
240
of the recipients intellectual capacity and less a matter of the time spent
gaining experience with the subject matter.
Tacit knowledge is personal knowledge embedded in individual
experience. It is a type of knowledge that is used by all people, but
people do not necessarily nd it easy to articulate. Most tacit knowledge
can never actually be articulated for much of it is not introspectable or
verbally articulable (relevant examples of the latter would include our
tacit knowledge of grammatical or logical rules, or even of most social
conventions) (Pylyshyn, 1981:603). Tacit knowledge is thus difcult to
codify or detach from individual researchers, employees or managers.
Therefore, it does not submit to easy transfer by written or other
verbalisation (Polanyi, 1967). The transfer of tacit knowledge requires
close, personal and ongoing interaction between instructor and learner
as well as strong efforts at assimilation. Although tacit knowledge is
difcult to dene, it exists and is important: the value to a rm of staff
with tacit knowledge (experience and know-how) is obvious to all
business leaders as a key determinant of a rms success. Tacit knowledge is harder to copy and is therefore likely to lead to more lasting
forms of competitive advantage than external, codied knowledge that
is bought in. Therefore, rms that have signicant tacit knowledge
resources have a competitive advantage over others because tacit
knowledge is less susceptible to copying or imitation (Kessler et al.,
2000). When applied to the research and technological learning process,
such forms of knowledge are correspondingly key to our understanding
of success in technological learning and catch-up.
Regardless of the abilities of the learner, tacit knowledge requires signicant time to acquire through experience and learning by doing. Once it
is acquired, a skilled worker intuitively knows the adjustment to be made
to a production process through a slight variance in materials, efforts or
technique in a way that an inexperienced new starter would not know and
would not be able to glean from an entirely rules-based approach.
Learning by being told and learning by observation are two key ways in
which to capture tacit knowledge from groups and individuals (Parsaye,
1989). The so-called soft technologies, such as the way things are done in
the rm, management practices and learning can be included within
characteristics of a tacit nature (Nelson and Winter, 1982).
In sum, tacit knowledge has a few distinctive characteristics. First, it
has strong embeddedness in that tacit knowledge is often embedded in
the person who owns it and takes a long time to acquire through
241
learning-by-doing. Second, tacit knowledge is difcult to transfer and

imperfectly imitable. These dening characteristics of tacit knowledge
suggest that it is difcult to transfer from one person (rm) to another
person (rm). Given the differences in the nature of the two types of
knowledge, the sources and mechanisms for learning and acquisition of
them are different. The acquisition of tacit knowledge requires strong
self-study, learning-by-doing and sustained interpersonal interaction
over time. Therefore, in-house R&D is a crucial source of tacit knowledge because of the strong learning function of R&D (Cohen and
Leventhal, 1989). The demand for sustained interpersonal interaction
in the transfer of tacit knowledge also suggests that imports of machineries and equipment are not an effective channel for the acquisition of
foreign tacit knowledge. Therefore, we have the following hypotheses:
H91: In industries where tacit knowledge is relatively more important, R&D activities are a more important source of tacit
knowledge.
H92: In industries where tacit knowledge is relatively more important, imports of foreign equipment/technology are a less important source of tacit knowledge.
Tacit technological knowledge, for example, experiences and knowhow, can have considerable industry specicity. Technological knowhow accumulated in peer rms can have great relevance to rms in that
industry. Such know-how can be diffused through movement of labour,
industry association conferences and meetings, social networks and
other formal or informal interpersonal interactions. Therefore, we
have the following hypothesis:
H93: In industries where tacit knowledge is relatively more important, peer rms in the same industry are a more important
source of tacit knowledge.
Compared to the other three sources (R&D, imports and linkages
with rms in the same industry), universities are likely to be sources of
both tacit and codied knowledge. Tacit knowledge can be gained via
collaborative research with universities and informal and formal consulting initiatives, whereas publications, patents, R&D project outputs
and licensing are important forms of codied university knowledge
(Etzkowitz and Leydesdorff, 1997; Hong, 2008; Eom and Lee, 2010).
242
Optical
Fibrecore
Raw
Materials
Profit
margin
Optical
Fibre
Optical
Cable
Make the
fibrecore
Draw the fibrecore

into fibre
Turn the fibre

into cable
7080%
1020%
1020%
Figure 9.1 The production chain in the optical bre and cable industry
In the process of transfer of tacit knowledge from universities, collaborative research and informal consultation play more of a role than
publications and patents (Hong, 2008). These insights suggest the
following hypotheses:
H94a: Universities are effective sources for the transfer of both codied and tacit knowledge.
H94b: The importance of universities is likely to be similar in industries in which tacit knowledge is relatively more important
and those in which it is not.
9.3 The Chinese optical bre and cable industry: an overview

The optical bre and cable industry is neither an assembly industry such
as automobiles nor a continuous process industry such as steel. Instead,
it is a batch process industry that consists of three independent continuous processes in the (1) brecore, (2) bre and (3) cable sub-industries
as shown in Figure 9.1. Although the Chinese optical bre and cable
industry originated in the late 1970s at the Wuhan Research Institute of
Post and Telecommunication, large-scale operations only began in
1984. From 1984 to 1986, seven optical bre manufacturers, all state
owned, including Beijing Glass Factory, Shanghai Quartz Glass Factory
and Shanxi Taiyuan Mine Wool Factory, introduced 34 forms of brecore manufacturing equipment and 17 draw-benches from British
Special Gas Control (SGC) at an average expense of US$20 million.
However, all seven pioneers left the market before 1991 mainly because
of the high cost of absorbing technical know-how and the lack of
complementary industries.
From 1991 onwards, the industry experienced large-scale growth and
expansion when technological learning became market driven and
243
started from the basis of the least technologically sophisticated subindustry of cable production rather than the most technology-intensive
brecore manufacturing sub-industry. Indigenous capabilities upgraded
gradually from cable manufacturing to bre manufacturing and then to
brecore manufacturing. Up to 2006, the market supply exceeded 20
million kilometres of core and total sales achieved were US$600 million,
meaning that the Chinese optical bre and cable industry underwent a
boom and became the second largest in the world.
The brecore is crucial to the performance and quality of both bre
and cable. The technical barrier and process requirements are high in
the manufacture of brecore. Initially, most indigenous SOEs attempted to adopt a relevant learning path involving digestion and absorption of the technology at the upstream brecore stage but were
unsuccessful. After this tremendous failure at the rst stage, most
indigenous rms, especially those private rms without government
support, started to operate at the downstream stage of the industry,
importing bre and producing cable from bre. The relatively low
technological requirements of this process enabled indigenous rms to
expand their production capacity and achieve economies of scale. At
the same time, some rms also attempted to set up joint ventures with
foreign companies to promote in-depth technological co-operation.
From 2000 on, market opportunities, technological learning and various collaborations allowed indigenous enterprises to engage successfully in brecore production.
9.4 Data and methodology

Data
This chapter uses rm-level survey data relating to all rms in the
optical bre and cable industries in China. Questionnaires were sent
to the presidents, vice-presidents or chief engineers of each enterprise
asking them to provide self-assessments on a list of questions regarding
their sources and experiences of technological learning. The answers to
these questions were scored using a Likert scale with labels of strongly
disagree assigned a numeric value of 1 and strongly agree assigned a
numeric value of 7.
In total, 103 questionnaires were returned representing a response
rate of 95 per cent. After data cleansing, the nal data set included
244
Table 9.1 Perception of the technological characteristics of the optical bre

and cable industry
Statements
Mean Std. Dev.
Tacit knowledge (operating techniques and know-how) in

brecore manufacturing is signicant
bre manufacturing is signicant
cable manufacturing is signicant
5.81
1.075
5.54
1.114
4.49
1.015
95 rms representing 87 per cent of the population. Most of the

respondents agreed that the production of cable helps a rm master
bre technology and that the production of bre has similar benets in
terms of the brecore technology. The scores on intensity of tacit knowledge show that it rises consistently as we move upstream in the production chain of the optical bre industry; that is, tacit knowledge is most
intensive in the bre core industry but relatively less signicant as we
move to consider production of bre and cable, respectively (Table 9.1).
Table 9.2 summarises the denition and the descriptive statistics of the
variables. The survey covers the whole population of rms in this
industry; therefore, we can expect the results to be consistent and
representative. However, as the data capture results at a single point
in time, the study does not provide any analysis of causal relationships.
Methodology
We test the effectiveness of different sources of tacit knowledge by
testing the importance of these sources in rms competitiveness in
industries of low and high intensity of tacit knowledge. The literature
has found a robust linkage between technological learning and rm
performance in the areas of prots and growth (Lee and Lim, 2001;
Geroski and Mazzucato, 2002), productivity (Keesing and Lall, 1992),
exports (Zahra et al., 2000; Ernst, 2004) and innovation (Zahra, 1996,
Padilla and Tunzelman, 2008). Therefore, we estimate the following
regression equation using an Ordered Logit model:
245
Table 9.2 Denition of variables

Variables Denition
comp
tacit1
tacit2
tacit3
tacit4
codied1
codied2
codied3
codied4
lnlab
lnass
hc
intechl
rd2
te
Competitiveness of rms proxied by rms

market leadership
Tacit knowledge: foreign
Tacit knowledge: other rms in the industry
Tacit knowledge: universities
Tacit knowledge: in-house R&D
Explicit knowledge: foreign
Explicit knowledge: other rms in the industry
Explicit knowledge: universities
Explicit knowledge: in-house R&D
Ln (no of employees)
Ln (total assets)
Level of human capital
Initial level of technology
Relative R&D strength score
Industry dummy, 1 = rms producing brecore,
0 = others
Mean Std. Dev.

2.5
0.9
3.9
1.1
1.2
3.7
3.8
1.2
1.2
3.8
6.2
8.3
0.3
2.8
2.7
0.19
2.1
1.4
1.2
1.4
2.1
1.4
1.3
1.4
0.8
1.5
0.2
1.5
1.1
0.4
ex
Ci B0 B1 TLin
i B2 TLi B3 Xi ui
9:1
where C represents the competitiveness of a rm, TLin represents sources of internal technological learning, TLex is a vector of sources of
external technological learning, and X is a vector of rm-specic control
variables.
The dependent variable, competitiveness, is a polytomous response
variable that is based on the market leadership of each rm. It is
measured by rms self-assessment of market leadership, ranked on a
scale of ve values ranging from minimum to maximum levels of market
leadership in each sub-industry. For rms producing in more than one
sub-industry, we take the average of the rank scores. Admittedly, rms
competitiveness revealed in their market leadership is not the only outcome of technological learning. Competitiveness can also be an outcome of other factors. Therefore, it is important to control for other
important rm characteristics. Although we have tried to control for
other factors affecting competitiveness, as a result of data availability,
246
the test may still have an omitted variable issue, and we should therefore
be cautious when drawing conclusions.
The independent variable TLin
i measures technological learning from
internal sources, composed of knowledge acquisition from internal
sources such as in-house R&D. Similarly the independent variable
TLex
i measures technological learning from external sources, composed
of knowledge acquisition from (1) foreign sources through imports and
licensing, (2) other rms in the industry and (3) universities. The equation also includes a vector of control variables represented by Xi, which
includes a rms size (measured as the log of asset holding), the rms
age (measured in years), ownership dummies (which includes the following categories: state-owned enterprises, shareholding, joint ventures, foreign-owned enterprises, collectively owned enterprises,
privately owned enterprises and others), level of human capital (measured as the percentage of technicians in total employees) and dummy
variables for sub-industry type.
In this chapter, we found an Ordered Logit model to be appropriate for
the analysis as the dependent variable is the self-assessment response of
market leadership of each rm, restricted to ordered values of 1 to 5.
Ordered Logit models are used to model relationships between a polytomous response variable and a set of repressor variables. These polytomous response models can be classied into two distinct types, depending
on whether the response variable has an ordered or unordered structure.
The measurement of learning and knowledge acquisition is not
straightforward. In this research, we use rms assessment of the importance of various sources for technological learning to proxy the importance of different knowledge sources. A total of eight self-assessment
factors regarding the importance of different sources of technology
were used, relating to the codied and tacit knowledge acquired from
foreign sources, other rms in the industry, universities and in-house
R&D. It is worth mentioning that the independent variables exhibit a
high degree of multicollinearity, and this may result in an imprecise point
estimate. Following Adler and Golany (2001), we use factor analysis to
summarise the variations in the eight variables into a few principal
components. This approach explains the variance structure of a matrix
of data through linear combinations of variables that capture a large
proportion of the variance in the data but, at the same time, reduce the
data to a few principal components. If most of the population variance
can be attributed to the rst few components, they can replace the original
247
Table 9.3 Factor analysis
Variable
Tacit knowledge:
foreign
Tacit knowledge:
other rms in
the industry
Tacit knowledge:
universities
Tacit knowledge:
in-house R&D
Codied
knowledge:
foreign
Codied
knowledge:
other rms in
the industry
Codied
knowledge:
universities
Codied
knowledge:
in-house R&D
Factor 1: acquire from Factor 2: Factor 3: acquire from

other domestic rms in in-house universities and
the industry
R&D
research institutions
0.647
0.538
0.520
0.980
0.039
0.073
0.062
0.111
0.962
0.047
0.947
0.140
0.715
0.437
0.510
0.956
0.088
0.058
.201
0.183
0.868
.047
0.922
0.111
Note: Figures highlighted in bold are the highest values obtained in the estimation and
therefore the main variables captured by each factor.
variables without much loss of information. The factor analysis results

are summarised in Table 9.3. We found that 96.3 per cent of the variation
in the eight tangible and intangible knowledge acquisition variables is
explained by three factors: (1) knowledge acquired from other domestic
rms in the same industry, (2) knowledge acquired via in-house R&D
and (3) knowledge acquired from universities and research institutions.
The factor analysis results indicate a consistent pattern in one aspect:
rms that use any of the domestic knowledge sources, either internal or
external, do not appear to use direct foreign knowledge sources. In other
248
words, rms do not appear to use foreign and domestic knowledge

sources in a complementary way.
Given the nature of tacit knowledge, direct measurement is challenging. In this chapter, we therefore use indirect measurement of tacit
knowledge. As discussed earlier, the three sub-industries in the optical
bre and cable industry represent three sectors with increasing intensity
of tacit knowledge. The brecore sector is the most intensive in tacit
knowledge and the cable sector is the least intensive in using tacit
knowledge. This is proven by the basic survey evidence. Moreover,
the rms in the sample are divided into two blocks: about 70 rms
produce cable and/or bre but do not have the technology to produce
brecore and more than 20 rms produce brecore with some also
producing bre.7 Therefore, the rms can be divided as follows: the
rst group of rms operates in the low tacit knowledgeintensive cable/
bre sector, and the second group has mastered the tacit knowledge
intensive bercore production technology. Therefore, we use a dummy
variable that equals 1 for rms in the more tacit knowledgeintensive
sector, and 0 for rms in the less tacit knowledgeintensive sector. We
use interaction terms to examine the effectiveness of different knowledge sources of tacit and codied knowledge.
9.5 Results
Table 9.4 reports the Ordered Logit estimation of the model. Column
(1) reports the estimated results of the full model where the three
knowledge source factors are entered into the model together.
Columns (2) to (4) report the results of the regressions where the three
factors are entered into the regression separately. The point estimate on
the university factor is found to be positive and statistically signicant in
the regression that has a full set of controls (column [1]) as well as in the
regression where the university factor is estimated separately (reported
in column [4]). This result suggests that rms that have universities as an
important knowledge source are more competitive in the market. To
note, given the cross-sectional nature of the data, claims regarding any
causal relationship would not be warranted. Knowledge sources from
other rms in the industry do not appear to have a signicant effect on
the competitiveness of rms in general. The estimated coefcient of
7
Three of them also produce cable.
249
Table 9.4 Knowledge sources and technological learning

Dependent variable: competitiveness
Factor 1 (other rms
in the industry)
Factor 2 (in-house R&D)
Factor 3 (universities)
rd2
intechl
lnass
hc
Industry dummy
ownership dummy
Constant
Log likelihood
Chi2
0.003
0.008
(0.013)
0.007
(0.013)
0.0214*
(0.012)
0.794***
(0.071)
0.002
(0.012)
0.001
(0.013)
0.099
(0.061)
y
y
y
105.7
56.53
(0.012)
0.009
(0.014)
0.791***
(0.069)
0.008
(0.011)
0.002
(0.014)
0.112*
(0.065)
y
y
y
109
53.23
0.795***
(0.069)
0.007
(0.011)
0.001
(0.013)
0.106*
(0.06)
y
y
y
108.8
53.46
0.0227**
(0.011)
0.787***
(0.07)
0.001
(0.011)
0
(0.012)
0.106
(0.065)
y
y
y
105.9
53.18
Note: 1. Robust standard errors in parentheses;

in-house R&D bears the expected positive sign but is not statistically
signicant. However, the relative R&D strength variable that indicates
a rms R&D activity relative to the rest of the rms in the industry
shows a positive effect on rms competitiveness and is statistically
signicant at the 1 per cent level. The size of the estimated coefcient
is much larger than the rest of the control variables, indicating an
important contribution of R&D to a rms competitiveness. This variable is likely to correlate with the internal knowledge source variable
and explain the insignicance of the internal knowledge source variable.
As regards the rest of the control variables, human capital has a positive
and signicant effect on competitiveness as expected. Firm size and the
initial technology level variable do not appear to have a signicant effect
on a rms competitiveness.
250
To test whether or not the embeddedness of tacit knowledge in the

industry matters, we estimate the Ordered Logit model using the three
factors as explanatory variables and also interact the factors with an
industry dummy (dummy = 1 if tacit knowledge is perceived to be
important for the industry and 0 otherwise). The results are reported
in Table 9.5. The regression result with a full set of controls (column [1])
nds that the point estimate on the interaction term on the industry
peers factor is 0.10 and is statistically signicant. In industries where
tacit knowledge is crucial, industry peers appear to be an important
source of knowledge accumulation. The point estimates on the other
interaction term and the factors are not found to be statistically signicant. We also estimate the model for each factor and the corresponding
interaction term separately (reported in columns [2] to [4]). The point
estimate on the interaction term on the industry peer factor declines
marginally to 0.09 but continues to be statistically signicant (column [2]). Similarly in column (3), the point estimate on the interaction
term on the in-house R&D factor is positive (0.06) and statistically
signicant. On the other hand, results in column (4) show that while the
point estimate university factor is positive (0.03) and statistically signicant, the interaction term on the university factor is not statistically
signicantly different from zero. This may indicate that universities are
an important actor in tacit knowledge accumulation irrespective of
which industry the rm belongs to. These ndings also support our
initial hypothesis on the relative effectiveness of different tacit knowledge acquisition channels.
Although summarising the knowledge sources into principal factors
offers useful advantages, analysis of the effect of the individual original
knowledge source indicators also provides rich in-depth information
about the effect of each source of the different types of knowledge.
Table 9.6 reports the estimated results of the regressions using the
original scores of individual tacit knowledge sources. We interact the
four tacit knowledge acquisition variables with an industry dummy that
equals 1 if tacit knowledge is perceived to be important for the industry,
and 0 otherwise. To avoid the problem of multicollinearity, which may
give rise to imprecise point estimates, we estimate the model for each
tacit knowledge variable and the corresponding interaction term separately (reported in columns [1] to [4]).
The regression results in columns (2) and (4) report statistically
signicant and positive point estimate on the interaction term on tacit
251
Table 9.5 Does the embeddedness of tacit knowledge in the industry

matter?
(1)
Factor 1 (other rms in the
industry)
Factor 2 (in-house R&D)
Factor 3 (universities)
te*fac1
te*fac2
te*fac3
te (industry dummy)
intechl
lnass
hc
rd2
ownership dummy
Constant
Log likelihood
Chi2
(2)
(3)
(4)

0.0104 0.0114
(0.0119)
0.0016
(0.0159)
0.0235
(0.0147)
0.101***
(0.0365)
0.0574
(0.0346)
0.02
(0.0304)
0.0803**
(0.0332)
0.01
(0.0149)
0.0027
(0.0141)
0.0488
(0.0608)
0.757***
(0.0819)
y
y
73.7
104
(0.0108)
0.0045
(0.0153)
0.0258**
(0.0126)
0.0909***
(0.0325)
0.0595*
(0.0355)
0.0396
(0.0359)
0.0058
(0.0121)
0.0020
(0.0134)
0.110*
(0.0607)
0.764***
(0.0702)
y
y
52.92
108.5
0.0107
(0.038)
0.0267
0.0138
(0.0385) (0.0448)
0.013
0.0036
(0.0133) (0.0115)
0.0006
0.0008
(0.0127) (0.012)
0.0604* 0.105
(0.0636) (0.066)
0.774*** 0.791***
(0.0729) (0.0735)
y
y
y
y
65.15
53.76
107.5
105.7
Note: Robust standard errors in parentheses; *signicant at 10 per cent; **signicant

at 5 per cent; ***signicant at 1 per cent.
knowledge from industry peer sources and the in-house R&D source
(0.057 and 0.056, respectively) indicating positive association. This
suggests that in industries where domestic knowledge is important,
domestic peers and R&D activities are important sources of
252
Table 9.6 Does the embeddedness of tacit knowledge in the industry

matter?
(1)
tacit1
te*tacit1
tacit2
te*tacit2
tacit3
te*tacit3
(2)
(3)

0.0020
(0.0077)
0.0314**
(0.0146)
0.0066
(0.0081)
0.0573*
(0.0302)
0.0087
(0.0094)
0.0096
(0.0342)
tacit4
te*tacit4
rd2
hc
0.766***
(0.0761)
0.168**
(0.0719)
0.0052
(0.0121)
0.0034
(0.0132)
0.0886
ownership dummy
Constant
Log likelihood
Chi2
y
y
69.87
108.7
te
intechl
lnass
(4)
0.755***
(0.0686)
0.0358
(0.0472)
0.0044
(0.0127)
0.0015
(0.0137)
0.125*
(0.0645)
y
y
53.46
109
0.788**
(0.0712)
0.012
(0.0442)
0.0055
(0.0122)
0.0011
(0.0123)
0.113*
(0.0645)
y
y
54.21
107.9
0.0042
(0.0101)
0.0560*
(0.0306)
0.771***
(0.0736)
0.191*
(0.112)
0.0144
(0.014)
0.0023
(0.013)
0.0558
(0.0601)
y
y
64.6
107.4
Note: Robust standard errors in parentheses; *signicant at 10 per cent; **signicant

at 5 per cent; ***signicant at 1 per cent.
253
technological learning. On the other hand, regression results in column

(1) report a statistically signicant but negative point estimate on the
interaction term on tacit knowledge from foreign sources (0.03) indicating negative association. This nding supports our hypothesis that
foreign sources of technology are less likely to be effective learning
channels. The point estimate on the interaction term on tacit knowledge
from the university source is found to be statistically insignicant (column [3]). As hypothesised earlier, empirical estimates show that universities are effective in transferring both tacit and codied knowledge.
9.6 Conclusions
This chapter analyses the effectiveness of various sources of technological learning in transferring and assimilating tacit knowledge using rmlevel data from the Chinese optical bre and cable industry. Evidence
from the research suggests that in industries where tacit knowledge is a
more important component of technological learning, internal R&D
activities and domestic peers are important knowledge sources for
technological learning. Moreover, universities have been an important
asset in creating learning organisations and effective knowledge sources
for technological learning. They play the same important role in transferring both tacit and explicit knowledge. However, imports of equipment and licensing are a less effective learning channel in the acquisition
of tacit foreign technology.
Moreover, our results show that the difculty in acquisition of tacit
knowledge has been a signicant barrier to technological learning,
which may be the reason that motivates rms to start in the cable subindustry, the sub-industry that requires the least tacit knowledge. As
rms gain technological knowledge and expertise, both tacit and
explicit, from producing less sophisticated products, they move up the
technology ladder to produce more technologically advanced and
sophisticated products. In the case of China, rms are nally capable
of producing the brecore that has the highest tacit knowledge requirement. In other words, the requirements of tacit knowledge and the
process of technological learning and accumulation have had a signicant impact on the evolution and trajectory of the Chinese optical bre
and cable industry.
Results from the research also indicate that rms that use any of the
domestic knowledge sources, either internal or external, do not appear
254
to use direct foreign knowledge sources. In other words, rms do not

appear to use foreign and domestic knowledge sources in a complementary way. Note, however, that in this survey and hence this study,
foreign knowledge sources refer to foreign knowledge acquisition
through imports and licensing. Foreign joint ventures or foreign subsidiaries in China are regarded as domestic peers in the same industry.
These foreign invested rms provide better proximity and greater likelihood for the transfer of tacit knowledge (e.g., better managerial practice, technological know-how). Therefore, future research is needed
to explore fully the role of foreign direct investment in the learning
process, and caution should be exercised in offering any conclusion as
regards the role of foreign knowledge sources.
Although this chapter has brought to light some interesting and
important insights on the relative effectiveness of different knowledge
sources/channels in technological learning in the Chinese optical bre
and cable industry, there are two caveats in the data. First, future
research should focus on analysis with panel data. Second, such work
should include objective measures of variables that allow us to control
for rm heterogeneity and dynamics in the learning process.
Appendix 9.1 Correlation coefcients
comp
codied1
codied2
codied3
codied4
tacit1
tacit2
tacit3
tacit4
intechl
lnass
hc
brecore
e
comp
codied1 codied2 codied3 codied 4 Tacit 1 Tacit 2 Tacit 3 Tacit 4 intechl lnass
1
0.039
0.060
0.090
0.082
0.032
0.076
0.099
0.150
0.063
0.114
0.090
0.117
0.026
1
0.657
0.483
0.432
0.939
0.601
0.484
0.443
0.404
0.318
0.024
0.399
0.141
1
0.114
0.099
0.710
0.921
0.291
0.008
0.433
0.348
0.138
0.249
0.196
1
0.266
0.491
0.147
0.817
0.188
0.360
0.172
0.020
0.291
0.291
1
0.282
0.127
0.276
0.819
0.140
0.012
0.209
0.101
0.224
1
0.689
0.511
0.418
0.442
0.430
0.031
0.307
0.203
1
0.224
0.064
0.462
0.356
0.121
0.242
0.189
hc
brecore e
1
0.284 1
0.346 0.098 1
0.144 0.185 0.295 1
0.074 0.217 0.201 0.323 1
0.248 0.012 0.501 0.107 0.289 1
0.248 0.075 0.016 0.117 0.073 0.174
10
Leapfrogging in green technology:

The solar-PV industry in China
and India
10.1 Introduction
Building a global green economy will require technology transition in
both developed and developing countries. Among the developing countries, some emerging economies have quickly established signicant
technological capabilities in elds related to the green economy. The
growth of production capacity and the diffusion of green technology in
these emerging economies have been dramatic. China and India, in
particular, have become global leaders in some of the emergent green
technology sectors such as solar photovoltaic (PV) panels, wind turbines and electric and hybrid electric vehicles. Given the rapid eminence
of these green sectors, what are the respective contributions to technological progress from learning by doing; indigenous research and development; and technology transfer through FDI, trade and other
channels? What is the role of national innovation systems in promoting
more effective technology acquisition, adaptation and development?
What lessons can other developing countries learn from the development of green sectors in China and India? Drawing on an analysis and
comparison of technology progress strategy in the solar-PV industry in
China and India, and considering the different role of technology transfer and indigenous innovation as well as the national innovation system
in the development, diffusion, application and adaptation of green
technologies, this chapter addresses the key determinants of green
technological capability in emerging economies. The major contributions of this chapter are twofold. First, it illustrates that the two-leg
forward strategy, that is, adopting both technology transfer and indigenous innovation simultaneously, is suitable for the technological progress and development of green technology sectors in developing
countries. It provides the right mix and sequence of technological
progress that can be applied to green technology industries in developing countries, showing the importance of each mechanism during
256
Leapfrogging in green technology
257
different stages of development. Second, it nds that national environmental innovation systems are important in assisting sustainable
technological progress and leapfrogging development towards a competitive green economy.
The chapter is organised as follows: Section 10.2 reviews the mechanisms of technology progress in developing countries, including conventional and unconventional mechanisms of technology transfer and
indigenous innovation and catching up. Section 10.3 provides an overview of the science and technology sector and solar-PV industries in
China and India, using case studies of leading solar-PV companies.
Section 10.4 analyses the technology progress strategies that have
been adopted in both countries. Section 10.5 discusses the role of public
policies and institutions as well as private actors in the innovation
systems with special reference to the environmental innovation system.
Section 10.6 presents conclusions and outlines policy implications for
developing countries seeking to enhance their technological capabilities
for a green economy.
10.2 Technology transfer, indigenous R&D and technological

progress in emerging economies
Innovation is costly, risky and path dependent. This may provide a
rationale for poor countries to rely on foreign technology acquisition
for technological development. In fact, most innovation activities are
largely concentrated in a few developed countries. International technology diffusion is therefore an important condition for the economic
growth of developing countries. If foreign technologies are easy to
diffuse and adopt, a technologically backward country can catch up
rapidly through the acquisition and more rapid deployment of the most
advanced technologies (Grossman and Helpman, 1991, 1994; Romer,
1994; Eaton and Kortum, 1995).
Technology can be diffused between rms and across regions and
countries through various transmission mechanisms, including (1) licensing; (2) movement of goods through international trade; (3) movement
of capital through inward and outward foreign direct investment (FDI
and OFDI); (4) movement of people through migration, travel and
foreign education of students and workers; (5) international research
collaboration; (6) diffusion through media and the Internet of disembodied knowledge; and (7) integration of benets into global value
258
chains from foreign technology transferred within the supply chain

(Fu, Pietrobelli and Soete, 2011). Some knowledge is transferred intentionally from the knowledge owner to the recipient, but a large proportion of knowledge spillovers take place as unintended knowledge
leakage. In recent years, the mode of innovation is becoming more and
more open and is making good use of external resources. International
knowledge diffusion can therefore benet a countrys or rms innovation at every stage of the innovation process.
FDI and technology transfer

Inward FDI has long been regarded as a major vehicle for the transfer
of advanced technology to developing countries (Lall, 1992; Dunning,
1994). Multinational enterprises (MNEs) have internal incentives
encouraging the transfer of technology across borders to share technology between parent companies and subsidiaries (Markusen,
2002). Therefore, it is expected that in the medium to long run, local
rms will benet from MNEs via (1) technology transfer within the
foreign invested joint ventures through imported machinery and
equipment and the training of labour, (2) horizontal technology spillovers to other rms in the same industry/region through demonstration effects and labour turnover, (3) vertical technology spillovers
within the value chain through forwards and backwards linkages;
and (4) the competition effect of pushing inefcient rms to exit
from the market and force others to innovate to maintain a competitive edge (Caves, 1974; Fosfuri, Motta and Ronde, 2001; Javorcik,
2004). As one of the largest recipients of inward FDI in the world,
China has introduced a set of policies to enhance linkages and knowledge transfer from foreign to indigenous rms. However, FDI may
also have a detrimental impact on the technological upgrading of local
rms. First, FDI may make the competing domestic rms worse off
and even crowd them out from the market (Aitken and Harrison,
1999; Hu and Jefferson, 2002). Second, strong competition from
foreign subsidiaries may reduce local rms R&D efforts (OECD,
2002). Moreover, foreign subsidiaries may remain as enclaves in a
developing country with a lack of effective linkages with the local
economy. As a result, empirical evidence on the effect of inward FDI
on the productivity and innovation capabilities of indigenous rms is
mixed, for instance, the positive impact in Buckley, Clegg and Wang
259
(2002) and the depressive effects in Hu and Jefferson (2002) and Fu

and Gong (2011) using Chinese industry/rm-level data, as well as
the negative impact in Sasidharan and Kathuria (2008) using Indian
rm-level data.
Licensing and technology transfer

The second conventional method of technology transfer is licensing,
which is an important source of international technology transfer
for developing countries (Correa, 2005). Licensing is particularly
useful in terms of transferring patented and codied knowledge.
From the aspect of the licensee, licensing is less costly than in-house
R&D and less risky and shortens the time required to bring new
products to the market. For the licensor, licensing is less risky than
FDI and highly protable but may lead to the risk of technology
leakage in the host country through copying or movement of labour
(Darcy et al., 2009). Therefore, some MNEs may prefer FDI (avoiding
engagement in licensing) or may transfer only lagging technologies
(Maskus, 2000). In addition, a successful transfer also depends on the
capability of licensees to learn, develop and market the technology
(Hoekman, Maskus and Saggi, 2005; Kamiyama, Sheehan and
Martinez, 2006).
Imports and technology transfer

Imports of machinery and equipment are another important channel
for countries to acquire advanced technology and enhance competitiveness (Fagerberg, 1994; Coe and Helpman, 1995). However, using
such machines to produce high-quality products does not mean that
developing countries necessarily master the skills required to design
and produce advanced machinery. Substantial technological learning
and reverse engineering are required to grasp the technologies embedded in the imported machinery. In the case of Chinas high-technology
industries, Li (2011) nds that investing in foreign technology alone
does not enhance innovation in domestic rms, unless it is coupled
with an industrys own in-house R&D effort. On the contrary, the
inverse is true; he nds that domestic technology purchases alone are
found to contribute to innovation, suggesting that domestic rms
much more easily absorb indigenous technology.
260
Outward FDI and technology transfer

For multinationals from emerging economies, one of the major motivations to directly invest in developed economies is knowledge sourcing
through the setting up of joint ventures with foreign rms, R&D labs,
research institutions and universities and greeneld new production
facilities with R&D functions. Mergers and acquisitions of local rms
and institutions that own the needed technology know-how or simply the
research personnel or potential also serve this function. Asset exploration
has become a major type of OFDI from the emerging economies
(Dunning, Kim and Park, 2007). The mode of innovation becomes
increasingly open. Active knowledge sourcing through such OFDI will
serve as an effective mechanism for enhancing rms innovation capabilities, especially for rms with the necessary absorptive capacity.
Indigenous innovations and catching up

The diffusion and adoption of technology is costly, requires certain
preconditions and is often difcult. First, technology producers are usually reluctant to share the underlying capabilities, and MNEs also try to
control knowledge leakage because these are the core competencies
needed to maintain their competitiveness (Mallett et al., 2009, Fu et al.,
2011). Second, many technologies are tacit and difcult to transfer; thus,
in-house R&D is crucial for the acquisition of tacit knowledge and universities may provide some assistance. Third, knowledge is cumulative
and path dependent, suggesting that indigenous R&D is an important and
necessary element for the effective assimilation and adaptation of transferred foreign technology and the development of indigenous technological capabilities for catching up. Finally, foreign technologies created in
developed countries can sometimes be unsuitable for developing countries
if we consider the notion that technological change is a localised learning
by doing process (Atkinson and Stiglitz, 1969). Therefore, effective technological capability building in developing countries should make use of
both the indigenous innovation efforts and foreign technology transfer
(i.e., a two-leg forward strategy), although the relative importance of each
driver varies according to the stage of industrialisation and development
in the concerned developing country (Fu and Gong, 2011; Fu et al., 2011).
Foreign technology transfer and indigenous innovation, in fact, reinforce each other: localised innovation is a prerequisite for developing
261
domestic absorptive and creative capabilities to ultimately benet from

transfer mechanisms. Unconventional technology transfer mechanisms,
such as international R&D collaboration and outward direct investment, are only possible once local industries have developed world-class
rms with international recognition and possess the resources and clout
to collaborate with, or acquire, foreign rms. Such emerging MNEs are
dubbed national champions and play an active role in the acquisition
of new technology and know-how by leveraging global value chains to
innovate within their network (Fu et al., 2011).
Another role of indigenous innovation is that it has the dual function
of creating knowledge and promoting learning and absorptive capacity
(Cohen and Levinthal, 1989; Aghion and Howitt, 1998). An important
component of absorptive capacity is the R&D activities carried out by
local rms. Li (2009), Fu (2008) and Fu and Gong (2011) all support
this hypothesis based on experiences from China. Foreign technology
will generate a positive effect on local rms technological change and
upgrading only insofar as sufcient indigenous R&D activities and
human capital are present.
The two-leg forward strategy is also suitable for the technological
progress of a green economy. Foreign technology transfer remains an
important driver and has been a crucial part of the global solution for
reducing greenhouse gas emissions under the United Nations Framework
Convention on Climate Change (UNFCCC) framework. Nevertheless,
Bell (1990) argues that low-carbon innovation capabilities are likely to
depend on indigenous investment in training, R&D and reverse engineering. Based on the experience of the wind power, solar energy and electric
and hybrid vehicles sectors in India and China, Lema and Lema (2012)
nd that conventional technology transfer mechanisms such as patent
licensing, inward FDI and imports were important for industry formation
and takeoff. However, other mechanisms such as indigenous R&D,
global R&D networks and acquisition of rms in the West become
more important after these sectors have taken off and start catching up
with global leaders.
10.3 Science and technology development and growth

in the solar-PV industry in China and India
Over the past three decades, tremendous economic growth in China has
been accompanied with signicant growth in R&D expenditure. When
262
compared to peers such as India and Brazil, China has had the largest
increase in R&D expenditure, at an annual rate of 19 per cent since
1995, and draws the largest number of its youth towards research and
science careers (OECD, 2008).1 In addition, foreign rms had established more than 1200 R&D centres in China by 2008 (Zhu, 2010).
Although there is still a gap between Chinas technological capabilities
and those of OECD countries, Chinas science and technology (S&T)
sector has produced considerable innovative accomplishments during
the 11th Five-Year Plan period (20062010). The R&D intensity in
China has increased from 0.6 per cent in 1995 to 1.7 per cent in 2009. In
2009, the number of higher education graduates (in science, engineering, agriculture and medicine only) was about 8 times the 1995 level,
granted patents about 13 times higher, and high-technology exports
more than 28 times higher (OECD, 2008; China National Bureau of
Statistics2).
Considerable breakthroughs in research on renewable energy have
also taken place. For example, the range of notable milestones in 2009
include substantial progress on several adjustable-speed wind energy
power plants including two 1.5 megawatt (MW) in early stages of
production as well as a 2.5MW and 3MW in later stages of installation; notable progress in solar energy and battery technologies including 21 cities using solar for illumination; collaboration with Japanese
technology companies to develop small generators for wind-powered
irrigation in arid areas; and collaboration with the leading U.S. solar
company First Solar to develop a 2GW solar power station in Inner
Mongolia. Substantial ongoing resource allocations supported these
efforts for 20092010, such as RMB 20 billion budgeted for the
development of solar power plants, RMB 200 million for the public
electric cars project in 13 cities, RMB 2 billion to boost related parts
for electric cars and 6 billion to support innovation in battery technology (MOST, 2010).
Chinas production capacity for renewable technologies has grown
rapidly in the since 2003. In the solar-PV industry, Chinas global share
increased from less than 1 per cent in 2003 to that of the worlds largest
producer in 2008 (Climate Group, 2009; Strangway, Liu and Feng,
1
As of 2006, China is only second to the United States with 1.2 million full-time
researchers (OECD, 2008).
http://www.stats.gov.cn/english/statisticaldata/yearlydata
263
Table 10.1 Renewable technology targets for 2020 in China

Type of power generation
2006 actual 2010 estimates 2020 target
Total water (MW)

Small-scale water (MW)
Wind (MW)
Biomass (MW)
Feed-in solar (MW)
Solar-powered water heaters (m2)
Ethanol for fuel (million tons)
Bio-diesel (million tons)
Biomass pellets (million tons)
Gas from biomass (million tons)
130,000
47,000
2,600
2,000
80
100
1
0.05
0
8
180,000
60,000
5,000
5,500
300
150
2
0.2
1
19
300,000
85,000
30,000
30,000
2,000
300
10
2
50
44
Source: Martinot and Li (2007).
2009). According to Solarbuzz, China produced 2570 MW of solar cells

in 2009, accounting for 37 per cent of worldwide production. Chinas
global market share of on-grid building-mounted segments rose from
33 per cent in 2008 to 88 per cent in 2009.3 However, around 95 per
cent of Chinas PV products have been exported. The domestic demand
of PV installation in China was only 228 MW in 2009, although an
impressive 552 per cent increase over 2008. Moreover, China has set
ambitious national goals for 2020 (Table 10.1). These targets translate
to a level of renewable energy generation by 2020 that is three times that
of 2006 and an increase in renewable energy as a percentage of all
power generation to 21 per cent from 16 per cent in 2005. Finally,
these forecasts envisage that solar-powered water heaters will be
installed in one-third of all households by 2020.
Following strong economic reforms, Indian growth has also
occurred at a fast pace, as has expenditure on education and R&D.
During the rst decade of the twenty-rst century, Indian public
expenditure on education increased by 126 per cent from 2000 to
2009 according to the Indian Ministry of Statistics and Programme
Implementation (MOSPI).4 From 2001 to 2008, there was an 80 per
3
4
www.solarbuzz.com
http://mospi.nic.in/Mospi_New/site/India_Statistics.aspx?status=1&menu_id=14
264
cent increase in the number of universities; a 90 per cent increase in the

training of research doctors; and a 48 per cent increase in higher
education graduates in science, engineering, agriculture and medicine.
During its 11th Five-year Plan (20072012), the Indian central government raised its expenditure on science, technology and environment by 194 per cent, and local governments raised such expenditure
by 539 per cent compared with the 10th Five-year Plan period
(20022007) (MOSPI). Support has also increased for indigenous
research, design and development of new and renewable energy in
India. In its 11th Plan Proposal, the Ministry of New and Renewable
Energy (MNRE) highlighted how important it is that domestic industry is in a position to provide the market with cost-effective state of the
art quality and reliable products and services without excessive reliance on imports (MNRE 2006, p. 8). An up to 100 per cent subsidy is
given to public R&D institutions and universities for achieving welldened outputs and 50 per cent is available to industry. More subsidies are provided to industry in the initial stage if the time horizon
for maturation of technology is relatively high, and R&D effort is
encouraged in production upgrading, introduction of standards and
so on. R&D collaboration with foreign companies is also supported if
a larger national interest is visible and is served. India has realised
that R&D is critical in the energy sector and has considered R&D as a
public good to be nanced by the government, particularly for
R&D in solar, bio-energy, alternate fuel and storage technologies
(MNRE, 2006).
India is also a key emerging country in terms of solar power. MNRE
estimated that overall production was more than 175 MW for solar
cells and 240 MW for PV modules during 2008 and 2009 (India
Semiconductor Association, 2010). Different from the rest of world,
Indian PV applications mainly focus on off-grid (rather than on-grid)
connectivity and small capacity applications, used mostly for
public lighting and domestic power back up in cities and small electrication systems and solar lanterns in the rural areas. Similar to
China, India has relatively limited domestic PV demand and exported
75 per cent of PV cells to international markets (India Semiconductor
Association, 2010; Lema and Lema, 2012). India also has set national
goals for 2022 (Table 10.2), which in total are about ve times that of
the 2007 level. However, these targets are much lower than those set
by China.
265
Table 10.2 Renewable technology targets for 2022 (end of 13th Plan) in
India (in MW)
Type of power generation

Wind power
Small hydro power
Bio power
Solar power (grid/off-grid)
Distributed/decentralised renewable
power system
2007
actual
2012
estimates
2022
target
5,333
522
669
1*
N/A
10,500
1,400
2,100
50
950
22,500
3,140
4,363
N/A
N/A
Note: *Grid only.

Source: MNRE (2006).
10.4 Technology transfer and indigenous innovation in the

solar-PV industry in China and India
A phenomenal feature of the development model of the solar-PV industry in China is a strong emphasis on indigenous R&D. All the major
rms are R&D intensive, although they are also engaged in licensing
agreements to access PV technologies (Lema and Lema, 2012). Up to
2009, than 500 solar-PV rms and R&D labs in China actively pushed
the frontiers of related technologies (Climate Group, 2009). In general,
the industry has not only invested greatly in R&D but has also invested
upstream and downstream in the value chain including the processing
of silicon materials.
Based on total shipments of megawatts in 2010, national champions
Suntech, Yingli Solar and Trina Solar are ranked as global top-10
companies in the industry (Lema and Lema, 2012). Detailed information on these rms appears in Table 10.3. Suntech has the worlds
largest solar power plants. It develops, manufactures and delivers
solar modules to more than 80 countries in the world. Suntechs creed
is that technology is the core competency and innovation is the soul.
Suntech adopts a mix of mechanisms of technology transfer and indigenous innovation. Different from other high-tech companies that usually introduce the whole set of foreign equipment and machines,
Suntech only licensed the core equipment and then internalised foreign
technology through integrating with domestically produced equipment.
266
Table 10.3 Details of leading solar-PV companies in China and India

Chinaa
Year founded
No. of
employees
Turnover
(million)
Total assets
(million)
Sales in MW
Exports
(percentage)
of sales
India
Suntech
2001
20, 231
1998
1997
2005
11, 435 12, 863 >7000
1989
>600
2007
100500
$2901.9 $1893.9 $1857.7 > $500
$250
$35c
$ 5217.1 $3664.9 $2132.1 >$1000
N/A
N/A
67.4
79.20
40
8190
1572
94.7
1061.6
94.00
Trina
Solar
Moser
Baer
Solar
Yingli
Solar
1057
96.20
300d
90
TATA
HHV
BP Solab Solar
Note: aall data for Chinese companies are as of the end of 2010; bturnover, sales, and
exports for TATA BP solar are as of 20082009 scal year; cthe 2010 goal set by HHV
Solar; dproduction capacity by 2010.
Source: Company websites, annual reports and various press news.
Besides this conventional technology transfer strategy, Suntech has

focused more on in-house innovation since it was established. It put
5 per cent of its annual revenue to R&D activities every year at the
takeoff stage (Jiang, 2009). It now has a world-class research team
including more than 450 experts and has collaborated closely with
universities and research institutions in China and abroad. It has developed and commercialised its own core silicon and cell technologies,
solar module designs and encapsulation methods, which resulted in a
great number of patents.5 Suntech has set up vertical strategic alliances
with upstream and downstream companies and gained nancial and
5
As of December 31, 2010, Suntech Power Holdings had a total of 87 issued

patents, 216 pending patent applications in China, and 4 issued patents and 44
pending patent applications outside China through Suntech China, and 52 issued
patents and 46 pending patent applications in Japan, 8 pending patent
applications outside Japan through Suntech Japan and 2 issued patents in
267
non-nancial support from local and central governments. In terms of

unconventional technology transfer mechanisms, Suntech has carried
out OFDI and overseas acquisitions. For example, in 2006 it took over
Japans largest PV module producer MSK, which at that time held
98 patents and had more than 20 years experience in designing and
implementing building-integrated PV (Jiang, 2009). Since 2009, it has
set up several subsidiaries in the United States and Europe. The benets
are twofold: it is closer to the suppliers of high-standard immediate
products and the main consumers of Suntechs nal products, and it is
easier for Suntech to grasp the latest technologies of solar cells in these
technologically advanced countries.
Similar to Suntech, Yingli Solar and Trina Solar are both innovation
focused. Moreover, they both have adopted vertical integration business models and so have established more strategic technological collaborations with downstream suppliers. Yingli Solar has also engaged
in overseas investment. In addition, Suntech, Yingli Solar and Trina
Solar have successfully listed on the New York Stock Exchange, allowing these companies to make use of global resources. As a result, at the
macro level, China has become a global leading location in solar-PV
research and production, which has attracted major MNEs to set up
R&D labs or joint R&D labs in China (Lema and Lema, 2012).
Therefore, the model of technological capability building in the
Chinese solar-PV industry is an advanced indigenous R&D-led model
with close links from industry to universities and research institutions,
and with increasing national and international R&D collaborations.
The Indian solar-PV sector is a mix of three major approaches including
patent licensing, joint ventures and acquisitions and in-house R&D
(Mallett et al., 2009), which matches the current development level of
the technology and production capacity in the Indian solar-PV sector.
The biggest solar manufacturers in India are Moser Baer Solar, Tata
Power and HHV Solar (see Table 10.3 for more details). Moser Baer
Solar has long-term R&D plans, focusing on improving efciencies and
reducing cost through innovations. It has set strategic alliances with
American solar technology companies focusing on distribution and
production of solar technologies. It has its own R&D facilities across
Germany through KSL-Kuttler. In addition, it has 5 pending patent applications
led under the Patent Cooperation Treaty, which provides a unied procedure for
ling patent applications to protect inventions internationally (Suntech Annual
Report 2010).
268
India and collaborative research programmes with internationally

reputed research institutions and industrial players in India and abroad.
HHV Solar also has a mature R&D centre and overseas partnerships;
Tata power employed a conventional technology transfer mechanism
by setting up a joint venture, Tata BP Solar, with BP Solar, one of the
leading solar companies in the world.
Figure 10.1 summarises the main technology transfer and creation
mechanisms used in China and India in the solar-PV industry. It shows
that companies in both countries have used a mix of technology transfer and indigenous innovation mechanisms and these mechanisms
change across the sequence stages of development in companies. At
the early stages or takeoff phase, these companies have mainly adopted licensing, joint venture, joint R&D and in-house R&D. The importance of these mechanisms is the same in China and India. At the later
stages the catch-up phase, in-house R&D, local technology linkages, overseas R&D labs, and mergers and acquisitions have played
increasingly important roles, although joint ventures, R&D collaboration and licensing have remained essential. Comparing China with
India, the evidence suggests that China has taken a more home-based
outside-in technology transfer and indigenous innovation model,
whereas India has taken a more go-global active technology acquisition model.
To sum up, Figure 10.2 provides a general path of development for
the solar-PV industries of China and India, although their paths are not
absolutely identical. The comparison between targets of India and
China (Figure 10.1 and Figure 10.2) may have implications for the
green technology sectors of emerging economies.
10.5 National innovation systems and technology acquisition,

adaptation and development
Indigenous innovation and technology transfer are both important for
the development of the solar-PV industries of China and India. Private
rms are the major force in undertaking R&D and transforming scientic inventions into production technologies and ultimately commercialising them for the market. However, this cannot be sustained without
a favourable external environment. Therefore, it is necessary to have a
supportive national innovation system (NIS), formed by market-driven
private rms, all levels of government agencies, research and training
269
Indigenous
technology
creation
Takeoff Stage (China and India)

Local tech linkage with universities,
In-house R&D
Conventional
tech
transfer
Active
technology
acquisition
Cross border merge and acquisition

Overseas R&D lab
Joint R&D with MNC
Tech licensing
FDI/joint venture
Trade imports
China and India
Indigenous
technology
creation
Local tech linkage with universities,

Active
technology
acquisition
Cross border merge and acquisition
India
China
Overseas R&D lab
China
Joint R&D with MNC
India
China
Conventional
tech
transfer
Catch-up Stage (China and India)
Tech licensing
India
China
India
China
In-house R&D
India
India
China
India
China
FDI/joint venture
India
China
Trade imports
0
India
China
Figure 10.1 Technology transfer and indigenous knowledge-creation mechanisms adopted by solar-PV industry in China and India
Note: Top Chinese national rms: Suntech, Yingli Solar and Trina Solar; top
Indian national rms: Moser Baer Solar, TATA BP Solar and HHV Solar. 1
denotes low importance; 2 denotes medium importance; 3 denotes high
importance.
Source: Company websites, annual reports, Lema and Lema (2012), Strangway
et al., Lewis.
270
Takeoff stage
A combination of in-house R&D
with outside-in simple conventional
mechanisms, for example, licensing and
foreign invested joint ventures
Domestic universityindustry linkage
Catch-up stage
A combination of in-house R&D with more active
and more selective mechanisms, for example,
overseas acquisistions, overseas R&D labs
International and domestic universityindustry
linkage
Figure 10.2 Mixing and sequencing of technology creation and acquisition

mechanisms in solar-PV industry in China and India
institutions and nancial intermediaries, as well as the linkages and

interactions between these actors (Balzat and Hanusch, 2004).
The NIS is sector and context specic. The effects of individual actors
depend on the systems conditions such as the regulatory framework,
which ultimately inuences market demand and underlying technological push-and-pull dynamics (Walz, 2008). As such, government regulations can guide the evolution of a countrys NIS and help determine
competencies along with the international competitiveness of domestic
industries. Environmental regulation, for example, is a key driver of
domestic demand for sustainable technologies in water, energy and
transportation. When coupled with funding and favourable policies
aimed at creating and strengthening indigenous capabilities and technological expertise, there are real possibilities for developing countries
to take alternative paths towards development and leapfrogging into an
internationally competitive low-carbon economy.
Therefore, an important feature of the NIS, including the environmental innovation system, is collaboration and coordination between
the relevant government departments and the introduction of a comprehensive set of complementary innovation policies that can effectively
guide and incentivise rms to innovate. Policies may include increasing
investment in science and technology and public research institutions,
271
targeted tax incentives, increasing R&D nancial support, specic government technology procurement to support innovation and strengthening intellectual property rights.
The second important feature of the NIS is university-industry linkage. Universities are widely regarded as a major contributor to advances
in basic scientic research and innovation. However, scientic research
in universities should not be separated from industrial production. It is
important to build linkage between the two. The leading solar-PV
companies in both China and India have adopted this strategy. In the
case of China, universities are historically important in its NIS and have
played a leading role in terms of R&D expenditure and patents of
inventions (Liu and White, 2001; Li, 2009). The Chinese government
has been advocating a use-driven science policy, encouraging universities to serve the national economy by solving practical problems for
industry (Hong, 2006). University-industry linkages are built through
licensing, consulting, joint or contract R&D and technology services. A
second form of use-driven innovation occurs as a result of universityafliated or university-run enterprises, given the market-oriented
reforms and the low absorptive capacity of industrial rms and underdeveloped intermediary institutions (Ma, 2004; Eun, Lee and Wu,
2006). Chinese universities have played a signicant role in the promotion of the diffusion of frontier technology and the creation of new
country- or rm-level innovation outcomes. However, their contribution to the creation of groundbreaking innovations is still limited
(Fu and Li, 2010).
Foreign rms can be active players in the NIS through knowledge
transfer, knowledge creation (sometimes) and the effects of competition. On average, MNEs use relatively cleaner technology than domestic rms in developing countries so that there is a possibility of clean
technology transfer from foreign to domestic rms. This is especially the
case in joint ventures in green energy sectors such as solar power, wind
power and hybrid/electric vehicles. However, there is also a possibility
that MNEs are looking for institutional voids where they can effectively
relocate to pollution heaven. Zhang and Fu (2008) nd that as a result
of its lower pollution standards and lack of enforcement, China has
selectively attracted more heavily polluting industries. FDI in such
industries tends to target regions with relatively weak environmental
regulations. Therefore, the role of FDI in the national environmental
innovation system is two sided.
272
In terms of the national environmental innovation systems in China

and India, Walz (2008) found that from 2000 to 2004, neither country
specically aimed at decoupling environment and resource consumption from economic development. Moreover, he found that India had
been experiencing an increasing shortage of young scientists, representing a large barrier for capacity development in sustainability research
and public research in general. However, looking at both countries
now, growth in green technology and the development of a green
economy have been dramatic.
Table 10.4 compares the environmental innovation systems in China
and India. In China, it is a state-led innovation system. Since 2006,
China has made sustainable technology a primary component of
Table 10.4 Sustainability-oriented innovation systems of China and India
Technological
specialisation
Framework
conditions
Government
policy
Sustainable
R&D
China
India
Solar-PV, wind turbines,

electric cars and other
renewable energies.
State-led innovation system,
strong university-industry
link, strong general
manufacturing and trade.
Strong policy U-turn now
emphasising development
of sustainable technologies
since 2006 whereby a set of
scal, nancial,
technology, trade and
industry policies have been
introduced and public
procurement used.
Rapid increase in R&D
investment in this sector
since 2005/6. Strong
encouragement of
international R&D
collaboration in green
technology.
Wind turbines, solar-PV,

biopolymers and electric
cars.
Resilient private sector;
functional overall
framework conditions for
general innovation.
Increasing government
support, increasing R&D
investment, but the private
sector plays a leading role.
Material efciency and water

technologies. Wind turbine
industries started in 1990s.
Strong R&D in the private
sector in this area.
273
Table 10.4 (cont.)

China
Sustainable IP
FDI
attractiveness
(trade policy)
Exports of
sustainability
technology
products
Summary and
implications
India
Largest number of
Low number of patents in
transnational patents in
contrast to high
absolute numbers.
capabilities and IP in other
Major breakthroughs in
sectors.
this area since 2003.
Most attractive and far ahead Lowest inows. Relative
in magnitude.
strength of FDI in services
FDI strong in manufactursector.
ing sector.
Highest exports in solar (PV) High exports in solar-PV
panels. Electric cars
sector, but exports of
catching up. Leader in
minor international
wind turbine exports.
importance in the rest of
green sector.
A rapidly growing role of
Increasing role of sustainable
sustainable technologies
technologies since 1998.
since mid-2000s. Strong
Possesses well-functioning
demand and strong
overall framework
conditions for innovation
government policy support
for green technology.
and a vivid private sector.
Acquisition of foreign
Legacy of weak
technology through
environmental protection.
licensing and joint venture
at the takeoff stage,
followed by heavy
indigenous R&D activities.
FDI strength implies China
possesses most absorptive
capacity for technology.
Source: Authors summary and Walz (2008).
national policy and has made strides in many elds including wind
turbine and solar-PV technologies. As a result, it has been the export
leader in wind turbines and solar-PV products. In contrast, the private
sector plays a leading role in India. The overall framework conditions
are functional for innovation, with increasing government support.
274
Indias solar-PV exports and intellectual protection, for instance, have

increased dramatically. Nowadays, both China and India stand at the
forefront of many environment-related industries, and both countries
have formed a clear innovation strategy for the development of
environment-related industries and the transition to a green economy.
This is accompanied by strong government policies and funding for
green technology innovation, especially in China. A rapid catch-up in
environment technology-related innovation systems has taken place in
these two countries that lagged behind even as recently as the start of the
twenty-rst century. This evidence demonstrates the possibility that
developing countries could take an alternative path of development
with green technology industries growing with leaps and bounds and
hence leapfrogging into an internationally competitive low-carbon
economy.
10.6 Conclusions
This chapter analyses the strategy of mixing and sequencing different
technology transfer and indigenous innovation mechanisms in the
solar-PV industry in China and India and discusses the role national
innovation systems play in sustaining technology acquisition, adaptation and development. Both China and India have grown dramatically
in the solar-PV industry in a short time period. Their successful leapfrogging in this industry has signicant implications for the growth of
developing countries. It suggests that developing countries have opportunities to catch up with developed countries in the emergent green
industries.
The development of the solar-PV industry in both China and India
has made good use of mixed and sequenced mechanisms of indigenous
innovation and international technology transfers. The importance of
different transfer mechanisms varies across the different levels of technology and production capabilities that the domestic industries have at
a given development stage. Most leading solar-PV companies in both
countries started from international technology transfer through licensing and joint venture with MNEs. At the same time, all of them have put
increasingly substantial effort into in-house R&D for the assimilation
and adaptation of transferred technologies and the development of
indigenous technological capabilities. However, once the basic production and technological capabilities were built up, they started more
275
active knowledge acquisition and creation through indigenous innovation, international R&D collaboration and cross-border mergers and
acquisitions. Such experiences suggest that accomplishing such catchup processes requires a combination of international technology transfer and indigenous innovation. Technology transfer will be a feasible
and evidence-proven entry point for developing countries, although this
should be accompanied with substantial indigenous efforts in assimilation and learning. In this respect, attractive bundles of trade, investment
and technology policies are important to leverage technology transfer
mechanisms.
The experience of China suggests that the state has a crucial role in
initiating the transition and maintaining the momentum of the catch-up
process. Given the nature of technology as well as that of the environment as a public good, government-funding support through focused
R&D programmes has been crucial in promoting technological breakthroughs and hence indigenous technological capabilities as well as
the acquisition of foreign green technology. The experience from China
also demonstrates that the development of a green economy requires
a set of complementary and coherent policies covering regulatory,
nance, technology and industrial policies to motivate and reinforce
the transition and ensure a substantial change. In addition, the connection between industry and universities and public research institutions is
important in transforming outputs of scientic research into applied
technologies and assisting rms to move closer to the technology
frontier.
Despite the rapid progress of solar-PV along with other green technology industries in China and India, substantial challenges remain.
First, in China, for example, although there is strong government support for green technology, the lack of core indigenous technological
capability still lingers in a wide range of sectors. Industry-academic joint
research is not strong despite the substantial government push for
greater research-industry linkage. Many research outputs from the
universities either do not meet the needs of industry or are put on the
shelf and not commercialised (Strangway et al., 2009). Thus, there is a
demand for increased investment in science and technology and in
education, along with improvement in government policy and services
to push research organisations into the market.
Second, in both China and India, environmental regulations are
relatively weak, as is the enforcement of these regulations, in general,
276
and with considerable variations across regions. A new problem that

emerged in the development of industrial capabilities in the green technology sector is the heavy pollution generated in the production process
of renewable energy or the manufacturing process of related green
energy equipment. For example, the production processes for some
important material used for making solar panels, especially the purication of polisilicon, is high in energy consumption and heavy in environmental pollution. Only six or seven rms in developed countries
have developed a cleaner production technology. Chinese and Indian
rms still use old and dirty technologies. Other waste products from
solar-PV systems also pollute earth and water (Du and Cao, 2010).
While China and India are, respectively, exporting more than 90 per
cent and 70 per cent of their solar-PV panels to international markets,
the waste is left behind. Therefore, signicant attention to these ethical
issues, adequate strategic planning and continued technological innovation to reduce the negative externalities in the production of the clean
energies should be emphasised. Developing countries need to avoid the
trap of becoming the new world manufacturing workshop of environmentally damaging industries for green products.
Finally, the transition from a traditional to a green economy requires
not only technology and nancial input but also involves considerable
transition cost. The availability of cheap coal and the signicant switching
cost suggests that the transition will not come naturally. Indeed, the
current reliance on coal in both developed and developing countries
indicates there is a long way to go. Middle-income countries, which
have already taken the traditional industrialisation route without sufcient technological and nancial capabilities, will face the biggest challenge in comparison to the high- and low-income countries. In low-income
countries, the sunk cost of transitioning from existing systems of production to a green economy is low. Combining this advantage with appropriate technology and nancial resources, the low-income and some
middle-income countries will be able to carry out the transition at a
relatively low cost.
part iii
11
Internationalisation, reverse learning

and capabilities upgrading: The case
of Huawei and ZTE*
xiaolan fu and zhongjuan sun
11.1 Introduction
In recent years, a signicant number of technology-intensive rms have
not only operated successfully inside China but also moved quickly
offshore, penetrating the market previously dominated by established
Western multinational companies (Zheng, 2014). While such internationalisation plays a disproportionately large role in the telecommunication industry of China as the locus of both technology development
and its corporate capability upgrading, few studies elucidate the learning process by which these internationalisation activities developed.
Related studies address Chinas internationalisation strategies
(Prange, 2012), innovation capability development (Fan, 2006; Zhou
and Li, 2008), telecommunication market and industry (Chang, Fang
and Yen, 2005; He and Mu, 2012), innovative capability and export
performance (Guan and Ma, 2003), technology learning and development (Jin and von Zedtwitz, 2008) and the interaction between R&D
and marketing in these rms (Li and Atuahene-Gima, 2001). However,
our understanding of the nature and the process of the learning in this
process and its impact on strategic capabilities of rms is still limited.
A substantial literature addresses learning and capabilities development as a critical challenge for latecomer multinational rms in developing countries, dislocated from centres of technological development
(Gassmann and von Zedtwitz, 1998; Boutellier, Gassmann and von
Zedtwitz, 2002). Over the past three decades, a signicant number of
* The research on which this chapter is based was supported by the British Academy
Grant No: SG122404. We would like to thank Jizhen Li and Jiangang Victor
Zhang for helpful discussions and Huawei Technologies Ltd. and Zhongxing
Telecommunications Equipment Corporation for their support for the eldwork.
279
280
studies have examined this issue at the project, rm, industry and
national levels of analysis in a developing country context, South
Korea in particular. These studies identify different modes and processes of technological capabilities upgrading based on development
motivation and source of initiation (Kim and Lee, 2003) and argue that
the technology capabilities development process in newly industrialised
economies follows three steps of transfer, absorb and diffuse and nally
move to innovate and develop (Kim, 1997; Kim and Nelson, 2000).
For China, a small number of studies describe learning activities in
Chinese organisations. Firms that have been studied include Lenovo
(Xie and White, 2004; Liu, 2007; Sun et al., 2014), Huawei (Nakai
and Tanaka, 2010) and Datang (Mu and Lee, 2005; Fan, 2006; Jin
and von Zedtwitz, 2008). Studies of Chinas telecommunication
industry have either focused on the rms entry modes for international markets and technology catch-up strategy (e.g., Fan, 2006)
or the role of the government (Mu and Lee, 2005) and capability
accumulation based on a single case study (e.g., Xie and White,
2004; Sun et al., 2014 on Lenova) that nds stage-wise development
from initial sales, distribution and service activities to manufacturing,
product and process design and nally developmental R&D. Huawei
has been a classical case for study, but most of the existing studies
focus on its general experiences of catch-up (Fan, 2006; Liu, 2010) or
its IPR strategy (e.g., Nakai and Tanaka, 2010). None of the existing
studies has specically examined the channel, mode and impact of
reverse learning in rms internationalisation process, especially for
MNEs originating from developing countries.
The objective of this chapter is to shed light on the reverse learning
and capability upgrading in the process of internationalisation of
Chinese multinational enterprises through an in-depth study of multiple
cases, specically, Huawei and Zhongxing Telecommunications
Equipment Corporation (ZTE), both leading global information and
communication technology (ICT) solutions providers. These two cases
provide a basis for conceptualising the process by which corporate
capability upgrading evolved from internationalisation of business
activities to become successful, global rms based on proprietary
technology and capability. We focus on the evolving nature of learning
by which the rms were able to realise such a transition and the
implications for both research and practice. This chapter contributes
to the literature by providing the rst study of reverse learning through
Internationalisation, reverse learning and capabilities upgrading
281
internationalisation and capability upgrading of emerging market

MNEs based on multiple case studies. In particular, we focus on the
channels of reverse learning and their impact on the upgrading of rms
overall corporate capabilities, which include both technological and
non-technological capabilities.
This chapter is structured as follows: Section 11.2 reviews the literature on internationalisation and knowledge sourcing. Section 11.3
provides an overview of internationalisation of Chinese rms.
Section 11.4 discusses methodology. Section 11.5 presents background
information of the rms under study. Section 11.6 discusses the reverse
learning and capability upgrading in the internationalisation process
based on a comparative analysis of two cases. Section 11.7 offers
conclusions and discussions.
11.2 Internationalisation and knowledge sourcing:

received wisdom
The extant literature related to internationalisation and technology
mostly focuses on the internationalisation of R&D/technology/innovation systems at the rm level (e.g., Granstrand, Hkanson and
Sjlander, 1993; Granstrand, 1999; Carlsson, 2006; Beersa, Berghll
and Poot, 2008; Dunning and Lundan, 2009) and the internationalisation of technology-based rms (e.g., Yli-Renko, Autio and Tontti,
2002; Johnson, 2004; Evangelista, 2005; Blomqvis et al., 2008;
Andersson, Curley and Formica, 2010). The research focusing on how
the process of internationalisation contributes to the accumulation of
organisational capabilities of the investing rms focuses mainly on the
impact of organisations networks on the internationalisation market
expansion.
Network relationships are an effective tool for international market
expansion because rms not only can access indirect network relationships from other members in the network but can also learn from
network partners, such as customers, collaborators and competitors.
Network relationships, particularly alliances with partners in the foreign market, constitute an effective strategy to overcome deciencies,
such as a lack of resources or foreign market knowledge, on entering
international markets (Lu and Beamish, 2001). They also act as a
strategic tool for overcoming the negative perceptions held by foreign
investors (Zain and Imm Ng, 2006). In addition, networks can help
282
entrepreneurs identify international opportunities and establish credibility and often lead to strategic alliances and other co-operative strategies (McDougall and Oviatt, 2005), which may provide information
about business opportunities, foreign market characteristics, obstacles
and problems involved in the process, resulting in decreased risk
(Barney and Hansen, 1994; Gulati, 1999; Iyer, 2002). Moreover,
according to the Uppsala model (e.g., Johanson and Vahlne, 1977)
and the network model (e.g., Johanson and Mattsson, 1988), internationalisation is an incremental process to accumulate knowledge via
learning by doing international operations, with rms increasing propensity to commit to international operations.
Knowledge of international markets is another type of knowledge
that rms acquire through internationalisation (Johanson and Vahlne,
1977). Market knowledge is important because it makes it easier to
cope with both opportunities and problems that might arise when
conducting operations abroad (Brdin and Lngnr, 2012). Firms can
acquire objective knowledge and experiential knowledge through internationalisation. The objective knowledge can be taught, whereas experiential knowledge can only been learned through personal experience
(Johanson and Vahlne, 1977).
Some research investigates the relationship between internationalisation and technological capabilities (e.g., Elango and Pattnaik, 2007;
Kafouros et al., 2008; Filatotchev and Piesse, 2009; Lowe, George and
Alexy, 2012). As an effective channel to form the direct linkages with
foreign partners, internationalisation has facilitated cross-border technology learning and assimilation. Enterprises are able to build capabilities in international operations through networks, learning, transfer of
knowledge and transfer of capabilities (Kogut and Zander, 1993;
Tsang, 1999; Elango and Pattnaik, 2007).
FDI has been recognised as an effective mechanism to facilitate crossborder knowledge ows. It not only brings technology from parent
companies to the host country through training, imported machinery
and equipment, demonstration effects and knowledge spillovers
(Drifeld and Love, 2007; Fu, 2012b), but it is also argued that direct
investment abroad directly can be used for strategic asset seeking. For
example, technological and managerial knowledge absorbed by subsidiaries in the host countries would reverse ow back to the home country
(Singh, 2007; Fu, 2012b). Investment abroad has been found to help
rms build up R&D and organisational capabilities through
283
organisational learning (Kuemmerle, 1997; Steensma et al., 2004). The

productivity of the subsidiaries is found to be associated with the
performance of the parent company, which may, to a certain extent,
reect the impact of knowledge ow from subsidiaries on the performance of the parent company, although the causality may go the other
way (Harzing and Noorderhaven, 2006; Drifeld et al., 2010; Marin
and Giuliai, 2011). The productivity of rms in the UK is linked with
their investment levels in the United States (Grifth et al., 2006). The
authors suggest that an underlying desire in the internationalisation of
production is not to exploit existing technology within the rm but to
acquire the leading-edge technology within a host economy.
In the context of developing countries, increasing research has found
evidence of reverse technology transfer impelled by MNEs from emerging economies (Child and Rodrigues, 2005; Liu and Buck, 2007; Lima
and Barros, 2009). Not all multinationals land in foreign markets with
ownership advantages that are traditionally possessed by MNEs from
industrialised economies (Fosfuri and Motta, 1999; Siotis, 1999). This is
likely the case for many Chinese MNEs, which, therefore, may not only
seek market expansion through direct investment in foreign countries
but also see acquisition of strategic assets that are not available in their
home markets for the upgrading of their technological capabilities.
11.3 Internationalisation of Chinese rms: an overview

Since the introduction of the go global strategy in the late 1990s,
internationalisation of Chinese rms through outward foreign direct
investment (OFDI) has increased rapidly. The total stock of Chinese
OFDI has increased from near zero in 1990 to around US$300 billion in
2010 (Figure 11.1). The growth of Chinese overseas investment is
consistent with its economic development according to the investment
development path theory developed by Dunning and Narula (1996).
After more than 30 years of rapid economic growth, China has become
the worlds second largest economy, the worlds largest exporter and a
major driving force for global economic growth. In parallel with its
continuous expansion in economic growth, Chinese rms OFDI has
also increased rapidly.
Although mining has been an important destination of Chinese overseas direct investment, it only accounted for 14 per cent of total Chinese
OFDI stock and 6 per cent of total overseas rms by 2010. In fact,
284
350000
300000
250000
200000
150000
100000
50000
0
1981 1983 1985 1987 1991 1993 1995 1997 1999 2001 2003 2005 2007 2009
Figure 11.1 Stock of Chinas outward direct investment, mil$

Source: MOFCOM.
manufacturing has been the largest sector in terms of total number of

overseas rms, accounting for 29 per cent of total Chinese invested
overseas rms by 2010. Wholesale and retail trades as well as leasing
and business services are the top two and three destination sectors of
Chinese OFDI. Four per cent of the overseas rms were in scientic
research by 2010 (Table 11.1).
Amongst the rms invested in the manufacturing sector, as
Figure 11.2 shows, evidence from a rm-level survey carried out in
Guangdong province in 2010 suggests that the industry dispersion of
the Chinese overseas investment is wide, ranging from textiles and
apparel to metal processing, automobile and electronic industries (Fu,
Liu and Li, 2013).
These Chinese rms going abroad are for a variety of motivations. Fu
et al. (2013) examine the overseas investment strategies of rms in
Guangdong Province of China based on a rm-level survey conducted
in 2010. They focused on the characteristics of rms that undertake
OFDI, the incentives for and the main obstacles of overseas investment,
the destination of OFDI from Guangdong, the strategies to conduct
OFDI and the impact of overseas investment on company performance
and future development. According to their ndings, rms in
Guangdong Province invested overseas for a wide range of motivations,
including to exploit international markets and to acquire advanced
technological and managerial knowledge. Figures 11.3 and 11.4 demonstrate some major objectives for surveyed rms investing in
285
Table 11.1 Industry distribution of Chinas OFDI: by FDI stock and

number of subsidiaries, 2010
Industry
OFDI
stock,
percentage
of total
Industry
Leasing and business 31

service
Banking
17
Number of
overseas
subsidiaries,
percentage of
total
Manufacturing
29
23
Mining
14
Wholesale and retail

trade
Transport, storage
and post
Manufacturing
Computer services
Real estate
13
Wholesale and retail

trade
Leasing and business
service
Construction
Mining
6
3
2
5
4
4
Construction
Scientic research
Public utility
Services to
households
Agriculture
Environment
management
Lodging & catering
Culture &
entertainment
Public services
Total
2
1
1
1
Agriculture
Scientic research
Transport, storage and
post
Services to households
Computer services
Real estate
Lodging & catering
1
0.4
Banking
1
Culture & entertainment 1
0.1
0.1
Public utility
Others
0.02
317210.59 Total
13
3
2
1
1
1
0.5
16107
Source: Ministry of Commerce (MOC) statistics, Peoples Republic of China.

http://english.mofcom.gov.cn/article/statistic
286
IT and
communication
industry
6%
Craftworks and
others
9%
Agricultural & food

production
4%
Textile, clothes and

shoes production
11%
Wood & paper-making
related industry
7%
Electronic
appliance
industry
9%
Printing-related industry
2%
Chemical industry
4%
Industry of medical products
13%
Automobile
industry
13%
Other industrial
equipment
4%
Electronic
equipment
production
2%
General
Industry of
equipment
metal
manufacturing products
2%
6%
Rubber- & plasticrelated industry

6%
Metallurgy &
smelting
2%
Industry of non-metal
construction products
2%
Figure 11.2 Manufacturing rms with overseas investment

Source: Fu et al. (2013).
Optimise the allocation of resources globally
Use local convenient financial system and international payments
Use the preferential policy of host country
Prevent trade friction (anti-dumping, trade quota)
Avoid (tariff and non-tariff) trade barriers
Reduce the cost of labour force
Transfer domestic excess production capacity and avoid domestic competition
Increase the global market share
Seize local market
Exploit international market
Use local power resource
Buy raw materials and parts at low price
Use local sales network
Use local brand resource
Use strategy resource such as local technology
Acquire talent resource
Acquire advanced technology and management experience
44.7%
41.6%
44.0%
41.1%
36.0%
28.4%
31.6%
65.0%
61.0%
76.7%
23.7%
28.4%
66.3%
56.4%
55.1%
53.1%
75.6%
0.0% 20.0% 40.0% 60.0% 80.0% 100%
Figure 11.3 Main incentives for investing in developed countries

developed and developing countries, respectively. For rms that

invested in developed countries, to explore international markets and
to acquire advanced technology and management knowledge are the
top two objectives. About 77 per cent and 76 per cent of the surveyed
Optimise the allocation of resources globally

Use local convenient financial system and international payments
Use the preferential policy of host country
Prevent trade friction (anti-dumping, trade quota)
Avoid (tariff and non-tariff) trade barriers
Reduce the cost of labour force
Transfer domestic excess production capacity and avoid domestic competition
Increase the global market share
Seize local market
Exploit international market
Use local power resource
Buy raw materials and parts at low price
Use local sales network
Use local brand resource
Use strategy resource such as local technology
Acquire talent resource
Acquire advanced technology and management experience
287
48.6%
31.3%
55.9%
39.7%
37.1%
42.95%
38.0%
69.45%
72.0%
70.7%
48.6%
51.4%
68.1%
35.2%
42.0%
40.6%
43.3%
0.0% 20.0% 40.0% 60.0% 80.0%
Figure 11.4 Main incentives for investing in developing countries

rms regarded these two as their objectives to invest in developed

countries, respectively. However, rms that invest in developing countries appear to have wider and somewhat different objectives than those
that invest in developed countries. For example, only 43 per cent of the
rms regard to acquire advance technology and management experience as an important objective for investing in developing countries.
Instead, market expansion appears to be the most important objective
to invest in developing countries, with about 72 per cent of the surveyed
rms regarding to seize the local market as an important motivation.
In the survey, the proportion of income from new products/technologies is positively related to the percentage of rms with overseas
investment. New products/technologies contribute more than half of
the income of 64 per cent of rms with overseas investment. We can see
that for rms that have more than 50 per cent of their revenues generated from new products/technologies, 55 per cent have overseas
investments. This gure drops to less than 30 per cent among rms in
which new products/technologies contribute less than 10 per cent of
their revenues.
On average, rms that go global through OFDI appear to have
signicantly higher innovation frequency and more investment in
R&D and have achieved a higher percentage of sales accounted for by
new or signicantly improved products. As Table 11.2 shows, the
proportion of innovators in the rms with OFDI was 94 per cent,
which is signicantly higher than that in the rms without OFDI at 82
288
Table 11.2 Innovativeness of rms: comparing rms with and without

OFDI
Non OFDI
OFDI
t-test
Innovation
New sales
R&D
0.819
0.938
2.698***
0.219
0.295
1.944*
3.076
4.624
3.972***
Note: Innovation here is measured by dummy, value 1 indicates innovator and 0

otherwise; t-test Ho: Mean(Non OFDI)Mean(OFDI) = 0.
Source: Authors estimate based on the Guangdong OFDI survey.
*** 1% signicance; ** 5% signicance; 10% signicance.
per cent. The average percentage of new product sales in total sales is 30
per cent, also signicantly higher than that in the rms without OFDI.
The comparison of R&D spending also reinforced the message. The
average R&D spending-to-sales ratio is 4.6 per cent in rms with OFDI,
which is signicantly higher than that in rms without OFDI. The t-test
statistics indicate that the differences between the two groups are statistically signicant at the 1 per cent or 10 per cent level.
This result indicates that internationalisation through OFDI is positively associated with the higher innovation capacity of the investing
rms. What is the direction of the causality? In what follows, this
chapter examines learning and capability upgrading in the process of
internationalisation of Chinese multinational enterprises based on case
studies of Chinese MNEs in the information and telecommunications
industry so as to shed light on the causality of the impact of internationalisation on capabilities upgrading in the Chinese MNEs and the transmission mechanisms.
11.4 Methodology
Research Method
We constructed a multiple-case exploratory design to develop the existing theory. The case studies focus on individuals and analyse dynamic
issues and are particularly suitable for in-depth study of one problem.
First, exploratory case studies allow researchers to construct theories in
a relatively new research area with rather limited literature; the study on
289
the learning mechanism and corporate capability upgrading in the

process of internationalisation of Chinese multinational enterprises is
appropriate for this method.
Second, the characteristics of the learning mechanism, corporate
capability upgrading and the process of internationalisation are comprehensive, complex and dynamic; hence, the case study is necessary for
comprehensive and in-depth analysis of the phenomenon. We attempt
to study the learning mechanism and corporate capability upgrading
focusing on the experience of two cases by exploring their internationalisation process.
Third, the interviews with key decision makers can enhance our
understanding of the problems and issues occurring in daily operations,
so as to apply the theories and framework that could interpret the
valuable experiences that occurred during the process of internationalisation, even to produce new theories on the learning mechanism and
corporate capability upgrading.
Research Design
Eisenhardt (1989) studied building theories from case studies and
its roadmap, which is constituted by several main steps including
the denition of research questions, selecting cases, collecting and
analysing data, shaping hypotheses, comparison with the literature
and nally drawing conclusions. This roadmap synthesised previous
work on qualitative methods (e.g., Miles and Huberman, 1984). In
this study, we will follow Eisenhardt (1989)s design of case study
research.
We introduce the research target in the introduction section and then
introduce the way to select and collect cases in the methodology section.
The introduction to cases analyses the case history of internationalisation, along with learning and corporations capability upgrading. Our
discussion on the comparable analysis between two case practices and
between the practice and the literature appears in the section on shaping
hypotheses.
Case selection
Over the past three decades, the telecommunication industry has sprung
up (Yu and Wang., 2008), and it plays a pivotal role in national growth
290
and rm competitiveness in the current knowledge economy (OECD,

2004; Takahashi et al., 2004; Ollo-Lopez and Aramenda-Muneta,
2012). Meanwhile, many Chinese multinational enterprises in telecommunications are classic examples of Chinese rms going out because
they have achieved great success in internationalisation. These rms not
only have rich experience to enter foreign markets but also learn relevant technology, obtain management knowledge and achieve upgrading of their corporate capability. Therefore, the communication
industry is our research eld.
In this study, the following qualication criteria were applied to
explore the learning mechanism and corporate capability upgrading in
the process of internationalisation of Chinese multinational enterprises. First, from the perspective of time, decades of internationalisation experience allow historical analysis of successes and failures, and
these are classic examples because of their growth, performance and
contribution to the world. Second, the depth and breadth of academic
analyses based on these rms and the archival data are available.
Third, because of their comparability, we can identify the relevant
conclusions including internationalisation practices, learning activities and corporate capability upgrading. Fourth, they are appropriate
for building the theory. Therefore, Huawei and ZTE are the two cases
selected for our study.
Both Huawei and ZTE are representative rms for at least three
reasons. First, the tremendous successes achieved by these two companies in fewer than 30 years provide a rich research eld and also are
well worth learning from for companies that are striving for internationalisation. They were built up after the Chinese economic reform
and opening up to the outside world. Starting their internationalisation in the 1990s, they have become two of the leading multinational
rms in the ICT industry in China and in the world. All of these
achievements are good examples for other Chinese rms undertaking
internationalisation. Second, they also have, to a degree, similar internationalisation paths but also slightly different ones in the past few
decades; hence, their experiences have strong comparability.
Accordingly, a comparison between them can lead to some interesting
results. Third, they employ good management practices with documented savings and le arrangements, so we can easily collect data
from them. Finally, Huaweis and ZTEs internationalisation in both
291
market expansion and corporations capability upgrading are also an

interesting area for researchers.
Data Collection
We collected extensive archival data on our two cases and the communication industry from 2011 to 2014. Most of these data are interviews
of the key projects president and senior-level ofcers or managers from
newspapers and business magazines. We believe that the interview
documentation is the most reliable public resource for our research
purpose. First, the interviews from mainstream conferences, newspapers and business magazines are in general credible. Second, the interview topics with senior-level managers or ofcers between the two cases
covered different angles and are representative of the research target.
Third, the senior-level managers or ofcers understand better the corporate strategy and corporations capability upgrading and internationalisation process, which ensures the efciency of our research.
To study the learning mechanism and corporate capability upgrading
in the process of internationalisation of Chinese multinational enterprises, we picked 17 interview documentations that satisfy two conditions: (1) the interview topics relate to the research problem and (2) the
interviews were conducted between 2011 and 2014 with the projects
president, senior-level ofcers or managers. The date, interviewer and
topics discussed are listed in Appendix 11.1.
11.5 Overview of the cases

Huawei and its internationalisation
Huawei Technologies P/L was founded in 1987, is the second largest
telecommunications equipment company in the world by revenue and is
poised to become the largest (Ahrens, 2013). It is an entirely employeeowned private company that provides customised network solutions for
telecom carriers around the world with a vision to enrich life through
communication. Huawei provides competitive solutions and services
via end-to-end capabilities across the carrier networks, enterprise and
consumer markets that have been deployed in more than 140 countries,
serving more than one-third of the worlds population.
292
Huaweis internationalisation takes the path of using the countryside

to surround the cities. In this case, it was developing countries instead
of the countryside and developed markets instead of the cities. It is the
rst internationalisation push to occupy developing countries markets
from 1996 to 2004, and surrounding developed countries has been its
second step since 2001.
The rst phase of internationalisation occupying developing markets
While business was booming in China in the 1990s, competition in
advanced technologies from international competitors was still erce,
and Huawei looked abroad for continued growth. The builder of
Huawei, Zhengfei Ren, likened Huaweis situation to that of a mountain goat needing to run faster and climb higher than a lion so as to
avoid being eaten. Subsequently, Ren said that Huawei needed to be a
wolf, a metaphor he used for years.
Huaweis rst international customer was Hutchison Telecommunications (owned by Li Ka-shing) in Hong Kong in 1996, which
purchased switches and related equipment for its xed-line network.
Hong Kong was a well-developed telecom market, yet it was close to
Shenzhen, providing Huawei with an excellent rst test case.
Then, the following year, Huawei formed a joint venture in Russia
with the Beto Corporation to produce switching equipment, essentially
assembling Huawei switches in Russia. Huawei was able to undercut
international prices by around 12 per cent, but its after-sales service was
what really impressed the Russians. Its rst sale amounted to only
US$12 million, but by 2001 its sales had reached US$100 million.
Soon after entering Russia, Huawei made sales in Thailand, Brazil,
and South Africa. Its pricing became more aggressive, often undercutting rivals by 30 per cent. With its status as a national champion,
Huaweis international marketing strategy is oriented to winning
many international friends, like Chinas diplomatic route, and
Huawei became an active partner under this strategy, which resulted
in a large increase in prots. In November 2000, Chinas vice-premier,
Wu Bangguo, travelled together with Ren Zhengfei during a trip to
Africa, laying the foundation for future business deals, including a
US$20 million contract in Ethiopia in 2003, and a US$200 million
code division multiple access (CDMA) project in Nigeria in 2005
(Kuo, 2006). Ghana, Mauritius, Morocco, Congo and Kenya followed
in 2006, along with another large Nigerian contract.
293
In the internationalisation process of occupying developing countries, Huaweis lower price for affordable technology is its biggest
advantage. This push for internationalisation led to growth such that
by 2004, its international revenues were higher than its domestic
revenues.
The second phase penetrating into the advanced markets
Following successes in Russia and a number of developing markets,
Huawei started to turn its attention to developed markets. In 2001,
Huawei made its rst major sales in Europe, to the Netherlands and
Germany. The wireless station product sold to the Dutch enabled multiple communications standards to be run, and upgrades were done by
software rather than hardware (Pomfret, 2010). The product was a
good example of cost innovation, in that it provided advanced features
at low cost, while saving the carrier money on hardware (Zeng and
Williamson, 2007). The Germans purchased optical network (SDH)
products.
Soon after, Huawei made a sale to Neuf, the French operator, not
only offering rock-bottom prices but also actually building part of it free
of charge and allowing the operator to run it for three months to test it
before purchase (Farhoomand and Ho, 2006). A subsequent sale to the
United Arab Emirates made that country the rst Arab state with 3G
technology.
In 2004, Huawei made sales to a Danish company (in Portugal) and
then made a major sale to the Netherlands for building out the 3G
network. In 2005, British Telecom (BT) included Huawei as a preferred
supplier for its massive next-generation network. Some analysts believe
that this was a crucial factor in raising Huaweis international prole,
because soon after it signed a global supplier agreement with Vodafone
(Conti, 2007).
On Valentines Day in 2001, Huawei entered the U.S. market, setting
up an ofce in Plano, Texas. Three years later, Huawei was still without
a single American customer (Prasso, 2011). Huawei has since had
success, however, and its Plano ofce is now its U.S. headquarters,
overseeing its 12 other ofces and 7 R&D centres. In the United
States, it has 1,100 employees, of whom 900 or so are Americans. It
has yet to get a tier-one customer, but it is gaining traction in middle
markets. Its North American sales were US$765 million in 2010, with
294
customers including Leap (equipment and devices); Best Buy; and,

purportedly, Level3 Communications.
Meanwhile, Huawei was also entering the emerging Internet data
communications market. The giant in this area had been Cisco Systems,
with 80 per cent of the Chinese router market. Three years later,
Huawei had chipped away at Ciscos share and had captured 12 per
cent to Ciscos 69 per cent. While lower prices certainly helped, Cisco
claimed that Huawei had stolen its software (as well as user interface
and manuals), and parts of Ciscos code were found in Huaweis products. By the time Huawei settled with Cisco, Huawei had about a third
of the market (Ahrens, 2013).
By 2002, Huawei had also overtaken Shanghai Bell, the dominant
China-based international joint venture (IJV) at the time. In 2004,
Huawei amped up its second major international push, with extensive
credit backing from the China Development Bank (CDB), which provided a credit line of US$10 billion, and the Export-Import Bank of
China, which provided an additional US$600 million. With this robust
backing, Huawei started to make a major global push. It slashed prices
well below those of its competitors, purportedly sometimes by as much
as 70 per cent, and provided vendor-nanced loans to the customers.
Sales in the rst half of 2005 skyrocketed to more than US$4 billion, an
85 per cent increase over the previous year. More than 50 per cent of
this value came from abroad. Sales growth in 2007 was more than 50
per cent; in 2008, when most suppliers saw negative sales growth,
Huaweis sales still grew more than 40 per cent. In 2011, Huawei and
Ericsson recorded approximate net revenues of US$32.9 billion and
US$32.4 billion, respectively. By July 2012, Huaweis sales had bested
those of Ericsson by US$500 million (based on half-year sales).
Huawei currently has around 110,000 employees worldwide, with
30 per cent or so in Shenzhen. Approximately two-thirds of its revenues
come from international markets, and, according to Huawei, it works
with 45 of the 50 largest carriers globally. Huawei has also led more
than 49,000 patents, as its focus on R&D continues. It has also played a
leading role in standards development, both in China and abroad. It is
the leading contributor to LTE core specications, and it holds 83
positions in various standards bodies (Huawei press release, 2011).
According to Huaweis website, it participates in The 3rd Generation
Partnership Project (3GPP), APT (Asia-Pacic Telecommunity), ARIB
(Association of Radio Industries and Businesses), ETSI (European
295
Telecommunication Standards Institute), IEEE (Institute of Electrical and

Electronics Engineers), IETF (Internet Engineering Task Force), ITU
(International Telecommunication Union), TIA (Telecommunication
Industry Association) and WWRF (Wireless World Research Forum).
Aside from allegations of intellectual property impropriety, Huawei has
undoubtedly become a technology leader.
ZTE and its internationalisation

The ZTE Corporation is a Chinese multinational telecommunications
equipment and systems manufacturer, headquartered in Shenzhen,
China. It was founded in 1985 in Shenzhen by a group of state-owned
enterprises associated with Chinas Ministry of Aerospace. Listed on
the Shenzhen Stock Exchange in 1997, since 2004 ZTE can also be
found on the Hong Kong Stock Exchange. Today, general investors
hold 67 per cent of stock shares, with overseas investors holding 18.3
per cent a number that continues to increase every year; the company
has 107 ofces worldwide and subsidiaries in more than 100 countries.
With a market share of 4.3 per cent in 2012, ZTE is the worlds fourth
largest mobile phone manufacturer measured by unit sales.
Furthermore, it is the worlds fth largest telecom equipment manufacturer measured by 2011 revenues. In 2012, ZTE became the fourth
largest smart phone manufacturer in the world.
ZTE started its internationalisation by selling its products in developing and threshold countries. It took its rst steps towards globalisation
when it established branches in Indonesia in 1996. In 1998, ZTE signed a
US$95 million contract, its rst big overseas project in Pakistan. The rst
engagement in Africa was in 1999, when the company deployed a video
conferencing system in Kenya. In 2001, ZTE considered its internationalisation strategy as a formal one and completely entered the overseas
market. The rst marketing sector was built in 2002, and the foreign
marketing strategy has become one of intensive cultivation rather than
extensive cultivation.
In 2003, ZTE became the largest code division multiple access
(CDMA) system provider to BSNL, Indias largest telecom services
company. The internationalisation strategy of the following years was
characterised by strategic partnerships especially in Europe and North
America. Better still, years of endeavours in the overseas market started
to pay back. The three growing product series and the overseas
296
expansion worked together to push the company to a higher level in

2003. On January 8, 2004, a board announcement showed a record of
new contract sales of RMB 25.19 billion in 2003, which represented a
yearly growth of 50 per cent. This included RMB 4.82 billion from the
mobile phone sales and represented a yearly increase of 66.3 per cent. It
also included US$610 million from the overseas market, which represented a yearly increase of more than 100 per cent.
In 2004, the overseas market for mobile phone and 3G was pointed
out as two principal growth sources of the company. Specically, the
company generated revenue of RMB 34 billion, of which RMB 13.6
billion (US$1.644 billion) was contributed by overseas contract sales,
up 169.5 per cent from the previous year, thus ushering in a new era in
the history of the company in building its international name. Over the
whole year, more than 10 million sets of mobile phones were sold, up
100 per cent from 2003. This year witnessed the tremendous growth of
the company in both international and domestic markets, proving that
it was the fastest growing and most robust telecom equipment manufacturer in China.
In 2005 ZTE began its internationalisation The company started
collaborations with Alcatel, Ericsson, France Telecom and Portugal
Telecom in 2005. Two years later, the company announced a strategic long-term partnership with France Telecom and signed a longterm contract with Canadas second largest telecommunications
operator Telus. To broaden the market abroad, ZTE considers
developing countries as its main market target, and Pakistan is the
rst localisation market after occupied domestic market position.
Then, ZTE entered Russia and the European market. In 2007,
it formed strategic partnerships with Vodafone, Telefonica and
Hutchison. The result of this continuous internationalisation
strategy is reected in ZTEs revenue stream, wherein international
operations accounted for 60 per cent of the companys total revenue
in 2011.
Moreover, ZTE rose to be the fourth largest smart phone manufacturer in the world during the third quarter of 2012. The overseas market
has become ZTEs strategic market. ZTE has already started partnerships with more than 150 operators in more than 60 countries around
the world. ZTE has denitely become the leader in Chinese multinational communication rms entering the international market.
297
11.6 Reverse learning and capability upgrading

Huawei and ZTE are now two leading multinational corporations in the
communication industry. When they started internationalisation and
going global, they had accumulated some technological capabilities in
the domestic market, but they were far from being industry leaders at that
time. Therefore, instead of beginning with core proprietary technologies
and exploiting them in the process of internationalisation, our study
reveals that they followed a technology and capability upgrading and
development process driven by reverse learning from foreign customers,
collaborators and subsidiaries. This section describes these processes.
Reverse learning from customers

Because information technology develops quickly, and customers expect
more choices and better services, any operations mechanisms must make
corresponding adjustments to improve operational efciency and remain
competitive. Therefore, customer-centric thinking has become the necessary method in the process of the reverse learning path.
Reverse learning
Huawei and ZTE represent the way in which a new entrant may challenge incumbents, especially foreign competitors, by developing resources
and capabilities that are especially adapted to the local market. On the
one hand, the payoffs from innovation activities are determined by
market processes that involve not only the activities of the innovator
but also the reactions of customers and competitors (Koellinge, 2007).
Huawei and ZTE strive to maintain their market-focused product innovations that proved hard for domestic competitors to replicate. Huawei
and ZTE have kept this strategy as they have extended their capabilities
into manufacturing and R&D; namely, a major objective of Huaweis
and ZTEs ongoing activities is to develop products that target specic
customer segments. This case clearly illustrates how the nature and
direction of business activities evolves in a rms accumulation of relevant
resources and capabilities. For example, every Huawei innovation stems
from close interaction with its customers to understand their needs and
market demands. Huawei annually invests an average of 10 per cent of its
annual revenue to R&D. In 2011, Huawei increased its R&D investment
to RMB 0.09 billion, equivalent to 13.7 per cent of sales in that year.
298
Currently, Huawei has 14 regional headquarters, 16 R&D centres, 28

joint innovation centres and 45 training centres worldwide. Customers
want more from Huawei; Huawei is a huge company, and customers tell
it what they want to buy, so what happens next is 4G, 5G, 6G. Therefore,
Huawei naturally progresses from being a fast follower to becoming the
leader and innovator.
The picky customer is an important motive to improve the standards
for the appearance, material and technical aspects of the product. Because
customers are both demanding and picky, Huawei has engaged in more
technology and process testing as it has searched for better solutions to
customer needs. For example, Huawei has produced a high-end waterproof phone, the Huawei Ascend D2. Through such a process, Huawei
has continuously pushed the frontiers of technological development.
Similarly, one of advantages of ZTE is satisfying customer needs
quickly. Through ZTEs 3G road show, it shows its customers ZTEs
latest R&D results and its capability in building up commercial 3G networks in Turkey. It is estimated that 250 customers attended the Turkish
leg of ZTEs road show on 22 and 23 March 2005 at the Crown Hotel,
Istanbul, including senior executives from the Turkish Transportation
Ministry, the Turkish Telecom Authority, Turk Telekom, Turk Cell,
Telsim and Avea. ZTE has established 14 ofces in Europe and has
made signicant steps into the European market.
Capability upgrading
The traditional value chain begins with the companys core competencies, its assets. It then moves to inputs and other raw materials, to an
offering, to the distribution channels, and then nally to the customer.
But the customer is the rst source of Huaweis and ZTEs learning,
with whom they have direct contact and provide their extensive solutions. The entire thought process has been reversed. Figure 11.5 shows
the process of capability upgrading beginning with the customer.
Customer
Priorities
Channels
Solutions
R&D
Inputs
Figure 11.5 The reverse learning chain from customers
Capability
Upgrading
299
In this reverse process of capability upgrading, understanding

customers demands and prioritising them underlies basic capability
learning. Next, according to customer demands, the rm builds capability by selecting and testing innovative ideas, product concepts and
product prototypes through improved channels, solutions and R&D
investment. In addition, in this process, management capability,
especially in controlling and managing the distribution network, and
some intangible assets including the brand name and rm reputation
have also been improved.
First, understanding what customers value can be quite complex and
is the rst demand on rms because customers select products, services
and providers of those products and services based in large part on how
well they create value for them. Managers or rms can gain some
understanding of customers through indirect means such as analysing
sales data and sales call reports or can gain the deepest insights from
direct interaction with customers from a research perspective, observing
their operations and meetings and listening openly to their views, their
market environment interpretations and their strategic discussions
through open-ended in-depth interviews. When Huawei UK and ZTE
entered the European market, their rst challenge was to establish
long-term customer relationships to understand subdivided markets
and diverse customer requirements and to monitor the market and
customer preferences. They were especially concerned with the package
of combining functional, service, and relationship benets as they relate
to market entry and market share. This kind of understanding of the
customer is based on direct and indirect communication with customers
and is the rst accumulated capability.
Second, selecting and testing innovative ideas, product concepts and
product prototypes make up the second capability accumulated by
multinational rms. For example, in China, people do not use a
technology product with point-point connection, but this is popular in
Europe; hence, Huawei developed and provided differentiated new
products for the European market that followed customers habits.
People in many countries such as in Africa like mobile Internet and do
not use xed telephony. When Huawei began to enter the African
market, it did not provide services very quickly. Instead, Huawei built
infrastructure because that was the real need in these countries. After
several years, Huawei became the dominant player in the communication industry market in Africa. This kind of capability of selecting
300
innovative product concepts is a form of capability learning through

totally understanding customers.
In addition, learning from the customer improves management capability, especially the capability of controlling and managing distribution
networks. Long-term customer relationships for understanding subdivided markets and diverse customer requirements and monitoring the
situation in a market and customer preferences according to customer
requirements and competitive goals can lead to enhanced ability to
control and manage a distribution network.
Finally, rms intangible assets can obviously be improved after
providing good service and technological assistance, such as improving
the brand name and rm reputation, shortening the time from trial
manufacturing to commercialising and formatting the virtuous circle
in communication with dominant customers and suppliers.
Reverse learning from co-operation

In recent years, even the largest innovative organisations have not been
able to rely solely on internal resources; they also acquire knowledge
and capability from external sources. Therefore, in the innovation and
global deployment phase, Huawei and ZTE gained external markets
and technology capability through alliances, collaboration and joint
R&D centres as external sources can facilitate the building of innovation capability.
Reverse learning
Huawei has focused on expanding its mobile technology and networking solutions through a number of partnerships while it has engaged in
internationalisation (see Table 11.3).
In 1993, Huawei achieved its primary threshold of knowledge accumulation on PBX and successfully made the rst breakthrough in
C&C08 digital telephone switch by effectively taking advantage of the
technology diffusions from Shanghai Bell (the rst Sino-foreign joint
venture in China). Huawei thereafter successfully monopolised the
Chinese rural market and small cities within China. Afterwards, with
higher product quality and improved product development, Huawei
started to compete with foreign enterprises in the Chinese urban market.
From 1998 to 2003, Huaweis management style evolved through
the inuence of IBM, especially in the eld of R&D and supply-chain
301
Table 11.3 Huaweis history of alliances

Date
Partners
Country
19891994
Shanghai Bell Telephone Manufacturing

Company Sino-foreign joint venture
Texas Instruments
BETO TELEKOM
Fujian Provincial Mobile
Qualcomm
NEC, Matsushita Communications
Agere System
Microsoft
NEC, Matsushita Communications
3Com
Avici Systems
Siemens Info& Common Mobile
Inneon
Nokia Siemens Networks
Intel
HP
Freescale Semiconductor
Motorola
Qualcomm
Global Marine Systems
Symantec
International Telecommunication Union
IBM
Microsoft
Inneon
Sun Microsystems
U.S. JV
20 Feb. 97
9 Apr. 97
26 Aug. 99
8 Jun. 00
27 Nov. 01
21 Oct. 02
21 Oct. 02
23 Oct. 02
19 Mar. 03
4 Jun. 03
29 Aug.03
16 Sep. 03
12 Feb. 04
25 Apr. 05
2 Mar. 06
31 May. 06
25 Jul. 06
13 Feb. 07
14 May. 07
21 May. 07
31 Oct. 07
19982003
12 Dec. 08
20 Mar. 09
9 Aug. 09
U.S.
Russia
China
U.S.
Japan
U.S.
U.S.
Japan
U.S.
U.S.
Germany
Germany
Germany
U.S.
U.S.
U.S.
U.S.
U.S.
UK
U.S.
Organisation
U.S.
U.S.
Germany
U.S.
Source: Zhang (2009); Nakai and Tanaka (2010); Huawei interviews in Shenzhen and
Reading (UK); http://www.huawei.com/en
management. It reorganised as an IT solution vendor from a mere

hardware wholesale company. It survived the collapse of the IT bubble
at the beginning of the 2000s, with advice from IBM consultants,
increasing prots through better supply-chain management, stronger
R&D and a more integrated corporate structure (Nakai and Tanaka,
2010).
302
In March 2003, Huawei and 3 Com Corporation formed a joint

venture company, 3Com-Huawei (H3C), which focused on the R&D,
production and sales of data-networking products. The company later
divested a 49 per cent stake in H3C for US$880 million in 2006.
In 2005, Huawei began a joint venture with Siemens, called TD Tech,
for developing 3G/TD-SCDMA mobile communication technology
products, and Huawei established a Shanghai-based joint R&D centre
with Motorola to develop UMTS technologies and a joint venture with
Telecom Venezuela, called Industria Electronica Orinoquia, for research
and development and the sale of telecommunications terminals after
2006 (Orinoco, 2010). In 2008, Huawei launched a joint venture with
a UK-based marine engineering company, Global Marine Systems, to
deliver undersea network equipment and related services (Wang, 2008).
Similarly, ZTEs internationalisation strategy was characterised by
strategic partnerships especially in Europe and North America. ZTE
started collaborations with Alcatel, Ericsson, France Telecom and
Portugal Telecom in 2005. Two years later, the company announced
a strategic long-term partnership with France Telecom and signed a
long-term contract with Canadas second largest telecommunications
operator Telus. In 2007, ZTE formed strategic partnerships with
Vodafone, Telefonica and Hutchison. The result of this continuous
internationalisation strategy is reected in ZTEs revenue stream
international operations accounted for 60 per cent of the companys
total revenue in 2011.
As a latecomer in the global knowledge economy, China can obtain
the critical resources and capabilities to move from the position of
late-follower to that of rapid-follower, or even leader through different
internationalisation routes (Child and Rodrigues, 2005; Deng, 2007).
One important route is the contract manufacturer/JVs (Child and
Rodrigues, 2005): many mainland Chinese companies, such as Huawei
and ZTE, choose to co-operate with foreign MNCs through JVs, contract
manufacturing or technology licensing. They gradually learn the
technologies or capabilities they need and move up the value chain.
First, building the global brand via collaborations with local famous
rms is a shortcut to upgrade rms reputations and competitive power.
As we discussed, Huawei set up many joint R&D labs with renowned
rms such as Texas Instruments and IBM, and created strategic
303
alliances with multinational giants such as Siemens and 3COM. In the

joint venture with 3COM, it learned about innovative marketing strategies and branding, for example, by selling under the Huawei-3COM
name to improve and diffuse its own brand. Huawei has already moved
well up the value chain of a typical original equipment manufacturer,
but to achieve the pricing power and prots of a Sony or a Nike, the
missing part of the jigsaw is the brand. Building a global brand takes
time and is expensive. Thus, collaborative branding proved successful;
in 2005, Huaweis contract sales were already US$8.2 billion, of which
nearly 58 per cent came from foreign markets.
Before 1991, ZTE only developed low-end systems based on traditional chip production methods. The specialised chips that supported
large capacity and complex systems were almost inaccessible to the
rm. For a long time, this low-price route was successful in the Chinese
market when local brands experienced their most prosperous period,
with local mobile phone manufacturers Ningbo Bird, Amoi, TCL, China
Kejian and Konka evenly matching their foreign counterparts, including
Nokia and Motorola. At that time, almost all top-selling mobile phones
were made by contract manufacturers of low-end products. These
brands, however, could only resort to low-price strategies to gain a
share of the Chinese cell phone market. When foreign mobile phone
producers also adopted medium- and low-price strategies, local mobile
phone brands lost large numbers of manufacturing contracts. Facing a
tough price war between 1995 and 1998, ZTE experienced difcult
times. For comparable products, ZTE prices were estimated to be around
25 per cent lower than those of Western rivals, but lacking in innovation
to distinguish themselves from other companies. However, continuous
effort on the technical innovation front has proved to be extremely
fruitful. The strategic alliance with international communication leaders
Alcatel and Nortel laid a rm foundation for ZTEs goal to strive to
build-up CDMA as a number one global brand. ZTE now has 14 wholly
owned R&D centres across the United States, Europe and Asia.
Second, rms build up competence through taking advantage of technological complementarity among different collaborators. Zhengfei Ren,
the president of Huawei, said that the goal of co-operation is to explore
new technology and corporate capability. Targets include quickly transferring technology into products and sharing high-end technologies
when Huawei collaborates with other rms in the communication
industry that have some advantages in the specic aspects, such as
304
rms providing retail services that are more familiar with markets, as
well as with companies that operate in network development layers and
are more powerful in regulation and infrastructure. Both Huawei and
ZTE have regular exchange meetings with their collaborative institutions, through which R&D input, data transferring, product exploration
and exploitation and appearance design capability have been improved
the most.
Third, collaboration is helpful to rms growth, in both technology
and managements. For example, UK companies have a long history of
strong product demand and rigorous quality and health-safety requirements. This has enabled ZTE to develop in these areas when it entered
the UK market through continuous contact with local rms. ZTE UK
not only learned how to regulate its operation management but also
received feedbacks that are related to the headquarters. Therefore,
technology and management capability for subsidiaries and headquarters is the third channel of capability upgrading.
Project management is also an important capability for rms to learn
from the collaborative experiences. Huawei and ZTE came to the consistent conclusion that establishing project targets, phases standards
and project managing regulations and adjusting organisation structure
exibly in the new innovation projects are important capabilities accumulated from previous collaborations.
Other corporate capabilities have been accumulated from collaboration practices, such as facilitating communication among R&D personnel, harmonising basic research, developing and commercialising,
promoting information ow and interconnection between different
function departments and perfecting production regulations and
systems.
Learning from subsidiaries

Learning from subsidiaries is the third characteristic for both Huawei
and ZTE. International marketing involves recognising that people all
over the world have different needs. Huawei and ZTE have brands that
are now recognised across the globe. While many of the products that
these businesses sell are targeted at a global audience using a consistent
marketing mix, it is also necessary to understand regional differences,
hence the importance of international marketing. They must accept that
differences in values, culture, languages and currencies will mean that
305
some products will suit only certain countries even within global markets. For example, there are important regional differences in cell phone
colour, wire colour and infrastructure x trial criteria for different
countries. Huawei produced the high-end waterproof phone Huawei
Ascend D2 within the European market environment where stringent
standards for the appearance of the product, material and technical
aspects; more technology and process testing; and picky customers
ensured that Huawei had to nd a better solution. Managers at ZTE
also point out that document management in UK rms is an important
aspect of the ZTE learning process.
Whats more, learning from each other through different subsidiaries
is another interesting learning activity. For Huawei, the UK is the rst
country (in Europe) that it entered; it also learned a great deal from
developing countries and about technology by establishing a relationship between Africa and the UK.
The UK is the rst developed country in which ZTE developed its
brand and enhanced its reputation to be able to respond to customer
needs. ZTE provides an interesting case about learning among different
subsidiaries. It met huge obstacles when it sold 3G phones to Africa
because of lack of charging power. To solve this question, it developed
new energy resources such as a storage battery utilising solar energy,
and this technology provides huge market prot not only in the African
market but also worldwide.
Third, learning a new operating vision in developed markets, such as
social responsibility, is an obvious advantage. Huawei met signicant
barriers in the United States; the U.S. government considered it untrustworthy. As a consequence, Huawei is extremely careful to obey local
laws in the country it is operating in, to accept social responsibility for
its actions and to promote good collaboration with local partners. All of
this will be good for Huawei in its future business activities. Similarly,
in the case of ZTE, advanced operational thinking is regarded as a
shortcut for operating its business in the future.
The headquarters and knowledge integration and diffusion
In the process of learning from subsidiaries, the headquarters serve as a
knowledge gatekeeper, integrator and mediator between subsidiaries.
The subsidiaries are the rst unit that creates or acquires a new technology. The parent company then acts as the mediating unit that collects
and selects knowledge transferred back to the headquarter, integrates it
306
with the companys existing capabilities and then disseminates the new
or integrated knowledge to other subsidiaries of the company group.
Finally, other subsidiaries adopt the new knowledge and technology
diffused from the headquarters and sometimes from regional headquarters. In this process, technology- or the knowledge-providing unit (a
subsidiary) is a point, and the headquarters is the hub. Huawei has a
capabilities centre in its headquarters in Shenzhen. This centre is responsible for (1) assessing the capabilities of the company and identifying
areas to improve; 2) collecting knowledge and information from its
subsidiaries all over the world, identifying the relevant knowledge that
is needed by the company and integrating it with the companys existing
technologies; and (3) diffusing the newly acquired or integrated knowledge to the departments or subsidiaries that are in need of it. The
regional headquarters of Huawei also serves to collect and diffuse
relevant new technologies or other non-technological knowledge to
other subsidiaries in the region directly. This function of the regional
headquarters also allows the reverse transfer and re-diffusion to be
faster and more efcient.
Most Chinese rms are deeply embedded in their institutional context.
Huawei and ZTE also have a deep understanding of their local markets,
including knowledge about customer preferences, adaptation requirements and price sensitivity. Changing rules of global competition
require Huawei and ZTE to constantly update their resource base and
explore new market space. Overseas markets are there to be acquired,
new technology and capability have to be absorbed and upgraded and
subsidiaries are the units that rst acquire technology; hence, communication among subsidiaries is helpful to capability improvement.
A rms long-term development strategy and management philosophy will obviously be improved based on learning from subsidiaries in
developed areas. Strategic capability normally includes connections
between technological strategy and business strategy, advanced decision systems, entrepreneurial spirit and intense innovation environment, adjusting innovation strategy accordingly and so on.
Developed markets look for something different from developing
countries; in the process, they become the market leader. Developing
countries look for solutions and provide services into their respective
countries. Both Huawei and ZTE said that they have learned some
307
advanced management ideas from the European market, such as healthsafety issues and corporate social responsibility, and this is a developing
trend for the Chinese headquarters and other subsidiaries. An investment
philosophy exists amongst most rms in Europe, whereas few rms in
China seem to adopt such an outlook. Therefore, the rms long-term
development strategy regarding management capability will be improved
when the subsidiaries in developed countries have absorbed and shared
this investment philosophy with other subsidiaries.
Both Huawei and ZTE also illustrate a rms accumulation of
relevant resources and capabilities in business operations because the
subsidiary in a new market is a new project or a wholly new entrepreneurship. Comparing developed countries to developing countries,
different subsidiaries in different countries have different production
lines, which does not imply that one is better than the other; it is just
different. The African market, which researches technology, presents a
different problem; a company needs to try solutions and then get
differentiated products. It is worthwhile to note that most of this
behaviour is learned from other subsidiaries in developed countries,
sometimes just a kind of imitation; hence, the ability of the China
mainland to re-innovate when facing the international market was
obviously improved. Huawei and ZTE also learned that subsidiaries
will feed back accumulated experiences and capabilities to headquarters, such as establishing project targets, phases standards and project
managing regulations and adjusting organisation structure exibly.
Therefore, this is the second contribution to learning among subsidiaries to accumulate business operation abilities.
Cultivating learning consciousness and investing in learning comprise
a kind of learning path combining unconscious absorption and planned
learning behaviours in the connection with subsidiaries in different submarkets. On the one hand, as we have already discussed, subsidiaries
provide technology or knowledge in the process of learning. They
transfer their information back to headquarters and achieve information exchange via the network of branches and subsidiaries led by the
headquarters. These are planned learning behaviours. On the other
hand, some informal information ow and interconnection between
different functional departments create opportunities for unconscious
learning, absorption and capability upgrading.
Besides the previously discussed aspects of capability upgrading,
companies management capability in human resource management,
308
total quality management and organisational capability of benchmark

system are also upgraded. Interviews with subsidiary managers of both
Huawei and ZTE also highlighted the gains in these aspects.
11.7 Conclusions
This study of Huawei and ZTE has several important ndings regarding
the relationship between a latecomer rms internationalisation strategy
and market expansion on the one hand, and its capability upgrading on
a particular learning path on the other. First, reverse learning generates
revenues from foreign customers, collaborators and subsidiaries.
Learning from these sources improves an MNEs capability in several
areas such as understanding customers needs, selecting and testing
innovation ideas, transforming customer needs and innovative ideas
into product concepts and product prototypes, controlling and managing the distribution network, business operations management, building global brand via collaboration, building competence based on
complementarities amongst collaborators and management philosophy
and cultivating learning consciousness. It also has signicant effect on a
rms long-term development strategy and the growth of its technology
and management capabilities. Therefore, the gains from internationalisation are the upgrading of the overall capabilities of these rms in
addition to technology upgrading as normally expected.
In addition, these two cases clearly illustrate that learning may have
its root in a project or in a collaboration, whereas capability upgrading
based on it may take place in the headquarters and through it also in
other subsidiaries of the company group. For example, in the case of
ZTE, the obstacle of lack of pricing power in the African market
induced the development of new energy resources such as a storage
battery using solar energy. This innovative technology provides ZTE
signicant prot not only in the African market but also worldwide. For
Huawei, the breakthrough in the Netherlands has contributed signicantly to the companys success in the continental European market.
Experiences from the Netherlands served as a model for the European
market, and the company learned and built up the capability to communicate with government, operators and customers and how to price
their products and services.
The two case studies illustrate an alternative path, namely reverse
learning, for a new entrant to build corporate capability. This learning
309
path includes several branches, including learning from customerrelated activities in marketing, sales and service; learning through collaboration with other rms, institutions and R&D centres; and also
learning from headquarters and other subsidiaries in the company
group. In each learning branch, specic learning activities lead to
upgrading in relevant capability areas.
Finally, the two cases examined also represent the way in which a new
entrant may challenge incumbents by developing resources and capabilities that are especially adapted to the local market. Huawei and ZTE
accumulated customer knowledge and created the strong customerpriority solution department that is very difcult for their Western
competitors to achieve because often in-house R&D leads the direction
of production and operation in these companies instead of customer
needs. They have continued with this customer-priority solution strategy even when their competitive advantage has moved from marketing
to manufacturing and R&D. A major objective of their ongoing activities is to develop products that are even more nely attuned to increasingly more specic customer segments. This consistent focus and
deepening capability in this regard have also emerged as a signicant
competitive advantage for them in the domestic and foreign markets.
Findings from the ZTE and Huawei cases have clear implications for
management. The two companies experiences in capability building
through reverse learning suggest to managers that they can build up
technological and overall capabilities of rms through operation in
foreign countries and acquire knowledge and capabilities through subsidiaries abroad. They should clearly link their existing set of resources
and capabilities in headquarters and subsidiaries to desired changes in
those features that they see as necessary to enhance competitiveness.
Furthermore, the customers are the primary source of Huaweis and
ZTEs learning. It is with the customers that Huawei and ZTE had
direct contact through their extensive distribution network. Innovating
and upgrading in response to highly demanding customers have become
major drivers of capabilities upgrading in these rms. Huaweis and
ZTEs rst success can arguably be attributed to their efforts to master
and understand customer needs before investing signicantly in R&D.
The case also shows how a rms development of new capabilities
requires different structures for learning. A rm can acquire different
capabilities through different means, for example, through communicating with customers, collaborating with a partner and learning
310
among subsidiaries. Furthermore, as a rm develops capabilities in

new functional areas or sub-markets or broadens the range of capabilities in a particular function, the organisation must be restructured
to support effective and efcient coordination of increasingly diverse
activities and improve capability to ensure new activities.
To note, although the research shed important insights on the benets
in learning alongside rms internationalisation process, considering
the exploratory nature of this study based on two cases, one needs to
be cautious in generalising the research ndings of this study. The
conclusions need to be validated with further research.
Appendix 11.1 Interviews coverage

Date
Name and Title
Topics
Chao Ai,VP Technology collaboration Huaweis global technology

Haibo Lin, head of technology
collaboration and knowledge
collaboration
integration
24 October 2013 Don Kelly, deputy COO at Huawei
Huawei UKs strategy and
Technologies
product solution paths and
strategy
24 Oct. 2013
Elton Yuxuan Huang, service director, Market entry strategy,
UK& I
technology-upgrading
strategy
25 Feb. 2013
Yang Shao, CMO at Huawei
Stand out with the Samsung/
Technologies
Apple competition
17 Jan. 2014
Jack Guo,
ZTEs strategy, production
senior product manager at ZTE
solutions, technology(UK) Limited
upgrading strategies
16 Oct. 2013
Stefano Cantarelli, CFO for network Channels of building
and carrier business units
technological capability
7 Feb. 2013
CEO of Huawei Technologies UK
Huawei story
23 June 2014
7 Feb. 2013
CEO of Huawei Technologies UK
Internationalisation and
innovation
Resources
Case interview in Shenzhen sponsored
by the British Academy project
Case interview sponsored by the project
http://phone-news-today.com/archives/
1239
Case interview sponsored by the

project
Oxford Department of International
Development (ODID) and the TMD
Centre co-hosted the Distinguished
Guest Lecture, the lecture organised
by a professor of this project
An informal discussion between
professors of this project and Victor
Zhang
Appendix 11.1 (cont.)

Date
Name and Title
20 Nov. 2011
Richard Ye Lihe, ZTEs senior director Internationalisation

Interview from the Mobile Asia
of wireless product operation,
distribution and rm strategy
Congress 2011.
product R&D system
The video can be accessed from
http://www.youtube.com/watch?
v=96htGIAbqs4
Lixin Cheng, CEO of ZTE (USA) Inc. Technology preference,
2012 International Consumer
development strategy in the
Electronics Show (CES). The video
coming year
can be accessed from
http://www.youtube.com/watch?
v=yGCz_rf1UxQ
Edward Zhou, CMO, Huawei
The introduction about Huawei Cambridge Wireless, 2011. The video
Technologies
UK, products, localisation
can be accessed from http://www.
and so on
youtube.com/watch?v=tTFgh
IAo5o4
James Lockett, vice-president, Huawei Does innovation foster trade or WTO Public Forum TV 2013. The
Technologies Co., Ltd.
does trade foster innovation?
video can be accessed from http://
www.tvballa.com/video-gallery/H
uawei-Technology#erfI2g9IDem
IKmrd.99
Zhu Shi, president
The background behind Exotic ZTE Malaysia GSMA Board Meeting
story of ZTEs growth
interview. The video can be accessed
from http://www.youtube.com/watc
h?v=dUjkdInrvY0#t=36
12 Jan. 2012
15 July 2011
2 Oct. 2013
1 Aug. 2011
Topics
Resources
7 March 2012
Javier Lopez, sales manager
New technologies and new

productions
15 Oct. 2012
Hu Xue Mei, CMO, ZTE, MEA
Industry change, opportunities

and challenges
14 May 2013
Zhengfei, Ren
Huawei founder gives rst

media interview
27 May 2013
Lixin Cheng, CEO of ZTE USA
Lixin Cheng on ZTEs U.S.

future at CTIA 2013
Mobile World Congress 2012.

Interview by Total Television. The
video can be accessed from http://
www.youtube.com/watch?
v=3KAJS2KKltY
ITU TELECOM WORLD 2012,
Dubai. The video can be accessed
from http://www.youtube.com/watc
h?v=Uo_iuqSaEiQ
Microblog Buzz, China. The video can
be accessed from http://www.you
tube.com/watch?v=jBgYTw8mU68
The Engadget Interview. The video can
be accessed from http://www.engad
get.com/2013/05/27/the-engadget-in
terview-lixin-cheng-on-ztes-us-futur
e-at-ctia-2
12
International collaboration and

radical innovation
12.1 Introduction
For latecomer economies, the acquisition of advanced knowledge
through imports, licensing and attraction of inward foreign direct investment is the most frequently used conventional channel for international
knowledge acquisition (Fu, Pietrobelli and Soete, 2011). China is no
exception in this regard. All these knowledge diffusion mechanisms are
followed by acquisition, assimilation and adaptation and hence are
closely linked to imitative innovation in terms of producing products or
using processes that are new to the country but not new to the world. In
recent years, in its pursuit of the transition from imitator to innovator,
China has increasingly employed various unconventional international
knowledge-sourcing mechanisms. These include the attraction of highly
skilled returning migrants, outward direct investment in advanced economies and international innovation collaboration to create or co-create
new knowledge. Will China become a frontier innovator through unconventional international knowledge sourcing? With internationalisation of
innovation and changes in the innovation mode from closed to open,
innovation is increasingly a collaborative and global undertaking.
Therefore, in this chapter we explore the role of international innovation
collaboration in Chinas transformation from imitator to innovator.
The rest of the chapter is organised as follows: Section 12.2 briey
overviews the patterns and trends in innovation collaboration amongst
Chinese rms. Section 12.4 presents the empirical results. Section 12.5
provides conclusions.
12.2 International collaboration and radical innovation:

the literature
The transformation from imitation to innovation requires increasing
novel or radical innovation that has greater novelty than the
314
315
diffusionary, imitative innovations that China mainly engaged with in

the takeoff and catch-up stages. Radical innovations represent major
departures from existing practices and involve the disruptive creation of
new insights (Ettlie, 1983). Accordingly, the launch of radical innovation requires an extension of both the depth and breadth of knowledge
(Zhou and Li, 2012). The high requirements to launch radical innovation lead rms to acquire knowledge not only from internal R&D but
also from external learning and collaboration (Birkinshaw, Bessant and
Delbridge, 2007; Bao, Chen and Zhou, 2012).
One of the mechanisms to tap into external innovation resources and
capacities is collaboration. External linkages, both public (including
universities) and private, benet rms innovation (Freeman and Soete,
1997). Through collaboration, rms can expand the range of expertise,
strengthen R&D capacities, acquire complementary resources and
enhance commercialisation capacities. Collaborating with other rms
may not only help a rm acquire knowledge from outside its boundary
but also combine different sources of knowledge that could be useful in
exploring uncertain worlds (Belderbos, Carree and Lokshin, 2004). By
being involved in collaborative relationships, a rm could open up gateways to new knowledge that departs from existing organizational memory (Bao et al., 2012) and obtain the opportunity to synergistically
integrate partners capabilities and resources. Moreover, heterogeneity
in technological knowledge and difference in market domains between
partners ensure diversity of inter-rm knowledge portfolios (Prabhu,
Chandy and Ellis, 2005), increasing the possibility of the development
of new architecture or functions. Collaborative innovation is of particular importance for radical innovation because it requires acquisition of
knowledge not only from internal R&D but also from external learning
and collaboration (Birkinshaw et al., 2007; Bao et al., 2012). Therefore,
companies should not innovate in isolation but cooperate with external
partners throughout the innovation process (OECD, 2008).
Innovation has increasingly become internationalised. Firms go across
the border for either exploitation or exploration of knowledge (Archibugi
and Iammarino, 2002). The globalisation of innovation has three components: (1) the international exploitation of nationally generated innovation, (2) the global generation of innovations by MNEs and (3) global
techno-scientic collaboration (Archibugi and Iammarino, 2002). These
activities can be either asset-exploiting R&D or asset-augmenting activity
(Dunning and Narula, 1995). In addition to MNE-led global innovation
316
generation, strategic technology partnering (STP) is argued to be a key

strategy complementary to internal R&D-based innovation (Narula and
Zanfei, 2004). This stream of literature nds that large rms tend to enter
into STP because of their strong technological capabilities and absorptive
capacity. This enables these rms to keep up with the technological
frontier (Cantwell, 1995). On the other hand, Narula (2002) nds that
small rms rely on non-internal sources because of their limitations in
managerial, nancial and knowledge resources. Hence, international
technology collaboration has changed from technology transfer to organisational learning (Niosi, 1999). Non-equity STPs can be a useful tool as
they are more exible and more suitable for knowledge development and
learning.
The wave of innovation internationalisation and the rise of the
emerging economies have induced a large number of R&D outsourcing
activities from the United States and EU in the emerging economies
(EEs) in particular, China and India. This encourages innovation
collaboration between the developed countries such as the UK and the
EEs. Moreover, the emerging economies, especially China and India,
have invested heavily in R&D and both have reservoirs of low-cost but
skilled labour (Fu et al., 2011). This also makes them attractive partners
for international innovation collaboration.
Despite the heavy investment in R&D and the large pool of educated
scientists and engineers, studies on middle-income country rms in
South East Asia engaging in radical innovation at the technology frontier suggest that rms from developing and emerging economies are
normally constrained by their level of technological capabilities and
human resources to achieve world-leading innovation performance
(e.g., Hobday, Rush and Bessant, 2004; Dantas and Bell, 2009).
Innovation through international collaboration is likely to break these
constraints. Evidence from Chapter 8 suggests that Chinese rms
engaged in innovation collaboration with universities in the newly
industrialised economies are more likely to produce novel innovations.
Therefore, international innovation collaboration may nurture more
radical innovation and change the technology trajectory in China.
12.3 Collaborative innovation in China: an overview

Using the national innovation survey carried out in 2008 by the
National Bureau of Statistics of China and Tsinghua University, we
317
Table 12.1 Collaborative innovation activities

Firms that have joined
collaborative innovation
Firms that have not joined

collaborative innovation
Firm
ownership
No. of
rms
Proportion
(percentage)
No. of
rms
Proportion
(percentage)
State-owned
Private
Foreign
invested
Total
64
443
161
4.6
32.0
11.6
58
399
260
4.2
28.8
18.8
668
48.2
717
51.8
Source: 2008 China National Innovation Survey by Tsinghua University and NBS.
have an overview of the pattern of innovation collaboration in Chinese

rms. As explained in Chapter 8, this sample is biased towards innovative rms. Therefore, the results may overestimate the actual breadth
and depth of innovation in the population of Chinese rms.
As Table 12.1 indicates, around 48 per cent of sampled Chinese rms
have engaged in collaborative innovation; 52 per cent have not. This is
on par with the share of rms engaged in collaborative innovation in the
United Kingdom from 2008 to 2010 (BIS, 2013). Interestingly, in the
domestic sector, a majority of state-owned (SOEs) and privately owned
rms (POEs) engaged in collaborative innovation: 64 SOEs versus
58 SOEs, and 443 POEs versus 399 POEs. In contrast, with respect to
foreign invested rms, a majority have not taken part in collaboration,
which could be indicative of their relative reluctance towards networking with local rms.
Firms do collaborate with various partners including other rms
within their company group; suppliers of equipment, raw material or
software; customers; competitors; consultants or private R&D institutes; and universities or public R&D research institutions. As
Table 12.2 reports, universities and public research institutions are the
sector that registered the largest number of collaborations, mainly
because of the high number of collaborations with domestic universities
and research institutions. Suppliers and customers also have seen a large
number of innovation collaborations. In particular, collaboration with
foreign customers is the most popular pattern of international collaboration with 187 cases registered in this category, followed by foreign
318
Table 12.2 Types of collaborators and regional distribution
Types of
collaborators
Other rms
within
afliated
group
Suppliers of
equipment,
raw material
or software
Users or
consumers
Competitors or
other rms
within the
same industry
Consultants or
private R&D
institutes
Universities or
public R&D
institutes
Collaboration
Collaboration with
Total number
with domestic international of
Do not
partners
partners
collaborations collaborate
315
119
434
223
316
162
478
204
291
187
478
217
222
106
328
318
223
48
271
362
473
35
508
157

Source: 2008 China National Innovation Survey by Tsinghua University and NBS.
suppliers, which also reported 162 cases of international collaboration.

With regard to collaboration with other partners, we still see a relatively
high proportion in comparison to internationally leading innovative
nations such as the UK in the same collaborative category.
12.4 Collaborative innovation and Chinas transformation

from imitator to innovator: empirical evidence
Using the China 2008 National Innovation Survey and similar methods
used in the analysis of the role of universities in Chinas industrial
319
Table 12.3 The impact of collaboration on innovation performance

in Chinese rms: Tobit model estimation results
1
Co_international
Co_domestic
Ln(inhouse R&D)
Ln(extramural R&D)
Age
Size4
Constraints in human capital
Coastal region
Foreign invested rm
Industry dummies
Constant
Observations
F statistics
LogLikelihood
Novel new sales
Imitative new sales
16.13***
(5.178)
1.277
(5.728)
1.326
(0.903)
1.515**
(0.642)
0.132
(0.157)
1.019
(5.57)
2.361
(5.13)
2.644
(4.892)
15.24***
(5.158)
yes
yes
819
4.476
1291
1.819
(3.610)
7.565*
(3.921)
2.776***
(0.590)
0.074
(0.462)
0.066
(0.062)
5.425
(3.868)
4.431
(3.913)
6.406*
(3.533)
5.787
(3.934)
yes
Yes
804
4.358
3320
Note: Robust standard errors in parentheses; *** p<0.01, ** p<0.05, * p<0.1.
innovation (details discussed in Chapter 8), this section presents the

empirical test results on the impact of international innovation collaboration on industrial innovation in China. As Table 12.3 reveals, innovation collaboration with foreign partners has had a positive and
signicant effect on novel innovation in rms. On average, the percentage of sales of products that are new to the world in rms that have
collaborated with international partners is 16 per cent higher than that
of those rms that have not. The estimated coefcient is statistically
signicant at the 1 per cent level. Not surprisingly, collaboration with
320
domestic partners does not exert a signicant effect on novel innovation

in Chinese manufacturing rms. The estimated coefcient of the extramural R&D variable is positive and statistically signicant at the 5 per
cent level, suggesting engaging with extramural R&D contributes to the
creation of radical innovation in China. Foreign rms are more creative
than domestic rms. The average share of sales on account of novel
innovation in foreign invested rms is 15 per cent higher than that of
domestic rms.
In contrast, collaboration with domestic partners appears to have a
positive impact on the generation of imitative diffusionary innovation
in China, although the estimated coefcient is only marginally signicant at the 10 per cent level. The main contributor to imitative innovation in China is in-house R&D activities. The estimated coefcient is
positive and statistically signicant at the 1 per cent level, suggesting
that a 1 per cent increase in in-house R&D spending increases the
percentage of new sales by 0.12 per cent. Firms in coastal regions
appear to have less imitative innovation than that of the rms in the
inland region. But the estimated coefcient is only marginally signicant at the 10 per cent level. The difference between the foreign and
domestically owned rms does not appear to be signicant in this
regard.
Distinguishing the collaborative partners according to the type of
their external partners, Table 12.4 reports the estimated results of
their impact on novel innovation. We have controlled for collaboration with domestic partners and other foreign partners to single out
the effect of the particular type of collaboration under concern. Except
for collaboration with partners within the company group, collaboration with all other types of foreign partners appears to have a
signicant positive effect on the creation of novel innovation in
Chinese rms. Collaboration with foreign customers appears to have
the largest impact on novel innovation. The average share of sales on
account of novel products in rms that collaborated with foreign
customers appears to be 33 per cent higher than for those that did
not. Collaboration with the suppliers also appears to have a signicant
contribution to novel innovation. But the size of the effect and the
signicance of the effect are considerably smaller than those of collaboration with customers. Collaboration with foreign universities
appears to have the second largest benets. The average sales of
novel products in rms that collaborate with foreign universities are
Table 12.4 The impact of international collaboration on novel innovation in China: Tobit model estimation results
1
Dep Var: Novel new sales percentage

Co_within company group_intl
Co_supplier_international
Co_customer_international
Co_competitor_international
Co_private res. institute_intl
Co_University&Pub res. inst_intl
0.0731
(7.715)
4.225
(6.483)
25.28***
(5.618)
4.499
(6.56)
8.297
(8.693)
14.49
(9.926)
Co_international
Co_domestic
Ln(inhouse R&D)
Ln(extramural R&D)
7.037
(5.344)
1.083
(0.883)
1.509**
(0.646)
6.114
(7.305)
11.74*
(7.116)
33.34***
(7.215)
15.47**
(7.004)
19.32**
(8.372)
16.02***
(5.614)
4.806
(5.639)
1.129
(0.91)
1.569**
(0.649)
11.89*
(6.733)
5.451
(5.595)
1.17
(0.908)
1.541**
(0.648)
2.714
(7.326)
5.332
(5.511)
1.118
(0.892)
1.640**
(0.636)
12.19**
(6.026)
4.723
(5.601)
1.06
(0.908)
1.470**
(0.653)
15.60***
(5.479)
4.75
(5.631)
1.132
(0.917)
1.508**
(0.646)
23.54**
(9.212)
15.60***
(5.37)
4.525
(5.603)
1.11
(0.904)
1.504**
(0.646)
Table 12.4 (cont.)

1
0.176
(0.165)
2.558
(5.617)
2.253
(5.136)
Yes
Yes
819
4.816
1289
0.169
(0.163)
2.307
(5.627)
1.011
(5.174)
Yes
Yes
819
4.936
1289
Dep Var: Novel new sales percentage

Age
Size4
Lack human capital
Sector dummies
Constant
Observations
F statistics
Loglikelihood
0.148
(0.158)
1.317
(5.586)
2.256
(5.014)
Yes
Yes
819
5.362
1278
0.178
(0.166)
2.762
(5.63)
2.056
(5.174)
Yes
Yes
819
4.274
1291
0.176
(0.164)
2.726
(5.665)
2.100
(5.166)
Yes
Yes
819
4.376
1290
Note: Robust standard errors in parentheses; *** p<0.01, **p<0.05, *p<0.1.
Dependant variable: percentage of sales that are new to the world.
0.167
(0.162)
2.178
(5.569)
2.924
(5.025)
Yes
Yes
819
5.836
1280
0.188
(0.168)
2.81
(5.663)
1.769
(5.157)
Yes
Yes
819
4.71
1289
323
23.5 per cent higher than for those that did not. Collaborations with
foreign private research institutions and foreign rms in the same
industry all appear to have brought considerable gains to the rms.
The estimated coefcients are all statistically signicant at the 5 per
cent level.
12.5 Conclusions
Given the fact that innovation is increasingly a collaborative task and
that more and more rms are adopting international collaboration in
todays globalisation era, this chapter analyses the contribution of
international collaboration to Chinas transformation from imitator
to (radical) innovator. Whereas collaboration between domestic partners only appears to be marginally benecial to the production of
diffusionary innovation in China, collaborations with foreign partners are found to have made a signicant positive impact on the
creation of novel innovation in Chinese rms. The type of foreign
partners that Chinese rms may benet collaborating with covers a
wide range, including foreign customers, suppliers, universities, private research institutions and rms in the same industry.
Collaboration with foreign customers generates the largest benets
for the creation of novel innovation. Collaboration with foreign universities also proves to be fruitful for the generation of innovations
that are new to the world.
Findings from this research have profound policy implications. To
achieve the important second-stage catch-up and move the economy
to the high-income group is the main development objective of China
and many other middle-income countries. This movement to the
world frontier requires high-level creative ideas and talents that are
quantitatively and qualitatively different from those that are required
for imitative innovation. It requires an open mind and a strategy for
path-breaking innovation instead of path-following incremental
innovation. It requires vision in identifying strategic direction, heavy
investment for such long-term R&D activities, partners to share the
risks and the high-level and complex skills that are required to carry
out such groundbreaking innovations. International collaborative
innovation is therefore an effective mechanism to address the challenges for the path-breaking radical innovations that are necessary to
achieve and maintain international innovation leadership. This is, in
324
fact, a widely observed trend among the OECD countries, which have
introduced a series of programmes to encourage and support international collaboration. Results from this chapter conrm that it is an
effective channel that will bring signicant gains to Chinese rms for
the creation of novel radical innovation.
13
Innovation efciency and the

cross-country gap in innovation
13.1 Introduction
The gap in innovation between European Union (EU) countries and the
United States has been well documented. Publications by the European
Commission and the UK Department of Trade and Industry on international research and development (R&D) expenditure show that Europe
lags behind the United States and some Asian economies in R&D
investment (DTI, 2005; EC, 2005). Hall (2004) and other studies
have also suggested an increasing gap between Europe and the United
States. Numerous academic and government studies investigating the
factors affecting national innovation performance and quantifying
national innovative capacity have identied gaps between the OECD
economies and the United States as the lead economy in this respect
(e.g., Porter and Stern, 1999; EC, 2002, 2005; Furman, Porter and
Stern, 2002; Mairesse and Mohnen, 2002; DTI, 2003a, 2003b,
2003c, 2005; Faber and Hesen, 2004; Furman and Hayes, 2004;
Jaumotte and Pain, 2005).
One of the most inuential contributions to this literature has been
based on a particular notion of innovation capacity. This is dened as
the ability of a country as both a political and economic entity to
produce and commercialise the ow of new-to-the-world technologies
over the long term (Furman et al., 2002, p. 900). This notion of
innovative capacity is not the realised level of innovative output per
se, but reects more fundamental determinants of the innovation process (ibid.). This notion, operationalised by the authors as a production
function for international patents, is they argue readily captured by
a small number of observable factors that describe a countrys national
innovative capacity. These are the (1) common innovation infrastructure, including overall science and technology policy, basic research
support mechanisms, higher education and the stock of cumulative
technological knowledge; (2) specic innovation environments, such
325
326
as industrial clusters of the kind identied by Porter (1985) in his work

on the sources of national competitive advantage; and (3) linkages
between the common innovation infrastructure and the clusters inuenced, amongst other things, by the nature of the university system and
the nature of funding sources for new ventures linked to the particular
industrial clusters.
In most developments of this approach, innovation output is proxied
by USPTO patenting per capita. The contribution or weight of each of
the potential drivers is then derived from a regression analysis. This
analysis attempts to explain the cross-country patterns of patenting in
terms of proxy variables designed to capture infrastructure, cluster and
linkage variables. On the basis of these estimated weights and the value
of the underlying innovation drivers, countries may then be ranked in
terms of per capita national innovation capacity (for a review of developments using this approach see Gans and Hayes, 2008).
Other multivariate regression approaches to the analysis of crosscountry patterns of innovation outputs have used different estimation
procedures and have augmented or moved away from patents as a
proxy for innovation output. Jaumotte and Pain (2005), for example,
model the determinants of business sector R&D and then separately
model the determinants of patenting. They then model the determinants
of R&D employment and real wages in an attempt to assess the potentially adverse impacts that an increased demand for R&D employees
might have on the cost of R&D to the private sector. Faber and Hesen
(2004) and Mairesse and Mohnen (2002) replace or augment patents as
an indicator of innovation with more direct measures arising from the
European Community Harmonised Innovation Surveys. These more
direct variables include measures of the share of innovative products
in nal sales. Unfortunately there are no comparable data for the United
States or Japan or other non-EU countries that can use these more direct
measures of innovation, so broader OECD comparisons continue to be
primarily based on patenting data (Gans and Hayes, 2008).
Each of these approaches has provided important insights into crosscountry patenting and innovation performance differences. However,
they fail to distinguish explicitly between the extent to which innovation
output differences across countries result from the lack of innovation
input per se, and the efciency with which those inputs are converted
into innovation outputs. The innovation performance of an economy
depends not only on how much it invests in innovation (e.g., R&D
Innovation efciency and the cross-country gap in innovation
327
investment and personnel) but also on how efciently it manages the

innovation process and successfully transforms the innovation inputs
into useful innovation outputs, such as patents. With low innovation
efciency, an increase in investment in, say, R&D or the science base
may not produce the expected increase in outputs. The importance of
such inefciency is implicit in many policy discussions. The EC (1995)
Green Paper on Innovation, for instance, states that
Europe suffers from a paradox. Compared with the scientic performance of
its principal competitors, that of the EU is excellent, but . . . its technological
and commercial performance in high-technology sectors such as electronics
and information technologies has deteriorated. One of Europes major weaknesses lies in its inferiority in terms of transforming the results of technological
research and skills into innovations and competitive advantages.
Although the innovation capacity approach includes innovation system variables that one might expect to inuence the translation of inputs
into innovation outputs, it is not able to systematically distinguish and
account for the relative contribution of these system effects that inuence efciency compared to inputs per se. In this chapter, we attempt to
do this by decomposing each countrys gap in patenting compared to an
estimated world-patenting frontier into basic patenting capacity based
on innovative inputs and the efciency with which those inputs are
used. We employ the stochastic frontier analysis (SFA) approach to
estimate the world patenting frontier. SFA is a widely used standard
efciency estimation approach in productivity analysis, which we apply
in the new context of knowledge production.1 The advantage of the SFA
approach lies in that it explicitly decomposes the observed patent performance into two components: the patenting potential of a country
given the best practice use of inputs and an efciency gap relative to the
best practice patenting frontier. This is different from the traditional
total factor productivity (TFP) approach that denes productivity as the
residual of the patenting production function. In contrast, SFA takes
account of measurement error and decomposes a countrys deviation
from the frontier into inefciency and a random error.
An alternative approach to SFA in the evaluation of cross-country
patenting efciency would be to apply data envelopment analysis
1
For examples of the application in traditional productivity analysis, see Banker,

Datar and Rajan, 1987; Hoer and Payne, 1993; Lovell, 1996; and Hiebert, 2002.
328
(DEA). Using this approach, and with R&D capital stock and personnel
as inputs and patents and academic publications as innovation outputs,
Wang and Huang (2007) evaluate the relative efciency of the R&D
activities of 30 countries. They nd that fewer than one-half of the
countries are fully efcient in R&D activities. Similarly, Hollanders and
Esser (2007) carry out a DEA analysis using scores from the European
Innovation Scoreboard (EIS) for 2007. Their model includes three
categories of innovation inputs, namely innovation drivers, knowledge
creation and innovation and entrepreneurship, and two broad categories
of outputs that they term applications and intellectual property.
Applications include, inter alia, the CIS-based direct innovation measures
and indirect measures based on high-tech activity. Their study also shows
a variety of efciency in innovation across countries. Both studies focus
on the estimation of innovation efciency and report relatively little on
innovation capacity. Moreover, the DEA method attributes all the deviation from the frontier to inefciency (see, e.g., the discussions in Fu, 2012
and Fritsch and Slavtchev, 2007), whereas SFA has the advantage of
controlling for statistical noise in the estimation of innovative efciency.
Using USPTO patenting as our innovation output proxy, we dene a
countrys patenting capacity as its predicted patent output if it were
located on the estimated patenting frontier. We dene patenting efciency as the ratio of observed patenting value to estimated patenting at
the frontier. The use of patents allows direct comparability with the
previously cited major studies of comparative international patent performance. We recognise that the use of the USPTO imparts a country
bias when comparing any one countrys performance relative to that of
the United States, but there is no reason for this bias to vary across
countries when comparing each to the United States.
This chapter contributes to the literature as the rst attempt to apply
the SFA approach to the analysis of cross-country innovation activity
provided by patents. In addition, we attempt to allow for structural
differences between economies that may affect the incidence of patenting.
As patenting varies signicantly across industries, we include industry
structure as one of the explanatory variables affecting observed patenting
gaps. Given the pressure of home country patent data bias, we focus
relatively more on the comparison of non-U.S. economies with each other
relative to the United States. However, given the fact that the United
States is widely regarded as one of the leading economies in innovation,
we include it in the construction of the world patenting frontier.
329
The rest of the chapter is organised as follows: Section 13.2 discusses

the theoretical framework. Section 13.3 presents some stylised facts.
Section 13.4 discusses the model, data and methodology. Section 13.5
reports the estimated results. Section 13.6 benchmarks China against
the world innovation frontier using data covering the 2005 to 2011
period, and Section 13.7 presents conclusions.
13.2 Determinants of national innovation performance

In this chapter, we seek to identify sources of differences in national
innovation performance as proxied by patenting, in terms of differences in countries innovation systems. Following existing work in
this eld, we dene a national innovation system as a set of agents, the
institutional framework within which they operate and the networks
and interrelationships among them. The national systems approach
argues that taken together, these system elements determine national
innovation performance. The institutional framework and patterns of
relationships condition, and co-evolve with, the motivation and abilities of rms to invest in R&D and human capital and to develop and
commercialise product designs and service activities and manufacturing and service production processes that are new to them (Freeman,
1987; Lundvall, 1992; Nelson, 1993). The innovation performance
of a nation is determined not only by the quantity of human and
capital factor inputs into innovation but also by institutional and
other system factors. These factors constitute constraints and/or
incentives for innovation (Pavitt, 1980; Furman et al., 2002; Faber
and Hasen, 2004). They also inuence the direction and nature of
innovation decisions taken by rms and hence determine the efciency of the national innovation system (Mairesse and Mohnen,
2002).
Within an innovation system perspective, we seek to estimate the
innovation capacity of a country based on its basic innovation inputs
of overall R&D expenditure, the stock of research workers and educational expenditure. We then seek to explain distance from the frontier
(or innovation inefciency) in terms of system variables such as the scale
of the economy, the IPR regime, the availability of venture nance, the
breakdown of R&D between the public and private sectors and economic openness. Our approach as summarised in Figure 13.1 is similar
in spirit to the approach of Furman et al. (2002) and subsequent papers.
330
-Direct
innovation
inputs
-Financial
resources
-Human
resources
Basic
Innovation
Capacity
Innovation Efficiency
Actual
Innovation
Performance
Innovation environment
Development of clusters
Strength of university-business links
Protection of IPR
Availability of venture capital
Openness
Competition
Government support
Figure 13.1 National innovation system: a system theoretic framework
However, here we associate innovation capacity with the basic inputs

and attempt to distinguish and quantify the impact of system effects
explicitly, in translating those inputs into performance.
Determinants of basic innovation capacity

We consider R&D inputs and human capital resources as key input
factors affecting patent production. Investment in R&D is often found
to be a signicant determinant of innovation performance. Countries
that invest more in R&D are more likely to produce more patents and/
or new products and processes. This is of particular importance at the
country level, although R&D is neither a necessary nor a sufcient
condition for innovation in a rm (Baldwin, 1997). Labour force skills
and the availability of qualied scientists and engineers, particularly
qualied research staff directly involved in R&D activities, are widely
recognised critical factors that also contribute to a rms innovation
performance (Hoffman et al., 1998; Porter and Stern, 1999).
In addition to input factors per se, we also allow for variations in
sectoral structures that may affect the role that each may play at an
aggregate level. The particular technological or input requirements of
sectors and patterns of supplier or customer behaviour and innovation
331
strategy may lead to variations in the requirements for particular combinations of inputs and outputs. This leads to variations in the incidence
of patenting across sectors but also in underlying rates of R&D activity
and other measures of innovation input and output (Malerba,
2006). Variations across countries in the sectoral distribution of activity may therefore affect innovation performance of a country, especially its patenting performance. For example, it has been argued that
the EU-U.S. innovation gap is the result of such differences: compared
to the United States and Asia, Europe as a whole is under-represented
in fast-growing R&D-intensive sectors such as IT hardware, software
and electronics, and over-represented in protable but less innovative
areas such as food manufacturing, oil and utilities (Cookson, 2005,
pp. 302).
Determinants of innovation efciency

The actual innovation performance of nations is determined not just
by factor inputs but also by the efciency of the inherently recursive
process by which innovation opportunities are recognised, and the
necessary capabilities and inputs to commercialise new ideas and
appropriate market value put into practice. Innovation efciency
thus reects a nations intangible capability for transforming direct,
tangible innovation inputs into nal commercially successful innovation outputs. It explains the difference between actual outcomes and
expected innovation outcomes based on inputs alone. Innovation
efciency is affected by the institutional and networking aspects of
an economy that encourage, cultivate and facilitate the creation and
exploration of new ideas (Furman et al., 2002; OECD, 2003b).
Government policy and support for innovation, openness of the economy, the relative involvement of the business and public sectors in
R&D, the linkages between the science base and industries, the information infrastructure of the economy and the strength of protection
for intellectual property are all important factors affecting the innovative efciency of nations.
The participation of the private sector in innovation is one of the
most important factors in a national innovation system. The private
sector is not only a nancier and creator of innovative ideas but also
the place where the commercialisation takes place. As Porter and Stern
(1999) argue, therefore, the extent of R&D funded by private rms is a
332
reection of whether or not wider system conditions are conducive to

commercialisation. As innovation has become more open and more
global, there has been an increasing emphasis on universities as partners for business. The nature of that contribution must, however, be
seen in the context of the much more intensive use of commercial
partners as innovation knowledge sources and in the wider set of
contributions through human capital and public space functions that
universities perform (HM Treasury, 2003; Hughes, 2008). Given that
caveat, universities as major producers of knowledge are a signicant element in determining the innovative efciency of a national
innovation system.
In addition to funding from government and the private sector,
venture capital is frequently attributed an increasingly important
role in nancing innovation activities and in affecting innovation
performance, particularly in the high-technology sector (Porter and
Stern, 1999; Lerner, Sorensen and Stromberg, 2008). The success of
the Silicon Valley model is often cited as evidence of the important role
played by informal venture capital. However, the role of government
defence expenditure and the labour market in that respect cannot be
underestimated (Lcuyer, 2006). Moreover, venture capital does
not appear to systematically change patent quantity (Lerner et al.,
2008) and tends to focus on relatively few industries, such as biotechnology, the Internet, and telecommunication rms in the late 1990s
(Gompers, 2006). Moreover, in the case of the United States and
the UK, little formal venture capital goes to early-stage hightechnology businesses (Hughes, 2008). Therefore, we might expect
that the share of venture capital per se may play a nuanced role in
overall innovation performance and perhaps more through large business and later-stage developments than through early-stage high-tech
developments.
The degree of openness to international trade and investment is
another institutional factor that it is argued may affect national innovative efciency. On the one hand, greater openness may increase
competitive pressure on domestic rms and compel them to innovate
(e.g., Geroski, 1994; Nickell, 1996; Blundell, Grifth and Van Reenen,
1999). Market augmentation through exporting may also increase the
returns to R&D investment and thereby provide the incentive for
innovation. Moreover, the diffusion of knowledge through foreign
contacts and foreign direct investment may contribute to domestic
333
innovation activities via knowledge spillovers. On the other hand, openness to imports and foreign direct investment (FDI) may crowd out local
small and medium enterprises, create certain degrees of brain drain of
skilled scientists and engineers to foreign rms or to foreign countries
and result in some dependence on foreign technology transfer (Jaumotte
and Pain, 2005; Fu and Gong, 2008). The ability to appropriate returns
to innovation-related investment is therefore important. Strong protection of intellectual property will, it is argued, encourage rms to invest
in innovative activities and signal the attractiveness of the country as a
site in which to locate innovative activity.
13.3 Cross-country differences in innovation activity:

the stylised facts
R&D investment, patent counts, and sales of new or signicantly
improved products are the most frequently used indicators in attempts
to shed light on the innovation activities of nations. In Figures 13.2 and
13.3, we present data on R&D and patenting respectively.
3.5
3.0
France
2.5
Germany
Italy
Japan
2.0
UK
US
1.5
EU15
1.0
2005
2003
2001
1999
1997
1995
1993
1991
1989
1987
1985
1983
1981
0.5
Figure 13.2 Total national R&D as percentage of GDP for major industrial
countries
Source: OECD (2008).
334
400
350
300
250
200
150
100
50
FRANCE
GERMANY
UK
US
ITALY
2005
2003
2001
1999
1997
1995
1993
1991
1989
1987
1985
1983
1981
1979
1977
JAPAN
EU15
Figure 13.3 Patents granted by USPTO, 19822006

Source: EUROSTAT.
As a key input factor, R&D investment has a harmonised denition.

Figure 13.2 shows that Japan and the United States have long been
among the leading major industrial countries in terms of R&D expenditure. The share of R&D investment as a proportion of GDP has also
been increasing since the mid-1990s in these two countries. In contrast,
in Europe, both the UK and France have experienced drops in the shares
of R&D spending over the same period. The UK experienced a mediumterm decline in its R&D spending share after 1981. The share of R&D
investment in Germany started to increase after 1999, after a decade of
decline and stagnation since reunication in 1989.
Data from other company-based sources conrm the gap. According
to the 2005 International R&D Scoreboard (DTI, 2005), European
companies invested no more in R&D in 2004 and 2005 than they had
on average over the previous four years. The increased corporate R&D
investment for 2004 and 2005 was 2 per cent in Europe but 7 per cent in
the United States and Asia. According to the 2002 and 2003 UK R&D
Scoreboard (DTI, 2002c, 2003c), the average R&D intensity of
European top rms in 2003 was 3.7 per cent, which was signicantly
lower than that of top U.S. or Japanese companies at 5.2 and 4.3 per cent,
335
respectively.2 The gure for the UK top 700 companies was only 2.2 per
cent, which was low even compared to the rest of Europe. In contrast,
many smaller European economies have signicantly increased spending
on R&D. For example, Sweden, Finland, Denmark and Ireland have all
accelerated spending as a share of output in the past 20 years by 72 per
cent, 172 per cent, 88 per cent and 104 per cent, respectively (DTI 2005).
A basic problem in relying on R&D investment as an innovation
performance indicator is that it is basically a measure of innovation
inputs, which could be used inefciently (Jensen, 1993), and says little
directly about innovation outputs. An alternative is to use the share of
innovative/new sales in a rms total sales, based on rm-level survey
data grossed up to the national level. This is a direct measure of
successful innovation. However, for present purposes, this measure is
effectively restricted to European comparisons because it is most widely
available on a consistent basis as a result of the regular Community
Innovation Surveys (CIS) and cannot be used to compare all countries
with either Japan or the United States.3
Patent counts are another widely used innovation measure suitable
for international comparison and are widely available for countries
outside Europe. They are a measure of non-trivial inventive technologies with commercial promise (Smith, 2004). They are available over
time and across countries so that cross-country performance can be
compared over time. But patent count data, too, have well-known
limitations: they represent only a portion of innovative activity, and
they do not reect the economic value of different patents (Griliches,
1990). While recognising these limitations, it has been argued that the
patent count is nonetheless a useful measure for international innovation comparison given the alternatives (Trajtenberg, 1990; Furman
et al., 2002), and that the analysis of patent data has proven very
fruitful (Smith, 2004 page 160). For this reason, we will use patent
count in this chapter as a proxy of innovation output, albeit imperfect.
2
The DTI R&D Scoreboard compares the innovation activities by the top 700 UK
and top 700 international R&D investing companies. Detailed comparison of
large-rm R&D activities based on DTI (2002 and 2003) is summarised in
Appendix 13.1.
A notable exception is Cosh, Hughes and Lester (2004), who provide a
comparison using this class of measures for the UK and the United States. The
comparison is for manufacturing and business services. There is no economy-wide
comparison.
336
On this basis, it is clear that Europe has long lagged behind the United
States and Japan in the number of patents granted by USPTO per
million population. Since the mid-1990s, while the number of patents
granted by USPTO per million population has increased signicantly
for the United States and Japan, the growth of patents in Europe has
instead been rather modest (with the exception of Germany). This has
led to a widening gap in patenting between Europe and the United States
(see Figure 13.2). The data on European patent applications also conrm this trend. Although the predominance of U.S. rms holding U.S.
patents reects a home advantage, the United States records a relatively
strong performance in the EU market as well. With regard to the UK, its
record on patent numbers is only a third of that of Germany, and less
than those of France, Japan and the United States.
13.4 Estimating the patenting frontier and patenting efciency:

model, data and methodology
Model and method
We use stochastic frontier analysis to construct an international patenting frontier. We then use distance from the frontier and the factors
affecting that distance to estimate the patenting efciency of national
innovation systems and the determinants of that efciency. In the spirit of
Jones (1995), we thus emphasise the distinction between efciency in the
use of inputs from the magnitude of the inputs themselves. Following
Griliches (1979), Jaffe (1989), and Eaton and Kortum (1999), we start
from a patenting (innovation) production function as follows:
Y it Ait Xit
13:1
where Yit denotes innovative/patenting output of a nation, Xit is a vector

of basic innovation inputs, and Ait evit uit is an efciency parameter,
with as a constant; vit is a random disturbance that captures the effects
of statistical noise and is distributed as N0; 2v ; uit is a one-sided error
term representing a variety of features that reect efciency and is
interpreted as patenting efciency by the denition stated earlier; u is
independent of v; and u 0, distributed as truncations at zero of the
Nit ; 2u distribution.
Based on empirical data, SFA constructs a patent production frontier,
which maximises patent production given the inputs. In applying SFA
337
to estimate national patenting efciency, we rewrite Equation (13.1) in

the following functional form:
yit f Xit expvit uit
13:2
The deterministic production frontier f(Xit) represents patenting production possibilities, that is, the estimated maximum innovative
capacity with given innovative inputs. The stochastic production frontier is f(Xit)exp(vit it). Patenting efciency is calculated as the ratio of
the observed patent output to its maximum potential at the deterministic frontier as follows:
Patenting efficiency iEit
yit
expuit
f Xit expvit
13:3
The determinants of patenting efciency can then be analysed using the

following equation:
0
IEit it zit
13:4
where IE is patenting efciency; zit in our study is a vector of factors,

other than basic innovation inputs, that affect the innovative efciency
of nations, that is, the innovation environment and the quality of linkages; 0 is a vector of parameters to be estimated; i is a vector of nationor time-specic effects. In this study, because the institutional and
structural effects are likely to be captured by country-specic effects,
we exclude the country dummies from the model so that the signicance
of the impact of the institutional and structural factors can be maintained. We experiment with both a traditional two-stage approach and
a simultaneous approach for the SFA estimation as a robustness check.4
It is important to note that while the level of innovative inputs affects
innovation outputs, it may itself depend on innovation output and the
broader institutional environment. Although the use of SFA allows new
Stochastic frontier analysis is normally conducted in a two-stage procedure.

However, this procedure suffers from a contradiction (Battese and Coelli, 1995).
The Battese and Coelli (1995) model overcomes this contradiction and allows the
simultaneous estimation of the parameters of the stochastic frontier and efciency
determinants. However, the estimation of the simultaneous model is extremely
problematic in all but the most favourable of cases (Greene, 2003). Therefore, we
experiment with both the traditional two-stage approach and the simultaneous
model while bearing in mind the limitations of both.
338
analytical insights into the relative role of inputs versus efciency, the
method has its limitations, in that it cannot take into account this
possible endogeneity problem.
Data and measurement

We focus this study on OECD countries and have constructed the data set
from OECD Main Science and Technology Indicators (MSTI), the World
Bank Development Indicators (WBDI) and the Institute for Management
Development (IMD) World Competitiveness Yearbook (WCY). As some
of the data are not available for some countries, especially the transition
economies and developing countries, we have included 21 countries in the
dataset from 1990 to 2002, for which publicly accessible data are available. The countries included are Austria, Belgium, Denmark, Finland,
France, Germany, Greece, Iceland, Ireland, Italy, Netherlands, Norway,
Portugal, Spain, Sweden, United Kingdom, Australia, Canada, Japan,
New Zealand and United States.
The dependent variable
Following Jaffe (1989) and Furman et al. (2002), we proxy innovation
output by the number of granted U.S. patents. The country origin of a
patent is determined by the residence of the rst-named inventor. In
keeping with Furman et al. (2002), we take patents in country j in year
(t+3), PATENTSj,t+3, as the dependent variable. In other words, we
assume that ideas production in a given year is reected in the patents
that are granted three years in the future. We also employ U.S. patents
to provide a comparable measure of innovations with substantial
commercial importance across the 21 OECD countries in our sample.
Given the potential home bias effect of using the number of U.S. patents,
we focus our analysis and discussion of the results on the non-U.S.
countries.
Explanatory variables for the knowledge production function
The rst innovation input to be considered is human capital resources
devoted to the ideas-producing sector. We measure these inputs by the
number of full-time equivalent scientists and engineers working in
R&D in an economy (FTE S&E). Substantial heterogeneity exists
among countries and over time in terms of underlying international
patenting efciency (PATENTS / FTE S&E), ranging from more than
339
70 patents per 1,000 R&D workers for the United States during the
1990s to less than 1 patent per 1,000 R&D workers for Greece and
Portugal. Innovation expenditure is another important input to the
ideas-producing sector. We measure it by aggregating R&D expenditure. Including both R&D expenditure and personnel in the regression
may involve some double counting because a proportion of R&D
expenditure is on wages and salaries. This should not affect the relative
rankings of countries based on the SFA estimates, but it will affect the
interpretation of each variable as the estimated coefcients and their
statistical signicance are affected. We also include the share of GDP
spent on education as an indicator of a nations commitment to human
capital investment.
As discussed earlier, the industry structure in an economy may affect
its aggregate innovation performance at the country level, and policy
debate has focused on the share of high-technology activity in particular. We therefore attempt to capture these effects by the value-added
share of high-technology industries in the total economy.5 This does
vary across countries. For example, Ireland had the highest average
value-added share of high-technology industries in the whole economy
at 9 per cent from 1990 to 1999. The gures for the UK, the United
States, Japan, Germany, Sweden and Finland varied between 6 to 7 per
cent in these industries over the same period.
Explanatory variables for the patenting efciency equation
To account for a nations patenting efciency, we included the percentage of R&D funded by private industry and the percentage of R&D
performed by higher education institutions as two of the explanatory
variables. Porter and Stern (1999) suggested the former as an indicator
of the cluster environment for innovation; the latter is an indicator of
the involvement of universities in the national innovation system. As
discussed earlier, openness is also an important factor affecting innovative efciency of nations. We measure the degree of openness of an
economy by the volume and intensity of its exports and FDI alternately.
5
The high-technology industries include pharmaceuticals; ofce, accounting and

computing machinery; electrical machinery and apparatus; radio, television and
communication equipment; medical, precision and optical instruments; aircraft
and spacecraft; and post and telecommunications. Note, however, that this gure
for industry output may not be a good indicator of high-technology activities or
inputs in the economy (von Tunzelmann and Acha, 2005).
340
In keeping with previous literature, the level of GDP per capita is also
included to capture the development level of the economy as a proxy for
the overall level of institutional and infrastructural development. The
strength of IPR protection and the availability of venture capital in the
economy, measured by expert ratings collected in surveys, are also
included as explanatory variables.
The empirical model for innovation capacity as proxied by patents is
therefore as follows:
LnPat i;t3 LnRDit #LnRDPit LnEDUit LnHITEC
it it
13:5
The empirical model for patenting efciency is therefore as follows:

LnIEit it ?LnBERDit ?LnRDHEit IP VC
LnOPEN it LnGDPPCit it
13:6
where Pat = number of patents, RD = gross R&D expenditure, RDP =

R&D personnel, EDU = education expenditure, HITEC = value-added
share of high-technology industries in the whole economy, = IE =
patenting/innovative efciency, GDPPC = GDP per capita, OPEN =
openness, BERD = business R&D expenditure, RDHE = R&D performed by higher education, IP = strength of intellectual property protection, VC = availability of venture capital. Denitions of variables and
sources of the data are listed in Table 13.1. Appendix 13.1 reports the
average values of the key variables of the 21 countries.
13.5 Results
The estimated results from the SFA are presented in Table 13.2. R&D
personnel exert a signicant positive contribution to the basic patenting
capacity of nations. The magnitude and the statistical signicance are
robust across specications. A 1 per cent increase in R&D personnel
increases patenting output by nearly 1 per cent. Aggregate R&D expenditure also exerts a signicant positive impact on basic patenting
capacity. The estimated coefcient of the education variable is positive
but not statistically signicant, probably because of the high correlation
between education expenditure and R&D personnel.
We also estimated the SFA using the simultaneous model approach
suggested by Battese and Coelli (1995). The estimated results are
341
Table 13.1 Denition and sources of variables

Variable Denition
Sources
Pat
IE
MSTI
Patents granted in USPTO 3 years later

Innovative/patenting efciency resulting from SFA
analysis
RD
Gross expenditure on R&D in GDP
RDP
Number of full-time equivalent researchers
BERD Percentage of R&D funded by private sector
RDHE Percentage of R&D performed by higher education
institutions
EDU
Total expenditure on education
GDPPC GDP per capita in 1995 constant prices (US$)
EXP
Value of total exports (US$m)
EXPS
Ratio of exports to GDP
FDI
Value of inward FDI stock (US$m)
FDIS
Ratio of FDI to GDP
IP
How adequately IPRs are protected (except for 1990,
when it is the extent to which protection by IPRs
encourages corporate R&D) (scale 0 least, 10 greatest)
VC
How readily venture capital is available for business
development (scale 0 least, 10 greatest)
HITEC Value-added shares of high-tech industries relative to the
total economy
MSTI
MSTI
MSTI
MSTI
WBDI
WBDI
WBDI
WBDI
UNCTAD
UNCTAD
WCY
WCY
OECD
Note: We have included 21 countries in the data set, from 1990 to 2002, for which
publicly accessible data are available. Detailed list of the countries included is given in
the text of the paper.
broadly consistent with the two-step estimates. The estimated coefcient of the R&D variable marginally probably loses statistical signicance in the simultaneous model because of the double-counting
problem between R&D expenditure and R&D personnel. The practical
difculty of the simultaneous approach as discussed earlier may also be
a factor responsible for this change. With regard to industrial structure,
the value-added share of high-technology industry in the total economy
exerts a signicant positive impact on its basic patenting capacity. This
result suggests that industrial structure does matter in shaping overall
national innovation capacity in patent production. Economies with a
greater proportion of industries in the high-technology sector are more
Table 13.2 Stochastic frontier analysis of basic patenting capacity

Dependent variable: number of patents granted 3 years later
2-step estimation
(1)
Constant
Log(RD)
Log(RDP)
Log(EDU)
Log(HITECH)
N
Year dummies
Log Likelihood
Simultaneous estimator
(2)
(3)
(4)
Coeff
p-value
Coeff
p-value
Coeff
p-value
Coeff
p-value
12.174***
0.356***
0.821***
0.093
0.000
0.002
0.000
0.438
11.197***
0.294***
0.881***
0.069
0.236**
205
Yes
84.657
0.000
0.009
0.000
0.558
0.032
9.925***
0.254
1.005***
0.001
0.279**
205
Yes
0.000
0.122
0.000
0.994
0.040
9.898***
0.249
1.005***
0.001
0.293**
205
Yes
0.000
0.129
0.000
0.994
0.036
210
Yes
87.005
Note: ***signicant at the 1 per cent level. **signicant at the 5 per cent level. *signicant at the 10 per cent level.
Independent variables of the inefciency equations in models (3) and (4) include LNRDHE, LNBERD, LNGDPPC, IP and LNOPEN.
343
likely to have a greater number of granted patents. Whether this is

simply an artefact of choosing industries such as pharmaceuticals that
are known to have exceptionally high propensities to patent must be left
to further research to assess.
Table 13.3 reports the estimated results of the determinants of patenting efciency. As a robustness check, we report both the Tobit and the
OLS estimates, which are consistent in general. The estimated coefcients of the GDP per capita variable are positive, large and statistically
signicant. The result is robust across specications and equations. This
suggests that the development level of an economy, such as the technology and institution advancement level and infrastructure development,
exerts a signicant effect on the patenting efciency of nations.
The share of R&D funded by the private sector also exerts a positive
impact on patenting efciency and is statistically signicant in most of the
specications. If the share of R&D funded by the private sector is taken as
a proxy for system-wide innovation environment effects (Porter and
Stern, 1999), then a higher share of R&D funded by the private sector
reects a more favourable innovation system environment and therefore
increases the patenting efciency of nations. R&D performed by higher
education institutions also has a robust signicant positive impact on
national patenting efciency. The magnitude of the estimated coefcients
is of similar scale as those for the business R&D expenditure variable.
This evidence conrms that universities have played a signicant role in
patenting. Thus, the higher the share of universities in total R&D, the
higher the patenting efciency of a nation will be.
As expected, the degree of intellectual property rights protection has a
signicant positive impact on patenting efciency. The higher degrees of
intellectual property rights protection motivate greater efforts in innovation and effective commercialisation of patenting inventions.
However, the availability of venture capital is negatively associated
with patenting efciency. This may be explained by the fact that, as
the OECD data and other work suggest, much or all of the formal
venture capital in the UK and the United States is largely focused on
the activity of large rms and management buyouts and related restructuring rather than on technology-intensive sectors or early-stage innovative rms (Hughes, 2008). Our result is also consistent with the
ndings by Lerner et al. (2008) that venture capital does not appear to
systematically change patent quantity. The signicant negative coefcient remains of concern because it implies that the size of the formal
Table 13.3 Determinants of patenting efciency

Tobit
Log(RDHE)
Log(BERD)
Log(GDPPC)
IP
VC
Log(EXPS)
Log(EXP)
Log(FDI)
Log(FDIS)
OLS
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
0.084***
0.006
0.121**
0.019
0.215***
0.000
0.012**
0.021
0.012***
0.002
0.070**
0.026
0.085***
0.006
0.123**
0.019
0.280***
0.000
0.012**
0.021
0.012***
0.003
0.074**
0.018
0.105**
0.041
0.204***
0.002
0.008*
0.099
0.010***
0.008
0.073**
0.019
0.105**
0.041
0.197***
0.001
0.008
0.102
0.010***
0.008
0.083***
0.005
0.092*
0.091
0.198***
0.000
0.010*
0.056
0.011***
0.004
0.054*
0.091
0.084***
0.005
0.092*
0.096
0.241***
0.002
0.010*
0.060
0.011***
0.006
0.073**
0.011
0.082
0.108
0.200***
0.002
0.008*
0.096
0.010**
0.011
0.073**
0.011
0.082
0.108
0.194***
0.001
0.007*
0.098
0.010**
0.011
0.063**
0.046
0.045
0.134
0.008
0.25
0.007
0.264
0.008
0.231
0.007
0.248
Constant
Country dummies
N
Log likelihood
R-squared
2.140***
0.000
Yes
205
348.107
1.150***
0.008
Yes
205
347.619
1.920***
0.000
Yes
205
346.278
2.051***
0.000
Yes
205
346.332
1.904***
0.001
Yes
205
1.158***
0.004
Yes
205
1.811***
0.001
Yes
205
1.934***
0.002
Yes
205
0.978
0.978
0.977
0.977
Note: ***signicant at the 1 per cent level. **signicant at the 5 per cent level. *signicant at the 10 per cent level.
Dependent variable: estimated patenting efciency of nations.
346
venture capital relative to GDP hinders rather than helps patenting

efciency. This may reect the extent to which formal venture capital
has been focused on the reorganisation of existing assets through buyouts and cost efciency drivers rather than investment in high-risk R&D
expenditures and/or enhanced patenting performance. This is an issue
that requires careful further analysis.
In keeping with the results from Jaumotte and Pain (2005), we nd
that the impact on patenting of openness to FDI is not signicant.
Openness to exports even shows a negative impact on patenting efciency. This result may be partly the result of the fact that, as far as
innovation is concerned, the openness of an economy serves as a twoedged sword. While competitive pressure may compel the local rms to
innovate to be competitive and foreign contacts may facilitate international knowledge transfer, the competitive pressure may also crowd out
local rms from product markets, and easy access to foreign knowledge
sources and advanced machinery may lead to certain degrees of dependence on foreign technology.
Benchmarking national patenting capacities

and patenting efciency
Figures 13.4 and 13.5 compare the estimated patenting capacities and
patenting efciency of six major industrialised countries obtained from the
SFA, with patents proxying for innovative output. By this measure, Japan
has been the top non-U.S. country with the highest patenting capacity
(Figure 13.4). Japan has not only done well in patenting output, it is also
one of the most efcient patenting countries in the world (Figure 13.5). In
terms of innovation capacities in patent production, the distance of the rest
of the G7 countries from the world leader United States is substantial, and
there is little sign of convergence in the 1990s. Compared with the G7
countries, the UKs innovation capacity in patent production is higher
than that for Italy, but lower than that for France and Germany, and far
below that for the United States and Japan. In terms of patenting efciency, the UK is consistently ranked the lowest among the G7 countries
and shows little sign of catch-up in this respect. Considering the UKs
strong performance in scientic publications and citations (King, 2004),
this suggests that issues of successful translation of ideas into commercialisation practices remain important. France is in a similar situation to the
UK, although its overall performance has been slightly better.
347
Patenting capacity of selected countries

Number of patents per million
population
300
250
200
150
100
50
0
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999
France
Germany
United Kingdom
Italy
Japan
United States
Figure 13.4 Patenting capacity of selected countries
Patenting efficiency of selected countries

1.10
1.00
0.90
US = 1.0
0.80
0.70
0.60
0.50
0.40
0.30
0.20
1990
1991
1992
1993
1994
1995
1996
1997
1998
France
Germany
Italy
Japan
United Kingdom
United States
1999
Figure 13.5 Patenting efciency of selected countries
Italy has been the weakest among the G7 countries in terms of

national patenting capacity. The number of U.S. patents it may produce
based on its basic innovation inputs is far below that of the world
leaders. However, Italy did fairly well in efciently transforming its
348
innovative inputs into outputs. Italys estimated patenting efciency has

started to catch up since 1992, and this has placed it close to the world
patenting frontier. This suggests that Italy has created an effective
innovation system and mechanism to transform the innovative inputs
into outputs. The main problem for Italy would appear to be the lack of
innovation inputs. Different from the UK, France and Italy, Germany
has been the best European country in terms of patenting capacities. It
has also demonstrated a trend towards catch-up in patenting efciency
to the world frontier. Its patenting efciency in 1999 was almost as high
as that of the United States and Japan.
We have also compared the country patenting ranking from this study
with those from several other studies that use similar data and variables
but employ different methodology. These studies do not decompose the
national innovation performance; instead, they incorporate innovation
inputs and environment indexes into one model. Tables that summarise
the rankings and the estimated efciency scores using SFA can be found
in Appendix 13.2. Both our capacity and efciency rankings are, to a
certain extent, correlated with the innovation rankings from other studies suggesting their consistency in the main. However, the decomposition
exercise we carried out in this study reveals some interesting insights that
distinguish our method from other methods. Japan has invested huge
resources in R&D and became the world leader in terms of patenting
capacity; its overall innovation performance ranked in other studies is
mostly one or two ranks lower because it is less efcient in innovation in
comparison with the worlds most efcient innovative economies. The
United States stands in the top ranks in terms of both patenting capacity
and patenting efciency. It is hence, with no surprise, ranked as the world
leader in innovation performance by most other studies. Sweden,
Canada, Finland and Denmark all have impressive patenting efciencies
that ranked higher than their positions in the innovation capacity league
table. Italy has the lowest patenting capacity but an impressive performance in patenting efciency. It therefore stands in the lower-middle rank
in most studies on overall performance.
13.6 Benchmarking China against the world

innovation frontier
Using data collected from OECD, Eurostat and the World Bank, following the same methodologies discussed earlier, we estimate the
349
expected innovation capacity of China and other emerging economies

for the period from 2005 to 2011 and compare it with the actual
performance of these countries in terms of number of patents granted
by the USPTO. The countries included in this analysis are China,
Russia, Korea, Singapore and Argentina, in addition to the major
industrialised countries such as the United States, the UK, Japan,
Germany, France, Italy and Spain. The objective of this exercise is to,
rst, benchmark the innovation (patenting) performance of China and
the other emerging economies against the world frontier and understand where China is in the global landscape after three decades of
catch-up. Second, we decompose the gap between China and the
world frontier into investment-based capacity and innovation efciency
and hence understand where the major gap lies.
As Figure 13.6 shows, the United States is the largest investor in R&D
and remained the worlds top innovator between 2005 and 2011 in
terms of its expected innovation capacity estimated based on a countrys
investment and human capital for innovation. Japan ranks second in
this respect. Amongst all the sampled countries, two have shown a
clearly upward catching-up trend in expected innovation capacity and
are ranked as the top fth and sixth investors in 2011 in this sample.
They are China and South Korea, which have invested heavily in R&D.
Expected innovation capacity: emerging economies

12.00
US, 11.87
Ln(No of patents granted)
11.00
Japan, 10.81
Germany, 10.45
France, 10.09
China, 9.75
Korea, 9.64
UK, 9.67
Italy, 9.00
Spain, 8.71
Russia, 8.44
10.00
9.00
8.00
Singapore, 7.12
7.00
Argentina, 6.98
6.00
5.00
2005
2006
2007
2008
2009
2010
2011
Figure 13.6 Expected innovation capacity: emerging economies
350
China also has the worlds largest pool of trained scientists and engineers. However, looking at the actual performance in terms of the
number of patents granted by the UPSTO, if we exclude the United
States, which is the home country of the patent-granting authority,
Japan is the largest owner of patents granted by USPTO. Korea is the
second largest owner of granted patents by the USPTO in 2011,
although its number of granted patents is about only a quarter of that
of Japan. The number of granted patents of China in 2011 was 3473,
which was about 8 per cent of that of Japan (Figure 13.7). However, as
reported in Chapter 2, the R&D expenditure to GDP in China was 1.98
in 2011, which was 59 per cent of that of Japan. In terms of total R&D
expenditure, Chinas was more than half (54 per cent) of that of Japan.
The differences between actual innovation performance and expected
capacity based on R&D investment and personnel suggest the
variations across countries in their efciency in innovation. Using the
stochastic frontier analysis method discussed earlier in this chapter, we
can estimate the innovation efciency of the sample countries. The
estimated innovation efciency scores are reported in Figure 13.8. As
the diagram shows, Japan was the most efcient innovator in the world
in most of the years over the sample period of 2005 to 2011. Korea was
Actual performance: patents granted by USPTO

(US doubles Japan's number)
50000
Japan, 46189
45000
No of patents granted
40000
35000
30000
25000
20000
15000
Korea, 12233
Germany, 11913
10000
France, 4553
UK, 4352
China, 3473
Italy, 1921
Spain, 492
Russia, 331
Singapore, 650
Argentina, 60
5000
0
2005
2006
2007
2008
2009
2010
Figure 13.7 Actual performance: patents granted by USPTO
2011
351
Innovation efficiency of emerging economies (Japan=100)

1.20
Efficiency score
1.00
Japan, 1.00
Korea, 0.85
0.80
US, 0.80
0.60
Singapore, 0.56
0.40
Germany, 0.37
UK, 0.29
Italy, 0.25
China, 0.22
France, 0.20
Spain, 0.09
Russia, 0.08
Argentina, 0.06
0.20
0.00
2005
2006
2007
2008
2009
2010
2011
Figure 13.8 Innovation efciency of emerging economies
the second most efcient in 2011, and the United States ranked third in
the sample. In 2011, Chinas innovation efciency score was only 22 per
cent of that of Japan. In sum, there is a signicant gap in innovation
performance between China and the world leaders. Although a gap in
R&D investment remains between China and the United States and
Japan, the major gap lies in the low innovation efciency in China.
13.7 Conclusions
This chapter has compared the determinants and performance of patenting among the OECD countries and benchmarked their patenting
performance against the world patent production frontier. It nds that
with respect to basic patenting capacity that is determined by physical
innovation inputs, Japan and the United States have been the world
leaders in the 1990s. The gap between the European countries and these
leaders are substantial and there is no signicant sign of convergence.
With respect to national patenting efciency, which is determined by
institutional and structural factors in the national innovation system,
Japan, Germany and Italy have been catching up during the sample
period. The gap in patent production between the UK and the world
leaders is substantial and lies in both basic patenting capacity and the
patenting efciency of the national innovation system. Both the
352
generation of inputs and greater efciency appear to be required. The

results from our study also suggest that institutional factors, such as the
involvement of universities in the national innovation system; the proportion of R&D funded by the private sector, which reects the effectiveness of the institutional system in encouraging industrial R&D
investment; and the development level of the economy have identiable
effects on patenting.
A number of policy implications are possible. To improve national
innovation performance and enhance productivity and competitiveness
of economies, considerable investment has been made in R&D.
Increasing the ratio of R&D to GDP of countries has also been a
major target of national governments and of the EU. The evidence
from this study suggests that in addition to increased R&D expenditure,
patenting performance is also signicantly affected by the patenting
efciency of nations. In assessing the balance of effort in policy development between enhancing inputs and enhancing their conversion into
outputs, countries should examine carefully the relative weakness in
each aspect. We regard our ndings as methodologically useful in
distinguishing between inputs and efcient transformations.
In terms of developing current policy, a number of important issues
arise. First, the development of better longitudinal data of an internationally comparable kind on innovation outputs is important. The
development of a CIS-like sweep in the United States and Japan
would, for example, help here. Second, we have presented relatively
simple models that are parsimonious in relation to the variables
included. The approach could be made richer if consistent data on the
other system aspects were developed. Finally, our data end in 2002 so
an update would inject interesting evidence as to whether this extensive
policy commitment to EU innovation policy in the past ve years
(200914) has been translated into either patenting capacity or efciency improvements.
Findings from the chapter indicate a signicant gap in innovation
performance between China and the world leaders. Although a gap in
R&D investment remains between China and the United States and
Japan, the major gap lies in the low innovation efciency in China. This
is the bottleneck that China has to tackle even if we take into account
the limitations in using number of patents as a proxy of a countrys
innovation performance.
Appendix 13.1 Mean values of key variables for sampled countries

Country
Pat
Edu
RD
BERD
RDHE
GDPPC
OPEN
VC
IP
HITEC
Australia
Austria
Belgium
Canada
Denmark
Finland
France
Germany
Greece
Iceland
Ireland
Italy
Japan
Netherlands
New Zealand
Norway
Portugal
Spain
Sweden
United Kingdom
United States
Total
445
283
289
2058
241
423
2811
7037
9
4
53
1138
23622
689
54
141
3
138
833
2163
62371
4991
5.37
5.02
4.02
7.16
7.89
7.22
5.47
4.34
2.77
4.67
5.63
4.46
3.67
4.11
5.09
6.91
6.31
4.61
8.04
5.36
5.36
5.40381
1.57
1.63
1.80
1.67
1.90
2.56
2.29
2.37
0.52
1.76
1.20
1.09
2.88
1.98
1.03
1.67
0.64
0.86
3.33
1.97
2.59
1.775381
46.05
44.86
66.08
44.25
52.22
61.21
49.50
62.38
22.75
35.27
65.36
43.83
71.17
47.31
32.31
47.87
21.42
45.62
65.34
49.40
61.85
49.3359
26.50
32.55
25.90
25.84
21.82
19.82
16.69
17.48
44.90
24.98
20.95
27.12
17.02
28.59
32.18
26.99
39.35
30.12
23.16
18.75
14.68
25.49429
20990
29887
27840
20186
34835
26805
27399
30414
11685
27361
20070
19322
42767
27687
16439
34139
11277
15368
27809
19777
28407
24784
19
40
72
37
37
35
24
27
19
35
76
24
10
58
31
40
30
23
38
27
11
33.86414
4.98
4.61
5.42
6.12
6.34
5.72
5.41
5.17
3.38
3.91
5.73
3.70
4.30
7.27
4.52
5.08
4.21
4.10
6.04
7.08
7.92
5.319507
7.22
7.82
6.91
7.19
7.28
6.93
7.33
7.91
5.55
6.34
6.57
5.51
6.67
7.33
7.14
6.92
5.00
5.40
6.91
6.91
7.54
6.790071
3.83
5.4
4.85
5.03
4.63
6.06
5.45
6.67
2.08
2.15
9.12
4.88
6.65
4.41
3.33
3.63
4.01
4.44
6.15
6.68
6.75
5.057143
Appendix 13.2a Estimated patenting capacity of sample countries (unit: patents per million population)
Australia
Austria
Belgium
Canada
Denmark
Finland
France
Germany
Greece
Iceland
Ireland
Italy
Japan
Netherlands
New Zealand
Norway
Portugal
Spain
Sweden
United Kingdom
United States
Average
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
45
31
34
52
40
54
74
118
4
14
18
32
229
39
14
47
7
19
64
63
187
56
50
30
36
51
42
53
78
111
4
14
19
31
226
39
16
47
8
19
68
67
184
57
49
27
35
51
42
54
75
99
5
15
21
29
215
40
16
47
8
18
68
64
172
55
51
31
39
54
45
59
77
96
5
16
24
29
216
43
15
49
10
20
75
67
172
57
54
34
40
55
51
64
78
98
5
20
26
28
227
44
15
52
11
20
82
71
177
60
69
44
51
75
63
91
96
117
11
22
35
35
254
54
22
65
15
26
105
81
225
74
69
47
54
78
69
107
95
119
11
37
34
33
263
61
25
68
17
28
112
81
239
78
71
51
61
78
74
130
98
123
14
43
36
31
278
62
26
70
20
33
121
88
251
84
73
56
66
80
78
151
104
134
17
47
38
32
280
66
27
72
21
33
130
91
266
89
71
57
67
79
80
165
106
134
20
52
41
32
269
64
26
72
23
40
136
89
269
90
Appendix 13.2b Estimated patenting efciency of sample countries
Australia
Austria
Belgium
Canada
Denmark
Finland
France
Germany
Greece
Iceland
Ireland
Italy
Japan
Netherlands
New Zealand
Norway
Portugal
Spain
Sweden
United Kingdom
United States
Average
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
0.478
0.829
0.698
0.858
0.800
0.843
0.662
0.694
0.220
0.883
0.664
0.658
0.746
0.819
0.778
0.631
0.220
0.233
0.827
0.575
0.852
0.665
0.540
0.826
0.664
0.865
0.804
0.856
0.611
0.704
0.298
0.844
0.608
0.655
0.750
0.854
0.589
0.655
0.220
0.220
0.848
0.524
0.862
0.657
0.550
0.856
0.730
0.870
0.788
0.872
0.634
0.744
0.220
0.756
0.669
0.622
0.757
0.831
0.697
0.678
0.220
0.227
0.868
0.566
0.871
0.668
0.533
0.843
0.744
0.866
0.819
0.891
0.615
0.761
0.280
0.707
0.674
0.658
0.774
0.802
0.777
0.675
0.220
0.220
0.866
0.561
0.882
0.675
0.502
0.832
0.791
0.871
0.855
0.886
0.639
0.761
0.229
0.554
0.683
0.678
0.757
0.805
0.877
0.677
0.220
0.231
0.851
0.572
0.877
0.674
0.557
0.781
0.769
0.859
0.837
0.875
0.639
0.786
0.220
0.726
0.572
0.709
0.809
0.828
0.857
0.704
0.220
0.264
0.872
0.601
0.878
0.684
0.545
0.796
0.742
0.865
0.844
0.868
0.656
0.786
0.220
0.815
0.684
0.694
0.802
0.795
0.809
0.718
0.220
0.220
0.886
0.607
0.874
0.688
0.533
0.763
0.769
0.874
0.827
0.829
0.653
0.806
0.220
0.737
0.797
0.759
0.788
0.805
0.773
0.738
0.220
0.220
0.893
0.565
0.866
0.687
0.611
0.781
0.773
0.879
0.834
0.839
0.665
0.812
0.220
0.816
0.723
0.751
0.803
0.783
0.812
0.737
0.220
0.220
0.895
0.582
0.859
0.696
0.614
0.743
0.732
0.875
0.798
0.836
0.646
0.811
0.220
0.774
0.688
0.756
0.825
0.903
0.863
0.718
0.220
0.220
0.887
0.547
0.852
0.692
Source: Authors estimation using the SFA.
Appendix 13.2c Comparison of country rankings in innovation
Study
Our study:
capacity
Our
study:
efciency
Porter & Stern Gans & Stern

(1999)
(2003)
OECD
(2003)
INSEAD
(2007)
EIU (2007) EC (2008)
Year
1998
1998
1995
1998
2007
2006
Indicator
Predicted
Predicted
Predicted
Triad
patents per mil Efciency patents per mil patents per mil patents per
pop
score
pop
pop
mil pop
Japan
1
United
2
States
Switzerland
Finland
3
Germany
4
Sweden
5
France
6
United
7
Kingdom
Canada
8
Denmark
9
Australia
10
Norway
11
2000
2007
8 pillars of Innovation Average scores

innovation enabled
of 25 indicators
9
3
3
1
2
1
3
6
4
1
1
3
6
9
4
8
1
16
18
2
6
5
4
8
14
3
5
7
4
9
13
1
4
5
2
12
14
6
13
2
12
5
3
2
5
6
4
12
18
2
3
7
1
14
8
2
5
17
14
9
7
12
10
10
6
16
8
16
9
17
15
8
11
17
25
13
7
21
16
16
5
17
18
Ireland
Netherlands
New
Zealand
Belgium
Austria
Iceland
Spain
Italy
Portugal
Greece
12
13
14
10
12
11
15
16
17
18
19
20
21
6
11
19
13
5
19
19
11
13
17
16
15
11
12
15
7
11
13
9
15
22
9
14
11
14
17
20
18
19
22
23
10
19
22
18
20
27
24
20
21
27
24
28
39
49
19
25
20
22
33
32
13
15
12
11
12
23
21
28
27
Sources: 1) Gans, Joshua and Scott Stern. (2003). Assessing Australias Innovative Capacity in the 21st Century. 2) OECD. (2003c). OECD
Science Technology and Industry Scoreboard. 3) INSEAD. (2007). The World Business/INSEAD Global Innovation Index 2007. 4) Economist
Intelligence Unit (EIU) 2007. Innovation: Transforming the Way Business Creates. 5) European Commission. (2008). European Innovation
Scoreboard 2007. [Relate to columns above]
14
Capabilities, incentives, institutions

and national innovation performance
14.1 Introduction
After more than 30 years of economic reforms and having transformed
from a low- to a middle-income country, China now faces signicant
challenges in moving from imitation to innovation. The success of this
transformation will be of crucial importance for China to avoid
the middle-income trap and sustain its long-term economic growth
(Wu, 2013).
The past fteen years has seen an exponential increase in research and
development (R&D) investment in China. Developing an innovationdriven economy in the coming decades is a major objective for China.
Although many research papers and government reports discuss national
innovation capabilities in China, most of them address this problem using
the national innovation system (NIS) approach (e.g., Chen and Liu, 2008;
OECD, 2008; MOST, 2010; Wu, 2010; Mu and Fan, 2011). The NIS
framework is powerful in mapping out the major players in the innovation system and allows policy makers to identify the objects for policy
intervention, but it does not explain how to increase the capabilities of
players and strengthen the linkages between them. The question of how
to enhance the innovation capabilities and performance of the country
requires in-depth and systematic analysis.
This chapter examines policy choices for the development of innovation capabilities in China using the national innovation performance
(NIP) framework developed by the OECD (1987) and Lall (1992) and
modied and extended by Fu (2014). The importance of policy in
enhancing national innovation performance vis--vis simply leaving it
to the market is validated by several factors. First, given the nature of
knowledge as a public good and the prevalence of externalities that may
benet other users of knowledge, government intervention is required
to protect R&D investors and to encourage innovation activities.
Second, innovation is risky and costly. Therefore, government nancial
358
359
support is needed to enable active innovation activities, especially in

basic research where uncertainty is high because it is difcult to control
the direction of the development of a technology and the cost of failure
of an experiment is also high. Third, the presence of market failure
caused by problems such as information asymmetries, externalities and
public goods calls for policy intervention in information and knowledge
provision. Finally, China is being forced towards a more skill-intensive
and technology-intensive growth path, with the amount of surplus
unskilled labour in China falling. This transformation can be realised
through the operation of market forces, but long-sightedness by rms
and government intervention are also required because of the long
gestation periods involved.
The present chapter is organised as follows: Section 14.2 discusses the
theoretical framework concerning the determinants of national innovation performance. Section 14.3 analyses the policy choices for boosting
innovation performance in China based on analysis of Chinas innovation performance reported in earlier chapters, especially Chapters 2
and 13. Section 14.4 examines the space for innovation policy in the
twenty-rst century. Section 14.5 presents conclusions.
14.2 Determinants of national innovation performance:

towards a theoretical framework
In a similar way to the overall growth performance of a country,
national innovation performance (NIP) is a complex interaction of
incentives, capabilities and institutional factors. It is the interplay of
all these factors, in particular country settings, that determines how well
countries use resources and develop their innovation capabilities. The
OECD (1987:153) denes the role of incentives, capabilities and institutional factors as follows:
The capabilities dene the best that can be achieved; while the incentives guide
the use of the capabilities and, indeed stimulate their expansion, renewal or
disappearance. In the advanced economies, the capabilities refer primarily to
the supplies of human capital, of savings and of the existing capital stock, as
well as to the technical and organizational skills required for their use; the
incentives originate largely in product markets and are then more or less
reected in markets for factors supplies thereby determining the efciency
with which capabilities are used. Both incentives and capabilities operate within
an institutional framework: institutions set rules of the game, as well as directly
360
National Innovation Performance

Capabilities
(Best can be
achieved)
Capital,
Human capital,
State capacity
Tech effort
Institutions
NIP
(Set rules,
intervene)
Incentives
(Guide use of
capabilities)
Legal, Industri
Training,
Technological
Openness
(global
resources &
knowledge)
Macro, Micro
Product market
Competition
Factor markets
Figure 14.1 National innovation performance
intervening in the play; they act to alter capabilities and change incentives; and
they can modify behavior by changing attitudes and expectations.
In this chapter, we rst develop a model of the determinants of

national innovation performance based on this three-pronged approach
developed by the OECD (1987) and Lall (1992), taking into account the
recent trend towards globalisation and the shift from a closed to an
open innovation paradigm. This modied model, which is depicted in
Figure 14.1, will create a framework to analyse the drivers of national
innovation performance in China.
Determinant 1: Capabilities
Innovation capabilities include (1) the traditional elements of capabilities such as physical investment, human capital and technological
efforts (Lall, 1992) and (2) the capabilities of the state.
Human capital. This term has a broad denition in this context. It not
only refers to the skills in mastering and operating machines or the
education level of the workforce but also includes creativity in the population the creativity not only in the industrial but also in the research/
university and the public sectors and not only codied knowledge but also
tacit knowledge (Fu, Pietrobelli and Soete, 2011). Skills can be developed
through formal education but also informal or on-the-job training.
Physical investment. This refers to general investment in production
facility, from investment in premier production equipment, production
361
material and related infrastructure. Dening innovation in a broad

sense to include the process from idea creation, R&D, production to
commercialisation, physical investment is key for the commercialisation process.
Technological efforts. These are the efforts of productive enterprises to
assimilate and improve on the relevant technology. This includes efforts
through monetary investment in R&D but also involves non-monetary
forms of effort, for example, risk taking and entrepreneurship.
Capabilities of the state. Because the state is regarded as a natural
actor in the national innovation system through its role in supporting,
facilitating and nancing innovation, its ability to introduce correct
intervention at the appropriate time is also important for the full functioning of the national innovation system.
In addition, because of information asymmetry, non-competitive
markets and the public good nature of knowledge, market failures
may occur, creating a justication for government intervention to remedy the failures. In light of the role of the state in the national innovation
system, the costly and risky nature of innovation processes and the
externalities that innovation will generate, the state may actively provide direct nance for innovation or support innovation through scal,
nancial, trade and industry policies. However, government policy does
not necessarily lead to positive or expected outcomes. Therefore, the
governments capacity to design and implement appropriate policies
will also affect the outcome.
Determinant 2: Incentives
Because innovation is costly and risky, knowledge may generate
externalities and the benets are not easy to appropriate without
stringent protection of intellectual property rights (IPR), innovation
requires strong incentives to promote. Incentives will affect the
(human and physical capital) resource allocation in the economy
and determine how much and what kind of resources will ow to
the innovative sector. The incentives of innovation include macroeconomic incentives: incentives (or disincentives) from product markets and competition; from factor markets; from economic system;
and from proactive or responsive monetary, scal, trade and science
and technology policy.
Macroeconomic incentives. Some elements of the macroeconomic
policies will affect the incentives for innovation; economic growth;
362
and monetary, scal, trade and science and technology policy. A fastgrowing economy indicates greater demand and markets for new
products and hence greater incentive to invest in innovation.
However, an adverse economic environment may also force rms to
innovate and use process innovation to cut costs in particular to
survive. A low interest rate will decrease the cost of borrowing and
hence be a greater incentive to invest in innovation. So, too, will a low
expansionary tax policy enable rms, small and medium enterprises
(SMEs) in particular, to innovate. Of course, an economic environment in which it is easy to survive may make some rms/entrepreneurs
lacking motivation begin to innovate. Empirical evidence is scarce in
this respect. Finally, greater export orientation is found to be positively associated with innovation activity at the rm level (Coe and
Helpman, 1995; Bustos, 2011; Kiriyama, 2012); therefore, an
exchange rate that favours exports is likely to stimulate more
innovation.
Competition and product market. Competition is a two-edged sword
for innovation. On the one hand, rents from monopoly power will give
rms greater incentives to innovate (Shumpeter, 1942). On the other
hand, competition in the market will force rms to innovate so as to
survive (Geroski, 1994). More recent research also argues that the
relationship between competition and innovation follows an inverted
U-shape (Aghion et al., 2005). Firms innovate to pursue prot or to take
strategic leadership in a new or existing market. A product market that
enables rms to appropriate the rents from innovation and minimises
the rents from monopoly power is conducive to maintain rms motivation to innovate and to direct nancial resources to ow to the
innovation sector.
The economic system of an economy may also affect the incentives for
innovation. When some sectors are open for free competition while
others have restrictions for entry, rms in the restricted sectors are
able to enjoy abnormal returns from monopoly rents. As a result,
capital and even talents may ow to the restricted sectors.
However, competition concerns domestic markets; it is also related
to a countrys trade policy. The debate on the role of industrial policy
is inclusive. A wealth of literature suggests benets from openness to
innovation and growth through competition effects and knowledge
transfer and spillover. The World Bank and a list of scholars have
argued for openness to trade (and foreign direct investment) based on
363
the success of the South East Asian economies and the recent rapid
growth of China and India since the reforms (Kruger, 1974; Lin,
2011). Alternatively, the success of Korea and Japan inspired the
infant industry argument and arguments for selective winnerpicking industry policy, which protects local industry by restricting
foreign entry based on the experience of Japan, Korea and Taiwan
(Wade, 1990; Chang, 2003).
Factor markets. The efciency of capital and labour markets will
signicantly affect the inter-rm and inter-industry resource allocation.
Efciency factor markets will direct and facilitate capital and labour
ow to the sectors with higher return. Efcient labour markets should
also transmit the higher returns to skills through wages and therefore
attract talent to the innovative sectors.
Determinant 3: Institutions
Both incentives and capabilities operate within an institutional framework, which means they operate within various market and non-market
institutions (OECD, 1987). With regard to national innovation capabilities, several sets of institutions are involved: the legal framework
concerning science, technology and innovation; industrial institutions;
education and training institutions; technological institutions; and IPR
institutions:
With regard to the legal framework, an intellectual property rights

protection regime is the most important factor. This includes not only
the availability of the relevant laws and regulations but also the
strength in the implementation of the law.
Industrial institutions refer to those that promote inter-rm linkages
and intra-industry knowledge diffusion and co-ordinate the production of public innovative knowledge that benets everyone in the
sector. They also include those that provide support to SMEs.
Training institutions include the formal higher education sector,
vocational training institutions and institutions that provide training
support for SMEs.
Technology institutions include those market and non-market institutions that promote the creation and diffusion of innovation within
the country and between countries, for example, the technology
exchange markets and technology transfer organisations that help
364
the diffusion of innovation. In recent years, government-supported

innovation platform or innovation alliances have been widely used in
the United States, Europe and China.
A culture that encourages risk taking and tolerance to failure and
does not over-emphasise short-term performance is also conducive to
innovation activities of rms/organisations and individuals.
Therefore, a countrys technological efforts, human capital and physical
investment dene the best that can be achieved in innovation, whereas
incentives at macro and micro levels will guide the use of these capabilities and stimulate their expansion, renewal and disappearance.
Incentives determine the efciency with which capabilities are used.
Both of these factors operate within an institutional framework: for
example, legal institutions, such as IPR protection and economic and
educational institutions. These institutions set the rules of the game and,
through these, alter capabilities and change incentives.
Moderating Factor 1: Interplay

It is the interplay of the previously noted factors that shapes the national
innovation capabilities of a country. Active and appropriate interactions between these factors require co-ordination and facilitators. In
economies with well-developed markets, markets and prices will serve
as moderating factors. If a market is unavailable (e.g., if there is no
formal market for the trade of technology such as through a technology
exchange or an intellectual property exchange), underdeveloped or
subject to market failures, the state will play an important role. As the
capabilities and initial level of development of countries and cultures
differ, policies for innovation should be structured differently.
Moderating Factor 2: Openness

In the present context, openness refers to the international openness of
innovation systems, primarily at the national level but also at the rm
level (Chesbrough, 2003). With increasing globalisation and the internationalisation of innovation activities, the openness of innovation
systems to foreign or transnational rms, organisations and researchers
inevitably has a signicant impact on all three determinants of innovation capabilities (Fu, 2008; Fu, Pietrobelli and Soete, 2011).
365
First, opening up the innovation system will affect capabilities

through inows and outows of talent. Second, multinational enterprises (MNEs) change the landscape of competition and other incentive
structures in the economy, leading to changes in institutions. Third,
opening up the innovation system at both the national and rm levels
also increases the importance of cross-border innovation collaboration,
transforming the role of developing countries from innovation taker/
acquirer to co-producer. These co-production activities will change the
innovation capabilities of a nation, although the interactions will
be dynamic and the causal relationships may operate in both directions
(Fu and Zhang, 2011).
14.3 Boosting innovation performance in China:

the policy choices
This section analyses the policy choices for boosting innovation performance in China based on analysis of Chinas innovation performance and
efciency reported in earlier chapters, Chapters 2 and 13 in particular.
Capabilities building
China has become one of the major global investors in R&D. This is a
necessary step to promote the transformation process of new innovation. The Chinese government needs to continue to increase its investment in R&D, particularly in basic research. As noted earlier, China is
now the third largest investor in R&D in the world. Moreover,
comparing China with the OECD countries, the proportion of R&D
investment by the private sector in gross domestic R&D expenditure
in China has increased rapidly, from 60 per cent in 2003 to 74 per cent
in 2011, and has taken over that of the major industrialised economies. For example, the proportion of R&D investment by the private
sector in gross domestic R&D expenditure was 60 per cent in the
United States in 2011, having fallen from a higher level of 70 per cent
in 2003.1 Therefore, in terms of encouraging private sector
1
The data come from Gross domestic expenditure on R&D by sector of

performance and source of funds collected from the OECD website (http://stats.
oecd.org). We cross-checked the China data in The 2nd National R&D census
available from the National Statistical Bureau website (http://www.stats.gov.cn),
indicating 72 per cent for China in 2009.
366
investment in R&D in China, the business environment has improved

over the past decade. In addition to having substantial investment in
innovation, China has a large pool of trained scientists and engineers.
The number of university graduates in science and engineering in
China is the highest in the world, with 6.7 million graduates receiving
undergraduate and postgraduate degrees in 2012.2 This is also an
advantage for Chinas innovation, particularly in terms of promoting
the diffusion of innovation. Although the proportion of scientists and
engineers in the total population in China is still comparatively low,
at only 0.03 per cent in 2008,3 their talents provide China the resources to make some breakthroughs in selected focused industries or
disciplines, especially in the manufacturing sector.
The bottlenecks in Chinas capabilities lie in the creativeness of
researchers and the labour force. Creativity is difcult to measure. In
general, from a learning perspective, the Chinese education system emphasises respect for and attention to existing knowledge and doctrine, rather
than fostering critical thinking and challenging existing limits. This
emphasis may hinder the development of creativity for the younger generation (Li, 2006; Zi, 2013). Therefore, reforming the current education
system is a crucial step in the development of national innovation capabilities over the long run. The development of creativity starts with nursery
education, but the reform process should start by making changes to the
university admissions process, followed by the exam system at middle and
primary schools. This is a long-term process that needs to start now.
China should also follow the best international practices in education,
not only the experiences from the United States and Europe but also from
Japan and Korea. Given the existing substantial literature and debate on
education reform in China (e.g., Cheng, 2004; Shi, 2008; Wu, 2013), the
present chapter will not expand the discussion in this regard.
Enhancing the skills of the labour force through on-the-job training
should be prioritised. Innovation in the modern industrial sector necessitates increases in training and re-training. Because of the presence of
externalities of knowledge and the lack of resources for training in the
private sector, especially in the SMEs, workforce training decisions and
2
This is calculated from data collected from the National Bureau of Statistics of
China website, available from http://data.stats.gov.cn.
This is calculated from data collected from the National Bureau of Statistics of
China website, available from http://data.stats.gov.cn.
367
actions cannot be left purely in the hands of business. There is a role

for policy intervention in the skill enhancement of the labour force.
Since 1960, South Korea has insisted that companies spend at least
56 per cent of their total budget on education and training programs,
hence involving the private sector in the education and innovation
process in a meaningful way (McMahon, 1987).
In addition to the long-term or medium-term effects of education and
training, the utilisation of international knowledge and talent will have
both immediate short-term and long-term effects. China has started
several programmes to attract highly skilled international researchers
since 2008 and should continue to do so, while also paying attention to
the necessary follow-up support. More research is needed on knowledge
transmission mechanisms to transfer information, knowledge and skills
from returnees to locally trained staff.
China needs to introduce effective foreign direct investment promotion policies to attract MNEs to invest more in R&D activities.
Although there have been news headlines regarding MNEs, such as
Microsoft having R&D labs in China and India, several pieces of
research commissioned by the EU nd that there has been no hollowing
out effect on R&D activities within the EU (EC, 2013). In other words,
most of the frontier R&D in the MNEs is still based in their headquarters and in OECD countries. The R&D intensity of the MNE
subsidiaries in China is still not high.
Another constraint to address is the inequality in access to innovation
resources and the need for greater support of SMEs (which form the
most dynamic sector in the economy), as well as the private sector in
general (the most efcient innovator in the economy). According to the
EC (2013), SMEs have been the major driver of innovation growth in
EU countries. In China, state-owned rms are currently the major
beneciary of government R&D programs. These rms also have
much better access to bank loans than the private and SME sectors.
Reform of the nancial sector started recently in China, including
liberalisation of interest rates. However, reforming the nancial sector
is a complex task that will take time to accomplish. Moreover, the
constraint in access to nancial resources by SMEs is a widely recognised problem even in developed economies. Therefore, even with a
more liberalised nancial sector in China, the government needs to set
up targeted SME innovation funds and information support systems to
promote the innovation activity of SMEs.
368
Strengthening incentives at multiple levels

As discussed in Section 14.2, the biggest gap between China and the
most innovative countries is not physical infrastructure and investment,
but the lack of effective incentives to stimulate efforts and creativity for
novel innovation and to guide efcient allocation of resources, as well
as the lack of soft skills in terms of innovation management and policy
co-ordination.
Incentives are of crucial importance in guiding physical and human
resources in the economy towards the innovative sectors and innovative
rms. Incentives act to stimulate the expansion, renewal or disappearance of resources. Moreover, incentives at the micro level will greatly
stimulate the efforts of researchers, managers and employers; reduce
X-inefciency4; and determine the efciency of innovation. Despite high
investment in R&D, Chinas incentive structure is not well developed or
set appropriately, in the following aspects.
1. Greater role of the market in incentivising
investment for innovation
The Chinese government needs to modify its role and learn to coordinate resource use more efciently using a diverse set of tools.
Problems in the funding allocation and administration system (to be
discussed later) have led to inefcient use of R&D resources. Sometimes
excessive intervention by various local governments leads to investment
behaviour that seriously distorts market demand: for example, the lack
of co-ordination among local governments and, hence, the overinvestment in the solar photovoltaic (PV) industry in China have led to a
serious surplus in production capacity in this industry. This recommendation does not mean a full withdrawal by the state but, rather, less
direct intervention than at present. Moreover, the government should
extend the allocation of public funding more widely to include the
private sector and SMEs and also use indirect tools in the labour market
or in product markets to ensure that skills and innovative products are
protected, for example, by inuencing the price of and demand for
skilled labour and innovative new products, such as subsidising
customers using solar PV panels. Finally, the Chinese government
4
X-inefciency indicates the inefciency that occurs in the absence of fear of entry
and rivalry; it may occur in a monopoly since there is no competitive pressure to
produce at the minimum possible costs.
369
should more actively employ policies to create an environment that

fosters and encourages innovation in addition to increasing direct
investment in R&D.
2. Enhance competition to motivate innovation
Reform of the economic system and competition policy is needed. Given
the barriers to market entry and the abnormal prots that result from
monopoly power and market failure, capital has tended to ow into
sectors of the economy that are less innovative, such as the real estate
sector and the stock market. Therefore, institutional reforms and marketisation that are intended to create a more efcient capital market will
direct investment to the more innovative and productive sectors.
Similarly, further increases in the openness of the economy and removal
of protection against competition of certain groups of rms, such as
state-owned enterprises (SOEs) in the transportation and telecommunication sectors, will have the same effect. Both Singapore and South
Korea have used an approach whereby the government offers no protection but intervenes heavily in guiding investment (Krause, 1988).
3. Better use of human resource management practices
Better use of human resource management practices is crucial because
of the entrenched unsuccessful practices currently used in many areas
related to innovation in universities and research institutes. Better management practices might include, rst, setting higher wages and rewards
for personnel involved in innovation, not only R&D staff (scientists and
engineers) but also production and marketing staff engaged with new
product commercialisation. People are the crucial players in the creation of new knowledge. Internal brain drain of talent from the innovative sectors to the non-science and non-technology sectors has become a
common problem faced by many countries, including China. This
measure will not only attract more talent to participate in innovation
and innovation-related activities in the business and public sectors but
will also attract more young people to study science and engineering at
universities.
Second, effective human resource management practices should be
introduced to promote creative and innovative activities in rms,
research institutions, universities and government departments. This
includes, in particular, appropriate appraisal and reward/promotion
systems. Of course, the emphasis should vary in different organisations.
370
Creativity comes from researchers and other innovation-related personnel, starting with top management and running through to R&D
researchers and marketing staff. The motivation of each individual to
foster innovation is one of the most important drivers of innovation.
This is also an area in which China needs to improve. For example, the
evaluation of top managers of rms (SOEs in particular) should be
based on their performance not only in terms of the growth of sales or
the total amount of prots but also in terms of the innovations (both
technological and managerial) that the rm has introduced under their
leadership.
With respect to researchers at universities and research institutes,
human resource management policies can play several roles:
1. Give higher weight to the quality than to the quantity of research
outcomes in appraisals. For example, in universities, the number of
publications is the main criterion to evaluate a researchers academic
achievement, whereas the quality of the publications has not
received enough attention. In fact, quality should be the main criterion instead of the quantity of papers, books and patents a researcher
has published. In recent years, there have been changes in some
research-oriented universities in this regard. However, such reform
should be scaled up to the wider academic community in the
country.
2. Use a researchers/institutions ability to attract external funding as
one of the appraisal criteria, but also use output as a more important
appraisal criterion.
3. Set achievable targets that suit high-quality research in research
performance appraisal. Major research performance evaluation
should be carried out every three to four years instead every year.
The emphasis should not be the quantity but the quality of research
publications.
4. Set knowledge transfer as one of the appraisal criteria. In this respect,
China can benet from learning from international best practice. For
example, in UK universities, the Research Excellence Framework
(REF) exercise is carried out every ve years, and only the four
highest quality research outputs are required to be submitted for
review by each academic staff member. Moreover, the REF exercise
requires reporting on the impact that the research has on the
economy and society.
371
5. Tolerate failure and have medium-term or longer-term vision in

appraisals. Three to ve years are necessary for high-quality research
to be published. Although annual evaluations can still take place,
major evaluations require a longer period of time. Of course, subject
specicity should be taken into consideration.
Third, reform the R&D funding grant system and project appraisal
methods:
1. Strengthen the appraisal of research projects.
2. The appraisal criteria should include not only the number of papers
published but also the number of patents; also (a must) the value of
knowledge transfer should be measured either in monetary terms or
in terms of change in behaviour, practice and policy of rms, people
and government.
Of course, for all these new policy and management practices to work,
honesty is crucial. The government should ensure that there is a signicant punishment if the code of conduct is broken. Moreover, a guide
on research ethics should be introduced to researchers, teachers and
students because there has been a weakness and lack of training in this
regard in China (Cao et al., 2013).
4. Improve intellectual property rights protection to strengthen
the internal motives for innovation
Given the high risk and costly nature of innovation, it is crucial to
improve the protection of IPR in China to strengthen the motivation
for innovation of investors and innovators. In the following subsection,
we explain this in detail and examine institutional development.
The incentive bottleneck is currently a major problem that China
faces. Breaking out of it will not only motivate innovative activities in
China but also greatly increase efciency in innovation.
Development of institutions
With respect to the development of institutions in building innovation
capabilities, the most important task for China is to continue to
strengthen the protection of IPR. China has made substantial progress,
which may not be widely recognised (Bound et al., 2013). Nevertheless,
relatively weak IPR protection in China has been widely cited by foreign
372
investors and MNEs as a major barrier preventing them from using the
most advanced technology in their production in China.
The strengthening of IPR protection will promote indigenous
innovation because it will improve the motivation of investors and
researchers to ensure that their investment and efforts will be properly
rewarded and their rights protected. Moreover, it will not only help
China attract R&D-intensive investment by multinationals but also
promote international innovation collaboration between Chinese
universities, rms and world-leading institutes. It will also incentivise
existing MNEs in China to engage more with local rms and institutions for technology transfer. Moreover, correct signals will be passed
to the factor markets and will further direct nancial and human
resources to the innovative sectors.
Investment in R&D, the commercialisation of new technology and
the diffusion of innovation need to be greatly enhanced through using a
scientic method for the pricing of IPR and intangible assets. The
nancing system for innovation is currently based on experience. For
example, the value of a technology and the decision to invest or not in a
technology commercialisation project is made mainly according to the
experience of the venture capitalists. The lack of a widely accepted
mechanism for the pricing of IPR has constrained the banks and other
nancial institutions from providing funding to innovation activities.
This also hinders the trade and diffusion of the existing technology.
With a new pricing system (e.g., with a more transparent and objective
price evaluation mechanism), rms with international patents would
more easily obtain funds for commercialisation. Investors, including the
banks, would have a clearer idea about the value of the innovative
technology to make a more informed decision. Investors in R&D and
owners of intellectual property would have a tool to discover the value
of an innovation outcome and a way to obtain the return at a market
price. This will greatly enhance the incentives and resource use for
innovation. This is a challenge facing rms not only in China but also
in the developed countries. Venture capital and experienced investors
have played a signicant role to ll part of this gap. Nevertheless,
developing a better system to nance innovation and commercialisation
of inventions remains an important task for the governments in the
world. Of course, progress in this area rst requires a breakthrough in
the research and the design of such an IPR pricing system, which is
373
beyond the scope of what government policies can achieve. This should
be a priority area for government support.
China needs to strengthen its technology market. The technology
market is an intermediate market place for technology transfer and
exchange. Although the technology market was introduced in China
during the 1990s, it has not really prospered in the country. Indeed, the
growth of the technology market should develop in parallel with the
strength of IPR protection. Only when IPR and knowledge are properly
valued will the technology market be able to serve as an effective marketplace to facilitate the transaction of technologies. In addition, when
developing a domestic technology market to facilitate the intra-country
diffusion of innovations, China should develop a reputable international
technology exchange market to provide a respected, high-end platform
for international technology transactions and exchange. This is an important institution to enhance the diffusion of innovation. Of course, the
effectiveness of this institutional platform depends on the strength of IPR
protection and the demand from the rms.
Strengthening co-ordination in policy making

and implementation
The strength of the innovation capabilities of a country also depends
on the interplay among the factors discussed earlier. The consistency
of incentives and capabilities, as well as the co-ordination of the
various policies, plays a key role in shaping the strength of the newly
industrialised countries. Admittedly, the Chinese government has
done a good job in treating the process of innovation as system
engineering. A series of complementary policies were introduced in a
sensible sequence to motivate and support innovation (Wu, 2010). A
wide range of government departments have introduced policies relevant to innovation: for example, the Guidelines on Key Areas of
Commercialisation of Prioritised High Technologies, jointly issued
by the National Development and Reform Commission, the Ministry
of Science and Technology, the Ministry of Commerce and the State
Intellectual Property Ofce in 2007, and the Preferential Policies on
Income Tax Related to Technology Innovation, jointly issued by the
Ministry of Finance and the State Bureau of Taxation in 2006.
Attempts at co-ordination in policy making and implementation
have been made. For example, the innovation funds for SMEs are
374
managed with co-ordination between the Ministry of Science and

Technology and the Ministry of Finance under the direct support of
the State Council. The government has also established a national
Science, Technology and Education Leading Group to co-ordinate
the policies between the Ministry of Science and Technology and the
Ministry of Education.
Nevertheless, because of the disparate interests of different government departments and the existence of information asymmetry
between different policy makers, overlaps and conicts among policies
still exist, and there is a lack of co-ordination between local governments (OECD, 2008; Wu, 2010, 2013). Therefore, the Chinese
government should consider, rst, setting up a central innovation
leading group (CILG) led by the prime minister. Its members should
include ministers of all the relevant departments, such as the National
Development and Reform Commission, the Ministry of Science and
Technology, the Ministry of Commerce, the Peoples Bank of China,
the Ministry of Finance, the State Administration of Taxation, the
General Administration of China Customs, the State Intellectual
Property Ofce, the Ministry of Education, the Ministry of Human
Resources and Social Security. Second, the government should set up
an ofce for the CILG. The CILG would meet once every six months
to make strategic decisions on the countrys innovation policy and
strategy and also to undertake co-ordinated initiatives requiring agreement at a high level. The ofce of CILG would be a regular agency
responsible for policy research and implementing, evaluating and coordinating innovation policies within different departments. The
ofce would report to the CILG and would also be responsible for
the implementation of the decisions of the CILG.
Opening up the innovation system

The shift of innovation mode from closed to open innovation, the
globalisation of the innovation system and the increasingly severe
trade environment for China offer both opportunities and challenges.
The Chinese government needs to actively use non-traditional modes to
integrate China into the global innovation system, in particular by
adopting the following measures.
First, the government must strengthen innovation collaboration and
actively participate in international knowledge co-production. It should
375
better provide some specied measures. Innovation has become a global

and collaborative undertaking (OECD, 2008).
Second, the government should encourage international technology
acquisition through cross-border mergers and acquisitions by offering
support not only in a monetary sense but also in providing soft skills
that are needed for successful knowledge acquisition. This could
involve sponsoring research on strategy and the process management
of mergers and acquisitions and also follow-up knowledge transfer to
Chinese MNEs.
Third, the government should gradually open up the innovation
system by actively attracting highly skilled world-leading researchers
for innovation projects in China. China introduced the thousand-talent
program in 2009, intended to attract overseas Chinese and foreign
talent to support innovations in both academia and industry. The
government should continue to invest in such programs and seek to
improve them through careful evaluation of their impacts.
Finally, there should be active participation in international innovation activities through involvement in the global standard-setting
activities of international organisations. Of course, active and
successful participation requires high-calibre talented professionals
who not only have expert knowledge in certain areas of science and
technology but also international vision and excellent communication
skills.
14.4 Space for innovation policy in the twenty-rst century

There remains a large role for the government to play in promoting the
technology and innovation capabilities of a country as the space for
selective industry policy in the twenty-rst century is becoming more
limited. The world trading system is moving towards more free trade
and away from protectionism. Countries are more capable of using
non-tariff barriers, such as anti-dumping and anti-subsidy tools, to
block the imports of goods that they regard as having benetted from
the exporting countrys government protection and subsidies.
Countries are now increasingly using anti-dumping and anti-subsidy
sanctions and non-tariff barriers to protect domestic industries and to
compete against major competitors. China has become one of the major
targets of trade disputes. In fact, some industry policies have been
rebranded and transformed into innovation policies, for example,
376
policies supporting innovation networks or policies encouraging technology transfer instead of protectionist industrial policy based on the
infant industry argument. Policy has been transformed from selective
picking of winners and protectionist policies into horizontal policies
that promote capabilities across sectors, such as those that improve
infrastructure, institutions and universal capabilities building.
Therefore, we have the following recommendations for policy design.
First, it is advisable to use pull rather than push policy measures to
attract R&D-intensive MNEs and highly skilled talents to China.
Examples of pull policies are improvements to IPR protection and the
competitive market environment. The policy space for forcing MNEs to
transfer technology and locate innovation activities in China has been
limited according to the World Trade Organisation (WTO) rules since
Chinas WTO entry in 2002. In this regard, continued strengthening of
IPR protection in China is key to the removal of this bottleneck.
Another relevant area involves helping MNEs and returning or foreign
talented professionals to quickly integrate into the local research and
social community and provide them with continued support after
relocation.
Second, it is advisable to use more horizontal policies (e.g., policies
supporting education, R&D, and improvement of infrastructure and
institutions) rather than selective policies to promote the growth of
certain industries. Selective policies, especially those through supplyside inputs (e.g., subsidies) are more likely to be a subject of trade
disputes. Of course, this does not mean the abolition of selective
policies. Rather, it means refraining from providing subsidies and
preferential policy based on selective supply-side approaches. The
Chinese government should allocate more resources and pay attention
to policies that encourage demand, develop markets and improve the
institutional environment. China should closely watch and take note
of the relevant new developments in innovation policies that are
introduced by other industrialised countries (e.g., the innovation platform, dedicated knowledge transfer funding programs and internship
schemes) and should introduce them into China with the appropriate
adaptation.
Third, government policy should focus on activities that suffer from
market failure: for example, basic research that faces high risks and
uncertainties and requires long-term investment to be sustained. The
government should refrain from intervening in market-led activities
377
such as the majority of the commercialisation process, especially the

scaling up of production activities.
14.5 Conclusions
Building an innovation-driven economy has been highlighted by the
Chinese government as one of its major tasks at the current stage of
development of the country. In the present chapter, we examined the
policy choices facing China when considering the goal of transforming
from an imitator to an innovator. We have approached this question
from the perspective of capabilities, incentives and institutions within
the innovation capabilities framework. The analysis is subject to the
limitation that the chapter only covers a few of the determinants of
national innovation capacity in a wide range of macroeconomic, microeconomic, political, social and even cultural factors.
Analysis presented in this chapter suggests possible policy recommendations to enhance Chinas innovation performance, including the
following.
First is to maintain the momentum of rapid growth in R&D investment and to ensure reforms will lead to more private sector investment
growth; a wider allocation of the resources to the private sector and the
SMEs sector; and a more efcient, market-driven funding allocation
process.
Second is to encourage fair competition through reforms of the
economic system, which should play a signicant role in incentivising
innovation at the rm level and in guiding the resources owing to the
innovation sectors to seek innovation rents instead of monopoly rents.
Third is to set appropriate incentives at multiple levels by introducing
new, or modifying existing, human resource and funding management
practices. This appears to be the key area for policy intervention at the
current stage. The policy options include correcting misleading policies
in appraisal and reward systems and in the management of research
funding, as well as the introduction of new policies including heavy
punishment for academic and professional dishonesty and wider criteria that will have an impact on the evaluation of researchers, projects
and institutions.
Fourth is to encourage and facilitate international knowledge sourcing and co-production using unconventional methods. These unconventional methods include direct investment abroad, attracting highly
378
skilled international talent and international innovation collaborations.

China has made appropriate efforts in this regard and should continue
to do so by rening the management process of these activities to ensure
that Chinese rms will genuinely benet from these methods. All these
policies, if well designed and properly introduced, will have a signicant
impact on the propensity for, and efciency of, innovation in Chinese
rms and research institutions.
Fifth is the creation of institutions to co-ordinate the efforts and
policies issued by different government departments; enhance the interplay between capabilities, incentives and institutions; and avoid possible contradiction and conict among policies. In particular, the Chinese
government should consider setting up a CILG led by the prime minister
and involving ministers of all the relevant government departments. A
regular agency such as the ofce of the CILG could also be set up to
carry out the research, undertake assessment of innovation policies, coordinate the innovation policy of the different government departments
and implement decisions made by the CILG.
Despite the important role that the state can play in fostering innovation capabilities, we have to realise that the space for innovation
policies in the twenty-rst century is smaller than what existed when
Japan, Korea and other newly industrialised economies emerged.
Therefore, the Chinese government should follow the global trend and
use more horizontal policies rather than old selective policies that
favour a few industries. More policies that address the pull factors
should be used because the space for using push policies is becoming
increasingly limited with respect to international knowledge sourcing
and also in terms of promoting innovation in domestic rms.
Finally, we would like to highlight our expectations of government
policy. As Stiglitz (1989, p. 202) argues, governments face information
and incentive problems no less than does the private sector . . . We need
to recognise both limits and strengths of markets, as well as the
strengths, and limits of government interventions aimed at correcting
market failures.
Conclusions
15
Conclusions: Open national

innovation system and Chinas path
to innovation
Chinas industrial capabilities have increased signicantly since the

1980s. Transforming the country into an innovation-driven economy
has now become its top priority. This book attempts to provide a
systematic, comprehensive and rigorous study of Chinas path to innovation from the 1980s to the present and for the future. This chapter
summarises the major ndings of the study and discusses the policy
implications for China and other countries.
Open national innovation system and Chinas path

to innovation
In the main, evidence reported in the book suggests that China has
followed a path of an open national innovation system (ONIS) since
the reforms and the take-off stage in the 1980s. An ONIS is a national
innovation system (NIS) that is opened up to international knowledge,
resources and markets. This model has the following characteristics:
(1) It has dual knowledge sources and uses a dynamic combination of
foreign and indigenous innovation that evolves over time. It has an NIS
making efforts to foster indigenous innovation capabilities while opening up to external knowledge or resources for knowledge production at
all stages of the innovation chain, from research to development to
commercialisation. (2) It has multiple driving forces: the state, the
private sector and the MNEs and uses a combination of the state and
the market to guide the direction of innovation and allocate the resources. (3) It encompasses both the outside-in and inside-out branches of
the open innovation model, though the inside-out branch becomes more
relevant at the later stage of technology development. (4) It employs
multiple knowledge diffusion channels and uses a combination of multichannel, multi-layer knowledge sourcing strategy, which is again
dynamic, evolving at different stages of technology development for
381
382
Conclusions
international and domestic knowledge diffusion. These characteristics

will be elaborated in the following sections of the chapter.
Before the reforms in the 1980s, the science and technology system
in China was a closed innovation system developed following the
former Soviet Union model (Xue, 1997). Since the opening up and
system reforms in the 1980s, following the ONIS path, Chinas
national innovation system has had a high degree of openness to
external knowledge from the start. There has been investment in
indigenous innovation over the years but with varying efforts at different stages of development: a small amount at the start that was mostly
spent on absorption of foreign technology and development, followed
by a lane change to emphasise indigenous innovation starting in the
late 1990s, after about two decades of openness, while maintaining
high openness to external knowledge. The system has even become
more active in knowledge sourcing, using unconventional channels
that are not often used in developing countries, for example, outward
direct investment, international innovation collaboration and attraction of highly skilled migrants.
In principle, this is an appropriate path for a developing country
along its development trajectory. China succeeded in embarking on a
fast industrialisation trajectory as a result of this technological development strategy. However, the emphasis on indigenous R&D and
innovation started later than it should have: policies encouraging
indigenous innovation were introduced around 2000 when the
dependency on foreign core technology acquisition had become a
signicant constraint in some industries. In comparison, although
the model of innovation and technology development in Japan and
South Korea is widely regarded as an inward looking, indigenous
capability buildingoriented model supported by strong employment
of import-substitution strategy and industrial policy, in fact, Japan
and South Korea have heavily relied on foreign technical assistance
and aid provided by the Western countries, especially after World
War II and the Cold War. In terms of knowledge sources for innovation and technical progress, they have not closed their doors, although
they have closed their markets to foreign competition. Moreover, in
the history of Japans industrial catch-up, technology transfer from
MNEs has also have played its role (Cantwell, 1992; Cantwell and
Zhang, 2009). Although MNEs have not played a prominent role in
the catch-up process in South Korea and Taiwan, international
Conclusions
383
knowledge transfer through diasporas and highly skilled returned

migrants has been crucial in building up local technological capabilities in these two economies (Dahlman, 1994; Wong, 1999).
Research presented in this book also suggests that ONIS is not a static
but a dynamic model, evolving with the advancement of the innovation
capability of a country. Leading Chinese rms have started to use unconventional methods of international knowledge sourcing, for example,
international innovation collaboration with foreign partners including
foreign universities and direct cross-border investment and acquisition.
Although the full effect may need some time to appear, these unconventional knowledge sourcing activities are signicantly associated with a
greater likelihood of creating radical novel innovations in the Chinese
rms. International operations have also allowed some Chinese MNEs to
upgrade their overall capabilities and competitiveness, as the case studies
of Huawei and ZTE reveal. In addition, China has also successfully
sourced international knowledge through human ows and social networks. In the 1980s and 1990s, diasporas have been important foreign
investors in China, through which superior technologies (though not the
most advanced) were transferred to China (Fu, 2004). Since 2000, especially the introduction of the indigenous innovation strategy in 2006,
highly skilled diaspora workers and many Chinese students educated
abroad returned to China. They have played an important role in the
development of some high-technology industries and the so-called strategic emerging industries such as renewable energy, electrical cars and
biotechnology (Wang, 2012). This type of sequencing is also observed
among the East Asian economies. As Lee, Jee and Eun (2011) nd, at the
macro and aggregate levels, China follows the East Asian sequencing
rather than the Washington Consensus.
Therefore, for the way ahead, a rational strategy for Chinas path to
innovation will be to continue to enhance innovation by harnessing internal and external resources as suggested by the ONIS model, and guiding it
with a well-dened incentive structure and fully developed institutions.
Relationship between foreign and indigenous innovation

in the ONIS model
The ONIS model suggests that opening up to foreign technology transfer and investment in indigenous innovation should go hand in hand.
Despite the potential offered by globalisation and a liberal trade regime,
384
Conclusions
the benets of international technology diffusion can only be delivered

with parallel indigenous innovation efforts and the presence of modern
institutional and governance structures and a dynamic innovation
system. In this sense, indigenous and foreign innovation efforts are
complementary. Without proactive indigenous innovation efforts, foreign technology remains only static technology embedded in imported
machines that will never turn into real indigenous technological capability. This conclusion is compounded by the expected inappropriateness of technologies created in the industrialised economies for
countries in the developing South, which calls for greater efforts to
develop indigenous innovation crucial for technological change and
catch-up, especially in the middle-income countries.
As regards diffusion and adoption then, rst, technology diffusion
and adoption are not costless and unconditional. The speed of diffusion
and adoption, and thereby of technological capabilities building,
depends on the rms absorptive capacity and complementary assets.
Empirical evidence from Chapter 3 illustrates this and is supported by
evidence from emerging economies (Fu, Pietrobelli and Soete, 2010).
Second, only in the presence of local innovation capacity will MNEs
adopt a more integrated innovation practice, which has greater linkages
with the local economy and thereby enables greater opportunities of
knowledge transfer (Franco et al., 2011). Hence, the encouragement of
indigenous R&D and innovation activities remains an indispensable
centrepiece of an innovation strategy targeting the assimilation and
adaptation of foreign technology and the acceleration of technological
learning and capabilities building.
Admittedly, developing countries face a dilemma of resource constraints to meet the high investment costs and high-risk challenges of
innovation (Erdilek, 1984; Hoekman et al., 2005). Experience from
China suggests that to maximise the benets from innovation and
accelerate catching-up, the explicit and well-focused encouragement
of indigenous innovation and acquisitions of foreign knowledge must
work in parallel. Neither autonomous innovations nor FDI-reliant
strategies can be used independently. Relying solely on one of them
would not be optimal for technological capability development and
catching-up. The Chinese ONIS model proposes a strategy to maximise
the benets for the developing country. How to select and shape the best
combinations at different stages of development and for different countries and industries is a question of utmost relevance for future research.
Conclusions
385
Evolution of the indigenous-foreign combination at different

stages of development
The ONIS model is a dynamic model with continuous adjustment of this
indigenous-foreign combination at different stages of development. It
suggests a high openness to external knowledge from the early stage of
technological development of a country and that indigenous innovation
capacity building should also start from an early stage instead of purely
depending on foreign technology transfer. However, given the resource
constraints and the skills available in the domestic economy of the
developing countries, the emphasis on indigenous innovation could
increase with the level of economic and technological development of
the developing country. The channels used for foreign technology transfer should also evolve with the advancement of technical progress at
different stages of development. Along Chinas path to innovation and
technology upgrading, imports of machinery and equipment and transfer of technology through licensing have been major channels of foreign
technology acquisition in the 1980s. Direct adoption of these technologies and learning how to use the imported machines were the main
efforts made by domestic rms and institutions.
In the 1990s, technology transfer through direct foreign investment
has been a new focus. Huge amounts of FDI since then have owed
into China. Chinas overall industrial and technological capabilities
increased signicantly, supported mainly by those subsidiaries of
MNEs especially in the high-technology industries. However, although
there has been some learning by the indigenous partner in the joint
ventures and spillovers to local companies in the same and linked
industries, as evidenced in the book, the benets to the technological
and innovation capabilities from the FDI are limited. Therefore, since
the late 1990s, acceleration of science and technology development has
been given priority policy attention. Since 2006, indigenous innovation
has been formally stated as one of the governments strategic priorities.
At the same time, the channels used by China for foreign knowledge
sourcing also started to change as analysed in Part III of the book. First,
the Chinese government modied its inward FDI policy to attract
knowledge-intensive FDI. Second, the go global strategy was introduced in 1998. Since then, China has started a more active international
knowledge-sourcing strategy. Third, the intensity of international innovation collaboration increased rapidly, especially in recent years when
386
Conclusions
more specic funding programmes were set up to encourage internationally collaborative research.
The experience of the solar-PV industry in China and India as
analysed in Chapter 10 provides a classic example of catch-up and
leapfrogging using a strategy of mixing and sequencing of different
technology transfer and indigenous innovation mechanisms at different
stages of technology development. Most leading solar-PV companies in
both countries started from international technology transfer through
licensing and joint ventures with MNEs. At the same time, all of them
have put increasingly substantial effort into in-house R&D for the
assimilation and adaptation of transferred technologies and the development of indigenous technological capabilities. However, once the
basic production and technological capabilities were built up, they
started more active knowledge acquisition and creation through indigenous innovation, international R&D collaboration and cross-border
mergers and acquisitions. Such experience suggests that to accomplish
such catch-up processes requires a combination of international technology transfer and indigenous innovation. Technology transfer will be
a feasible and evidence-proven entry point for developing countries,
although this should be accompanied with substantial indigenous
efforts in assimilation and learning. In this respect, attractive bundles
of trade, investment and technology policies are important mechanisms
to leverage technology transfer.
The effectiveness of knowledge-sourcing channels

at different stages
Given the availability of a variety of channels for international knowledge sourcing, choosing the most effective channels for countries of
different industry structures and at different stages of development is
important. Chinas experience in the past three decades offers some
examples and lessons.
First, international knowledge and innovation exchange and collaboration, through inter-rm and intra-rm networks and global value
chains (GVC), is now regarded as an important route for developing
country rms to access knowledge and enhance learning and innovation. Integration into the GVC through engaging with processing and
assembly activities is a widely used method by many developing countries, with the expectation of substantial learning and consequent
Conclusions
387
development of technological capabilities in the local rms in related

industries. The evidence from this book demonstrates that processing
trade-FDI has generated signicant positive information spillover effect
(e.g., market intelligence, export management) on the export performance of indigenous rms. However, the technology spillover effect on
the development of international competitiveness in indigenous rms is
limited and in fact exerts a signicant depressive effect on the propensity
to export in these rms. Indigenous innovation, economies of scale and
productivity are found to be the main drivers of export performance in
indigenous rms in the high-technology industries.
Chinas experience in the acquisition of tacit knowledge within the
Chinese optical bre and cable industry also suggests that in industries
where tacit knowledge is a more important component of technological
learning than codied knowledge, internal R&D activities and domestic
peers are important knowledge sources. Additionally, universities are
shown to be an important asset in creating learning organisations and
providing effective knowledge sources of both tacit and codied knowledge. However, imports of equipment and licensing are a less effective
learning channel in the acquisition of tacit foreign technology.
Second, the experience of Chinas leading technology-intensive rms,
Huawei and ZTE, and evidence from a survey of rms in Guangdong
Province on the impact of going global indicate that rms can benet
from employing more active knowledge-sourcing channels for catch-up
at a more advanced stage. For example, international operations
allowed some of these leading Chinese MNEs to upgrade their overall
capabilities, not only technological capabilities but also design and
marketing capabilities for greater control at the two ends of value
chains, where most of the value added is generated. Moreover, international innovation collaboration with foreign partners including foreign
universities and direct cross-border investment and acquisition are signicantly associated with a greater likelihood of creating radical novel
innovations in the Chinese rms.
Leading players in the ONIS model: is the Chinese ONIS

model state led?
Findings of the book also indicate that although the state-owned enterprises (SOEs) have been a major force in the Chinese economy, especially before and at the early stage of the reforms, in respect to
388
Conclusions
innovation, the foreign invested and domestic privately owned rms

are not only investing more in R&D than the SOEs, they are also
more efcient in innovation and are producing more innovative outputs. Moreover, instead of having the SOEs taking a lead in pushing
forward the technology frontier among Chinese rms, foreign invested
(FIEs) and privately owned (POEs) enterprises are the leading players
in the high-technology and low- and low-medium technology industries, respectively. That said, SOEs remain a leader in the high-medium
technology industries. So the SOEs, POEs and FIEs all have played their
roles in Chinas technology upgrading and innovation. Therefore,
instead of a pure market-driven model of innovation or the often
assumed state-led model of innovation, Chinas path to innovation
follows a multi-driver model led by a mix of players the state, the
private sector and the MNEs, with each of them playing a leading role in
different segments of the economy and the innovation system.
Therefore, the ONIS model in China is led by a combination of the
state and the market, although the actual practice so far is neither
perfect nor mature.
Of course, the multiple-driver model of ONIS does not undermine the
role of the state in ONIS and the role of policies in promoting innovation. Analysis of Chinas innovation policy in the past in Chapter 6 and
policy choices for the future in Chapter 14 both suggest a role for the
state in guiding and supporting the innovation process. The nature of
innovation as public goods, the externalities that it may generate for
long-term growth and its risky and costly nature all suggest a role for the
state. Therefore, the state may play a more important role in areas of
greater externality (e.g., supporting basic research) and in areas of
greater capital intensity and uncertainty (e.g., the capital-intensity
heavy industries). However, in areas where the market functions well
(e.g., commercialisation and labour-intensive industries), the private
sector will be more efcient in guiding the allocation of innovation
resources and incentivising innovation.

and the role of universities
Developing country rms often face resource, capabilities and institutional constraints for innovation. Evidence from the book shows how
Chinese rms use open innovation as a response to the constraints and
Conclusions
389
risks of innovation that they face. It is found that institutional, nancial

and knowledge- and skills-related risks and constraints are all signicantly associated with rms depth and breadth of openness in innovation. The responses, however, vary across rms of different ownership
types. Foreign invested rms appear to be most responsive and take
action to widen and deepen their openness in innovation. Privately
owned rms have made signicant responses to market- and institutionrelated and nance- and risk-related impediments but not to knowledge- and skills-related impediments. State-owned rms appear to be
least responsive to the use of open innovation. Firm size and industryspecic effects also appear to have signicant moderating effect on
rms responses to the various constraints. These ndings are consistent
with those from an in-depth study of the Chinese semiconductor industry by Teece and Chesbrough (2005).
Among the various possible partners that rms can collaborate
with for innovation, universities are ones that have received substantial attention. Research reported in the book nds that domestic
universities have played a signicant role in the promotion of the
diffusion of frontier technology and the creation of new products or
processes that are new to the country or the rm, although their
contribution to the creation of groundbreaking innovations is limited
in China. International innovation collaboration with foreign universities, especially those in the newly industrialised economies and
the emerging South, appears to be fruitful in enhancing the creation
of groundbreaking innovations in Chinese rms. The experience of
the Chinese optical bre and cable industry also suggests that universities are an important asset in creating learning organisations
and provide effective knowledge sources for both tacit and codied
knowledge.
Incentives and institutions and the efciency of innovation

Heavy investment in R&D is not equal to a strong innovation performance. There is a need for appropriate incentives to guide the use of the
resources and stimulate their expansion, renewal or disappearance to
transform potentially strong into actually strong performance.
Institutions also affect this by setting the rules, modifying behaviour
by changing attitudes and expectations, altering capabilities and changing incentives (OECD, 1987).
390
Conclusions
Research on the innovation potential and actual performance of the

advanced economies indicates that institutional factors such as competition, openness and international property rights (IPR) protection are
signicantly associated with the patenting efciency of an economy.
Comparing China and the OECD and other emerging economies in
recent years reveals that the gap in patenting performance between
China and the world frontier is the result of relative underperformance
mainly in the efciency of innovation production. Therefore, developing institutions and reforming incentives are important areas to enhance
innovation performance in China. In fact, evidence at sub-national level
from Chinese regions suggests that the strength of technology spillovers
to the domestic Chinese rms depends on the availability of absorptive
capacity and the presence of innovation-complementary institutions in
the host region. The type and quality of FDI inows and the strength of
local absorptive capacity and complementary assets in the host regions
are crucial for FDI to serve as a driver of knowledge-based development. These innovation-complementary institutions include not only
technology exchange markets but also training and nancial institutions for innovation in start-ups and SMEs, for example, incubators,
IPO market, venture capital, polytechnic institutions and information
provision services for SMEs.
One of the areas that is of particular importance for the ONIS to
work is the introduction, development and the implementation of
regulation on IPR protection. Given the high risk and costly nature
of innovation, effective protection of IPR will not only incentivise the
investors and innovators to invest in innovation but also will ensure
that the owners of advanced technology that their entitled rights are
protected when they transfer and use advanced technologies in foreign
countries or invest in R&D in foreign host countries. Relatively weak
IPR protection in China has been widely cited by foreign investors and
MNEs as a major barrier preventing them from using the most
advanced technology in their production in China, although substantial progress has been made in the country (Bound et al., 2013). Of
course, it is also argued that the requirement made by the industrialised countries and the MNEs for strong IPR protection in host
developing countries is actually somewhat unfair given that Western
countries also have practised imitation in their own catchup several
centuries ago (Chang, 2003). Therefore, the design of the IPR protection regulation may need to take into account the nature of the
Conclusions
391
technology and the stage of development of the recipient country.

In-depth research in this area is needed.
The role of South-South technology transfer

in the developing world
The analysis of the role of indigenous innovation and foreign technology presented in the book points to the inappropriateness of foreign
technology in developing countries. Supplementing the earlier literature
on appropriate technology and directed technical change that discusses
the general appropriateness of foreign technology at the aggregate level,
research in this book moves the analysis down to the industry level,
taking into account the different characteristics in different industries.
It argues that technologies created in labour-abundant countries may be
unskilled-labour augmenting. In low-technology industries that use
unskilled labour intensively, labour-augmenting indigenous technology
will be more efcient than foreign technology. In contrast, foreign
technology from industrial countries will be skilled-labour augmenting,
and it will be more efcient than indigenous technology in the
technology-intensive sector that uses skilled labour intensively. This is
testied to by the empirical evidence presented in Chapter 5, where
collective indigenous innovation efforts are found to be a major driver
of indigenous technical change in China, and the indigenous rms are
the leading forces on the technology frontier in the low- and mediumtechnology industries, whereas the MNEs dominate the frontier of the
high-technology industries.
Therefore, middle-income countries such as China can reap the gains
from South-South technology transfer, and the less developed countries
can benet from acquiring more appropriate technology. In fact, SouthSouth technology transfer and the possible gains from it are highly
consistent with the inside-out branch of open innovation. The innovative South can reap the gains of innovation from commercialisation of
the innovative ideas/technologies not only in the domestic market but
also in the market of other developing economies. Moreover, technologies created and rst used in China and other middle-income countries
may also later be diffused to the industrialised countries through reverse
innovation (Immelt, Govindarajan and Trimble, 2009). Such SouthNorth technology transfer is also benecial to producers and customers
in both the South and the North.
392
Conclusions
Findings of the book suggest that there are multiple and multi-tier
choices of technology rather than the simple bi-dimensional NorthSouth divide. Technologies developed in labour-rich emerging economies will be more appropriate to the factor endowments mix in other
populous developing countries; technologies created in land- and
resource-rich emerging economies will be more appropriate to other
land- and resource-abundant countries. They will also be easier to
diffuse and absorb by other local rms. Following this rule of thumb,
South-South trade and FDI will represent effective vehicles for the
diffusion of these technologies, and policies should consequently follow
suit on a consistent basis.
Growth and developmental impact of innovation in China

Innovation is one of the most important drivers of long-term economic
growth. It is not only relevant for the developed economies but also
important for developing countries. Evidence based on regional data
from China presented in Chapter 3 of the book attested to its role as an
important determinant of economic growth in Chinese regions.
Innovation also has important developmental effects. The unbalanced
development in innovation and technological capabilities across
Chinese regions has been a factor contributing to the increasing regional
income inequalities in China. A review of the recent development of the
regional distributions of R&D expenditure and patents documented in
Chapter 2 suggests the increasing regional gap in innovation capability
between the coastal and the inland regions in China. As innovation is
increasingly playing an important role in Chinas economic growth, we
can expect the pressure of continued and increasing regional income
inequalities amongst Chinese regions. Therefore, how to accelerate the
domestic diffusion of innovation from the leading to the following
regions and how to encourage the type of innovation that leads to
more inclusive and sustainable development are important topics for
future research.
Policy choices for China at the crossroads

For China at the crossroads of structural change and great transformation into an innovation-driven economy, there are several policy priorities for the Chinese government to consider.
Conclusions
393
First, China should continue to increase its investment in R&D and in

education. The government should maintain the momentum of rapid
growth in R&D investment and to ensure that reforms will lead to more
private sector investment growth; a wider allocation of the resources to
the private sector and the SMEs sector; and a more efcient, marketdriven funding allocation process.
Second, an attempt should be made to strengthen the incentive system
at the macro, meso and micro levels. This strengthening may include
reforms to release the power of competition and guide resources
towards innovative sectors; adopt appropriate human resource management, such as appraisal and remuneration systems; create effective
policies for research funding management; and evaluate the efciency of
research to encourage the creativity of researchers, managers and
employees.
To be more specic, the Chinese government should encourage fair
competition through reforms of the economic system, which should
play a signicant role in incentivising innovation at the rm level and
in guiding the resources owing to the innovation sectors to seek
innovation rents (i.e., the Schumpeterian rents that occur during the
period of time between the introduction of an innovation and its successful imitation and are earned by innovators) instead of monopoly
rents. China also needs to set appropriate incentives at multiple levels by
introducing new, or modifying existing, human resource and funding
management practices. This appears to be the key area for policy
intervention at the current stage. The policy options include correcting
misleading policies in appraisal and reward systems and in the management of research funding, as well as the introduction of new policies
including heavy punishment for academic and professional dishonesty
and wider criteria that will have an impact on the evaluation of
researchers, projects and institutions.
Third, China should encourage and facilitate international knowledge sourcing and co-production using unconventional methods, as it
has already done through policy and nancial support. These unconventional methods include direct investment abroad, attracting highly
skilled international talents and international innovation collaborations. China has made appropriate efforts in this regard and should
continue to do so by rening the management process of these activities
to ensure that Chinese rms will genuinely benet from these methods.
All these policies, if well designed and properly introduced, will have a
394
Conclusions
signicant impact on the propensity for, and efciency of, innovation in

Chinese rms and research institutions.
Finally, China should consider creating institutions to co-ordinate the
efforts and policies issued by different government departments;
enhance the interplay between capabilities, incentives and institutions;
and avoid possible contradictions and conict among policies.
Therefore, China and other developing countries should set up a highlevel policy co-ordination group headed by the top leader of the country
and composed of the heads of the relevant ministries. The strategic
policy making and policy co-ordination of such a group will be more
effective than a post such as a chief science and technology advisor in the
government.
Overall, China is on the right track with its pursuit of becoming an
innovative nation, although some areas need improvement, as previously discussed. What is also crucial is the determination and patience
of all key players as innovation and capabilities development are never
achieved in a short time.
Limitations and issues for future research

This book has conducted a comprehensive and systematic analysis of
Chinas path to innovation in the past three decades. Nevertheless, there
are still areas that have not been discussed systematically in this book
given the fact that innovation is sophisticated system engineering. These
areas include the role of human mobility in international knowledge
transfer, the role of IPR protection and innovation nancing. Several
important issues await further research. First is the impact of the incentive structure on the propensity, intensity and quality of innovation at
macro, meso and micro levels. At the macro level, what is the impact of
competition and marketisation on innovation in a transition economy
such as China? What is the role of the state in the innovation system and
in what areas, using what model, to what extent and in what sequence
should policies be introduced to correct market failures? At the meso
level, what are the factors that affect the efciency of innovation at
project and industry levels? What incentive structure should be introduced in the non-private sector? Where are the constraints or disincentives at the micro level towards innovation in China? Would a direct
introduction and implementation of the incentives, for example, stock
options, used in the Western countries be effective in promoting the
Conclusions
395
propensity and intensity of innovation in rms and non-business sectors? What is the impact of culture on fostering innovation?
Second, the next stage of the transformation from imitation to innovation in China requires the harnessing of internal and external resources for more and better innovation. Challenges also exist for greater
capabilities of commercialisation of innovation. All this requires greater
capabilities to integrate knowledge from different sources and from
different disciplines. To fully understand this process, rms need to
know what strategies and pathways they can adopt: whether to acquire
or to co-produce, how to choose the best partner, how to develop the
capabilities to manage such sophisticated integration, how to share the
value added between partners and how to protect intellectual property
rights created in these open collaborative innovation activities. These
are all important areas for future research to ensure that Chinas pursuit
to build an innovation-driven economy is successful and is undertaken
along an efcient and sustainable path.
Third, the role of nancing and new nancial institutions for innovation, especially for entrepreneurial technology start-ups, is likely to be
of growing importance now and in the future. These institutions include
venture capital, IPO markets and institutions that support innovations
in SMEs and private enterprises in addition to the science parks and
incubators. Therefore, research on reforms of the current statedominated banking system and state-driven venture capital funding
system is also of urgent need. Finally, as discussed earlier, understanding the nature of inclusive innovation and developing policies that
encourage the type of innovation that leads to more inclusive and
sustainable development are important tasks for future research.
References
Acemoglu, D. (2002). Directed technical change, Review of Economic

Studies, 69(4), 781810.
and Zilibotti, F. (2001). Productivity differences, Quarterly Journal of
Economics, 116(2), 563606.
Luc, A. and Varga, A. (2002). Patents and innovation counts as measures
of regional production of new knowledge, Research Policy, 31(7),
106986.
Acs, Z. J. and Audretsch, D. B. (1990). Innovation and Small Firms.
Cambridge, MA: MIT Press.
Adler, N. and Golany, B. (2001). Evaluation of deregulated airline networks
using data envelopment analysis combined with principal component
analysis with an application to Western Europe, European Journal of
Operational Research, 132(2), 26073.
Aghion, P., Bloom, N., Blundell, R., Grifth, R. and Howitt, P. (2005).
Competition and innovation: an inverted U relationship, Quarterly
Journal of Economics, May, pp. 70128.
and Howitt, P. (1992). A model of growth through creative destruction,
Econometrica, 60(2), 323351.
(1998). Endogenous Growth Theory. Cambridge, MA: MIT Press.
and Tirole, J. (1994). The management of innovation, The Quarterly
Journal of Economics, 109(4), 1185209.
Ahrens, N. (2013). Chinas Competitiveness: Myth, Reality, and Lessons for
the United States and Japan (Case study: Huawei), CSIS report,
Washington, DC.
Aitken, B., Hanson, H. G. and Harrison, A. E. (1997). Spillovers, foreign
investment, and export behavior, Journal of International Economics,
43(1), 10332.
and Harrison, A. E. (1999). Do domestic rms benet from direct foreign
investment? Evidence from Venezuela, American Economic Review,
89(3), 60518.
Almeida, P. and Kogut, B. (1997). The exploration of technological diversity
and the geographic localization of innovation, Small Business
Economics, 9(1), 2131.
396
References
397
Amsden, A. (1989). Asias Next Giant. South Korea and Late

Industrialization. Oxford, UK: Oxford University Press.
(1992). Asias Next Giant: South Korea and Late Industrialization. New
York: Oxford University Press.
(2001). The Rise of the Rest. Challenge to the West from Lateindustrializing Economies. Oxford, UK: Oxford University Press,
chapters 1 and 610.
Andersson, T., Formica, P. and Curley, M. (2010). Knowledge-Driven
Entrepreneurship. New York: Springer.
Anselin, L., Varga, A. and Acs, Z. (1997). Local geographic spillovers
between university research and high technology innovations, Journal
of Urban Economics, 42, 42248.
Arcelus, F. J. and Arocena, P. (2000) Convergence and productive efciency
in fourteen OECD countries: a non-parametric frontier approach,
International Journal of Production Economics, 66(2), 10517.
Archibugi, D. and Iammarino, S. (2002). The globalization of technological
innovation: denition and evidence, Review of International Political
Economy, 9(1), 98122.
Arellano, M. and Bond, S. (1991). Some tests of specication for panel data:
Monte Carlo evidence and application to employment equations,
Review of Economic Studies, 58(2), 27797.
Arrow, K. J. (1962). Economic welfare and the allocation of resources for
invention, in Nelson, R. R. (Ed.), The Rate and Direction of Inventive
Activity: Economic and Social Factors. Princeton, NJ: Princeton
University Press for N.B.E.R., 60925.
Athreye, S. and Cantwell, J. (2007). Globalisation and the emergence of new
technology producers, Research Policy, 36(2), 20926.
Atkinson, A. B. and Stiglitz, J. E. (1969). A new view of technological
change, Economic Journal, 79(315), 5738.
Audretsch, D. and Feldman, M. (1996). R&D spillovers and the geography
of innovation and production, American Economics Review, 86(3),
63040.
Bai, J. (2009). Technology acquisition, open independent innovation and overseas investment strategy of Chinese enterprises, REFORMATION &
STRATEGY (Chinese), 25(6), 4045.
Balasubramanayam, V. N., Salisu, M. and Sapsford, D. (1996). Foreign
direct investment and economic growth in EP and IS countries,
Economic Journal, 106(434), 92105.
Baldwin, J. R. (1997). The Importance of Research and Development for
Innovation in Small and Large Canadian Manufacturing Firms,
Research Paper Series No. 107. Ottawa: Statistics Canada.
398
References
Baldwin, J. R. and Lin, Z. 2002. Impediments to advanced technology adoption for Canadian manufacturers. Research Policy, 31(1): 118.
Balzat, M. and Hanusch, H. (2004). Recent trends in the research on national
innovation systems, Journal of Evolutionary Economics, 14(2),
197210.
Banker, R. D., Datar, S. M. and Rajan, M. V. (1987). Measurement of productivity improvements: an empirical analysis, Journal of Accounting,
Auditing & Finance, 2(4), 31947.
Bao, Y., Chen, X. and Zhou, K. Z. (2012). External learning, market dynamics, and radical innovation: evidence from Chinas high-tech rms,
Journal of Business Research, 65(8), 122633.
Barlow, T. (2013) Between the Eagle and the Dragon: Who is Winning the
Innovation Race? California: Hansen House Publishing.
Barney, J. B. and Hansen, M. H. (1994). Trustworthiness as a source of
competitive advantage, Strategic Management Journal, 15, 17590.
Barrios, S., Grg, H. and Strobl, E. (2003). Explaining rms export behaviour: R&D, spillovers and the destination market, Oxford Bulletin of
Economics and Statistics, 65(4), 47596.
Basu, S. and Weil, D. N. (1998). Appropriate technology and growth,
Quarterly Journal of Economics, 113(4), 102554.
Battese, G. E. and Coelli, T. J. (1995). A model for technical inefciency
effects in a stochastic frontier production function for panel data,
Empirical Economics, 20(2), 32532.
Bayona, C., Garca-Marco, T. and Huerta, E. (2001). Firms motivations for
cooperative R&D: an empirical analysis of Spanish rms, Research
Policy, 30(8), 12891307.
van Beers, C., Berghllb, E., and Poot, T. (2008). R&D internationalization,
R&D collaboration and public knowledge institutions in small economies:
evidence from Finland and the Netherlands, Research Policy, 37(2):
294308.
Belderbos, R., Carree, M. and Lokshin, B. (2004). Co-operative R&D and
rm performance, Research Policy, 33(10), 147792.
Bell, M. (1990). Continuing Industrialisation, Climate Change and
International Technology Transfer. Brighton: Science Policy Research
Unit, Sussex University.
and Pavitt, K. (1993). Technological accumulation and industrial growth:
contrasts between developed and developing countries, Industrial and
Corporate Change, 2(2), 157209.
Bernard, A. and Jensen, B. J. (1999). Exceptional exporter performance:
cause, effect, or both?, Journal of International Economics, 47(1),125.
Bernard, A. and Jensen, B. (2004). Why some rms export, Review of
Economics and Statistics, 86(2), 62839.
References
399
Bernard, A., Eaton, J., Jensen, B. and Kortum, S. (2003), Plants and productivity in international trade, American Economic Review, 93(4), 126890.
Bessant, J., Caffyn, S. and Gilbert, J. (1996). Learning to manage innovation,
Technology Analysis & Strategic Management, 8(1), 5970.
Bhagwati, J. (1982). Directly unproductive prot seeking (DUP) activities,
Journal of Political Economy, 90(5).
Birkinshaw, J., Bessant, J. and Delbridge, R. (2007). Finding, forming, and
performing: creating networks for discontinuous innovation. California
Management Review, 49(3), 6785.
BIS. (2013) First ndings from the UK innovation survey 2011. UK
Department of Business, Innovation and Skills. Available at www.gov.
uk/government/uploads/system/uploads/attachment_data/le/200078/12P106A-UKIS_2011First_ndings_Apr13.pdf.
Blomqvis, K., Hurmelinna-Laukkanen, P., Nummela, N. and Saarenketo, S.
et al. (2008). The role of trust and contracts in the internationalisation of
technology-intensive Born Globals, Journal of Engineering and
Technology Management, 25(12), 12335.
Blomstrm, M. and Kokko, A. (1998), Multinational Corporations and
Spillovers, Journal of Economic Surveys, 12, 3, 24777.
Blundell, R., Grifth, R. and Van Reenen, J. (1999). Market Share, Market
Value and Innovation in a Panel of British Manufacturing Firms. The
Review of Economic Studies, 66(3), 529554.
Brdin, V. and Lngnr, M. (2012). A Study on Actual Internal Changes due
to Major External Internationalization in SMEs. Bachelor Thesis,
Gteborgs Universitet, Handelshgskolan.
Bound, K., Saunders, T., Wildson, J. and Adams, J. (2013). Chinas
Absorptive State. London: NESTA.
Boutellier, R., Gassmann, O. and von Zedtwitz, M. (2002). Future
Competitiveness: Research and Analysis of Cases on Global R&D
Management (in Chinese). Guangzhou: Guangdong Economics Publisher.
Braczyk, H. J., Cooke, P. and Heidenreich, M. (1998). Regional Innovation
Systems: The Role of Governance in a Globalized World. London; New
York: Routledge.
Breschi, S. and Lissoni, F. (2001). Knowledge spillovers and local innovation
systems: a critical survey, Industrial and Corporate Change, 10(4),
9751004.
Buck, T., Liu, X., Wei, Y. and Liu, X. (2007). The trade development path
and export spillovers in China: a missing link?, Management
International Review, 47(5), 683706.
Buckley, P. J., Clegg, J. and Wang, C. (2002). The impact of inward FDI on
the performance of Chinese manufacturing rms, Journal of
International Business Studies, 33(4), 63755.
400
References
Bustos, P. (2011). Trade Liberalization, Exports, and Technology Upgrading:

Evidence on the Impact of MERCOSUR on Argentinian Firms, American
Economic Review, 101(1), 30440.
Cantwell, J. (1992). Japans industrial competitiveness and the technological
capabilities of the leading Japanese rms, in Arrison, T. S., Bergsten, C. F.,
Graham, E. M. and Harris, M. C. (Eds.), Japans Growing Technological
Capability: Implications for the US Economy. Washington DC: National
Academy Press, 16588.
(1995). The globalisation of technology: what remains for the product
cycle model, Cambridge Journal of Economics, 19, 15574.
and Piscitello, L. (2002). The location of technological activities of MNCs
in European regions, Journal of International Management, 8(1),
6996.
and Santangelo, G. (1999). The frontier of international technology networks: sourcing abroad the most highly tacit capabilities, Information
Economics and Policy, 11(1), 10123.
and Zhang, Y. (2009). The co-evolution of international business connections and domestic technological capabilities: lessons from the Japanese
catch-up experience, Transnational Corporations, 18(2), 3768.
Cao, C., Li, N., Li, X. and Liu, L. (2013) Reforming Chinas S&T system,
Science, 342(6145), 46062.
Carlsson, B. (2006). Internationalisation of innovation systems: A survey of
the literature, Research Policy, 35(1): 5667.
Cassiman, B. and Veugelers, R. (2006). In search of complementarity in
innovation strategy: internal R&D and external knowledge acquisition,
Management Science, 52(1), 6882.
Caves, R. E. (1974). Multinational rms, competition and productivity in
host-country markets, Economica, 41(162), 17693.
(1996). Multinational Enterprise and Economic Analysis. 2nd ed.
Cambridge: Cambridge University Press.
Chang, H.-J. (2003). Kicking Away the Ladder: Development Strategy in
Historical Perspective. London: Anthem Press.
Chang, J., Fang, X. and Yen, D. C. (2005). Chinas telecommunication
market for international investors: opportunities, challenges, and strategies, Technology in Society 27, 10521.
Chang, Y., Shih, C., Luh, Y. and Wu, S. (2006). MNEs global R&D strategy
in developing countries: a study of foreign-afliated R&D centres in
Taiwan, paper presented at IAMOT 2006, Tsinghua University, Beijing,
2226 May.
Charnes, A., Cooper, W. W. and Rhodes, E. (1978). Measuring the efciency
of decision making units, European Journal of Operational Research,
2(6), 42944.
References
401
Chen, J. (1994), Learning model: from technology introduction to indigenous

innovation, Science Research Management (Chinese), Vol. 2, pp. 1620.
Chen, J. and Qu, W. G. (2003). A new technological learning in China,
Technovation, 23, 861867.
and Liu, X. (eds.) (2008). Indigenous Innovation and Prosperity of
Country, Beijing: China Science Press (zhong guo ke xue chu pan she).
Chen, K. and Kenney, M. (2007). Universities/research institutes and
regional innovation systems: the cases of Beijing and Shenzhen, World
Development, 35(6), 105674.
Chen, Y. F. (2009). Improving the indigenous innovation capabilities in
Chinese enterprises through open innovation. SCIENCE OF SCIENCE
AND MANAGEMENT OF S&T (Chinese), 4, 8186.
and Chen, J. (2008). The inuence of openness to innovation performance,
STUDIES IN SCIENCE OF SCIENCE (Chinese), 26(2), 41926.
Chen, Y. T., He, L. and Si, C. (2007). A study on relationship between open
innovative culture, market driven and innovative performance of hightechnology enterprises: empirical study on Jiang/Zhe/Hu/Min regions,
STUDIES IN SCIENCE OF SCIENCE (Chinese) 25(3): 56772.
Chen, Y. T., Ning, Z. and Si, C. L. (2006). Indigenous innovation modes of
integrating external innovation sources: an empirical study on 241 rms
in China. Economic Management (Chinese), 17, 1115.
Cheng, V. M. Y. (2004). Progress from traditional to creativity education in
Chinese societies, in Lau, S., Anna, N., Hui, N. and Ng, G. Y. C. (Eds.),
Creativity: When East Meets West. Singapore: World Scientic
Publishing Company, 13768.
Chesbrough, H. (2003). Open Innovation: The New Imperative for Creating
and Proting from Technology. Cambridge, MA: Harvard Business
School Press.
(2006). Open Business Models: How to Thrive in the New Innovation
Landscape. Cambridge, MA: Harvard Business School Press.
and Crowther, A. K. (2006). Beyond high tech: early adopters of open
innovation in other industries, R&D Management, 36(3), 22936.
Vanhaverbeke, W. and West, J. (2006). Open Innovation: Researching a
New Paradigm. London: Oxford University Press.
Cheung, K. and Lin, P. (2004). Spillover effects of FDI on innovation in China:
evidence from the provincial data, China Economic Review, 15, 2544.
Child, J. and Rodrigues, S. B. (2005). The internationalisation of Chinese
rms: A case for theoretical extension. Management and Organization
Review, 1(3): 381410.
Christensen, J. F., Olesen, M. H. and Kjar, S. (2005). The industrial dynamics
of open innovation: evidence from the transformation of consumer
electronics. Research Policy, 34(10), 153349.
402
References
Climate Group. (2009). Chinas clean revolution II: opportunities for a low
carbon future. The Climate Group. Available at http://www.theclimate
group.org/_assets/les/Chinas-Clean-Revolution-II.pdf.
Coe, D. and Helpman, E. (1995). International R&D spillovers, European
Coelli, T. (1996). A Guide to DEAP Version 2.1, CEPA working paper,
96/08.
Cohen, W. and Levinthal, D. (1989). Innovation and learning: two faces of
R&D, Economic Journal, 99, 56996.
(1990). Absorptive capacity: A new perspective on learning and innovation, Administrative Science Quarterly, 35, 12852.
Nelson, R. R. and Walsh, J. P. (2002). Links and impacts: The inuence of public research on industrial R&D. Management Science,
48, 123.
Conti, J. P. (2007). Prole Huawei: from China with love. Communications
Engineer, (8), 2631.
Cookson, C. (2005). R&D spending falls further behind target, Financial
Times, 24 May 2005.
Cooper, W. W., Seiford, L. M. and Tone, K. (2000). Data Envelopment
Analysis: A Comprehensive Text with Models, Applications, References
and DEA-Solver Software. Boston/Dordrecht/ London: Kluwer Academic
Publishers.
Correa, C. M. (2005). Can the TRIPS agreement foster technology transfer
to developing countries?, in Maskus, K. E. and Reichman, J. H. (Eds.),
International Public Goods and Transfer of Technology under a
Globalised Intellectual Property Regime. Cambridge: Cambridge
University Press, 22756.
Cosh, A., Fu, X., and Hughes, A. (2012) Organisation, Structure and
Innovation Performance in Different Environments, Small Business
Economics, 39(2), 301317.
Cosh, A., Fu, X., and Hughes, A. (2004). How much does informality in
management matter for SME innovation, 35th Entrepreneurship,
Innovation and Small Business (EISB) Conference, Barcelona, Spain,
September 2005, European Best Paper Award.
and Hughes, A. (2001). Innovation. The Contribution of European SMEs,
in Enterprises in Europe: Sixth Report. Luxembourg: Eurostat.
and Lester, R. K. (2006). UK PLC: Just How Innovative Are We?
Cambridge MA; Cambridge UK: MIT Institute, Centre for Business
Research and Industrial Performance Centre.
Cragg, John G. (1971). Some Statistical Models for Limited Dependent
Variables with Application to the Demand for Durable Goods.
Econometrica 39(5): 829844.
References
403
Criscuolo, C., Haskel, J. and Slaughter, M. (2005). Why Are Some Firms
More Innovative? Knowledge Inputs, Knowledge Stocks and the Role of
Global Engagement, NBER Working Paper No. 11479 (June).
Dahlman, C. J. (1994). Technology strategy in East Asian Developing
Economies, Journal of Asian Economics, 5(Winter), 54172.
Darby, M. R., Zucker, L. G. and Wang, A. (2003). Universities, Joint
Ventures, and Success in the Advanced Technology Program. NBER
working paper no. 9463.
Darcy, J., Krmer-Eis, H., Guellec, D. and Debande, O. (2009). Financing
technology transfer, EIB Papers, 142: 5473.
Dasgupta, P. and David, P. A. (1994). Toward a new economics of science,
Research Policy, 23, 487521.
De Bondt, R. (1997). Spillovers and innovative activities, International
Journal of Industrial Organisation, 15(1): 128.
Deng, P. (2007). Investing for strategic resources and its rationale: The case of
outward FDI from Chinese companies, Business Horizons, 50(1): 7181.
Department of Trade and Industry (DTI). (2002). The 2002 R&D
Scoreboard: Commentary and Analysis. London: Department of Trade
and Industry.
Department of Trade and Industry (DTI). (2003a). Competing in the Global
Economy The Innovation Challenge, DTI Economics Paper No. 7.
Department of Trade and Industry (DTI). (2003b). Competing in the Global
Economy The Innovation Challenge, DTI Innovation Report-Overview.
Department of Trade and Industry (DTI). (2003c). The 2003 R&D
Scoreboard: Analysis.
Department of Trade and Industry (DTI). (2005). The 2005 R&D
Scoreboard: Analysis.
Department of Trade and Industry (DTI). (2008). Persistence and Change in
UK Innovation, 20022006. London, UK: Department for Business,
Innovation & Skills.
Drifeld, N. and Love, L. H. (2007). Linking FDI motivation and host
economy productivity effects: conceptual and empirical analysis.
Journal of International Business Studies, 38(3), 46073.
and Love, J. (2003). Foreign direct investment, technology sourcing and
reverse spillovers, The Manchester School, 71(6), 65972.
Drucker, P. F. (1985). Innovation and Entrepreneurship. New York:
Harper & Row.
Du, P. and Cao, Y. (2010). Ethical reection on new energy: the case of PV
industry, in Chinese Academy of Science (Ed.), High Technology
Developement Report of China. Beijing: Science Publisher.
Dunning, J. (1993). Multinational Enterprises and the Global Economy.
Reading, MA: Addison Wesley.
404
References
(1994). Multinational enterprises and the globalization of innovatory

capacity, Research Policy, 23, 6788.
Kim, C. and Park, D. (2007). Old Wine in New Bottles: A Comparison of
Emerging Market TNCs Today and Developed Country TNCs Thirty
Years Ago. Oxford University TMCD (Technology and Management
Centre for Development) Working Paper, No 011.
and Lundan, S. (2009). The internationalisation of corporate R&D:
a review of the evidence and some policy implications for home countries. Review of Policy Research, 26(12): 1333.
and Narula, R. (1995). The R&D Activities of Foreign Firms in the United
States. International Studies of Management & Organization,
25: 3973.
and Narula, R. (1996), The Investment Development Path Revisited: Some
emerging issues, in Dunning, J. H. and Narula, R., eds., Foreign Direct
Investment and Governments: Catalysts for Economic Restructuring.
London and New York: Routledge.
Lim, C. and Kim, J. (2001). Incorporating trade into the investment development path: a case study of Korea and Taiwan, Oxford Development
Studies, Vol. 29, pp. 145154.
Eaton, J. and Kortum, S. (1995). Engines of Growth: Domestic and Foreign
Sources of Innovation, NBER Working Papers 5207, National Bureau of
Economic Research, Inc.
(1996). Trade in ideas: patenting and productivity in the OECD, Journal
of International Economics, 40(34), 25178.
(1999). International technology diffusion: theory and measurement,
International Economic Review, 40(3), 53770.
Economist Intelligence Unit (EIU) (2007). Innovation: Transforming the way
Business Creates. Available at http://graphics.eiu.com/upload/portal/Cisco
InnoSmallFile.pdf.
EC (European Commission). (1995) Green Paper on Innovation, December,
EC, DG XIII.
(2002). European Innovation Scoreboard. Available at http://trendchart.
cordis.lu/Reports/index.cfm?fuseaction=ReportInnovationHome.
(2005). European Innovation Scoreboard. Available at http://trendchart.
cordis.lu/Reports/index.cfm?fuseaction=ReportInnovationHome.
(2008). European Innovation Scoreboard 2007. Luxembourg: Ofce for
Ofcial Publications of the European Communities, 2008.
(2013). Innovation Scoreboard, 2013. Brussels: European Commission.
Elango, B. and Pattnaik, C. (2007). Building capabilities for international
operations through networks: a study of India, Journal of International
Business Studies, 38(4), 54155.
References
405
Enkel, E., Gassmann, O. and Chesbrough, H. (2009). Open R&D and open
innovation: exploring the phenomenon, R&D Management, 39(4),
31116.
Eom, B.-Y. and Lee, K. (2010). Determinants of industry-academy linkages
and their impact on rm performance: the case of Korea as a latecomer in
knowledge industrialization, Research Policy, 39, 62539.
Erdilek, A. (1984). International Technology Transfer in the Middle East and
North Africa, Management Decision, 22(2), 4549.
Ernst, D. (2006). Innovation off-shoring: Asias emerging role in global
innovation networks. East-West Center Special Reports 10, 150.
et al. (2004). Searching for a New Role in East Asian Regionalization:
Japanese Production Networks in the Electronics Industry, Economics
Study Area Working Papers 68. Geneva: East-West Center.
Ettlie, J. E. (1983). Organizational policy and innovation among suppliers to
the food processing sector. Academy of Management Journal, 26, 2744.
Etzkowitz, H. and Leydesdorff, L. (1997). Universities and the Global
Knowledge Economy: A Triple Helix of University-Industry-Government
Relations. London: Continuum.
2000, The dynamics of innovation: from national systems and Mode 2
to a triple helix of universityindustrygovernment relations, Research
Policy, 29(2), 10923.
Eun, J.-H., Lee, K. and Wu, G. S. (2006). Explaining the University-run
enterprises in China: a theoretical framework for universityindustry
relationship in developing countries and its application to China,
Eurostat. (2004). Available at www.europa.eu.int/comm/eurostat.
Evangelista, F. (2005). Qualitative insights into the international new venture
creation process, Journal of International Entrepreneurship, 3(3), 17998.
Faber, A., Kemp, R. and Van der Veen, G. (2008). Innovation policy for the
environment in the Netherlands and the EU, in Nauwelaers, C. and
Wintjes, R. (Eds.), Innovation Policy in Europe: Measurement and
Strategy. Northampton, MA: Edward Elger, chapter 6.
Faber, J. and Hesen, A. B. (2004). Innovation capabilities of European
nations: cross-national analyses of patents and sales of product innovations, Research Policy, 33(2), 193207.
Fagerberg, J. (1994). Technology and international differences in growth
rate, Journal of Economic Literature, 32(3), 114775.
Mowery, D. C. and Nelson, R. R. (2005). The Oxford Handbook of
Innovation. New York: Oxford University Press.
Fan, P. L. (2006). Catching up through developing innovation capability:
evidence from Chinas telecom-equipment industry. Technovation, 26,
35968.
406
References
Fang, X. (2007). Formation, evolution and reform of the Chinese S&T

system, in Fang, X. (Ed.), Chinese Technological Innovation and
Sustainable Development (in Chinese). Beijing: Science Press.
Fre, R., Grosskopf, S., Norris, M. and Zhang, Z. (1994). Productivity
growth, technical progress, and efciency change in industrialized countries, The American Economic Review, 84(1), 6683.
Farhoomand, A. F. and Ho, P. (2006). Huawei: Ciscos Chinese Challenger.
University of Hong Kong Case HKU599 (Available through Harvard
Business School, pp. 910).
Feldman, M. P. (1999). The new economics of innovation, spillovers and
agglomeration: a review of empirical studies, Economics of Innovation
and New Technology, 8, 525.
Feng, M. T. (2009) Technology Transfer from University to Industry: Insight
Into University Technology Transfer in the Chinese National Innovation
System. London: Adonis & Abbey Publishers Ltd.
Figueiredo, P. N.. (2003). Learning processes features: how do they inuence
inter-rm differences in technological capability-accumulation paths and
operational performance improvement? Industrial and Corporate
Change, 12(3), 60743.
Filatotchev, I. and Piesse, J. (2009). R&D, internationalisation and growth of
newly listed rms: European evidence, Journal of International Business
Studies, 40(8), 126076.
Findlay, R. (1978). Relative backwardness, direct foreign investment and the
transfer of technology: a simple dynamic model, Quarterly of Journal of
Economics, 92(1), 116.
Fosfuri, A. (2006). The licensing dilemma: understanding the determinants of
the rate of technology licensing, Strategic Management Journal, 27(12),
114158.
and Motta, M. (1999). Multinationals without advantages, The
Scandinavian Journal of Economics, 101(4), 61730.
and Ronde, T. (2001). Foreign direct investment and spillovers through
workers mobility, Journal of International Economics, 53(1),
20522.
Franco, E., Ray, S. and Ray, P. K. (2011). Patterns of innovation practices of
multinational-afliates in emerging economies: Evidences from Brazil
and India, World Development, 39(7), 124960.
Freeman, C. (1987). Technology and Policy and Economic Performance:
Lessons from Japan. London: Pinter.
(1994). The economics of technical change, Cambridge Journal of
Economics, 18(5), 463514.
(1995). The national system of innovation in historical perspective,
Cambridge Journal of Economics, 19, 524.
References
407
and Soete, L. (1997). The Economics of Industrial Innovation. London:

Pinter.
Fritsch, M. and Slavtchev, V. (2007). What Determines the Efciency of
Regional Innovation Systems? Jena Economic Research Paper No.
2007006. Available at SSRN: heep://ssrn.com/abstract=1018593.
Fu, X. (2004a). Limited linkages from growth engines and regional disparities in China, Journal of Comparative Economics, 32(1), 14864.
(2004b). Exports, Foreign Direct Investment and Economic Development
in China. London and New York: Palgrave McMillan.
(2005). Exports, technical progress and productivity growth in Chinese
manufacturing industries, Applied Economics, 37(7), 72539.
(2007). Trade-cum-FDI, human capital inequality and the dual economy
in China: the Signer perspective, Journal of Economic Change and
Restructuring, 40(1), 13755.
(2008). Foreign direct investment, absorptive capacity and regional innovation capabilities: Evidence from China, Oxford Development Studies,
36(1), 89110.
(2012a). How does openness affect the importance of incentives for innovation? Research Policy, 41(3), 51223.
(2012b). Managerial knowledge spillovers from FDI through the diffusion
of management practices, Journal of Management Studies, 49(5), 97099.
(2014). What potential does open innovation hold for Asia, The
Economist, Oct 10, 1216.
and Balasubramanayam, V. N. (2003). Township and village enterprises in
China, Journal of Development Studies, 39(4), 2746.
Cosh, A., Yang, Q. and Hughes, A. (2006). World innovation frontier and
the EU-US innovation gap, paper presented at the 9th North America
Workshop of Efciency and Productivity Analysis, Stern Business
School, New York University, New York.
Cosh, A., Hughes, A., De Hoyos, R. and Eisingerich, A. (2006). The
experiences of UK mid-corporate companies in emerging Asian economies, UK Trade & Investment, London. 2006 (URN 06/1137).
and Gong, Y. (2008). Indigenous and Foreign Innovations Efforts and
Drivers of Technological Upgrading: Evidence from China, SLPTMD
Working Paper No 018, University of Oxford.
and Gong, Y. (2010). Absorptive Capacity and the Benets from Global
Reservoirs or Knowledge: Evidence from a Linked China-OECD
Dataset, SLPTMD Working Paper 31, University of Oxford.
(2011). Indigenous and foreign innovation efforts and technological
upgrading in emerging economies: rm-level evidence from China,
World Development, 39(7), 121325.
Liu, S. and Li, T. (2013). Determinants and Impact of Outward Direct
Investment from China: Evidence from a Firm-level Survey in
408
References
Guangdong Province. TMD Working Paper No: TMD-WP-49,

University of Oxford.
Pietrobelli, C. and Soete, L. (2011). The role of foreign technology and
indigenous innovation in the emerging economies: technological change
and catching-up. World Development 39(7), 120312.
and Soete, L. (2010). The Rise of Technological Power in the South.
London and New York: Palgrave Macmillan.
and Xiong, H. (2011). Open innovation in China: policies and practices,
Journal of Science & Technology Policy in China, 2(3), 196218.
and Yang, Q. (2009). World innovation frontier: exploring the innovation
gap between the EU and the US, Research Policy, 38(7), 120313.
and Zhang, J. (2011). Technology transfer, indigenous innovation and
leapfrogging in green technology: the solar-PV industry in China and
India, Journal of Chinese Economic and Business Studies, 9(4), 32947.
Funke, M. and Niebuhr, A. (2000). Spatial R&D Spillovers and Economic
Growth Evidence from West Germany, HWWA Discussion Paper No 98.
Furman, J. and Hayes, R. (2004). Catching up or standing still? National
innovative productivity among follower countries, 19781999,
Porter, M. and Stern, S. (2002). The determinants of national innovative
capacity, Research Policy, 31, 899933.
Galia, F., and Legros, D. (2004). Complementarities between obstacles to
innovation: evidence from France, Research Policy, 33(8), 118599.
Gans, J. S., Stern, S., (2003). Assessing Australias Innovative Capacity in the
21st Century. Intellectual Property Research Institute of Australia
Working Paper.
Gans, J. and Hayes, R. (2008). Measuring innovative performance.
Melbourne Review, Vol 2, No 1. 2006. 7077.
Gao, X., Liu, J., Chai, K. H. and Li, J. (2007). Overcoming latecomer
disadvantages in small and medium-sized rms: evidence from
China, International Journal of Technology and Globalization, 3(4),
36483.
Gassmann, O. (2006). Editorial: opening up the innovation process towards
an agenda, R&D Management, 36(3), 22328.
Gassmann, O., Enkel, E. and Chesbrough, H. 2010. Editorial: The future of
open innovation, R&D Management, 40(3), 213221.
and von Zedtwitz, M. (1998). Organization of industrial R&D on a global
scale, R&D Management, 28(3), 14761.
Geroski, P. A. (1990). Innovation, technological opportunity, and market
structure, Oxford Economic Papers, 42(3), 586602.
(1994). Market Structure, Corporate Performance, and Innovative
Activity. Oxford: Oxford University Press.
References
409
and Mazzucato, M. (2002). Learning and the sources of corporate

growth, Industrial and Corporate Change, 11(4), 62344.
Gilman, D. (2010). The New Geography of Global Innovation, Global
Market Institute, Goldman Sachs, report.
Girma, S. (2005). Absorptive capacity and productivity spillovers from FDI:
a threshold regression analysis, Oxford Bulletin of Economics and
Statistics, 67, 281306.
Gong, Y. and Grg, H. (2009). What determines innovation activity in
Chinese state-owned enterprises? The role of foreign direct investment,
Greenaway, D. and Wakelin, K. (2001). Who benets from foreign direct
investment in the UK? Scottish Journal of Political Economy, 48, 1933.
Goes, J. B. and Park, S. H. (1997). Interorganizational links and innovation:
the case of hospital services, Academy of Management Journal, 40(3),
67396.
Gompers, P. (2006). Venture capital, in Eckbo, B. E. (Ed.), Handbook of
Corporate Finance: Empirical Corporate Finance. Amsterdam; Boston:
Elsevier/North-Holland, chapter 9.
Grg, H. and Greenaway, D. (2004). Much ado about nothing? Do domestic
rms really benet from foreign direct investment? World Bank
Research Observer, 19, 17197.
and Strobl, E. (2001). Multinational companies and productivity spillovers: a meta-analysis, Economic Journal, 111(475), F72339.
Granstrand, O. (1999). Internationalisation of corporate R&D: A study of
Japanese and Swedish corporations. Research Policy, 28(Special issue),
275302.
Bohlin, E., Oskarsson, C. and Sjoberg, N. (1992). External technology
acquisition in large multi-technology corporations, R&D Management
22, 11133.
Hkanson, L. and Sjlander, S. (1993). Internationalisation of R&D a
survey of some recent research, Research Policy, 22(56), 41330.
Greenaway, D. and Kneller, R. (2004). Exporting and productivity in the
UK, Oxford Review of Economic Policy, 20(3), 35871.
Sousa, N. and Wakelin, K. (2004). Do domestic rms learn to export from
multinationals? European Journal of Political Economy, 20(4), 102743.
Upward, R. and Wright, P. (2002). Sectoral and Geographic Mobility of
Labour Markets and Structural Adjustment, mimeo, University of
Nottingham.
Greene, W. (1997). Frontier production functions, in Pesaran, M. H. and
Schimidt, P. (Eds.), Handbook of Applied Econometrics, vol. II:
Microeconomics. Oxford: Blackwell Publishers.
Greene, W. (2003). Econometric analysis, 5th Edition. London: Prentice-Hall.
410
References
Grifth, R., Harrison, R. and Reenen, J. V. (2006). How special is the special
relationship? Using the impact of U.S. R&D spillovers on U.K. rms as a
test of technology sourcing, The American Economic Review, 96(5),
185975.
Redding, S. and Van Reenen, J. (2003). R&D and absorptive capacity:
theory and empirical evidence, Scandinavian Journal of Economics,
105(1), 99118.
Griliches, Z. (1979). Issues in assessing the contribution of R&D to productivity growth, Bell Journal of Economics, 10(1), 92116.
(1990). Patent statistics as economic indicators. Journal of Economic
Literature. 28 (4), 16611707.
Grossman, G, M. and Helpman, E. (1991). Innovation and Growth in the
Global Economy. Cambridge, MA: MIT Press.
(1994). Technology and trade. NBER Working Papers 4926, National
Bureau of Economic Research, Inc.
Guan, J. and Ma, N. (2003). Innovative capability and export performance
of Chinese rms, Technovation, 23, 73747.
(2007). Chinas emerging presence in nanoscience and nanotechnology: A
comparative bibliometric study of several nanoscience giants,
Research Policy, 36(6), 88086.
Gulati, R. (1999). Network location and learning: the inuence of network
resources and rm capabilities on alliance formation, Strategic
Management Journal, 20, 397420.
Hagedoorn, J. (2002). Inter-rm R&D partnerships: an overview of major
trends and patterns since 1960, Research Policy, 31(4), 47792.
Hall, B. (2004). Exploring the Patent Explosion, Centre for Business
Research, University of Cambridge, Working Paper no WP291.
Hansen, L. (1982). Large sample properties of generalized method of
moments estimators, Econometrica, 50(4), 102954.
Harzing, A. and Noorderhaven, N. (2006). Knowledge ows in MNCs:
an empirical test and extension of Gupta and Govindarajans typology of subsidiary roles, International Business Review, 15(3),
195214.
Hayhoe, R. (1996). Chinas Universities, 18951995: A Century of Cultural
Conict. New York: Garland Publishing.
He, X. and Mu, Q. (2012). How Chinese rms learn technology from transnational corporations: A comparison of the telecommunication and
automobile industries, Journal of Asian Economics, 23, 27087.
He, Y. (2008). Shenyang shukong: open innovation. Peoples Daily (Chinese),
4 May.
Hershberg, E., Nabeshima, K. and Yusuf, S. (2007). Opening the ivory tower
to business: university-industry linkages and the development of
References
411
knowledge-intensive clusters in Asian cities, World Development, 35(6),

93140.
Hiebert, D., 2002. The determinants of the cost efciency of electric generating plants: a stochastic frontier approach. Southern Economic Journal
68 (4), 935946.
Hippel, E. von and Krogh, G. von. (2003). Open source software and the
private-collective model: issues for organization science, Organization
Science, 14(2), 20923.
HM Treasury. (2003). Lambert review of business-university collaboration,
available at www.lambertreview.org.uk.
Hobday, M. G. (1995). Innovation in East Asia: The Challenge to Japan.
Cheltenham, UK: Edward Elgar.
Rush, H. and Bessant, J. (2004). Approaching the innovation frontier in
Korea: the transition phase to leadership, Research Policy, 33, 143357.
Hoekman, B. M., Maskus, K. E. and Saggi, K. (2005). Transfer of technology
to developing countries: unilateral and multinational policy options,
Hoffman, K., Parejo, M., Bessant, J. and Perren, L. (1998). Small rms,
R&D, technology and innovation in the UK: a literature review,
Technovation, 18(1), 3973.
Hoffmann, W. and Schlosser, R. (2001). Success factors of strategic alliances
in small and medium-sized enterprises: an empirical survey, Long Range
Planning, 34(3), 35781.
Hoer, R. and Payne, J. E. (1993). Efciency in social versus private agricultural production: The case of Yugoslavia, The Review of Economics and
Statistics, 75(1), 15357.
Hollanders, H. and Esser, F. C. (2007). Measuring Innovation Efciency,
INNO-Metrics Thematic paper, available at http://www.pedz.uni-man
nheim.de/daten/edz-h/gdb/07/eis_2007_Innovation_efciency.pdf
Hong, W. (2006). Technology transfer in Chinese universities: is mode 2
sufcient for a developing country? in Law, P., Fortunati, L. and
Yang, S. (Eds.), New Technologies in Global Societies. Singapore:
World Scientic Publishers, 2150.
(2008). Decline of the center: The decentralizing process of knowledge
transfer of Chinese universities from 1985 to 2004, Research Policy,
37(4), 58095.
Hoskisson, R. E., Hitt, M. A., Johnson, R. A. and Grossman, W. (2002).
Conicting voices: the effects of institutional ownership heterogeneity
and internal governance on corporate strategies. Academy Management
Journal, 45, 607716.
Hou, C. (2009). Weichai: open innovation, CHINESE MANUFACTURING
INFORMATION (Chinese), 8, 5253.
412
References
Hsiao, C. (2003). Analysis of Panel Data, Cambridge University Press,

Cambridge.
Hu, A. and Jefferson, G. (2002). FDI impact and spillover: evidence from
Chinas electronic and textile industries, World Economy, 38(4), 106376.
and Qian, J. (2005). R&D and technology transfer: rm-level evidence
from Chinese industry, The Review of Economics and Statistics, 87(4),
78086.
Hu, A. (2011). China in 2020: A New Type of Superpower. Washington, DC:
Brookings Institution Press.
Huang, Y. (2003). Selling China. New York: Cambridge University Press.
Huawei Press. (2011). Huawei is the leading contributor to LTE standards, Huawei press release, 29 November. Available at www.huawei.
com/en/about-huawei/newsroom/press-release/hw-104732-huaweilte
standards.htm.
Hughes, A. (2003). Knowledge Transfer, Entrepreneurship and Economic
Growth: Some Reections and Implications for Policy in the
Netherlands. University of Cambridge Centre for Business Research
Working Paper no 273.
(2008). Innovation policy as cargo cult: myth and reality in knowledge-led
productivity growth, in Bessant, J. and Venables, T. (Eds.), Creating
Wealth from Knowledge. Meeting the Innovation Challenge. Cheltenham,
UK: Edward Elgar.
(2010). The multifaceted role of universities, ESRC Society Now IN
FOCUS, no. 8.
and Scott Morton, M. S. (2006). The transforming power of complementary assets, MIT Sloan Management Review, Summer, 47, 5058.
Hwang, Y., Kim, S., Byun, B., Lee, G. and Lee, H. (2003). Strategies of
Promoting Industry-Academia-Research Institute R&D Partnerships to
Cooperation with New Technologies: Focusing on Industry-Research
Institute Inter-rm R&D Partnerships. Science & Technology Policy
Institute.
Immelt, J. R., Govindarajan, V. and Trimble, C. (2009). How GE is disrupting itself, Harvard Business Review, October.
India Semiconductor Association. (2010). Solar PV industry 2010: contemporary scenario and emerging trends. India Semiconductor Association.
Available at http://www.isaonline.org/documents /ISA_SolarPV Report_
May2010.pdf.
INSEAD. (2007). The World Business/INSEAD Global Innovation Index
2007.
International Labour Organization. (2003). Report of the Committee on
Employment and Social Policy, GB.286/15, ILO, Geneva.
References
413
Iwasaki, I., Csizmadia, P., Illessy, M., Mako, C. and Szany, M. (2010),
Foreign Direct Investment, Information Spillover, and Export
Decision: The Concentric-circle Model with Application to Hungarian
Firm-level Data, Institute for Economic Research, Hitotsubashi
University, working paper, no 527.
Iyer, K. N. S. (2002). Learning in strategic alliances: an evolutionary perspective, Academy of Marketing Science Review, 10, 116.
Jaffe, A. (1989). Real effects of academic research, American Economic
Review, 79, 95770.
Trajtenberg, M. and Henderson, R. (1993). Geographic localization of
knowledge spillovers as evidenced by patent citations, The Quarterly
Journal of Economics, 108, 57798.
Jakobson, L., (Ed), 2007. Innovation with Chinese Characteristics, HighTech Research in China, Palgrave Macmillan/Finish Institute of
International Affairs, Hampshire UK/New York, USA.
Jaumotte, F. and Pain, N. (2005). From Ideas to Development: The
Determinants of R&D and Patenting, OECD Working Paper, ECO/
WKP, 44.
Javorcik, B. S. (2004). Does foreign direct investment increase the productivity of domestic rms? In search of spillovers through backward linkages,
American Economic Review, 94(3), 60527.
(2008). Can survey evidence shed light on spillovers from foreign direct
investment? World Bank Research Observer, 23(2), 13959.
Jensen, M. C. (1993). The Modern Industrial Revolution, Exit, and the
Failure of Internal Control Systems. Journal of Finance. 48: 83180.
Jiang, J. H. (2009). Research on the cost determinants of knowledge transfer
and inuencing mechanisms for the latecomer rms, Science Research
Management, 306, 18.
Jiang, X. J. (2004). Comprehension of technical globalization,
MANAGEMENT WORLD (Chinese), 6, 413.
Jin, J. and von Zedtwitz, M. (2008). Technological capability development in
Chinas mobile phone industry, Technovation, 28(6), 32734.
Johanson, J. and Mattsson, L. (1988). Internationalization in industry system: a network approach, in Hood, N. and Vahlne, J.-E. (Eds.),
Strategies in Global Competition. London, UK: Routledge, 287314.
and Vahlne, J.-E. (1977). The internationalization process of the rm a
model of knowledge development and increasing foreign market commitments. Journal of International Business Studies, 8(1), 2332.
Johnson, J. E. (2004). Factors inuencing the early internationalisation of
high technology start-ups: US and UK evidence, Journal of International
Entrepreneurship, 2(12), 13954.
414
References
Jones, C. I. (1995). R&D-based models of economic growth, Journal of

Political Economy, 103(4), 75981.
Kafouros, M., Buckley, P., Sharp, J. and Wang, C. (2008). The role of internationalisation in explaining innovation performance, Technovation,
28(12), 6374.
Kamiyama, S., Sheehan, J. and Martinez, C. (2006). Valuation and
Exploitation of Intellectual Property, OECS Science, Technology and
Industry Working Paper No 2006/5.
Katila, R. and GautAm, A. (2002). Something old, something new: a longitudinal study of search behavior and new product introduction,
Academy of Management Journal, 45(6), 118394.
Kaufmann, A. and Todtling, F. 2001. Science-industry interaction in the
process of innovation: the importance of boundary-crossing between
systems, Research Policy, 30, 791804.
Keesing, D. B. and Lall, S. (1992). Marketing, manufactured exports from
developing countries: learning sequences and public support, in
Helleiner, G. K. (Ed.), Trade Policy, Industrialization, and Development:
New Perspectives. Oxford: Clarendon Press.
Kessler, E. H., Bierly, P. E. and Gopalakrishnan, S. (2000). Internal vs. external learning in new product development: effects on speed, costs and
competitive advantage, R&D Management, 30(3), 21323
Keupp, M. M. and Gassmann, O. (2009). Determinants and archetype users
of open innovation, R&D Management, 39(4), 33141.
Kim, L. (1997a). Imitation to Innovation: The Dynamics of Koreas
Technological Learning. Boston: Harvard Business School Press.
(1997b). The dynamics of Samsungs technological learning in semiconductors, California Management Review, 39(3), 86100.
and Nelson, R. R. (Eds.). (2000). Technology, Learning and Innovation:
Experiences of Newly Industrializing Economies. Cambridge: Cambridge
University Press.
Kim, Y. and Lee, K. (2003). Technological collaboration in the Korean
electronic parts industry: patterns and key success factors, R&D
Management, 33(1), 5977.
King, D. A. (2004). The scientic impact of nations, Nature, 430, 31116.
Kiriyama, N. (2012). Trade and Innovation: Synthesis Report, OECD Trade
Policy Papers 135, OECD Publishing.
Kitson, M., Howells, J., Braham, R. and Westlake, S. (2009). The Connected
University: Driving Recovery and Growth in the UK Economy. London:
NESTA.
Kleinknecht, A. (1996). New indicators and determinants of innovation: an
introduction, in Kleinknecht, A. (Ed.), Determinants of Innovation. The
References
415
Message from New Indicators. Hampshire and London: Macmillan,

112.
Kneller, R. and Pisu, M. (2007). Industrial linkages and export spillovers
from FDI, World Economy, 30, 10534.
and Stevens, P. A. (2006). Frontier technology and absorptive capacity:
evidence from OECD manufacturing industries, Oxford Bulletin of
Economics and Statistics, 68, 121.
Koellinge, P. (2007). Why are some entrepreneurs more innovative than
others, Small Business, Economics, 31, 2137.
Kogut, B. and Zander, U. (1993). Knowledge of the rm and the evolutionary
theory of the multinational corporation, Journal of International
Business Studies, 24(4), 62546.
Kokko, A., Tansini, R. and Zejan, M. C. (1996). Local technological capability and productivity spillovers from FDI in the Uruguayan manufacturing sector, Journal of Development Studies, 32(4), 60211.
(2001). Trade regimes and spillover effects of FDI: evidence from
Uruguay, Weltwirtschaftliches Archiv, 37(1), 12449.
Koopman, R., Wang, Z. and Wei, S. J. (2008). How Much of Chinese Exports
Is Really Made in China? Assessing Domestic Value-Added When
Processing Trade Is Pervasive, NBER Working Paper 14109.
Koruna, S. (2004). External technology commercialization policy guidelines, International Journal of Technology Management, 27(2/3),
24154.
Koschatzky, K. (2001). Networks in innovation research and innovation
policy an introduction, in Koschatzky, K., Kulicke, M. and
Zenker, A. (Eds.), Innovation Networks: Concepts and Challenges in
the European Perspective. Heidelberg: Physica Verlag, 323.
Krause, L. (1988). Hong Kong and Singapore: twins or kissing cousins?
Economic Development and Cultural Change, 36(3), 4566.
Kroll, H. and Liefner, I. (2008). Spin-off enterprises as a means of technology
commercialization in a transforming economy: Evidence from three universities in China, Technovation, 28, 298313.
Krueger, A. (1974). The political economy of rent-seeking society, American
Krugman, P. (1991). Increasing returns and economic geography, Journal of
Political Economy, 99(3), 48399.
Kuemmerle, W. (1997) Building effective R&D capabilities abroad,
Harvard Business Review, 3/4, 6170.
Kuo, K. (2006). China pursues Africa deals, red herring, 11 December.
Available at www.siemens.be/cmc/newsletters/index.aspx?id=1362117892.
416
References
Lall, S. (1982). Developing Countries as Exporters of Technology. London:

Macmillan Press.
(1987). Learning to Industrialise: The Acquisition of Technological
Capability by India. London: Macmillan.
(1992). Technological capabilities and industrialization, World
(1996). Learning from the Asian Tigers Studies in Technology and
Industrial Policy. London: Macmillan.
(2001). Competitiveness, Technology and Skills. Cheltenham, UK: Edward
Elgar.
(2003). Foreign direct investment, technology development and competitiveness: issues and evidence, in Lall, S. and Urata, S. (Eds.),
Competitiveness, FDI and Technological Activity in East Asia, published
in association with the World Bank, Cheltenham, UK: Edward Elgar.
Lall, S., and Urata, S. (Eds.) (2003). Competitiveness, FDI and technological
activity in East Asia. Cheltenham, UK: Edward Elgar.
Laursen, K. and Salter, A. (2006). Open for innovation: the role of openness
in explaining innovation performance among UK manufacturing rms,
Strategic Management Journal, 27(2), 13150.
Lcuyer, C. (2006). Making Silicon Valley: Innovation and the Growth of
High Tech, 193070, MIT Press, Cambridge, MA.
Lee, J., Bae, Z. and Choi, D. (1988). Technology development processes: a
model for a developing country with a global perspective, R&D
Jee, M. and Eun, J. H. (2011). Assessing Chinas economic catch up at the
rm level and beyond: Washington consensus, East Asian consensus and
the Beijing model, Industry and Innovation, 18(5), 487507.
and Lim, C. (2001). Technological regimes, catching-up and leapfrogging: ndings from the Korean industries, Research Policy, 30(1),
45983.
Lee, S., Park, G., Yoon, B. and Park, J. (2010). Open innovation in SMEs an
intermediated network model, Research Policy, 39(2), 290300.
Lema, R. and Lema, A. (2012). Whither technology transfer? The rise of
China and India in green techonology sectors, Innovation and
Leonard, D. and Sensiper, S. (1998). The role of tacit knowledge in group
innovation, California Management Review Berkeley, Spring (electronic). Available at http://connection.ebscohost.com/c/articles/738860/
role-tacit-knowledge-group-innovation.
Leornard-Barton, D. (1995). Wellsprings of Knowledge: Building and
Sustaining the Sources of Innovation. Cambridge, MA: Harvard
Business School Press.
References
417
Lerner, J., Sorensen, M. and Stromberg, P. (2008). Private Equity and LongRun Investment: The Case of Innovation, in Globalization of Alternative
Investments Working Papers Volume 1: The Global Economic Impact of
Private Equity Reports: World Economic Forum, 2742.
Levinthal, D. A. and James, G. M. (1993). The myopia of learning, Strategic
Management Journal, 14(S2), 95112.
Lewis, J. I. (2007). Technology acquisition and innovation in the developing
world: wind turbine development in China and India, Studies in
Comparative International Development, 42(34), 20832.
Li, H. and Atuahene-Gima, K. (2001). The impact of interaction between
R&D and marketing on new product performance: an empirical analysis
of Chinese high technology rms, International Journal of Technology
Management, 21(1/2), 6175.
Li, J. and E. Martinot. 2007. Powering Chinas development: the role of
renewable energy. Worldwatch Institute. http://www.worldwatch.org/
node/5491.
Li, J. and Kozhikode, R. K. (2009). Developing new innovation models: shifts
in the innovation landscapes in emerging economies and implications for
global R&D management, Journal of International Management, 15,
32839.
Li J. L. (2007). Little Swan: building open innovation systems, Economic
Times (Chinese), 17 November.
Li J. T. (2010). Global R&D alliances in China: collaborations with universities and research institutes, IEEE Transactions on Engineering
Li, K. (2006). Please train talents that are needed in the 21st century,
Zhongguo Jiaoyu Xinwen (China Education News), 9 November.
Available at www.jyb.cn/high/gdjyxw/200611/t20061109_47908.html.
Li, X. B. (2009). Chinas regional innovation capacity in transition: an empirical approach. Research Policy 38, pp. 338357.
(2011). Sources of external technology, absorptive capacity, and innovation capability in Chinese state-owned high-tech enterprises. World
Development 39(7), 124048.
Lichtenthaler, U. (2008). Open innovation in practice: an analysis of strategic
approaches to technology transactions, IEEE Transaction on Engineering
(2009). Outbound open innovation and its effect on rm performance:
examining environmental inuences, R&D Management, 39(4),
31730.
and Holger, E. (2007). External technology commercialization in large
rms: results of a quantitative benchmarking study, R&D Management,
37(5), 38397.
418
References
and O. de Barros. (2009). The growth of Brazils direct investment abroad

and the challenges it faces. Columbia FDI Perspectives 13(17 August).
Lin, J. Y. (2011). New structural economics: a framework for rethinking
development, World Bank Research Observer, World Bank Group,
26(2), 193221.
Liu, C. Z. (2007). Lenovo: an example of globalization of Chinese enterprises, Journal of International Business Studies, 38, 57377.
Liu, L. (2008). The Evolution of Chinas Science and Technology Policy
(19752007), OECD Review of Innovation Policy: China, 38193.
Liu, X., Strangway, D. W. and Feng, Z. (2012) Environmental Innovation in
China. Southampton: WIT Press.
Liu, X. H. and Buck, T. (2007). Innovation performance and channels for
international technology spillovers: Evidence from Chinese high-tech
industries, Research Policy, 36(3), 35566.
Liu, X. L (2010). Chinas development model: an alternative strategy for
technological catch-up, in Fu, X. and Soete, L. (Eds.), The Rise of
Technological Power in the South. London: Palgrave MacMillan,
89106.
(2001). Comparing innovation systems: a framework and application to
Chinas transitional context, Research Policy, 30, 1091114.
Lokshin, B., Belderbos, R. and Carree, M. (2007). The Productivity Effects of
Internal and External R&D: Evidence from a Dynamic Panel Data
Model, UNU-Merit Working Paper Series 2007026, United Nations
University: Maastricht Economic and Social Research and Training
Centre in Innovation and Technology.
Lovell, C. A. K. (1996). Applying efciency measurement techniques to the
measurement of productivity change, Journal of Productivity Analysis,
7(2/3), 32940.
Lowe, M., George, G. and Alexy, O. (2012). Organizational identity and
capability development in internationalisation: transference, splicing and
enhanced imitation in Tescos US market entry, Journal of Economic
Geography, 12, 102154.
Lu, J. W. and Beamish, P. W. (2001). The internationalization and performance of SMEs, Strategic Management Journal, 22, 56584.
Lu, Qiwen. (2000). Chinas Leap into the Information Age: Innovation and
Organization in the Computer Industry. Oxford: Oxford University
Press.
Lundvall, B. (1992). National Systems of Innovation: Towards a Theory of
Innovation and Interactive Learning. London: Pinter.
Ma, A. and Van Assche, A. (2010). The role of trade costs in global production networks: Evidence from Chinas processing trade regime, paper
presented at 22nd CEA annual conference, Oxford.
References
419
Ma, W. (2004). From Berkeley to Beida and Tsinghua: The Development and
Governance of Public Research Universities in the US and China. Beijing:
Educational Science Press.
MacDonald, G., Yit-Seng, Y. and Xing, L. (2008) Innovation in China:
The Dawning of the Asian Century. London: Adonis & Abbey
Publishers Ltd.
Mairesse, J. and Mohnen, P. (2002). Accounting for innovation and measuring innovativeness: an illustrative framework and an application,
American Economic Review Papers and Proceedings, 92(2), 22630.
Malerba, F. (2002). Sectoral systems of innovation and production,
Research Policy, 31(2), 24764.
(2006). Innovation and the evolution of industries, Journal of
Evolutionary Economics, 16(1), 323.
Mallett, A., Ockwell, D. G., Pal, P., Kumar, A., Abbi, Y. P., Haum, R.,
MacKerron, G., Watson, J. and Sethi, G. (2009). UK-India collaborative
study on the transfer of low carbon technology: phase II nal report.
Sussex University and Institute of Development Studies. Available at
https://www.sussex.ac.uk/webteam/gateway/le.php?name=decc-uk-indiacarbon-technology-web.pdf&site=264.
Manseld, E. and Lee, J. Y. (1996). The modern university: contributor to
industrial innovation and recipient of industrial R&D support,
Marin, A. and Giuliani, E. (2011). MNC subsidiaries position in global
knowledge networks and local spillovers: evidence from Argentina,
Innovation and Development, 1(1), 91114.
Markusen, J. R. (2002). Multinational Firms and the Theory of International
Trade. Cambridge, MA: MIT Press.
Martinot, E. and Li, J. (2007). Powering Chinas Development: the Role of
Renewable Energy. Washington, D.C.: Worldwatch Institute.
Maskus, K. E. (2000). Intellectual Property Rights in the Global Economy.
Washington, DC: Institute for International Economics.
McDougall, P. and Oviatt, B. (2005). Dening international entrepreneurship and modeling the speed of internationalization, Entrepreneurship
Theory and Practice, 29(5), 537 S.
McMahon, W. W. (1987). Education and industrialization, Background
Paper for the 1987 World Development Report. Washington, DC:
World Bank.
Melitz, M. J. (2003). The impact of trade on intra-industry reallocations and
aggregate industry productivity, Econometrica, 71(6), 1695725.
Metcalfe, S. (1997). Technology systems and technology policy in an evolutionary framework, in Archibugi, D. and Michie, J. (Eds.), Technology,
420
References
Globalisation and Economic Performance, Cambridge: Cambridge

University Press.
Meyer, K. E. (2004). Perspectives on multinational enterprises in emerging
economies, Journal of International Business Studies, 35(4), 25976.
Milberg, W. (2007). Export Processing Zones, Industrial Upgrading and
Economic Development: A Survey, Background paper for ILO
Governing Board.
Miles, M. and Huberman, A. (1984). Qualitative Data Analysis. Beverly
Hills, CA: Sage Publications.
Ministry of Education. (1999). The Regulation Regarding the Protection and
Management of Intellectual Properties in Higher Education Institutions.
Act 3, No. 8120.
MNRE (Ministry of New and Renewable Energy, government of India).
(2006). XIth plan proposals for new and renewable energy. Available
at www.mnre.gov.in/pdf/11th-plan-proposal.pdf.
MOF (Ministry of Finance of China). (2001). Accounting System for Business
Enterprises. Available at www.mof.gov.cn.
MOFCOM (Ministry of Commerce of China). (2010). Statistics of foreign
investment. Available at www.mofcom.gov.cn.
MOFTEC (2000). Almanac of Chinas Foreign Economic Relations and
Trade, China Foreign Trade Press, Beijing.
Mohr, J. and Spekman, R. (1994). Characteristics of partnership success:
partnership attribute, communication behavior and conict resolution
techniques, Strategic Management Journal, 15(2), 13552.
MOST (Ministry of Science and Technology of China). (2010a). Statistics of
science and technology. Available at www.most.org.cn.
MOST (Ministry of Science and Technology). (2010b). Survey of Major
Specic Areas of S&T work and Studies on the 12th Five-year Plan
S&T Development Strategy, Ministry of Science and Technology,
Beijing, China (in Chinese).
MOST (Ministry of Science and Technology, China). (2010c). 2009
Innovation in environmental protection and energy conservation industries, in Ministry of Science & Technology 2010 Annual Report, 10.
Moulton, B. (1990). An illustration of a pitfall in estimating the effects of
aggregate variables on micro units, Review of Economics and Statistics,
72(2), 33438.
Mowery, D. C., Oxley, J. E. and Silverman, B. S. (1996). Strategic alliances
and interrm knowledge transfer, Strategic Management Journal, 17,
7791.
and Sampat, B. N. (2005). Universities in national innovations systems, in
Fagerberg, F., Mowery, D. C. and Nelson, R. R. (Eds.), The Oxford
Handbook of Innovation. New York: Oxford University Press, 20939.
References
421
Mu, Q. and Lee, K. (2005). Knowledge diffusion, market segmentation and

technological catch-up: the case of the telecommunication industry in
China, Research Policy, 34(6), 75983.
Mu, R. and Fan, Y. (2011). Framework for building national innovation
capacity in China, Journal of Chinese Economic and Business Studies,
9(4), 31727.
Nakai Y. and Tanaka, Y. (2010). Chinese companys IPR Strategy: How
Huawei Technologies succeeded in dominating overseas market by
sideward-crawl crab strategy, Technology Management for Global
Economic Growth (PICMET) 2010 Proceedings, 1822 July, Phuket,
Thailand.
Narula, R. (2002). Innovation systems and inertia in R&D location:
Norwegian rms and the role of systemic lock-in, Research Policy, 31,
795816.
Narula, R. (2003) Globalisation and Technology: Interdependence,
Innovation Systems and Industrial Policy. Cambridge: Polity Press.
Narula, R. and Zanfei, A. (2004). Globalisation of innovation: the role of
multinational enterprises, in Fagerberg, J., Mowery, D. C. and
Nelson, R. R. (Eds.), Handbook of Innovation. New York: Oxford
University Press, 31845.
National Bureau of Statistics (NBS) (2013). China Statistical Yearbook,
Beijing: China Statistics Press.
Needham, J. (2004). Science and Civilisation in China. Cambridge:
Cambridge University Press.
Nelson, R. R. (1959). The simple economics of basic scientic research,
Journal of Political Economy, 67, 297306.
(1986). Institutions supporting technical advance in industry, American
Economic Review, 76, 18689.
(Ed.). (1993). National Innovations Systems: A Comparative Analysis.
Oxford: Oxford University Press.
and Winter, S. (1982). An Evolutionary Theory of Economic Change.
Cambridge, MA: Harvard University Press.
Nickell, S. J. (1996). Competition and corporate performance, The Journal
of Political Economy, 104(4), 72446.
Niosi, J. (1999). The internationalization of industrial R&D: from technology transfer to the learning organization, Research Policy, 28, 10717.
Nolan, P. (2001). China and the Global Business Revolution. Basingstoke,
UK: Palgrave.
Norman, M. and Stoker, B. (1991). Data Envelopment Analysis: The
Assessment of Performance. Chichester, UK: John Wiley.
OECD. (1987). Structural Adjustment and Economic Performance. Paris:
OECD.
422
References
OECD. (1997). Proposed Guidelines for Collecting and Interpreting

Technological Innovation Data: The Oslo Manual, Organization for
Economic Development and Co-operation, Paris.
OECD. (2002). Frascati Manual 2002: Proposed Standard Practice for Surveys
on Research and Experimental Development, The Measurement of
Scientic and Technological Activities, OECD Publishing.
OECD. (2003a). The Sources of Economic Growth in OECD Countries.
Paris: OECD.
OECD. (2003b). OECD Science Technology and Industry Scoreboard.
OECD. Paris.
OECD. (2004). OECD Information Technology Outlook 2004. Available at
http://www.oecd.org/internet/ieconomy/oecdinformationtechnologyout
look2004.htm.
OECD. (2005). OSLO Manual: Guidelines for Collecting and Interpreting
Innovation Data, Paris: OECD.
Pack, H. and Saggi, K. (2006). Is there a case for industrial policy? a critical
survey, The World Bank Research Observer, 21(2), 26797.
Pack, H. and Westphal. L. E. (1986). Industrial strategy and technological
change: theory versus reality, Journal of Development Economics,
22(1), 87128.
Pacter, P. and Yuen, J. (2001). New comprehensive accounting system adopted in China, China Financial Reporting Update, 2.
Padilla, R. and von Tunzelmann, N. (2008). Technological capabilities and
global-local interactions. The electronics industry in two Mexican
regions, World Development, 36(10), 19802003.
Parsaye, K. (1989). Intelligent Databases. New York: Wiley.
Pavitt, K. (1980). Industrial R&D and the British Economic Problem, R&D
Management, 10, 149158.
Pearce, R. D. (1999). Decentralised R&D and strategic competitiveness:
globalised approaches to generation and use of technology in multinational enterprises, Research Policy, 28(23), 15778.
(2005). The globalisation of R&D: key features and the role of TNCs, in
Globalisation of R&D and Developing Countries: Proceedings of an
Expert Meeting. New York and Geneva: UNCTAD, United Nations.
Polanyi, M. (1967). The Tacit Dimension. London: Routledge & Kegan
Paul.
Pomfret, J. (2010). History of Telecom Company illustrates lack of strategic
trust between U.S., China,Washington Post, 8 October. Available at w
ww.washingtonpost.com/wp-dyn/content/article/2010/10/07/AR2010
100707210.html?wprss=rss_business.
Porter, M. E. (1985) Competitive Advantage, Free Press, New York, 1985.
(1990). The Competitive Advantage of Nations. New York: Free Press.
References
423
and Stern, S. (1999). The New Challenge to Americas Prosperity: Findings

from the Innovation Index. Washington, DC: Council on Competitiveness.
Powell, W. W. and Grodal, S. (2005). Networks of innovators, in
Fagenberg, J., Mowery, D. C. and Nelson, R. R. (Eds.), The Oxford
Koput, K. W. and Smith-Doerr, L. (1996). Inter-organizational collaboration and the locus of innovation: networks of learning in biotechnology,
Administrative Science Quarterly, 41(1), 11645.
Prabhu, J. C., Chandy, R. K. and Ellis, M. (2005). The impact of acquisitions
on innovation: poison pill, placebo, or tonic? Journal of Marketing,
69(1), 11430.
Prange, C. (2012). Ambidextrous internationalization strategies: the case of
Chinese rms entering the world market, Organizational Dynamics, 41,
24553.
Prasso, S. (2011). What Makes China Telecom Huawei So Scary? Fortune,
July 28.
Pylyshyn, Z. (1981). The imagery debate: Analogue media versus tacit
knowledge. Psychological Review, 88, 1645.
Rigby, D. and Zook, C. (2002). Open-market innovation, Harvard Business
Review, 80(10), 8089.
Roberts, M. and Tybout, J. (1997). The decision to export in Colombia: an
empirical model of entry with sunk costs, American Economic Review,
87, 545564.
Rodrik, D. (2004). Industrial policy for the twenty-rst century. Available at
www.hks.harvard.edu/fs/drodrik/Research%20papers/UNIDOSep.pdf
Romer, P. (1994). The origins of endogenous growth, Journal of Economic
Perspectives, American Economic Association, 8(1), 322.
Romer, P. M. (1990). Endogenous Technological Change, Journal of Political
Economy, 98 (5), part II, S71S102.
Royal Society. (2011). Knowledge, Networks and Nations: Global Scientic
Collaboration in the 21st Century. London: Royal Society.
Ruane, F. and Sutherland, J. (2005). Foreign direct investment and export
spillovers: how do export platforms fare?, IIIS Discussion Paper No 58.
Sainsbury, Lord of Turville. (2007). Race to Top: Sainsbury Review of
Science and Innovation., London: HM Treasury.
Sasidharan, S. and Kathuria, V. (2008). Foreign Direct Investment and R&D:
Substitutes or Complements a Case of Indian Manufacturing after
1991 Reforms. Oxford University SLPTMD Working Paper, no 021.
Schumpeter, J. (1942). Capitalism, Socialism, and Democracy. New York:
Harper & Bros.
Schumpeter, J. A. (1994). Capitalism, Socialism and Democracy. London:
Routledge.
424
References
Schwaag, S. (2006). China, from shop oor to knowledge factory, in

Karlsson, M. (Ed.), The Internationalization of Corporate R&D.
Stursund: Swedish Institute for Growth Policy Studies, 22766.
Shi, Z. (2008). Does Chinese traditional culture handicap the cultivation of
creative talents? Zhongguo Jiaoyu Xuekan (Journal of the Chinese
Society of Education), 8, 16.
Simon, D. F. and Cao, C. (2009). Chinas Emerging Technological Edge.
Cambridge: Cambridge University Press.
Siotis, G. (1999). Foreign direct investment strategies and rms capabilities,
Journal of Economics & Management Strategy, 8(2), 25170.
Sjoholm, F. (1999), Technology gap, competition and spillovers from direct
foreign investment: evidence from establishment data, Journal of
Development Studies, 36(1), 5373.
Skrondal, A. and Rabe-Hesketh, S. (2004). Generalised Latent Variable
Modeling: Multilevel, Longitudinal and Structural Equation Models.
Boca Raton, FL: Chapman & Hall/CRC.
Smith, K. (2004). Measuring innovation, in Fagerberg, J., Mowery, D. C.
and Nelson, R. R. (Eds.), The Oxford Handbook of Innovation. New
York: Oxford University Press, 14877.
Soete, L. (1985). International diffusion of technology, industrial development and technological leapfrogging. World Development, 13(3),
409422.
Someren, T. C. R, and Someren-Wang, S. (2013). Innovative China:
Innovation Race between East and West. Berlin: Springer.
State Statistical Bureau of China (SSB). (20002006). China Statistical
Yearbook. Beijing: China Statistics Press.
Steensma, H. K., Barden, J. Q., Dhanaraj, C. and Lyles, M. (2004). The
inuence of power, learning, and conict on the internationalzation of
international joint ventures, Academy of Management Proceedings, 1,
L1L6.
Stewart, F. (1983). Macro-policies for appropriate technology: an introductory classication, International Labour Review, 122(3), 279.
Stiglitz, J. (1989). The Economic Role of the State. London and New York:
Wiley-Blackwell.
Stockdale, B. (2002), UK Innovation Survey, Department of Trade and
Industry, London, available at: www.dti.gov.uk/iese/ecotrends.pdf
October 2004.
Strangway, D., Liu, X. and Feng, Z. (2009). Policy report of China Council
for International Cooperation on Environment and Development, in
Chinese Academy of Science (Ed.), Building an Environmentally-friendly
Society through Innovation: Challenges and Choices. Mimeo.
References
425
Sun, H. and Xu, Z. Q. (2009). On paths from open to indigenous innovation:

a case study of coal liquefaction technological innovation, S&T AND
ECONOMY (Chinese), 127(22), 79.
Sun, Z., Xie, W., Tian, K. and Wang, Y. (2014). Capability accumulation
and the growth path of Lenovo, in Yu, F.-L. T. and Yan, H.-D. (Eds.),
Handbook in East Asia Entrepreneurship. London: Routledge. Available
at www.tmd-oxford.org/content/publications.
Symeonidis, G. (2001). Price Competition, Innovation and Protability:
Theory and UK Evidence. CEPR Discussion Paper No. 2816. London:
Centre for Economic Policy Research.
Takahashi, K. I., Tatemichi, H., Tanaka, T., Nishi, S. and Kunioka, T. (2004).
Environmental impact of information and communication technologies
including rebound effects, International Symposium on Electronics and
the Environment (ISEE04), 1316.
Tan, Y. (2011) Chinnovation: How Chinese Innovators are Changing the
World. Singapore: John Wiley & Sons.
Teece, D. J. (1986), Proting from technological innovations, Research
Policy, 15(6), 285306.
and Chesbrough, H. W. (2005). Globalisation of R&D in the Chinese
Semiconductor Industry. Report to the Alfred P. Sloan Foundation.
and Shuen, A. (1997). Dynamic capabilities and strategic management.
Strategic Management Journal, 18(7), 50933.
Tether, B. (2002). Who co-operates for innovation, and why: an empirical
analysis, Research Policy, 31(6), 94767.
Tobin, J. (1958). Estimation of relationship for limited dependent variables,
Econometrica, 26(1), 2436.
Trajtenberg, M. (1990). Patents as Indicators of Innovation, Economic
Analysis of Product Innovation. Cambridge (MA): Harvard University
Press.
Tsai, W. (2001). Knowledge transfer in intraorganizational networks:
effects of network position and absorptive capacity on business unit
innovation and performance, Academy of Management Journal,
44(5), 9961004.
Tsang, E. W. (1999). A preliminary typology of learning in international
strategic alliances, Journal of World Business, 34(3), 21129.
Tushman, M. L. and Anderson, P. (1986). Technological discontinuities and
organization environments, Administrative Science Quarterly, 31(3),
43965.
UNCTAD. (1997). World Investment Report: Transitional Corporations,
Market Structure and Competition Policy. New York and Geneva:
United Nations.
426
References
UNCTAD. (2005a). World Investment Report: Transnational corporations

and the internationalization of R&D. New York and Geneva: United
Nations.
UNCTAD. (2005b). Globalisation of R&D and Developing Countries:
Proceedings of an Expert Meeting. New York and Geneva: United Nations.
UNCTAD. (2014). Trade Statistics Database, available at www.unctad.org.
Varum, C. A. and Huang, C. (2007) China: Building An Innovative
Economy. UK: Chandos.
Von Hippel, E. (1988). Sources of Innovation. New York: Oxford University
Press.
Von Tunzelmann, N. and Acha, V. (2005). Innovation in low-tech industries,
in Fagerberg, J., Mowery, D. C. and Nelson, R. R. (Eds.), The Oxford
Vrande,V. van de, de Jong, J. P. J., Vanhaverbeke, W. and de Rochemont, M.
(2009). Open innovation in SMEs: trends, motives and management
challenges, Technovation, 29(6/7), 42337.
Wade, R. (1990). Governing the Market: Economic Theory and the Role of
Government in East Asian Industrialization. Princeton, NJ: Princeton
University Press.
Walz, R. (2008). Technological competencies in sustainability technologies in
BRICS countries 12, paper presented at SLPTMD conference for
Confronting the Challenge of Technology for Development: Experiences
from the BRICS, 2930 May. Oxford, UK.
Wang, E. C. and Huang, W. (2007). Relative efciency of R&D activities: A
cross-country study accounting for environmental factors in the DEA
approach, Research Policy, 36, 26073.
Wang, H. (2012). Globalizing China: The Inuence, Strategies and Successes
of Chinese Returnees. Bradford, UK: Emerald Publishing.
Wang, H. Y. and Liu, X. (2007). Changes, decision-making and trends of
Chinese S&T policy, in Fang, X. (Ed.), Chinese Technological
Innovation and Sustainable Development (Chinese). Beijing: Science Press.
Wang, L. 2006. CHUN-LAN: building open innovation platform,
Economic Times (Chinese), 12 April.
Wang, X. (2008). Huawei, Global Marine Systems in telecom JV, China
Daily, 18 December.
Willoughby, K. W. (1990). Technology choice: a critique of the appropriate
technology movement (Book), Futurist, 24(4), 45.
Wong, P. K. (1999). National innovation systems for rapid technological
catch-up: an analytical framework and a comparative analysis of
Korea, Taiwan and Singapore, paper presented at the DRUID Summer
Conference on National Innovation Systems, Industrial Dynamics and
Innovation Policy. Rebild, Denmark.
References
427
World Bank. (1996). Industrial Restructuring: Experience, Future Challenges.

A World Bank Operations Evaluation Study. Washington, DC: World
Bank.
(2005). Economic Growth in the 1990s: Learning from a Decade of
Reform. Washington, DC: World Bank.
World Bank and Development Research Center of the State Council. (2013).
China 2030: Building a Modern, Harmonious and Creative Society.
Washington DC: World Bank.
Wu, G. (2010). Research on the content and development trend of science
and technology policy, economic policy and innovation policy, Centre
for Innovation Research, Tsinghua University, unpublished project
report.
Wu, G. S., Wang, Y., Xiong, H. R. and Wang, Z. F. (2009). Study on the
contents and trends of Chinese S&T policy, economy policy and innovation policy, Internal Research Report (Chinese), Beijing Municipal
Science & Technology Commission.
Wu, J. (2013). Choices of Chinas Growth Model. Shanghai: Shanghai Far
East Publisher (in Chinese).
Wu, W. P. (2007). Cultivating research universities and industrial linkages in
China: the case of Shanghai, World Development, 35(6). 107593.
Xie, W. (2004). Technological learning in Chinas color TV (CTV) industry,
Technovation, 24, 499512.
Xie, W. and White, S. (2004). Sequential learning in a Chinese spin-off: the
case of Lenovo Group Limited, R&D Management, 34(4), 40722.
Xie, W. and Wu, G. (2003). Differences between learning processes in small
tigers and large dragons learning processes of two color TV (CTV) rms
within China, Research Policy, 32(8), 146379.
Xiong, H. R. and Li, J. Z. (2008). Innovative strategy of resource distribution
for latecomers: perspective from opening, China Economic Herald
(Chinese), 3 June.
Xu, B. (2000). Multinational enterprises, technology diffusion, and host
country productivity growth, Journal of Development Economics,
62(2), 47793.
Xu, Z., Chen, Y. T. and Ke, W. (2009), The mechanism of corporate venture
capital for open innovation strategy, Science of Science and Management
of S&T (Chinese), 4, 1304.
Xue, L., (1997). A historical perspective of Chinas innovation system reform:
a case study, Journal of Engineering and Technology Management, 14,
6781.
Yli-Renko, H., Autio, E. and Tontti, V. (2002). Social capital, knowledge,
and the international growth of technology-based new rms,
International Business Review, 11(3), 279304.
428
References
Yu, K. L. and Wang, Y. M. (2008). The inuence of open innovation on

indigenous innovation: a M&A perspective, MANAGEMENT
WORLD (Chinese), 4, 15059.
Yu, M. (2010). Processing Trade, Firms Productivity, and Tariff
Reductions: Evidence from Chinese Products, Peking University, China
Center for Economic Research,Working Paper No. E2010007.
Yuan, J. (2002). The Institutional Logic of University Start-Ups. Xuzhou,
China: The Chinese Mining University Press.
Zahra, S. A. (1996). Governance, ownership, and corporate entrepreneurship: the moderating impact of industry technological opportunities,
Academy of Management Journal, 39(6), 171335.
Ireland, R. and Hitt, M. A. (2000). International expansion by new venture
rms: international diversity, mode of market entry, technological learning and performance, Academy of Management Journal, 43(5), 92550.
Zain, M. and Ng, S. I. (2006). The impacts of networks relationships on
SMEs internationalization process, Thunderbird International Business
Review, 48(2), 183205.
Zeng, M. and Williamson, P. (2007). Dragons at Your Door: How Chinese
Cost Innovation Is Disrupting Global Competition. Cambridge, MA:
Harvard Business School Press.
Zhang, J. (2003). The Development of High-tech Enterprises in Chinas
Universities. Wuhan: Huazhong Science and Technology University
Press.
Zhang, K. H. (2009). Rise of Chinese Multinational Firms. The Chinese
Economy, 42(6): 8196.
Zhang, Y. (2009). Alliance-based Network View on Chinese Firms Catchingup: Case Study for Huawei Technologies Co. Ltd. UNU-MERIT
Working Papers, ISSN 18719872.
Zheng C. (2014). The inner circle of technology innovation: a case study of
two Chinese rms, Technological Forecasting & Social Change, 82,
14048.
Zheng, G., He, Y. B., Chen, J. et al. (2008). How Chinese manufacturing
enterprises improve the international competitiveness by indigenous
innovation: A case study from CIMC, SCIENCE RESEARCH
MANAGEMENT (Chinese), 29(4), 7, 95102.
Zhou, C. and Li, J. (2008). Product innovation in emerging market-based
international joint ventures: an organizational ecology perspective,
Journal of International Business Studies, 39, 111432.
Zhou, K. Z. and Li, C. B. (2012). How knowledge affects radical innovation:
knowledge base, market knowledge acquisition, and internal knowledge
sharing, Strategic Management Journal, 33, 1090102.
References
429
Zhou, X., Zhao, W., Li, Q. and Cai, H. (2003). Embeddedness and contractual relationships in Chinas transitional economy, American
Sociological Review, 68, 75102.
Zhou, Y. (2006). Features and impacts of the internationalisation of R&D by
transnational corporations: Chinas case, in Globalisation of R&D and
Developing Countries. New York and Geneva: UNCTAD, United
Nations.
Zhu, Y. (2010). An Analysis on Technology Spillover Effect of Foreign Direct
Investment and Its Countermeasures, International Journal of Business
and Management, 5(4), 178192
Zhu, Z. H. and Chen, J. (2008), Explorative learning and exploitative learning: antinomy or synergy?, Studies in Science of Science (Chinese), Vol.
26 No. 5, pp. 105260.
Zi, Z., (2013). The purpose of education, Zhongguo Jiaoyu Bao (China
Education Paper), 9 December. Available at www.xxyedu.net/Article
Detail.aspx?articleid=4743.
Index
absorptive capacity of FDI, regional

innovation performance, 4852
Directed Technical Change theory, 112
European Union (EU) innovation
efciency, cross-country gap, 325
expenditure for acquisition of foreign
technology, 232
export processing zones (EPZs), foreign
direct investment (FDI), 745
foreign direct investment (FDI)
absorptive capacity, regional
innovation performance, 4852
in China, 529
Chinas outward, by industry
distribution, 284
in export processing zones (EPZs),
745
indigenous rms, export
competitiveness and, 10
innovation capacity and, 10
and innovation capacity, empirical
evidence, 5965
innovation in China and, 11213
innovation-growth linkage, coastal
and inland regions, 69
intensity, foreign asset share in total
industrial assets, 534
regional FDI stock distribution,
534
regional innovation capacity impact
of, 63, 68
regional innovation efciency and,
658
technology transfer and, 2589
technology upgrading and, 10812
trade in China, 523
430
foreign direct investment (FDI),

spillovers
exports in technology-intensive
industries and, 802
from processing-trade-FDI (PT-FDI),
7880
theory, literature, 767
foreign technology imports, 234
Huawei Technologies P/L. See
internalization, reverse learning,
capabilities upgrading
imitation to innovation, 34, 31823
indigenous rms, export
competitiveness, 10
industrial competitiveness, 3
industrial output, 3
innovation. See also open innovation;
radical innovation; regional
innovation
capacity, empirical evidence, 5965
Chinas capabilities, literature and
research on, 79
constraints, risks to, 1956
denition, sources of, 47
foreign direct investment (FDI) and,
11213
foreign, indigenous enterprises, 59
growth, developmental impact, 392
inputs, R&D and research personnel,
1624
invention applications, regional
distribution, 546
national innovation performance,
determinants, 32930
national innovation systems,
technology acquisition, adaptation
and development, 26874
Index
new products sales, exports across
ownership structures, 379
output across industries, new product
sales and export values, 323
output across industries, patent
applications, 323
outputs by rm ownership, 34
ownership structures and, 358
performance, patents, published
journals, high-tech export, 249
quality, science citation ranking,
279
R&D expenditure, industry
distribution, 2934
R&D expenditure, regional disparity,
3943
regional economic growth and, 689
regional efciency, 658
role of state, industrial policy and, 7,
14, 2323
sales of new products, regional
distribution, 546
sector development, 7
university role, 12
innovation capacity
basic innovation capacity
determinants, 3301
dened, 3256
emerging economies, 34950
emerging economies, innovation
efciency, 3501
empirical evidence, foreign direct
investment (FDI), 5965
foreign direct investment (FDI), 10
innovation activity, cross-country
differences, 3336
innovation efciency determinants,
3313
patenting determinants and, 3267,
328
stochastic frontier analysis (FSA) and,
3278
innovation efciency, cross-country gap
European Union (EU) and, 325
innovation-growth linkage, coastal and
inland regions, 69
internalization, reverse learning,
capabilities upgrading, 297
Chinese rms, overview, 2838
rms innovativeness, OFDI, 2878
431
Huawei Technologies P/L, 2915
Huawei Technologies P/L, history of
alliance, 3002
investment in developing countries,
2847
knowledge sourcing and, 27980
learning from subsidiaries, 3048
learning process, studies, 2813
manufacturing rms, overseas
investment, 284
outward directed investment, China,
283
outward FDI, 284
research, criteria for case inclusion,
28991
research, data collection, 291
research design, 289
research method, 2889
reverse learning from cooperation,
3004
reverse learning from customers,
297300
ZTE Corporation, 2956
international collaboration, radical
innovation, 31416
collaborative innovation activities,
317
collaborative innovation in China,
overview, 316
collaborative types, regional
distribution, 31718
innovation performance impact,
31920
novel innovation and, 3203
policy implications, 3234
scientic literature, 31416
national innovation performance (NIP)
capabilities building and, 3657
China, 360
competition, product market and,
3623
competition to motivate innovation
and, 369
determinants, theoretical framework,
35965
factor markets and, 363
human capital and, 360
human resource management
practices and, 36971
432
national innovation performance (cont.)
incentives strengthening, multiple
levels, 36871
innovation system, closed vs. open,
3645, 3757
institutions and, 3634
institutions development and, 3713
intellectual property rights (IPR) and,
371
investment for innovation, market
role, 3689
macroeconomic incentives and, 3612
national innovation performance
(NIS) framework, 3589
national innovation system (NIS)
approach, 358
physical investment and, 3601
policy choices, 36575
policy-making, implementation
coordination, 3734
state capabilities and, 361
technological efforts and, 361
21st century space, 3757
National Innovation Strategy (NIS), 15,
3589
ONIS. See Open National Innovation
System
open innovation, 1112
breadth of openness determinants,
ordered-Logit model estimates, 182
challenges, 1667
dened, 141
depth of openness determinants,
ordered-Logit model estimates,
1857
determinants of, 1713
rm size and industry, moderating
effect, 1934
globalization and, 168
government policy, rm practices,
1469
inbound, policies towards, 1502
inbound vs. outbound, 141
innovation, constraints, risks, 1737
measurement, control variables, 1801
measurement, dependent variables,
1789
measurement, independent variables,
17980
Index
open innovation networks, policies
towards, 1536
outbound, policies towards, 1523
overview, 1426
ownership effect, 1903
policies affecting, 14956
practices, 15863
public policies for, 1568
research data, methodology, 1778
research, descriptive statistics and
correlations, 182
research, OLS robustness check,
18790
research on, 170
selected cases, 159
trends, 1635
types, mode and frequencies, 163
Open National Innovation System
(ONIS), 14
Chinas path to innovation and,
3813
constraints and risks, universities role,
3889
foreign, indigenous innovation
relationship in, 3834
incentives and institutions, efciency
of innovation, 38991
indigenous-foreign
combination, stages of
development, 3856
knowledge sourcing and, 3867
leading players, state-owned
enterprises (SOEs), 3878
optic ber, cable industry
production chain, 242
research, denition of variables, 244
research, factor analysis, 2478
technology characteristics perception,
244
patents, patent performance, 1314,
246. See also innovation capacity
applications vs. granted, 246
mean patent application numbers,
412
ownership structure and, 37
patenting capacity, selected countries,
3468
patenting efciency determinants,
343
Index
patenting efciency, selected
countries, 3468
patenting frontier estimation, 33640
patents granted by USPTO, 336, 350
triadic patent families, across BRICS,
26
triadic patent families, China and
other OECD economies, 26
policy choices, implications, 3924
innovation, 3234
national innovation performance
(NIP), 36575
open innovation, government policy,
rm practices, 1469
policy-making, implementation
coordination, national innovation
performance (NIP), 3734
processing trade-FDI (PT-FDI),
1067
role of state, industrial policy, 7, 14,
2323
processing trade-FDI (PT-FDI),
745. See also technology,
technology-intensive industries
export performance, indigenous
export growth, 1013
policy implications, 1067
spillovers, domestic rms export
value, 98100, 101
spillovers, indigenous rms export
performance, 1035
radical innovation, 13. See also
innovation
R&D
collaboration with partners,
universities, 21517
composition, by funding sources, 19
composition, by types of activities,
1819
expenditure, above scale
manufacturing industries, 301
expenditure, across ownership
structures, 352
expenditure, China, 19952012,
1617
expenditure, China vs. other
economies, 19952012, 17
433
expenditure, growth rate, 568
expenditure, R&D/GDP ratios,
3941
funding sources, 423
globalization of, 72
government spending on, 15
indigenous, technology transfer and
progress in emerging economies,
25761
industrial, regional distribution, 546
international R&D stock analysis,
summary statistics, 11417
investment, composition of, 1619
personnel, across sectors, 31
personnel, by executive entity, 20
personnel, comparison in per
thousand employees, 22
personnel, compositions of, 204
research personnel and, 16
sectoral pattern, 114
total national R&D, percentage
GDP for major industrial countries,
3345
total number personnel, 20
regional FDI stock distribution, 534
regional innovation
foreign direct investment (FDI),
absorptive capacity, innovation
performance, 4852
multinational enterprises (MNEs)
and, 48
research on, 47
regional innovation capacity
impact, foreign direct investment
(FDI), 63, 68
regional innovation efciency, foreign
direct investment (FDI), 658
solar PV industry, India and China
leading companies, details, 2657
mixing, technology creation and
acquisition sequencing, 268
national innovation systems,
technology acquisition,
adaptation and development,
26874
renewable technology targets, China,
2645
renewable technology targets, India,
263
434
solar PV industry, India and China
(cont.)
science, technology development,
2614
sustainability oriented innovation
systems, 2724
technology transfer, adopted
indigenous knowledge creation
mechanisms, 268
technology transfer, indigenous
innovation, 2658
South-South technology transfer,
developing world, 38991
tacit knowledge. See also optic ber,
cable industry
acquisition, 12
emboldens of, 250
external vs. internal sources of,
23842
research data, 2434
research methodology, 12
technological learning, 1689, 171,
196. See also optic ber, cable
industry
impact of, 236
industrialization and, 2367
knowledge sources and, 24850
technology, technology-intensive
industries
efciency improvement, GMM
estimates, 132
export performance, top 10, 924
exports, foreign direct investment
(FDI) spillovers, 802
exports, innovation indicators,
domestic and foreign rms, 902
rm-level production, export data,
867
high-tech products, export and
import volume, 334
indigenous vs. foreign, 67
Probit model estimates, export
decision, 94
processing exports percentage,
electronic industry, 92
processing exports share, China total,
812
product-level trade data, 8790
technical change determinants, GMM
estimates, 132
Index
technical change, efciency
improvement, 1246
technology acquisition expenditure, 54
technology diffusion, 56
upgrading, developing countries,
1011
upgrading, drivers, 1236
technology transfer
foreign direct investment (FDI) and,
2589
imports and, 259
licensing and, 259
outward foreign direct investment
(FDI) and, 260
technology transfer, foreign direct
investment (FDI) and, 2589
technology upgrading, foreign direct
investment (FDI) and, 10812
total factor productivity (TFP)
foreign, indigenous R&D efforts and,
11921
growth determinants and technical
change, GMM estimates, 1268
growth estimation, 11719
growth, technical efciency, 1213
trade in China, 523
United Kingdom (UK) industrial
innovation, university role, 22631
universities, role, 12, 2012
external resources utilization, 21517
industrial innovation in China and,
2078, 21723
industrial innovation in China,
university-concentrated cities, 2246
industrial innovation UK, 22631
literature, theoretical framework,
2026
Open National Innovation System
(ONIS), 3889
research, control variables, 21011
research data, 21315
research, dependent variable
measurement, 209
research, direct university
measurement, 210
research methodology, data, 20815
ZTE Corporation. See internalization,
reverse learning, capabilities
upgrading

China's Path To Innovation

Uploaded by

Copyright:

Available Formats

China's Path To Innovation

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

China's Path To Innovation

Uploaded by

Copyright:

Available Formats

Chinas Path to Innovation

Professor Fu has integrated a variety of scholarly articles from the rm level to

University Printing House, Cambridge CB2 8BS, United Kingdom

Innovation in China since the reforms: An overview

Part I International knowledge transfer and technological

Foreign direct investment, absorptive capacity and regional

Processing trade, FDI and international competitiveness of

Indigenous and foreign innovation efforts and

Development of indigenous innovation capacity

The role of state policy in shaping innovation practices: The

Open innovation as a response to constraints and risks

The dual role of universities in industrial innovation:

Technological learning, tacit knowledge acquisition and

Leapfrogging in green technology: The solar-PV industry

Towards a global innovation leader

Internationalisation, reverse learning and capabilities

International collaboration and radical innovation

Innovation efciency and the cross-country gap in

Capabilities, incentives, institutions and national

Conclusions: Open national innovation system and Chinas

2.1 Chinas R&D expenditure, 19952012

2.18 Export and import volume of high-tech products,

5.1 Technical efciency by ownership and technology

3.1 FDI intensity: share of foreign assets in total industrial

6.3 The importance of information sources in innovation

11.1 Industry distribution of Chinas OFDI: by FDI stock and

This book is an outcome of my research on innovations in China over

Indigenous and foreign innovation efforts and drivers of technological

National innovation performance

modern sector. China now faces signicant challenges in moving from

Innovation and its sources

industrialisation and catch-up in developing countries (Romer, 1990). In

a learning process but a large proportion of knowledge spillovers take

empirical evidence on the effects of the sources of indigenous or foreign

processes and drivers of innovation at different stages of development and

The objective and structure of the book

at the early stage of the reforms. Part II analyses the development of

ones. State-owned rms appear to be least responsive in the use of open

in innovation (proxied by patenting). A patent production frontier is rst

Innovation in China since the reforms:

2.2 Innovation inputs: R&D and research personnel

R&D investment and composition of R&D

Research and development involves the creative work undertaken on a systematic

Innovation in China since the reforms

100 million RMB

R&D spending as percentage of GDP

Figure 2.1 Chinas R&D expenditure, 19952012

R&D spending in GDP

Figure 2.2 R&D expenditure in China and other economies, 19952012

Basic research is experimental or theoretical work undertaken primarily to acquire

Innovation in China since the reforms

Figure 2.3 Composition of R&D by types of activities: basic research, applied

big proportion of the increase in R&D expenditure in the past decades

Other funding may include self-raised funding, in particular for independent

Figure 2.4 R&D composition by funding sources, 20032011

Innovation in China since the reforms

Researchers per 10,000 people

Figure 2.5a Total number of R&D personnel

Figure 2.5b R&D personnel by executive entity, 2009