aggregate and regional productivity growth in …aggregate and regional productivity growth in...
Post on 10-Aug-2020
2 Views
Preview:
TRANSCRIPT
Aggregate and regional productivity growth in Chineseindustry, 1978-2002Citation for published version (APA):Wang, L. (2009). Aggregate and regional productivity growth in Chinese industry, 1978-2002. TechnischeUniversiteit Eindhoven. https://doi.org/10.6100/IR641134
DOI:10.6100/IR641134
Document status and date:Published: 01/01/2009
Document Version:Publisher’s PDF, also known as Version of Record (includes final page, issue and volume numbers)
Please check the document version of this publication:
• A submitted manuscript is the version of the article upon submission and before peer-review. There can beimportant differences between the submitted version and the official published version of record. Peopleinterested in the research are advised to contact the author for the final version of the publication, or visit theDOI to the publisher's website.• The final author version and the galley proof are versions of the publication after peer review.• The final published version features the final layout of the paper including the volume, issue and pagenumbers.Link to publication
General rightsCopyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright ownersand it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.
• Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain • You may freely distribute the URL identifying the publication in the public portal.
If the publication is distributed under the terms of Article 25fa of the Dutch Copyright Act, indicated by the “Taverne” license above, pleasefollow below link for the End User Agreement:www.tue.nl/taverne
Take down policyIf you believe that this document breaches copyright please contact us at:openaccess@tue.nlproviding details and we will investigate your claim.
Download date: 06. Nov. 2020
Aggregate and Regional Productivity Growth inAggregate and Regional Productivity Growth inAggregate and Regional Productivity Growth inAggregate and Regional Productivity Growth in
Chinese Chinese Chinese Chinese Industry,Industry,Industry,Industry, 1978 1978 1978 1978----2002200220022002
Lili Wang Lili Wang Lili Wang Lili Wang
Aggregate and Regional Productivity Growth in Chinese Industry, 1978-2002 / by Lili
Wang. – Eindhoven: Technische Universiteit Eindhoven, 2009. - Proefschrift.-
A catalogue record is available from the Eindhoven University of Technology Library
ISBN: 978-90-386-1562-2
NUR: 781
Keywords: Productivity growth / Regional inequality/ Structural change/
Technological spillovers
Cover design: Joep van Son
Photos: Niels Philipsen
Printed by Eindhoven University Press
Aggregate and Regional Productivity Growth in
Chinese Industry, 1978-2002
Proefschrift
ter verkrijging van de graad van doctor aan de
Technische Universiteit Eindhoven,
op gezag van de
Rector Magnificus, prof.dr.ir. C.J. van Duijn,
voor een commissie aangewezen door het College voor
Promoties in het openbaar te verdedigen op
woensdag 11 maart 2009 om 16.00 uur
door
Lili Wang
Dit proefschrift is goedgekeurd door de promotor:
prof.dr. A. Szirmai
Acknowledgements
The last four years of working on my PhD thesis have been like an unforgettable
journey which did not only enhance my academic career, but also enriched my
experience in many other ways. Although I cannot mention everyone who has
supported me during the past years, I would like to express sincere thanks to some of
them.
First of all, I am grateful to my promoter, Professor Eddy Szirmai. His
enthusiasm in research and his constructive suggestions helped me all the time
through the writing of my PhD. He is not only a wise professor but also an
understanding gentleman. I appreciated his assistance during the last stages of my
work while I was pregnant and afterwards had to combine the finishing of my thesis
with a new job and taking care of a baby. Working with Eddy has been a great
pleasure for me.
Special thanks go to Dr. Huub Meijers and Dr. Thomas Ziesemer for their
valuable suggestions on econometrics. Their comments and suggestions have been
very helpful in improving my thesis. I thank Dr. Micheline Goedhuys for her great
effort in discussing frontier models with me. Thanks also to Alessandro Nuvolari who
has always been available to help me with all kinds of questions.
It was a challenging job to deal with Chinese regional data. Therefore, I would
like to thank Dr. Jianbing Liu, who not only offered me tremendous help in finding
missing data, but also in answering all puzzling China-related questions. Furthermore,
I have been very happy that Selin Ozyurt (University of Paris-Dauphine, France) also
struggled along with me to analyze the FDI spillovers in Chinese regions. Thanks to
her for having valuable discussions and sharing experience with me.
I would also like to express my gratitude to my former colleagues in ECIS
(Eindhoven Centre for Innovation Studies) and the capacity group of Technology and
Policy at the Eindhoven University of Technology. In particular, Marianne Jonker
offered me great help on many aspects. In Eddy’s words, she is really like a “super-
mother” who takes good care of everything for us. Thanks for the support from Chris
Snijders, Letty Calame, Bert Sadowski, Saskia Repelaer, Henny Romijn. During my
pregnancy, Letty was so thoughtful that she arranged a folded bed in my office to
offer me a good rest during lunch breaks. I am grateful for all the help I received from
our former “T&B” group.
ii
I was more than happy to share my office with Sjoerd van der Wal and
Andriew Lim, who helped me to get involved in our group easily. Michiel van Dijk,
Jojo Jacob, Frank Vercoulen, Ted Clarkson and Yongwei Liu helped making it easy
and enjoyable for me to start a new life in Eindhoven. Besides them, our nice lunch
and Thursday-drink group also included Rudi Bekkers, Arianna Martinelli, Mei Ho,
Christoph Meister and Effie Kesidou. In addition, I would like to thank Gergely
Mincsovics and Yu Da, not only for sharing their knowledge on data software, but
also for their good friendship.
Furthermore, I wish to thank the reading committee: Prof. Bart van Ark, Prof.
Geert Duysters, Prof. Chris Snijders, Prof. Bart Verspagen, and Prof. Yanrui Wu, for
their valuable time and their comments.
Dr. Ad van de Gevel (Tilburg University), deserves a special mention. If he
had not introduced ECIS to me in 2004, I would have missed the chance to have my
defence in the Netherlands.
I am honoured to have the cover of this book designed by Joep van Son. Joep
is not only talented when he plays music with my husband, but also creative in
designing.
Besides academic aspects, I am indebted to the love and support from my
family. The unconditional love from my father and my late mother has been a
backbone for me in pursuing my interests. I am also very lucky to have the great
support from my parents-in-law, Toine and Nelly Philipsen. Many thanks to them for
being supportive and considerate all the time. The birth of my son Timo has greatly
enriched my life. This little one has witnessed the important last stages of writing my
thesis both before and after he was born. Finally, I owe millions of thanks to my
husband, Niels. Without his true love and endless support, I could not have achieved
anything I have. I admire and appreciate his talent in many ways. He is the one who
provided insightful suggestions when I encounter any academic questions, he is the
one who uses his humour to cheer up my life when I am down, he is the one who
guides me through all western music, and he is the one who realizes all my dreams.
Being with him is the most wonderful thing!
Lili Wang
Maastricht
January, 2009
iii
Contents
Chapter 1 Introduction ……………………………………………….…..…….………..1
Chapter 2 Literature Review …………………………………………….….….………..9
Chapter 3 Economic Reform, Institutional Change and Economic Development
in China……………………………..………………………..………………27
Chapter 4 Data and Statistical Problems…………………………….………………….51
Chapter 5 Regional Capital Inputs in Chinese Industry and Manufacturing ………….59
Chapter 6 Productivity Growth and Structural Change in Chinese Manufacturing,
1980-2002 ……………............………………………………….…………109
Chapter 7 Regional Performance and Productivity Efficiency …………….….…...…133
Chapter 8 Contribution of Technological Spillovers to Industrial Growth in
Chinese Regions …………….…...………………………...…….…………153
Chapter 9 Conclusions…………………………………………………………………177
Appendix ………………………………………………………………………………183
References ………………………………………………………….………………….195
Summary ………………………………………………………………………………211
Curriculum Vitae ……………………………………...…..………………………….215
iv
List of Tables
Table 3.1 Ownership Categories of Industrial Enterprises in 1985, 1995 and 2004.........37
Table 3.2 Percentages of Ownership Categories of Industrial Enterprises in 1985,
1995 and 2004…................................................................................................38
Table 3.3: Value Added of Industry in TVEs, 1995-2005.................................................41
Table 3.4: Gross Value Added and Employment of Industry in TVEs, 1987-1999..........42
Table 3.5: Education Expenditure in China, 1980-2004....................................................44
Table 3.6: FDI as percentage of TIFA and GDP, and FDI per Capita, 1981-2005...........47
Table 5.1: Capital Concepts...............................................................................................62
Table 5.2: Breakdown of Investment in Real Estate Development, Total Economy,
1997-2003........................................................................................................75
Table 5.3: Breakdown of Investment Types by Content Categories.................................77
Table 5.4: Content of Investment by Type of Investment ................................................78
Table 5.5: Comparison of Newly Increased Fixed Assets and Accumulation of
Fixed Assets......................................................................................................83
Table 5.6: Proportions of Investment Categories in TIFA (%), Total Economy,
1981-2003 .......................................................................................................96
Table 5.7: Productive NIFA and Estimated Productive Capital Stock (100
million yuan)...................................................................................................103
Table 5.8: Estimated Productive Capital Stock in Manufacturing by Region,
1985-2003 .......................................................................................................107
Table 5.9: Estimated Productive Capital Stock in Industry by Region...........................108
Table 6.1: Labour Shares in Manufacturing, 1980-2002 (%)..........................................118
Table 6.2: Decomposition of Manufacturing Productivity - Contribution of
Sectoral Shifts, 1980-2002.............................................................................119
Table 6.3: Decomposition of Manufacturing Productivity - Contribution of Shifts
between Technology Classes..........................................................................120
v
Table 6.4: Decomposition of Industrial Productivity - Contribution of Shifts in
Ownership.......................................................................................................124
Table 6.5: Industrial Productivity: Shift-Share by Ownership and Region,
1992-2002 - Contribution of Institutional Shifts by Region............................125
Table 6.6 Regional Shares in Industrial Employment and Value Added in
1978 and 2002..................................................................................................127
Table 6.7: Decomposition of Industrial Productivity - Contribution of Regional
Shifts, 1985-2002.............................................................................................128
Table 6.8: Industrial Productivity: Shift-Share by Region and Ownership, 1992-2002
- Contribution of Regional Shifts by Institutional Categories (%)..................129
Table 7.1: Standard Deviation, Mean and Coefficient of Variation of per Capita
GDP in Chinese Regions, 1978-2005..............................................................136
Table 7.2: GDP per Capita in Chinese Regions (at 1978 Constant Prices) ....................137
Table 7.3: Standard Deviation, Mean and Coefficient of Variation of Industrial
Labour Productivity in Chinese Regions........................................................141
Table 7.4: Industrial Labour Productivity in Chinese Regions (at 1978 Constant
Prices) ............................................................................................................141
Table 7.5: Labour Productivity in Industry, by Geographical Location .........................142
Table 7.6: Beta Convergence...........................................................................................143
Table 7.7: Regional Efficiency Scores, 1978-2002.........................................................148
Table 8.1: Summary of Variables of R&D and FDI Spillovers in China........................164
Table 8.2: Estimates on R&D and FDI Spillovers in Chinese Regions, with Cutoff
Distance at 1520 km (All Regions), 1991-2002.............................................169
Table 8.3: Estimates on R&D and FDI Spillovers in Chinese Regions, with Cutoff
Distance at 760 km (All Regions), 1991-2002...............................................171
Table 8.4: Estimates on R&D and FDI Spillovers in Chinese Regions, with Cutoff
Distance at 1520 km, 1991-2002 (with Regional Dummies).........................172
vi
Table 8.5: Estimates on R&D and FDI Spillovers in Chinese Regions, with Cutoff
Distance at 760 km, 1991-2002 (with Regional Dummies)...........................174
Table D-1: Employment in Three Industries and Percentages........................................186
Table D-2: GDP in Three Industries and Percentages.....................................................187
Table D-3: Comparisons of Discrepancy between National and Regional
Resources in Industry, 1989...........................................................................188
Table D-4: Comparisons of Discrepancy between National and Regional Resources
in Industry, 2003.............................................................................................189
Table D-5: Industrial Employment in Chinese Regional, 1978-2002 …………………190
Table D-6: Industrial Value Added in Chinese Regional, 1978-2005 ………...……….191
Table D-7: NIFA in Basic Construction and Technical Renovation (100 mill yuan).....192
Table D-8: Newly Invested Fixed Assets (NIFA) in Total Economy (100 mill yuan)....193
Table D-9 Productive Ratio in Newly Invested Fixed Assets (NIFA) in Total
Economy.........................................................................................................193
vii
List of Figures
Figure 2.1 Conditional and Absolute Convergence ..........................................................16
Figure 2.2 The Potential Contribution of Spillovers to Catch up and Growth..................21
Figure 3.1: GDP per Capita in China, 1978-2005............................................................30
Figure 3.2: Value Added in Chinese Industry, 1978-2005...............................................34
Figure 3.3: Ratio of R&D Expenditure to GDP in China, 1990-2004 .............................43
Figure 3.4: FDI in China, 1983-2005................................................................................46
Figure 3.5: Total Investment in Fixed Assets from FDI...................................................46
Figure3. 6: China's Exports, 1952-2005............................................................................48
Figure 3.7: Ratio of Exports to GDP in China, 1952-2005...............................................49
Figure 5.1: Total Investment in Fixed Assets by Type of Investment, Total Economy,
1980- 2003......................................................................................................74
Figure 5.2: Total Investment in Fixed Assets by Content Category, Total Economy,
1981-2004.......................................................................................................76
Figure 5.3: Total Investment in Basic Construction by Content Category,
1950-2003........................................................................................................79
Figure 5.4 Total Investment in Technical Renovation by Content Category,
1980-2003.........................................................................................................80
Figure 5.5: Other investment by Content Category, 1985-2003........................................80
Figure 5.6: Structure of Newly Increased Fixed Assets, 1981-2003.................................81
Figure 5.7: Productive Newly Increased Fixed Assets in Total Economy, Industry and
Manufacturing, 1953-2003..............................................................................94
Figure 5.8: The Estimate Process on Capital Input in Non-residential Fixed Structures,
Machinery and Equipment in Industry: by Region 1953-2003 at Constant
1952 Prices.....................................................................................................102
viii
Figure 5.9: Estimates of the Capital Stock in Total Economy, Industry and
Manufacturing, 1952-2003 (at 1952 Prices)..................................................104
Figure 7.1: Coefficient of Variation of GDP per Capita in Chinese Regions,
1978-2005.......................................................................................................138
Figure 7.2a: Kernel Density of GDP per Capita in Chinese Regions, 1980- 2005........ 139
Figure 7.2b: Kernel Density of GDP per Capita in Chinese Regions, 1978-1992..........139
Figure 7.2c: Kernel Density of GDP per Capita in Chinese Regions, 1992-2005...........140
Figure 7.3: Coefficient Variation of Industrial Labor Productivity in Chinese
Regions, 1978-2002........................................................................................142
Figure 7.4: Kernel Distribution of Industrial Labour Productivity in Chinese
Regions, 1978-2005 (10000 yuan/person)....................................................144
Figure 7.5a: Regional Labour Productivity as Percentage of Shanghai,
1978-1990......................................................................................................144
Figure 7.5b: Regional Labour Productivity as Percentage of Shanghai,
1990-2002......................................................................................................144
Figure 7.5c: Regional Labour Productivity as Percentage of Shanghai, 1978-
2002................................................................................................................145
Figure 7.6: Technical Efficiency of Industry in Chinese Regions, 1978-2002
(by DEA Model)............................................................................................147
Figure 7.7: Coefficient of Variation of Technical Efficiency in Industry in 31
Chinese Regions, 1978- 2002........................................................................148
Figure 7.8: Growth Rate of TFP, Technical Efficiency and Technological Progress ....148
CHAPTER 1
Introduction
Since 1978, the year economic reform started, the Chinese economy has been growing
rapidly. This growth has been characterised by major institutional changes and great
disparities between regions. In the past, most statistically grounded studies on Chinese
industrial performance have focused on aggregate performance, i.e. on the national
growth in specific industries. There is an increasing need for more disaggregated
approaches and careful measurement and analysis of economic performance at the
regional levels (Amiti and Wen, 2000; Démurger et al. 2001, Wu, Y. 2000a, b). This
thesis hopes to fill part of this gap.
1.1 Background of This Research
Chinese industry1 has maintained a high rate of growth since the beginning of the
reform period in 1978. In particular after the mid-1990s, productivity growth in
Chinese manufacturing has been accelerating dramatically. Productivity growth
accelerated to 14.8 per cent per year during the period 1992-2002. It was no less than
19.6 per cent per year between 1996 and 2003. The period 1980-92 can be
characterised as a period of growth without catch up. Productivity growth was
respectable, but the gap relative to the world productivity leader (the United States)
remained about the same. In 1992, productivity relative to the U.S. stood at 5.5
1 Chinese statistics normally do not distinguish manufacturing from industry, which also includes
mining and utilities. Most tables in this thesis therefore focus on industry. However, where possible we
also present data for manufacturing.
Chapter 1
2
percent of the US level. By 2002, it had reached 13.7 per cent of the US level
(Szirmai et al., 2005)2. There has been a debate on the growth rate of China’s
economy. China’s National Bureau of Statistics (NBS) estimates that the average GDP
growth is 9.6 per cent per year under the revised GDP calculation system (and 9.4 per
cent according to the old GDP system). However, Wu (2000) presents an annual GDP
growth rate of 8.46 per cent during 1978-1997. Maddison (1998, 2006) shows an
average growth rate of 4.4 per cent between 1952 and 1978, 7.5 per cent between
1978 and 1995, and 7.9 per cent during 1990-2003.
Empirical research into the productivity growth of the Chinese economy and its
industries has been hampered by the lack of consistency in the published data.
However, the quality and accessibility of Chinese statistics are improving rapidly. The
statistical system has made considerable progress in shifting to the System of National
Accounts (SNA) and more and more statistics are becoming available for research and
scrutiny (see Hsueh and Li, 1999; NBS, 1997; NBS/Hitotsubashi, 1997; OECD 2000).
However, the numerous changes in concepts, approaches and coverage in combination
with rapid changes in economic structures and patterns of ownership create major
problems for the consistency of time series. Without time series which are consistent
in concepts and coverage, any attempt to analyse the Chinese industrial growth
experience becomes meaningless. Szirmai, Bai and Ren (2005) and Szirmai and Ren
(2007) offer a major assessment of concepts, coverage and consistency in time series
for Chinese manufacturing. Statistical problems identified by those authors include:
enormous discrepancies between employment figures in the industrial census and
other sources (in 1995, this discrepancy for the “social labour force” in total industry
was no less than 37.4 million people), limited coverage of the time series for output,
different coverage of employment series and output series, changes in concepts from
net material product to gross value added, lack of detail on township and village
enterprises, and organisation of data by ownership rather than sector. Using detailed
2 See Chapter 6 for a more detailed discussion.
Introduction
3
information from the 1995 census and the 1995 and 1997 IO table (Census 1995, IO
Table 1995, and IO Table 1997), they made a series of adjustments to the data which
are consistent in concepts and coverage. These time series are broken down by sector
of manufacturing, but do not cover the total manufacturing sector. In addition,
therefore, the authors present aggregate time series with more complete coverage. The
study showed that while the statistical problems are huge, there is substantial scope
for improvement. As to capital input, given that published data on capital investment
in Chinese statistics is not yet consistent with the SNA framework, considerable
progress has been made in estimating capital stocks at national level (e.g. Holz, 2006;
Wu and Xu, 2002; Chow, 1993; Huang et al. 2002; and Cao et al. 2007).
Analyzing regional economic growth without having a sound database is misleading.
Therefore our research in this thesis starts with the construction of regional long-term
time series which have consistent coverage. Hitherto, there are no officially published
data on capital inputs at the level of Chinese regions. We have made a particular effort
to estimate regional and national capital inputs in industry according to SNA concepts,
which make the estimates comparable at the international level.
Along with the reforms and rapid growth, the Chinese economy has experienced
dramatic changes in its institutional structure. The share of state-owned enterprises
which used to play a dominant role in the whole economy has decreased greatly, and
various new ownership types have emerged, such as private enterprises, shareholding
enterprises and (because of the opening up to foreign investment) foreign-funded
enterprises. The so-called collective sector includes a wide range of ownership
arrangements, ranging from private ownership in all but name, collective ownership,
semi-private ownership and public ownership by towns, villages and municipalities.
In particular, the township and village enterprises (TVEs) have been growing rapidly
from the late 1980s till the mid 1990s. Since late 1990s, the TVEs changed their
nature again and many turned into private companies. Finally, there are millions of
tiny individually-owned enterprises, about which rather little is known and which
Chapter 1
4
seem to bear more resemblance to the informal sector in developing countries, than to
modern manufacturing. The changes in ownership structures of Chinese industries and
the productivity differentials between different types of enterprises have attracted
much attention from scholars.
There are large technology and performance gaps between the advanced coastal
regions and the backward interior regions. The open-door policy has brought high
inflows of foreign direct investment (FDI) to and exports from coastal regions, which
according to some observers have been the main sources of growth for the coastal
regions. How much different regions in China benefit from technological spillovers
from the more advanced regions is still remaining inconclusive.
1.2 Aim of This Thesis
This research analyses the growth experience in Chinese industry and manufacturing,
with a special emphasis on the decomposition of growth, structural change, regional
divergence and convergence, and technology spillovers. The decomposition analysis
focuses on three dimensions: sectoral, regional and institutional. The thesis examines
regional productivity differentials and in regional productivity convergence or
divergence. It includes an analysis of the regional, institutional and technological
sources of growth.
First, due to the scarcity of regional data and inconsistencies in the published data, it
is important to construct a regional database with a long-term time series that is
consistent in coverage. This thesis provides a number of adjustments to the data series
to achieve a long-term time series of value added and employment in Chinese regions.
Meanwhile, because of the lack of capital input data series from any official
(published) sources, we construct a new regional capital stock database.
Introduction
5
Secondly, the Chinese economy has experienced dramatic structural changes during
the period of its rapid growth. How much did these structural changes contribute to
the aggregate growth? This question has often been asked by economists and
politicians. In particular, what is the contribution of changes in ownership structures
to productivity performance? The value added of Chinese state-owned enterprises
increased from 130 billion yuan in 1980 to 576 billion yuan in 2002 (at 1980 constant
prices), but its share in the national economy declined from 81 per cent to 48 per cent.
The gross output of state-owned enterprises accounted for 64.9 per cent of the
national total and dropped to 10.6 per cent in 2004, while the share of foreign funded
enterprises increased to more than 30 per cent in 2004. Using shift-and-share
techniques, this thesis examines three types of structural change: changes in the
sectoral structure of production, changes in the regional structural of production and
changes in the ownership structure.
Thirdly, the inequality between Chinese regions has been a topic of hot debate in
recent years. It is well recognized that a fast increasing GDP is often connected with
increasing disparities between regions. Many empirical studies have proved the
Kuznets U-curve relationship between per capita income and inequality. Disparities
are often expected to rise in the beginning of economic development, and decline
afterwards. If the inequality level in a nation remains high in the long term, it might
cause economic and social instability. We examine whether this is indeed the case for
China. Are the Chinese regions converging or diverging in terms of per capita income,
productivity and technical efficiency?
Finally, this thesis evaluates the contribution of technological spillovers to the process
of catching-up. With the open-door policy in China, a great deal of foreign direct
investment (FDI) flows to Chinese coastal regions. Are the technological spillovers
from FDI the engine of the rapid growth of coastal regions? Are they the main
resource for China's catching-up? What is the contribution of spillovers from coastal
regions to interior regions? We will attempt to answer those questions in our
Chapter 1
6
technological spillover model, combining both international and interregional
spillovers.
1.3 Thesis Structure
The structure of this thesis is as follows.
Chapter 2 provides a general review of the literature on regional disparity, structural
change and technological spillovers. In later chapters we will apply the insights from
this literature to the analysis of China's regional industrial performance.
In Chapter 3, we provide a summary of the aggregate growth in China since 1978.
This summary describes the main stages of the reform process and the corresponding
institutional changes. In addition, Chapter 3 also presents a survey of the
developments of technology indicators and education levels in China. Along with the
openness policy, the changes in foreign investment and trade are also discussed.
Chapter 4 tackles data issues and statistical problems at both the national level and
regional levels. Adjustments for value added and labour are made to create consistent
time series under comparable coverage.
In Chapter 5, we provide new estimates of capital inputs in the Chinese economy.
Estimates are made for the total economy (1953-2003), for the industrial sector
(1978-2003) and for the manufacturing sector (1985-2003). The estimates for industry
and manufacturing are broken down into thirty regions. This chapter makes a
systematic attempt to apply SNA concepts to the estimation of Chinese capital inputs,
according to the Perpetual Inventory Method. It makes a clear distinction between
capital services and wealth capital stocks. After a general discussion of theoretical
Introduction
7
issues in capital measurement, we provide a detailed analysis of the relevant Chinese
statistical concepts and data.
Chapter 6 focuses on the contribution of structural change to aggregate manufacturing
performance in China. Since the start of the reform period the booming Chinese
economy has experienced rapid structural change. Using shift-and-share techniques,
this chapter examines three types of structural change: changes in the sectoral
structure of production, changes in the regional structural of production and changes
in the ownership structure.
Chapter 7 explores the extent to which there is regional productivity divergence or
convergence in Chinese manufacturing. Traditional regression methods are based on
the relationships between productivity growth rates and initial productivity levels.
Instead of these methods, we use the stochastic kernel density approach, which
provides a better view of distribution dynamics. Besides the commonly used variables
like GDP per capita and labour productivity, we will use Data Envelopment Analysis
to measure the productive efficiency of manufacturing in Chinese regions relative to
best regional practice. The evolution of regional productivity performance can thus be
compared among thirty Chinese regions.
In Chapter 8, the aim is to analyze the contribution of technological spillovers in the
process of industrial growth and catching-up in Chinese regions. Concerning the
sources of technological spillovers in Chinese regions, we distinguish between the
regional level and the international level. The former refers to R&D inputs in other
regions, the latter concerns international R&D investment which is embodied in FDI.
Our analysis covers the impact of spillovers from R&D in other regions, from FDI in
the own region, as well as FDI in other regions.
Chapter 9 concludes the whole thesis, with a brief discussion of the main results of
our analysis.
CHAPTER 2
Literature Review
The fast growth of Chinese industry has been characterised by large disparities
between regions, changes in the structure of production and dramatic changes in
China's institutional structure. The sharp decrease in the number of state-owned
enterprises and the booming of private and foreign enterprises have attracted lots of
attention from scholars. Along with the opening up of China, the massive inflow of
foreign investment has been regarded as a great benefit for coastal regions. In the
literature, technological spillovers are seen as playing a crucial role in the catching-up
process at both the national and the regional levels. It is necessary to find out whether
and to what degree the growth of Chinese regions has benefited from technological
spillovers. That is, to fully understand the aggregate growth of Chinese industry, it is
of importance to explore its regional, institutional and technological sources.
This chapter provides a general review of studies of structural change, regional
disparities, convergence or divergence, and technological spillovers. A more technical
discussion of the literature can be found in the relevant chapters, 5, 6, 7 and 8. Our
focus is to apply these theories to analyze the growth, regional inequality and catch up
of Chinese industry.
2.1 Structural Changes
It has often been argued that structural changes are among the key sources of
economic growth in developing economies. One of the stylised facts in economic
development is the decreasing share of agriculture in employment and value added
Chapter 2
10
and the increasing shares of manufacturing and services. At later stages of
development, there are major structural changes within the manufacturing sector and
the service sector.
There are numerous studies in the literature illustrating the positive relationships
between structural change and economic growth, in particular with regard to Asian
economies (e.g. Van Ark and Timmer, 2003; Nelson and Pack, 1999; Timmer and
Szirmai, 2000). Kuznets (1979) explains Taiwan's economic growth from the
structures of production and use, household income and distribution, and population
patterns. The Fei and Ranis (1964) two sector model focuses on the positive effects of
reallocation from agriculture to industry.
However, the fact that structural change and growth are connected does not
necessarily mean that a so-called structural change bonus occurs all the time. Not all
structural changes are beneficial to economic growth and productivity growth
(Peneder 2003). Fagerberg (2000) argues that rapid growth does not need to be
accompanied by major structural changes. Even worse, some structural changes may
impede economic growth, which is referred to as a "structural burden". Baumol
(1967), using a two-sector model, argues that the opportunities for productivity
increase in services are limited, so that the expansion of the service sector will result
in an aggregate productivity slowdown. In a recent paper Van Ark and Timmer (2003)
find some evidence of the Baumol effect in East Asia, but its impact is partly offset by
the fact that productivity levels in some of the service sectors are even higher than in
manufacturing. These authors also find positive effects of manufacturing on total
economic performance. Manufacturing is still an engine of growth in developing
countries.
2.1.1 Structural Change and Productivity Growth
Two strands of literature can be distinguished in the extensive literature on the
relationship between structural change and productivity growth.
From the demand side, structural change is understood as the result of a variety of
exogenous factors. The structure of production is influenced by changes in domestic
Literature Review
11
demand (e.g. the Engel effect1) and changes in export demand (expansion of exports,
changes in the composition of demand for exports). Furthermore, technological
changes result in changes in broader life styles and consumption styles which in turn
affect the structure of production. A well-known example of the impact of demand on
economic structure is the shift of employment from agriculture to industry in response
to the low income elasticity of demand for agricultural products. (Syrquin, 1988). At a
later stage in the process of economic development, the role of the service sector
becomes more important, as the share of services in consumption increases.
From the supply side, the neo-classical growth perspective interprets structural change
as a way of re-allocating productive resources – labour and capital – to more efficient
uses. Differences in marginal productivity and differentials in productivity growth
induce movements of input factors from sectors with lower productivity to those with
higher productivity (Harberger, 1998; Lucas, 1993). The reallocation of resources is
one of the important sources of growth identified by Denison in Why Growth Rates
Differ (1967). Much research based on supply-side perspectives has focused on
explaining the relationship between factor reallocation and productivity growth,
disregarding the changes of external environmental parameters (Chenery, 1960).
Differential rates of technological change within sectors can also affect the sectoral
structure of production.
Traditionally, the term structural change primarily refers to changes in the sector
structure of the economy. In this thesis, we interpret structural change in a broader
sense, which also includes changes in the institutional (i.e. ownership) structure of
production and changes in the regional structure of production.
Sectoral change is driven by both demand- and supply-side factors. In the case of
regional change, supply-side factors (the relative efficiency of regions) are the more
important ones. But demand factors can also affect the interregional structure of
production, through changes in regional demand. In the case of institutional change,
supply-side factors (the relative efficiency of firms in different institutional categories)
also predominate. Of course, institutional change is also heavily influenced by
1 The Engel effects refers to changes in commodity demands by people while their incomes are rising,
e.g. when a family's income increases, the proportion of money they spend on food decreases.
Chapter 2
12
changes in political ideologies, power relationships and government policies. Thus,
the share of foreign owned firms in the Chinese economy would not have increased, if
the government had not opened up the economy to foreign investment.
In China, the rapid industrial growth has witnessed structural changes in many aspects,
e.g. sectoral changes, ownership changes and regional changes. Overall, structural
change has been an important source of the growth in Chinese industry. This will be
further discussed in Chapter 6.
2.1.2 Decomposition Methods
In measuring the contribution of structural change to productivity growth, it is of
importance to find an appropriate approach to distinguish the separate effects of
intersectoral shifts from those of intrasectoral productivity growth2. The shift-share
method is a powerful tool to decompose productivity growth into various resources.
There are two streams of shift-share applications. One stream concerns regional
output and employment changes. The shift-share model decomposes a regional
economic variable into three components: the national share, the proportional shift
(which measures the industrial composition), and the differential shift (which
measures the change in a particular industry). The main goal is to examine whether
(and to what extent) the difference in growth between each region and the national
average is due to the region performing uniformly better than the average of all
industries, or to the fact that the region is specialized in fast growing sectors (see also
Knudsen, 2000; Haynes and Dinc, 1997; Esteban, 2000). This shift-share model has
been widely used in studies on regional growth and employment changes (Harrison
and Kluver, 1989; Toft and Stough, 1986). Accompanied by its increasing number of
applications, a group of modifications emerged from the basic model. Esteban (1972,
2000) decomposes regional labour productivity differences into three components:
regional industry mix, productivity differentials and an allocative component3, which
2 See Chapter 6 for more details.
3 The regional industry mix component measures the differential productivity from one region's
specific sectoral composition; the productivity differential component presents the contribution of
sectoral productivity differences to the shift between regional and national average productivities; and
the allocative component measures the covariance between the two previous components, i.e. it is an
Literature Review
13
shows that regional differences in productivity per worker can be fully explained by
the existence of region-specific productivity differentials and uniform productivity
shifts across sectors4.
In the other stream, shift-share is adopted to measure structural changes. The shift-
share model is used to distinguish between the contribution of shifts between sectors
and the contribution of productivity growth within sectors. Namely, it separates intra-
and intersectoral effects on output and productivity. This method was first developed
by Fabricant (1942), and has been widely applied by many others, e.g. Syrquin (1984),
Paci and Pigliaru (1997), Timmer and Szirmai (2000), Van Ark and Timmer (2003),
Fagerberg (2000), Peneder (2003). Methodological details and some of the potential
shortcomings of this group of shift-share methods will be discussed in chapter 6. Here
we would like to mention the fact that shift and share methods are not suited to
capture the effects of intersectoral technology spillovers. Therefore they tend to
provide a lower bound for the impacts of structural change. In chapter 8, we will
substitute the shift-share analysis by a spillover analysis using regression models.
2.1.3 Discussion
As regards the aggregate growth of the Chinese economy, we will explore (see
Chapter 6) the dramatic changes in its institutional structure, sectoral structure and
regional structure of production and their implications for economic development.
Resource reallocation brings higher productivity and faster growth. Apart from the
shifts between sectors, the Chinese economy has witnessed dramatic changes in the
structure of ownership, as part of the wider transition from plan to market. The
decline of shares in State-owned enterprises, the booming of township and village
enterprises, and the rapid increases in foreign and private enterprises have been
interesting features of the Chinese growth experience. Moreover, the regional
composition of industrial employment in China also changed over time. Labour shifts
from inland to coastal regions have contributed to productivity growth as well.
indicator of the efficiency of each region in allocating its resources over the different industrial sectors
(see Esteban, 2000, p.356-357). 4 Such uniform increases in regional productivities can result from sectoral technologies, infrastructures
and human capital.
Chapter 2
14
2.2 Regional Disparities and Convergence or Divergence
2.2.1 Aggregate Growth and Regional Disparities
Fast growth is often accompanied by an increase in regional disparities. In a market-
oriented economy, labour, capital and other mobile factors of production are more
likely to flow to the areas where firms can gain higher returns. Such regional
reallocation is admittedly good for aggregate productivity growth. However, one
region’s prosperity is usually at the expense of other regions, which will result in an
increasing gap between rich and poor regions: rich regions will become richer and
poor regions will stay poor or become even poorer.
There is a rich literature on the relationship between (income or regional) inequality
and economic growth. Many studies using cross section data of economic
development seem to confirm the inverted U-shaped curve for the relationship
between per capita income and inequality, as put forward by Kuznets (1955). Namely,
disparities rise in the early stage of economic development and then decline with the
increase of per capita income. Williamson (1965) extended the Kuznets curve into a
regional or spatial context. Based on cross-sectional data for 24 countries, Williamson
proposed the hypothesis that when per capita income increases, relative regional
disparities first widen, then remain steady and subsequently decline. He concluded
that during the early stages of development an increase in regional inequality is
generated, while mature growth produces regional convergence or a reduction in
differentials. Barrios and Strobl (2006) have examined the evolution of regional
inequalities within European countries. Using a panel of European countries, they
conclude that there is an inverted U-shaped relationship between national income per
capita and the degree of regional inequality. Tamura (1996) and Lucas (2000) also
argue that the regional inequalities first rise and then decline in the course of
economic development. Forbes (2000) finds that in the short and medium term, an
increase in a country's level of income inequality has a significant positive
relationship with its economic growth. Banerjee and Duflo (2003) criticise the linear
parametric methods used in many empirical studies, pointing out that the changes in
Literature Review
15
inequality (in both increasing and decreasing directions) are associated with reduced
growth in the next period.
In the early stages of economic development, due to industrialization and urbanization
labour and investment flow to the more dynamic areas, which inevitably results in
more regional inequality. As indicated in Perroux's theory of economic growth poles
(1949), one region's growth takes place at the cost of other regions. According to
Perroux, growth does not happen everywhere and all at once. Growth poles consisting
of groups of geographically clustered leading industries can grow faster than other
regions. The leading areas can have both negative (e.g. pollution) and positive effects
(e.g. technological spillovers) on surrounding regions. Temporary increases in
disparities between regions could be acceptable to the people of a country, if there are
hopes for improvement of the poorer regions in the longer run. Hirschman (1973)
demonstrates that "society’s tolerance" for disparities exists in the expectation that
eventually the disparities will narrow again. “If this does not occur, there is bound to
be trouble and, perhaps, disaster” (Hirschman, 1973, p. 29). In other words, a long-
term high level of inequality might cause polarization and social instability, which
would have a negative effect on a country’s growth prospects. The extent of regional
inequality is closely related to the national economy. Lucas (2000) presents a growth
model in order to predict the income inequality trend, linking the inequality level and
economic growth. He also points out that the postwar period growth rates vary less
among advanced economies than among developing economies.
2.2.2 Convergence or Divergence
According to the older neo-classical growth theories, diminishing marginal returns to
capital offer opportunities for capital flows from rich regions (or countries) with high
capital intensities to poorer regions (or countries) with lower capital intensities. In
other words, there is an equilibrating mechanism which tends to reallocate capital.
This theoretical perspective predicts long-run convergence between countries and
regions. Factor mobility contributes to the convergence process by moving labour or
capital from regions where they are relatively abundant to regions where they are
relatively scarce. The assumption of free access to technological knowledge means
Chapter 2
16
that technological change will not result in divergence (Barro and Sala-i-Martin, 1992,
2004). The joint impact of decreasing returns (in the convex production function) and
exogenous technological change predicts convergence between regions. Some
empirical studies of catch up also find such convergence, showing that countries with
lower initial levels of income per capita grow faster than those with higher initial
levels (e.g. Mankiw et al., 1992; Barro and Sala-i-Martin, 1991). This could also
operate at the level of regions within a country.
Barro and Sala-i-Martin (2004) also stress the difference between conditional and
absolute convergence. The lagging regions (or countries) can have a higher growth
rate in their catching-up process. However, if the rich and poor have different steady
state growth rates, catch up will result in a so-called conditional convergence, in
which the poor country growth rates converge on a steady state growth rate, which
may be lower than that of the set of rich countries (see the following figure). Thus,
convergence within groups of countries could co-exist with divergence between rich
and poor countries.
Figure 2.1: Conditional and Absolute Convergence
In contrast, new growth theories and endogenous growth theories predict long-run
divergence of per capita incomes, rather than convergence. They point to increasing
returns to the primary inputs (capital and labour). Romer (1986) proposes a long-run
growth model with increasing marginal productivity of knowledge inputs. Technology
and innovation are the keys to augmenting the marginal returns to inputs of capital
and labour. As rich countries (or regions) have more advanced technologies than poor
ones, rich countries (or regions) tend to become richer and the poor tend to lag behind.
same steady
states
depends on the
similarity of tastes
and technologies
No
Yes absolute convergence: the poor and
the rich will converge on the same
steady state.
conditional convergence: the growth
rate of the poor country will slow
down when it approaches its steady
state.
Literature Review
17
Romer (1986) and Lucas (1988) show that the divergence in the long term is also due
to the increasing returns to scale.
In evolutionary strands of endogenous theories of growth catch up and convergence
are also seen as possible outcomes. They will occur if technologies spill over from
leading countries and regions to backward countries and regions. Whether this
happens, depends on the size of technology gaps, on the one hand. If such gaps are
small there is little potential for catch up. If they are too large, the obstacles to
diffusion of technology can become insurmountable. On the other hand, catch up
crucially depends on the learning capabilities and absorptive capacities of the laggard
regions. If the laggards have certain abilities to absorb technologies developed in the
most advanced economies or regions, they can grow more rapidly than the leaders.
They can profit from the advantages of backwardness. If, on the contrary, their
absorptive capacities are too low, and the technology gaps are too large, advanced
technology may not be appropriate for the laggards. They might be caught in a
poverty trap and fall behind. Thus convergence or divergence depends on the balance
between the rate of technological advance in the lead countries (or regions) and
technology diffusion to the less advanced countries (or regions) (see e.g. Verspagen,
1991; Fagerberg and Godinho, 2005; Szirmai, 2005, 2008).
Interestingly, convergence and divergence happen at the same time in some regions.
For instance, in the EU regions, some authors found that there is a paradox of
emerging convergence and divergence at the same time: convergence of economic
growth between the EU partners is accompanied by a divergence within its regions
(Martin, 1999; Fujita and Thisse, 1996). Especially in those poor countries, which are
growing faster and converging towards the others, there is a process of domestic
regional divergence. In an empirical study of the European Union, Fagerberg,
Verspagen and Caniëls (1997) argue that convergence between European regions is
not taking place, in spite of the expectations of increased homogeneity as a result of
economic integration. Caniëls and Verspagen (2001) suggest that the impact of spatial
proximity on the diffusion of technological knowledge may be responsible for this
paradoxical situation. With an evolutionary interpretation of OECD patterns,
Verspagen (2001) argues in favour of the possibility of divergence in the OECD area
in the near future. He points out that "[w]hile convergence to the sample mean is still
Chapter 2
18
going on [...], divergence is taking place for the indicator based on differences relative
to the leading countries".
As the biggest developing country, China's economic development has been
accompanied by large regional inequalities. Much literature tends to predict its
convergence or divergence trend. From the perspective of geography and policy,
Démurger et al. (2002) argue that two Chinese characteristics, i.e. the household
registration system (known as hukou)5 and the monopoly state bank system, inhibit
income convergence between Chinese regions. The former impedes labour movement
from poor regions to rich areas, and the latter results in most funds flowing to
traditional customers and very few to western provinces. Hsueh and Li (1999) argue
that per capita incomes among Chinese provinces are diverging since the open door
reforms. Tsui (1991) finds that interprovincial income gaps increased from 1952 to
1985. Kanbur and Zhang (2005) present that openness and fiscal decentralization
contributed to the increasing inland-coastal disparity in the 1980s and 90s. Applying
an augmented Solow growth model, Chen and Fleisher (1996) find evidence on
conditional convergence of per capita production, measured by national income
and/or gross domestic product, across China’s provinces between 1978 and 1993.
Bhalla, Yao and Zhang (2003) conclude that the inequality level in China is large by
international standards and this has been getting worse during 1952-1999. They
predict that the divergence trend will continue in China.
2.2.3 Discussion
The fast economic growth of a country, in particular a big country, is often
accompanied by large disparities among regions. Regions with better infrastructure
attract more investment and more skilled labour. This would happen in the early phase
of growth. However, if the inequality between regions stays large, this may pose a
threat to the further prospects of economic growth. The widening gap can lead to
polarization, social and political instability, and in an even worse case, it might cause
a nation to split in two.
5 Hukou is a household registration system which identifies a person as a resident of one area. Hukou is
an obstacle for a person from a rural area to go to urban regions because they are not entitled with the
health care, housing, and education. Even their children can not go to school in the unauthorized region.
Literature Review
19
In growth studies on developing countries, it is of importance to find out the growth
path and the convergence or divergence trends between regions. China's rapid
economic growth has also been characterized by the above-mentioned large regional
inequalities. However, whether there is convergence or divergence of growth and
productivity performance between regions has been a question of interest for
economists and politicians.
Many studies on regional disparities have focused mainly on income differentials and
GDP per capita. Our research will focus primarily on regional disparities in labour
productivity and technical efficiency (see Chapter 7).
2.3 Technological Spillovers
In the process of regional or international catching-up, technological spillover is one
of the most important sources of growth. As emphasized in the Gerschenkronian
tradition, one of the important potential sources of catch up in technologically
backward economies are international technology and knowledge spillovers from the
advanced economies. A similar reasoning can be applied to regions. Technologically
more backward regions can profit from spillovers from technologically more
advanced regions.
In neo-classical theory, which emphasizes the importance of factor accumulation (in
particular physical capital accumulation) for economic growth, technology has been
treated as an exogenously determined factor. All countries have access to the same
rate of technological advance. In recent decades, however, along with the introduction
of endogenous growth theory6 (Romer, 1990; Grossman and Helpman, 1991; Romer,
1986), the importance of innovation and technological change has been greatly
emphasized. Some empirical studies have shown that, despite the decreasing marginal
productivity of physical capital, accumulation of knowledge has an increasing
6 In explaining the knowledge contribution to economic growth, the main difference between the
endogenous growth model and earlier literature (on knowledge increasing returns and externalities), is
that in the endogenous model knowledge is treated as a capital good with an increasing marginal
product (see Romer, 1986).
Chapter 2
20
marginal productivity, and will result in an increase of the growth rate in the long run.
As shown by Romer (1990), "Technological change provides the incentive for
continued capital accumulation, and together, capital accumulation and technological
change account for much of the increase in output per hour worked".
Research and development activities (R&D), the creative work to increase the stock of
knowledge and apply it into use, is of importance to economic growth in terms of its
both direct and indirect effects. The direct effect is that a firm which carries out R&D
will profit from the innovation of its new products, processes, materials or
organization. This helps this firm to increase its market share and to make more
profits. The indirect effect refers to the possibility that other firms can imitate and
learn from the existing technology, no matter whether it is embodied in new products,
processes or organizational routines. Such technological spillovers (externalities) will
have increasing returns to scale at the macro level, even if technological investment in
one firm might have decreasing marginal returns (see Mohnen, 1996, for an excellent
survey on R&D externalities).
2.3.1 Spillover Types and Contributions
Griliches (1979, 1992) distinguished two different types of technological spillovers,
namely rent spillovers and knowledge spillovers. The first type originates from
product innovation under market competition. Rent spillovers occur when technology-
intensive inputs are sold to other industries at a price less than "their full quality price".
As stated by Griliches (1979), this is related to issues in the measurement of capital
equipment and materials and their prices and is not really a case of pure knowledge
spillover. "They are just consequences of conventional measurement problems."
(Griliches, 1979, pp. 103 and 104). This type of spillover is closely related to the
product demand and supply relationship7.
The second type, pure knowledge spillover, is also called technological spillover. This
is a type of externality available without any business transactions occuring.
7 In the domestic economy rent spillovers are just a measurement issue: how is value added allocated to
different sectors or regions. In the international economy, a developing country can profit from rent
spillovers. Then it is not just a measurement issue.
Literature Review
21
Technological spillovers can happen through many channels, like imitation of
innovations, labour mobility of skilled personnel, reverse engineering, infringing on
patents, access to international scientific literature, or communicating with R&D
personnel, etc. The spillovers analyzed in chapter 8 of this thesis are mainly the
second type of knowledge spillovers. We will examine both inter-regional spillovers
and international spillovers embodied in foreign direct investment.8
Spillovers exist at various levels, i.e. between firms, industries, regions and countries.
They can contribute to both growth and catch-up. Figure 2.2 shows the three levels at
which spillovers can be analysed. At the intermediate level, we can focus on spatial
levels (regions) and industrial levels (sectors).
Figure 2.2: The Potential Contribution of Spillovers to Catch up and Growth
micro-level intermediate-level macro-level
Note: At all the three levels, unit ① is always assumed to have higher level of technology than unit ②. Firm ① and ② belongs to industry (or region) ①; industry (or region) ① and ② belong to country ①. Source: author's own summary.
It is well recognized that technological spillovers play an important role in the process
of catching-up. New technologies leaking from advanced units not only offer free
knowledge for lagging units to use directly, but also provide good examples for them
to generate their own innovations. Nevertheless, knowledge spillover does not happen
automatically: how much lagging units can benefit from technological spillovers
8 FDI is not the only channel of international technology spillovers. However it has been identified as
one of the most important channels at international levels. In particular in developing countries, local
firms are expected to learn advanced technologies from the subsidiaries of multinational enterprises.
Country
spillo
vers
catching-
up
spillo
vers
catching-
up
Country
industry (or region)
industry (or region)
firm ①
firm ②
growth
growth
spillo
vers
catching-
up convergence
and world
development
Chapter 2
22
depends on many factors, such as the size of the technology gaps, the capabilities of
the learning units or geographical distance. These factors will be discussed in the
following subsection.
2.3.2 Factors Influencing Knowledge Spillovers
Technology gap: A technology gap is regarded as the precondition for lagging
regions or countries to receive positive knowledge spillovers from the technology
frontier, and thus to catch up rapidly. This is also explained by the advantages of
backwardness (Gerschenkron, 1962; Griffith, Redding and Van Reenen, 2004).
Backward regions or countries can borrow low-cost and low-risk technologies from
more advanced economies, and they can also imitate successful development models
from their leaders.
At plant level, using a plant-level panel for UK manufacturing, Haskel et al (2002)
conclude that FDI spillovers are more important to plants at lower levels of
technology than those at higher levels. Firms or regions which are already near to the
frontier cannot benefit much from FDI. Girma and Wakelin (2001) demonstrate that
highly skilled establishments do not benefit from FDI, because they are already very
close to the technology frontier.
Girma, Greenaway and Wakelin (2001) investigate FDI in the UK and find that firms
with small technology gaps (compared to the technology frontier), and/or with high
levels of skills can benefit from FDI presence. However, firms with large technology
gaps and low levels of skills are negatively influenced by FDI.
At the regional level, technology gaps are beneficial for the lagging regions. They
allow them to leap to higher levels of economic and technological development in a
short time. However, such gaps should not be too big. Using an equilibrium model,
Glass and Saggi (1998) explain that a big technology gap between host country and
the country of origin limits the transfer of advanced technology via FDI. Girma (2005)
divides the UK into 14 regions and presents a distance-weighted measure of foreign
presence outside the region but within the same sector.
Literature Review
23
At the national level, Fagerberg and Verspagen (2002) present the contribution of
innovation and diffusion in the process of convergence or divergence from an
evolutionary point of view. (see also Verspagen, 2001; Dosi, et al. 1998) As Szirmai
(2008, p. 19) states, "Evolutionary theory postulates a race between technological change
and domestic spillovers in the lead countries resulting in divergence and international
diffusion of technology to follower countries resulting in catch up."
Absorptive capacity is a crucial factor closely related to technology gap issues in
technological spillovers. In order to imitate or utilize the know-how spilling over from
the leader, the followers need certain abilities (background knowledge) to understand
and use it, otherwise the available technological knowledge would mean nothing to
the followers. Abramovitz (1986) states that "a country's potential for rapid growth is
strong not when it is backward without qualification, but rather when it is
technologically backward but socially advanced." (Abramovitz, 1986, p. 388). A
similar reasoning can be applied to technologically backward regions and firms.
As such, very large technology gaps already indicate a low absorptive capacity in the
lagging firms, regions or countries. However, for a given technology gap there may be
substantial differences in absorptive capacity.
Harhoff (2000) discusses differences between firms with regard to technological
spillovers. He shows that high-technology firms gain more from spillovers than firms
which are less technology-oriented. This result contradicts the aforementioned
conclusions by Haskel et al. (2002) and Girma and Wakelin (2001). The debate
focuses on the question which firms benefit more from technological spillovers, low-
level-capacity firms (with big gaps relative to the frontier) or high-level-capacity
firms (close to the frontier)?
Analysing this issue at the regional level, Girma (2005) applies a threshold regression
model and illustrates the existence and significance of threshold levels of absorptive
capacity. To have a positive spillover effect from more advanced regions, the
receiver's absorptive ability should be higher than the minimum threshold.
Chapter 2
24
At the national level, Verspagen (1991) also stressed the necessity of "learning
capability" for a country in order to benefit from technological spillovers. His model
presents the learning capability as a combination of an "intrinsic" learning capability
and the technological distance between the leader and the recipient. This is consistent
with our interpretation of Abramovitz given above.
In addition to imitating new innovations created by outside sources, absorptive
capacity also provides an ability to exploit outside knowledge (Cohen and Levinthal,
1989, 1990)9. In this sense, a successful spillover can be a training lesson for the
imitators, who can be inspired to take further steps and generate their own innovations.
In other words, absorptive capacity is not only an important factor for catching-up in
the short term, but also necessary for independent innovation and sustained growth in
the long run.
Congruence/Homogeneity: At the national level, congruence is another factor
influencing the utilization of external sources of internal technology. It is easier for
knowledge spillovers to take place if there is technological congruence between
technology leaders and followers. As proposed by Abramovitz (1986), countries
whose economic conditions and factor proportions are congruent with those of the
leaders are more likely to able to exploit the leader’s path of technological progress.
In contrast to the country level, firms at the regional level seem more likely to profit
from interregional technological spillovers, because there are generally more
similarities in culture, policies, or economic structures between regions than between
countries.
Geographic distance is also an important element in analyzing the effects of
technological spillovers. Spillovers could be more beneficial to more adjacent regions
if spatial interaction decreases with the geographical distance. Many empirical studies
(Funke and Niebuhr, 2005; Orlando, 2004; Caniëls and Verspagen, 2001) show that
the spillover effects do decrease with distance. Through a spatial analysis of
productivity effects of G-5 countries' R&D spending in other OECD countries, Keller
9 Cohen and Levinthal (1989) also distinguish absorptive ability from learning-by-doing. They argue
that learning-by-doing is an automatic process which makes a firm "more practiced" and "more
efficient" in doing certain things. Absorptive capability, however, through acquiring outside knowledge,
can lead a firm to do things in an innovative ("different") way. (Cohen and Levinthal, 1989, p. 570).
Literature Review
25
(2002) finds that "the distance at which the amount of spillovers is halved is about
1,200 kilometers" and he concludes that "technology is to a substantial degree local,
not global, as the benefits from spillovers are declining with distance." López-Bazo,
Vayá and Artís (2004) argue that the externalities across EU regions are locally
bounded in the sense that they occur only within distances below 600 kilometers.
2.3.3 Discussion
There have been mixed findings on the contribution of technological spillovers. Some
studies have confirmed the importance of spillover effects, while others failed to find
positive results, or even pointed out some negative effects.
FDI is often regarded as one of the important mechanisms of international knowledge
spillovers contributing to the growth of developing countries. In the Chinese case,
FDI has been increasing dramatically since the 1990s. It increased from 4.4 billion US
dollars in 1991 to 11 billion US dollars in 1992, and 27.5 billion US dollars in 1993: a
growth rate of no less than 150% each year. The FDI stock per capita changed from
0.6 US dollar per person in 1983 to 58.9 US dollar per person in 2005. Foreign
companies have been entitled to a number of privileges from the Chinese government,
partly policy makers considered FDI to be a positive resource of technological
spillovers to Chinese regions. However, whether and/or to what degree Chinese
regions actually benefit from the spillovers of FDI still needs to be examined.
Moreover, it is also important to combine the analysis of international spillovers
associated with FDI with an analysis of interregional technological spillovers in order
to find out their combined effects on regional catch-up or falling behind.
International technological spillovers can be regarded as a potential source of China's
catching-up and are seen as contributing to the fast growth of coastal regions. Inter-
regional knowledge spillovers may play an important role in diminishing regional
disparities and achieving a balanced development in China. These issues will be
analysed in chapter 8 of this thesis.
Chapter 2
26
2.4 Summary
This chapter summarizes the general literature on regional disparity, convergence/
divergence, structural change and the contribution of technological spillovers to
growth and catch-up. Disaggregating economic growth, i.e., exploring the
contribution of various sources to economic growth, is of importance for a better
understanding of the mechanisms underlying economic growth.
It is well known there have been large disparities in economic performance between
Chinese regions during the fast growth process. However, whether these disparities
are increasing or decreasing has been a long-debated question. Using our newly
constructed regional data, our analysis will provide a detailed analysis on the
convergence/divergence trends of Chinese regions.
Secondly, structural changes in the fast growth of China involve many levels. Besides
the commonly analyzed sectoral changes, institutional changes and regional shifts are
also important in the Chinese development process. Hence it is important to
disaggregate the growth rate along various key dimensions, such as sectoral,
institutional and regional shifts.
Finally, as a crucial aspect of catch-up theories at both the national and the regional
level, technological spillovers in China have to be analyzed. Foreign direct investment
(FDI) is a potential source of growth for coastal regions and a factor of China's
catching-up, while technological spillovers at the regional level might be important in
the process of regional catching-up and convergence. In order to fully understand
these two different types of technological spillovers, and in order to compare their
effects, we will combine them (international and local spillovers) in one empirical
model.
CHAPTER 3
Economic Reform, Institutional Change and Economic
Development in China
Since 1978, China has experienced a series of economic changes which resulted in an
acceleration of economic growth. As will be shown in section 3.1, this reform process
involved many changes at different levels of the economy. As the largest transition
economy in the world, China's reform has been carried out through step-by-step
experimentations. From conventional points of view, the reform process might be
puzzling to some researchers. However, the growth resulting from the reforms is
unmistakable. China's GDP has grown more than 9 per cent per year after 1978
according to the official data provided by China's National Bureau of Statistics (NBS).
Section 2 of this chapter surveys the main institutional changes since the reform. It
discusses the reorganization of State-owned enterprises (SOEs), the dynamics of
township and village enterprises (TVEs) and the emergence of private and joint stock
enterprises. Section 3 provides a survey of the development of technology and
education levels in China. Finally, in section 4, the changes in foreign investment and
trade, which resulted from China's increased openness, are discussed.
3.1 China's Reform and Industry Growth
3.1.1 China's Reform: from Plan to Market, from Rural to Urban
In 1978 the 11th National Congress of the CPC (Communist Party of China)(shiyijie
sanzhong quanhui)denoted the beginning of the era of China's reform: a transition
from a planned economy to a market economy, combining features of both systems.
The market system was introduced as an "assistant hand" to the planned economy.
Chapter 3
28
From the beginning of the reform until mid-1984, elements of central planning and
the market system were simply combined, each having separate functions. Although
government planning was still the essence of China's economy, the market was
beginning to be introduced. As an important part of the early reform, a household
responsibility system was introduced in agriculture. Releasing farmers from collective
teams provided them with incentives to work more efficiently. This system was
initially carried out on an experimental basis in a few places only, but it showed great
success with surprising increases in agricultural output and farmers' income.
Following the success of this experiment, 45 per cent of collective agricultural teams
were abolished by 1981. By 1983, 98 per cent of collective teams had adopted the
household responsibility system.
With the introduction of the market system in non-agricultural sectors, non state-
owned enterprises also began to develop. Sachs and Woo (1994) argue that starting
the reform in rural areas was indeed a correct decision, as it made the beginning of
China’s reform easier than that of the former Soviet Union. After all, the majority of
labour in China was located in rural areas1, whereas in the Soviet Union the share of
the urban and industrial sectors in total employment was far higher. China
experienced a “classic economic development”, by transferring workers from low-
productivity agriculture to higher-productivity industry. This is much easier than the
industrial adjustments that had to be made in Eastern Europe and the Soviet Union,
where employment in inefficient and subsidized industry was cut and new jobs in
efficient industry and services were created (see Sachs and Woo, 1994, p. 103).
In October 1984, the Third Plenary Session of the 12th National Congress of the CPC
(shierjie sanzhong quanhui) made further decisions on economic reform. The main
focus of the economic reform was shifted from rural to urban areas. Subsequently, the
13th National Congress of the CPC (shisan da) held in 1987 stressed the importance of
the unification of planning and market forces. It pointed out that using only the central
planning system no longer corresponded to the requirements of economic
development in China. Since then, enterprise autonomy had been enlarged and non-
state enterprises began to emerge rapidly. A dual-track price system was established
1 In 1978, 71% of the Chinese national total employment was in primary industry, and 15% in industry
(see Appendix Table D-1).
Economic Reform, Institutional Change and Economic Development
29
in the mid-1980s, which stimulated the increase of extra output produced by farmers
and the development of non state enterprises. Prices were liberalized at the margin,
while the planned prices and planned quotas were maintained. Farmers and companies
were allowed to sell their extra output beyond the planned quotas. This has been
regarded as a unique Chinese solution which improved economic efficiency on the
one hand while maintaining political stability on the other (Qian, 2003). Meanwhile,
the contract responsibility system was introduced in enterprises in the late 1980s,
which intended to clarify the authority and responsibilities in enterprises. This
contract system provides the firm manager with legal rights to operate the firm during
an agreed (contract) period, e.g. 3-5 years. Such contracts normally specified the
distribution of value-added between the state and the firm, performance targets,
production plans, etc (see also Xu, 2000). As pointed out by Huang and Woo (1998),
China's reform process between the 1980s and the 1990s was dominated by the
manager-responsibility system and the contract system, with the former dominating in
the 1980s and the latter in the 1990s. Huang and Woo also show that by 1994, about
90 per cent of the surveyed SOEs claimed to have the right to make production
decisions, and about 60 per cent had the right to make decisions on investment,
export/import and employment.
The southern tour2 of Deng Xiaoping in early 1992 was a milestone in China's
economic reform. Deng Xiaoping made a famous statement: "the planned economy is
not equal to socialism, planning exists in capitalism as well; the market economy is
not equal to capitalism, there is also a market in socialism", "plans or markets are not
the criteria to judge socialism or capitalism"3. Based on Deng Xiaoping’s theories, the
14th National Congress of the CPC (shisi da) in October 1992 set up a reform
framework called "constructing a socialist market economy" (jianshe shehui zhuyi
shichang jingji). The market liberalization involved more changes in the price system,
ownership responsibilities, etc. The market was proposed to play a crucial role in
promoting China's economic development. Enterprises were entitled with self-
management and self-responsibility with regard to production quantity, profit
distribution, the right to hire or fire employees, etc. More and more private enterprises
emerged, and private entrepreneurs become more motivated and more active.
2 Visiting Guangzhou, Shenzhen, Zhuhai and Shanghai.
3 The speech is available at http://www.oklink.net/lszl/dangdai/dxp01.html .
Chapter 3
30
Inspired by Deng's proposition that "some regions can get rich first", coastal areas of
China achieved remarkable growth. Generally speaking, China's reform process has
been characterized by its gradual and step-by-step experimentations, which was
portrayed in Deng Xiaoping's famous saying "crossing the river by groping for
stones" (Qian and Wu, 1999; Chow, 2004).
However, as many scholars argued, China's economic reforms succeeded without
complete liberalization and privatization (Qian, 2003, p.298). Also, in contrast to the
transitions in Eastern Europe and the former Soviet Union, China's reforms were
undertaken without fundamental changes in the political system (Qian and Wu, 1999;
Chow 2004).
Although China's reforms were not carried out according to conventional recipies4, its
rapid growth is unmistakable. According to official GDP and population data
collected from regional yearbooks, GDP per capita increased from 362 yuan/person in
1978 to 3679 yuan/person5 in 2005, with an average growth rate of 9.0 per cent each
year (see Figure 3.1).
Figure 3.1: GDP per Capita in China, 1978-2005
0
500
1000
1500
2000
2500
3000
3500
4000
1978
1980
1982
1984
1986
1988
1990
1992
1994
1996
1998
2000
2002
2004
yuan/person
Note: at 1978 constant prices.
Source: Various Chinese regional yearbooks. (Instead of using data from the national yearbooks, here
we use the sum of regional data, in order to keep consistent with the analysis in Chapter 7, Table 7.1
and Figure 7.1).
4 For instance, a well-designed reform blueprint, complete liberalization, well-defined property rights,
well-defined intellectual property rights, privatization and democratisation (see Qian, 2003, p.299). 5 Calculated at 1978 constant prices.
Economic Reform, Institutional Change and Economic Development
31
In the early years the Soviet accounting system, which is also known as the material
product system (MPS), was applied in the calculation of Chinese national accounts
statistics. However, partly due to its centrally-planned economic system, China's
service sectors were not covered well by the MPS accounting method. The system of
national accounts (SNA) has been used in China since 1992. During 1985-1992, MPS
and SNA co-existed in the Chinese accounting system, namely, national income was
calculated on the basis of MPS while GDP was based on SNA. One of the most
important differences between MPS and SNA is that MPS excludes non-material
services (including passenger transport, housing, health, education, entertainment,
banking, insurance, personal services, government, party administration and the
military) whereas SNA does not6. Another important difference is that between the
MPS concepts of gross and net material product and the SNA concept of value added.
In 2006, China's National Bureau of Statistics (NBS) issued China's first national
economic census (Economic Census 2004), which revised in particular the GDP level
of service sectors7. However, the revised GDP calculation system does not have much
impact on industry statistics, which are the main focus of this thesis. According to the
new estimation, the value added of tertiary industry in 2004 was 6501.8 billion yuan,
i.e. 2129.7 billion more than the calculation from old Chinese statistics8.
There has been disagreement on the real growth rate of China's economy. According
to the official estimation by NBS, the average GDP growth is 9.6 per cent per year
under the revised GDP calculation system (and 9.4 per cent according to the old GDP
system). However, there is much debate on the reliability of these official estimates.
The reasons for this are, first, related to the inadequacies of China’s statistical
reporting system. Most of the historical statistics were destroyed during the Great
Cultural Revolution. The first published Chinese Statistical Yearbook dates only from
1981. Secondly, the transition from MPS to SNA makes long-term comparisons
difficult. The non-productive sectors, which were left out of the MPS calculations, are
included in total GDP in the framework of SNA. And economic growth might be
6 See also Maddison and Wu (2008) and “The Historical National Accounts of the People’s Republic of
China 1952-1995”, http://www.ier.hit-u.ac.jp/COE/Japanese/online_data/china/china.htm . 7 The re-estimation of service is up to 93 per cent.
8 See the report from the Chinese government, http://english.gov.cn/2005-12/21/content_133044.htm ,
and the announcement by NBS with regard to the GDP revision, www.stats.gov.cn/tjdt/zygg/t20060109
_402300176.htm.
Chapter 3
32
overestimated in the MPS system, considering that MPS involves double counting9
(see Maddison and Wu, 2008). Thirdly, the often adopted “comparable prices”
(instead of “constant prices”) understated China’s inflation and thereby overstated real
growth. “Comparable prices” are prices reported by enterprises based on some certain
(inadequate) price manuals. As discussed in Wu (2000), Maddison (1998) and
Maddison and Wu (2008), in order to meet the high growth targets set by the
government, state-owned enterprises are more likely to exaggerate their real output,
and “there are substantial possibilities for exaggerating the volume of output when
new products are incorporated into the reporting system at so-called ‘comparable’
prices” (Maddison and Wu, 2008, p. 26). Wu (2000) presents an annual growth rate of
8.46 per cent during 1978-1997. Maddison (1998) makes new estimates of China’s
GDP growth rate by industry, agriculture, industry and non-productive services. And
he constructs a GDP time series at constant prices. His study shows an average GDP
growth rate of 4.4 per cent between 1952 and 1978, and 7.5 per cent between 1978
and 1995. More recent work by the same author shows a growth rate of 7.9 per cent
between 1990 and 2003 (see Maddison, 2006, p. 122, Table 2). Holz (2006) argues
that Maddison’s method under-estimated China’s growth rate “due to his assumption
of lower than official growth rates in ‘other services’ and in industry, and a larger
base year weight for ‘other services’ and agriculture.” Holz states that the official
growth rate is not over-estimated if an alternative price deflator (deflator for a net
output, instead of the one for gross output value) is applied to Maddison’s estimates.
In a recently published paper, Maddison and Wu (2008) summarize the critical views
that were presented in different contributions to the book Debates on the Rate of
Growth of the Chinese Economy10.
Despite the different estimations of the exact growth rate of GDP, all authors agree
that China witnessed very rapid growth after the onset of reforms.
3.1.2 The Growth in Chinese Industry
9 Inter-sector transfers of inputs are not deducted from the gross output.
10 For details, see Maddison and Wu (2008, p. 16).
Economic Reform, Institutional Change and Economic Development
33
Although Chinese agriculture accounts for more employment than industry, GDP in
industry accounts for a major part of total GDP. Industrial GDP was on average 41
per cent of national GDP during 1978-2004 (see Appendix Table D-2). Chinese
industry has witnessed a fast growth since the beginning of the economic reform
period, which can be divided into two different stages. The first stage is characterized
by a gradual growth, the second by a dramatic growth.
The first stage: releasing industrial enterprises from government control (1978-
1992).
Before the reform Chinese industry consisted mainly of state-owned enterprises, in
which the production and profit distribution were completely controlled by the central
government. Since the issue of the "regulations enlarging the managerial autonomy in
state-owned industrial enterprises" (guanyu kuoda guoying gongye qiye jingying
guanli zizhuquan de ruogan guiding) in 1979 by the State Department, industrial
enterprises began the journey towards more self-governance and self-responsibility.
As a result of the reform and the new regulations, industrial enterprises were endowed
with more rights to manage their production plans. Fulfilling obligations as specified
by the government plan is no longer their only goal. The enterprises have more
incentives to operate well in that they are allowed to sell their products and
redistribute profits. Therefore their performance has been closely connected with their
own income. In the first ten years after the reform China’s industry maintained a high
rate of growth. Using the regional data we collected for all state-owned and non-state-
owned above designated size11 industrial enterprises, we have the growth rate of value
added in Chinese industry during 1978-2005 in Figure 3.2. The value added of
industry increased from 137 billion yuan in 1978 to 364.7 billion yuan12 in 1992. The
average growth rate in this period was 7%.
The second stage: building modern enterprises (1992-2005).
After 1992 (the year of Deng Xiaoping’s important speech) industrial enterprises were
introduced to a further reform, towards a modern enterprise mechanism, in accordance
with the requirements of a market economy. A modern enterprise mechanism means
11 This covers all the enterprises with annual sales revenue over five million yuan after 1998.
12 Calculated at 1978 constant prices.
Chapter 3
34
to further clarify the property ownership, clearly define rights and responsibilities,
free enterprises from government manipulation, and set up a scientific management
system13. By the end of the 1990s Chinese industry had gradually set up the new
market economy system, with enterprises being the principal part of the economy.
Using the aggregated regional data published in CSY, CIESY and regional
yearbooks14, we present the growth rate of value added of Chinese industry in Figure
3.2 (series 1). It shows a dramatic increase during the second stage, i.e. after 1992.
The value added of industry grows at an average rate of 13% per year during 1992-
2005, which is almost twice as high as that before 1992. This growth trend is similar
to the estimate of Szirmai, Ren and Bai (2005). Applying their own deflation index
for each manufacturing sectors, Szirmai et al. have the industry growth rate at 7.4 per
cent during 1980-1992, and 10.3 per cent during 1992-2002. Their result also shows a
higher industry growth in later years (see Figure 3.2, series 2).
Figure 3.2: Value Added in Chinese Industry, 1978-2005
0
2000
4000
6000
8000
10000
12000
14000
16000
18000
20000
19781980198219841986198819901992199419961998200020022004
100 miil yuan
Series1
Series2
Note: at 1978 constant prices. Coverage is enterprises with independent accounting system at township
and above before 1998, and all state-owned and non-state-owned above designated size industrial
enterprises after 1998.
Source: Series 1 is the sum of regional data collected from China Statistical Yearbooks, various issues,
CIESY, various issues and China regional yearbooks, various issues. Series 2 is from Szirmai, Ren and
Bai (2005). Series 1 is unadjusted, but series 2 are national aggregate data adjusted by Szirmai, Ren and
Bai. These two series use the same deflator.
13 In Chinese, it is "chanquan fenming, zequan mingque, zhengqi fenkai, guanli kexue".
14 Due to the fact that data of value-added in Chinese industry from national yearbooks are incomplete,
we use complementary data from regional yearbooks; subsequently we aggregate all the regional data
to attain the national total. This is also a way to keep consistent with regional analysis for later chapters.
Economic Reform, Institutional Change and Economic Development
35
3.2 Institutional (Ownership) Change
Before the reform, China’s economy had been called a “solo-mechanism” (danyi
tizhi), which means that state-owned enterprises (SOEs) had dominant shares in the
national economy. The second largest category was the collective enterprises. The
difference between state-owned and collective enterprises is that the former is
financed mainly by state, while the latter is sponsored mainly by county, town or
village collective organizations. There were also various types of joint ventures
between state and collective enterprises.15
Non-state-owned or non-collective
enterprises had very modest shares in GDP. They included private enterprises, foreign
funded enterprises, and enterprises funded by Hongkong, Macao and Taiwan.
Through the implementation of the reform plans (from a planned to a market economy)
the solo-mechanism has been successfully transformed into a combination of various
ownership types. Lin et al (1996) argue that the dynamism of China's economy came
mainly from the swift entry of new, small, non-state enterprises.
The following tables present a survey on the institutional changes in three benchmark
years: 1985, 1995 and 2004. The gross output of SOEs accounted for 64.9% of
national total gross output in 1985, then dropped to 32.6% in 1995, and subsequently
to 10.6% in 2004. The percentage of collective enterprises experienced a small
increase in the beginning, from 32.1% of total gross output in 1985 to 35.5% in 1995;
however, it dropped greatly afterwards, till 4.4% in 2004.
In Chinese industry as a whole, the share of private enterprises has increased
dramatically, with an output at 2.8% of national total in 1995, to 22.4% in 2004 (see
Table 3.1 and Table 3.2).
In Table 3.1 and Table 3.2, the category of joint venture can also be called "only
collective- and state-joint venture" for it excludes foreign firms. Private enterprises
refer to economic units financed or controlled (by holding the majority of the shares)
15 Joint ventures include: 1) joint ownership of two (or more) state-owned enterprises; 2) joint
ownership of two (or more) collective enterprises; 3) joint ownership of state-owned and collective
enterprises.
Chapter 3
36
by natural persons who hire labour for profit-making activities. This is different from
the category of sole proprietorship. Sole proprietorship is a type of enterprise with
only one owner. Since the owner of a sole proprietorship does not have partners, there
is no need to pay corporate tax. However, the sole proprietor has to pay income tax
instead. Sole proprietorship in China is mostly related to manual labour or to low
levels of mechanisation, and is characterized by its decentralized investment and
small scale. This category is no longer included in the Chinese Economic Census
2004. For further explanations of ownership categories, see Appendix B at the end of
this thesis.
Economic Reform, Institutional Change and Economic Development
37
Table 3.1: Ownership Categories of Industrial Enterprises in 1985, 1995 and 2004
Enterprise
nr.
Gross
output
Empl.(year-
end)
Enterprise
nr.
Gross
output
Empl.(year-
end)
Enterprise
nr.
Gross
output
Empl.(year-
end)
unit
(100 mill
yuan)
(10 000
persons) unit
(100 mill
yuan)
(10 000
persons) unit
(100 mill
yuan)
(10 000
persons)
1985
Total Industrial enterprises IAS at township level and above Others
Total 5185300 9716.47 9682.05 358701 8434.72 6604.50 4826599 1281.75 3077.55
1. State owned 93700 6302.12 70342 6167.09 3858.19 23358 135.03
2. Collective 1740939 3117.19 286570 2149.24 2689.26 1454369 967.95
4. Sole
proprietorship 3347804 179.75 831.83 3347804 179.75 831.83
5. Joint venture 1126 80.79 48.05 1126 80.79 48.05
9. Foreign and HK,
MC, TW funded 516 36.65 7.81 516 36.65 7.81
10. others 1215 na. na. 147 0.95 1.19 1068 na. na.
1995
Total Industrial enterprises IAS at township level and above Others
Total 7341517 82296.63 14735.51 510381 54946.86 8575.58 6831136 27349.77 6159.93
1. State owned 118000 26840.51 4652.23 87905 25889.93 4464.65 30095 950.58 187.58
2. Collective 1465628 29253.29 5858.26 363840 15839.33 3088.93 1101788 13413.96 2769.33
3. Private 287483 2338.90 490.64 2708 146.5 16.52 284775 2192.4 474.12
4. Sole
proprietorship 5403643 9632.53 2576.40 5403643 9632.53 2576.4
5. Joint venture 5903 666.63 87.40 5493 652.76 85.39 410 13.87 2.01
6. Incorporated
Enterprise 5873 2750.34 254.81 5559 2727.01 253.04 314 23.33 1.77
9 Foreign and HK,
MC, TW funded, of
which 54045 10722.16 807.81 44293 9612.53 660.53 9752 1109.63 147.28
# Foreign funded 17692 4744.96 274.82
# HK, MC and TW
funded 26601 4867.57 385.71
10. others 942 92.27 7.96 583 78.8 6.52 359 13.47 1.44
2004
Total Industrial enterprises above designated size* Others
Total 1375263 222315.93 9303.94 276474 201722.19 6622.09 1098789 20593.74 2681.85
1. State owned 25339 23519.12 892.41 23417 23424.99 883.96 1922 94.13 8.45
2. Collective 141772 9819.04 688.08 18095 7865.41 334.89 123677 1953.63 353.19
3. Private 902647 49705.23 3225.14 119357 35141.25 1515.43 783290 14563.98 1709.71
4. Sole
proprietorship
5. Joint venture 6547 1033.43 43.97 1439 931.93 27.86 5108 101.5 16.11
6. Share-holding
Incorporated 50097 3396.73 205.61 8215 2641.38 116.77 41882 755.35 88.84
7. Share-holding
Ltd. companies 17427 23120.84 506.51 7171 22901.60 472.87 10256 219.24 33.64
8. Ltd. companies 102392 44042.82 1693.26 41234 42675.04 1508.27 61158 1367.78 184.99
9 Foreign and HK,
MC, TW funded, of
which 106165 67137.76 1991.41 57165 65995.21 1755.26 49000 1142.55 236.15
# Foreign funded 51255 42751.35 953.27 28766 42247.22 862.58 22489 504.13 90.69
# HK, MC and TW
funded 54910 24386.41 1038.14 28399 23747.99 892.68 26511 638.42 145.46
10. others 22877 540.94 57.55 381 145.37 6.79 22496 395.57 50.76
Note: 1) Gross output is calculated at current prices. 2) This table represents all state-owned and non-state-
owned industrial enterprises with annual sales revenue over 5 mill yuan. 3) Category 7 (Share-holding Ltd.
companies) and 8 (Ltd. companies) are not available in Chinese Industrial Census 1985 and 1995. Category 6
(incorporated enterprises) did not exist in 1985.
Source: Chinese Industrial Census 1985; China Statistical Yearbook 1993; and Szirmai, et al. 2005; Chinese
Industrial Census 1995; Chinese Economic Census 2004.
Chapter 3
38
Table 3.2: Percentages of Ownership Categories of Industrial Enterprises
in 1985, 1995 and 2004
Enterprise
nr.
Gross
output
Empl.
(year-end)
Enterprise
nr.
Gross
output
Empl.
(year-end)
Enterprise
nr.
Gross
output
Empl.(year-
end)
unit
(100 mill
yuan)
(10 000
persons) unit
(100 mill
yuan)
(10 000
persons) unit
(100 mill
yuan)
(10 000
persons)
1985
Total Industrial enterprises IAS at township level and above Others
Total 100% 100% 100% 100% 100% 100% 100% 100% 100%
1. State owned 1.8% 64.9% 19.6% 73.1% 58.4% 0.5% 10.5%
2. Collective 33.6% 32.1% 79.9% 25.5% 40.7% 30.1% 75.5%
3. Private
4. Sole
proprietorship 64.6% 1.8% 8.6% 69.4% 14.0% 27.0%
5. Joint venture 0.02% 0.8% 0.5% 0.3% 1.0% 0.7%
9. Foreign and HK,
MC, TW funded 0.01% 0.4% 0.08% 0.1% 0.4% 0.1%
10. others 0.02% 0.04% 0.01% 0.02% 0.02%
1995
Total Industrial enterprises IAS at township level and above Others
Total 100% 100% 100% 100% 100% 100% 100% 100% 100%
1. State owned 1.6% 32.6% 31.6% 17.2% 47.1% 52.1% 0.4% 3.5% 3.0%
2. Collective 20.0% 35.5% 39.8% 71.3% 28.8% 36.0% 16.1% 49.0% 45.0%
3. Private 3.9% 2.8% 3.3% 0.5% 0.3% 0.2% 4.2% 8.0% 7.7%
4. Sole
proprietorship 73.6% 11.7% 17.5% 79.1% 35.2% 41.8%
5. Joint venture 0.1% 0.8% 0.6% 1.1% 1.2% 1.0% 0.0% 0.1% 0.0%
6. Incorporated
Enterprise 0.1% 3.3% 1.7% 1.1% 5.0% 3.0% 0.0% 0.1% 0.0%
9 Foreign and HK,
MC, TW funded, of
which 0.7% 13.0% 5.5% 8.7% 17.5% 7.7% 0.1% 4.1% 2.4%
# Foreign funded 3.5% 8.6% 3.2%
# HK, MC and TW
funded 5.2% 8.9% 4.5%
10. others 0.0% 0.1% 0.1% 0.1% 0.1% 0.1% 0.0% 0.0% 0.0%
2004
Total Industrial enterprises above designated size Others
Total 100% 100% 100% 100% 100% 100% 100% 100% 100%
1. State owned 1.8% 10.6% 9.6% 8.5% 11.6% 13.3% 0.2% 0.5% 0.3%
2. Collective 10.3% 4.4% 7.4% 6.5% 3.9% 5.1% 11.3% 9.5% 13.2%
3. Private 65.6% 22.4% 34.7% 43.2% 17.4% 22.9% 71.3% 70.7% 63.8%
4. Sole
proprietorship
5. Joint venture 0.5% 0.5% 0.5% 0.5% 0.5% 0.4% 0.5% 0.5% 0.6%
6. Share-holding
Incorporated 3.6% 1.5% 2.2% 3.0% 1.3% 1.8% 3.8% 3.7% 3.3%
7. Share-holding
Ltd. companies 1.3% 10.4% 5.4% 2.6% 11.4% 7.1% 0.9% 1.1% 1.3%
8. Ltd. companies 7.4% 19.8% 18.2% 14.9% 21.2% 22.8% 5.6% 6.6% 6.9%
9 Foreign and HK,
MC, TW funded, of
which 7.7% 30.2% 21.4% 20.7% 32.7% 26.5% 4.5% 5.5% 8.8%
# Foreign funded 3.7% 19.2% 10.2% 10.4% 20.9% 13.0% 2.0% 2.4% 3.4%
# HK, MC and TW
funded 4.0% 11.0% 11.2% 10.3% 11.8% 13.5% 2.4% 3.1% 5.4%
10. others 1.7% 0.2% 0.6% 0.1% 0.1% 0.1% 2.0% 1.9% 1.9%
Source: See Table 3.1.
Economic Reform, Institutional Change and Economic Development
39
3.2.1 Ownership Reform in SOEs
Since the mid-1980s, when the State Department issued regulations to further enlarge the
managerial autonomy of state industrial enterprises16, the management structure of SOEs
began to change rapidly. That was the early stage of the reform in order to release enterprises
from government control, known as the beginning of government decentralization.
As discussed above, there were two important events in 1992 which promoted China's
economic reform to a higher stage: Deng Xiaoping's southern tour and the 14th National
Congress of the CPC. Both stressed the role of market forces in China's economy. Following
that, the Third Plenary Session of the 14th National Congress of the CPC (shisijie sanzhong
quanhui) in 1993 appealed to construct a modern enterprise mechanism. Compared with the
1980s, the reform of SOEs had become more fundamental, and responsibility in enterprises'
operation was now emphasized.
In the 1990s another SOE reform policy was adopted, known as "invigorating large-scale
enterprises while relaxing control over small enterprises" (zhuada fangxiao). This was
deemed as a "safe" transition policy that could avoid any big turbulence in SOE's reform.
China did not want to follow the risky strategy of direct and large-scale privatization, like in
Eastern Europe and Russia. Instead, China intended to set up some big and powerful state-
controlled industrial groups with competitive advantages, which were to be the backbone of
China's growth. Korea and Japan are examples of countries having successfully developed
similar industrial groups. Greatly supported and controlled by the government, however,
China's SOEs ended up as large-sized but badly-operated enterprises. Due to the monopoly
power and administrative power that was granted to them, these state-controlled (i.e. planned)
enterprises failed to improve their efficiency and innovation abilities. Some scholars call the
reform of large-scale SOEs the most significant failure of the economic reform in China
(Qian, 2003, p.306)
In September 1997, the 15th National Congress of the CPC (shiwu da) suggested a
reconstruction of SOEs, while encouraging a variety of different ownership types.
16 This document from May 1984, called "guanyu jinyibu kuoda guoying qiye zizhuquan de zanxing guiding", is
available at http://www.pt.fjaic.gov.cn/law_show.asp?law_type=GSQY1217 .
Chapter 3
40
Managerial reform and market competition gave a strong incentive for middle-sized and
small SOEs to improve their efficiency. Ownership diversification has no doubt been a very
successful part of the economic reform. Li and Wu (2002) conclude in their paper that
ownership diversification is more important than managerial reform in improving the
performance of SOEs. Lin et al. (1996) point out that the efficiency of SOEs was improved
through greater autonomy and by meeting competition from the non-state sectors.
3.2.2 Township and Village Enterprises (TVEs)
Township and village enterprises (TVEs) originated from the Community Funded Enterprises
named by Chairman Mao in 1959. The formal name “Township and Village Enterprises” was
introduced in 1984, and covers township funds, village funds, joint ventures, and sole
proprietorships. In the beginning, the majority of TVEs were collective enterprises. The share
of collective enterprise numbers decreased greatly in the late 1980s, but the shares of
collective output and employment remained dominant, given that joint ventures and sole
proprietorships both operated on a very small scale. Since the early 1990s, the number of sole
proprietorships and joint ventures developed dramatically, and their share in the total number
of TVEs, as well as their share in total value added and employment of TVEs, exceeded 50
per cent until the end of the 1990s.
The first law on TVEs, The Law of Township and Village Enterprises in the People's
Republic of China, was issued on 1 January 199717, following the emergence of rural-share
holding, cooperatives and foreign-funded enterprises. This law provided the first legal
definition of TVEs: those enterprises mainly funded by rural collective economic
organizations or farmers. In other words, rural collective economic organizations or farmers
should have more than 50% of the whole investment, or if this is not the case, they at least
should play an important role in the share-holding and practical operation.
Township and village enterprises (TVEs) have been regarded as an engine of China's growth.
The value added of industrial TVEs accounts for 45-50% of the total industrial value added
during 1998-2005.
17The content of the law is available from http://www.gov.cn/banshi/2005-06/01/content_3432.htm
Economic Reform, Institutional Change and Economic Development
41
Table 3.3: Value Added of Industry in TVEs, 1995-2005
1995 1996 1997 1998 1999 2000
Value added of industry
total (100 mill yuan) 24951 29448 32921 34018 35861 40034
Value added of industrial
TVEs (100 mill yuan) 10804 12628 11985 15530 17374 18812
Share of TVEs in total
industry 43.3% 42.9% 36.4% 45.7% 48.4% 47.0%
2001 2002 2003 2004 2005
Value added of industry
total (100 mill yuan) 43581 47431 54946 65210 76913
Value added of industrial
TVEs (100 mill yuan) 20315 22773 25745 29359 35662
Share of TVEs in total
industry 46.6% 48.0% 46.9% 45.0% 46.4%
Note: Value added is at current prices. The concept of TVEs in this table includes enterprises with township
funds, village funds, (small) joint ventures, and sole proprietorships.
Source: Value added of industry total is from CSY2006, Table 3-1. Value added of industrial TVEs is from
TVE yearbooks, 1996-2006.
The categories of TVE have changed over time in the statistical yearbooks. The data
presented in TVE statistical yearbooks until 1996 generally make a distinction between four
types of enterprise: township funds, village funds, (small) joint ventures, and sole
proprietorships (TVE 1997)18. In the TVE yearbooks 1998-2000, enterprises are classified
according to three categories only: township and village collective enterprises, joint-venture
enterprises, and sole proprietorships. Since TVE 2001 and onwards, the categories have
changed to 1) collective, 2) share cooperative, 3) joint venture, 4) share-holding Ltd. 6)
private, and 7) others. Most TVEs have become privatised, so that after 2000 the TVE is no
longer the same hybrid private-public enterprise that it was before 2000. Those changes make
it difficult to have a complete and consistent TVE time-series. Hence Table 3.4 provides the
breakdown by township enterprises and village enterprises only till 1999. Chinese statistical
sources do not specify the relationships between the TVE yearbooks and the industrial census
and survey data. There is a substantial overlap between the TVEs in Table 3.4 and the
category of collective enterprises in the first three columns of Table 3.1 (see Szirmai and Ren,
2007, p. 107, for a discussion on the comparisons between TVEs and collective enterprises in
1995).
18 Although in some parts of these yearbooks, the terms township collective and village collective are used, they
have the same meaning as township and village funded enterprises
T
able
3.4
: G
ross
Val
ue A
dded
and
Em
ploy
men
t of
Ind
ustr
y in
TV
Es,
198
7-19
99
19
87
1988
19
89
1990
19
91
1992
19
93
1994
19
95
1996
19
97
1998
19
99
Tow
nshi
p an
d V
illag
e E
nter
pris
es in
Ind
ustr
y
Num
ber
of E
nter
pris
es
9691
77
9972
70
9823
38
9352
39
9284
42
9727
69
1069
133
1053
891
1030
000
9817
99
8440
50
7046
22
6141
15
Gro
ss V
alue
Add
ed (
100
mill
yua
n)
13
52.0
0 16
69.0
0 25
02.0
0 42
75.0
0 61
46.0
0 76
01.0
0 81
95.2
0 81
61.6
1 81
04.4
6 80
45.1
9
Gro
ss V
alue
of
Out
put
(100
mill
yua
n)
2610
.21
3438
.22
4613
.55
5240
.16
6528
.47
9852
.82
1696
2.25
25
524.
72
3474
3.68
35
538.
76
3607
1.06
35
566.
95
3494
3.54
Staf
f an
d W
orke
rs (
year
en
d, 1
0 00
0 pe
rson
s)
3338
.95
3507
.22
3451
.67
3399
.76
3549
.31
3820
.93
4239
.30
4305
.54
4440
.00
4338
.96
3979
.78
3534
.51
3209
.66
Tow
nshi
p E
nter
pris
es in
In
dust
ry
Num
ber
of E
nter
pris
es
2572
16
2669
04
2625
93
2557
37
2548
92
2631
30
2911
59
2906
45
2900
00
2760
43
2373
13
1981
12
1726
65
Gro
ss V
alue
Add
ed (
100
mill
yua
n)
73
7.00
90
5.00
13
41.0
0 22
27.0
0 31
22.0
0 37
72.0
0 40
80.4
0 40
63.6
7 40
35.2
2 40
05.7
1
Gro
ss V
alue
of
Out
put
(100
mill
yua
n)
1413
.88
1840
.36
2485
.62
2807
.95
3540
.01
5281
.06
8836
.90
1296
6.26
17
547.
66
1798
7.36
18
256.
78
1800
1.63
17
686.
10
Staf
f an
d W
orke
rs (
year
en
d, 1
0 00
0 pe
rson
s)
1574
.15
1661
.05
1632
.14
1630
.94
1713
.22
1827
.28
1995
.82
2033
.39
2091
.00
2037
.26
1868
.62
1659
.55
1507
.03
Vill
age
Ent
erpr
ises
in
Indu
stry
Num
ber
of E
nter
pris
es
7119
61
7303
66
7197
45
6795
02
6735
50
7096
39
7779
74
7632
46
7400
00
7057
56
6067
37
5065
10
4414
50
Gro
ss V
alue
Add
ed (
100
mill
yua
n)
61
5.00
76
4.00
11
61.0
0 20
48.0
0 30
24.0
0 38
29.0
0 41
14.8
0 40
97.9
4 40
69.2
4 40
39.4
8
Gro
ss V
alue
of
Out
put
(100
mill
yua
n)
1196
.33
1597
.86
2127
.93
2432
.22
2988
.46
4571
.75
8125
.35
1255
8.46
17
196.
02
1755
1.40
17
814.
29
1756
5.32
17
257.
44
Staf
f an
d W
orke
rs (
year
en
d, 1
0 00
0 pe
rson
s)
1764
.80
1846
.18
1819
.53
1768
.82
1836
.09
1993
.65
2243
.48
2272
.15
2349
.00
2301
.70
2111
.16
1874
.96
1702
.64
Not
e: T
he c
once
pt o
f T
VE
s in
this
tabl
e is
sm
alle
r th
an th
at in
Tab
le 3
.3.
In th
is ta
ble,
TV
Es
refe
r to
ent
erpr
ises
wit
h to
wns
hip
fund
s an
d vi
llage
fun
ds, e
xclu
ding
(sm
all)
jo
int v
entu
res,
and
sol
e pr
opri
etor
ship
s.
Sour
ce: v
ario
us y
earb
ooks
on
tow
nshi
p an
d vi
llage
ent
erpr
ises
.
42
Chapter 3
Economic Reform, Institutional Change and Economic Development
43
3.3 R&D Expenditure and Education
Compared to the growth of GDP, China's R&D expenditure level only started increasing
recently. In 1990, national total R&D expenditure was only 12.5 billion yuan, and the ratio of
R&D to GDP was 0.71 per cent. In the early 1990s this ratio even decreased, while it
fluctuated in the mid and late 1990s (see Figure 3.3). A clear increasing trend emerged only
after 2000. In 2002 the R&D/GDP ratio exceeded 1% for the first time. The R&D/GDP level
in 2005 was 1.24%19, which is still much lower than that of some western countries.
Figure 3.3: Ratio of R&D Expenditure to GDP in China, 1990-2004
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
percentage
Source: China Statistical Yearbook on Science and Technology, various issues.
Regional technology levels show great geographic disparities. The Eastern regions20 of China
generated 67% of national total R&D expenditure in 1999 in China, and this ratio even
increased to 72% in 2004.
Educational expenditures
As stated by Shen Shituan, a member of the National Committee (zhengxie weiyuan), China
is fostering 20% of the secondary and elementary students in this world, while using only
0.78% of educational expenditure. The total educational expenditures in China increased
from 73.2 billion yuan in 1991 to 724.3 billion yuan in 2004. Accordingly the education/GDP
19 It is 1.34% according to the official publications, which is slightly different from our own calculation.
20 Eastern regions include Beijing, Tianjin, Shanghai, Hebei, Liaoning, Jiangsu, Zhejiang, Fujian, Shandong,
Guangdong, Guangxi and Hainan 12 provinces (or cities).
Chapter 3
44
ratio increased from 3.45% in 1991 to 4.42% in 2004. However, this increase mainly resulted
from the raises in tuition and miscellaneous fees21. Government educational expenditure as a
percentage of GDP did not show any improvement over these years. Table 3.5 shows that the
ratio of government expenditure on education to GDP had even been decreasing from 2.6%
in 1980 to 2.3% in 2004. This reveals the limited nature of educational expenditure in China.
Table 3.5: Education Expenditure in China, 1980-2004
1980 1985 1990 1995 2000 2004
Total education expenditure
(billion yuan) - - - 187.8 384.9 724.3
Government expenditure on
education (billion yuan) 11.4 22.7 46.3 141.2 256.3 446.6
Ratio of total education
expenditure to GDP (%) - - - 2.6 3.4 3.8
Ratio of government expenditure
on education to GDP (%) 2.6 2.6 2.5 2.0 2.3 2.3
Note: Total education expenditure consists of 1) government expenditure on education; funds of social
organizations and citizens for running schools; 3) donations and fund-raising for running schools; 5) tuition and
miscellaneous fee, and 6) other educational fee.
Source: China Statistical Yearbook 2006, Table 21-36, and China Statistical Yearbook 2003, Table 20-35.
3.4 FDI and Trade
China's opening up started in the coastal regions, which have advantageous geographical
locations in particular in exporting and importing. In order to strengthen the interactions with
foreign economies Special Economic Zones (SEZs) have been set up. The development of the
SEZ has been one of the most important aspects in China's economic reforms. The first group
of SEZs including Shenzhen, Zhuhai, Shantou and Xiamen, was established in 1980. Those
areas were provided preferential policies, such as reduced custom duties and special subsidies.
SEZs were meant to be the "window" of China in communicating with foreign investors,
exporting products and importing advanced technologies. In addition, China's State Council
established the first Free Trade Zone (FTZ) in Shanghai Waigaoqiao in 1990, followed by
other 14 FTZs22. FTZ is a special type of SEZ. Tariffs and other taxes or duties are not levied
when goods and materials enter the FTZ from outside the country. That is, goods can enter
this zone free of customs duties as long as they stay there. If the goods are finished and enter
21 Total education expenditure consists of 1) government expenditure on education; 2) funds of social
organizations and citizens for running schools; 3) donations and fund-raising for running schools; 5) tuition and
miscellaneous fees, and 6) other educational fees. 22 See also http://www.kejianhome.com/lunwen/436/519/118098.html.
Economic Reform, Institutional Change and Economic Development
45
the market in China from one of the FTZs, a low tax is levied, but if the goods are re-
exported from this zone to other foreign countries, no customs or taxes are levied (Firoz et al,
2003).
Overall, the establishment of SEZs and FTZs greatly improved the process of absorbing
foreign investment and developing foreign trade for China.
3.4.1 FDI
The “Open Door Policy” brought China a remarkable inflow of FDI. The first law on foreign
investment was issued in 1979, known as the “Law on Chinese-Foreign Equity Joint
Ventures” (zhongwai hezi jingying qiye fa). This law guarantees the rights and interests of
foreign firms. Foreign investment first appeared only in some coastal regions in the early
1980s, such as Guangdong, Fujian, and Tianjin. In the mid-1980s, however, FDI had
gradually expanded into almost all areas in China. It made a great leap in the early 1990s. It
increased from 4.4 billion US dollars in 1991 to 11 billion US dollars in 1992, and 27.5
billion US dollars in 1993: a growth rate of no less than 150% each year. In 2005 it had
reached 603.25 million US dollars. FDI per capita increased from 0.6 US dollar per person in
1983 to 58.9 US dollar per person in 2005.
A remarkable benefit provided to foreign companies is the preferential tax policy. Namely,
foreign investors do not have to pay tax during the first and second year of making a profit,
whereas they only have to pay half of the normal tax rate during the third and fourth year.
The normal tax rate applies from the fifth year.23 Moreover, foreign-funded banks are
allowed to do RMB business.24
In 1999, FDI experienced a sharp decline of 11.3% compared with 1998, which has often
been explained by two main features. One is that the Asian crisis during 1997 -1999 caused a
big drop in foreign investment in Asia. The other is that because of market saturation and
fierce competition, foreign investors in small and middle-sized companies had to drop out of
the game25.
23 See also at http://www.chinaunique.com/business/sez.htm .
24 http://www.projectsmonitor.com/detailnews.asp?newsid=11583 .
25 See also at http://www.people.com.cn/GB/channel3/21/20001020/279329.html .
Chapter 3
46
Figure 3.4: FDI in China, 1983-2005
0
100
200
300
400
500
600
700
1983
1985
1987
1989
1991
1993
1995
1997
1999
2001
2003
2005
year
100 mill US dollars
at current prices at 2000 constant price
Source: From China Statistical Yearbook, 2006, 1993 and 1986. The deflator for US dollars is from the
International Monetary Fund (IMF) database, http://www.imf.org/external/data.htm.
As regards investments in fixed assets, foreign investment in 2005 was 30 times the value of
1981 (at 1980 constant prices). Especially during the period 1994-1997, the share of FDI in
total investment in fixed assets (TIFA) was more than 10%. After 1997, FDI decreased to its
lowest level in 2000. Afterwards it increased again. The share of FDI in TIFA did not change
much between 2000 and 2005, lying between 4 and 5%.
Figure 3.5: Total Investment in Fixed Assets from FDI
0
500
1000
1500
2000
2500
3000
3500
4000
4500
1981
1983
1985
1987
1989
1991
1993
1995
1997
1999
2001
2003
2005
year
100 mill yuan
at current prices at 1980 constant price
Source: DSIFA (Department of Statistics on Investment in Fixed Assets National Bureau of Statistics of China),
(2002), Statistics on Investment in Fixed Assets of China, 1950-2000, China Statistics Press. 2002
Economic Reform, Institutional Change and Economic Development
47
Table 3.6: FDI as percentage of TIFA and GDP, and FDI per Capita, 1981-2005
FDI in
TIFA Percentages
FDI per
capita
FDI in TIFA FDI in GDP (yuan/person)
1981 36.36 3.78 0.74 3.63
1982 60.51 4.92 1.14 5.96
1983 66.55 4.65 1.12 6.49
1984 70.66 3.86 0.98 6.83
1985 91.48 3.60 1.01 8.64
1986 137.31 4.40 1.34 12.77
1987 181.97 4.80 1.51 16.65
1988 275.31 5.79 1.83 24.80
1989 291.08 6.60 1.71 25.83
1990 284.61 6.30 1.52 24.89
1991 318.89 5.70 1.46 27.53
1992 468.66 5.80 1.74 40.00
1993 954.28 7.30 2.70 80.52
1994 1768.95 10.38 3.67 147.60
1995 2295.89 11.47 3.78 189.55
1996 2747.41 11.96 3.86 224.48
1997 2683.89 10.76 3.40 217.10
1998 2617.03 9.21 3.10 209.76
1999 2006.78 6.72 2.24 159.54
2000 1696.24 5.15 1.71 133.83
2001 1730.73 4.65 1.58 135.61
2002 2084.98 4.80 1.73 162.31
2003 2599.35 4.68 1.91 201.15
2004 3285.70 4.66 2.06 252.77
2005 3978.80 4.48 2.17 304.29
Note: at current prices, 100 mill yuan. TIFA is short for total investment in fixed assets.
Source: From various China Statistical Yearbooks.
FDI can contribute both directly and indirectly to the host country. The direct effect implies
that capital input from FDI increases economic growth at the local level. FDI also provides
more employment opportunities. The indirect contribution of FDI refers to the transfer of
knowledge or technology. This is commonly referred to as “knowledge spillovers”. This
concept will be discussed extensively in Chapter 8.
Zhang and Felmingham (2002) emphasize the contribution of FDI to economic growth in
China, both in its direct effects and its externality effects. On the contrary, Qian (2003, p.299)
states that the role of FDI has been "vastly overstated", because FDI, even at its peak, was
only one tenth of the total investment in China, and FDI per capita is still not high according
to international standards. This will be further examined in Chapter 8.
Chapter 3
48
3.4.2 Exports
Exports are often regarded as another important source of economic growth in China,
together with increased openness and FDI inflows. Before the economic reforms, exports had
remained stable at around 4% of GDP till the end of the 1970s. In 1986, exports for the first
time reached a value higher than 100 billion yuan, whereas they exceeded 1000 billion yuan
in 1994 and onwards. The exports to GDP ratio increased from 4.6% in 1978 to 34.2% in
2005: an increase of more than 7 times in 27 years. However, if we convert GDP into US
dollars by purchasing power parities (PPPs) and exports by exchange, the ratio of exports to
GDP will become much lower. This is mainly due to the fact there is a large non-trade sector.
Lin (2004) states, "China’s economic growth and opening up, followed by continuing
integration into the global economy, is indispensably linked with the systemic change
oriented towards the market system, on the one hand, and export-led growth, on the other."
However, different from the export-led growth hypothesis, some literature suggests China's
reform and economic growth has often been misinterpreted by overstating the driving force
of FDI and exports. Qian (2003, p.300) argues that "it is not exports that drive growth, but
that the same forces of domestic change drive both exports and domestic growth." Hence, he
argues that the rapid growth happens not only in coastal regions but also in the inland of
China.
Figure3.6: China's Exports, 1952-2005
0
10000
20000
30000
40000
50000
60000
70000
195219541956195819601962196419661968197019721974197619781980198219841986198819901992199419961998200020022004
year
100 mill yuan
Source: China Statistical Yearbook 2006 (Table 18-5) and China Statistical Yearbook 1993 (p.573).
Economic Reform, Institutional Change and Economic Development
49
Figure 3.7: Ratio of Exports to GDP in China, 1952-2005
0
5
10
15
20
25
30
35
40
1952
1954
1956
1958
1960
1962
1964
1966
1968
1970
1972
1974
1976
1978
1980
1982
1984
1986
1988
1990
1992
1994
1996
1998
2000
2002
2004
year
percentage
Source: China Statistical Yearbook 2006 (Table 3-1 and Table 18-5), China Statistical Yearbook 1999 (Table 3-
1), and China Statistical Yearbook 1993 (p.573).
3.5 Conclusions
This chapter provides a general overview of economic reform in China since 1978. Through a
step-by-step trial and error reform process, China succeeded in transforming from a planned
to a market economy. This reform has been successful in the sense that it resulted in a high
GDP growth rate. China's GDP has grown at a 9 per cent average rate through 1978-2005,
and GDP per capita in 2005 was 10 times as high as that in 1978. The evolution of economic
growth in the whole of Chinese industry can be divided into two stages: there was a “gradual
growth” during 1978-1992, while there was a “dramatic increase” of economic growth in the
period 1992-2005.
China's growth has also been accompanied by a series of changes, e.g. institutional change,
technology improvement, and openness to FDI and trade. Institutional change is
characterized by a big drop of the share of state-owned enterprises and an increase of various
different ownership types. Township and village enterprises (TVEs) which started from the
negligible small community-funded enterprises have showed an impressive growth and
contribution to China's economy.
Chapter 3
50
The R&D intensity in China has improved as well. However, it improved later than the
growth of GDP and industry. The ratio of R&D expenditure to GDP shows a clear raise only
after 2000. Education expenditure is still rather low in China. Compared with the aggregate
growth of China, investments in education are lagging far behind. FDI and exports have
increased remarkably since the mid-1990s.
CHAPTER 4
Data and Statistical Problems
Chinese time series on manufacturing suffer from a number of problems which need
to be taken into account when analysing Chinese manufacturing performance (Szirmai
et al. 2005; Wu, 2004; Maddison, 1998; Holz and Lin, 2001). This chapter and
Chapter 5 provide a discussion of data issues in published Chinese statistical sources
and explain the construction of our own database with consistent long-run time series,
broken down by region.
4.1 Data Problems
Szirmai et al. (2005) have made a variety of adjustments to the published sectoral time
series for industry and manufacturing for the total economy. The same adjustments
have been made in this thesis so that the series broken down by region and by
ownership categories are consistent with the macro series of the previous paper by
Szirmai et al. (2005).1 We have applied the same adjustment factors to all regions. We
have not made region-specific adjustments. Problems relevant for this thesis include
the following:
1. Frequent changes in output concepts, such as the shift from net industrial output to
gross value added in 1993. To achieve consistency in the whole time series, we
have adjusted the net industrial output concept used prior to 1993 to the gross
value added concept introduced in 1993.
2. Changes in employment concepts, which make it difficult to create consistent
series of productivity. These changes are especially marked at sectoral level. Also,
1 For a detailed discussion of the adjustments, the reader is referred to the earlier paper.
Chapter 4
52
the coverage of employment by sector is not consistent with the coverage of
output after 1993. Based on the Szirmai et al. 2005 series, we have adjusted the
employment series for these differences in coverage.
3. Changes in coverage over time, such the shift from enterprises at township level
and above prior to 1998, to all state enterprises, plus all non-state enterprises with
more than five million yuan in annual sales from 1998 onwards. No adjustments
have been made for this. The breaks in the series are not very dramatic (Szirmai,
et al, 2005; Holz, 2001).
4. Incomplete coverage. Most data are available for the state-owned sector, which is
becoming less important over time. Individual ownerships and enterprises at
village level are not covered. The coverage of the detailed time series is declining
over time.
5. The manufacturing sector in Chinese statistics is not clearly distinguished from
industry (which also includes mining and utilities). Breakdown by region and
ownership is only available for total industry. Only for sectoral analysis we can
provide specific data for manufacturing. This implies that some of the shift-share
analyses below will be performed for industry, others for manufacturing.
6. Incomplete integration of regional data and national data. National data contain
some regional breakdown, but these data are not broken down by sector or
ownership categories. Detailed regional data are published in regional statistical
yearbooks, but these data are not always consistent in concepts and coverage with
the national data.
4.2 Construction of Data
Sectoral data
For the sectoral data on value added and employment, we use the adjusted
manufacturing time series 1980-2002 from Szirmai et al (2005), which are corrected
for changes in concepts and coverage. Given the lack of data for hours worked, labour
productivity refers to value added per person engaged. As a first step towards the
analysis of technology classes, we also break down the data into high-tech and low-
tech sectors.
Data and Statistical Problems
53
Note that the coverage of these and subsequent series refers to enterprises at township
level and above, with independent accounting systems. In 1995, these enterprises
accounted for 87.5 per cent of total value added in manufacturing (Szirmai et al., 2005,
Table 8). In 1998, coverage of the detailed time series shifted to all state enterprises,
plus all enterprises with more than five million yuan in sales (Holz and Lin, 2001).
The exact implication of these breaks in the series are not known as there are no
overlapping years, but the breaks are not very dramatic (Szirmai et al., 2005). Over
time, the coverage of the time series is declining.
Institutional data
We have consistent time series (1980-2002) for the following ownership categories:
state-owned and state-holding enterprises, collective-owned enterprises, share-holding
corporations ltd., private enterprises, foreign-funded enterprises, and enterprises
funded from Hong Kong, Macao and Taiwan. Since the categories "state-owned and
state-holding enterprises" and "share-holding corporations Ltd." have some overlap,
the two categories cannot be used together in one table. In the ownership tables, we
classified "state-owned and state-holding" as one category and put all other share-
holding enterprises into the category of "others"). Institutional breakdown is only
available for total industry (including mining and utilities), not for manufacturing. The
sum of value added and employment by institution from data obtained directly from
Chinese publications is different from the sectoral national total. In order to keep it
consistent in the shift-share analysis in Chapter 6, the data of value added and
employment by institution are adjusted to be consistent with Szirmai et al (2005).
Regional data
We make use of output and employment data by region from the China Statistical
Yearbooks (CSY), and the China Industrial Economy Statistical Yearbooks (CIESY).
For the period 1985-2002, these are only available for total industry, not for
manufacturing. Crosstabulated data for regions and ownership categories can only be
constructed for an even shorter period, namely 1992-2002. A breakdown by
ownership categories for 1992 was not available. We found that the aggregate
productivity figures for 1993 are extreme outliers. Therefore, we decided to use the
Chapter 4
54
published totals for 1992 and break these down by the ownership and regional and
institutional proportions of 1993 (see Table 6.4 and Table 6.5).2
Price indices of industrial products from the CSY, 2003 (p.313) are used to deflate the
current price series of value added (for all sectoral, institutional and regional series) to
constant 1980 prices.
It is often assumed that there are serious discrepancies between data from national
sources and those from regional sources. This is not correct. We systematically
compared the data from both national statistics and regional yearbooks for a number
of benchmark years. The differences between the two sources were negligible (see
Appendix Table D-3 and D-4 for the benchmark years 1989 and 2003). Therefore, we
use regional sources to supplement data in the national sources where the regional
breakdowns in the latter are incomplete. Our sources for the national data are the
CSY, the CIESY, the industrial censuses of 1995 and 1995, the China Labour
Statistical Yearbooks (CLSY) and the Statistics of China’s Industry and Transport,
1949-1999 (SCIT, 2000). Data from regional yearbooks are used whenever regional
information in the national sources is lacking.
Szirmai, Ren and Bai (2005) and Szirmai and Ren (2007) provide a detailed
discussion of the problems of long-term time series of value added and employment
for the national economy. There are major inconsistency problems in the series, in
terms of concepts and coverage. The authors have made a large number of
adjustments to achieve consistency in these series, which will not be further discussed
in this thesis. Our aim has been to make the regional series of employment consistent
with the adjusted aggregated national series for total industry. In order to achieve this,
we have applied the adjustment factors for the national level to all regional data. For
the adjustments, the reader is referred to the two publications cited above.
2 This means we slightly underestimate the impact of ownership changes for 1992-97, and
overestimate them for 1985-1992. The reason for doing this is that the totals for 1993 are outliers
which seriously distort the shift and share analysis.
Data and Statistical Problems
55
Employment data
The choice of employment data for the years after 1999 needs to be elaborated further
here. We have found great inconsistencies between two sets of employment data.
One set of data derives from the staff and workers time series in the CLSY (2006,
p.25 and p.29). This series has been adjusted according to Szirmai and Ren (2007).
There are two upward adjustments, one for the fact that the coverage of the staff and
worker series is limited to urban workers, while the output data include output of rural
township workers. Another adjustment corrects for the fact that after 1998 millions of
so-called not-on-post workers are suddenly excluded, leading to a break in the series.
We refer to this adjusted time series as series I.
The second employment series – series II - derives from the CSIEY. This series
probably has another employment concept, but the two series roughly are comparable
up to 1997. If we make the same adjustment for not-on-post workers as for series I,
the two series are even roughly comparable up to 2002. Both series indicate that
industrial employment is shrinking (see also Banister, 2005).
However, the series II data suddenly stop declining in 2002 and explode upwards
from 55.2 million workers in 2002 to 69.0 million workers in 2005, an increase of 13
million workers in three years. In series I, one can also discern an upturn in
employment after 2002 but it is much more modest, from 59.4 million workers in
2002 to 61.9 million in 2005.
This results in a fundamental difference between the two series. Series I shows a
substantial net decline in employment from 1999 to 2005: 70.5 to 61.9 million
workers. This is consistent with other assessments of jobless growth in Chinese
industry. Series II shows at net increase in employment between 1999 and 2005, 58.1
to 69.0 million. Such discrepancies have immense implications for the analysis of
productivity trends. Until this mystery is sorted out, we have decided to limit the time
span of the detailed analysis of total factor productivity to the period from 1980 till
20023. We have chosen for the adjusted series of staff and workers deriving from the
3 The two series are comparable up to 2002, if we apply the same adjustment for not-on-post workers.
Chapter 4
56
labour statistics yearbooks.4 Appendix Table D-5 presents the employment data used
in our analysis in later chapters. The adjustment is consistent with Szirmai and Ren,
2007.
R&D expenditure
There are two types of R&D variables often used in the literature: R&D stocks and
R&D expenditure figures. In the first approach, R&D is treated as an investment
which is accumulated from each year with a certain depreciation rate. Applying the
perpetual inventory method, the R&D capital stock of region i at time t ( itRD ) can be
calculated by ititit rdRDRD +−=−
)1(1 δ , where δ is the depreciation rate and itrd is
the R&D expenditure of year t. The depreciation rate is normally assumed to be either
15% (Los and Verspagen, 2000; Griliches, 1990; Raut, 1995) or 5% (Coe and
Helpman, 1995). The R&D stock in the initial year can be estimated through dividing
the R&D expenditure in the initial year by the sum of growth rate of R&D
expenditure and the depreciation rate, i.e. )/(0 δ+grd i (Coe and Helpman, 1995; Los
and Verspagen, 2000). Another method consists of using the flow of R&D
expenditure over GDP5. We will use the second type of R&D variable in our research,
namely its ratio to GDP. There are two reasons for doing this. One is that our spillover
model in chapter 8 includes both R&D spillovers and FDI spillovers, so in order to
treat them in a comparable fashion, it is better to measure both variables in the same
way, i.e. as flows. The other reason is that because actual R&D data in China are only
available for a relatively small number of recent years, it is almost impossible to
derive the R&D stocks. Instead expenditure on science and technology (S&T)
activities6 has been used as a replacement. S&T expenditure covers R&D expenditure
and also some other related expenses. Using S&T expenditure to estimate the R&D
stock is obviously inappropriate, but their ratios to GDP are rather reliable.
Considering the lag of contribution of technological investment, we use one year lags.
4 In case of the coefficient of variation of labour productivity, we have included the years 2003-2005,
because we are interested in the regional distribution, rather than the level of productivity. 5 See also the discussions in Los and Verspagen (2000). 6 According to the China Statistical Yearbook on Science and Technology 2005 (p.436), the
expenditure of science and technology activities refers to the actual total expenses spent in this
particular unit on S&T activities in the report period, including expenses on wages, research business,
research management, fixed assets in non-basic construction, and other S&D activities; not including
the expenditure on productive activities, paying off loans or money transferred to other units.
Data and Statistical Problems
57
Data on science and technology are taken from the China Statistical Yearbook on
Science and Technology (various issues). The S&T expenditure generally (except in
some earlier S&T yearbooks) is the sum of four categories: independent research
institutions, large & medium-sized industrial enterprises, institutions of higher
education, and others. The reason for using the sum of S&T expenditure instead of
that of mere industrial enterprises, is that we intend to measure both intra- and inter-
industry knowledge spillovers. Innovations used in industries can originate not only
from laboratories of industrial enterprises, but also from independent research
institutions, universities and other industries.
Foreign Direct Investment
Our FDI variable is represented by the ratio of FDI to GDP in a particular region7.
One year lags are applied, i.e. we use the ratio for year t-1 as the FDI variable. Capital
input and labour input of foreign companies are already included in the variables of
capital (K) and labour (L), so these should not be double counted by adding extra
variables for them. Our purpose in chapter 8 is to use the FDI/GDP ratio to explain the
technological spillover effects from FDI.
Capital input
Published information on capital investment in Chinese statistics is not yet consistent
with the SNA framework, which implies that such data cannot directly be used in
productivity calculations. In recent years, considerable progress has been made in
estimating capital stocks at the national level (e.g. Holz, 2006; Wu and Xu, 2002;
Chow, 1993; Huang et al. 2002; and Cao et al. 2007). However, so far there has been
little research on regional capital stocks, except Wu (2004). Hence we will use our
own regional data to estimate the fixed capital stock for 30 regions. The procedures
are discussed in chapter 5).
7 The published FDI data in Chinese yearbooks are in US dollars (at current prices). Before calculating
the ratios, we adjusted FDI to Chinese yuan (at current prices), and further to the same constant price as
total investment in fixed assets.
CHAPTER 5
Regional Capital Inputs in Chinese Industry and
Manufacturing1
5.1 Introduction
This chapter provides new estimates of capital inputs in the Chinese economy.
Estimates are made for the total economy, the industrial sector and the manufacturing
sector. The estimates for manufacturing are broken down by 30 regions for the period
1985-2003.
The measurement of capital inputs is fraught with difficulties. Otherwise than labour
inputs, fixed assets are produced inputs that can be used repeatedly in the production
process over longer periods. Varying service lives and the decline of the productive
capabilities of fixed assets over time make it hard to measure capital inputs accurately.
Partly due to the difficulty of observing capital services directly, the productive
capital stock and the wealth capital stock are often confused in practice. These need to
be distinguished. As fixed assets age, the decline in their productive capability is
represented by their age-efficiency profiles. The decline in the market value of assets
is represented by their age-price profiles. The age-efficiency profile is used in
estimates of capital services in productivity analysis, The age-price profile is relevant
to the measurement of the net capital stock and consumption of fixed capital in wealth
accounting (OECD, 2001a, p.15; OECD, 2001b, p.53; Hulten and Wykoff, 1996,
p.13).
1 This chapter is based on Wang and Szirmai (2008b).
Chapter 5
60
Capital services are primary inputs into the production process. To use the wealth
capital stock (either gross or net) in production analysis is theoretically wrong,
because capital service, like labour input, is a flow rather than a stock. When referring
to capital inputs, we will use the term volume indices of capital service (OECD, 2001a,
p.21, 84; Triplett, 1996, 1997).
In the case of China, things are further complicated by the lack of sufficient published
data on investment in fixed assets and a measurement system that still deviates from
the SNA. In Chinese statistics, fixed assets acquired in different years are normally
valued at their historical acquisition prices. According to the SNA, the capital cost in
the production process should “reflect underlying resource costs and relative demands
at the time the production takes place. It should therefore be calculated using the
actual or estimated prices and rentals of fixed assets prevailing at the time and not at
the times the goods were originally acquired” (SNA, 1993, Par 6.180).
Some of the earlier attempts at measurement of capital inputs are constrained by
inappropriate conceptual frameworks. Several of those studies simply use the wealth
capital account derived from information about gross fixed assets and economic
depreciation (Chow, 1993; Chen et al., 1988; Holz, 2006).
Our estimates of capital inputs into Chinese manufacturing will be based on the
production and productivity analysis in Jorgenson's work, in line with the framework
of 1993 SNA, and complementary literature on the difference between productive
capital stock and wealth capital stock (Triplett, 1996, 1997; Hulten,1990; Hulten and
Wykoff, 1996).
The chapter is structured as follows. In section 2, we discuss measurement issues with
regard to capital inputs. In particular, we focus on the distinction between capital
services and the wealth capital stock. Capital services are the appropriate inputs for
productivity analysis. The wealth capital stock is the appropriate concept for national
accounts. In section 3, we introduce the basic concepts used in Chinese statistics. In
section 4 we discuss the different ways in which investment in fixed assets are broken
down in Chinese statistical practice. Earlier estimates of Chinese capital inputs are
Regional Capital Inputs
61
discussed in section 5, in the light of the theoretical and practice issues and problems
raised in the sections 2 to 4. Section 6 presents new estimates of capital services for
the total economy, for industry and for manufacturing. The final section 7 presents
regional capital estimates.
5.2 Measuring Capital Inputs into the Production Process
5.2.1 Capital Services vs. Wealth Capital Stocks
Based on the concepts of 1993 SNA, Triplett (1997) and Hulten and Wykoff (1996)
make an important distinction between the concept of "productive capital " used in
productivity analysis and the concept of "wealth capital stock" used in wealth
accounting (OECD, 2001b, p.53).
In the SNA, the term depreciation normally equals the consumption of fixed capital,
which denotes "the reduction in the value of the fixed assets used in production during
the accounting period resulting from physical deterioration, normal obsolescence or
normal accidental damage" (SNA-Glossary, 1993). This is a wealth accounting
concept. In the SNA framework, the same term depreciation is also used to refer to
the decay of the productive capacity of fixed assets in the production process. This
relates to the quantity of productive capital services, rather than to figures in the
balance sheets in the business sector. In order to avoid confusion, in this chapter we
consistently use decay in the context of productivity analysis, and depreciation in the
context of wealth accounting.
Volume indices of capital services measure the contribution of capital to the
production process. They reflect the productive capability of capital and are used in
productivity analysis. Thus, for the analysis of total factor productivity (TFP), the
index of capital services is the most appropriate capital input. In contrast, the wealth
capital stock reflects the market valuation of fixed assets, used especially in business
accounts. The main differences between the different capital concepts are listed in the
following table.
Chapter 5
62
Table 5.1: Capital Concepts
Capital services Productive capital
stock
Wealth capital stock
Object Flows of capital
services
Productive capital
stock
Stock of capital goods
Field of
Application
Production and
productivity analysis
Production and
Productivity analysis
Income and wealth,
business accounts
Focus of
Measurement
Capability or
efficiency
Capability or
efficiency
Capital value
Deterioration Decay in productive
capability
Decay in productive
capability
Economic
depreciation/capital
consumption
Value Age-efficiency profile Age-efficiency
profile
Age-price profile
Price weights for
aggregation
Rental prices (or user
cost)
Deflated acquisition
prices of fixed assets
Acquisition prices of fixed
assets adjusted to current
values
The decay of productive capabilities and economic depreciation are separate concepts,
but they are not independent of each other. In the following section, we will discuss
the relationships between rental prices and asset prices. We shall see that capital
service is an important basis for determining the value of fixed assets (in wealth
accounting), but not the other way around.
5.2.2 Measuring Capital Services
In theory, to construct volume indices of capital services (VICS), different types of
fixed assets of different ages first need to be converted into standard efficiency units
(known as the quantity of capital services). Next, these units are multiplied by the
rental prices (or user costs) which are the unit prices of such services (OECD, 2001, p.
21). Rental prices are the appropriate weights for the construction of volume indices
of capital service inputs. This is the preferred solution. In practice, however, it is
difficult to find quantities and prices for capital services and indirect estimation
methods of user cost require additional assumptions. Therefore some researchers
continue to use the productive capital stock as a proxy for capital services, assuming
that capital services are proportional to the productive capital stock. Thus, capital
services are measured as the use of a stock of fixed assets during a specified period
Regional Capital Inputs
63
(e.g. a year). The productive stocks are calculated by cumulating the values of gross
fixed capital formation (GFCF) multiplied by their age-efficiency coefficients.
In productivity studies, the capital stock2 (K) can be expressed as the sum of
productive investment (IN) in the production process. The productive capabilities of
each type of investment (fixed assets) should be converted into standard efficiency
units.3
TtTttt INININK −− +++= φφφ ...110
or ∑=
−=T
s
stst INK
0
φ (1)
where, T is the average service life of investment, s is the age of a fixed asset, and φ
is the productive capability (or efficiency) coefficients of an asset. 1=φ for a new
asset at time t, and 0=φ when its service life is over. φ is also equal to the ratio of
the marginal product of currently used assets to the marginal product of a similar new
asset (under the condition that the capital- labour ratio remains constant) (Hulten and
Wykoff, 1996, p.14).
Taking into account retirement patterns (scrapping patterns) and price indices (OECD,
2001b, p.132), the productive capital stock can be written as
0,0 st
sts
T
s
stp
INFK
−
−
=
⋅=∑φ (2)
where 0,stp − is the price index of year st − relative to year 0, sF is the retirement
rate of assets at age s. 4
In obtaining volume indices of capital stocks, estimating the efficiency coefficient φ
is an important step. With a given decay function, the Perpetual Inventory Method
estimates the decay in efficiency of capital assets. The volume index of capital
services is assumed to be proportionate to the index of the capital stock.
2 In the remainder of this chapter, the term capital stock will be used as shorthand for the productive
capital stock, as long as we remain within the scope of productivity analysis. 3 Triplett writes that capital stocks should not be called capital inputs (Triplett, 1998 p. 2). Unless we
are explicitly referring to capital flows, we will use the term "capital stock" rather than capital input. 4 A further refinement in capital input measurement is to take the mortality function into account. The
mortality function refers to the distribution of retirements around the average service life of an asset.
Mortality functions are not taken into account in this chapter.
Chapter 5
64
Choosing a proper decay function is a key to this method. An important point worth
noting is that the term depreciation used in SNA 1993 represents the declining
productive capability of fixed assets (in productivity analysis), rather than the
allocation of the costs of fixed assets over the successive accounting years (in the
business accounts) (SNA-glossary, p.13). The decline of productive capability is
different from depreciation in the wealth accounting field (OECD, 2001b, p.53). In
order to avoid the confusion surrounding the use of the term depreciation, we use the
term decay of efficiency in the context of production analysis and the term economic
depreciation in the context of wealth accounting.
There are five main decay patterns for the loss of efficiency of the capital stock in its
contributions to the production process, each of them based on different assumptions:
One-hoss shay, straight-line decay, geometric decay, the double declining balance
method (Hulten, 1990) and the hyperbolic decay pattern. The decay patterns apply to
both capital stocks and capital services.
• The One-hoss-shay efficiency method
The one-hoss-shay pattern assumes that fixed assets provide constant productive
services throughout the whole service life (T) of the asset
≥
−==
Ts
Tss
0
1,...1,01φ (3)
where s is the age of the fixed asset.
In such an efficiency profile, fixed assets are able to operate as efficiently as new ones,
as long as they exist. There is no productivity decay. The typical example of this
pattern is the computer. In this pattern, the productive capital stock is equal to the
gross capital stock. The one-hoss-shay pattern seems inappropriate for the aggregated
capital estimation (Triplett,1997, p.14).
• Straight-line depreciation
The straight-line depreciation model assumes that the productive efficiency of an asset
decays by an equal amount every year of its service life.
Regional Capital Inputs
65
≥
−=−=Ts
TsT
s
s0
1,...1,01φ (4)
• The geometric model
The geometric assumes that efficiency decay takes place at a constant rate every year.
10 =φ (5)
)1(1 δφφ −⋅= −ss 1,...,1 −= Ts
where δ is the decay rate(Jorgenson, 1990). This pattern is used in Canada.
However it is not appropriate in the sense that assets will be used infinitely without
ever being retired though with very small values in the late stages. In practice, this
approach is therefore sometimes substituted by the double declining balance method.
• The double declining balance method
The double declining balance method is a combination of the geometric model and the
straight line method. The double declining method uses a geometric decay rate based
on doubling the decay rate for the first service year of an asset as calculated according
to the straight-line method. This decay rate is applied to subsequent years. When the
efficiency rate calculated using this decay rate drops below the efficiency rate
calculated with the straight-line method, the method switches to the straight-line
decay rate for the final years.
• The hyperbolic pattern.
)/()( sTsTs βφ −−= (6)
In this pattern, the productive efficiency falls slowly in early periods and more rapidly
in later stages (Triplett, 1997, p.14). The Bureau of Labor Statistics (BLS) in the US
and the Australian Bureau of Statistics (ABS) use the hyperbolic-efficiency function.
As a matter of fact, this pattern is a very suitable pattern for an increasing number of
high-tech fixed assets, like computers (or software). Such fixed assets normally don't
lose much of their capabilities in the early stages. The values used for the slope
coefficient ( β ) are 0.5 for equipment, and 0.75 for structures (OECD, 2001a, p.86).
Chapter 5
66
In determining the shape of the efficiency functions, Jorgenson (1990) and Hulten
(1990) use relative marginal products, while Triplett prefers engineering information
(Triplett, 1997, p.12).
5.2.3 The Relationship between Age-Efficiency Profiles and Age-Price Profiles
As stated above, the age-efficiency profile describes the pattern of decline of
productive efficiency of assets, while the age-price profile portrays the pattern of
changes in asset values. The latter is appropriate for the estimation of the net capital
stock and the consumption of fixed assets in national accounts. Age-efficiency and
age-price profiles are related, but not identical to each other.
For instance, obsolescence is an important factor in age-price profiles but not in age
efficiency profiles. Obsolescence reduces the value of an asset in wealth accounting.
It does not affect the amount of capital services provided by the fixed assets in the
production process. Thus, the introduction of a newly invented (similar) fixed asset
will reduce the value of an existing fixed asset considerably. However, the capital
service of the existing asset will remain unchanged.
The pattern of decline over time may also differ. The market value of fixed assets will
often decline rapidly in the first years of use, while the productive capability declines
much less in the initial period.
Using depreciation figures directly from published yearbooks implicitly denotes a
choice for the wealth accounting concept. Unfortunately in practice, depreciation is
often used to measure the decay of productive capacity.
Hulten (1990) and Hulten and Wykoff (1996) discuss the links between productive
capability (efficiency) and economic depreciation in wealth accounting - the
relationship between φ and δ - (see Wu, 2002, p.13; Jorgenson, 1973; Hulten, 1990,
p.128; and Hulten & Wykoff, 1996, p.14).
Regional Capital Inputs
67
One way to estimate efficiency declines is to use the marginal products of assets with
different ages or rental prices, if market data are available (Hulten, 1990, p.127;
Schreyer, et al. 2003, p.9). Rentals are equal to quantity of capital services multiplied
by the unit price of services.
TsP
Ps
t
st,...2,1,
0,
, == φ (7)
stP , is the rental, i.e. the income, from a s-year-old fixed asset at time t. The ratio of
stP , to the rental price of a comparable new machine 0,tP indicates the relative
marginal productivity of two vintages, which equals the efficiency coefficient (φ ) of
productive capability of assets with age s. However, it is very difficult to obtain the
rental prices of certain fixed assets, given that most fixed assets are used by their
owners.
Jorgenson (1963, 1990) discusses the use of the concept of "user cost" instead of
rental price. Under perfect competition, marginal productivity is equal to the rental
price. The rental price represents the revenue the fixed asset can obtain in a given year.
The value of a fixed asset at year t should be equal to all the income gained in the
remaining years of its service life, discounted to the present year. Therefore it has
three main determinants: the rental prices of this fixed asset, a discount rate5 and the
scrap value.
Assume the income generated by this fixed asset at time t is stP , . With an interest rate
r, the value of a fixed asset at age s should be
∑∞
=+++
+=
01
,,
)1(ττττ
r
PV
stst (8)
where τ is the number of years starting from year t.
5 According to OECD (2001a, p. 16), the discount rate is "often taken as the interest rate on long-term
bonds", and it is also stated that the discount rate in "real terms" is "a nominal rate of interest minus the
rate of general inflation" (OECD, 2001a, p. 17).
Chapter 5
68
Given the productive capability variableφ , the value stV , of a fixed asset at age s can
be expressed as a fraction ( τφ ) of the value of a new fixed asset.6 Thus we get
∑∞
=+++
+=
01
0,,
)1(ττττφ
r
PV
ts
st (9)
The two equations above do not take the scrap value into account. This means they
assume that a fixed asset will stay in the capital stock forever without being discarded
(T=∞ ), even though its productive contribution is very small in the far future.
If we take into consideration the retirement of fixed assets at the end of their service
lives T, then we can get
sT
sTst
str
valueScrap
r
PV
−
−−
=+++
++
+= ∑
)1()1(
1
01
,,
ττττ
= sTsT
TsTtstst
r
valueScrap
r
P
r
P
r
P
−−−−+++
++
+++
++
+ )1()1()1(1
,1)(
2
1,1,L (10)
The economic depreciation rate equals
st
stst
V
V
,
1,, 1
+−=δ . (11)
Thus we have
0,,,, )( tsststst PPVr φδ ==+ (12)
which connects the economic depreciation rate (δ ), the value of a fixed asset (V ) and
the rental price ( P ). From the above equation, one sees that the economic
depreciation rate (δ ) and efficiency decay rate (φ ) are only the same in the geometric
pattern. If the depreciation rate is constant over time, we can get ss )1( δφ −= . (For
the derivations see Appendix A, see also Wu, 2002, p.13).
In all other decay (or depreciation) functions, δ and φ are not the same and cannot be
substituted for each other. OECD (2001b, p.58-67) provides examples of the different
shapes of age-price profiles and age-efficiency profiles
6 The new vintage doesn't have to be identical to the old one. It can represent a technologically more
advanced version of the same type of asset. Thus the efficiency rate (φ ) also incorporates the influence
of technological obsolescence.
Regional Capital Inputs
69
5.3 Measurement of Capital Stocks in China: Basic Concepts
Measuring capital stocks in China is even more difficult than in other countries, as the
National Bureau of Statistics (NBS) in China uses a framework which deviates from
the SNA (see also Chapter 4), and because the published statistics are not consistent
over time.
Basic concepts and variables with regard to fixed assets in China:
The commonly used variables related to capital estimates are the following:
- Total Investment in Fixed Assets (TIFA). TIFA includes the "volume of activities in
construction7 and purchases of fixed assets and related fees" (China Statistical
Yearbook 2005). However, this term is broader than the formation of fixed assets in
two ways. First, it includes "activities" that will never be transformed into fixed
assets.8 Next, besides productive fixed assets according to the SNA conception, TIFA
also includes the non-productive part of investment, such as inventories and the
residential capital stock.9
TIFA data are available from 1950 to present. Between 1950 and 1979, the data only
refer to state-owned units. Prior to 1996, TIFA had a coverage of enterprises with
investment of more than 50 thousand yuan per year. However, except for investments
in real estate development, rural collective investment and individual investment, the
coverage changed to more than 500 thousand yuan from 1997 onwards. The data for
1996 are published for the two types of coverage. They show that the investment with
the more limited coverage is only 0.26% lower than the investment with the more
extended coverage of the earlier series. This is not a serious discrepancy and can be
disregarded (see, CSY 2005, Table 6-2).
7 The term “construction” used in Chinese yearbooks is a potential source of confusion. It does not
refer to construction activities, or the construction sector, but rather to the creation of fixed assets in
general.
8 Compared to the average for the total economy, state-owned units normally have higher proportions
of TIFA investment that will not be turned into productive fixed assets (DSIFA, 2002, p.77). 9 Like the stock of infrastructure, the residential capital stock is part of the productive capital stock at
the level of the total economy. It is not part of the productive capital stock from the perspective of the
industrial sector.
Chapter 5
70
- Newly Increased Fixed Assets (NIFA). NIFA is defined in the China Statistical
Yearbooks as "the newly increased value of fixed assets, constructed or purchased,
that have been transferred to the investors". This concept is narrower than TIFA
because investment activities that are not transformed into fixed assets are excluded.
Hence NIFA is an useful concept according to the SNA framework. One should note
that NIFA still includes the non-productive part of investment in fixed assets. NIFA is
published in statistical yearbooks of fixed assets investment in China since 1981.
From 1952-1980, the NIFA data are only available for investment in basic
construction in state-owned units (for an explanation of basic construction see section
4).
-The Rate of Projects of Fixed Assets Completed and Put into Operation. This
concept refers to “the ratio of the newly increased fixed assets to the total investment
made in the same period" (China Statistical Yearbook 2005, p.252). On first sight this
ratio could be used to calculate NIFA from the "Total investment in fixed assets".
However, in practice, it is based on incomparable data. The realization of fixed assets
in the current year resulting from the investment undertaken some years ago is
expressed as a percentage of the total investment in the current year. Given that it
might take quite some years for an investment to result in fixed assets, it is misleading
to apply this ratio to estimate the newly increased fixed assets (NIFA) from total
current investment (TIFA).
- The Original Value of Fixed Assets (OFA). OFA represents the stock of fixed
assets valued at their historical acquisition prices. Hence, the OFA of total fixed assets
is a cumulated value of assets purchased in different years at different prices. Using
historic valuation results in a stock of assets valued at a mixture of prices. Therefore,
the OFA data published in statistical yearbooks cannot be used directly to estimate
gross capital stocks, However, the difference between OFA in two subsequent years
can be used to derive the annual investment figures in the intervening period, as has
been done by Chen et al., (1988) (See section 5 of this chapter for a more detailed
explanation of this method). Data on OFA in industry are available for 1952, 1957
and from 1963 onwards.
Regional Capital Inputs
71
- The Net Value of Fixed Assets (NFA): NFA is the value of OFA minus cumulative
depreciation.
∑=
−=t
i
itt ondepreciatiOFANFA0
(13)
The difference between OFA and NFA is equal to depreciation. Unfortunately,
Chinese statistical yearbooks do not provide any information about the depreciation
rates used to derive the depreciation figures. Furthermore, the use of depreciation
rather than decay implies wealth accounting, rather than production analysis. It is not
clear for which years published NFA data are available.
- Accumulation of Fixed Assets (AFA). Accumulation of fixed assets refers to "the
value of the increased fixed assets (including the value of major repairs) in a certain
period minus the values of basic depreciation and major repair fund of the fixed
assets". This concept is found in the older statistical series prior to 1993 based on the
Material Product System. One way of calculating AFA is by deducting the net value
of fixed assets (NFA) at the beginning of the accounting period from the net value of
fixed assets (NFA) at the end of the accounting period. The other way is to subtract
the values of basic depreciation and major repairs of fixed assets from the value of the
newly increased fixed assets (NIFA) (i.e. the actual investment in fixed assets minus
the costs that do not increase the value of fixed assets, see DSIFA, 1997, p.451). Time
series of AFA are available from 1952 till 1992). Accumulation of fixed assets is also
a wealth accounting concept.
5.4 The Structure of Total Investment (TIFA) and Newly Increased
Fixed Assets (NIFA)
This section provides an analysis of the different ways in which investment can be
broken down into subcategories. The analysis serves as the analytic background for
our discussion of existing Chinese capital stock estimates in section five and for our
new estimates of regional capital services inputs in industry in section 6.
Chapter 5
72
5.4.1 Types of TIFA10
Total investment in fixed assets (TIFA) includes four types of investment:
1) Investment in basic construction11
2) Investment in technical renovation
3) Investment in real estate development
4) Other investment
Investment in basic construction refers to "the new construction projects or extension
projects and the related work of the enterprises, institutions or administrative units
mainly for the purpose of expanding production capacity or improving project
efficiency covering only projects each with a total investment of 500,000 yuan and
over".12
Investment in technical renovation refers to "the renewal of fixed assets and
technological innovation of the original facilities by the enterprises and institutions as
well as the corresponding supplementary projects and the related work (excluding
major overhaul and maintenance projects) covering only projects each with a total
investment of 500,000 yuan and over".13
Investment in real estate development refers to "the investment by real estate
development companies, commercial building construction companies and other real
estate development units of various types of ownership in the construction of
buildings, such as residential buildings, factory buildings, warehouses, hotels,
guesthouses, holiday villages, office buildings, and the complementary service
10 There are two different breakdowns of TIFA in Chinese statistics: breakdown into different types of
investment (section 4.1) and breakdown into different content categories (section 4.2). The terms type
and content have been introduced by us, to avoid confusion between the two breakdowns. They are not
found explicitly in Chinese statistical sources. We use the term "types of investment" to refer to the
breakdown into categories such as basic construction, technical renovation, real estate development and
other investment. We use the term "content of investment" to break down investment into substantive
categories such as fixed structures, machinery and equipment and other investment. For instance, Chen
et al. (1988) distinguish three types of investment and four content categories. 11 In some yearbooks, basic construction is also referred to as capital construction.
12 The definitions for these four categories are from CSY, 2000, p. 233. The coverage was all
enterprises with more than 50,000 yuan in investment prior to 1996. 13 Technical renovation is sometimes also referred to as innovation which is not the most appropriate
term, or in some publications as technical updates and transformation.
Regional Capital Inputs
73
facilities and land development projects, such as roads, water supply, water drainage,
power supply, heating, telecommunications, land levelling and other projects of
infrastructure. It excludes the activities in pure land transactions". Unfortunately, the
two investment types, investment in basic construction and investment in real estate
development, are not mutually exclusive. Basic construction also includes some
investment in non-residential fixed structures.
Other investment in fixed assets-
According to the China Statistical Yearbook 2000 (p. 234), this category includes:
A) The following projects of the state-owned units with the total planned (or actually
needed) investment of 500,000 yuan and over, which are not included in the plan of
capital construction (i.e. type 1) and the plan of innovation (i.e. type 2): (1) projects
of oil fields maintenance and exploitation with the oil fields maintenance funds and
petroleum development funds; (2) opening and extending projects with the
maintenance funds in coal, ore and other mining enterprises and logging enterprises;
(3) project of reconstruction of the original highways and bridges with the highway
maintenance funds in the department of communication; (4) projects of construction
of warehouses with the funds of simple construction in the commercial department.
B) Investment in fixed assets by urban collective units. This refers to: projects of
construction and purchases of fixed assets with the planned total investment of
500,000 yuan and over by all collective units in cities and county towns and in
townships which are approved by the State Council or provincial governments,
excluding investment by collective units under township enterprise administration
offices.
C) The projects of construction and purchases of fixed assets by the enterprises,
institutions or individuals other than those mentioned above with total investment of
500,000 yuan and over, which are not included in the plan of capital construction and
the plan of innovation.
Thus, other investment is a mixed residual category which includes investment in
exploitation of natural resources, investment in infrastructure, investment in non-
Chapter 5
74
residential fixed structures as well as other investments which are not included in
basic construction or technical renovation.
Before 1980, the third and fourth types of investment (real estate development and
other) were included in the first two (basic construction and technical renovation).
Data for real estate development are available since 1986, data on other investment
since 1985. The complete breakdown into four types is only available since 1986. For
the years between 1980 and 1986, we can reconstruct a residual category of "real
estate plus other investment" by deducting basic construction and technical renovation
from total investment. In Figure 5.1, we merge the data for real estate development
and other after 1986, to get a consistent breakdown into three types – basic
construction, technical renovation, real estate development plus other - for the whole
period 1980-2003.
Figure 5.1: Total Investment in Fixed Assets by Type of Investment,
Total Economy, 1980-2003
0%
10%
20%
30%
40%
50%
60%
70%
1980
1982
1984
1986
1988
1990
1992
1994
1996
1997
1999
2001
2003
year
percentage
basic construction technical renovation real estate development and others
Source: CSY2004, Table 6-4, Table 6-6, CSY2002, Table 6-6, and DSIFA 1997, pp20, pp.71.
The share of basic construction in TIFA decreases slightly between 1981 and 1990,
while the share of other investment increases substantially. After that the shares
remain stable. Real estate development is a special investment category in China,
which became more important from the early 1980s onwards. Before the 1980s,
housing investment was included in the basic construction category, which was
carried out by normal production companies or organizations. In the process of
Regional Capital Inputs
75
enterprise reform (zhufang zhidu gaige), investment in residential fixed structures (i.e.
housing) was transferred to real estate companies, for which a separate statistical
category was created.
The investment in real estate development as published in recent Chinese Statistical
Yearbooks mainly consists of four component parts: residential buildings, office
buildings, housing for business use, and others. The productive part of real estate
development (investment in office buildings and housing for business use) is rather
small, as showed in the table below. This implies that most of investment in non-
residential fixed structures will still be found in the category "basic construction".
Table 5.2: Breakdown of Investment in Real Estate Development,
Total Economy, 1997-2003
Total investment
(100 mill. yuan) Percentage shares of
Productive
Investment (%)
residential
buildings
(1)
office
buildings
(2)
houses for
business
(3)
other
use
(4)
Col. 2 +3 as % of
total
(5)
1997 3178.37 48.4 12.2 13.4 25.9 25.6
1998 3614.23 57.6 12.0 13.2 17.2 25.2
1999 4103.20 64.3 8.3 11.8 15.6 20.1
2000 4984.05 66.5 6.0 11.6 15.9 17.6
2001 6344.11 66.5 4.9 11.9 16.8 16.8
2002 7790.92 67.1 4.9 12.0 16.0 16.9
2003 10153.80 66.7 5.0 12.8 15.4 17.8
Source: CSY 2004, Table 6-44.
Between 1953 and 1980, the officially published time series for basic construction and
technical renovation for the total economy only cover the state-owned units (CSY,
2002, Table 6-6; CSY, 2004, Table 6-6 & DSIFA, 1997, p. 20, p.71). They exclude
investment by non-state firms, such as e.g. collectively owned enterprises.
5.4.2 Breakdown of TIFA by Content of Investment9
By content, all categories of total investment (TIFA) can be classified into three
categories:
1) Investment in Fixed Structures, referred to as "construction and installation"14
14 Construction and installation represents various investments in houses, buildings and foundations,
etc. It has a different meaning from the term basic construction discussed in section 4.1 of this chapter.
Chapter 5
76
2) Investment in Machinery and Equipment, referred to as the "purchase of equipment
and instruments"15.
3) Other investment.
Figure 5.2: Total Investment in Fixed Assets by Content Category,
Total Economy, 1981-2004
0%
10%
20%
30%
40%
50%
60%
70%
80%
1981
1983
1985
1987
1989
1991
1993
1995
1996
1998
2000
2002
2004
year
percent
construction and installation purchase of equipment and instruments others
Source: DSIFA 1997, pp. 26-27, and CSY 2005, Table 6-2.
The share of construction and installation in TIFA decreased by more than 10
percentage points from 1981 to 2004, i.e. from around 70 to 60 per cent. The share of
other investment increased from less than 5 per cent to more than 15 per cent.
We are not only interested in the aggregate proportions of the three content categories
in TIFA. We are also interested in the proportions within each of the four types of
investment distinguished in section 4.1. In the published yearbooks, only two of the
According to the CSY, 2000, p. 235 "Construction and installation" refers to the construction of various
houses and buildings and installation of various kinds of equipment and instruments, including
construction of various houses, equipment foundations and industrial kilns and stoves, preparation
works for project construction, and clearing up works post-project construction, pavement of railways
and roads, drilling of mines and putting up of oil pipes, construction of projects of water conservancy,
construction of underground air-raid shelters and construction of other special projects, installation of
various machinery equipment, testing operation for pre-testing the quality of installation projects. It is
the Chinese equivalent of the investment in fixed structures. The value of equipment installed is not
included in the value of installation projects. Equipment belongs to the investment in machinery and
equipment.
15 Purchase of equipment and instruments refers to the total value of equipment, tools, and vessels
purchased or self-produced which meet the standards for fixed assets. Equipment, tools and vessels
purchased or self-produced for new workshops by newly established or expanded units are categorized
as "purchase of equipment and instruments", no matter whether they come up to the standards for fixed
assets or not (from CSY, 2000, p. 235).
Regional Capital Inputs
77
types of investment - basic construction and technical renovation – are broken down
by content. There is no breakdown for real estate development or other investment.
Table 5.3 indicates what breakdown is available in the published sources and how
some of the gaps in the data can be filled.
Table 5.3: Breakdown of Investment Types by Content Categories Type of investment
Basic
construction
(1)
Technical
renovation
(2)
Real estate development
(3)
Other
(4)
Total
(5)
Construction
and installation
Published Published Assumption that all real
estate development
belongs to construction
and installation
Calculated
as residual
Published
Purchase of
equipment and
instruments
Published Published Calculated
as residual
Published
Content of investm
ent
Other expenses Published Published Calculated
as residual
Published
Total Published Published Published Published Published
We may safely assume that investment in real estate development can be classified
fully as construction and installation, since it involves only housing or office
construction. As we know the totals for each content category (see Figure 5.2), we can
thus subtract basic construction (col 1), technical renovation (col 2) and real estate
development (col 3) from total investment within each content category. The residual
equals "other investment" as indicated in column 4. Thus, we can derive a full
crosstabulation of types of investment and content categories of investment.
This method has been applied in Table 5.4. In each content category, the residual is
calculated by deducting the published categories from the content totals. The residuals
equal investment type "other investment". Column XIV represents the sum of those
three residuals. The figures in this column exactly equal the published data for the
fourth type category, other investment (from CSY, 2004, Table 6-6). Hence, our
method provides us with a reliable breakdown of the type category other investment
by content of investment (columns V, IX and XIII of Table 5.4, see also column (4) of
Table 5.3).
T
able
5.4
: Con
tent
of
Inve
stm
ent
by T
ype
of I
nves
tmen
t
C
onst
ruct
ion
and
Inst
alla
tion
P
urch
ase
of M
achi
nery
and
equ
ipm
ent
Oth
er I
nves
tmen
t su
m o
f 3
resi
dual
s
Tot
al
from
BC
fr
om T
R
tota
l RE
re
sidu
al 1
T
otal
fr
om B
C
from
TR
re
sidu
al 2
T
otal
fr
om B
C
from
TR
re
sidu
al 3
(I
) (I
I)
(III
) (I
V)
(V)
(VI)
(V
II)
(VII
I)
(IX
) (X
) (X
I)
(XII
) (X
III)
(X
IV)
1978
300.
85
165.
78
34
.36
19
79
1980
381.
07
74.2
1
136.
53
59.6
5
41
.29
3.52
19
81
689.
83
223.
64
47
.54
1982
87
1.12
29
1.41
67.8
7
19
83
993.
32
358.
31
78
.43
1984
12
17.5
8
50
9.23
106.
06
1985
16
55.4
6 72
6.71
19
6.23
732.
52
718.
08
217.
39
224.
94
275.
75
169.
65
130.
27
27.9
7 11
.41
1019
.68
1986
20
59.6
6 77
0.6
267.
8 10
0.96
92
0.30
85
1.95
26
0.34
30
8.58
28
3.03
20
8.99
14
5.17
42
.83
20.9
9 12
24.3
2 19
87
2475
.65
856.
76
349.
18
149.
88
1119
.83
1038
.78
325.
19
353.
29
360.
30
277.
26
161.
15
56.1
1 60
.00
1540
.13
1988
30
99.6
6 10
10.1
5 47
7.76
25
7.23
13
54.5
2 13
05.3
7 37
2.61
43
0.87
50
1.89
34
8.77
19
1.55
71
.92
85.3
0 19
41.7
1
1989
29
94.5
9 99
8.73
37
7.25
27
2.65
13
45.9
6 11
15.3
1 38
0.94
35
5.89
37
8.48
30
0.00
17
2.07
55
.64
72.2
9 17
96.7
3 19
90
3008
.72
1045
.37
372.
91
253.
25
1337
.19
1165
.54
453.
76
397.
36
314.
42
342.
74
204.
69
59.9
2 78
.13
1729
.74
1991
36
47.6
8 13
08.8
3 42
6.33
33
6.16
15
76.3
6 14
60.1
9 52
1.22
51
3.35
42
5.62
48
6.63
28
5.76
83
.54
117.
33
2119
.31
1992
51
63.3
7 18
89.3
9 62
0.64
73
1.2
1922
.14
2125
.14
667.
34
715.
34
742.
46
791.
58
455.
92
125.
12
210.
54
2875
.14
1993
82
01.2
1 30
18.7
4 94
5.27
19
37.5
1 22
99.6
9 33
15.9
2 89
9.55
10
70.9
3 13
45.4
4 15
55.1
8 69
7.22
17
9.65
67
8.31
43
23.4
4 19
94
1078
6.52
41
23.8
9 12
58.7
2 25
54.0
8 28
49.8
3 43
28.2
6 14
02.8
4 14
19.7
9 15
05.6
3 19
28.0
8 91
0.01
24
0.09
77
7.98
51
33.4
4
1995
13
173.
33
4641
.13
1343
.62
3149
.02
4039
.56
4262
.46
1635
.04
1682
.2
945.
22
2583
.48
1127
.44
273.
53
1182
.51
6167
.29
1996
15
153.
41
5345
.27
1396
.77
3216
.4
5194
.97
4940
.79
1861
.15
1900
.5
1179
.14
2879
.83
1404
.42
325.
47
1149
.94
7524
.05
1997
15
614.
03
6215
.22
1540
.77
3178
.37
4679
.67
6044
.84
2060
.6
2033
.74
1950
.50
3282
.25
1641
.2
347.
43
1293
.62
7923
.79
1998
17
874.
53
7695
.75
1681
.38
3614
.23
4883
.17
6528
.53
2101
.83
2445
.24
1981
.46
4003
.10
2118
.84
390.
13
1494
.13
8358
.76
1999
18
795.
93
8543
.598
16
67.6
1 41
03.2
44
81.5
2 70
53.0
4 21
32.2
97
2465
.7
2455
.04
4005
.74
1779
.389
35
1.76
7 18
74.5
8 88
11.1
4 20
00
2053
6.26
89
36.8
11
1943
.25
4984
.05
4672
.15
7785
.62
2457
.876
27
76.4
1 25
51.3
3 45
95.8
5 20
32.5
86
387.
93
2175
.33
9398
.81
2001
22
954.
90
1015
4.63
22
06.0
6 63
44.1
1 42
50.1
0 88
33.8
0 24
73.2
93
3297
.88
3062
.62
5424
.80
2192
.176
41
9.82
28
12.8
0 10
125.
53
2002
26
578.
90
1186
5.82
25
98.0
3 77
90.9
2 43
24.1
3 98
84.5
0 27
80.1
8 36
35.3
2 34
69.0
0 70
36.6
0 30
20.6
23
517.
193
3498
.78
1129
1.91
20
03
3344
7.20
15
426.
44
3420
.13
1015
3.8
4446
.83
1268
1.90
34
95.7
77
4460
.51
4725
.62
9437
.50
3986
.383
74
4.22
3 47
06.8
9 13
879.
34
Not
e: B
C:
basi
c co
nstr
ucti
on,
TR
: T
echn
olog
ical
Ren
ovat
ion,
RE
Rea
l E
stat
e D
evel
opm
ent:
100
mil
l yu
an a
t cu
rren
t pr
ices
. (V
)=(I
)-(I
I)-(
III)
-(IV
); (
IX)=
(VI)
-(V
II)-
(VII
I);
(XII
I)=
(X)-
(XI)
-(X
II);
XIV
=(V
)+(I
X)+
(XII
I).
S
ourc
e: D
SIF
A, 2
002,
p.2
88; C
SY, 2
005,
p.1
86; C
SY, 2
004,
Tab
le 6
-8 a
nd T
able
6-2
1.
Chapter 5
78
Regional Capital Inputs
79
Table 5.4 provides us with the shares of the three content categories within the other
investment type category. Figures 5.3, 5.4 and 5.5 present the breakdown of the
different investment types by the three content categories. No separate figure is
included for real estate development as this only consists of construction and
installation investment.
The most detailed published data are available for basic construction, for which the
series can be traced back to 1950. A breakdown for technical renovation is available
since 1980. For other investment we have estimated the breakdown for the period
since 1985.
Figure 5.3: Total Investment in Basic Construction by Content Category,
1950-2003
0%
10%
20%
30%
40%
50%
60%
70%
80%
1950
1953
1956
1959
1962
1965
1968
1971
1974
1977
1980
1983
1986
1989
1992
1995
1998
2001
year
percentage
construction and installation purchase of equipment and instrument others
Source: DSIFA, 1997, pp.97; and CSY 2004, Table 6-8.
Chapter 5
80
Figure 5.4 Total Investment in Technical Renovation by Content Category,
1980-2003
0%
10%
20%
30%
40%
50%
60%
70%
year
percentage
construction and installation purchase of equipment and instrument others
Source: DSIFA, 1997, PP. 249; and CSY 2004, Table 6-21.
Figure 5.5: Other investment by Content Category, 1985-2003
0%
10%
20%
30%
40%
50%
60%
70%
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
year
percentage
construction and installation purchase of equipment and instruments others
Source: DSIFA, 2002, pp. 28; CSY 2005, pp.186; and own calculations.
The decline in the aggregate share of fixed structures (construction and installation)
since 1981, as documented in Figure 5.2, primarily takes place in the technical
renovation and other investment types. In basic construction, there is not all that much
change in the share of fixed structures.
For machinery and equipment (purchase of equipment and instruments), the main
changes are found in basic construction and technical renovation. In basic
construction the share of machinery and equipment declines after 1980, in technical
Regional Capital Inputs
81
renovation it increases. On balance this results in the more or less stable share of
machinery and equipment in Figure 5.2.
This analysis of types and content categories will be useful, when we try to break
down investment by content in section 6.
5.4.3 Newly Increased Fixed Assets (NIFA)
As indicated in section 3, total investment in fixed assets (TIFA) is broader than the
real investment in the formation of the capital stock. The more appropriate concept is
newly increased fixed assets (NIFA). NIFA shares the same type classification as
TIFA, i.e. basic construction, technical renovation, real estate development and others.
Figure 5.6: Structure of Newly Increased Fixed Assets, 1981-2003
0%
10%
20%
30%
40%
50%
60%
1981
1983
1985
1987
1989
1991
1993
1995
1997
1999
2001
2003
year
percentage
basic construction technical renovation real estate+others
Source: DSIFA, 1997, p. 62; DSIFA, 2002, p. 77; CSY2004, Table 6-14, 15, and 23.
5.5 Discussion of Chinese Capital Estimates in the Literature
In this section, we discuss literature on Chinese capital input estimates in the light of
the theoretical, empirical and conceptual issues discussed in paragraphs 2, 3 and 4.
Chapter 5
82
5.5.1 Productive Capital versus Wealth Accounting
In section 2.1, we discussed the differences between the concepts of “productive
capital services” and "productive capital stock" used in productivity analysis and the
concept of "wealth capital stock" used in wealth accounting ( OECD, 2001a, b, p.53).
One of the shortcomings of many earlier estimates of Chinese capital inputs is that
they tend to use wealth accounting capital stock concepts and apply them in TFP
analysis, for which they are not really appropriate (Chow, 1993; Hsueh and Li, 1999,
Wu, 2000; Jefferson et al., 2000; Wang and Yao, 2003; Holz, 2006). Wu (2004)
constructed Chinese regional capital stock 1953-2000, by introducing a "backcasting
approach" to substitute the use of initial value of capital stock, but his general
framework is still that of wealth and accounting.
Huang et al. (2002) are among the few researchers who give explicit consideration to
the difference between depreciation and the efficiency decline of fixed assets (called
"replacement" in their paper). Sun and Ren (2007) estimate the flow of capital service
using capital service prices, and construct the capital input indices by industry in
China (1980-2000). This is one of the most consistent efforts to create a capital input
index for China which is consistent with the SNA framework. Compared to the
present chapter, less attention is paid to nature and coverage of the investment data
and there are no attempts to estimate the capital inputs at a regional level.
5.5.2 Choice of Investment Concepts
• TIFA versus NIFA
The published investment figures in Chinese official reports or yearbooks are usually
the total investment in fixed assets (TIFA).16 For instance, Hsueh and Li (1999),
Wang and Yao (2003) and Huang, Ren and Liu (2002) use TIFA to construct the
gross capital stock, which would overestimate the final size of the capital stock. Our
preferred concept is NIFA.
16 As mentioned in the former section, TIFA is not the real investment in fixed assets. Not all
investment is transformed into productive assets, which is better denoted by the term NIFA.
Regional Capital Inputs
83
• Accumulation of fixed assets (AFA) versus NIFA
The data on accumulation of fixed assets consist of fixed assets and circulating funds.
The accumulation of fixed assets excluding circulating funds is the part needed for
estimating the capital stock (see Chow, 1993, p. 816-817). In principle, the productive
part of accumulation of fixed assets is equal to the productive part of investment in
newly increased fixed assets (NIFA) minus depreciation. Chow (1993) uses the
accumulation of fixed assets variable to derive a series for newly increased fixed
assets.
Holz (2006, p. 143) states that the published data on accumulation of fixed assets (for
instance, used by Chow, 1993, 1994), seem to have a zero depreciation rate. To check
whether this is indeed the case we put together a table comparing AFA and NIFA for
years in which both figures are available.
Table 5.5: Comparison of Newly Increased Fixed Assets and Accumulation of
Fixed Assets
Accumulation of
Fixed Assets (AFA)
Newly Increased Fixed
Assets (NIFA)
Rate of
depreciation and
major repair
Total
(1)
#Productive
(2)
total
(3)
#Productive
(4) [(4)-(2)]/(4)
1981 778 393 824.53 473.28 16.96%
1982 969 487 992.47 569.68 14.51%
1983 1125 586 1187.23 681.47 14.01%
1984 1453 829 1490.96 855.81 3.13%
1985 1883 1156 1950.03 1119.32 -3.28%
1986 2196 1350 2633.52 1767.09 23.60%
1987 2718 1690 3100.73 2080.59 18.77%
1988 3360 2012 3808.64 2555.60 21.27%
1989 2835 1701 3758.43 2521.91 32.55%
1990 3008 1685 3995.34 2680.87 37.15%
1991 3768 2176 4649.8 3110.72 30.05%
Note: at current prices, 100 million yuan.
Source: AFA is from CSY1992, P.40; and CSY1993, p.43; and NIFA is from CISFA, 1950-1995, p.
10.
According to the concepts discussed in CISFA 1950-1995 (page 451), the difference
between productive NIFA (col. 4) and the accumulation of fixed assets (col 2) should
be equal to basic depreciation and the major repair fund in fixed assets in that year.
The comparison of accumulation (col.2) and productive NIFA (col.4) shows that the
depreciation and major repair as a percentage of fixed assets is greater in later years
Chapter 5
84
than in earlier ones. Though we cannot disentangle major repairs and depreciation,
Chow does not seem to have used a zero depreciation rate. However, the increase in
the rate of depreciation may suggest that Chow's investment figures for the early years
are underestimated and for the later years overestimated.
• OFA versus NIFA
In Chen et al. (1988, p. 244), the original value of fixed assets at year t is stated as
previous year's original value of fixed assets plus the newly increased fixed assets in
the current year.17 Direct NIFA data are not yet published prior to 1981. Chen et al.
estimate newly increased fixed assets in each year by IN(t)= OFA(t) - OFA(t-1).18 The
same method is used by Jefferson et al. (1992, p.261, equation A4; 1996, p.174; 2000).
A problem with the Chen et al. estimates is that they do not discard assets at the end
of their life times.
5.5.3 Is the Neglect of Scrap Values a Problem?
The scrap value is the value of an asset discarded at the end of its service life. Some
researchers neglect it because they think the scrap value is a negligible proportion of
the total capital stock. Disregarding scrap values simplifies the capital calculations,
but may introduce a bias in the estimates.
The equation used by Chen et al. (1988) disregards scrap values. Scrap values are also
neglected in the publications of Jefferson et al. (1992, p.261, equation A4;1996, p.174;
2000). Like Chen et al., these authors obtain investment through deducting the
original value of fixed assets (OFA) in year t-1 from OFA of year t, disregarding
scrap values.
In theory, the difference of original value of fixed assets in two continuous years, is
equal to the NIFA minus the scrap value in that year, i.e. OFAt – OFAt-1= NIFAt-
17 Chen et al. (1988) use the term "newly-commissioned fixed assets" which has the same meaning as
NIFA. 18 Chen et al. (1988), write that KFO(t)=KFO(t-1)+I(t). In order to keep those concepts consistent with
others in this chapter, we rewrite this equation using OFA and IN instead of KFO and I respectively.
Regional Capital Inputs
85
Scrapt We can get scrap values for each year by the following equation: Scrapt =
NIFA –( OFAt - OFAt-1)
Holz (2006, p. 148) includes the scrap value in the equation IN(t)= OFA(t)- OFA(t-1),
as follows OFA(t)- OFA(t-1) = IN(t) – scrap value(t). He criticises Chen et al. (1988)
for disregarding scrap values. In this section, we argue that Holz overestimates the
significance of the scrap value in his criticism of Chen.
If our purpose is to construct the gross capital stock, the scrap value has to be
deducted from investment before deriving the gross capital stock, as follows:
Disregarding price deflation, the gross capital stock, can be estimated by
∑=
−+=t
i
iiGt scrapINOFAK
1
0 )( (14)
where G
tK is gross capital stock at time t; OFA0 is the initial gross capital stock, which
is approximated as the original value of fixed assets in 1952; iIN is (real) investment
at time i.
If a price index is involved, the investment should be deflated at year-i prices, while
the scrap value should be deflated to a price at T (service life of fixed assets) years
earlier than i. Then we have
∑= −
−+=t
i Ti
i
i
iGt
p
scrap
p
INOFAK
1
0 )( (15)
Given that investment can be estimated from OFA and the scrap value.
1−−=− tttt OFAOFAscrapIN (16)
If there is no price deflator influence, we shouldn't make an adjustment for the scrap
value at all in constructing the gross capital stock, not because scrap value is very
small, but it is already incorporated when using OFA to estimate investment.
))(()(
1
10
1
0 ∑∑=
−=
−+−+=−+=t
i
tttt
t
i
iiGt scrapscrapOFAOFAOFAscrapINOFAK (17)
Combining with price deflation, we have (see also Holz, 2006, p.151)
))11
((
1
10 ∑
= −
− −+−
+=t
i Tiii
i
iiGt
ppscrap
p
OFAOFAOFAK (18)
Chapter 5
86
The last item on the right hand of above equation shows the effect of the price index.
It is a T-year lagged price influence on the scrap value. For instance, we use the price
index of 2004 and 1991, which is Tii pp −
−11
=1/1.9677-1= -0.4918 . This means that
neglect of the scrap value only leaves out only half of the scrap value. Therefore, the
neglect of the scrap value by Chen et al (1988) is less of a problem than suggested by
Holz. The scrap value is only around 3-5% of OFA t-1. It is not necessary to make
adjustments for such a modest figure in the approximate estimation of capital services.
Another point worth making is that the scrap value is more important in the
calculation of the gross capital stock than the net capital stock or the capital service.
For instance, if one uses the gross capital stock (e.g. Holz, 2006), the scrap value will
be the original price of a fixed asset, which is normally not negligible. However, if we
use net capital stock or capital service series, after deducting for the decay of a fixed
asset, the residual part will be very small at the end of its service life. Thus the scrap
value will not make all that much difference.
5.5.4 Gross or Net Fixed Assets?
Besides the confusion between productive capital service and wealth capital stock,
there is a controversy about the use of gross or net fixed asset concepts. This has to do
with the choice of the decay pattern of fixed assets. The use of a gross capital stock
concept assumes that there is no decay of productive capacity during the life time of
an asset.
Holz (2006) argues that net fixed assets should not be used as the capital stock in the
production function. Instead, he argues in favour of measuring fixed assets at the
original purchasing value of all fixed assets. He says "...the appropriate fixed asset
measure is a count of the fixed assets used during the production period. ... Even a
machine that is completely written off is included in the account "original value of
fixed assets", at its purchasing prices, as long as it is still in use; as long as the
machine is still in use, it is likely to potentially operate at the same capacity as at its
Regional Capital Inputs
87
purchasing data" (Holz, 2006, p.144-145). Thus, Holz opts for the one-hoss-shay
pattern where there is no productivity decline during the life time of an asset.
We disagree with this choice for two reasons. Admittedly there are certain types of
assets that may "contribute as much to production as a new machine of the same
quality" as in the case of the computer example (p.144). However, most of other
assets truly deteriorate and age over time during the production process. Therefore,
the applicability of the one-hoss-shay efficiency pattern (assets contribute fully as new
ones as long as they are still in use) is limited to very few fixed assets. OECD (2001b,
p.62) shows that using the gross capital stock in productivity analysis generally results
in over-estimation of the volume of capital services.
Next, Holz seems to confuse the concept of depreciation from wealth accounting with
the concept of productive decay. It is correct that sometimes a machine is written off
by a certain depreciation method in the "balance sheet" while it might be still in use in
the production. But, this observation is mainly based on a business account concept of
depreciation. The meaning of depreciation in business accounts - "allocating the costs
of past expenditures on fixed assets over subsequent accounting periods" - is different
from the one used in SNA, which refers to the decay of capital services, as discussed
in the first part of this chapter.
5.5.5 Non-Productive Fixed Investment
It is important to distinguish between productive and non-productive investment.
Within industry and manufacturing, the non-productive part of fixed assets in industry
includes the residential housing stock, but also other non-productive investment such
as investment in infrastructure. According to most China statistical yearbooks, the
non-productive part of gross investment is 30% of all TIFA in the total economy,
while the residential part accounts for little more than 10% of TIFA. If we only
exclude investment in the residential capital stock from the total investment in fixed
assets in industry, this will result in overestimation of the productive capital stock in
industry.
Chapter 5
88
To create a series of NIFA in industry, Chow et al. (1988) and Wu and Xu (2002) use
original fixed assets (OFA) data from the published yearbooks (see discussion in Wu
and Xu, 2002, p.16).
1−−= ttt OFAOFANIFA (19)
Using OFA to create NIFA, following Chen et al. (1988), solves a practical problem
of the lack of direct data on industrial investment in fixed assets prior to 1981.19 In
estimating the productive (or efficient) NIFA, Chen et al. (1988) deduct the residential
housing part from total NIFA. However, this is not an adequate solution because the
non-productive part of capital stock includes a variety of other non-productive assets
along with the residential housing stock.
The classification of investment into basic construction and technical renovation only
in Chen et al (1988, p. 260, Table A2) is also somewhat misleading. Figure 5.6 shows
that investment in real estate development plus other investment is a very substantial
proportion of total NIFA, though the productive part of the real estate category may
be small.
Holz includes non-productive fixed assets in the total capital stock in his "economy-
wide output" analysis (Holz, 2006, p.145). For the total economy, this is less
problematic than for the industrial sector. From the perspective of the total economy,
investment in residential fixed structures and investment in infrastructure are
productive investments.
Non-productive fixed assets, as part of consumption material, are the infrastructures in
residential buildings, schools, hospital and other welfare structures.20
5.5.6 Types of Investment
Since 1986, total investment in fixed assets (TIFA) consists of four types, basic
construction, technical renovation (also called technical updates and transformation in
19 As explained in section 5.3, this method ignores scrap values. But we have argued that scrap values
are negligible and can be neglected in estimating investment. 20 See also from http://old.ynce.gov.cn/content.asp?ARTID=5575&COLID=174 .
Regional Capital Inputs
89
recent yearbooks), real estate development and others.21 However, the data on newly
increased fixed assets in published sources are available mostly only for two of the
categories: basic construction and technical renovation. The real estate development
category is usually neglected by researchers. Chen et al. (1988) distinguish only three
categories: basic construction, technical renovation and miscellaneous. In Figure 5.2,
we have made estimates for basic construction, technical renovation and other
(including real estate development).
5.5.7 Breakdown of Investment in Fixed assets into Different Content
Categories
Different kinds of fixed assets are not homogeneous. For reasons of simplicity, some
studies (Jefferson et al., 2000; Chow and Li, 2002), only consider one aggregate type
of investment. Although it is almost impossible to distinguish all different categories
of fixed assets, the use of an aggregated capital series may produce rather big errors
because of different types of fixed assets have different price deflators and the
different service lives.
Chen et al. (1988) decompose newly increased fixed assets into four content
categories: non-residential construction, equipment, housing and others. The
proportions of these categories within industry are not known. The authors use
proportions from the total economy to break down industrial investment into these
four categories of fixed assets within industry. They consider housing as non-
productive investment. But as we have argued above, the concept of non-productive
assets is broader than that of residential housing alone.
5.5.8 Revaluation Problems
After 1993, many fixed assets have been revalued. As a result, the original value of
fixed assets (OFA) is a mix of assets valued at their historical acquisition prices and
21 Before 1980, the category of others was included in the technical renovation category. Figures on
Real estate development are available from 1986 onwards. Before that year they were included in basic
construction.
Chapter 5
90
revalued assets. Therefore the published data on the original value of fixed assets and
cumulative depreciation have to be used with caution (for a good discussion, see Holz,
2006, p.145 and 148). Chen et al (1988) present a very good method to estimate
newly increased fixed assets (which is called investment in their paper) by deducting
the original values of assets (OFA) in two successive years. However, this method
cannot be directly used from the 1990s onward because the large revaluations of fixed
assets in the enterprises will result in overestimation of the annual investment figures.
5.6 New Estimates of Capital Service Inputs in China (Total
Economy, Industry and Manufacturing)
5.6.1 Introduction
In this section, we explain the choices we made in constructing indices of capital
service inputs in the Chinese economy, the industrial sector and the manufacturing
sector, in the light of the theoretical, empirical and conceptual discussions in the
previous sections. To estimate capital services in productivity analysis, we need the
following data: investment series (or gross fixed capital formation), the service lives
of fixed assets, decay coefficients, an estimate of the initial capital stock and price
indices.22 We estimate the gross capital stock from the accumulated investment series
and an initial capital stock, according to the Perpetual Inventory Method. We take
age-efficiency patterns into account. The resulting capital stock series is used as a
proxy for the index of capital service inputs.
5.6.2 Investment Series
5.6.2.1 Newly increased fixed assets (NIFA)
As explained in section 2 of this chapter, newly increased fixed assets (NIFA) is the
investment variable that is consistent with the SNA. Data on newly increased fixed
22 When the scrap value is taken into account, one also needs information about retirement patterns
(mortality functions) around the average service life. As explained in section 5.3 of this chapter, the
scrap value as percentage of investment is very low. We will disregard it.
Regional Capital Inputs
91
assets in the total economy are published for the period 1981-present. Prior to 1981,
we only have NIFA data on basic construction for the state-owned sector.
a. NIFA 1981-present, total economy
NIFA data are published from 1981 onwards. Using the proportions from TIFA, these
can be broken down by different types of investment. From 1981 onwards, the
statistical yearbooks provide data on two types: basic construction and technical
renovation in NIFA. From 1986, data are provided for all four types: basic
construction, technical renovation, real estate development and other.
Between 1981 and 1986, the difference between total NIFA and basic construction
and technical renovation equals real estate development plus other investment. Thus,
from 1981 onwards, we can reconstruct series of NIFA for three types of investment:
1. basic construction, 2. technical renovation, 3. real estate development and other
investment (see Figure 5.6 in section 5.4.3).
b. Reconstructing NIFA 1953-1980, total economy
From 1953 to 1980, the only data we have are on NIFA-basic construction in state-
owned units (NIFA-BAC-SOU). In this period real estate development was included
in the basic construction figures. For TIFA, we do have data on both basic
construction and technical renovation (CSY, 2004, Table 6-6; DSIFA, 1997, p. 20 and
p. 71). We can reconstruct total NIFA by applying the proportions of basic
construction and technical renovation from TIFA. Thus we derive estimates for total
NIFA in SOUs.
c. Coverage adjustments State-owned units – total economy 1953-1980
The NIFA estimates under (b) refer only to state-owned units. Thus we have to make
a coverage adjustment to get an estimate for the total economy, 1953-1990.
Using available information about the original value fixed assets (OFA) in the
industrial sector, we calculate the ratios of non-SOU/SOU OFA in industry (ratio-a)
for the period 1953-1999. From 1980 to 1996, we have time series of TIFA in state-
owned units as well as for the total economy including non-state owned units. Thus
we can calculate the ratio of non-SOU/SOU TIFA for all years between 1980 and
Chapter 5
92
1996 (ratio-b). We can compare the ratios a and b for the overlapping years 1980-
1996. Ratio-b was twice ratio-a in 1980. It increases to four times ratio a by 1985, due
to the decline of the share of the state owned sector.
Considering that the industry structure in China didn't change too much before 1980,
therefore we rely on the 1980 ratio of 2 to 1 to adjust the earlier ratios of non-
SOU/SOU upward. These adjusted ratios are used to make the coverage adjustment
non-state owned state-owned NIFA from 1953-1980.
d. Change in coverage in 1997
Prior to 1997, the published TIFA and NIFA series include investment of sums of 50
thousand yuan or more. After 1997, the coverage changes to investment of 500
thousand yuan and above. In order to maintain consistency in coverage for the whole
time series, we adjust the NIFA data before 1997 to a coverage of 500 thousand yuan
and above. We opt for leaving the more recent data unchanged, since recent
investment has a higher weight in PIM than investment of the earlier years.
The result of steps a, b, c and d is a time series of NIFA investment in the total
economy from 1953 till 2003.
5.6.2.2 Productive NIFA (P-NIFA) in the Total Economy
To be consistent with SNA concepts, the non-productive part (e.g. residential housing
and other non-productive investment) has to be deducted from NIFA. There are two
ways to derive the productive part of NIFA:
(1) NIFA data in four categories (basic construction, technical renovation, real estate
development and others) are multiplied by the shares of productive investment in
those four groups respectively, and then summed. This method requires detailed
information on shares of productive investment in each of the categories. For instance,
it is known that the productive ratio in real estate development is very low, about
17%-26% during 1997-2003. However, the shares for the other investment types of
NIFA are not known.
Regional Capital Inputs
93
(2) The ratio of productive to total investment for total NIFA is applied to each of the
investment types. The share of productive investment to total investment is published
once every five years from 1953 to 1995. After 1995, we use the 1995 ratio.
This results in a time series of P-NIFA in the total economy from 1953 to 2003.
5.6.2.3 Productive NIFA in Industry (P-NIFA-Industry)
The next step is to construct a series of P-NIFA in the industrial sector. In section 5.5
of this chapter, we have explained that the method used by Chen et. al. (1988) and Wu
and Xu (2002) will tend to overestimate productive investment, because it only
deducts housing investment from total NIFA. Our procedure is to derive productive
NIFA in industry from the time series of productive NIFA in the total economy. This
is done as follows:
a. For the period 1985-2003, we calculate the ratio of capital investment in industry
(investment in NIFA-basic construction plus NIFA technical renovation) to capital
investment in the total economy (investment in NIFA-basic construction plus
NIFA technical renovation).23
b. We apply this ratio to P-NIFA in the total economy and thus derive an estimate of
productive NIFA in the industrial sector for the period 1985-2003.24
c. We compare the P-NIFA-industry series with the series of NIFA derived by
deducting OFA(t-1) from OFA(t) for the period 1985-2003. The difference
between the two series provides an estimate of non-productive investment. The
average share of non-productive investment for the whole period is 12.6 %.25
23 Given that there are no NIFA data for the category "other" , which is actually only a small part in
productive -NIFA, our calculation is mainly based on the two major categories: investment in NIFA-
basic construction and NIFA -technical renovation. 24 An alternative would be to apply the ratio NIFA-industry/NIFA-total for investment in basic
construction + technical renovation to NIFA-total. But, this would lead to biased results because real
estate development (non-productive) has a rather big share in NIFA total. 25 This confirms that the use of an average ratio of residential to total fixed assets of 8.2% in Chen et al
(1988) results in an overestimation of the real productive NIFA in industry.
Chapter 5
94
d. We apply this ratio to the NIFA-industry data estimate from OFA for the period
1953-1984. This gives us an estimate of P-NIFA in industry between 1953 and
1984.
For manufacturing, we use the same method as for P-NIFA in industry. After
obtaining productive NIFA for the total economy, we apply the ratio of investment in
manufacturing to total investment for NIFA in the categories basic construction plus
technical renovation. However in the case of manufacturing, we can only estimate
productive NIFA from 1985 onwards. The OFA data used to estimate NIFA in
industry for the earlier years, are not available for the manufacturing sector.
Figure 5.7: Productive Newly Increased Fixed Assets in Total Economy, Industry
and Manufacturing, 1953-2003
0
5000
10000
15000
20000
25000
30000
19531955195719591961196319651967196919711973197519771979198119831985198719891991199319951997199920012003
Year
100 mill yuan
Total economy Industry Manufacturing
Note: at current prices. The coverage is 500 thousand yuan and above.
Source: Various China Statistical Yearbooks; DSIFA 1997 and 2002; and our own calculations.
5.6.2.4 Breakdown of Productive Investment in Industry into Non-residential
Fixed Structures, Machinery and Equipment and Other
We want break down the productive NIFA data in industry into three content
categories: non-residential fixed structures, machinery and equipment and other
investment. However, direct information for this is not available.
Regional Capital Inputs
95
As explained in section 4.2, we do have data on TIFA in the total economy for three
content categories of investment in fixed assets: construction and installation,
purchase of equipment and instruments, and other investment (see Figure 5.2).
Proportions from TIFA from the total economy can be used to make a breakdown of
productive NIFA in industry into three content categories.
In order to do this, we first have to make an adjustment for investment in residential
fixed structures. These are productive from the perspective of the total economy, but
non-productive from the perspective of the industrial sector.
In the total economy, we start by excluding real estate development which consists
mainly of non-productive investment. We also disregard the less important type
"other" about which we have insufficient information. We focus on the main types:
basic construction and technical renovation. For the years 1995-2003, each of these
can be broken down by content into construction and installation, purchase of
equipment and instruments and others, (CSY 1996-2004, see also figures 5.3 and 5.4).
In addition, we have data on investment in residential construction in both basic
construction and technical renovation. We use this information to calculate the shares
of non-residential fixed structures (FS), machinery and equipment (ME) and other
(OT) in TIFA as follows:
FS share in TIFA =ialTRresidentTRialBCresidentBC
ialTRresidentTRFSialBCresidentBCFS
−+−−+− )()( (20)
ME share in TIFA =ialTRresidentTRialBCresidentBC
TRMEBCME
−+−+ (21)
Other share in TIFA = ialTRresidentTRialBCresidentBC
TROTBCOT
−+−+ (22)
Unfortunately, from CSY 2005 onwards, the TIFA and the NIFA data are only
available for urban investment, so we cannot extend our series beyond 2003. The
results of this exercise are reproduced in Table 5.6. These proportions are
subsequently applied to the NIFA data for industry.
Chapter 5
96
Table 5.6: Proportions of Investment Categories in TIFA (%),
Total Economy, 1981-2003
Non-residential
construction and
installation
Machinery
and
equipment
Others
1981 59.2 33.6 7.1
1982 58.9 33.4 7.8
1983 56.9 35.3 7.7
1984 55.0 37.2 7.8
1985 53.3 37.8 8.9
1986 54.7 36.4 8.9
1987 53.8 36.5 9.7
1988 53.6 36.6 9.8
1989 56.0 34.7 9.3
1990 55.0 34.8 10.2
1991 53.4 35.0 11.6
1992 54.2 33.4 12.4
1993 52.9 32.0 15.0
1994 52.7 32.7 14.6
1995 55.2 27.9 16.9
1996 52.3 30.9 16.8
1997 52.2 29.7 18.2
1998 51.5 30.9 17.6
1999 51.1 30.7 18.1
2000 50.4 30.7 18.9
2001 49.7 29.4 20.9
2002 48.5 29.5 22.0
2003 49.2 30.3 20.5
2004 52.3 30.9 16.8
Note: Proportions are calculated from Basic construction and Technical renovation after the deduction
of Residential housing investment. The Residential housing construction data are only available from
1981-2000; we apply the average residential ratio in TIFA (1996-2000) to 2001-2004.
Source: Statistics on Investment in Fixed Assets of China, 1950-2000, p.30, and China Statistical
Yearbook, 2005, p.186.
We apply the proportions of Table 5.6 to P-NIFA in industry. This results in time
series of investment in machinery and equipment, non-residential fixed structures and
other assets in industry for the period 1981-2003.
We apply the average of the proportions 1981-2003, to break down P-NIFA for the
earlier period 1953-1980. The same procedure is followed for manufacturing.
5.6.3 Price Deflators
Price indices also play an important role in measuring the value of fixed assets at
constant prices, given that NIFA is available at acquisition prices. We apply specific
price indices for each of the three categories: construction and installation, purchase
Regional Capital Inputs
97
of machinery and equipment, and other investment for the period 1992-2004. For the
period 1953-1991, we have used the aggregate price index for fixed assets, as specific
deflators for the three investment categories were not available.
5.6.4 Initial Capital Stock
To assess the initial or benchmark level of the capital stock (e.g. year 1952 in this
chapter), PIM requires the use of a long time series of investment preceding the initial
year. Such series are unavailable. Therefore, we need to estimate the initial capital
stock by proxy methods (e.g. Huang et al., 2002).
Timmer (1999) has estimated initial capital stocks by applying the average of
incremental value added-output ratios in the initial years to total value added in the
initial year. Osada (1994) has used incremental capital-output ratios (ICORs) for this
purpose (Osada, 1994). The assumption underlying these procedures is that the
capital-output ratio is sufficiently stable, so that incremental capital output ratios
approximate the average capital-output ratios.
Another method to estimate the initial capital stock is to use the average growth rate
of investment and the depreciation rate (Reinsdorf and Cover, 2005). The initial
capital stock Vo can be expressed as
dg
gINV
++
⋅=1
00 (23)
where g is the average growth rate of investment before the initial year, and d is the
constant geometric rate of depreciation.
In the literature, various estimates of the benchmark capital stock have been made.
Based on Chow (1993, p. 822 and 823), Chow and Li (2002) estimate the initial stock
in the total economy at 221.3 billion yuan at the end of 1952, reaching 1411.2 billion
yuan by the end of 1978, Wang and Yao (2003) at 175 billion yuan. Chen et al. (1988)
have a 14.88 billion initial capital stock in industry in 1952.26 Jefferson, Rawski and
26 The original cost of fixed assets of independent accounting units within state-sector industry is taken
as the initial industrial capital stock.
Chapter 5
98
Zheng (1992) estimate a net value of productive fixed assets in industry in 1980 of
228.59 billion yuan (at 1980 prices).27
Applying the ICVAR method proposed by Timmer, we average the ICVARS for 1952
to 1957 and use these to calculate the initial capital stock. We find a capital stock in
1952 of 84.3 billion yuan (at 1952 prices) in the total economy and a capital stock of
19.9 billion yuan (at 1952 prices) in industry. This is much lower than Chow and Li
and Wang and Yao, but in the same ballpark as Chen et al.
Due to the lack of data in manufacturing in the early years, we construct a capital
stock series for manufacturing from 1986 onwards, with an initial capital stock at
407.8 billion yuan in 1986 (at 1952 prices).
5.6.5 Service Lives
Service lives are difficult to estimate (see Erumban, 2006). In 1985, the State
Department of China issued the Regulation of fixed assets and depreciation in State-
owned enterprises28, which is so far the most informative document on service lives of
fixed assets. It offers service lives for three types of fixed assets: ordinary machinery,
special purpose machinery and construction. The average service life is 16 years for
machinery and equipment, and 30 years for construction. There is no information on
the category of others. We assume a service life of 7 years for this category.
On fixed assets in industry, there is a widely used document, Regulation of Industrial
Enterprises29, published by the Financial Department of China and valid since 1993.
With the data from this source, we find an average service life of 14 years for
machinery and equipment, and 27 years for construction in industrial fixed assets.
These service lives are somewhat shorter than the ones for state-owned enterprises.
There are two possible explanations for this difference. One is that the first estimates
27 Jefferson et al. (1992) estimate the net value of productive fixed assets at end of 1979 through the net
value of fixed assets and the ratio of productive to total fixed assets (i.e. NFA*productive ratio).
28 http://www.86148.com/chinafa/shownews.asp?id=1247 issued on 26 April 1985. 29 http://www.bjab.gov.cn/flfg/showsingle.asp?which=99 issued on 30 December 1992, and valid since
1993.
Regional Capital Inputs
99
are for the total economy, while the second are for industry. Service lives may be
somewhat shorter in industry. The second explanation may be that service lives of
fixed assets are getting shorter as time progresses. The second regulation is from 1992
while the first one dates from 1984. The later set of estimates are more appropriate for
recent years, mainly because there are more product innovations in the market and
obsolescence rates are increasing especially in high-tech sectors related to computing
and the internet (OECD, 2001a, p.50).
Summarizing, for the period 1952-1989, we assume a service life of 30 years for
construction, 16 years for machinery and equipment, and 7 years for other assets.
From 1990 onward, we use service lives of 27 years for construction, 14 years for
machinery and equipment, and 6 years for other assets.
5.6.6 Age-Efficiency Patterns
As explained in section 2.3 of this chapter, efficiency coefficients can be obtained
either through assuming a certain pattern, or by means of tracking the relationships
between age-efficiency and age-price profiles if rentals and economic depreciation
rates are available. In this chapter, we apply hyperbolic decay functions as proposed
by the Bureau of Labour Statistics and the Australian Bureau of Statistics to derive the
efficiency of fixed assets in Chinese total economy, industry and manufacturing. The
hyperbolic age-efficiency function used by BLS and ABS is
)/()( sTsTs βφ −−=
T is the service life of a fixed asset, and s is the age of current fixed asset, and β is a
parameter determining the hyperbolic shape, which takes a value of 0.5 for equipment
and of 0.75 for structures.
The age efficiency function is used to derive the efficiency coefficients for the NIFA
of each year and to express the productive capability of investments in standardised
efficiency units. PIM is applied to efficiency adjusted investment series.
Chapter 5
100
5.6.7 Estimates of the Capital Stock: Summary
We can summarise our procedures for construction a capital stock for the industrial
sector as follows:
1. We construct a time series for investment for the total economy for the period
1953-2003, using the concept of Newly Increased Fixed Assets (NIFA). Data on
total NIFA are published from 1980 onwards. From 1953 to 1980, there are
published data on NIFA in basic construction in state owned enterprises. We use
proportions of total investment to investment in basic construction from TIFA to
adjust the published NIFA series upwards.
2. The NIFA series for the total economy are adjusted for changes in coverage in
1980 and 1997.
3. We derive estimates of productive NIFA (P-NIFA) in the total economy from the
NIFA series in 2.
4. We apply the ratio of investment in industry to investment in the total economy to
P-NIFA in the total economy. For this, we use investment ratios in the combined
categories basic construction and technical renovation. This gives us a series of
productive NIFA in the industrial sector.
5. Applying ratios of investment in machinery and equipment, non-residential fixed
structures, and other assets from TIFA for the total economy, we decompose
productive NIFA in industry into these three categories since 1981. We apply the
average of the content proportions 1981-2003, to break down P-NIFA for the
earlier period 1953-1980.
6. The investment in the three categories is deflated using appropriate deflators.
7. An estimate of the initial capital stock is estimated for industry for 1952, using
incremental capital value added ratios.
8. Investments are standardised for productive efficiency using a hyperbolic decay
function.
9. We apply a PIM with appropriate service lives for the three asset categories,
resulting in an efficiency standardised capital stock series. For the period 1952-
1989, we assume a service life of 30 years for construction, 16 years for
machinery and equipment, and 7 years for other assets. From 1990 onward, we
Regional Capital Inputs
101
use service lives of 27 years for construction, 14 years for machinery and
equipment, and 6 years for other assets.
10. The capital stock series are used as a proxy for the index of capital service inputs
The same procedures have been applied for manufacturing. Here we estimate an
initial stock for 1985. Data to construct productive NIFA in manufacturing prior to
this year are not available.
Figure 5.8 shows the whole estimate process, and the final results are reproduced in
Table 5.7 and Figure 5.9.
Chapter 5
102
Figure 5.8: The Estimate Process on Capital Input in Non-residential Fixed
Structures, Machinery and Equipment in Industry: by Region 1953-2003 at
Constant 1952 Prices
NIFA-BC (total economy) (SOU)
(1953-80)
NIFA (SOU) (total economy)
(1953-80)
NIFA (Total economy) (1953-80)
+
published NIFA for 1981-present
P-NIFA (total economy)
P-NIFA-Industry (by three categories of content,
(at constant 1952 prices)
P-NIFA-industry (by region)
P-NIFA-total economy (by region)
P-NIFA-Industry (broken down by fixed structure,
machinery and others)
P-NIFA-Industry (at current prices)
The content categories (non-residential fixed
structure, machinery and equipment, and
others) are not available in published NIFA. We
make an estimate of the content categories,
using proportions from BC and TR in TIFA for
the period 1981-2003. We apply the average of
the 1981-2003 ratios to the data for 1953-80.
There is no published NIFA on technical renovation. We have to
use the ratios of BC and TR in TIFA, 1953-1980. (See also
CSY04-6-6 & DSIFA,1997, p20 and p71).
1) NIFA-BC in SOU is changed to NIFA (BC+TR) in SOU.
2) We have NIFA-SOU since 1952, but NIFA in total economy is available only from
1981, so we have to transfer the NIFA-SOU (1953-80) to NIFA total economy:
In the "original FA" file, we have the ratio of non-SOU/SOU in industrial OFA during
1953-99. (The ratio for the total economy is not available, but industry is a very
substantial part of total NIFA.) Then we calculate the non-SOU/SOU ratio (called ratio-
a). In the "Investment" file, we can get the ratio of non-SOU/SOU TIFA during 1980-96
(called ratio-b). Comparing the overlapping years, 1981-96, we get ratio-b which was 2
times of ratio-a in 1980, and it increases gradually to 4 times in 1985. Considering the
industry structure in China didn't change too much before 1980, therefore we rely on the
1980 ratio to interpret TIFA-total from TIFA-SOU during 1953-1979. So,
correspondingly, we can have the ratio of non-SOU/SOU TIFA during 1953-1980. We
assume NIFA and TIFA share the same non-SOU/SOU ratios, so we apply this to get the
NIFA-total. Applying this, we can transfer NIFA-SOU to NIFA-total economy 1953-80.
Applying the productive ratio (a ratio is published
from 1953-1995 once every five years).
1) For 1985 onwards: using the ratio (of
industry/total) in NIFA in basic construction and
technical renovation). We use the ratio
industry/total only for the combining categories,
BC+TR, the ratio in real estate and others is less
relevant for our purpose, because we know that the
productive part of these categories are very low.
2) During 1953-1984, we rely on OFA-industry.
Applying price deflator to three
content categories (fixed
structure, machinery and
equipment, and others)
The regional content
categories (non-residential
fixed structure, machinery
and equipment, and others)
are assessed by published
regional BC+TR in TIFA
(1995-2003). For years
before 1995, we apply the
average proportions 1995-
1999 in each region.
P-NIFA-Industry (by region,
at constant 1952 prices)
Apply national
price deflators.
Regional Capital Inputs
103
Table 5.7: Productive NIFA and Estimated Productive Capital Stock
(100 million yuan) Productive NIFA (at current prices) Estimated productive stock (at 1952 prices)
Total Economy Industry Manufacturing Total Economy Industry Manufacturing
1952 843.23 198.68
1953 50.26 22.63 867.43 215.30
1954 55.93 38.36 893.37 246.88
1955 60.83 26.15 921.83 265.59
1956 81.53 31.30 857.23 262.47
1957 95.24 46.64 912.38 301.37
1958 179.47 90.23 1050.05 385.47
1959 219.31 122.83 1198.31 490.79
1960 247.01 132.51 1349.78 597.63
1961 100.32 70.35 1405.80 651.20
1962 59.64 48.82 1406.57 677.38
1963 76.97 30.86 1409.41 681.01
1964 110.64 55.18 1432.06 702.61
1965 164.11 80.81 1483.18 740.25
1966 153.52 73.32 1500.72 761.93
1967 81.33 49.17 1408.04 748.58
1968 59.62 46.73 1411.25 763.78
1969 112.12 62.24 1472.70 797.15
1970 208.24 155.09 1630.10 928.82
1971 196.07 144.45 1762.35 1041.16
1972 200.71 198.07 1884.01 1199.69
1973 266.27 211.66 2068.77 1367.90
1974 262.41 158.75 2252.24 1479.48
1975 314.07 219.86 2485.06 1649.18
1976 256.51 215.76 2647.72 1808.66
1977 328.99 252.11 2865.43 1986.66
1978 426.00 316.80 3171.54 2223.77
1979 512.47 284.37 3540.13 2408.11
1980 531.06 287.86 3888.82 2577.02
1981 472.05 324.91 4154.83 2769.78
1982 568.20 366.14 4478.68 2976.26
1983 679.70 412.08 4880.24 3205.18
1984 853.59 450.35 5377.76 3436.36
1985 1116.41 561.29 350.49 5998.60 3708.92 4078.00
1986 1762.51 987.30 685.27 7006.94 4257.63 4450.29
1987 2075.19 1241.01 799.83 8167.49 4945.75 4852.15
1988 2548.97 1545.89 1009.56 9414.86 5692.92 5285.87
1989 2515.37 1521.78 886.25 10481.62 6331.08 5038.04
1990 2673.92 1665.10 968.82 11526.36 6984.20 5292.59
1991 3102.65 1949.69 1166.95 12625.96 7684.87 5550.46
1992 4173.32 2414.39 1444.18 13923.63 8427.53 5785.04
1993 6191.31 3304.21 2113.47 15435.00 9196.51 5968.36
1994 7948.13 4155.70 2641.27 17225.75 10080.75 6510.51
1995 9689.84 4917.67 3025.86 19302.84 11069.14 7065.49
1996 12334.39 6175.27 3866.35 21811.42 12232.95 7726.70
1997 13852.79 6540.39 3772.03 24738.37 13502.12 8317.15
1998 15138.93 6757.67 3466.76 27876.46 14763.66 8723.56
1999 16480.21 6971.54 3574.19 31216.08 16008.96 9022.82
2000 17957.43 7319.64 3320.63 34774.64 17261.40 9060.13
2001 18855.68 7513.00 3881.73 38426.29 18495.34 9698.09
2002 21611.51 8793.31 4775.20 42586.55 19962.93 10501.95
2003 25242.71 11314.43 7028.38 47410.23 21997.95 11872.08
Chapter 5
104
Source: 1) P-NIFA in total economy from DSIFA, 1997, p.62; DSIFA, 2002, p.77; and productive ratio
from DSIFA, 1997, p.98. 2) P-NIFA in industry (1953-1984) is from the (CIESY04-p.25, CIESY95-
p.53) after applying a (calculated) industry productive ratio; the P-NIFA in industry (1985-2003) is
derived from P-NIFA-total using the ratio (of industry/total) in NIFA in basic construction and
technical renovation. (CSY04-6-27, & 6-28). (CSY04, 6-14& 6-15).
3) P-NIFA in manufacturing is from the P-NIFA-total wit using the ratio (of manufacturing/total) in
NIFA in basic construction and technical renovation (CSY04-6-27, & 6-28). (CSY04, 6-14& 6-15).
Figure 5.9: Estimates of the Capital Stock in Total Economy, Industry and
Manufacturing, 1952-2003 (at 1952 Prices)
0
500
1000
1500
2000
2500
3000
3500
4000
4500
5000
1952
1955
1958
1961
1964
1967
1970
1973
1976
1979
1982
1985
1988
1991
1994
1997
2000
2003
year
billion yuan
Total economy Industry Manufacturing
Source: Table 5.7.
5.7 Regional Capital Input Estimates, 1978-2003
The regional capital input estimates for 30 Chinese regions are derived from the
aggregate national estimates of productive NIFA (P-NIFA) in industry and
manufacturing discussed in section 6.2.3 (see Table 5.7 and Figure 5.8). Regional
shares in investment are used to calculate P-NIFA at regional levels. The procedures
for estimating regional capital inputs can be summarised as follows:
1. Estimating regional series of Productive NIFA, 1953-1980.
There are published regional data on TIFA in basic construction (BC) from 1953 to
1980. Regional shares in TIFA-BC in the total economy are applied to the
aggregate national series of P-NIFA in industry. This provides us with regional
estimates of P-NIFA in industry, 1953-1980. The reason for using total economy
Regional Capital Inputs
105
regional shares, rather than regional shares in industrial investment, is that data on
the latter are not available. As can be seen in Figure 5.9, prior to 1980 industrial
investment accounts for most of aggregate investment, so there is no major bias
involved.
2. Estimating regional series of Productive NIFA, 1980-1994.
There are published regional data on TIFA in basic construction (BC) and technical
renovation (TR) in the total economy from 1980 till 1994. Regional shares in
TIFA(BC) + TIFA(TR) are applied to break down the aggregate series on P-NIFA
in industry, by region. This provides us with regional estimates of P-NIFA in
industry, 1980-1994.
The P-NIFA series in manufacturing start in 1985. Regional shares in TIFA(BC) +
TIFA(TR) are applied to break down the aggregate series of P-NIFA in
manufacturing. This provides us with regional estimates of P-NIFA in
manufacturing, 1985-1994. In this step, the use of regional shares in total
investment, rather than industrial investment to breakdown the series is a second
best solution. Data on regional industrial investment is not available.
3. Estimating regional series of Productive NIFA, 1995-2003.
There are published regional data on NIFA in basic construction (BC) and technical
renovation (TR) from 1995 to 2003 in both industry and manufacturing. Regional
shares in NIFA(BC) + NIFA(TR) in industry are applied to the aggregate national
series on P-NIFA in industry; regional shares in manufacturing are applied to P-
NIFA in manufacturing. This results in regional series of productive NIFA in both
industry and manufacturing, 1995-2003.
4. As in the estimates for the national economy, we apply regional content proportions
from TIFA for the total regional economy to break down regional P-NIFA into
three content categories: non-residential construction and installation, purchase of
equipment and instruments and other investment. At the regional level the
breakdown of TIFA into content categories and the information on residential fixed
structures is only available from 1995 to 2003 (CSY 1996- 2004). For the years
prior to 1995 (industry 1978-1995 and manufacturing 1985-1995), we apply the
Chapter 5
106
average content proportions for the period 1995-1999. From CSY 2005 onwards,
the TIFA and the NIFA data are only available for urban investment, so we cannot
extend our series beyond 2003.
5. Regional Investment in the three categories is deflated using the national deflators
for the three categories, as described in section 6.
6. Estimates of the initial regional capital stocks in industry are made for 1953, using
incremental value added ratios. Estimates of the initial regional capital stocks in
manufacturing are made for 1985, applying the same methodology.
7. The regional investment series are adjusted for efficiency decay, using the
hyperbolic efficiency decay function.
8. We apply a PIM with appropriate service lives for the three asset categories,
resulting in a efficiency standardised regional capital stock series.
9. The capital stock series are used as a proxy for the index of regional capital service
inputs.
Table 5.8 provides the resulting regional time series for capital stocks in
manufacturing. Table 5.9 provides the same information on industry by region30.
30 In Table 5.9, we only report the regional capital input estimates for industry from 1978 onwards.
Tab
le 5
.8: E
stim
ated
Pro
duct
ive
Cap
ital
Sto
ck in
Man
ufac
turi
ng b
y R
egio
n, 1
985-
2003
1985
19
86
1987
19
88
1989
19
90
1991
19
92
1993
19
94
1995
19
96
1997
19
98
1999
20
00
2001
20
02
2003
T
otal
41
17.6
8 44
94.7
6 49
02.0
2 53
41.8
3 50
75.7
6 53
35.5
6 55
99.3
4 58
40.1
9 60
27.1
0 65
71.4
0 71
30.3
7 77
92.4
0 83
87.6
9 87
98.6
6 91
02.5
6 91
39.6
8 97
86.5
7 10
600.
80 1
1989
.81
Bei
jing
2
08.
09
228.
88
25
2.65
2
76.5
6 2
50.
27
264
.55
27
5.75
2
85.4
4 2
91.
24
342
.13
34
5.27
3
75.1
7 3
83.
31
385
.77
37
8.77
3
51.0
7 3
64.
73
367
.85
370.
07
Tia
njin
1
10.
63
123.
22
13
4.13
1
44.2
9 1
35.
11
140
.77
14
9.81
1
56.6
0 1
55.
85
168
.05
20
1.84
2
18.7
2 2
29.
21
248
.88
26
8.12
2
78.8
8 2
95.
38
309
.04
321.
97
Heb
ei
16
6.92
18
1.26
1
96.
53
214
.54
20
6.52
2
16.6
2 2
25.
04
234
.39
24
0.48
2
60.6
7 2
96.
62
337
.23
37
8.51
4
10.1
7 4
64.
09
480
.57
51
7.09
5
60.8
0 65
4.57
Sh
anxi
1
39.
40
155.
57
16
9.63
1
81.6
4 1
71.
04
180
.23
18
9.68
1
95.4
0 1
95.
72
202
.56
20
3.88
2
05.8
1 2
07.
52
204
.77
19
5.03
1
90.6
1 2
02.
64
243
.37
286.
78
Inne
r M
ongo
lia
75.
74
83.3
5 9
0.0
6 96
.85
90
.67
96.4
1 1
03.
72
112
.24
12
0.14
1
30.1
9 1
48.
16
150
.73
15
3.98
1
52.9
6 1
48.
07
138
.57
14
5.01
1
70.5
2 21
3.96
Lia
onin
g 2
99.
18
326.
66
35
8.52
3
92.8
3 3
77.
17
393
.61
41
0.24
4
21.9
4 4
30.
22
458
.21
50
4.17
5
55.0
9 5
96.
02
614
.68
59
3.07
5
94.1
9 6
22.
82
650
.17
719.
26
Jilin
8
9.0
0 97
.32
10
6.67
1
17.9
5 1
12.
47
117
.35
12
2.57
1
27.5
7 1
32.
60
142
.80
15
4.29
2
13.0
4 2
47.
66
277
.42
28
9.41
2
95.2
3 3
12.
75
337
.18
378.
26
Hei
long
jian
g 1
74.
02
191.
31
20
9.53
2
27.8
6 2
17.
24
227
.06
23
4.80
2
39.1
8 2
37.
87
249
.34
25
6.99
2
68.3
2 2
75.
34
278
.77
27
6.45
2
65.6
3 2
85.
23
296
.69
329.
05
Shan
ghai
2
93.
25
317.
60
34
6.73
3
80.9
3 3
64.
98
383
.21
39
8.12
4
06.2
9 4
20.
28
469
.39
51
5.52
5
66.2
2 6
34.
13
666
.77
74
9.50
7
64.2
0 8
03.
38
823
.09
888.
23
Jian
gsu
20
5.91
22
6.02
2
49.
98
275
.36
26
2.38
2
72.6
6 2
84.
23
298
.56
30
8.66
3
31.4
8 3
87.
77
437
.79
48
4.16
5
29.0
6 5
53.
03
579
.49
66
4.17
7
49.8
8 88
7.38
Z
heji
ang
10
9.80
12
0.21
1
31.
19
142
.49
13
6.18
1
42.7
5 1
49.
27
156
.71
16
5.00
1
83.4
3 2
02.
45
227
.66
25
8.03
2
81.1
0 3
06.
86
315
.81
35
8.95
4
15.0
2 49
3.95
A
nhui
1
04.
48
116.
21
12
6.56
1
36.6
3 1
29.
76
135
.51
14
0.84
1
46.4
4 1
48.
08
157
.69
17
8.00
2
06.8
9 2
27.
89
233
.17
24
7.75
2
51.6
2 2
71.
38
293
.21
336.
83
Fuj
ian
89.
28
97.6
5 1
07.
01
115
.74
10
9.98
1
15.5
2 1
21.
19
126
.78
13
2.74
1
47.0
3 1
55.
42
165
.11
17
8.78
1
89.5
8 2
01.
05
223
.47
24
3.88
2
77.1
3 31
3.47
Ji
angx
i 6
8.5
9 75
.21
81
.17
87.6
0 8
3.3
8 87
.96
92
.32
95.9
2 9
8.5
7 1
06.5
0 1
17.
97
124
.21
13
6.78
1
40.6
6 1
41.
30
142
.75
14
7.11
1
56.5
6 18
4.48
Sh
ando
ng
22
7.27
24
6.55
2
68.
86
297
.28
28
3.46
2
97.5
0 3
12.
11
326
.98
33
6.50
3
62.7
2 4
00.
90
439
.08
47
7.98
5
19.1
6 5
67.
88
581
.85
64
9.72
7
62.8
6 97
1.78
H
enan
1
41.
47
155.
03
16
7.82
1
84.1
6 1
76.
44
184
.78
19
5.93
2
04.8
3 2
11.
34
232
.29
25
8.61
2
93.9
2 3
42.
18
356
.90
37
1.40
3
70.3
5 3
98.
91
450
.94
509.
64
Hub
ei
14
3.70
15
8.09
1
73.
26
189
.78
17
8.97
1
87.3
1 1
94.
90
203
.92
21
1.94
2
34.7
5 2
59.
63
343
.18
39
3.44
4
36.5
0 4
56.
45
466
.60
50
5.10
5
40.2
8 60
5.15
H
unan
1
06.
19
116.
44
12
6.89
1
38.1
1 1
31.
35
137
.80
14
5.28
1
54.3
7 1
59.
03
171
.82
18
8.77
2
06.5
8 2
29.
03
232
.90
23
0.33
2
33.7
0 2
54.
85
273
.72
315.
37
Gua
ngdo
ng
35
8.39
39
1.89
4
22.
21
458
.57
43
4.30
4
59.1
5 4
84.
66
515
.94
54
5.20
6
12.8
5 6
31.
26
669
.47
70
3.05
7
68.4
0 7
99.
21
803
.87
84
7.14
9
45.4
4 1
031
.75
Gua
ngxi
6
9.5
9 76
.70
84
.22
93.2
1 8
9.0
9 92
.43
96
.40
101
.87
10
9.91
1
22.6
3 1
52.
72
159
.68
16
4.46
1
67.0
5 1
66.
07
168
.64
17
5.98
1
81.5
5 21
0.07
H
aina
n 2
5.7
9 24
.97
24
.05
26.5
5 2
6.9
5 30
.84
34
.40
38.8
2 4
6.3
2 56
.23
63
.96
74.4
1 8
2.4
3 84
.92
81
.52
77.0
7 7
5.6
4 73
.74
75.9
0 Si
chua
n 2
27.
72
247.
20
26
9.30
2
91.7
9 2
78.
69
293.
86
30
9.83
3
22.5
6 3
28.
60
351
.15
38
4.90
4
06.0
9 4
37.
07
449
.96
45
2.28
4
49.6
8 4
79.
16
520
.10
610.
17
Gui
zhou
5
2.3
0 57
.34
62
.06
67.1
3 6
3.9
1 67
.30
70
.54
73.2
1 7
3.7
6 77
.92
82
.52
86.2
0 9
1.7
2 94
.32
10
5.26
1
08.4
6 1
19.
60
125
.85
136.
11
Yun
nan
71.
36
78.3
0 8
4.8
0 92
.17
87
.25
91.8
9 9
8.0
5 1
04.4
0 1
11.
80
123
.96
14
7.49
1
68.5
0 1
88.
17
203
.11
20
8.62
2
07.6
2 2
16.
64
224
.51
228.
73
Tib
et
9.5
5 10
.50
11
.18
11.8
8 1
1.0
9 11
.87
12
.89
13.6
6 1
4.1
8 15
.25
14
.72
14.4
6 1
3.6
0 12
.64
11
.27
9.81
9
.49
9.07
9
.03
Shaa
nxi
10
3.30
11
2.75
1
22.
76
132
.84
12
6.95
1
33.5
2 1
38.
56
140
.66
14
2.25
1
49.2
8 1
58.
10
168
.07
16
8.71
1
76.7
8 1
84.
22
179
.90
19
4.74
2
15.7
0 24
3.36
G
ansu
6
9.5
7 75
.79
82
.72
90.0
9 8
5.4
0 89
.93
93
.71
95.6
6 9
3.7
7 96
.85
10
3.02
1
12.6
4 1
20.
34
123
.15
13
8.86
1
41.3
9 1
53.
41
158
.10
189.
86
Qin
ghai
3
0.0
1 33
.65
37
.14
40.6
7 3
8.4
4 39
.71
40
.68
40.8
3 4
0.4
1 41
.39
41
.07
40.6
3 4
0.8
0 41
.04
38
.85
35.8
0 3
6.1
8 37
.06
38.8
2 N
ingx
ia
25.
84
28.7
2 3
1.6
0 33
.55
31
.51
33.0
6 3
4.6
0 35
.60
35
.34
36.9
7 3
8.1
0 38
.74
40
.01
39.0
7 3
9.1
9 43
.17
54
.36
57.8
3 64
.89
Xin
jian
g 8
6.9
3 93
.44
99
.59
107
.10
10
1.24
1
08.8
0 1
16.
47
126
.12
13
4.28
1
46.7
3 1
65.
85
172
.71
18
2.63
1
87.4
6 1
83.
70
180
.28
18
3.85
1
89.5
8 20
0.81
N
ot
clas
sifi
ed
23
4.43
24
6.95
2
73.
21
295
.69
28
3.57
3
01.6
1 3
22.
75
337
.30
35
5.02
3
89.1
4 3
70.
40
346
.07
32
0.79
2
91.5
8 2
54.
96
209
.38
19
7.30
1
83.9
6 17
0.11
Not
e: a
t 10
0 m
ill y
ua
n, a
t 19
52 c
onst
ant
pri
ces.
Ch
ongq
ing
is in
clud
ed in
Sic
hua
n.
The
nat
ion
al t
ota
l is
slig
htly
diff
ere
nt f
rom
the
aggre
gate
da
ta in
Ta
ble
5.7
. T
his
is d
ue t
o th
e f
act
tha
t we
app
lied
regi
ona
l pro
port
ions
in t
he c
alc
ulat
ions
. S
ourc
e:
auth
ors’
est
ima
tes
.
Regional Capital Inputs
107
Tab
le 5
.9: E
stim
ated
Pro
duct
ive
Cap
ital
Sto
ck in
Ind
ustr
y by
Reg
ion
19
78
1979
19
80
1981
19
82
1983
19
84
1985
19
86
1987
19
88
1989
19
90
1991
19
92
1993
19
94
1995
19
96
1997
19
98
1999
20
00
2001
20
02
2003
T
otal
22
33
2420
25
91
2786
29
95
3226
34
61
3736
42
91
4985
57
39
6384
70
42
7748
84
98
9272
10
161
1115
8 12
324
1360
4 14
878
1613
8 17
402
1865
3 20
141
2220
3 B
eiji
ng
84
93
104
114
122
133
145
161
192
230
270
302
335
364
394
425
508
521
555
581
610
641
667
711
711
721
Tia
njin
52
60
67
75
84
94
10
2 11
3 13
1 14
9 16
6 18
2 19
7 21
9 23
9 25
1 27
1 32
2 35
4 37
7 41
3 45
4 47
8 51
0 54
0 57
3 H
ebei
10
7 11
9 12
9 13
9 14
9 16
1 17
0 17
9 19
9 22
4 25
4 27
9 30
4 32
9 35
9 38
8 42
0 47
6 54
6 60
7 67
0 77
2 84
3 92
9 98
2 10
84
Shan
xi
74
80
86
91
98
106
119
132
155
178
198
219
240
263
282
297
310
331
347
384
420
434
474
506
624
685
Inne
r M
ongo
lia
37
43
47
51
56
63
71
78
88
99
110
121
133
149
168
188
204
242
257
267
292
334
332
339
384
481
Lia
onin
g 14
5 15
4 16
5 18
0 19
3 20
6 22
0 23
8 28
1 33
6 39
6 44
9 49
4 54
4 59
1 64
1 68
9 75
0 82
4 89
7 97
0 99
1 10
43
1093
11
38
1220
Ji
lin
59
64
69
74
79
85
89
94
106
121
140
153
165
179
194
212
228
247
312
354
396
422
456
494
534
585
Hei
long
jian
g 10
4 11
2 12
2 13
5 14
8 16
2 17
4 18
4 20
9 23
9 26
9 29
4 31
9 34
2 36
3 38
1 39
9 43
0 45
8 50
9 57
1 61
5 66
3 72
4 78
2 85
3 Sh
angh
ai
67
76
88
104
123
143
164
185
226
281
344
397
448
496
540
600
683
747
838
935
1013
12
22
1315
13
49
1351
14
10
Jian
gsu
66
76
85
94
104
116
129
144
176
219
265
299
329
365
408
450
489
566
641
709
800
896
999
1106
12
31
1408
Z
heji
ang
35
40
44
50
57
63
70
78
94
114
135
154
172
191
214
241
272
308
350
417
470
525
581
694
828
915
Anh
ui
60
64
68
73
78
84
90
98
115
132
150
165
180
196
214
229
245
275
329
378
393
421
464
490
519
565
Fuj
ian
31
34
38
42
46
51
57
64
77
94
110
125
140
156
174
194
218
236
256
298
338
379
422
459
506
543
Jian
gxi
41
44
47
50
54
59
65
69
78
89
99
110
121
132
144
156
169
186
198
226
238
247
263
270
285
331
Shan
dong
96
10
6 11
6 12
6 13
8 15
0 16
1 17
5 20
5 24
5 29
5 33
2 37
0 41
1 45
7 50
0 54
5 61
3 68
1 79
1 87
9 97
5 11
06
1232
13
99
1702
H
enan
97
10
4 10
9 11
8 12
6 13
4 14
3 15
2 17
1 19
2 21
8 24
1 26
1 28
8 31
4 34
0 37
3 43
2 50
2 59
3 65
9 70
6 74
0 80
9 89
0 98
4 H
ubei
13
8 14
7 15
5 16
4 17
1 17
8 18
5 19
3 21
0 23
3 25
9 27
5 29
4 31
3 33
8 36
4 39
9 42
6 51
8 58
2 66
3 71
6 76
4 81
0 86
3 10
11
Hun
an
74
79
84
89
95
101
107
114
128
145
163
178
194
213
236
256
276
298
328
362
379
403
432
457
480
543
Gua
ngdo
ng
90
98
106
120
138
157
180
213
269
328
401
459
526
598
685
778
892
975
1079
11
55
1271
13
46
1437
14
99
1656
17
70
Gua
ngxi
45
48
51
54
57
60
63
67
77
90
10
5 11
6 12
5 13
6 15
1 17
1 19
2 23
2 24
8 25
9 27
0 28
3 32
6 33
9 34
8 38
9 H
aina
n 2
2 2
1 2
2 3
3 4
4 5
14
23
32
43
58
75
90
103
112
116
118
120
118
116
120
Sich
uan
163
171
178
186
194
204
213
226
253
290
328
364
401
442
482
517
553
608
647
709
826
889
941
986
1058
11
63
Gui
zhou
54
57
59
62
63
65
67
70
76
83
90
98
10
5 11
3 12
1 12
7 13
2 14
7 16
5 17
5 18
3 21
5 24
4 26
4 28
4 30
8 Y
unna
n 60
64
68
71
76
80
84
89
98
10
8 11
9 12
8 13
9 15
2 16
8 18
7 20
5 23
6 27
0 29
5 32
4 34
5 35
8 38
7 42
0 44
1 T
ibet
6
7 7
7 8
8 10
12
13
14
15
16
18
20
22
24
26
28
28
28
27
28
30
30
31
32
Sh
aanx
i 88
92
97
10
1 10
6 11
0 11
4 11
9 13
1 14
7 16
3 17
8 19
4 20
8 22
0 23
4 24
5 25
5 27
6 29
2 31
5 35
1 37
4 40
5 44
3 49
7 G
ansu
67
70
72
74
76
78
81
84
91
10
2 11
3 12
2 13
2 14
2 15
0 15
4 15
9 16
6 18
0 20
5 22
0 24
3 26
8 29
4 31
1 35
5 Q
ingh
ai
25
30
32
34
36
38
41
43
47
52
58
61
64
67
69
71
72
76
84
98
105
113
116
126
131
130
Nin
gxia
17
19
20
20
21
22
23
25
29
34
38
41
45
49
53
55
58
60
64
77
82
84
94
10
9 11
9 13
4 X
inji
ang
43
49
55
61
67
73
78
84
94
104
117
130
147
165
189
211
231
271
298
341
374
402
466
511
551
607
Not
cl
assi
fied
20
5 21
7 22
2 22
5 23
1 23
7 24
2 25
1 26
7 30
9 34
6 38
4 42
5 47
4 51
7 57
0 62
4 60
7 58
8 59
2 59
2 56
9 58
8 60
4 62
8 64
4
Not
e: a
t 100
mil
l yua
n, a
t 195
2 co
nsta
nt p
rice
s. C
hong
qing
is in
clud
ed in
Sic
huan
. The
nat
iona
l tot
al is
sli
ghtl
y di
ffer
ent f
rom
the
aggr
egat
e da
ta in
Tab
le 5
.7. T
his
is d
ue to
th
e fa
ct th
at w
e ap
plie
d re
gion
al p
ropo
rtio
ns in
the
calc
ulat
ions
. S
ourc
e: o
wn
esti
mat
es.
Chapter 5
108
CHAPTER 6
Productivity Growth and Structural Change in Chinese
Manufacturing, 1980-20021
6.1 Introduction
Since the mid-nineties productivity growth in Chinese manufacturing has been
accelerating dramatically. Between 1980 and 1992, labour productivity growth
averaged 3.4 per cent per year for enterprises at township level and above. Between
1992 and 2002, productivity growth accelerated to 14.8 per cent per year. Between
1996 and 2003 it was no less than 19.6 per cent per year. The period 1980-92 can be
characterised as a period of growth without catch up. Productivity growth was
respectable, but the gap relative to the world productivity leader (the United States)
remained about the same. In 1992, productivity relative to the U.S. stood at 5.5 per
cent of the U.S. level. By 2002, it had reached 13.7 per cent of the U.S. level (Szirmai
et al., 2005). This is a spectacular example of productivity catch up.2
Chinese productivity growth can be explained by a variety of factors, including high
domestic rates of investment, the opening up of the economy to foreign direct
investment, a massive shakeout of non-productive labour in the state-owned
1 An earlier version of this chapter has been published in Industrial and Corporate Change, see Wang and Szirmai (2008a). 2 These figures refer to the enterprises at township level and above, for which long run time series are available. After 1998, the series refer to enterprises with more than five million yuan in sales. Much less is known about the millions of very small enterprises (individual enterprises and sole proprietorships), which are not systematically documented in Chinese statistics. It is clear, however, that productivity levels and growth rates are much lower for these enterprises. Szirmai and Ren (2007) make a rough estimate of productivity growth in total manufacturing including small-scale enterprises. It was only 7.8 per cent between 1995 and 2002. This chapter focuses on an analysis of the time series for enterprises at township level and above until 1998 and enterprises with more than five million in annual sales until 2002.
Chapter 6
110
enterprises, a succession of efficiency enhancing economic market reforms, the
emergence of new dynamic types of ownership such as joint stock companies, village
and township enterprises and foreign owned enterprises, and structural changes within
manufacturing. Rapid productivity growth in manufacturing has been achieved in the
context of shrinking employment. In 1980, the manufacturing sector at township level
and above employed 41.9 million persons. This increased to 71.3 million persons in
1995, but subsequently dropped to 48.7 million persons in 2004 (Szirmai and Ren,
2007, see also Banister, 2005; Deng and McGuckin 2005; McGuckin and Spiegelman,
2004). Lay-offs were particularly pronounced in the state-owned sector. In spite of the
drop in employment since 1995, manufacturing value added continued to expand at
15.1 per cent per year between 1995 and 2003. Part of the shedded labour was
reabsorbed in smaller enterprises. After 1995, the social labour force in total
manufacturing did shrink somewhat, but only by some three million workers. Excess
labour was also absorbed in the service sector (McGuckin and Spiegelman, 2004).
This chapter focuses on the contribution of structural change to aggregate
manufacturing and aggregate industrial performance. Since the start of the reform
period in 1978, the booming Chinese industrial sector has experienced rapid structural
change. The sector structure of production has been changing; the ownership structure
has been changing, with more scope for foreign funded enterprises, private enterprises
and reforms of state-owned enterprises. Finally, the regional structure is also
undergoing change. Using shift-share techniques, we will examine the effects of three
types of structural change: changes in the sectoral structure of production, changes in
the ownership structure and changes in the regional structural of production.3
The aim of this chapter is to analyse the interplay of structural, regional and
institutional change in the context of the rapid growth of manufacturing and industry.
3 The sectoral analysis is performed for manufacturing, the institutional and regional analysis for the broader industrial sector. There is no breakdown of the manufacturing figures by regions and ownership category.
Productivity Growth and Structural Change
111
6.2 Structural Change and Productivity Growth
6.2.1 Sectoral change
Industrialization is regarded as a crucial engine of growth in the process of economic
development. There is a vast literature discussing the productivity differentials and
employment changes in agriculture, industry and services (the primary, secondary and
tertiary sector). Many researchers believe that growth in developing countries is
driven by increases in the shares of manufacturing and declines in the shares of
agriculture. As productivity levels in manufacturing (and in industry) are much higher
than in agriculture, structural change provides a productivity bonus, the structural
change bonus.
A more negative assessment of structural change is offered by Baumol (1987).
Baumol formulated a two-sector model in which the possibilities for productivity
growth in the service sector are limited. Therefore the increasing share of services in
production will result in an aggregate productivity slowdown. Van Ark and Timmer
(2003) find some evidence of the Baumol effect in East Asia, but its impact is partly
offset by the fact that productivity levels in some service sectors are higher than in
manufacturing. In the same paper, the authors find positive effects of manufacturing
on total economic performance. Manufacturing is still an engine of growth in
developing countries.
Most of the structural change literature, especially with regard to developing countries,
focuses on the major sectoral shifts such as that from agriculture to industry, or
industry to services. Less attention has been paid to the study of shift effects within
the manufacturing sector (for exceptions see Timmer, 2000; Timmer and Szirmai,
2000; Fagerberg, 2000; Peneder, 2003; Vial, 2006). Much has been written about the
role of sectoral and technological upgrading in growth and development, especially
with regard to Korea and Taiwan, but this literature is often more descriptive in
nature.
In China, the manufacturing sector is one of the most important sectors of the
economy in terms of value added and employment. It is the main engine of the present
Chapter 6
112
growth process. Some studies have indicated that China’s economic growth will
continue in future years, though perhaps not at the same rate as in the past, because of
its healthy pattern of growth, characterised by structural change, catching up and
factor price equalization (Holz, 2005; Wu, 2000). It is of great importance to examine
the relationship between productivity and structural change within Chinese
manufacturing sectors in more detail. This chapter takes a first step in this direction.
6.2.2 Institutional change
Besides sectoral change, rapid institutional change is a typical characteristic of recent
Chinese economic history. Since the onset of reforms in 1978, state-owned companies
in China are experiencing the greatest changes. The number of state-owned
enterprises (at township level and above) decreased from over 80,000 in 1980 to
41,000 in 2002. See Table 3.1 for the 2004 figures. In 2004 there were only 23,417
state-owned enterprises above designated size accounting for 11 per cent of gross
output. The gross output share of foreign funded enterprises increased from 0.4 per
cent in 1985 to above 32.7 per cent in 2004. The value added produced by
state-owned enterprises increased from 130 billion yuan in 1980 to 576 billion yuan
(at 1980 constant prices) in 2002, but the SOE share in value added declined from 81
per cent to 48 per cent. The shares of private enterprises, joint ventures, joint stock
enterprises and especially foreign-funded enterprises have been increasing (see
Szirmai et al. 2005).
According to most studies, the state-owned enterprises have very low or even zero
growth of total factor productivity (Woo et al., 1994; Jefferson et al., 1992). Township
and village enterprises (TVE) - a subsector of the collectively owned sector - have had
positive TFP growth rates4. The TVE sector has been one of the most dynamic sectors
in Chinese manufacturing. Huang (2004) focuses on the role of foreign financed
enterprises in Suzhou and Zhejiang. He concludes that policy biases against domestic
private firms result in a high preference for FDI. Ai and Wen (2005) examine the
performance of different ownership categories in industry between 1996 and 2002.
Foreign financed enterprises and domestic private enterprises seem to have the highest
4 See more in Chapter 3.
Productivity Growth and Structural Change
113
productivity levels. The same conclusion is reached by McGuckin and Spiegelman
(2004). Though there are several studies of the growth performance in specific
ownership categories, comprehensive overviews of the structural shifts between all
the different ownership categories (state-owned, collective, foreign funded etc.) are
still scarce. This chapter hopes to increase our knowledge in this area.
6.2.3 Regional shifts
The regional composition of industrial employment is also changing over time. For
instance, the industrial employment share in Guangdong changed from 4 per cent in
1978 to 12 per cent in 2002. The industrial employment share in Shandong also
increased, from 5 per cent in 1978 to 10 per cent in 2002. However, the share in
Liaoning decreased from 8 per cent in 1978 to 5 per cent in 2002.
We will examine the impact of interregional shifts in employment on aggregate
productivity growth, as well as the relative contributions of coastal, middle and
western regions to productivity growth.
6.3 Decomposition Methods
To measure the contribution of structural change to productivity growth, it is crucial
to distinguish between the contributions of shifts between sectors and the
contributions of productivity growth within sectors. In analyzing the effects of
structural change, one should ideally analyze the impacts of shifts in both capital and
labour on total factor productivity. This is known as "complete measurement"
(Syrquin, 1984; Denison, 1967). However, given the lack of data, in many cases the
analysis has focused on the shift of one input factor (labour). This is referred to as
"partial measurement".
The shift-share method was first used by Fabricant (1942), who measured labour
requirements per unit of output. Since then, it has been widely applied in the analysis
of economic growth, though later attention has switched to productivity growth issues.
The shift-share model decomposes productivity growth into its different sources. It
Chapter 6
114
highlights the impact of shifts of factor inputs from the supply side. The standard
shift-share model (e.g. Syrquin 1984; Paci and Pigliaru, 1997; Timmer and Szirmai,
2000) has three terms.
0
1
00
0
1
00
0
1
00
0
0))(()()(
P
PPSS
P
PSS
P
SPP
P
PP
n
i
i
t
ii
t
i
n
i
ii
t
i
n
i
ii
t
it ∑∑∑===
−−
+
−
+
−
=−
(1)
tP is the aggregate labour productivity at year t;
0P is the aggregate labour productivity at year 0;
t
iP is the labour productivity of branch i (sector, ownership or region) at year t;
0
iP is the labour productivity of branch i at year 0;
t
iS is the employment share of i branch at year t;
0
iS is the employment share of i branch at year 0.
The first term in the left part of equation (1) denotes the effect of productivity growth
within sectors (or ownership categories, or regions). The second term measures the
static effect of reallocation of labour between sectors with differing levels of labour
productivity. The last term is an interaction effect of productivity growth and labour
shifts. It can be interpreted as the dynamic effect of shifts towards sectors with higher
than average or lower than average productivity growth. This term will have a
positive impact on productivity growth if labour shifts to sectors where productivity is
improving more rapidly than the average. It will have a negative contribution if labour
moves to sectors where productivity is increasing less rapidly than average
productivity.
This conventional shift-share method is applied in much research on growth and
structural change. Nevertheless a number of unavoidable shortcomings need to be
mentioned. In the first place, this method cannot tell us anything about the role of
demand in structural change. In the second place, the analysis is usually restricted to
labour productivity. Only if sufficient data on sectoral capital stocks are available can
one decompose aggregate total factor productivity (Timmer and Szirmai, 2000). In the
third place, shift-share methods disregard economies of scale. If factor inputs are
reallocated to given sectors, economies of scale could result in increasing productivity
in those sectors (Verdoorn’s law). This is an indirect effect of structural change, but it
Productivity Growth and Structural Change
115
will not be identified as a shift effect by the shift-share methods. Timmer and Szirmai
(2000) showed that it is possible to incorporate Verdoorn’s law in shift and share
methods by using branch specific elasticities of TFP relative to growth of output. But
they concluded that "the inclusion of the Verdoorn effect has little influence on the
decomposition results". A fourth and most important point of criticism regards the
assumption that marginal productivity equals average productivity. This may not be
the case. In traditional agriculture, there is frequently surplus labour with low or even
close to zero marginal productivity. Reallocation of this non-productive labour to
industry results in increases in productivity in the agricultural sector, due to the
shedding of unproductive labour. In the shift-share method this is measured as
intra-sectoral productivity growth in agriculture, while in reality it is a consequence of
the reallocation of labour. Thus, the shift-share model will underestimate the shift
effect. (For a recent attempt to adjust for this effect for agriculture, see van Ark and
Timmer, 2003). As surplus labour is less important in manufacturing than in
agriculture, we have not made adjustments for this in this chapter. But, adjusting for
surplus labour in the state-owned sector is one of the possibilities for future research.
Finally, shift-share methods cannot capture the effects of intersectoral technology
spillovers. For instance, the importance of the electronics sector in developing
countries may be greater than indicated by shift-share methods, because this sector
has such strong externalities (cf. Fagerberg, 2000; Peneder 2003). In conclusion,
shift-share methods provide fruitful and systematic insights in the sectoral
contributions to growth, but one should see them as lower-bound estimates of the
importance of structural change.
Most models decompose productivity growth from the beginning to the end of a given
period, using only the data at the beginning and the end of the time series. Van Ark
and Timmer (2003) use a shift-share method utilising all the intervening data points
and using mean labour shares as weights. As a result of mean weights and the use of
the data for all intervening years (in continuous time), the dynamic third term in
equation (1) disappears (see also Syrquin, 1984). One only distinguishes between the
within effect and the shift effect.
In standard formulations, the shift effects derive both from sectors with increasing and
with decreasing labour shares. For instance, agriculture will contribute negatively to
Chapter 6
116
the shift effect when its productivity is below average and its labour share is
decreasing. This makes it harder to interpret the shift effects analytically. Van Ark and
Timmer come up with a model (equation (2)) which differentiates between the shift
effects of expanding and shrinking sectors. They reallocate all the shift effects of
shrinking sectors to the shift effects of expanding sectors.
Suppose K is the set of sectors which expand their labour shares, and J is the set of
sectors with declining labour shares. The increase in the labour share of the expanding
sectors equals the decline of the labour share of the shrinking sectors. Therefore, for
expanding sectors we can use ))(( 0
Jii
T
i PPSS −− to express the combined shift
effect from shrinking and expanding sectors. This effect will be positive if average
productivity iP in expanding sectors is higher than average productivity jP in the
shrinking sectors. Thus the contribution of sector i to the aggregate labour
productivity becomes
))(()( 00int
Jii
T
iii
T
i
shift
i
ra
ii PPSSSPPCCC −−+⋅−=+= Ki∈
ii
T
i
ra
ii SPPCC ⋅−== )( 0int Ji∈ (2)
where 0S , TS are the labour share at year 0 and year T respectively, and S is
average productivity for the whole period; 0P and TP are the labour productivity
at year 0 and year T, P is the average productivity level. The average labour
productivity over all shrinking sectors is
∑
∑
∈
∈
−
−
=
Ji
i
T
i
Ji
ii
T
i
JSS
PSS
P)(
)(
0
0
(3)
In this chapter, we use two shift-share models to analyze the shift contribution to
productivity growth. We use the model of Van Ark and Timmer (2003) (equation 2)
for the analysis of sectoral change (Table 6.2), technology shifts (Table 6.3),
institutional change (Table 6.4) and regional change (Table 6.8). For the combined
effect of ownership and region (Table 6.5 and Table 6.9) we use the discrete model
(equation 1). An advantage of the discrete model is that it allows for a distinction
between static and dynamic shift effects. An important advantage of the Van Ark and
Timmer model is that it allocates the shift effects to the expanding sectors, which
makes it easier to interpret the results. It also utilises the information for all the
Productivity Growth and Structural Change
117
intervening years of a time series. The main reason for applying the discrete model in
tables 6.5 and 6.9 is data availability. In crosstabulations of region and ownership, not
all data are available for intervening years. The differences between the two models
are minor. The shift effect in the continuous models is approximately equal to the sum
of static and dynamic shifts in the discrete model. In the present chapter, we disregard
the issues of surplus labour or disguised employment.5
6.4 Results
6.4.1 Sectoral Change
We apply the modified shift-share model discussed above (eq. 2 and 3) to our 21
manufacturing sectors in the period 1980-2002. The basic data on sectoral labour
shares are reproduced in Table 6.1. Note that the changing shares are the net result of
job creation and job destruction. Even small changes may reflect stronger underlying
dynamics. Next, the shares should be interpreted in the context of increasing
aggregate employment till 1996, followed by very dramatic declines after this year
(see bottom row of table). These aggregate changes reflect the vast restructuring and
shedding of redundant labour, primarily by state-owned enterprises.
Thirteen of the 21 sectors are rather stable with less than a percentage point of change
in their shares over 22 years. A few of the changes are quite striking. The textile
industry first expands its labour share, but then shrinks to well below its 1980 levels.
In contrast, the clothing industry, which is less amenable to automation, gradually
expands its share. The most dramatic change is that of the machinery sector, where the
share declines from 18.7 to 10.8 per cent. Fabricated metals is also characterised by
quite substantial declines. The electronic and telecom sector doubles its share, which
is consistent with the increasing importance of high-tech activities. Increases are also
registered in beverages, leather products, chemicals, transport equipment and
electrical machinery. Table 6.2 presents the results of the sectoral shift and share
5 Disguised unemployment and marginal productivity much lower than average productivity may be
relevant for the analysis of institutional shifts from the state-owned sector to other sectors. It is likely that the state-owned sector in past years was characterised by surplus labour. This is not taken into account in this chapter.
Chapter 6
118
analysis. For each period, the table registers the contribution to total productivity
growth of each sector (column Total). The figure in this column is the sum of the
contribution of intrasectoral productivity growth (intra) and intersectoral shift effects
(shift). Sectors with highly positive or negative shift effects are the expanding sectors.
It is worth noting that some expanding sectors contribute negatively to productivity
growth. If a sector has a consistently shrinking labour share, the shift effect using
equation 2 will be equal or close to zero.6
Table 6.1: Labour Shares in Manufacturing, 1980-2002 (%)
1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002
Food manufacturing 5.2 5.4 5.7 5.7 5.7 5.7 5.6 5.8 5.9 5.9 5.7 5.4
Beverages 1.3 1.4 1.6 1.9 2.0 2.0 2.0 2.1 2.2 2.4 2.5 2.4
Tobacco processing industry 0.3 0.4 0.4 0.4 0.4 0.5 0.5 0.4 0.5 0.5 0.6 0.5
Textile industry 11.4 12.1 12.7 13.4 14.0 14.3 13.8 12.7 12.0 11.4 11.1 10.6
Clothing industry 3.2 3.4 3.6 3.6 3.4 3.5 3.6 4.0 3.8 3.9 4.3 5.2
Leather and fur products 1.4 1.5 1.5 1.6 1.5 1.6 1.7 2.0 2.1 2.0 2.1 2.4
Wood products 1.3 1.4 1.5 1.4 1.3 1.4 1.3 1.6 1.6 1.5 1.4 1.3 Paper, paper products and printing industry 4.0 4.0 4.0 4.0 4.1 4.1 4.1 4.1 4.1 4.1 3.8 3.7 Oil refining, coal, coking and coal products 0.9 0.8 0.8 0.9 1.0 1.1 1.2 1.1 1.2 1.3 1.3 1.4 Chemical industry, excluding oil refining 8.1 8.0 7.9 7.8 8.1 8.6 8.9 9.1 9.5 9.7 9.8 9.5
Rubber and plastic products 3.2 3.3 3.4 3.5 3.5 3.6 3.7 3.5 3.6 3.6 3.5 3.6 Building materials and other non-metallic minerals 10.9 11.2 11.6 12.0 11.8 10.8 10.6 10.7 11.1 11.0 10.8 10.4
Basic metals 6.7 6.5 6.2 6.1 6.2 6.5 6.6 7.2 7.2 7.4 7.6 7.4
Fabricated metals 5.2 5.0 4.8 4.7 4.5 4.4 4.3 4.1 3.8 3.8 3.5 3.6
Machinery 18.7 17.7 16.8 15.8 15.4 14.9 14.7 12.3 12.1 12.1 11.5 10.8
transport equipment 5.8 5.5 5.2 4.9 4.7 4.8 4.9 5.5 5.8 6.1 6.2 6.4 Electrical machinery and equipment 4.1 4.0 4.0 4.1 4.1 4.1 4.3 4.2 4.4 4.4 4.4 4.7 Electronic and telecom machinery 2.6 2.6 2.5 2.5 2.4 2.5 2.6 2.6 2.7 3.0 3.6 4.5
Instruments 1.6 1.5 1.4 1.3 1.2 1.2 1.2 1.4 1.3 1.3 1.3 1.4
Furniture 1.4 1.3 1.3 1.2 1.1 1.0 0.9 0.7 0.7 0.7 0.6 0.8
Other manufacturing 2.8 2.8 2.9 3.3 3.5 3.6 3.5 4.9 4.5 4.2 4.3 4.1
Total 100 100 100 100 100 100 100 100 100 100 100 100
Persons engaged (10000) 4186 4641 5161 5860 6381 6472 6777 7103 6922 6246 5350 4798
Source: Szirmai et al. 2005, Table 11. 1985: 1985 Industrial Census; 1980-1992, Yearbook of Industrial Statistics 1993, p. 90 ff. 1993-1994: from China Statistical Yearbook, 1996 (original source CLSY, various issues); 1995: 1995 Industrial Census; 1995-2002 China Statistical Yearbook, 2000 (original source China Labour Statistical Yearbook, various issues). The data for 1980-1992 and for 1995 refer to staff and workers in enterprises with independent accounting enterprises at township level and above. The coverage of the original series 1993-2002 is limited to staff and workers in urban enterprises excluding employment in rural enterprises at township and above. From 1998 onwards the staff and workers concept is restricted to on-post staff and workers. The data from 1998 onwards have been adjusted to the older concept of staff and workers. The data for 1993-2002 have been adjusted in coverage to township level and above (with independent accounting systems).
6 Using equation 2, a sector with shrinking shares in all years of a period will have a zero shift effect, as all shift effects have been reallocated to expanding sectors. As the data of all years are utilised and
sectors usually do not shrink in every single year, there are only a few sectors with zero shift effects.
Tab
le 6
.2:
Dec
ompo
siti
on o
f M
anuf
actu
ring
Pro
duct
ivit
y -
Con
trib
utio
n of
Sec
tora
l Shi
fts,
198
0-20
02
19
80-1
990
1990
-200
2 19
80-2
002
L
P le
vel
(Po)
19
80
Ann
ual
LP
grow
th
rate
Intr
a sh
ift
Tota
l L
P le
vel
(Po)
19
90
Ann
ual
LP
grow
th
rate
intr
a sh
ift
Tota
l
LP
leve
l (P
o)
1980
Ann
ual
LP
grow
th
rate
intr
a sh
ift
tota
l L
P le
vel
(Pt)
20
02
Food
man
ufac
turi
ng
3094
4.
21%
8.
72%
0.
14%
8.
86%
40
63
17.0
8%
6.42
%
0.06
%
6.48
%
3094
10
.64%
6.
52%
0.
07%
6.
58%
20
689
Bev
erag
es
4356
3.
50%
2.
22%
1.
37%
3.
59%
50
07
12.0
7%
2.33
%
0.21
%
2.53
%
4356
7.
79%
2.
32%
0.
25%
2.
58%
19
637
Toba
cco
proc
essi
ng
indu
stry
38
988
3.55
%
12.5
9%
9.04
%
21.6
3%
5763
5 11
.57%
4.
29%
0.
65%
4.
94%
38
988
7.56
%
4.64
%
1.00
%
5.64
%
1749
32
Text
ile
indu
stry
41
76
-2.6
5%
-19.
09%
3.
75%
-1
5.34
%
3242
17
.08%
7.
91%
0.
00%
7.
91%
41
76
7.21
%
6.78
%
0.16
%
6.94
%
1329
7 C
loth
ing
indu
stry
18
45
3.89
%
4.63
%
-0.7
5%
3.88
%
2675
14
.96%
2.
05%
-0
.28%
1.
77%
18
45
9.42
%
2.16
%
-0.3
0%
1.86
%
1022
3 L
eath
er a
nd f
ur p
rodu
cts
1950
1.
76%
0.
71%
-0
.33%
0.
38%
22
31
18.7
3%
1.22
%
-0.1
3%
1.08
%
1950
10
.24%
1.
19%
-0
.14%
1.
05%
10
998
Woo
d pr
oduc
ts
1799
-4
.14%
-1
.88%
-0
.57%
-2
.45%
93
2 24
.05%
0.
78%
-0
.07%
0.
71%
17
99
9.95
%
0.67
%
-0.0
9%
0.58
%
9072
P
aper
, pap
er p
rodu
cts
and
prin
ting
indu
stry
26
39
0.62
%
0.42
%
0.00
%
0.42
%
2711
16
.57%
2.
89%
0.
00%
2.
88%
26
39
8.60
%
2.78
%
0.00
%
2.78
%
1355
4
Oil
ref
inin
g, c
oal,
coki
ng
and
coal
pro
duct
s
2334
5 -4
.99%
-1
4.45
%
6.39
%
-8.0
6%
1380
2 1.
14%
0.
04%
0.
17%
0.
20%
23
345
-1.9
2%
-0.5
7%
0.43
%
-0.1
4%
1342
5
Che
mic
al in
dust
ry,
excl
udin
g oi
l ref
inin
g
4181
3.
64%
20
.60%
4.
42%
25
.02%
58
31
14.8
5%
13.4
9%
0.28
%
13.7
7%
4181
9.
24%
13
.79%
0.
45%
14
.24%
26
393
Rub
ber
and
plas
tic
prod
ucts
37
80
0.60
%
0.50
%
0.60
%
1.10
%
3877
15
.19%
3.
17%
0.
01%
3.
19%
37
80
7.90
%
3.06
%
0.04
%
3.10
%
1700
5
Bui
ldin
g m
ater
ials
and
ot
her
non-
met
allic
m
iner
als
1817
1.
77%
1.
83%
-2
.05%
-0
.22%
19
21
12.3
6%
3.83
%
-0.0
9%
3.73
%
1817
7.
06%
3.
74%
-0
.17%
3.
57%
71
14
Bas
ic m
etal
s
4313
0.
26%
-0
.06%
1.
87%
1.
81%
43
31
13.5
8%
5.89
%
0.11
%
6.00
%
4313
6.
92%
5.
65%
0.
18%
5.
83%
15
930
Fabr
icat
ed m
etal
s
1962
1.
02%
2.
28%
0.
00%
2.
28%
22
47
17.4
6%
2.32
%
-0.0
1%
2.31
%
1962
9.
24%
2.
32%
-0
.01%
2.
31%
11
466
Mac
hine
ry
2086
4.
33%
23
.95%
0.
00%
23
.95%
29
37
17.5
6%
11.8
3%
0.02
%
11.8
5%
2086
10
.95%
12
.34%
0.
02%
12
.36%
17
488
tran
spor
t equ
ipm
ent
2122
7.
49%
11
.95%
0.
11%
12
.05%
35
55
21.3
1%
12.2
3%
0.57
%
12.8
0%
2122
14
.40%
12
.22%
0.
55%
12
.77%
32
990
Ele
ctri
cal m
achi
nery
and
eq
uipm
ent
27
60
4.09
%
8.56
%
0.25
%
8.80
%
4118
17
.74%
5.
32%
0.
17%
5.
49%
27
60
10.9
2%
5.46
%
0.17
%
5.63
%
2175
1
Ele
ctro
nic
and
tele
com
m
achi
nery
22
19
8.87
%
10.2
0%
0.50
%
10.7
0%
4791
19
.99%
7.
01%
2.
04%
9.
05%
22
19
14.4
3%
7.14
%
1.97
%
9.11
%
3630
7
inst
rum
ents
25
31
2.08
%
0.78
%
0.00
%
0.78
%
2791
14
.50%
0.
84%
-0
.03%
0.
81%
25
31
8.29
%
0.84
%
-0.0
3%
0.81
%
1222
2 Fu
rnitu
re
1339
0.
45%
-0
.15%
0.
00%
-0
.15%
12
39
21.5
1%
0.41
%
-0.0
5%
0.36
%
1339
10
.98%
0.
38%
-0
.04%
0.
34%
98
33
Oth
er m
anuf
actu
ring
23
62
2.21
%
1.56
%
-0.5
9%
0.98
%
2715
14
.23%
2.
43%
-0
.30%
2.
13%
23
62
8.22
%
2.39
%
-0.3
1%
2.09
%
1131
4 T
otal
man
ufac
turi
ng
3090
2.
32%
75
.86%
24
.14%
10
0%
3727
15
.92%
96
.69%
3.
31%
10
0%
3090
9.
12%
95
.82%
4.
18%
10
0%
1833
4 N
ote:
Val
ue a
dded
in a
t con
stan
t 198
0 yu
an. L
abou
r pr
oduc
tivi
ty (
LP
) le
vel,
Po,
is in
yua
n/pe
rson
. Gro
wth
rat
es a
re a
vera
ges
of th
e ye
ar to
yea
r gr
owth
rat
es.
Sou
rce:
Tim
e se
ries
fro
m S
zirm
ai e
t al.,
200
5. C
over
age:
ent
erpr
ises
at t
owns
hip
leve
l and
abo
ve w
ith
inde
pend
ent
acco
unti
ng s
yste
ms.
Aft
er 1
998:
sta
te e
nter
pris
es
plus
ent
erpr
ises
of
desi
gnat
ed s
ize
wit
h m
ore
than
fiv
e m
illio
n yu
an.
119
Productivity Growth and Structural Change
Chapter 6
120
From 1980 to 2002, labour productivity grew by 9.1 per cent per year on average,
with low productivity growth in the first decade of growth without catch up and
rapidly accelerating productivity growth in the catch up period after 1990. The
contribution of shift effects differs notably between the period 1980-1990 and the
period 1990-2002. In the first period, there is a major shift effect, accounting for 24.1
per cent of aggregate productivity growth. Just when productivity growth accelerates
in the second period, the shift effect drops to a mere 3.3 per cent. Almost all
productivity growth in the second period is accounted for by within sector
productivity sectors. With the exception of oil refining sector, all sectors have double
digit productivity growth after 1990. Over the whole period 1980-2002, the greatest
contributions to aggregate productivity growth derive from chemicals, transport
equipment machinery and electronic machinery and telecom equipment. These are the
dynamic sectors driving productivity growth.
Table 6.3: Decomposition of Manufacturing Productivity
- Contribution of Shifts between Technology Classes Annual growth
rate Intra effect
Shift effect Total
1980-1991 high-tech 9.2% 77.5% 1.9% 79.4%
low-tech 7.6% 19.8% 0.8% 20.6%
total 8.4% 97.4% 2.7% 100.0%
1992-2002 high-tech 21.9% 58.9% 1.1% 59.9%
low-tech 21.4% 40.1% 0.0% 40.1%
total 21.6% 98.9% 1.1% 100.0%
1980-2002 high-tech 15.8% 59.3% 1.1% 60.4%
Low-tech 14.8% 39.6% 0.02% 39.6%
total 15.3% 98.9% 1.1% 100.0%
Sources: see Table 6.2.
In Table 6.3, we classify the 21 sectors into two groups according to their technology
intensity7: high-tech sectors (including high tech and medium-high tech industries
according to the standard OECD classification) and low-tech sectors (including
low-tech and medium-low tech industries). Table 6.3 indicates that the high-tech
sectors account for most of productivity growth. Shifts between low and high tech
sectors are unimportant for productivity growth.
7 It is hard to apply the OECD classifications directly to the available Chinese manufacturing data, due
to lack of detail in the Chinese series. At this stage, we distinguish only two groups instead of the usual three.
Productivity Growth and Structural Change
121
6.4.2 Institutional Change8
We decompose productivity growth in industry into six ownership categories, with
state-owned and collective ownership primarily representing the older types of
collective ownership deriving from a centrally planned economy and foreign and
Hong Kong, Macao and Taiwan funded 9 , private and joint stock companies
representing newer forms of ownership. Within the collective sector, Township and
Village enterprises represent a dynamic emerging semi-public semi-private sector.10
The discrepancies between the national totals as published and sum of the five
ownership categories mentioned above are categorised as “other”. This category was
not important in the 1980s, but is increasingly important in the later periods of our
study.
Table 6.4 distinguishes four sub-periods, 1980-85, 1985-1992, 1992-97 and
1997-2002. The periods represent different phases of the Chinese growth and reform
experience, but the exact choice of years is also determined by data availability and
data consistency within the periods.
For manufacturing the period 1980-85 represents the pre-reform period. The Chinese
reform process started in 1978 in agriculture, but the basic institutions in industry
remained unchanged till the end of 1984. There were some limited efforts to enlarge
enterprise autonomy and increase financial incentives within the framework of the
traditional system. An important change in this period was the opening up to foreign
direct investment from 1981 onwards. The Open Door Policy brought China
remarkable inflows of FDI, increasing from 0.6 billion $ in 1983 to 1.7 billion $ in
1985.
In the period 1985-1992 the reform process took off in earnest. In October 1984, the
Third Plenary Session of 12th National Congress of CPC (shierjie sanzhong quanhui)
8 Some of the topics discussed in this section have also been introduced in Chapter 3. 9 We combine Foreign funded and Hong Kong, Macao and Taiwan funded as one category (Foreign + HK, MC and TW) in this sector because of their overlapping data from published yearbooks. 10 As the coverage of the data in table 6.4 on the next page only refers to enterprises at township level and above, the collective enterprises only include the TVEs at township level. Village level TVEs are not included in this dataset (see chapter 4 on data issues).
Chapter 6
122
made sweeping further decisions about further economic reforms. From the
mid-eighties onwards, the managerial independence of state enterprises was enhanced.
Government control was relinquished or decentralised. In 1987, private enterprises
were formally recognised. They were legalised in 1988. Between 1987 and 1992 a
contract responsibility system was introduced and firms were increasingly exposed to
market influences. The concept of Township and Village enterprises was introduced in
198411. TVEs are an interesting hybrid of private and public entrepreneurship which
turned out to be one of the motors of Chinese manufacturing growth. After 1989, an
austerity programme was imposed in response to accelerating inflation, which
dampened growth performance in the short run.
1992-1997 was a period of acceleration of the reform process. Reform accelerated
after 1992, in response to Deng Xiaoping’s journey through the South of China. In
response to Deng's theory, the 14th National Congress of CPC (shisi da) set up a
reform framework of "Constructing a socialism market economy mechanism". In
1994 there were initiatives to introduce a modern corporate system. A new company
law was approved in 1993 and implemented in 1994. The enterprise became an
independent legal entity, with separation of ownership and management. Two new
forms of ownership were introduced: Limited liability companies (without size limits)
and shareholding companies. Many state owned companies were transformed into
shareholding companies. The trade regime was also reformed in 1994, with
unification of the dual exchange rate system and introduction of a managed float
system. The new trade and market environment encouraged foreign investors.
Between 1992 and 1997, there was an almost tenfold increase in the inflow of FDI.
The period 1997-2002 represents the maturing of the reform process after a temporary
slowdown of the economy caused by the Asian crisis of 1997. In 1998 a constitutional
amendment stipulated that private ownership should be promoted and protected.
Private enterprises were formally sanctioned in 1999. The Law of Township and
Village Enterprises in the People's Republic of China, issued on 1 January 1997
formalised the position of TVEs. This period was also a period of dramatic
restructuring, especially of state-owned enterprises. Some 23 million workers were
11 For more details, see Chapter 3 (section 3.2.2).
Productivity Growth and Structural Change
123
made redundant after 1995, with the peak of the layoffs between 1997 and 1999. With
the access of Zhu Rongji to the prime ministership in 1998, a decisive programme of
restructuring was implemented. The size of the central government was cut by a third.
The process of privatising small and medium enterprises was speeded up. Foreign
investment continued to growth rapidly, though not as rapidly as in the previous
period.
In the first period (1980-85), the highest labour productivity levels are found in the
state-owned and foreign-owned sectors. Productivity growth in the foreign-owned
sector is much higher than in all other sectors. In the next three periods (1985-2002),
the foreign-owned sector keeps continuing to have the highest productivity level. But
for period 2 (1985-92) and period 3 (1992-97) the growth of productivity growth is
highest in the private sector. In the final period (1997-2002), productivity growth in
state-owned and joint-ownership sectors accelerates dramatically.
The decomposition results presented in Table 6.4 are extremely interesting. In the first
period, there are substantial negative shift effects (-24.85%), indicating shifts to less
productive ownership categories. Most of this is caused by the shift from the
state-owned to the collective-owned category. In the three other periods there are
substantial positive shift effects, of which the highest are found in period 3 (1992-97):
23.13 per cent. Since 1985, the expanding foreign-funded sector accounts for most of
the shift contribution, i.e. 20.43 % out of 21.52 % total during 1985-92, 15.88 % out
of 23.13 % total during 1992-97, and 8.90% out of 10.15% total during 1997-2002.
Thus structural change contributes substantially to productivity growth after 1985.
In order to analyse the impact of institutional changes at regional level, we break
down the institutional effects by region in Table 6.5. Due to the paucity of data on
ownership by region, this analysis is limited to data for the period since 1992. We
decompose the aggregate productivity growth of industrial enterprises (at township
level and above) by six ownership categories and thirty regions.12 Due to lack of data
for all intervening years, we use the discontinuous shift-share model of equation 1 for
this table.
12 Since 1997, Chongqing is an independent city, so we combine its data with those of Sichuan province to which it belonged before 1997.
T
able
6.4
: D
ecom
posi
tion
of
Indu
stri
al P
rodu
ctiv
ity
- C
ontr
ibut
ion
of S
hift
s in
Ow
ners
hipa
19
80-1
985
1985
-199
2b
L
P le
vel
(Po)
A
nnua
l LP
In
tra
Shi
ft
Tot
al
LP
leve
l (P
o)
Ann
ual
LP
In
tra
Shif
t T
otal
grow
th r
ate
grow
th
rate
Stat
e-ow
ned
4118
3.
66%
83
.00%
0.
00%
83
.00%
49
28
0.30
%
6.32
%
0.00
%
6.32
%
Col
lect
ive
1674
6.
65%
40
.01%
-2
5.00
%
15.0
1%
2310
6.
90%
58
.00%
0.
00%
58
.00%
For
eign
+H
K,M
C,T
W
4090
16
.05%
0.
49%
0.
32%
0.
81%
86
10
2.58
%
3.61
%
20.4
3%
24.0
3%
Pri
vate
12
68
8.88
%
0.01
%
-0.0
7%
-0.0
5%
1940
16
.16%
0.
16%
-0
.03%
0.
12%
Join
t-ow
ners
hip
3453
6.
13%
1.
32%
-0
.10%
1.
23%
46
50
2.96
%
1.02
%
0.31
%
1.33
%
Oth
ers
NA
N
A
0.00
%
0.00
%
0.00
%
NA
NA
9.
37%
0.
82%
10
.19%
Tot
al in
dust
ry
3263
3.
43%
12
4.85
%
-24.
85%
10
0.00
%
3864
3.
17%
78
.48%
21
.52%
10
0.00
%
1992
-199
7b 19
97-2
002
L
P le
vel
(Po)
A
nnua
l LP
In
tra
Shi
ft
Tot
al
LP
leve
l (P
o)
Ann
ual
LP
In
tra
Shif
t T
otal
grow
th r
ate
grow
th
rate
Stat
e-ow
ned
5034
5.
90%
37
.94%
0.
32%
38
.25%
67
03
23.1
3%
56.7
5%
0.00
%
56.7
5%
Col
lect
ive
3685
8.
67%
29
.37%
0.
00%
29
.37%
55
83
17.6
2%
15.4
2%
0.00
%
15.4
2%
For
eign
+H
K,M
C,T
W
1029
1 6.
47%
9.
88%
15
.88%
25
.75%
14
076
10.7
6%
13.1
0%
8.90
%
22.0
0%
Pri
vate
55
37
10.1
6%
0.22
%
0.83
%
1.05
%
8982
7.
37%
2.
69%
1.
34%
4.
03%
Join
t-ow
ners
hip
5705
3.
91%
0.
49%
0.
00%
0.
49%
69
13
21.7
8%
0.93
%
0.00
%
0.93
%
Oth
ers
9638
-1
.10%
-1
.02%
6.
10%
5.
09%
91
18
2.25
%
0.96
%
-0.1
0%
0.86
%
Tot
al I
ndu
stry
48
08
8.27
%
76.8
7%
23.1
3%
100.
00%
71
53
19.2
4%
89.8
5%
10.1
5%
100.
00%
N
otes
: a.
B
reak
dow
n by
ow
ners
hip
cate
gori
es i
s on
ly a
vaila
ble
for
tota
l in
dust
ry, i
nclu
ding
min
ing,
man
ufac
turi
ng a
nd u
tiliti
es. C
over
age:
Ent
erpr
ises
at
tow
nshi
p le
vel a
nd a
bove
, wit
h in
depe
nden
t acc
ount
ing
syst
ems.
Val
ue a
dded
in a
t con
stan
t 198
0 yu
an. L
P le
vel (
Po)
is in
yua
n/pe
rson
. b.
O
wne
rshi
p da
ta f
or 1
992
are
lack
ing.
199
2 to
tals
are
bro
ken
dow
n us
ing
1993
pro
port
ions
. S
ourc
e: 1
985
cens
us; 1
995
cens
us; C
IESY
, 199
8, 2
002
from
CSY
200
3.
Chater 6
124
Tab
le 6
.5:
Indu
stri
al P
rodu
ctiv
ity:
Shi
ft-S
hare
by
Ow
ners
hip
and
Reg
ion,
199
2-20
02
- C
ontr
ibut
ion
of I
nsti
tuti
onal
Shi
fts
by R
egio
na
19
92-1
997b
1997
-200
2
P o
A
nnua
l gro
wth
ra
te
Int
ra %
In
ter
%
Dyn
amic
%
Po
Ann
ual g
row
th
rate
I
ntra
%
Inte
r%
Dyn
amic
%
Tot
al
4808
8.
27%
78
.80%
16
.70%
4.
49%
10
0%
7153
19
.24%
89
.93%
12
.57%
-2
.50%
10
0%
Bei
jing
60
84
9.15
%
58.6
7%
33.4
4%
7.88
%
100%
94
26
19.0
5%
84.9
8%
10.5
3%
4.49
%
100%
T
ianj
in
3860
11
.04%
60
.13%
28
.81%
11
.06%
10
0%
6517
25
.29%
81
.50%
17
.04%
1.
46%
10
0%
Heb
ei
4255
11
.40%
93
.55%
4.
25%
2.
20%
10
0%
7301
16
.36%
91
.75%
6.
43%
1.
83%
10
0%
Shan
xi
3437
7.
41%
97
.13%
4.
20%
-1
.33%
10
0%
4914
15
.41%
82
.31%
-1
.61%
19
.29%
10
0%
Inne
r 29
42
11.4
4%
93.1
7%
2.85
%
3.98
%
100%
50
57
22.1
4%
75.7
9%
5.95
%
18.2
6%
100%
L
iaon
ing
4901
0.
13%
-5
02.4
375
7.31
%
-154
.88%
10
0%
4934
26
.32%
77
.14%
12
.14%
10
.71%
10
0%
Jilin
37
03
3.70
%
68.0
0%
48.4
2%
-16.
42%
10
0%
4440
30
.63%
84
.13%
3.
64%
12
.23%
10
0%
Hei
long
jian
4192
9.
95%
10
1.40
%
-0.0
2%
-1.3
7%
100%
67
35
26.2
1%
81.6
6%
5.92
%
12.4
1%
100%
Sh
angh
ai
8078
11
.81%
57
.99%
53
.76%
-1
1.75
%
100%
14
1115
.83%
90
.29%
10
.94%
-1
.23%
10
0%
Jian
gsu
5008
8.
73%
81
.39%
10
.20%
8.
41%
10
0%
7609
20
.39%
95
.06%
13
.91%
-8
.97%
10
0%
Zhe
jian
g 47
63
9.86
%
90.3
9%
25.3
4%
-15.
73%
10
0%
7624
17
.12%
15
2.90
%
-1.5
9%
-51.
31%
10
0%
Anh
ui
4229
10
.29%
90
.10%
-0
.13%
10
.03%
10
0%
6902
14
.09%
62
.19%
10
.63%
27
.18%
10
0%
Fuj
ian
4729
12
.08%
77
.15%
8.
60%
14
.25%
10
0%
8363
17
.82%
10
1.14
%
7.75
%
-8.8
9%
100%
Ji
angx
i 35
37
2.86
%
68.1
9%
27.7
0%
4.10
%
100%
40
72
21.8
4%
88.5
5%
12.3
6%
-0.9
2%
100%
Sh
ando
ng
6001
6.
70%
86
.81%
7.
53%
5.
65%
10
0%
8302
16
.94%
10
7.15
%
25.7
6%
-32.
91%
10
0%
Hen
an
3649
11
.09%
95
.11%
-1
.16%
6.
04%
10
0%
6173
15
.00%
98
.01%
1.
32%
0.
67%
10
0%
Hub
ei
5186
8.
38%
91
.03%
10
.42%
-1
.45%
10
0%
7755
16
.17%
82
.81%
4.
85%
12
.35%
10
0%
Hun
an
2887
13
.25%
94
.15%
8.
09%
-2
.23%
10
0%
5378
20
.01%
86
.51%
23
.33%
-9
.84%
10
0%
Gua
ngdo
ng
7426
8.
78%
66
.29%
20
.62%
13
.08%
10
0%
1130
11.5
4%
119.
82%
15
.93%
-3
5.75
%
100%
G
uang
xi
5316
2.
21%
82
.98%
1.
83%
15
.19%
10
0%
5930
16
.97%
87
.48%
6.
48%
6.
03%
10
0%
Hai
nan
6012
0.
07%
0.
30%
34
9.74
%
-250
.04%
10
0%
6035
24
.60%
97
.05%
-0
.61%
3.
56%
10
0%
Sich
uan
3792
5.
26%
75
.93%
34
.20%
-1
0.12
%
100%
49
01
23.5
4%
84.2
8%
9.53
%
6.18
%
100%
G
uizh
ou
4750
3.
25%
94
.47%
11
.64%
-6
.11%
10
0%
5574
16
.58%
86
.25%
4.
53%
9.
23%
10
0%
Yun
nan
7381
11
.09%
10
0.90
%
-1.0
3%
0.13
%
100%
12
4817
.29%
82
.08%
12
.59%
5.
33%
10
0%
Tib
et
4183
19
.37%
88
.06%
0.
01%
11
.94%
10
0%
1013
1.58
%
103.
74%
-2
0.65
%
16.9
1%
100%
Sh
aan
xi
3879
3.
08%
60
.12%
45
.36%
-5
.48%
10
0%
4513
24
.54%
87
.46%
6.
05%
6.
49%
10
0%
Gan
su
4121
5.
73%
88
.50%
10
.27%
1.
23%
10
0%
5446
17
.14%
99
.91%
2.
20%
-2
.11%
10
0%
Qin
ghai
47
75
5.64
%
68.3
2%
90.7
5%
-59.
06%
10
0%
6284
22
.25%
93
.51%
2.
53%
3.
96%
10
0%
Nin
gxia
33
58
9.31
%
74.5
4%
4.54
%
20.9
2%
100%
52
41
15.6
1%
87.6
5%
10.7
3%
1.62
%
100%
X
inji
ang
4656
14
.65%
95
.34%
1.
56%
3.
10%
10
0%
9222
23
.10%
86
.69%
4.
25%
9.
06%
10
0%
Not
e: a
. T
he r
egio
nal
prod
ucti
vity
(P
o) i
s ca
lcul
ated
fro
m t
he s
um o
f fi
ve o
wne
rshi
p ty
pes
(sta
te-o
wne
d, c
olle
ctiv
e, f
orei
gn f
unde
d, p
riva
te,
join
t-ow
ners
hip
com
pani
es).
Due
to th
e fa
ct th
at s
ome
"unc
lass
ifie
d" e
nter
pris
es a
re n
ot in
clud
ed h
ere,
the
Po
colu
mns
are
dif
fere
nt f
rom
thos
e in
Tab
le 6
.7.
b.
Cho
ngqi
ng u
sed
to b
e a
part
of
Sich
uan
prov
ince
and
onl
y be
cam
e cl
assi
fied
as
an i
ndep
ende
nt c
ity
sinc
e 19
97.
We
have
mer
ged
its
data
for
199
7 an
d 20
02 in
to S
ichu
an in
ord
er to
be
cons
iste
nt w
ith
1992
. Val
ue a
dded
at 1
980
cons
tant
pri
ces.
LP
leve
l (P
o) is
in y
uan/
pers
on.
c. D
ue to
lack
of
deta
il f
or 1
992,
reg
iona
l agg
rega
tes
for
1992
are
bro
ken
dow
n in
to o
wne
rshi
p ca
tego
ries
usi
ng p
ropo
rtio
ns o
f 19
93.
S
ourc
e: C
IESY
and
CSY
var
ious
issu
es.
Productivity Growth and Structural Change
125
Chapter 6
126
From 1992 till 1997, the intra-sectoral effect accounts for 78.80 per cent of aggregate
productivity growth (see top row of Table 6.5).13 The regions with the strongest
effects of structural change are basically coastal regions: Beijing, Tianjin, Jilin,
Shanghai and Guangdong, But structural change is also important in some of the
non-coastal regions such as Shaanxi, Qinghai, Ningxia, Jiangxi and Sichuan. Liaoning
and Hainan have a very high negative dynamic shift effects, but the percentages are
somewhat misleading because the aggregate growth rates are very low. In these two
regions, the positive effects of reallocation to more productive ownership categories,
are counterbalanced by slow growth within each category and negative dynamic
effects.
In the period 1997-2002, the institutional shifts become less important and the
intra-category effects increase. All sectors are increasing their productivity and within
effects predominate. But at regional level, some regions see much more effects of
structural change than others. This is not limited to coastal regions. In inner regions
such as Yunnan and Inner Mongolia institutional change contributes very positively to
productivity growth, just as in coastal regions such as Liaoning, Tianjin, and
Heilongjiang. In three regions – Zhejian, Shandong and interestingly enough
Guangdong - there are very strong negative dynamic shift effects, which counteract
high productivity growth within the ownership categories. This indicates that the
expanding sectors are growing less rapidly than the regional average.
6.4.3 Regional Shifts
Due to the Hukou system, people do not change their jobs frequently in China. In the
long run, nevertheless, there is still some evidence of labour shift among regions. The
following table presents the shares of coastal, middle and western regions in terms of
labour and value added.
13 There is a slight discrepancy between the aggregate results in Table 6.4 and Table 6.5. There are two
reasons for this discrepancy. First, the shift and share analysis in Table 6.5 is based on the sum of the ownership categories, excluding ‘other’. The reason for this is that the residual category ‘others’ is negative for some regions. Table 6.4 is based on the published national data, which do include other forms of ownership. The other reason is that the two tables use the two different shift-share models. If one combines the static and dynamic shift effects from Table 6.5, the results are actually very close to the shift effect in Table 6.4.
Productivity Growth and Structural Change
127
Table 6.6 Regional Shares in Industrial Employment and Value Added
(1978 and 2002)
1978 2002
coast middle western coast middle Western
10000 persons 2142 1244 594 3625 1581 734 Employment
share (%) 0.54 0.31 0.15 0.61 0.27 0.12
100 mill yuan 818 354 201 5997 1974 1003 Value added
share (%) 0.60 0.26 0.15 0.67 0.22 0.11
Note: Value added is at 1978 constant prices. Source: CSY, various issues, CIESY, various issues.
Table 6.7 examines the contributions of regional shifts to aggregate productivity
growth for three periods – 1985-92, 1992-97 and 1997-2002.14 The column Total
shows the percentage contribution of each region. The columns intra and shift break
down the regional contribution into that of intraregional growth and interregional
shifts. The main conclusion derived from this table is that interregional shifts
contribute only very modestly to aggregate growth.
As before, there are striking differences between the sub-periods. In the first period,
there is a small negative shift effect, indicating that the expanding regions have lower
levels of productivity than the contracting regions. In the second and third periods
there are positive shift effects, increasing to 6.18 per cent between 1997 and 2002.
The three periods are characterized by different productivity champions. In the first
period (1985-92), Guangdong is the most prominent contributor (accounting for no
less than 24.86 per cent of total productivity growth), followed by Jiangsu and
Zhejiang. In the second period (1992-97), Guangdong, Shandong and Heilongjiang
are the three regions with the highest contributions. In the third period, Guangdong,
Shandong, Shanghai, Jiangsu, Zhejiang, Heilongjiang and Liaoning together explain
54 per cent of total productivity growth. Without a single exception the productivity
champions are the coastal regions.
14 No regional data were available for the years 1980-1985.
Tabl
e 6.
7: D
ecom
posi
tion
of
Indu
stri
al P
rodu
ctiv
ity
- C
ontr
ibut
ion
of R
egio
nal S
hift
s, 1
985-
2002
19
85-1
992
1992
-199
7 19
97-2
002
L
P le
vel
(Po)
198
5
Ann
ual L
P gr
owth
ra
te
Intr
a Sh
ift
Tota
l
LP
leve
l (P
o)
1992
Ann
ual L
P gr
owth
ra
te
Intr
a Sh
ift
Tota
l
LP
leve
l (P
o)
1997
A
nnua
l LP
grow
th r
ate
Intr
a Sh
ift
Tota
l B
eiji
ng
5587
2.
82%
3.
18%
0.
00%
3.
18%
67
86
6.91
%
2.82
%
0.21
%
3.04
%
9477
18
.92%
2.
78%
0.
01%
2.
79%
T
ianj
in
4726
-1
.01%
-0
.83%
0.
00%
-0
.83%
44
01
13.2
0%
3.37
%
0.00
%
3.37
%
8181
19
.71%
2.
47%
0.
07%
2.
54%
H
ebei
38
66
1.18
%
1.64
%
-0.2
5%
1.39
%
4196
11
.83%
6.
43%
-0
.07%
6.
36%
73
40
16.2
3%
3.93
%
-0.0
1%
3.92
%
Shan
xi
3181
-0
.25%
-0
.20%
-0
.10%
-0
.30%
31
26
9.58
%
2.59
%
-0.1
4%
2.45
%
4940
15
.29%
1.
77%
-0
.18%
1.
59%
In
ner
Mon
golia
20
79
3.40
%
1.34
%
-0.4
4%
0.89
%
2627
14
.12%
2.
12%
-0
.05%
2.
07%
50
84
22.0
1%
1.56
%
-0.1
9%
1.37
%
Lia
onin
g 37
79
0.30
%
0.75
%
0.00
%
0.75
%
3860
5.
14%
3.
98%
-0
.04%
3.
94%
49
60
26.1
8%
5.99
%
0.00
%
5.99
%
Jilin
24
85
3.49
%
2.80
%
0.00
%
2.80
%
3159
7.
16%
2.
05%
-0
.02%
2.
03%
44
64
30.5
0%
3.30
%
-0.1
3%
3.17
%
Hei
long
jian
g 26
16
4.34
%
6.08
%
-0.6
3%
5.45
%
3521
13
.97%
7.
37%
-0
.02%
7.
35%
67
71
26.0
8%
6.10
%
-0.2
2%
5.88
%
Shan
ghai
78
01
1.49
%
4.05
%
0.00
%
4.05
%
8650
7.
22%
5.
95%
0.
00%
5.
95%
12
257
19.1
5%
5.84
%
0.77
%
6.61
%
Jian
gsu
4528
6.
66%
18
.97%
0.
00%
18
.97%
71
10
1.47
%
2.21
%
2.12
%
4.33
%
7649
20
.27%
9.
38%
1.
13%
10
.51%
Z
heji
ang
4865
6.
35%
10
.41%
0.
00%
10
.41%
74
87
0.47
%
0.36
%
1.36
%
1.73
%
7665
17
.00%
3.
99%
1.
74%
5.
73%
A
nhui
32
19
1.14
%
0.94
%
-0.4
4%
0.49
%
3486
14
.76%
5.
38%
-0
.13%
5.
24%
69
38
13.9
7%
1.86
%
0.00
%
1.86
%
Fuj
ian
3008
5.
63%
3.
24%
-0
.44%
2.
79%
44
13
13.7
6%
3.89
%
0.09
%
3.98
%
8408
17
.69%
3.
30%
0.
02%
3.
31%
Ji
angx
i 26
01
2.15
%
1.18
%
0.00
%
1.18
%
3019
6.
28%
1.
17%
-0
.12%
1.
05%
40
94
21.7
1%
1.41
%
0.00
%
1.41
%
Shan
dong
43
04
4.75
%
10.3
6%
-0.4
2%
9.94
%
5957
6.
98%
7.
52%
0.
73%
8.
25%
83
46
16.8
1%
8.83
%
0.47
%
9.30
%
Hen
an
2976
4.
02%
4.
94%
-0
.48%
4.
46%
39
21
9.61
%
5.12
%
-0.2
4%
4.88
%
6205
14
.88%
3.
73%
-0
.17%
3.
55%
H
ubei
37
89
1.43
%
1.98
%
0.00
%
1.98
%
4184
13
.25%
7.
07%
0.
00%
7.
06%
77
97
16.0
4%
3.51
%
0.00
%
3.51
%
Hun
an
3718
0.
68%
0.
71%
-0
.07%
0.
63%
38
99
6.75
%
2.46
%
-0.2
8%
2.18
%
5406
19
.89%
2.
36%
0.
00%
2.
36%
G
uang
dong
37
46
11.8
6%
24.6
4%
0.22
%
24.8
6%
8209
6.
73%
8.
53%
1.
96%
10
.49%
11
369
11.4
2%
6.39
%
3.27
%
9.67
%
Gua
ngxi
31
20
5.35
%
2.61
%
-0.0
7%
2.54
%
4494
5.
81%
1.
10%
0.
04%
1.
14%
59
61
16.8
5%
1.20
%
-0.0
5%
1.15
%
Hai
nan
973
21.9
0%
0.79
%
-0.0
5%
0.75
%
3894
13
.06%
0.
33%
0.
01%
0.
34%
71
93
20.3
0%
0.27
%
-0.0
1%
0.26
%
Sich
uan
4205
-1
.81%
-3
.54%
-0
.06%
-3
.60%
37
00
5.90
%
3.37
%
-0.2
5%
3.11
%
4927
23
.41%
5.
18%
0.
00%
5.
18%
G
uizh
ou
3765
0.
94%
0.
38%
0.
00%
0.
38%
40
21
6.86
%
0.92
%
-0.0
3%
0.89
%
5604
16
.46%
0.
84%
-0
.11%
0.
73%
Y
unna
n 32
51
10.9
3%
5.88
%
0.00
%
5.88
%
6721
13
.31%
3.
54%
0.
26%
3.
80%
12
553
17.1
7%
2.24
%
0.11
%
2.36
%
Tib
et
1325
12
.00%
0.
06%
0.
00%
0.
06%
29
28
28.3
4%
0.09
%
0.00
%
0.09
%
1019
3 1.
47%
0.
01%
0.
00%
0.
00%
Sh
aanx
i 30
99
1.12
%
0.72
%
0.00
%
0.72
%
3350
6.
25%
1.
33%
-0
.23%
1.
11%
45
37
24.4
1%
2.13
%
-0.1
1%
2.02
%
Gan
su
4318
-1
.64%
-0
.81%
-0
.22%
-1
.03%
38
47
7.31
%
1.09
%
-0.0
8%
1.02
%
5475
17
.01%
1.
09%
-0
.14%
0.
95%
Q
ingh
ai
3230
-0
.22%
-0
.02%
0.
00%
-0
.02%
31
81
8.41
%
0.22
%
-0.0
1%
0.21
%
4764
29
.21%
0.
38%
-0
.01%
0.
36%
N
ingx
ia
3203
0.
11%
0.
01%
-0
.13%
-0
.12%
32
28
10.3
0%
0.36
%
-0.0
2%
0.34
%
5269
15
.49%
0.
25%
-0
.05%
0.
20%
X
inji
ang
3115
5.
20%
1.
53%
-0
.20%
1.
33%
44
44
15.8
4%
2.19
%
0.02
%
2.21
%
9271
22
.97%
1.
75%
-0
.03%
1.
72%
To
tal
3864
3.
17%
10
3.79
%
-3.7
9%
100.
00%
48
07
8.27
%
94.9
3%
5.07
%
100.
00%
71
54
19.2
4%
93.8
2%
6.18
%
100.
00%
N
ote:
Val
ue a
dded
at 1
980
cons
tant
pri
ce. L
P (l
abou
r pr
oduc
tivi
ty)
leve
l (P
o) in
yua
n/pe
rson
. Cho
ngqi
ng is
incl
uded
in S
ichu
an p
rovi
nce.
S
ourc
e: C
SY
, var
ious
issu
es, C
IES
Y, v
ario
us is
sues
.
Chapter 6
128
Productivity Growth and Structural Change
129
Using the same dataset as Table 6.5 (paragraph 5.2), we finally analyse the
contribution of regional shifts within each of the ownership categories. The results are
reproduced in Table 6.8. In the first period (1992-1997) negative effects of regional
change predominate. In four of the six categories the combined effects of static and
dynamic shifts are negative. Net positive effects of regional change are found for
collective enterprises and foreign-funded enterprises. Very large positive static shift
effects are found for private enterprises, but these are more than compensated for by
even larger negative dynamic shift effects. The same holds for the joint ownership
category.
Table 6.8: Industrial Productivity: Shift-Share by Region and Ownership,
- Contribution of Regional Shifts by Institutional Categories, 1992-2002
1992-1997 1997-2002
LP level (Po) 1992
Annual growth rate Intra Inter Dynamics
LP level (Po) 1997
Annual growth rate Intra Inter Dynamics
State-owned 5034 5.90% 105.69% -2.42% -3.27% 100% 6703 23.13% 100.92% 0.72% -1.65% 100%
Collective 3685 8.67% 97.16% -0.10% 2.94% 100% 5583 17.62% 93.24% 2.93% 3.83% 100%
Foreign+
HK, M and TW 10291 6.47% 14076 10.76%
Foreign
funded 13064 4.23% 90.99% 17.60% -8.59% 100% 16069 13.75% 101.01% 2.52% -3.53% 100%
HK, M & T 8301 7.67% 101.97% -8.62% 6.65% 100% 12012 8.47% 112.90% 1.20% -14.10% 100%
Private 5537 10.16% 134.89% 59.59% -94.48% 100% 8982 7.37% 92.43% 20.09% -12.52% 100%
joint-ownership 5705 3.91% 109.63% 11.51% -21.14% 100% 6913 21.78% 82.41% 5.42% 12.17% 100%
Others NA NA
Note: Value added is at 1980 constant prices. LP level (Po) is in yuan/person. Source: CIESY and CSY various issues.
In the second period 1997-2002, the private enterprises again have the highest static
and dynamic shift effects. In this period the net effect of static and dynamic shifts in
this category is positive. It is worth noting that foreign-funded enterprises have the
highest labour productivity in both periods, but within this category regional shift
effects do not seem to be very important in the second period. Regional shifts have
quite substantial positive effects on productivity growth in the collective sector, the
private sector and the joint-ownership sector. There are large negative dynamic shift
effects in private and Hong Kong Macao and Taiwan funded enterprises.
Chapter 6
130
6.5 Conclusions
This chapter focused on the contribution of structural change to industrial productivity
in the period 1980-2002. Using shift and share techniques three dimensions of change
were examined - sectoral, institutional and regional.
Overall productivity growth was slow in the 1980s, but accelerated dramatically from
1990 onwards. In the 1980s we found clear evidence of a structural change bonus at
sectoral level, with sectoral shifts contributing 24 per cent to overall productivity
growth in manufacturing. However, just when productivity growth accelerated in the
1990s, the contribution of the shift effect dropped to a mere 3.3%. Our interpretation
of this phenomenon is that the structural changes in the early reform period of the
1980s resulted in a more efficient economic structure, which provided a foundation
for rapid intra-sectoral productivity growth after the 1990s.
In marked contrast to sectoral changes, changes in the ownership structure contributed
negatively to overall productivity growth in the early 1980s. There was a negative
shift effect of around 25 per cent. This turned positive after 1985, reaching a peak of
23 per cent in the period 1992-1997, just when the shift effects of sectoral change
were negligible. The conclusion is that the reform of the ownership structure
contributed very substantially to the acceleration of productivity growth after 1992.
This is consistent with other more descriptive accounts of the Chinese reform process.
Institutional change has been especially important in the coastal regions. The
interesting contrast between the timing of the effects of sector structure and ownership
structure merits further examination.
In terms of contributions to productivity growth, the importance of the coastal regions
is confirmed by our analysis for all periods. For instance, between 1997 and 2002,
seven regions - Guangdong, Shandong, Shanghai, Jiangsu, Zhejiang, Heilongjiang
and Liaoning - together account for 54 per cent of total productivity growth.
The effects of regional change are much more modest than those of sectoral and
institutional change. Regional shifts contributed negatively to aggregate productivity
Productivity Growth and Structural Change
131
growth before 1992, and positively after 1992. During the period 1997-2002, there
was a positive shift effect of 6.18 per cent. Therefore, like institutional change,
regional change contributed positively to the acceleration of productivity growth.
Combining regional and ownership changes in this period, we found that positive
effects of regional change were found in the joint-ownership category, the
private-owned category and the collective-owned category. Foreign-funded
enterprises had the highest productivity levels, but these were hardly affected by
regional shifts.
CHAPTER 7
Regional Performance and Productivity Efficiency
7.1 Introduction
Rapid aggregate growth in Chinese industry is accompanied by large disparities
across regions. This is embodied both in terms of the personal income distribution and
in terms of regional income per capita. In 2004, the ratio of GDP per capita in the
richest region (Shanghai) to the poorest region (Guizhou) was more than 10 to 1.
Average GDP per capita in the five richest regions was 5.3 times as high as that of the
five poorest regions (CSY, 1996 and CSY, 2005). The benefits of growth are not
spread evenly. Rural areas in particular are being left behind.
One would expect the regional income differentials to be mirrored by regional
productivity differentials in manufacturing. Thus, one would expect regional
productivity differentials to be increasing over time. On the other hand, there is also a
tendency for manufacturing activities to shift land inwards, which may partially
counteract the increase in regional inequality. The hypothesis of increasing
interregional productivity differentials will be examined in this chapter.
A vast amount of empirical research indicates that regional disparities in GDP per
capita and other indicators of economic performance tend to increase in the course of
economic development (see also the literature review presented in chapter 2). Many
studies of regional inequality in China focus on income differentials and/or GDP per
capita. Jian et al (1996) explore China's regional disparities in terms of per capita
income during 1951-1993. They find convergence from 1952 to 1965 and from 1978
to 1990. Between 1965 and 1978 they find a clear divergence trend. The data set of
Hsueh and Li (1999) shows that per capita incomes among Chinese provinces are
Chapter 7
134
diverging since the open door reforms. Chen and Fleisher (1996) discuss the
convergence, conditional on investment in physical capital, employment growth,
human-capital investment, FDI and coastal location, across 25 provinces in China
from 1978-1993. They conclude that overall regional inequality as measured by the
coefficient of variation is likely to decline modestly but that the coast/non-coast
income differentials tend to increase somewhat. Convergence is primarily occurring
"within" the coast and non-coast groups rather than happening "between" them. Tsui
(2007) also finds a sustained decline in within-region inequality from the mid 1970s
to the end of the 1990s. Kanbur and Zhang (2005) use econometric analysis to study
the determinants of regional inequality. They find that regional inequality is explained
by three variables: the share of heavy industry in gross output value, the degree of
fiscal decentralization, and the degree of openness. The importance of these factors
varies from policy period to policy period. The authors conclude that "the heavy-
industry development strategy played a key role in forming the enormous rural–urban
gap in the pre-reform period, while openness and decentralization contributed to the
rapid increase in inland–coastal disparity in the reform period of the 1980s and 90s."
Using 3-dimensional panel data (covering benchmark years 1995 and 2004, 28
industries and 30 provinces), Chen, van Ark, and Wu (2008) find that there is a strong
convergence trend in labour compensation, productivity and unit labour cost across
Chinese provinces and regions. However, there is a divergence trend in capital
intensive and high skill intensive industries. From the perspective of regional
efficiency performance, using panel data of 27 Chinese provinces during 1981-1995,
Wu (2000) finds that the technical efficiency in Chinese regions has converged
rapidly since the early 1980s.
The main goal of this chapter is to provide an examination of the evolution of
productivity differentials in Chinese industry. In line with the method employed by
Wu (2000), we would like to examine total factor productivity and efficiency
differentials in Chinese industry. This requires estimates of regional capital inputs.
Estimations of capital stocks in China are fraught with difficulties, in part due to the
lack of long-run series of capital investment consistent with system of national
accounts (SNA). Several research groups have been working on capital stock
estimation for China (e.g. (e.g. Holz, 2006; Wu and Xu, 2002; Chow, 1993; Huang et
Regional Performance and Productivity Efficiency
135
al. 2002 and Wu (2004). In chapter five of this thesis, we explained how we had
constructed new regional capital stocks. In this chapter we will use these regional
capital stocks to analyse the regional differences in total factor productivity (TFP) and
efficiency.
7.2 Methods
Conventional convergence studies use regression methods to calculate the catching-up
rate (convergence coefficient β ) and dispersion rate (σ) (Barro and Sala-i-Martin,
2004). If poorer countries (or regions) grow faster than rich ones, then we say there is
a β convergence. The coefficient β also shows the rate of convergence. There is a
negative relationship between the growth rate (of GDP or income per capita) and the
initial level of income per capita. σ measures the dispersion, based on the standard
deviation of observations. The Coefficient variation is also used for σ measurement.
The shortcoming of these convergence indicators is that the distribution is described
with a single indicator. This does not give an adequate picture of the growth dynamics
of all the regions or countries involved. Hence mapping the entire cross-section
distribution is needed (Bianchi, 1997; Desdoigts, 1994).
Distribution dynamics (distribution density) has been accepted as a better solution in
analyzing a rather broad or the whole scale of cross-country distribution (Quah, 1996;
Lopez-Bazo, 1999; Bianchi, 1997; Fiaschi and Lavezzi, 2003; Bulli, 2001; Desdoigts,
1996). In addition to traditional measures such as the coefficient of variation this
chapter uses the distribution density to illustrate the distribution of regional GDP per
capita, productivity and growth.
7.3 Analysis
7.3.1 Regional Differentials in GDP per Capita
Chapter 7
136
There are huge regional differentials with regard to GDP per capita. In 1978, GDP per
capita in Shanghai, the top region, is 13 times higher than the bottom region (Guizhou)
in 1978. In 2005, it is 8.7 times higher. Over the whole period 1978-2005, the ratio of
the top five to the bottom five regions declines from 5.59 in 1978 to 4.82 in 2005.
Zhejiang, Guangdong, Fujian, and Shandong maintain growth rates of more than 11%
per year.
Table 7.1: Standard Deviation, Mean and Coefficient of Variation of per Capita
GDP in Chinese Regions, 1978-2005 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987
standard
deviation 444 445 458 450 451 469 510 551 544 558
mean 457 489 508 526 559 605 689 754 778 832
coefficient of
variation 0.97 0.91 0.90 0.85 0.81 0.77 0.74 0.73 0.70 0.67
1988 1989 1990 1991 1992 1993 1994 1995 1996 1997
standard
deviation 557 504 525 606 712 842 900 963 1060 1214
mean 866 816 875 966 1107 1266 1354 1452 1592 1768
coefficient of
variation 0.64 0.62 0.60 0.63 0.64 0.67 0.66 0.66 0.67 0.69
1998 1999 2000 2001 2002 2003 2004 2005
standard
deviation 1370 1539 1737 1918 2125 2461 2816 3023
mean 1952 2137 2372 2620 2919 3339 3849 4491
coefficient of
variation 0.70 0.72 0.73 0.73 0.73 0.74 0.73 0.67
Note: at 1978 constant price, yuan/person.
Source: GDP and population from regional statistical yearbooks, various issues, price deflator from
CSY 2006 (Table 9-1) and CSY 1996 (Table 8-1).
Regional Performance and Productivity Efficiency
137
Table 7.2: GDP per Capita in Chinese Regions (at 1978 Constant Prices)
1978 1980 1985 1990 2000 2005
National
average 362 416 642 892 2178 3679
Beijing 1248 1431 2044 2537 5074 10686
Tianjin 1141 1286 1703 1935 4572 8461
Hebei 362 395 558 801 2129 3517
Shanxi 363 409 639 815 1587 2973
Inner Mongolia 318 339 634 812 1649 3897
Liaoning 675 750 1097 1492 3152 4528
Jilin 381 415 680 959 1982 3181
Heilongjiang 559 642 825 1110 2386 3443
Shanghai 2484 2531 2992 3242 9614 12288
Jiangsu 427 501 818 1151 3270 5845
Zhejiang 330 437 827 1166 3744 6548
Anhui 242 268 501 639 1392 2096
Fujian 271 321 565 946 3210 4435
Jiangxi 273 316 462 619 1348 2246
Shandong 315 372 688 979 2651 4779
Henan 231 293 449 594 1512 2694
Hubei 330 396 621 834 2003 2725
Hunan 285 338 485 670 1571 2457
Guangdong 367 444 796 1373 3598 5806
Guangxi 223 256 364 582 1205 2087
Hainan 310 325 565 866 1902 2578
Chongqing 255 296 427 562 1436 2619
Sichuan 261 298 443 621 1332 2146
Guizhou 173 202 325 438 739 1266
Yunnan 223 247 376 666 1287 1863
Tibet 372 435 694 688 1252 2164
Shaanxi 292 312 470 671 1273 2358
Gansu 346 358 471 599 1084 1779
Qinghai 426 439 632 859 1425 2388
Ningxia 366 397 569 766 1338 2427
Xinjiang 316 384 642 986 2059 3092
Note: yuan/person.
Source: Same as Table 7.1.
Chapter 7
138
Figure 7.1: Coefficient of Variation of GDP per Capita in Chinese Regions,
1978-2005
0.00
0.20
0.40
0.60
0.80
1.00
1.20
1978
1980
1982
1984
1986
1988
1990
1992
1994
1996
1998
2000
2002
2004
year
cv
Source: Table 7.1.
The coefficient of variation is the highest at the beginning of the period, in 1978. It
declines to its lowest level in 1990. It increases somewhat during 1991-2004,
dropping again in 2005. The coefficient of 2005 is 69% of that in 1978. Thus, there
seems to be strong convergence from 1978 to 1990, some divergence between 1990
and 2000, stability from 2000 to 2004 and a drop in regional inequality in 2005. In the
long run regional inequality seems to be decreasing. This will be further examined in
the coming paragraphs.
The kernel density distribution – reproduced in figures 7.2a, 7.2b and 7.2c – provides
us with a more complete picture of the changes. In the early years we see a bimodal
distribution, with two small groups of leading regions and a concentration of lagging
regions. In the later years, the distribution becomes unimodal, which is consistent with
the convergence trend. The shift from the bimodal to the unimodal distribution takes
place gradually. The two leading groups are still visible in 1992. By 2005 the
distribution has become unimodal.
As the average regional GDP increases, the distribution becomes flatter. This should,
however, not be interpreted as a sign of divergence. It is primarily caused by the
Regional Performance and Productivity Efficiency
139
increase in the mean. In Table 7.1, one can see that the standard deviation increases
less than the mean, so that the coefficient of variation is declining.
Figure 7.2a: Kernel Density of GDP per Capita in Chinese Regions, 1978-2005
1980
1985
1990
1995
2000
2005
0 2000 4000 6000 8000 10000 12000 14000 16000
0
0.5
1
1.5
2
2.5
3
x 10-3
Figure 7.2b: Kernel Density of GDP per Capita in Chinese Regions, 1978-1992
1978
1980
1982
1984
1986
1988
1990
1992
0 500 1000 1500 2000 2500 3000 3500 4000
0
0.5
1
1.5
2
2.5
3
x 10-3
Chapter 7
140
Figure 7.2c: Kernel Density of GDP per Capita in Chinese Regions, 1992-2005
1992
1994
1996
1998
2000
2002
2004
02000
40006000
800010000
1200014000
16000
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
x 10-3
Note: GDP per capita at 1978 constant prices, yuan/ person.
Source: CSY, various issues.
7.3.2 Regional Differentials in Labour Productivity
Table 7.3 and Figure 7.3 provide information about regional disparities in industrial
productivity.1 The trends in value added per worker are rather similar to those for total
GDP per capita. The coefficient of variation declines substantially between 1978 and
1989. Between 1989 and 1994, disparities increase quite rapidly. Between 1994 and
2000, the trend stabilises. After 2000, the coefficient of variation declines. Overall,
regional inequalities have declined in this period. The coefficient of variation in 2005
is 30 per cent lower than in 1978. In general, productivity disparities are much less
pronounced than income disparities (Table 7.3 versus Table 7.1). Table 7.4 shows
regional labour productivity in industry in selected years.
1 We would have preferred to perform this analysis for manufacturing, but Chinese regional statistics
tend not to distinguish between manufacturing and industry.
Regional Performance and Productivity Efficiency
141
Table 7.3: Standard Deviation, Mean and Coefficient of Variation of Industrial
Labour Productivity in Chinese Regions 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 standard deviation
0.13 0.12 0.12 0.11 0.11 0.10 0.11 0.10 0.10 0.10 0.10 0.09 0.10
mean 3205 3315 3320 3164 3224 3445 3597 3495 3446 3705 3973 4020 3944
coefficient of variation
0.40 0.38 0.36 0.35 0.33 0.30 0.29 0.29 0.28 0.26 0.24 0.23 0.25
1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 standard deviation
0.11 0.12 0.18 0.21 0.18 0.21 0.24 0.23 0.27 0.37 0.41 0.44
mean 4102 4559 6124 5759 5412 6025 6987 6948 8318 10462 12270 14630
coefficient of variation
0.26 0.27 0.30 0.36 0.34 0.35 0.35 0.33 0.33 0.35 0.33 0.30
Note: The mean of labour productivity is calculated at yuan/person, at 1978 constant prices.
Sources: CSIEY, various issues; SCIT (2000); CLSY, various issues; 31 regional yearbooks, various issues.
Table 7.4: Industrial Labour Productivity in Chinese Regions
(at 1978 Constant Prices) 1978 1980 1985 1990 2000 2002
National
average 3205 3320 3495 3944 10462 14630
Beijing 5410 5050 5181 5612 15172 19753
Tianjin 4015 4431 4648 4473 12452 17628
Hebei 3428 3461 3584 3532 9973.3 13646
Shanxi 2347 2662 3068 2893 5547.3 8815
Inner Mongolia 1772 1918 2204 2642 7780.2 12046
Liaoning 4199 3476 3601 3752 9607.9 13905
Jilin 3653 3849 3010 3097 8739.7 14803
Heilongjiang 3583 3379 3057 4155 14763 18899
Shanghai 7696 7313 6651 6083 19554 25792
Jiangsu 2064 2482 2932 3589 11937 16865
Zhejiang 2476 2976 3064 3560 11465 14724
Anhui 2048 2117 2342 3151 7411.2 11690
Fujian 2834 3107 3194 4295 12172 16635
Jiangxi 2164 2625 2346 2477 5885.7 9579
Shandong 4546 4875 4405 4132 11592 15906
Henan 2797 3446 3012 3061 7681.5 10880
Hubei 3113 3880 3714 3391 10432 14377
Hunan 3037 2857 3179 3488 7523.5 11732
Guangdong 3370 3515 2694 5514 14198 17110
Guangxi 2182 2439 3114 3951 8430.5 11378
Hainan 1913 1765 2790 4408 12519 15879
Chongqing 3788 4360 5827 3893 7521.5 12358
Sichuan 2117 2283 3510 4425 7541.6 10518
Guizhou 3434 3511 3937 6741 16379 24291
Yunnan 1823 1756 3174 3108 7524.2 9609
Tibet 2978 2833 2931 3398 7814.1 11848
Shaanxi 5072 5184 4208 3745 6370.2 10523
Gansu 2809 2661 2810 4125 9779.2 15033
Qinghai 2949 2696 3227 3442 7877 9824
Ningxia 2515 2704 3428 4175 18208 22840
Xinjiang 3450 3579 3576 3908 10850 15111
Note: yuan/person.
Source: Same as Table 7.3.
Chapter 7
142
Figure 7.3: Coefficient Variation of Industrial Labor Productivity in Chinese
Regions, 1978-2002
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.35
0.40
0.45
1978
1980
1982
1984
1986
1988
1990
1992
1994
1996
1998
2000
2002
year
cv
Source: from Table 7.3.
To further explore the productivity disparities from a geographical perspective, we
have classified 31 regions into three groups2: West regions (9 provinces/cities):
Sichuan (including Chongqing), Guizhou, Yunnan, Tibet, Shaanxi, Gansu, Qinghai,
Ningxia, Xinjiang. Middle regions (9 provinces): Shanxi, Inner Mongolia,
Heilongjiang, Jilin, Anhui, Jiangxi, Henan, Hubei and Hunan. Eastern coastal regions
(12 provinces/cities): Beijing, Tianjin, Hebei, Liaoning, Shanghai, Jiangsu, Zhejiang,
Fujian, Shandong, Guangdong, Guangxi and Hainan (see also Appendix C with a map
of China). The results are reproduced in Table 7.5.
Table 7.5: Labour Productivity in Industry, by Geographical Location
1978 1985 1992 1997 2002
AV LP 3054 3494 4816 7417 15632
LPwest/LPtot 0.89 1.03 0.98 0.98 0.93 West(9)
LPwest/LPEC 0.83 0.96 0.87 0.86 0.85
AV LP 2724 3406 4236 6558 13744
LPmiddle/LPtot 0.79 0.81 0.81 0.80 0.83 Middle(9)
LPmiddle/LPEC 0.74 0.75 0.71 0.70 0.76
AV LP 3678 3584 4545 6863 14272 Coastal(12)
LPEC/LPtot 1.07 1.07 1.13 1.14 1.10
Source: see Table 7.3.
Notes: Labour productivity is in yuan/person at constant 1978 prices. AVLP is the average labour
productivity the geographical regional groups. LPi/LPtot is the ratio of AVLP to national total
productivity. LPi/LP(EC) is the ratio of AVLP to the productivity of the coastal regional group.
During the whole period 1978-2002, the coastal regions had much higher productivity
levels than other regions. They maintained their productivity leadership over time.
2 This classification is slightly different from the one in Chapter 8.
Regional Performance and Productivity Efficiency
143
The middle regions have the lowest relative productivity. In 2002, their productivity
was only 0.76 of the coastal level. Table 7.5 does indicate that over the whole period,
the coastal regions have marginally increased their productivity leadership since 1978.
But the trends are not unambiguous. Since 1997, the Middle and in particular the
West regions have been catching up. In sum, the results of Table 7.5 confirm the
conclusion derived from Table 7.3 that there are no systematic long-run trends
towards increasing regional productivity disparities.
The rate of convergence can be calculated from the following regression equation
Tii
TiiT y
T
e
T
yy,00
0 )log(1)/log(
ωαβ
+−
−=−
(1)
Table 7.6: Beta Convergence Number of
Years (T)
coefficient of
log( )0iy Std. error
β (rate of
convergence) Coefficient α
1978-90 12 -0.053* 0.009 0.084 0.194
1990-2002 12 -0.007 0.014 0.007 0.073
1978-2002 24 -0.027* 0.005 0.044 0.129 * Significant at 1% level
Source: see Table 7.3.
The beta coefficients in Table 7.6 confirm the rapid rate of convergence in the early
period and the slower convergence over the whole period. The regression coefficient
in the second period, 1990-2002 does not deviate significantly from zero.
The kernel distribution in Figure 7.4 indicates that in contrast to the regional
distribution, the distribution of industrial productivity is unimodal over the whole
period. There are no clearcut gaps between a club of leading regions and the follower
regions. In Figure 7.4, we have chosen to reproduce the density of labour productivity
divided by mean productivity. This avoids the flattening of the distribution as a result
of increases in the mean, which incorrectly suggests that disparities are widening.
Chapter 7
144
Figure 7.4: Kernel Distribution of Industrial Labour Productivity in Chinese
Regions, 1978-2005 (10000 yuan/person)
1980
1985
1990
1995
2000
2005
0
0.5
1
1.5
2
2.5
0
0.5
1
1.5
2
Note: labour productivity is calculated at 1978 constant prices, and divided by the mean.
Source: various national and regional yearbooks.
Shanghai has the highest labour productivity in almost all years. The following figures
show the distribution dynamics of regional productivity ratios relative to the leading
region Shanghai during the period 1978-1990 and the period 1990-2002.
d i st r i bu t i on f o r pe r c e n t a ge o f S ha ngha i
0.00
0.20
0.40
0.60
0.80
1.00
1.20
0.00 0.20 0.40 0.60 0.80 1.00 1.20
1978
d i st r i bu t i on f o r p e r c e n t a ge o f S ha ngha i
0.00
0.20
0.40
0.60
0.80
1.00
1.20
0.00 0.20 0.40 0.60 0.80 1.00 1.20
19 9 0
Figure 7.5a: Regional Labour
Productivity as Percentage of Shanghai,
1978-1990
Figure 7.5b: Regional Labour
Productivity as Percentage of Shanghai,
1990-2002
Regional Performance and Productivity Efficiency
145
Figure 7.5c: Regional Labour Productivity as Percentage of Shanghai,
1978-2002
distribution for percentage of Shanghai
0.00
0.20
0.40
0.60
0.80
1.00
0.00 0.20 0.40 0.60 0.80 1.00
19 7 8
Between 1978 and 1990 almost all regions have improved their performance relative
to the leader, confirming the rapid beta convergence in Table 7.5. In the later period
1990-2002, performance relative to the leader in 1990 worsens somewhat consistent
with the divergence trends documented above. However, a great many regions show
very little change in their relative performance. Panel c confirms the convergence
trends over the whole period, with most regions improving their comparative
performance.
From 1978 to 1990, all regions, except only Qinghai, locate above the diagonal.
Yunnan has the biggest jump from 45% to 110% of Shanghai from 1978 to 1990.
Guangdong and Beijing also have significant increases, from 44% to 99%, and 70% to
92% respectively.
7.3.3 Comparative Efficiency Trends using DEA
As we have estimated regional capital stocks in addition to the regional data on value
added and employment, we can go beyond comparative trends in labour productivity
and examine regional differentials in total factor productivity. We will do this using
frontier analysis for the 31 regions, which decomposes regional growth into growth at
the frontier and changes in efficiency.
Chapter 7
146
TFP change can be decomposed into various sources, e.g. changes of technical
efficiency, technological progress, scale efficiency, allocative efficiency3, etc. This
decomposition can be done using either a parametric approach (stochastic frontier
analysis), or a non-parametric approach (data envelopment analysis). SFA includes a
"noise" term in its model. Thus, the efficiency of a firm (industry or region) is
compared with a best practice level which is estimated econometrically. DEA
compares the performance of a firm (industry or region) with the observed best
practice. Given that the outcomes of DEA are heavily influenced by outliers most
researchers prefer stochastic frontier analysis. We have experimented with SFA
analysis, but the results of the stochastic models examined so far were so unstable,
that we will only present the DEA results here.
The DEA approach makes use of a Malmquist index approach. The Malquist TFP
index was introduced by Caves Christensen and Diewert (1982) and has been often
used for the measurement of TFP indexes and efficiency scores. The Malmquist
productivity index has two main advantages over the Törnqvist and the Fisher indexes.
One is that price data are not necessary for aggregation and the other is that it can
decompose TFP into various sources4.
The output-oriented distance function of production set at time t, (based on the
technology at time t) is ),( tt
t
o yxD . Likewise, it will be ),( 11
1
+++
tt
t
o yxD for production
set at time t+1, (based on the technology of time t+1). Taking the benchmark
technology at time t, the output-oriented Malmquist productivity index is
),(
),(),,,( 1111
1,
tt
t
o
tt
t
otttt
tt
yxD
yxDyyxxMTFP ++
+++ = (2)
Considering that technology at time t and time t+1 can be used in the production of
time t+1 and time t, the Malmquist TFP index can be expressed by the following
geometric mean
5.0
1
11
1
11
11
1,
),(
),(
),(
),(),,,(
•=
+++
+++
+++
tt
t
o
tt
t
o
tt
t
o
tt
t
o
tttt
tt
yxD
yxD
yxD
yxDyyxxMTFP (3)
3 Allocative efficiency can only be estimated if input or output prices are also available. This is not the
case in this study. 4 See Färe et al. (1994) for more comparisons between MPI and other index approaches.
Regional Performance and Productivity Efficiency
147
Färe et al. (1994) and Farrel (1957) decompose TFP into technological progress and
change of technical efficiency, which has been widely adopted by many researchers5.
Using the Malmquist TFP index with panel data, the output-oriented linear program
DEA model can be expressed as
[ ] φλφ ,
1
0 max),( =−
tt
t yxD
subject to
0
0
0
≥
≥+−
≥−
λ
λφ
λ
tit
tit
Yy
Xx
(4)
where, X is the input (k*n) vector, Y is the output (m*n) vector, and
T
n ),...,,( 21 λλλλ = . Equation (4) shows a distance function for production point
),( itit yx at technology t. To get the geometric Malmquist TFP index from time t+1 to
t, both ),( itit yx and ),( 11 ++ itit yx will be calculated at technologies in both time periods
t and t+1. Therefore, there will be 3 parallel linear programmes (see also in Coelli,
1996, p.27; Färe et al., 1994, p.75). Using the DEA program introduced by Coelli
(1996), we have estimated technical efficiency in industry in 31 Chinese regions. This
is reproduced in Figure 7.6 and Table 7.7.
Figure 7.6: Technical Efficiency of Industry in Chinese Regions, 1978-2002
(by DEA Model)
0.000
0.200
0.400
0.600
0.800
1.000
1.200
1978
1980
1982
1984
1986
1988
1990
1992
1994
1996
1998
2000
2002
year
5 There has been some criticism on further decomposition of the technical efficiency change into pure
technical efficiency change and scale efficiency change. Namely, the first step of such decomposition
(on technology progress and change of technology efficiency) is taken under the assumption of
constant return to scale (CRS). However, on the second step, the technical efficiency is decomposed
under the condition of various returns to scale (VRS). In this chapter, to avoid this inconsistency, we
assume only constant return to scale, and do not take consideration of the scale efficiency change.
Tab
le 7
.7: R
egio
nal E
ffic
ienc
y Sc
ores
, 197
8-20
02
Chapter 7
148
Regional Performance and Productivity Efficiency
149
Shanghai and Chongqing are always at the efficiency frontier throughout 1978-2002.6
Zhejiang, Jiangsu, Guangdong also are near the top ranks at a later stage. Figure 7.7
presents the coefficient of variation of the regional efficiency scores. Again the long
convergence trend is clearly visible. Efficiency scores converge until 1990, diverge
quite substantially until 2001, and then converge again.
Figure 7.7: Coefficient of Variation of Technical Efficiency in Industry in 31
Chinese Regions, 1978- 2002
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
1978
1980
1982
1984
1986
1988
1990
1992
1994
1996
1998
2000
2002
year
Figure 7.8: Growth Rate of TFP, Technical Efficiency and Technological
Progress
-0.2
-0.15
-0.1
-0.05
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001
technical efficiency Technological progress TFP
6 Chongqing figures should be treated with caution. Many assumptions had to be made to estimate the
capital stock. Before 1997, the Chongqing data were included in Sichuan. The level of capital and
labour input and the level of labour productivity are very low.
Chapter 7
150
Figure 7.8 plots the average growth rates of TFP, technical efficiency and
technological progress in industry in 31 Chinese regions. In the earlier years, changes
in technical efficiency contribute much more to the TFP growth than technological
progress. This is reversed after 1991, when technological progress is far more
important than changes in technical efficiency (with the exception of one year: 1996).
7.4 Summary and Conclusions
This chapter attempts to analyze the regional productivity and the trend of regional
convergence and divergence. We have analysed a wide range of indicators including
GDP per capita, labour productivity and comparative efficiency scores, using a
variety of techniques.
According to public perceptions of the Chinese growth experience, China is
characterised by increasing regional inequality. This was also our initial working
hypothesis, when we embarked on this research project. The empirical results point in
the opposite direction. There is no long-run divergence trend between Chinese regions
since 1978. On the contrary, there has been substantial regional convergence from
1978 to around 1990. This has been followed by a period of modest divergence up till
around 2001. After 2001, convergence trends resumed. Thus we conclude that during
the period of accelerated economic growth from the 1990s onwards, we observe the
pattern of the inverted u-curve: an increase in regional inequality followed by a
decrease.
However, whatever indicator was used, the degree of regional inequality in the latest
years was substantially lower than at the beginning of the reform period. Coastal
regions did have much higher productivity than inland regions, but there was no clear
tendency for coastal regions to forge ahead relative to regions in the west and in the
middle.
An analysis of the relative importance of technological change and efficiency
provides an interesting interpretation of the Chinese reform experience. In the early
stages of the Chinese reform process efficiency changes predominate. Once efficiency
Regional Performance and Productivity Efficiency
151
differentials between regions have been reduced in the process of efficiency
convergence, technological change at the frontier becomes more important as a driver
of growth in Chinese industry.
CHAPTER 8
Contribution of Technological Spillovers to Industrial
Growth in Chinese Regions
8.1 Introduction
Technological spillovers can play a crucial role in catching-up and convergence
theories at both regional and national levels. The lagging regions or countries benefit
from imitating, learning from or even using for free the new technologies invented by
the leaders. Having saved money from risky R&D expenditures, lagging regions or
countries can sometimes take large leaps in economic development. In development
and growth studies this is referred to as the advantages of backwardness.
(Gerschenkron, 1962; Abramovitz, 1989, see also the discussion in Szirmai, 2008).
Chapter 7 has shown that from from 1978 to 2002 there has been a net long-run
convergence trend in Chinese regions with regard to GDP per capita, labour
productivity and comparative efficiency scores in industry. In the convergence
process of industrial growth in Chinese regions, it is interesting to examine whether
and/or to what degree knowledge spillovers play a role in regional performance and
catching-up. Concerning the sources of knowledge spillovers in Chinese regions, we
can distinguish between the regional level and the international level. The former
refers to R&D inputs in other regions, the latter concerns international R&D
investment which is often embodied in foreign direct investment (FDI).
In this chapter we will explore the contribution of R&D knowledge spillovers to
industrial growth in Chinese regions. Our analysis covers the impact of spillovers
from R&D in other regions, FDI, as well as FDI from other regions.
Chapter 8
154
The chapter is organized as follows. Section 2 reviews the existing literature on
knowledge spillovers. Section 3 provides a survey on regional spillovers and FDI
spillovers in China. In section 4, different types of spillover models are discussed.
Section 5 presents our methodology, data and empirical results. Conclusions are
drawn in section 6.
8.2 Spillover Findings and Models
8.2.1 Spillover Findings
Spillover studies have been carried out at many levels. There have been studies of
domestic spillovers (between regions, firms or industries) and studies of international
spillover (Coe and Helpman, 1995). With regard to domestic spillovers, spillovers
occur both within and between industrial sectors. Intra-industry spillovers are
beneficial to the firms in the same industry. Inter-industry spillovers offer free
knowledge to firms from other industries, e.g. suppliers and customers can learn from
each others’ advanced technologies, practices or managerial skills.
Many researchers emphasize the significance of spillovers. Coe and Helpman (1995)
show that, in G7 countries, a R&D investment of one per cent has an average rate of
return of 1.55 per cent in the 22 countries in their study1. Bernstein and Nadiri (1989)
demonstrate there are significant intra-industry spillover effects in all industries in
their study. The empirical work of Girma and Wakelin (2007) on plant-level data in
the electronics sector in the UK indicates important intra-industry as well as inter-
industry spillovers. Funke and Niebuhr (2005) find that regional growth in 71 regions
in western Germany is positively influenced by the R&D activities in nearby regions.
In a less optimistic way, Aitken and Harrison (1999) argue that FDI spillovers are
ownership limited. In an analysis of panel data of Venezuelan plants, they find that
1 They also show that the average private rate of return from R&D investment in G7 countries is 1.23
per cent. Therefore, their results indicates a larger social return (international spillovers) than private
return (R&D return in own countries).
Contribution of Technological Spillovers
155
foreign investment is beneficial only to the productivity of joint ventures or affiliates;
domestic plants do not benefit from it. Foreign investment even has a negative
influence on the productivity of domestically-owned firms. Aitken and Harrison find
that in the short run, the entry of FDI reduces the productivity of domestic firms. Hu
and Jefferson (2002) document similar results in a study of China's electronics and
textile industries2.
Girma and Wakelin (2007) conclude that FDI spillovers are geographically limited.
Through their analysis of plant-level data, they show there is no relationship between
domestic productivity and FDI in other regions, although there are significant intra-
industry and inter-industry spillovers from FDI invested directly in a particular region.
Going even further, Girma and Wakelin (2002) suggest that domestic firms benefit
positively from foreign firms in the same sector and in the same region, but that FDI
from outside the region even has a negative impact on productivity.
8.2.2 Spillover Models
Various models have been used in the literature regarding technological externalities
and spillovers. Below four different groups of spillover models will be discussed
briefly.
Group one: General technological spillovers
The well-known approach to measure spillover effects is the augmented Cobb-
Douglas production function, proposed by Griliches (1979), which is also integrated
with the endogenous growth model of Romer (1990):
iteRRLKAY stititititit
εµγβα= (1)
where itY stands for the value added in region (industry or firm) i at time t, K, L and R
are the physical capital input, labour input and technological knowledge input
respectively. iR is the direct science and technology input (normally embodied by
2 Aitken and Harrison (1999) and Hu and Jefferson (2002) are similar to each other in terms of
methodology and findings, except that the former focuses only on the short-run effects in which FDI
reduces the market shares of domestic firms. The latter study also includes the long-run effects, namely
that domestic firms which survive the competition resulting from FDI, will be able to benefit from
technology spillovers.
Chapter 8
156
R&D expenditures or stocks) in unit i, while sR indicates the indirect technological
inputs (aggregated R&D expenditures) flowing from all other units to unit i. The
technology contribution to the output of unit i derives not only from its own R&D
expenditures, but also from the expenditures of other firms. The coefficient γ
represents the direct contribution of R&D. γ and µ represent the elasticity of direct
R&D and indirect R&D on output. itA captures the level of total factor productivity
apart from the impact of R&D (known also as the level of disembodied technology).
Using logarithms, equation (1) can be written as
itstitititit RRrLKaY εµβα +++++= )ln()ln()ln()ln()ln( (2)
This model has been widely adopted in measuring R&D spillover effects (see Raut,
1995, p. 5; Los and Verspagen, 2000, p. 129; Coe and Helpman, 1995).
Group 2: FDI spillovers
The contribution of FDI to economic growth comes not only from its direct
contribution to capital formation or employment creation, but also from its knowledge
spillovers. In developing countries, the technological spillover effect3 from FDI is of
great potential importance for total factor productivity (TFP) growth and technology
contributions, since foreign investment in developing countries, in most cases,
embodies advanced technologies.
Spillovers from FDI are among the technological spillovers which were discussed
more generally in group 1. Most existing literature on FDI spillovers is also based on
equation (2). Given that the technology level of one region can be a result of both the
direct technological influence from the foreign investment that has taken place in this
region, and spillovers from FDI in other regions, two spillover effects can be defined:
itFDI from direct FDI in region i, and stFDI from aggregated spillovers from other
regions. If both R&D and FDI are regarded as technology source, the knowledge input
factor will be
3 Estimates on FDI spillover effects are different from those of direct effects of FDI on productivity and
growth. In the latter case, foreign capital has to be separated from the domestic capital stock. In
estimating spillover effects, however, the main issue is on the external effect of FDI, which is an
indirect impact, hence it is no longer necessary to decompose the capital stock.
Contribution of Technological Spillovers
157
.),,,,( etcFDIFDIRRfY stitstitit = (3)
This has been used by numerous authors in measuring FDI spillovers (Liu et al.
LPVW, 2001; Wei and Liu, 2006; Cheung and Lin, 2004).
Group 3: Accounting for the endogeneity of FDI
As pointed out by Hale and Long (2007), most of the empirical studies on FDI
spillovers suffer from an upward bias due to the endogeneity of FDI. They argue that
most of the FDI contribution in China is normally biased upward. If FDI firms are not
distinguished from domestic firms, the positive (total) coefficient will be more likely
explained as the contribution from FDI. Hale and Long conclude that “empirical
evidence of FDI spillovers on Chinese domestic firms is mixed, largely because data
limitation has hampered the effort to control for the endogenous location of FDI.” The
endogeneity problem is caused by a misinterpretation of the impact of FDI, and the
estimates are likely to be biased upward (Hale and Long, 2007, p. 10).
Decisions on investment by foreign firms do not happen randomly. In order to
maximize profits, foreign companies tend to select regions (or industries) with higher
potential growth capabilities and higher technology levels than the FDI recipient. In
other words, FDI more likely takes place in a region (or industry) with higher
productivity and/or capable for more rapid growth. Therefore the FDI spillover effect
will be exaggerated if parts of regional or industrial growth are misinterpreted as a
FDI effect (Hale and Long, 2007; Hu and Jefferson, 2002). Ignoring the endogeneity
of FDI is one of the explanations of an upward-bias in the FDI spillover effect.
To account for the endogeneity, Aitken and Harrison (1999), and Hu and Jefferson
(2002) include an interaction term, firm-level FDI multiplied by industry-level FDI,
which indicates that an industry with high a level of productivity normally will attract
more FDI. After all, FDI gravitates towards more productive industries. The
correlation between the presence of foreign firms and the productivity of domestic
firms will lead to an overestimation of the positive impact of foreign investment (see
Aitken and Harrison, 1999, p. 606).
Chapter 8
158
Meanwhile, Aitken and Harrison also interpret the coefficients of firms and industries
with caution, taking into account the endogeneity factor. The positive and significant
coefficient of firm level FDI "might be correlated with, but not caused by, foreign
presence" (Aitken and Harrison, 1999, p.611). Also foreign firms could simply be
more efficient than their domestic counterparts. A high coefficient of FDI does not
have to mean a positive effect of knowledge spillovers.
In line with this approach, Buckley et al. (2002) also take into account the
interrelationship between labour productivity and foreign investment. Taking a
broader perspective than the model of Hu and Jefferson (2002), the foreign presence
in Buckley et al. (2002) is assumed to be related to labour productivity, export
intensity, market size, R&D intensity and labour quality in the studied industry.
Following Aitken and Harrison (1999) and Hu and Jefferson (2002), we will also
include an interaction term in our model to account for endogeneity of FDI.
Group 4: Distinguishing between effects of foreign and domestic investment
Balasubramanyam et al.(1996) separate the effects of FDI from domestic investment
effects in the model
xfkly φψγβα ++++= (4)
where y, l, k, f, and x are the growth rate of GDP, labour, domestic capital, foreign
capital and exports respectively, and β ,γ ,ψ andφ are their output elasticities. Being
aware of, and having stressed the impact of spillover and externalities from FDI, they
show that FDI has a more important contribution to growth processes than domestic
investment, through comparing the coefficients of γ andψ . However, given that the
two effects of FDI (knowledge spillovers and capital investment) are included
together inψ , their work does not measure the spillover effect separately.
Following the model of Feder (1982), which measures the sectoral externalities
separately, Zhang and Felmingham (2002) divide the whole economy into export
sectors and non-export sectors, and the capital input into foreign capital (from FDI)
and local capital. Their study shows that FDI flows stimulate growth in eastern,
central as well as western regions in China.
Contribution of Technological Spillovers
159
Most literature on inter-industry or international spillovers uses weights to aggregate
external R&D stocks, which are based on the correlation or similarity between the
destination and the origin of spillovers (Griliches, 1979; Aiello and Cardamone, 2005).
In international spillover studies exports and imports between two countries can be
taken as such an indicator (Coe and Helpman, 1995, p.860; Jacob and Szirmai, 2007).
In regional studies, however, it is difficult to quantify these weights. Some researchers
have used the unweighted aggregate R&D capital expenditures (Raut, 1995; Wei and
Liu, 2006).
8.3 Technological Spillovers in China
8.3.1 Regional Spillovers
Given the existing dramatic disparities in productivity between regions in China,
interregional spillovers are well recognized as an important source of catch-up for
poorer regions. In terms of regional spillover studies, there is one group focusing on
knowledge spillovers between regions (including this chapter); other studies are
examining output spillovers4 (Groenewold, et al. 2006; Zhang and Felmingham, 2002).
The degree to which knowledge can spill over from one unit (firm, industry or region)
to its peers depends on the closeness relationship between the knowledge source and
its recipients. As mentioned in the previous paragraph, in some studies export/import
ratios or input-output relationships between industries have been adopted as weights
in estimating spillover amounts. We argue that these weights are more appropriate
when estimating rent spillovers, and should not be used for knowledge spillover cases.
This chapter deals with knowledge spillovers, hence no such product-related weights
will be used in our analysis. Besides a product-related weight, a closeness indicator in
the technological sense can also be helpful in exploring the impact of spillovers
(Jabob and Szirmai, 2007). However, given the difficulty in quantifying technological
4 This branch focuses on the inter-regional growth spillovers, not technological spillovers.
Chapter 8
160
relationship and congruence between regions, we instead use geographic distance as
the weight of closeness.
8.3.2 Spillovers from FDI
FDI spillovers, as a type of international knowledge spillovers, have been regarded as
a very important source of knowledge in developing countries.
FDI has been believed to promote China's economic growth, which may explain why
the Chinese government has enacted much favourable legislation for foreign
companies. The contribution of FDI to the Chinese economy not only results from
capital investments, but also from technological spillovers. There are two main
channels by which local firms can benefit from FDI related technological spillovers.
One is through imitating products invented by foreign companies, given that local
firms can easily learn from the new design of a market product. The other channel is
through mobility of employment. Employees trained or hired by foreign companies
can bring knowledge to local firms when they switch jobs5.
Admittedly, the entry of foreign affiliates through FDI may also have negative effects
on domestic companies, considering that FDI reduces the market shares and takes
away talented employees from domestic companies.
The Chinese government provides considerable privileges to foreign firms in order to
attract foreign investment in China. This seems to indicate that the positive effects of
FDI to economic growth are assumed to outweigh its negative effects. Since the
presence of FDI in China, it has always shown a very uneven distribution in
geography, ownership and industries6. In recent years there have been many studies
on FDI in China undertaken from different points of view.
5 In addition to these two main types of spillovers, there are competition and demonstration effects.
Upon the market entry of foreign competitors equipped with more advanced technologies, local
companies can be pushed to innovate and adopt more new technologies. However, in a strict sense,
competition effects are not really spillovers, though they can stimulate firms to learn from their foreign
competitors. For a further discussion on spillover channels, see Madariaga and Poncet (2007, p.841)
and Cheung and Lin (2004, p. 26). 6 Hu and Jefferson (2002) give a very good example of the FDI impact difference by industry, though
generally the number of studies on FDI impact by industry is lower than those on FDI impact by
geography or ownership.
Contribution of Technological Spillovers
161
Geographical dimension:
One strand of research focuses on interregional spillovers. Cheung and Lin (2004),
using provincial data during 1995-2000, find positive effects of FDI on patent
applications. Madariage and Poncet (2007) argue that FDI flows to surrounding
locations should also be concluded in addition to the direct FDI in one location. They
believe that FDI is beneficial not only to the region that receives FDI, but also to its
neighbouring regions. Although Girma and Wakelin (2007) have showed that in the
UK electronics industry there are no interregional FDI spillovers, there is some
literature showing that in China FDI contributes both directly to one region and
indirectly to other regions.
Cheung and Lin (2004) find that the FDI spillover effect is stronger in western regions
than in eastern and central areas. They explain this as the result of higher spatial
concentration of FDI in the west which produces a stronger spillover effect than
"evenly distributed" FDI in provinces located in the east and center. The authors
furthermore state that "during the second half of the 1990s over 60% of the total FDI
inflow to the west region went to the Sichuan (Chongqing City included) and Shaanxi,
whereas the foreign investments to the east and central region were more evenly
distributed among the provinces within the regions. Since location proximity is of
crucial importance to technology/knowledge spillovers [...], higher degree of spatial
concentration of FDI tends to yield a stronger spillover effect" (Cheung and Lin, 2004,
p.42).
Ownership dimension:
Besides the differences in geographic effects of FDI, it also appears that spillovers
vary by ownership type. In China, it is well known that FDI is more likely to benefit
private instead of state-owned enterprises. Hale and Long (2006) find that FDI has a
strong positive effect on private firms in China, while SOEs receive no or even
negative impacts from FDI. Hu and Jefferson (2002) test the FDI impact on SOEs in
two industries (electronics and texiles). They find that FDI significantly depresses the
productivity of SOEs in the textile industry. They also conclude that, in the short run,
FDI enhances the productivity of foreign firms (who receive FDI) and reduces the
productivity of domestic firms (who do not receive FDI). However, Buckley et al.
Chapter 8
162
(2007) find that both state-owned enterprises and other locally owned enterprises can
benefit from inward FDI7.
Branstetter and Feenstra (2002) argue that Chinese SOEs are more likely to oppose
FDI because the entry of foreign firms brings more competition and threat to SOEs.
However, a new competition law (the 2007 Anti-Monopoly Law) has recently become
effective. Therefore, the level of FDI probably will not be determined by SOEs, but
will continue to increase in China.
Industrial dimension:
Using 29 manufacturing sectors from 1993-1998 in Shenzhen, Liu (2002) concludes
that there is a significant and positive relationship between the average level of FDI in
manufacturing and the productivity of its component sectors. He states that a 1%
increase in the average FDI in manufacturing raises the productivity growth among
manufacturing sectors by 0.5%. However, he does not find a significant direct
relationship between the amount of FDI in an industry and its productivity.
On the contrary, using a sample of enterprises in the electronics and textile industries
in China, Hu and Jefferson (2002) find that FDI has a strong impact on the directly
receiving firms, but that the spillover effects from FDI in an industry are negative for
its domestic firms. Using a dataset from the Chinese electronics industry for 1996 and
1997, Liu et al (2001) conclude that FDI has a positive impact on labour productivity
in the Chinese electronics industry.
8.3.3 Data and Variables Used in the Literature
R&D expenditure and R&D personnel are the commonly used indicators to represent
technology or knowledge levels in regions. In the case of China, national total R&D
expenditure has been available since 1990, but R&D expenditure by region has only
been available since 1998. Instead, many researchers have used science and
technology expenditures, for which regional data are available since 1990.
7 They suggest that both types of firms (SOEs and other locally owned firms) benefit more in
technology-intensive industries than in labour-intensive industries.
Contribution of Technological Spillovers
163
There is great variety in the choice of the FDI variable in the regression studies. Some
papers use FDI value (Cheung and Lin, 2004), some use the ratio of FDI over GDP
(Madriaga and Poncet, 2007), or the share of foreign equity (Hu and Jefferson, 2002).
A summary of variables used in the literature on spillovers in China is presented in
Table 8.1.
Tab
le 8
.1: S
umm
ary
of V
aria
bles
of
R&
D a
nd F
DI
Spill
over
s in
Chi
na
de
pend
ent
vari
able
F
DI
vari
able
R
&D
va
riab
le
othe
r in
depe
nden
t va
riab
les
liter
atur
e re
sult
s
valu
e ad
ded
perc
enta
ge o
f re
gist
ered
ca
pita
l ow
ned
by f
orei
gn
inve
stor
s in
indu
stry
i; a
nd
shar
e of
for
eign
equ
ity
part
icip
atio
n in
all
indu
stri
es.
Liu
(20
02)
The
re is
insi
gnif
ican
t rel
atio
nshi
p be
twee
n FD
I an
d it
s di
rect
rec
ipie
nt in
dust
ries
. H
owev
er th
ere
is a
sig
nifi
cant
and
pos
itiv
e im
pact
fro
m F
DI
in m
anuf
actu
ring
indu
stri
es
to it
s co
mpo
nent
indu
stri
es.
outp
ut
7 ty
pes
of m
easu
res
on F
DI
ra
tio
of
inta
ngib
le
asse
ts to
fi
xed
asse
ts
expo
rts
W
ei a
nd L
iu
(200
6)
The
re a
re p
ositi
ve in
ter-
indu
stry
spi
llov
ers
from
R&
D a
nd e
xpor
ts. B
oth
posi
tive
intr
a-
and
inte
r-in
dust
ry p
rodu
ctiv
ity
spill
over
s fr
om
FDI,
but
lim
ited
with
in r
egio
ns.
pate
nt
FDI
in y
ear
t-1
S&T
pe
rson
nel,
S&T
ex
pend
itur
e,
fore
ign
fund
ed e
xpor
t to
its
gros
s ou
tput
, GD
P p
er c
apita
C
heun
g an
d L
in
(200
4)
Pos
itive
eff
ects
of
FDI
on th
e nu
mbe
r of
do
mes
tic p
aten
t app
licat
ions
.
labo
ur
prod
ucti
vity
fo
reig
n ca
pita
l to
tota
l cap
ital
ca
pita
l int
ensi
ty, l
abou
r qu
alit
y,
firm
siz
e L
iu, P
arke
r,
Vai
dya
and
Wei
(2
001)
FDI
has
a po
siti
ve im
pact
on
labo
ur
prod
ucti
vity
in th
e C
hine
se e
lect
roni
cs
indu
stry
.
valu
e ad
ded
1)
for
eign
sha
re o
f eq
uity
, 2)
sale
s-w
eigh
ted
FDI
in
indu
stry
Hu
and
Jeff
erso
n (2
002)
labo
ur
prod
ucti
vity
1)
shar
e of
for
eign
cap
ital,
2)sh
are
of e
mpl
oym
ent i
n fo
reig
n fi
rms.
R&
D
inte
nsit
y la
bour
qu
alit
y,
firm
siz
e,
expo
rt
inte
nsit
y,
Buc
kley
, Cle
gg
and
Wan
g (2
002)
FDI
effe
ct o
n FD
I re
ceiv
ing
firm
s. S
pill
over
ef
fect
is s
tron
g, b
ut th
e ef
fect
fro
m th
e ag
greg
ated
FD
I in
the
who
le in
dust
ry is
ne
gati
ve.
sale
s th
e fo
reig
n ca
pita
l sha
re in
in
dust
ry (
wit
h on
e-ye
ar la
g)
m
anag
emen
t cos
t per
wor
ker,
sal
e fa
re p
er w
orke
r, a
nd n
ext f
ixed
as
sets
per
fir
m
Buc
kley
, Wan
g an
d C
legg
(20
07)
The
re a
re g
reat
er p
ositi
ve s
pill
over
s fr
om F
DI
in te
chno
logy
-int
ensi
ve in
dust
ries
than
in
labo
ur-i
nten
sive
indu
stri
es.
Chapter 8
164
G
DP
per
ca
pita
ra
tio o
f F
DI t
o G
DP
savi
ng
rate
, un
ive
rsity
pop
ula
tion
shar
e, p
opul
atio
n gr
owth
ra
te,
exo
gen
ous
rate
of
tech
nica
l ch
an
ge, d
epr
ecia
tion
rate
, la
gge
d de
pend
ent
vari
abl
e a
t T y
ea
r a
go M
ada
ria
ga a
nd
Pon
cet (
2007
)
Chi
nese
citi
es
bene
fit f
rom
not
onl
y fr
om t
heir
ow
n F
DI i
nflo
ws
but a
lso
from
FD
I flo
ws
rece
ive
d b
y th
eir
neig
hbou
ring
citi
es.
Val
ue-a
dde
d (T
FP
)/
labo
ur
prod
uctiv
ity
fore
ign
sha
re
va
riou
s se
t (se
e pa
ge 1
9-21
) H
ale
and
Lon
g (2
007)
The
re
is
no
evi
denc
e
of
FD
I sp
illov
ers
to
d
ome
stic
fir
ms
in C
hina
.
labo
ur
prod
uctiv
ity
1) r
atio
of
fore
ign
firm
s'
em
ploy
me
nt t
o to
tal
em
ploy
me
nt
2)
ratio
of
fo
reig
n fir
ms'
a
sse
ts t
o to
tal a
sse
ts
R&
D
per
em
ploy
ee
la
bou
r q
ualit
y (r
atio
of
em
plo
yee
s w
ith a
de
gree
of
colle
ge le
vel o
r a
bove
to
tota
l em
ploy
me
nt)
Li,
Liu
and
P
ark
er
(200
1)
SO
Es
bene
fit f
rom
the
FD
I co
mpe
titio
n, w
hile
ot
her
loca
l-ow
ned
firm
s (c
olle
ctiv
e o
r pr
iva
te-
owne
d)
gain
ed
from
the
de
mon
stra
tion
and
co
nta
gion
fro
m f
ore
ign
pre
senc
e.
outp
ut o
f re
gion
al
ind
ustr
y
loca
l FD
I/loc
al o
utp
ut
e
xpor
ts,
loca
l spe
cial
iza
tion,
co
mpe
titio
n, in
dust
rial d
ive
rsity
, w
orke
r qu
alit
y, tr
ans
port
atio
n,
tele
com
mu
nica
tion
s, s
tate
se
ctor
sh
are
Ga
o (2
004)
FD
I ha
s a
pos
itive
eff
ect
on r
egi
onal
indu
stria
l gr
owth
.
Sou
rce
: A
utho
r's o
wn
sum
ma
ry.
Contribution of Technological Spillovers
165
Chapter 8
166
8.4 Methodology and Empirical Results
8.4.1 Methodology
In this chapter, the main goal is to capture the effects of regional direct R&D input,
R&D spillovers from other regions, knowledge spillovers from direct FDI
(correlation), and spillovers from FDI in other regions.
The spillover model used in this chapter is derived from Aitken and Harrison (1999)
and Hu and Jefferson (2002). As explained in the previous section, the merit of the
models by Aitken and Harrison (1999) and Hu and Jefferson (2002) is that they tackle
the endogeneity issue by including a FDI clustering term. Elaborating on their
methodology, we also include technological ability (embodied in R&D level) in
explaining spillover effects. Considering that most foreign companies have more
advanced equipment and technology, they preferably choose a region which is
capable of cooperating with them. Thus, we distinguish two endogenous effects on
FDI: one is due to the clustering effect of FDI (FDI flows to regions that have higher
levels of FDI); the other is due to the technological capability of this region. Hence,
our extended model (equation 5) involves not only the interaction term between FDI
and the aggregate FDI from other regions, but also the interaction term between FDI
and the regional technology level. In order to capture technological change over time,
we also include a technology parameter, i.e. 1
0
ta
t eaA⋅⋅= . Taking the logarithm, we
get taaAt ⋅+= 10ln)ln( . Therefore, we have the following model:
ittititt
ittitititit
FDISRDeFDIFDIS
FDIRDSeRDeLKtaaY
εγγ
γγγβα
+++
+++++⋅+=
)ln(*)ln(*)ln()ln(
)ln()ln()ln()ln()ln()ln(
54
32110 (5)
Where
itY --- the industrial value-added of region i at time t;
itK --- the industrial capital input of region i at time t;
itL --- employment numbers of industry in region i at time t;
Contribution of Technological Spillovers
167
itRDe --- levels of R&D (i.e. the ratio of S&T expenditure/GDP at year t-1)8 in region
i at time t;
itFDI --- FDI level (i.e. FDI/GDP ratio)9 in region i at time t; the coefficient of this
term measures international spillovers;
tRDSe --- aggregated R&D levels from other regions; the coefficient of this terms
measures interregional spillovers;
tFDIS --- aggregated FDI levels from other regions; the coefficient of this term
measures international spillovers coming via other regions.
In equation (5) the interregional spillover effects of R&D and FDI are represented by
the coefficients of tRDSe and tFDIS . The interaction term of
)ln(*)ln(*)ln( titit FDISRDeFDI represents the correlation between FDI and its
environment. A significant positive coefficient 5γ indicates that FDI more likely
takes place in regions with high R&D levels and/or a favourable FDI environment (i.e.
high levels of FDI in its neighbouring regions). A favourable FDI environment
indicates that FDI is likely to cluster in areas10 having a good reputation on openness,
a good location (e.g. favourable foreign environment), and which are familiar to
foreign companies. In sum, our model aims to measure the R&D contribution to
regional growth, effects of R&D spillovers from other regions, direct FDI spillovers,
and FDI spillovers from other regions.
To judge the extent to which technologies can spill over from a knowledge resource to
its recipients, different weight systems can be used. Jacob and Szirmai (2007) adopt
structural congruence (structural similarity) of the Indonesian manufacturing and its
trading partners in other countries. We argue that such weights are less appropriate for
pure knowledge spillovers. But, as discussed in earlier sections, geographic proximity
is an important factor in the explanation of interregional spillovers. The closer two
regions are to each other, the greater the possibilities for communication and business
and labour flows. Hence there will be more opportunities for knowledge spillovers. In
8 See section 4.2 for a detailed explanation.
9 See section 4.2 for a detailed explanation.
10 See also Aitken and Harrison, 1999; Hu and Jefferson, 2002.
Chapter 8
168
this chapter, geographic distance is used as weight in aggregating spillovers from
outside regions, with regard to both R&D and FDI, as follows:
∑≠
=ij
jtijt RDwRDS
∑≠
=ij
jtijt FDIwFDIS (4)
Equation (4) measures the external R&D stock i as equal to the weighted sum of all
other regions' R&D stock. Likewise, FDI from outside regions received by region i
equals the weighted sum of all other regions' FDI. ijw is the standardized spatial
weight between region i and j. Suppose the geographical distance between region i
and j is ijd , the spatial weight matrix (see also Ertur, Gallo and Baumont, 2006) can
then be expressed as follows:
0* =ijw if i=j
2* /1 ijij dw = if cutoffd ij ≤ (5)
0* =ijw if cutoffd ij >
in which the distance weight is taken as the inverse of squared distance between
region i and j11. To standardize it, we can use ∑=
j
ijijij www ** / . Thus the sum of each
row in the matrix is equal to 1. The cutoff parameter has been chosen differently by
some researchers (see Baumont, Ertur and Le Gallo, 2000, for taking different
distance parameters as well as working without cutoff). We argue that a cutoff
distance should be considered given that a region is less likely to be influenced by
distant regions. We have experimented with two different cutoff distances. One takes
1520 kilometers as the cutoff distance12. This guarantees that each region in China has
at least one region with which it interacts. The other cutoff is taken at half of this
distance, namely 760 kilometers.
11 Distance of provinces is measured by their capital cities, considering that a capital city is usually the
central business and technology center of each province. 12 Madariaga and Poncet (2007) use the same method by taking a distance at 1624 km, but our data on
regional distance are slightly different from what they collected.
Contribution of Technological Spillovers
169
8.4.2 Empirical Results
Our empirical analysis covers effects of technological spillovers on output in 29
Chinese regions13 during 1991-2002. We run both fixed-effect
14 and random-effect
15
regressions, and then the Hausman test is applied to choose the more efficient model.
Two different cutoff distance settings are included in our analysis. In the following
tables, RDe refers to the direct effect of R&D input, FDI refers to international FDI
spillovers, RDSe and FDIS are the spillover effects of R&D and FDI from other
regions. FDIRDFDIS is the interaction term of FDI with local R&D expenditure and
FDI spillovers from other regions.
Table 8.2: Estimates on R&D and FDI Spillovers in Chinese Regions,
with Cutoff Distance at 1520 km (All Regions), 1991-2002
(1)K,L
(2) K, L, RD, RDS
(3)K,L,RD,RDS,FDI,FDIS
(4) ALL capital 0.317 0.381 0.354 0.356
(0.102)** (0.089)** (0.092)** (0.091)** labour 0.433 0.533 0.532 0.548 (0.060)** (0.054)** (0.061)** (0.060)** RDe 0.14 0.143 0.193 (0.038)** (0.039)** (0.043)** RDSe 0.245 0.241 0.234 (0.053)** (0.053)** (0.053)** FDI -0.016 0.022 (0.012) (0.018) FDIS 0.018 0.053 (0.015) (0.020)** FDIRDFDIS -0.002 (0.001)** t 0.053 0.056 0.058 0.059 (0.010)** (0.009)** (0.010)** (0.010)** Constant 0.789 -2.112 -1.965 -2.408 (0.588) (0.612)** (0.620)** (0.637)** Observations 348 348 348 348 Number of regions 29 29 29 29 R-squared 0.75 0.82 0.82 0.82 Hausman Test FE P>0.0002 FE P>0.0000 FE P>0.0000 FE P>0.0000
Note: 1) RDe is the ratio of S&T expenditure/GDP at year t-1; FDI is the ratio of FDI/GDP at year t-1.
2) FE means fixed effect regression, and RE means random effect regression. The Hausman test
is applied to choose a better regression out of FE and RE.
3) Standard errors in parentheses, * significant at 5% level, and ** significant at 1% level.
13 Chongqing is included in Sichuan, and Tibet is deleted because of its non FDI data.
14 Which assumes that omitted variables differ between cases but are constant over time.
15 Which assumes that omitted variables are constant over time but vary between cases.
Chapter 8
170
Table 8.2 shows the results from step-by-step regressions. In these four cases (with
different numbers of variables), the elasticity of technological change on value added
is always between 5 and 6%. This also indicates that total factor productivity ( tA ) is
growing at a rate of 5-6% per year. The regression which includes all variables is
showed in the last column. The coefficient of R&D is 0.19 (at 1% level of
significance), suggesting a 19% contribution of R&D on regional growth. The
coefficient of interregional R&D spillovers is 0.23 (at 1% level of significance) with a
cutoff distance at 1520km. This indicates that a 1% increase in weighted R&D input
outside of this region will contribute to the growth of regional industrial value added
by 0.23%.
In contrast to the R&D spillovers (RDSe) in the table, both direct FDI spillovers (FDI)
and indirect FDI spillovers via other regions (FDIS) are less important. The
coefficient of direct FDI spillovers is insignificant. The indirect FDI spillover from
other regions is significant, but with a very low value, i.e. 0.053. The coefficient of
the interaction term (FDIRDSFDIS) is significant but with a very small value, which
suggests it does not have much influence on regional value added.
When the cutoff distance changes to 760 km, the coefficient of interregional R&D
spillovers is lower, 0.10 (at 5% level of significance), while the coefficient of direct
R&D investment is higher, 0.23 (at 1% level of significance). Both direct FDI
spillovers and indirect FDI spillovers via other regions are insignificant. This
indicates that the interregional spillover effect declines when the cutoff distance gets
smaller, whereas more contribution is captured by direct R&D investment. Comparing
the two tables with different distance cutoffs, we prefer a cutoff distance at 1520 km,
as in Table 8.2, because the results shown there are more significant than those in
Table 8.3.
Contribution of Technological Spillovers
171
Table 8.3: Estimates on R&D and FDI Spillovers in Chinese Regions,
with Cutoff Distance at 760 km (All Regions), 1991-2002
(1)K,L
(2) K, L, RD, RDS
(3)K,L,RD,RDS,FDI,FDIS
(4) ALL capital 0.317 0.351 0.336 0.343
(0.102)** (0.091)** (0.093)** (0.093)** labour 0.433 0.495 0.499 0.500 (0.060)** (0.055)** (0.064)** (0.064)** RDe 0.217 0.212 0.231 (0.033)** (0.035)** (0.040)** RDSe 0.100 0.103 0.100 (0.043)* (0.043)* (0.043)* FDI -0.014 -0.003 (0.012) (0.016) FDIS 0.012 0.029 (0.015) (0.023) FDIRDFDIS -0.001 (0.001) t 0.053 0.056 0.058 0.058 (0.010)** (0.009)** (0.010)** (0.010)** Constant 0.789 -1.311 -1.218 -1.401 (0.588) (0.599)* (0.605)* (0.632)* Observations 348 348 348 348 Number of regions 29 29 29 29 R-squared 0.75 0.81 0.81 0.81 Hausman Test FE, P>0.0002 FE, P>0.0000 FE, P>0.0000 FE, P>0.0000
Note: Same as Table 8.2.
In order to have a more detailed understanding of spillover contributions in different
geographic locations, we classify the national total into three groups: coastal, middle
and western regions (see Table 8.4 and Table 8.5). Three types of regional slope shift
dummies (dummies for interregional R&D spillovers, dummies for FDI spillovers,
and dummies for indirect FDI spillovers via other regions) are introduced. In order to
avoid the problem of using too many dummies in one model for 348 observations, we
include only one type of dummy each time. Namely, first we include the interregional
R&D spillover dummies (D1 is the RDSe dummy for middle regions, and D2 is the
RDSe dummy for western regions). Secondly, regional dummies on FDI spillover
effects are included (D3 is the FDI dummy for middle regions, and D4 is the FDI
dummy for western regions). Lastly, dummies for indirect FDI spillovers via other
regions are included (D5 is the FDIS dummy for middle regions, and D6 is the FDIS
dummy for western regions).
Chapter 8
172
Table 8.4: Estimates on R&D and FDI Spillovers in Chinese Regions,
with Cutoff Distance at 1520 km, 1991-2002 (with Regional Dummies) (1) With regional RDSe dummies
(2) With regional FDI dummies
(3) With regional FDIS dummies capital 0.322 capital 0.225 capital 0.257
(0.092)** (0.097)* (0.094)** labour 0.558 labour 0.557 labour 0.538 (0.061)** (0.060)** (0.060)** RDe 0.204 RDe 0.268 RDe 0.296 (0.044)** (0.048)** (0.052)** RDSe 0.169 RDSe 0.202 RDSe 0.202 (0.066)* (0.053)** (0.053)** FDI 0.028 FDI 0.129 FDI 0.047 (0.019) (0.037)** (0.020)* FDIS 0.059 FDIS 0.084 FDIS 0.143 (0.020)** (0.021)** (0.034)** FDIRDFDIS -0.002 FDIRDFDIS -0.004 FDIRDFDIS -0.004 (0.001)** (0.001)** (0.001)** DDDD1111 0.051 DDDD3333 -0.055 DDDD5555 -0.072 (0.086) (0.022)* (0.057) DDDD2222 0.176 DDDD4444 -0.090 DDDD6666 -0.075 (0.091) (0.027)** (0.024)** t 0.063 t 0.071 t 0.067 (0.010)** (0.010)** (0.010)** Constant -2.340 Constant -2.150 Constant -2.273 (0.637)** (0.632)** (0.627)** Observations 348 Observations 348 Observations 348 Number of regions 29 Number of regions 29 Number of regions 29 R-squared 0.82 R-squared 0.83 R-squared 0.83 Hausman Test FE
P>0.0000 Hausman Test FE P>0.0000 Hausman Test FE
P>0.0000 Note: 1) Coastal regions are Beijing, Tianjin, Shanghai, Hebei, Liaoning, Jiangsu, Zhejiang, Fujian,
Shandong, Guangdong, Gungxi, Hainan. Middle regions are Shanxi, Inner Mongolia, Jilin,
Heilongjiang, Anhui, Jiangxi, Henan, Hubei, Hunan. Western regions are Sichuan (including
Chongqing), Guizhou, Yunnan, Shanxi, Gansu, Qinghai, Ningxia, Xinjiang. Tibet is not included in
this analysis.
2) RDe is the ratio of S&T expenditure/GDP at year t-1; FDI is the ratio of FDI/GDP at year t-1.
3) FE means fixed effect regression, and RE means random effect regression. The Hausman test
is applied to choose a better regression out of FE and RE.
4) Standard errors in parentheses, * significant at 5% level, and ** significant at 1% level.
In Table 8.4, column 2 shows that the interregional R&D effect does not have
significant geographic implications. However, the regional dummies on the FDI
spillover effect are both negatively significant (see column 4). This means that FDI
spillover has a higher contribution to industrial value added in coastal regions, but a
lower contribution in both middle and western regions. (The elasticity of FDI is 0.129
in coastal regions, 0.07416 in middle regions, and 0.039
17 in western regions). The
16 0.074 =0.129-0.055.
Contribution of Technological Spillovers
173
last column indicates that the regional dummies on indirect FDI spillovers (via other
regions) are both negative again (insignificant for middle regions and significant for
western regions). The elasticity of FDIS is 0.143 in coastal regions, 0.07118 in middle
regions, and 0.06819 in western regions. This means that indirect FDI spillovers are
present mainly in coastal regions. In any case, the interregional R&D spillovers
always have more significant contributions than both direct FDI spillovers and
indirect FDI spillovers. Namely, a percentage change in RDS has a higher impact on
the change of value added than a percentage change in FDI and FDIS.
We now apply the same method to the case with a cutoff distance at 760 km (see
Table 8.5). All the coefficients for RDS, FDI and FDIS are smaller than those in
Table 8.4. The elasticity of RDS is not significant. The elasticity of FDI is 0.08 (at 5%
significance level) in coastal regions, 0.03920 (insignificant) in middle regions, and
0.00321 (at 5% significance level) in western regions. The elasticity of FDIS is 0.118
(at 1% significance level) in coastal regions, 0.08722 (insignificant) in middle regions,
and 0.02523 (at 1% significance level) in western regions. Given that the results are
less statistically significant than those in Table 8.4, with capital and constant
coefficients being often at the 5% significance level, we prefer the results with a
cutoff distance at 1520 km (Table 8.4).
17 0.039=0.129-0.090.
18 0.071=0.143-0.072.
19 0.068=0.143-0.075.
20 0.039=0.08-0.041.
21 0.003=0.08-0.077.
22 0.087=0.118-0.031.
23 0.025=0.118-0.093.
Chapter 8
174
Table 8.5: Estimates on R&D and FDI Spillovers in Chinese Regions,
with Cutoff Distance at 760 km, 1991-2002
(with Regional Dummies) (1) With regional RDSe dummies
(2) With regional FDI dummies
(3) With regional FDIS dummies capital 0.321 capital 0.249 capital 0.212
(0.095)** (0.100)* (0.100)* labour 0.505 labour 0.512 labour 0.494 (0.064)** (0.064)** (0.063)** RDe 0.232 RDe 0.291 RDe 0.289 (0.040)** (0.046)** (0.043)** RDSe 0.049 RDSe 0.079 RDSe 0.084 (0.053) (0.044) (0.043) FDI 0.002 FDI 0.08 FDI 0.025 (0.016) (0.038)* (0.018) FDIS 0.034 FDIS 0.06 FDIS 0.118 (0.023) (0.026)* (0.035)** FDIRDFDIS -0.001 FDIRDFDIS -0.002 FDIRDFDIS -0.002 (0.001) (0.001)* (0.001)** DDDD1111 0.079 DDDD3333 -0.041 DDDD5555 -0.031 (0.079) (0.024) (0.024) DDDD2222 0.165 DDDD4444 -0.077 DDDD6666 -0.093 (0.098) (0.031)* (0.027)** t 0.06 t 0.067 t 0.069 (0.010)** (0.011)** (0.010)** Constant -1.377 Constant -1.323 Constant -1.03 (0.632)* (0.630)* -0.633 Observations 348 Observations 348 Observations 348 Number of regions 29 Number of regions 29 Number of regions 29 R-squared 0.81 R-squared 0.81 R-squared 0.82 Hausman Test FE
P>0.0000 Hausman Test FE P>0.0000 Hausman Test FE
P>0.0000 Note: Same as Table 8.4.
8.5 Conclusions
This chapter explores the impact of technological spillovers on the growth of
industrial output in Chinese regions. Our empirical analysis focuses on two types of
technological spillovers, interregional R&D spillovers and international level FDI
spillovers.
The results of the analysis for the whole of China indicate that R&D spillover effects
are much stronger than FDI spillover effects. FDI contribution, if there is any, is
mainly located in coastal regions. Interregional R&D spillovers contribute more than
Contribution of Technological Spillovers
175
20 per cent to regional value added. This is a promising sign for middle and western
regions to catch up.
We therefore conclude that FDI spillovers are not the driving force of economic
growth in Chinese regional industries. In addition, interregional R&D spillover plays
a more important role than FDI. We furthermore conclude that the fast growth of
Chinese industry relies mainly on the regional R&D expenditure and interregional
R&D spillovers, not on FDI. The high coefficient of interregional R&D spillover
seems to indicate a regional catching-up process for middle and western regions.
Policy implications from this analysis are that, to keep a sustained economic growth,
the Chinese government should put emphasis on R&D improvement and on
facilitating the communication and transportation possibilities between regions.
CHAPTER 9
Conclusions
The economic reform of China has been widely regarded as a puzzling but successful
process. Along with China's sustained rapid growth and dramatic structural changes
since 1978, also the rise of regional inequalities has attracted much attention from
scholars. In 2004, the ratio of GDP per capita in the richest region (Shanghai) to the
poorest region (Guizhou) was more than 10 to 1, against 7.5 to 1 in 1991. Average
GDP per capita in the five richest regions was 5.3 times as high as that of the five
poorest regions, as against 3.8 to 1 in 1991 (CSY, 1996 and CSY, 2005). Obviously,
the benefits of growth are not spread evenly among different regions.
9.1 Summary and Conclusions
The goal of this research has been to provide an in-depth understanding of regional
disparities in Chinese manufacturing/industry. In order to do so, we formulated a
number of research questions, which were dealt with throughout this thesis. Are the
industrial productivity gaps among Chinese regions increasing or decreasing? Is there
a trend of convergence or divergence in the growth and productivity performance
between regions? How do ownership categories differ in their productivity
performance? How much do structural changes (sectoral, institutional and regional
shifts) contribute to the manufacturing/industrial growth? Are there contributions of
technological spillovers across Chinese regions or from foreign investment? And if so,
which level of technological spillovers (regional or international) is more important to
the growth or catch up of Chinese regions?
Considering the lack of consistently published data in China, our research started by
constructing a new regional database (Chapter 5), in particular focusing on capital
Chapter 9
178
input by region. The measurement of capital inputs is fraught with difficulties. Unlike
labour inputs, fixed assets are produced inputs that can be used repeatedly in the
production process over longer periods. Variation of service lives and the decline of
the productive capabilities of fixed assets over time make it hard to measure capital
inputs accurately. In the case of China, things are further complicated by the lack of
sufficient published data on investment in fixed assets and a measurement system that
still deviates from the System of National Accounts (SNA). In Chinese statistics,
fixed assets acquired in different years are normally valued at their historical
acquisition prices. Naturally, empirical research should be based on a sound database;
otherwise it is likely to lead to unreliable results.
This thesis attempts to solve the problem of unavailability of capital inputs at the level
of Chinese regions. We provided new estimates of capital inputs by Chinese regions,
including total economy (1953-2003), industry (1978-2003) and manufacturing
(1985-2003). The estimates for industry and manufacturing are broken down into
thirty regions. This thesis made a systematic attempt to apply SNA concepts to the
estimation of Chinese capital inputs, according to the Perpetual Inventory Method. It
also clearly distinguished between capital services and wealth capital stocks. After
having presented a general discussion of theoretical issues in capital measurement, we
provided a detailed analysis of the relevant Chinese statistical concepts and data. We
went on to discuss previous capital estimates in the light of modern conceptual and
theoretical discussions. Our database construction ended with an explanation of the
procedures followed in constructing the national and regional capital input series.
Based on our newly constructed database, in this thesis we analyzed the contribution
of structural changes to industrial productivity (Chapter 6), regional disparity and the
convergence/divergence trend (Chapter 7) and effects of technological spillovers on
regional industrial productivity growth and catch up (Chapter 8).
Using the shift-share decomposition methods, we find clear evidence of a structural
change bonus at the sectoral level, with sectoral shifts contributing 24 per cent to
overall productivity growth in manufacturing in the 1980s. However, just when
productivity growth accelerated in the 1990s, the contribution of the shift effect
dropped to a mere 3.3%. Our interpretation of this phenomenon is that the structural
Conclusions
179
changes in the early reform period of the 1980s resulted in a more efficient economic
structure, which provided a foundation for rapid intra-sectoral productivity growth
after the 1990s.
In marked contrast to sectoral changes, changes in the ownership structure contributed
negatively to overall productivity growth in the early 1980s. There was a negative
shift effect of around 25 per cent. This turned positive after 1985, reaching a peak of
23 per cent in the period 1992-1997, just when the shift effects of sectoral change
were negligible. The conclusion is that the reform of the ownership structure
contributed very substantially to the acceleration of productivity growth after 1992.
This is consistent with other more descriptive accounts of the Chinese reform process.
Institutional change has been especially important in the coastal regions. The
interesting contrast between the timing of the effects of sector structure and ownership
structure merits further examination.
Coastal regions did have higher productivity than inland regions, but there was no
clear tendency for coastal regions to forge ahead relative to regions in the west and in
the middle. In terms of contributions to productivity growth, the importance of the
coastal regions is confirmed by our analysis for all periods. For instance, between
1997 and 2002, seven regions - Guangdong, Shandong, Shanghai, Jiangsu, Zhejiang,
Heilongjiang and Liaoning - together account for 54 per cent of total productivity
growth.
The effects of regional change are much more modest than those of sectoral and
institutional change. Regional shifts contributed negatively to aggregate productivity
growth before 1992, and positively after 1992. During the period 1997-2002, there
was a positive shift effect of 6.18 per cent. Therefore, like institutional change,
regional change contributed positively to the acceleration of productivity growth.
Combining regional and ownership changes in this period, positive effects of regional
change were found in the joint-ownership category, the private-owned category and
the collective-owned category. Foreign-funded enterprises had the highest
productivity levels, but these were hardly affected by regional shifts.
Chapter 9
180
According to public perceptions of the Chinese growth experience, China is
characterised by increasing regional inequality. This was also our initial working
hypothesis, when we embarked on this research. However, our empirical results point
in the opposite direction. There has been no long-run divergence trend between
Chinese regions since 1978. On the contrary, there had been substantial regional
convergence from 1978 to around 1990. This was followed by a period of modest
divergence up until around 2001. After 2001, convergence trends resumed. Whatever
indicator was used, the degree of regional inequality was substantially lower than at
the beginning of the reform period.
An analysis of the relative importance of technological change and efficiency
provides an interesting interpretation of the Chinese reform experience. In the early
stages of the Chinese reform process efficiency changes predominated. Once
efficiency differentials between regions had been reduced in the process of efficiency
convergence, technological change at the frontier became more important as a driver
of growth in Chinese industry.
Considering the technology gaps between regions and the vast amount of increase of
foreign investment in China, our analysis (Chapter 8) explored the impact of
technological spillovers on the industry growth of Chinese regions. Our empirical
analysis involves two types of technological spillovers, national level (interregional
R&D spillovers) and international level (FDI spillovers). The results indicate that
there are stronger R&D spillover effects than FDI spillover contributions at the
national level. Regional spillovers are an important explanation of the catching-up in
middle regions of China.
Policy implications from this analysis are that, to keep a sustained economic growth,
the Chinese government should put emphasis on R&D improvement, and on
facilitating the communication and transportation possibilities between regions.
Conclusions
181
9.2 Implications for Further Research
This thesis presented new estimates of capital inputs by Chinese regions, including
total economy, industry and manufacturing (Chapter 5). This may provide a good
example for other research aimed at constructing a database of capital inputs by
manufacturing sector. Hence similar empirical research on manufacturing sectors
could be carried out.
Efficiency change and technological change are examined in Chapter 7 with data
envelopment analysis (DEA). By this non-parametric method, each region is
compared with the best performing region. This approach can demonstrate the gaps of
productive efficiency among regions. However, due to the fact that the frontier
formed by the DEA approach relies only on the observed best units, the average
efficiency scores for other regions might be sensitive to the outlier and noise problems
in the observations. On this aspect, the parametric stochastic frontier analysis (SFA)
has an advantage over the DEA approach. SFA includes a "noise" term in its model,
which captures measurement errors and random events. The frontier by SFA is based
on statistical methods instead of observations. Although SFA also has its
shortcomings as an arbitrary method greatly relying on stochastic models1, it is a good
supplementary technique to DEA.
Our results on the efficiency trend and decomposition of total factor productivity
(TFP) through the DEA method have been confirmed by the SFA approach. However,
due to the fact that the efficiency scores formulated by SFA are very unstable in
different production function models, the SFA estimates are not presented in this
thesis. Nevertheless, it would be interesting to apply SFA and compare its results with
those of DEA in a future study.
As our results in Chapter 7 showed, the efficiency change was important in the early
stages of the growth of Chinese industry. However, technological change at the
frontier becomes more important as a driver of growth in the later stage. This calls for
1 Some have argued that economic theories or models cannot always represent the real world situation.
Chapter 9
182
further research on technology and education changes and their contribution to
economic growth in China.
The contribution of technological spillovers to regional industry has been examined in
Chapter 8. The analysis is based on the relationship between technological indicators
and regional industrial output (value-added). We find that FDI spillover effects are
much less important than R&D spillover effects. In future research, if there are
sufficient data, it would be of interest to explore the connection between technological
spillovers and the regional technological level, e.g. using innovation or patent
numbers as the dependent variable.
Appendices
183
Appendix A:
Relationships between Rental Price and Value of Fixed Assets
The value of a s-year-old fixed asset at time t should be equal to all the profits gained
by this fixed asset in the following service years, deflated into the present value of t.
Assuming r is the discount rate, stP , is the rental price at year t aged s, the value of the
fixed asset can be expressed as
sTsT
TsTtstst
str
Scrap
r
P
r
P
r
PV
−−
−−+++
++
+++
++
+=
)1()1()1(1
,1)(
2
1,1,
, L (1)
The last term on right hand in eq.(1) is the deflated crap value of this particular fixed
asset when it is discarded at the end of its service life (year T).
Accordingly, after one year in use, its value will be
11
,1)(
2
2,21,1
1,1)1()1()1(1 −−−−
−−+++++
+++
++
+++
++
=sTsT
TsTtstst
str
Scrap
r
P
r
P
r
PV L (2)
(1)-(2)*(1/(1+r))
r
P
rVV
st
stst+
=+
⋅− ++11
1 ,
1,1, (3)
then we have
stststst PVVrV ,1,1,, =−⋅+ ++ (4)
If st ,δ is assumed as the depreciation rate of this s-year-old fixed asset at time t, then
st
stst
stV
VV
,
1,1,
,
++−=δ (5)
Then eq.(4) can produce
ststst PrV ,,, )( =+⋅ δ (6)
If we apply the efficiency rate sφ to substitute the stP , , furthermore, we will get
0,,, )( tsstst PrV ⋅=+⋅ φδ (7)
For a s+1 year-old fixed asset, we will get
0,1,1, )( tsstst PrV ⋅=+⋅ ++ φδ (8)
If we assume that the depreciation rate remains constant, dividing (8) by (7), we have
s
s
φ
φδ 11 +=− (9)
Thus the depreciation rate and the productive efficiency can be connected (see also,
Hulten, 1990 and OECD, 2001a, p.87).
Appendices
184
Appendix B:
Ownership Categories
(1) State-owned and state-holding enterprises refer to industrial enterprises where the
means of production or income are owned by the state, and the enterprises in
which the state holds majority shares.
(2) Collective-owned enterprises refer to industrial enterprises where the means of
production are owned collectively, including urban and rural enterprises financed
by collectives and some enterprises which were formerly owned privately but
have been registered in the industrial and commercial administration agency as
collective units through raising funds from the public. Some of the dynamic
village and township enterprises fall under this category.
(3) Share-holding corporations Ltd. refer to economic units registered in accordance
with the Regulation of the People's Republic of China on the Management of
Registration of Corporate Enterprises, with total registered capital divided into
equal shares and raised through issuing stocks. Each investor bears limited
liability to the corporation depending on the holding of shares, and the
corporation bears liability to its debt to the maximum of its total assets.
(4) Private enterprises refer to economic units financed or controlled (by holding the
majority of the shares) by natural persons who hire labour for profit-making
activities. Included in this category are private limited liability corporations,
private share-holding corporations Ltd., private partnership enterprises and
private sole investment enterprises registered in accordance with the Corporation
Law, Partnership Enterprise Law and Tentative Regulation on Private Enterprises.
(5) Foreign funded enterprises refer to all industrial enterprises registered as a join-
venture, cooperative, sole (exclusive) investment industrial enterprise, or limited
liability corporation with foreign funds.
(6) Enterprises with funds from Hong Kong, Macao and Taiwan refer to all industrial
enterprises registered as a joint-venture, cooperative, sole (exclusive) investment
industrial enterprise or limited liability corporation with funds from Hong Kong
Macao and Taiwan.
Appendices
185
Appendix C:
Regions in China
Ap
pen
dix
Tab
le D
-1:
Em
ploy
men
t in
Thr
ee I
ndus
trie
s an
d P
erce
ntag
es
N
umbe
r of
Em
ploy
ed P
erso
ns a
t th
e Y
ear-
end
Com
posi
tion
in P
erce
ntag
e
Nat
iona
l T
otal
P
rim
ary
indu
stry
Se
cond
ary
indu
stry
In
dust
ry
Con
stru
ctio
n
Ter
tiar
y In
dust
ry
Pri
mar
y in
dust
ry
Seco
ndar
y in
dust
ry
Indu
stry
C
onst
ruct
ion
Ter
tiar
y In
dust
ry
1978
40
152
2831
8 69
45
6091
85
4 48
90
0.71
0.
17
0.15
0.
02
0.12
19
80
4236
1 29
122
7707
67
14
993
5532
0.
69
0.18
0.
16
0.02
0.
13
1985
49
873
3113
0 10
384
8349
20
35
8359
0.
62
0.21
0.
17
0.04
0.
17
1989
55
329
3322
5 11
976
9569
24
07
1012
9 0.
60
0.22
0.
17
0.04
0.
18
1990
64
749
3891
4 13
856
9698
24
24
1197
9 0.
60
0.21
0.
15
0.04
0.
19
1991
65
491
3909
8 14
015
9947
24
82
1237
8 0.
60
0.21
0.
15
0.04
0.
19
1992
66
152
3869
9 14
355
1021
9 26
60
1309
8 0.
59
0.22
0.
15
0.04
0.
20
1993
66
808
3768
0 14
965
1046
7 30
50
1416
3 0.
56
0.22
0.
16
0.05
0.
21
1994
67
455
3662
8 15
312
1077
4 31
88
1551
5 0.
54
0.23
0.
16
0.05
0.
23
1995
68
065
3553
0 15
655
1099
3 33
22
1688
0 0.
52
0.23
0.
16
0.05
0.
25
1996
68
950
3482
0 16
203
1093
8 34
08
1792
7 0.
51
0.23
0.
16
0.05
0.
26
1997
69
820
3484
0 16
547
1076
3 34
49
1843
2 0.
50
0.24
0.
15
0.05
0.
26
1998
70
637
3517
7 16
600
9323
33
27
1886
0 0.
50
0.24
0.
13
0.05
0.
27
1999
71
394
3576
8 16
421
9061
34
12
1920
5 0.
50
0.23
0.
13
0.05
0.
27
2000
72
085
3604
3 16
219
8924
35
52
1982
3 0.
50
0.22
0.
12
0.05
0.
27
2001
73
025
3651
3 16
284
8932
36
69
2022
8 0.
50
0.22
0.
12
0.05
0.
28
2002
73
740
3687
0 15
780
9155
38
93
2109
0 0.
50
0.21
0.
12
0.05
0.
29
Not
e: 1
0 00
0 pe
rson
s, a
t yea
r-en
d.
Sou
rce:
fro
m C
SY 2
005,
Tab
le 5
-2, 5
-6.
Appendices
186
Ap
pen
dix
Tab
le D
-2:
GD
P in
Thr
ee I
ndus
trie
s an
d P
erce
ntag
es
G
ross
Dom
esti
c P
rodu
ct
Com
posi
tion
in P
erce
ntag
e
N
atio
nal
Tot
al
Pri
mar
y in
dust
ry
Seco
ndar
y in
dust
ry
Indu
stry
C
onst
ruct
ion
Tert
iary
In
dust
ry
Pri
mar
y in
dust
ry
Seco
ndar
y in
dust
ry
Indu
stry
C
onst
ruct
ion
Tert
iary
In
dust
ry
1978
36
24
1018
17
45
1607
13
8 86
1 0.
28
0.48
0.
44
0.04
0.
24
1979
40
38
1259
19
14
1770
14
4 86
6 0.
31
0.47
0.
44
0.04
0.
21
1980
45
18
1359
21
92
1997
19
6 96
6 0.
30
0.49
0.
44
0.04
0.
21
1981
48
62
1546
22
56
2048
20
7 10
61
0.32
0.
46
0.42
0.
04
0.22
19
82
5295
17
62
2383
21
62
221
1150
0.
33
0.45
0.
41
0.04
0.
22
1983
59
35
1961
26
46
2376
27
1 13
28
0.33
0.
45
0.40
0.
05
0.22
19
84
7171
22
96
3106
27
89
317
1770
0.
32
0.43
0.
39
0.04
0.
25
1985
89
64
2542
38
67
3449
41
8 25
56
0.28
0.
43
0.38
0.
05
0.29
19
86
1020
2 27
64
4493
39
67
526
2946
0.
27
0.44
0.
39
0.05
0.
29
1987
11
963
3204
52
52
4586
66
6 35
07
0.27
0.
44
0.38
0.
06
0.29
19
88
1492
8 38
31
6587
57
77
810
4510
0.
26
0.44
0.
39
0.05
0.
30
1989
16
909
4228
72
78
6484
79
4 54
03
0.25
0.
43
0.38
0.
05
0.32
19
90
1854
8 50
17
7717
68
58
859
5814
0.
27
0.42
0.
37
0.05
0.
31
1991
21
618
5289
91
02
8087
10
15
7227
0.
24
0.42
0.
37
0.05
0.
33
1992
26
638
5800
11
700
1028
5 14
15
9139
0.
22
0.44
0.
39
0.05
0.
34
1993
34
634
6882
16
429
1414
4 22
85
1132
4 0.
20
0.47
0.
41
0.07
0.
33
1994
46
759
9457
22
372
1936
0 30
13
1493
0 0.
20
0.48
0.
41
0.06
0.
32
1995
58
478
1199
3 28
538
2471
8 38
20
1794
7 0.
21
0.49
0.
42
0.07
0.
31
1996
67
885
1384
4 33
613
2908
3 45
31
2042
8 0.
20
0.50
0.
43
0.07
0.
30
1997
74
463
1421
1 37
223
3241
2 48
11
2302
9 0.
19
0.50
0.
44
0.06
0.
31
1998
78
345
1455
2 38
619
3338
8 52
31
2517
4 0.
19
0.49
0.
43
0.07
0.
32
1999
82
067
1447
2 40
558
3508
7 54
71
2703
8 0.
18
0.49
0.
43
0.07
0.
33
2000
89
468
1462
8 44
935
3904
7 58
88
2990
5 0.
16
0.50
0.
44
0.07
0.
33
2001
97
315
1541
2 48
750
4237
5 63
75
3315
3 0.
16
0.50
0.
44
0.07
0.
34
2002
10
5172
16
117
5298
0 45
975
7005
36
075
0.15
0.
50
0.44
0.
07
0.34
20
03
1173
90
1692
8 61
274
5309
3 81
81
3918
8 0.
14
0.52
0.
45
0.07
0.
33
2004
13
6876
20
768
7238
7 62
815
9572
43
721
0.15
0.
53
0.46
0.
07
0.32
N
ote:
100
mill
yua
n, a
t cur
rent
pri
ces.
The
GD
P nu
mbe
r in
tert
iary
indu
stry
200
4 is
65,
018
in th
e C
hina
Eco
nom
ic C
ensu
s 20
04.
Sou
rce:
fro
m C
SY 2
005,
Tab
le 3
-1.
Appendices
187
Appendices
188
Appendix Table D-3:
Comparisons of Discrepancy between National and Regional Resources in
Industry, 1989
1989
Number of
enterprises (unit)
GVO (100 mill
yuan)
(current prices)
Value added (100
mill yuan)
(current prices)
Employment-staff
and workers (year-average)
National
sources
Regional
sources
National
sources
Regional
sources
National
sources
Regional
sources
National
sources
Regional
sources
Beijing 5336 6175 602.7 622.4 175.2 175.2 171.8 178.9
Tianjin 5345 493.6 493.6 124.2 124.2 157.4
Hebei 19400 19400 705.9 705.9 347.2
Shanxi 11249 11249 390.1 390.1 121.5 121.5 237.5
Inner
Mongolia 7349 7350 218.2 67.8 142.0
Liaoning 21344 21344 1277.3 1277.3 381.7 572.4
Jilin 12290 459.0 459.0 134.8 246.2
Heilongjiang 14652 14625 730.3 730.3 261.5 374.6
Shanghai 10282 10282 1373.5 1373.5 371.8 351.9 351.9
Jiangsu 37919 37919 1891.9 1891.9 422.1 726.0 726.0
Zhejiang 41046 41046 993.5 993.5 253.4 396.7 396.7
Anhui 21511 21511 503.5 503.5 129.5 256.1 256.1
Fujian 12341 367.5 367.5 119.1 108.6 152.0
Jiangxi 14697 14697 335.8 335.8 194.8
Shandong 22571 22571 1371.2 1371.2 503.6 503.6
Henan 16372 16372 669.9 669.9 191.5 339.4 339.4
Hubei 19953 19953 807.5 807.5 249.6 360.9 360.9
Hunan 21233 21233 569.1 569.1 174.6 168.9 293.5
Guangdong 24623 24632 1321.3 342.5 342.5 387.9
Guangxi 9128 279.8 279.8 83.5 122.3
Hainan 832 832 31.6 31.6 10.7 9.4 14.0
Chongqing 10976 10976 299.1 299.1 158.8
Sichuan 36239 36239 934.6 934.6 346.6 284.1 495.9
Guizhou 5239 5239 173.3 173.3 71.1 63.9 90.5
Yunnan 6055 6055 271.0 271.0 110.2 103.1 103.1
Tibet 208 208 2.4 2.4 1.1 1.9 1.9
Shaanxi 11964 336.0 336.0 115.5 84.3 184.2 188.5
Gansu 5663 5665 223.8 77.5 112.5
Qinghai 1485 1202 54.2 50.6 20.1 27.8 24.4
Ningxia 1617 1617 52.0 52.0 18.1 17.1 27.8
Xinjiang 4069 4069 153.7 153.7 51.6 70.5 70.5
Note: Coverage is IAS (independent accounting system) industrial enterprises.
Source: National columns are from China Statistical Yearbook, 1990. Regional columns are from
regional yearbooks.
Appendices
189
Appendix Table D-4:
Comparisons of Discrepancy between National and Regional Resources in
Industry, 2003
2003
Number of enterprises
(100 mill yuan) Gross output
(100 mill yuan) Value added
(100 mill yuan)
(current prices) (current prices)
National
sources
Regional
sources
National
sources
Regional
sources
National
sources
Regional
sources
Beijing 4019 4019 3810.36 3810.37 1012.53 1012.53
Tianjin 5341 5341 4049.61 4049.61 1074.78 1074.79
Hebei 7923 7923 5708.76 5708.80 1801.75 1801.70
Shanxi 3613 3613 2439.30 2439.30 908.71 908.71
Inner
Mongolia 1653 1653 1355.70 1354.45 516.72 515.86
Liaoning 6842 6842 6112.96 6112.96 1715.92 1715.92
Jilin 2284 2284 2662.27 2662.27 814.83 814.83
Heilongjiang 2567 2567 2909.98 2910.00 1363.10 1363.10
Shanghai 11098 10956 10342.82 10342.81 2832.88
Jiangsu 23862 23861 18036.74 18034.12 4670.58 4670.41
Zhejiang 25526 25526 12864.23 12864.23 3097.62 3097.61
Anhui 4158 4158 2610.03 2610.21 881.47 881.52
Fujian 9208 9208 4953.74 4953.74 1448.50 1448.50
Jiangxi 3051 3051 1472.33 1472.33 446.78
Shandong 16177 16177 15379.54 15379.55 4701.10 4701.10
Henan 9091 9091 5365.65 1740.11 1754.08
Hubei 6271 6271 4030.11 4030.11 1364.76 1364.76
Hunan 5967 5967 2611.45 2611.45 888.56 888.56
Guangdong 24494 24494 21513.46 21513.46 5718.14 5718.14
Guangxi 2871 2871 1436.43 1436.43 446.48 446.48
Hainan 619 620 333.46 333.60 96.31
Chongqing 2241 2243 1588.00 1588.99 447.63 477.85
Sichuan 5448 5448 3387.43 3387.46 1165.69 1165.70
Guizhou 2129 2129 977.64 977.64 346.49 346.49
Yunnan 1995 1995 1557.17 1557.17 745.97 734.34
Tibet 325 325 21.39 21.39 12.39 12.39
Shaanxi 2493 2493 1879.26 1833.98 674.35 674.35
Gansu 2884 2884 1147.52 1147.68 388.10 346.25
Qinghai 400 400 247.90 247.90 95.23 95.23
Ningxia 418 420 352.81 394.79 109.41 122.84
Xinjiang 1254 1254 1113.14 1113.14 463.36 463.36
Note: The coverage is all state enterprises, plus all non-state enterprises with more than five million
yuan in annual sales. Some regional yearbooks show a slightly different coverage. In Shanxi Statistical
Yearbook 2004, the coverage is state-owned, large & medium-sized enterprises and non-state-owned
enterprises with sales over 5 million yuan. In Jilin 2004, the coverage is output value over 5 million
yuan industrial enterprises. In Tibet Statistical Yearbook 2004, the coverage is independent accounting
system at township level and above.
Source: National columns are from China Statistical Yearbook, 2004. Regional columns are from regional
yearbooks.
Ap
pen
dix
Tab
le D
-5:
Indu
stri
al E
mpl
oym
ent
in C
hine
se R
egio
nal,
1978
-200
2
19
78
1979
19
80
1981
19
82
1983
19
84
1985
19
86 1
987
1988
19
89
1990
19
91
1992
19
93
1994
19
95
1996
19
97
1998
19
99
2000
20
01
2002
N
atio
nal T
otal
39
80
4074
43
83
4640
48
57
5038
55
09
6401
68
28 7
103
7351
74
62
7507
77
90
7919
82
85
8511
84
57
8195
78
93
7570
70
52
6535
61
13
5939
B
eiji
ng
115
117
141
153
159
162
165
166
170
171
171
172
173
172
175
209
179
174
167
158
154
151
133
122
116
Tia
njin
12
1 12
2 12
7 13
5 14
2 13
8 14
4 14
9 15
2 15
8 15
7 15
7 15
9 15
6 16
0 17
8 17
8 17
4 16
6 15
9 17
2 15
5 14
1 13
3 13
0 H
ebei
20
8 21
2 21
6 22
4 23
1 23
4 26
7 28
5 30
7 32
2 33
9 34
7 34
7 35
8 37
0 39
9 41
6 40
7 39
0 38
1 35
9 34
1 31
7 29
3 28
2 Sh
anxi
14
1 14
6 15
4 16
3 16
9 17
2 17
7 20
3 21
7 22
2 23
3 23
7 24
3 24
5 25
4 26
8 26
2 26
9 26
3 25
7 25
4 22
9 21
6 20
1 19
5 In
ner
Mon
golia
95
98
10
2 10
7 10
9 11
2 11
4 12
5 13
1 13
1 13
7 14
2 14
8 15
3 15
6 15
1 14
9 14
6 15
4 14
4 11
9 11
0 10
0 91
85
L
iaon
ing
323
363
421
443
448
462
490
542
548
557
567
572
564
579
578
591
609
592
593
552
434
379
347
301
269
Jilin
11
1 11
2 11
5 11
9 12
5 12
7 12
9 22
2 23
3 23
6 24
0 24
6 25
2 25
6 25
6 26
8 26
0 26
2 24
4 23
7 20
8 17
9 15
9 13
7 12
3 H
eilo
ngji
ang
236
260
284
296
310
324
338
348
360
363
369
375
381
392
387
410
385
409
355
354
309
265
229
201
182
Shan
ghai
24
1 24
6 26
3 27
8 29
0 29
9 31
1 32
6 33
6 34
2 34
7 35
2 35
4 35
7 35
6 33
6 34
5 33
6 30
6 27
7 29
3 26
6 24
1 22
8 22
5 Ji
angs
u 45
5 40
7 45
7 49
6 51
9 53
8 57
0 62
1 67
5 71
2 73
2 72
6 71
8 76
1 75
8 76
9 77
9 76
0 73
6 70
7 70
7 66
8 60
9 57
9 57
2 Z
heji
ang
153
167
186
228
243
278
354
368
383
400
410
397
388
411
430
426
445
413
396
353
384
372
380
406
444
Anh
ui
116
118
127
131
140
145
160
221
227
237
253
256
255
267
273
296
313
317
325
302
226
208
191
172
161
Fuj
ian
71
77
74
80
94
101
113
127
135
141
149
152
155
158
161
167
194
194
187
196
192
180
183
185
193
Jian
gxi
86
87
88
91
95
103
114
170
181
187
195
195
194
202
202
206
219
219
209
197
163
146
128
113
103
Shan
dong
19
2 19
6 20
0 20
4 23
1 24
4 27
7 36
9 40
6 44
6 47
7 50
4 50
7 53
9 56
6 56
9 64
3 64
6 64
5 63
4 66
9 65
0 61
4 59
5 59
9 H
enan
16
1 15
8 16
5 16
9 17
1 18
5 20
4 28
4 30
9 32
0 33
6 33
9 34
2 36
0 37
6 40
3 41
6 45
1 43
8 44
2 46
5 43
5 40
6 36
8 34
7 H
ubei
16
0 16
2 17
3 18
6 19
8 20
3 29
2 30
7 33
4 34
4 35
6 36
1 35
9 36
7 35
8 35
8 38
1 38
0 37
0 36
3 33
5 30
4 27
1 24
1 22
1 H
unan
13
8 15
1 17
4 18
5 19
0 19
2 19
7 24
5 26
0 28
3 28
7 29
4 29
7 30
5 31
4 34
8 31
5 31
4 32
1 30
0 22
8 21
6 19
6 17
6 16
4 G
uang
dong
15
8 15
9 16
9 17
6 18
1 18
4 22
8 38
5 30
9 34
0 38
2 38
8 39
0 43
3 45
1 47
8 53
8 53
8 52
9 52
3 67
0 67
3 67
3 67
2 69
3 G
uang
xi
92
91
91
94
98
99
99
102
109
112
117
122
123
126
133
139
145
155
142
142
131
118
107
96
89
Hai
nan
13
13
13
13
13
13
14
14
14
14
14
14
15
16
16
19
16
17
16
15
16
14
14
14
14
Cho
ngqi
ng
95
98
100
113
115
118
121
125
147
151
155
159
161
165
166
175
169
172
189
176
142
122
107
95
88
Sich
uan
119
123
127
125
134
137
137
148
307
323
329
337
342
361
368
427
467
409
358
361
309
279
244
220
206
Gui
zhou
68
72
74
75
77
80
81
88
90
91
92
91
91
93
93
94
97
10
3 10
1 92
89
86
80
74
70
Y
unna
n 57
62
64
71
73
76
82
98
10
7 10
6 10
1 10
3 10
5 11
0 11
1 11
3 11
1 11
8 11
5 11
1 11
0 99
91
80
74
T
ibet
2
3 3
2 2
2 2
2 2
2 1
2 2
2 2
2 2
2 2
2 3
3 3
3 3
Shaa
nxi
112
119
127
135
143
152
162
165
172
176
181
184
192
194
197
200
203
202
204
196
175
166
147
132
123
Gan
su
59
58
61
63
65
67
70
91
98
103
107
112
120
118
119
137
130
125
124
120
115
111
107
95
88
Qin
ghai
20
19
20
20
20
20
21
25
26
26
27
28
28
28
28
29
30
30
30
28
23
23
19
16
15
N
ingx
ia
17
18
19
19
19
20
20
22
24
25
27
28
29
29
30
32
32
33
33
32
32
31
26
24
23
Xin
jian
g 45
42
48
49
51
51
56
59
60
62
63
70
72
75
76
85
85
88
87
84
83
74
55
48
44
N
ote:
100
00 p
erso
ns.
Cov
erag
e is
ent
erpr
ises
at t
owns
hip
leve
l and
abo
ve w
ith
inde
pend
ent a
ccou
ntin
g sy
stem
s. A
fter
199
8, it
cha
nges
to s
tate
ent
erpr
ises
plu
s en
terp
rise
s of
de
sign
ated
siz
e w
ith
mor
e th
an f
ive
mill
ion
yuan
. The
adj
ustm
ent f
or 1
998-
2002
is c
onsi
sten
t wit
h Sz
irm
ai a
nd R
en, 2
007.
(Se
e m
ore
deta
ils
in C
hapt
er 4
).
Sou
rce:
CIE
SY
var
ious
issu
es, S
CIT
(20
00),
and
adj
uste
d ac
cord
ing
Szi
rmai
and
Ren
, 200
7.
Appendices
190
Ap
pen
dix
Tab
le D
-6:
Indu
stri
al V
alue
Add
ed in
Chi
nese
Reg
iona
l, 19
78-2
005
1978
19
79
1980
19
81
1982
19
83
1984
19
85
1986
19
87
1988
19
89
1990
19
91
1992
19
93
1994
19
95
1996
19
97
1998
19
99
2000
20
01
2002
20
03
2004
20
05
Nat
iona
l tot
al
1373
14
48
1569
15
72
1659
17
96
2030
22
89
2406
26
95
2834
29
51
2934
31
68
3647
52
80
4906
46
40
5126
55
59
5463
61
44
7090
77
51
8974
11
132
1358
8 17
228
Bei
jing
62
66
71
69
71
75
83
86
95
102
98
102
97
103
119
169
193
138
128
147
148
167
202
206
229
268
312
400
Tia
njin
49
44
56
57
57
59
62
69
70
73
74
72
71
69
75
91
91
10
5 98
10
2 12
7 14
0 17
6 19
9 22
9 28
5 34
6 43
8 H
ebei
71
74
75
72
74
79
96
10
2 87
10
3 11
5 12
4 12
3 13
5 15
5 22
6 21
6 21
1 24
4 27
4 25
0 27
8 31
6 34
0 38
4 47
8 61
0 75
6 Sh
anxi
33
38
41
39
44
48
53
62
65
69
64
71
70
73
93
12
2 10
5 10
4 11
4 12
5 11
2 11
4 12
0 13
7 17
2 24
1 30
8 41
9 In
ner
Mon
goli
a 17
17
19
19
23
25
26
27
28
31
35
39
39
43
49
63
59
57
64
72
63
67
78
84
10
2 13
7 19
3 29
6 L
iaon
ing
136
138
146
134
137
150
170
195
220
235
223
228
211
216
250
385
347
250
260
269
241
267
333
344
375
455
559
742
Jili
n 41
43
44
47
50
53
62
67
68
82
85
83
78
81
94
13
2 11
6 96
10
6 10
4 10
0 11
7 13
9 16
1 18
2 21
6 24
7 27
9 H
eilo
ngjia
ng
85
87
96
93
96
100
108
106
126
144
150
152
158
163
191
229
219
205
227
235
219
266
339
330
343
361
402
514
Shan
ghai
18
6 18
7 19
2 19
3 19
6 19
9 20
2 21
7 23
8 23
8 22
1 22
1 21
5 23
2 26
5 35
9 35
6 32
3 30
5 38
6 37
5 43
9 47
1 54
4 58
0 75
1 85
0 98
4 Ji
angs
u 94
98
11
3 11
7 12
3 13
4 15
1 18
2 19
0 23
2 24
6 25
8 25
8 27
6 35
8 51
2 50
8 48
4 51
7 53
1 53
9 63
7 72
7 80
5 96
5 12
38
1599
19
38
Zhe
jiang
38
43
55
62
64
75
89
11
3 12
5 14
1 13
8 13
9 13
8 15
6 19
1 27
0 24
1 24
1 26
5 26
5 30
9 36
1 43
6 51
3 65
4 82
1 10
35
1153
A
nhui
24
25
27
28
31
37
45
52
57
63
72
78
80
83
91
16
7 14
7 13
5 18
2 20
5 12
5 14
1 14
2 15
8 18
8 23
4 26
8 35
4 Fu
jian
20
21
23
25
27
29
32
41
42
49
62
69
67
67
75
106
126
124
138
162
169
189
223
239
320
384
458
547
Jian
gxi
19
21
23
23
25
28
35
40
43
45
51
52
48
52
59
71
86
64
78
79
65
71
75
84
99
118
153
211
Shan
dong
87
89
98
10
1 10
8 11
5 14
8 16
3 16
4 19
9 18
4 20
0 20
9 22
9 28
4 45
4 33
7 41
0 48
8 51
9 53
1 59
8 71
2 79
4 95
2 12
46
1611
22
38
Hen
an
45
50
57
57
61
69
76
86
99
111
108
108
105
114
151
195
203
218
246
269
273
283
312
342
377
461
578
807
Hub
ei
50
54
67
68
72
80
99
114
117
133
120
125
122
131
149
247
225
184
211
277
253
270
282
293
318
362
413
479
Hun
an
42
47
50
50
54
58
63
78
81
89
98
102
104
110
110
134
131
121
157
159
123
132
147
166
192
236
297
389
Gua
ngdo
ng
53
55
59
67
72
78
86
104
113
141
183
199
215
269
323
518
445
465
541
584
687
794
956
1023
11
86
1516
17
57
2247
G
uang
xi
20
21
22
23
25
25
27
32
37
42
46
48
49
54
60
98
98
84
87
83
78
80
90
94
101
118
148
190
Hai
nan
2 2
2 2
2 3
3 4
4 5
5 6
7 7
8 12
12
9
9 9
14
16
18
18
22
26
25
36
Cho
ngqi
ng
19
21
21
22
23
26
29
33
30
33
40
44
38
39
55
76
70
58
66
67
59
68
79
84
98
119
144
157
Sich
uan
62
75
78
76
85
96
109
126
104
116
167
158
158
174
133
228
177
155
173
183
167
181
185
216
266
309
383
516
Gui
zhou
14
17
17
17
19
24
30
31
30
33
39
41
40
41
44
59
47
46
49
51
51
56
61
65
74
92
10
9 14
0 Y
unna
n 20
21
23
24
29
33
36
38
43
50
57
66
71
75
83
11
1 12
7 12
6 13
5 13
7 14
6 14
0 14
8 15
9 17
9 19
8 21
8 23
8 T
ibet
0
0 1
0 0
1 1
1 1
1 1
1 1
1 1
1 1
1 2
2 2
2 3
3 3
3 4
4 Sh
aanx
i 33
35
36
33
36
39
42
48
49
54
61
67
65
71
67
95
79
76
84
87
80
99
11
5 12
6 14
5 17
9 21
6 31
5 G
ansu
30
31
32
27
29
32
37
38
41
41
42
45
45
49
52
67
63
61
60
65
58
64
68
81
93
10
3 12
5 13
4 Q
ingh
ai
6 7
5 5
5 6
6 7
8 9
11
12
12
12
13
15
16
16
15
16
14
17
18
20
22
25
33
45
Nin
gxia
5
5 5
4 5
5 6
7 7
8 9
11
10
10
12
14
14
14
16
17
17
18
21
23
23
29
36
51
Xin
jiang
11
13
13
13
13
17
18
20
22
24
28
30
30
33
37
53
50
57
57
76
68
73
10
0 10
0 10
2 12
3 15
3 21
2 N
ote:
At 1
978
cons
tant
pri
ces.
Cov
erag
e is
ent
erpr
ises
at t
owns
hip
leve
l and
abo
ve w
ith
inde
pend
ent a
ccou
ntin
g sy
stem
s. A
fter
199
8, it
cha
nges
to s
tate
ent
erpr
ises
plu
s en
terp
rise
s of
des
igna
ted
size
wit
h m
ore
than
fiv
e m
illio
n yu
an.
Sou
rce:
CIE
SY
, CS
Y, a
nd r
egio
nal y
earb
ooks
, var
ious
issu
es.
Appendices
191
Appendices
192
Appendix Table D-7:
NIFA in Basic Construction and Technical Renovation (100 mill yuan)
Total
economy
(100 mill
yuan)
Industry
(100 mill
yuan)
Manufacturing
(100 mill yuan)
percentage of
industry in total
economy
(%)
percentage of
manufacturing in
total economy
(%)
1980 534.64
1981 526.64
1982 601.05
1983 675.71
1984 775.74
1985 1049.95 527.87 329.62 0.503 0.314
1986 1398.46 783.37 543.73 0.560 0.389
1987 1541.7 921.97 594.21 0.598 0.385
1988 1802.66 1093.27 713.97 0.606 0.396
1989 1815.91 1098.61 639.81 0.605 0.352
1990 2085.55 1298.71 755.64 0.623 0.362
1991 2357.03 1481.15 886.51 0.628 0.376
1992 3080.05 1781.9 1065.85 0.579 0.346
1993 4290.86 2289.97 1464.73 0.534 0.341
1994 5922.35 3096.52 1968.08 0.523 0.332
1995 7237.54 3673.11 2260.08 0.508 0.312
1996 9170.83 4591.42 2874.70 0.501 0.313
1996' 9124.96 4584.9 2869.62 0.502 0.314
1997 10693.7 5048.87 2911.83 0.472 0.272
1998 12190.18 5441.42 2791.51 0.446 0.229
1999 13163.97 5568.69 2854.98 0.423 0.217
2000 14543.16 5927.95 2689.27 0.408 0.185
2001 14524.16 5787.12 2990.02 0.398 0.206
2002 16682.85 6787.93 3686.18 0.407 0.221
2003 19247.73 8627.33 5359.19 0.448 0.278
Note: at current prices. Prior to 1996, total investment in fixed assets had a coverage of enterprises with
investment of more than 50 thousand yuan per year. However, except for investments in real estate
development, rural collective investment and individual investment, the coverage changed to more than
500 thousand yuan from 1997 onwards. Therefore the data for 1996 are published for the two types of
coverage.
Source: DSIFA1997, p.62 and DSIFA 2002, p.77.
Appendices
193
Appendix Table D-8:
Newly Invested Fixed Assets (NIFA) in Total Economy (100 mill yuan)
year NIFA (total economy) year
NIFA (total economy)
1953 75.22 1981 824.53
1954 83.69 1982 992.47
1955 91.03 1983 1187.23
1956 122.00 1984 1490.96
1957 142.52 1985 1950.03
1958 210.70 1986 2633.52
1959 257.47 1987 3100.73
1960 290.00 1988 3808.64
1961 117.78 1989 3758.43
1962 70.02 1990 3995.34
1963 97.19 1991 4649.8
1964 139.71 1992 6254.37
1965 207.23 1993 9278.63
1966 183.68 1994 11911.50
1967 97.30 1995 14521.72
1968 71.33 1996 18484.99
1969 134.14 1997 20706.71
1970 249.14 1998 22629.19
1971 238.28 1999 24634.09
1972 243.92 2000 26842.19
1973 323.59 2001 28184.88
1974 318.90 2002 32304.20
1975 381.69 2003 37732.01
1976 348.00 2004 45783.91
1977 446.34
1978 577.96
1979 695.26
1980 720.49
Note: at current prices.
Source: Data for NIFA during 1953-1980 are estimated based on stated-owned NIFA. Data for NIFA
during 1981-2004 are from DSIFA 1997, p. 62, and DSIFA 2002, p. 77.
Appendix Table D-9:
Productive Ratio in Newly Invested Fixed Assets (NIFA) in Total Economy
year 1953-
57
1958-
62
1963-
65
1966-
70
1971-
75
1976-
80
1981-
85
1986-
90
1991-
95
productive
ratio 0.670 0.854 0.794 0.838 0.825 0.739 0.547 0.671 0.669
Note: This ratio is applied to get the productive NIFA in total economy.
Source: from DSIFA, 1997, p.98.
References
Abramovitz, M. (1986), Catching up, forging ahead, and falling behind, Journal of
Economic History, Vol. 46 (2), pp. 385-406.
Ai, C. and M. Wen, (2005), Ownership and sector performance in current China - A
regional study, Annals of Economics and Finance, Vol. 6, pp. 303-317.
Aiello, F. and P. Cardamone, (2005), R&D spillovers and productivity growth:
evidence from Italian manufacturing microdata, Applied Economics Letters,
Vol. 12, pp. 625-631.
Aitken, B.J. and A.E. Harrison, (1999), Do domestic firms benefit from direct foreign
investment? Evidence from Venezuela, American Economic Review, Vol. 89
(3), pp. 605-618.
Ao, X. and L. Fulginiti (2003), Productivity Growth in China: Evidence from Chinese
Provinces, Paper presented at the American Association of Agricultural
Economics Meetings, Tampa, August, 2002.
Ark, B. van, and M.P. Timmer, (2003), Asia’s Productivity Performance and Potential:
The Contribution of Sectors and Structural Change, University of Groningen
and Conference Board, http://www.rug.nl/staff/h.h.van.ark/asia_paper4.pdf.
Balasubramanyam, V.N., M. Salisu, and D. Sapsford, (1996), Foreign direct
investment and growth in EP and IS countries, The Economic Journal, Vol.
106, pp. 92-105.
Banerjee, A.V. and E. Duflo, (2003), Inequality and growth: What can the data say?
Journal of Economic Growth, Vol. 8, pp. 267-299.
Banister, J. (2005), Manufacturing employment in China, Monthly Labor Review, July,
2005, pp. 11-29.
Banister, J. (2007), Manufacturing in China today: Employment and labor
compensation, The Conference Board, Economics Program Working Paper,
EPWP #07-01.
Barrios, S. and E. Strobl, (2006), The dynamics of regional inequalities, Documento
de Trabajo, 2006-01, http://www.fedea.es/.
Barro, R.J. and X. Sala-i-Martin, (1991), Convergence across states and regions:
Comments and Discussion, Brookings Papers on Economic Activity, Vol. 1, pp.
107-158.
References
196
Barro, R.J. and X. Sala-i-Martin, (1992), Convergence, The Journal of Political
Economy, Vol. 100 (2), pp. 223-251.
Barro, R.J. and X. Sala-i-Martin, (2004), Economic Growth, MIT.
Battese, G.E. and T.J. Coelli, (1992), Frontier production functions, technical
efficiency and panel data: With application to paddy farmers in India, The
Journal of Productivity Analysis, Vol. 3, pp. 153-169.
Battese, G.E. and T.J. Coelli, (1995), A model for technical inefficiency effects in a
stochastic frontier production function for panel data, Empirical Economics,
Vol. 20, pp. 325-332.
Baumol, W.J. (1967), Macroeconomics of unbalanced growth: The anatomy of urban
crisis, American Economic Review, Vol. 57 (3), pp. 415-426.
Baumont, C., C. Ertur and J. Le Gallo, (2000), Geographic spillover and growth: A
spatial econometric analysis for European regions, LATEC-Document de
travail-Economie, No. 2000-07.
Bernstein, J.I. and M.I. Nadiri, (1989), Research and development and intra-industry
spillovers: an empirical application of dynamic duality, Review of Economic
Studies, Vol. 56, pp. 249-269.
Bhalla, A. S. Yao, and Z. Zhang, (2003), Causes of inequalities in China, 1952 to
1999, Journal of International Development, Vol. 15 (8), pp. 939-955.
Bianchi, M. (1997), Testing for convergence: Evidence from non-parametric
multimodality tests, Journal of Applied Econometrics, Vol. 12 (4), pp. 393-409.
Branstetter, L.G. and R.C. Feenstra, (2002), Trade and foreign direct investment in
China: a political economy approach, Journal of International Economics, Vol.
58, pp. 335-358.
Buckley, P.J., C. Wang, and J. Clegg, (2007), The impact of foreign ownership, local
ownership and industry characteristics on spillover benefits from foreign direct
investment in China, International Business Review, Vol. 16, pp. 142-158.
Buckley, P.J., J. Clegg, and C. Wang, (2002), The impact of inward FDI on the
performance of Chinese manufacturing firms, Journal of International
Business Studies, Vol. 33 (4), pp. 637-655.
Bulli, S. (2001), Distribution dynamics and cross-country convergence: A new
approach, Scottish Journal of Political Economy, Vol. 48, No. 2, pp. 226-243.
References
197
Caniëls, M. and B. Verspagen, (2001), Barriers to knowledge spillovers and regional
convergence in an evolutionary model, Journal of Evolutionary Economics,
Vol. 11, pp. 307-329.
Cao, J., M.S. Ho, D.W. Jorgenson, R. Ren, L. Sun and X. Yue, (2007), Industrial and
aggregate measures of productivity growth in China, 1982-2000, International
Conference on Experiences and Challenges in Measuring National Income and
Wealth in Transition Economies, IARIW, Beijing, China, September 2007.
Chen, J and B.M.Fleisher, (1996), Regional income inequality and economic growth
in China, Journal of Comparative Economics, Vol. 22, pp. 141-164.
International Conference on Experiences and Challenges in Measuring
National Income and Wealth in Transition Economies, IARIW, Beijing, China,
September 2007.
Chen, K., G.H. Jefferson, T.G. Rawski, H.Wang and Y. Zheng, (1988), New estimates
of fixed investment and capital stock for Chinese state industry, The China
Quarterly, Vol. 114, pp. 243-266.
Chen, V.W., H.X. Wu, and B. van Ark, (2008), Measuring changes in
competitiveness in Chinese manufacturing industries across regions in 1995 –
2004: An unit labor cost approach, The Conference Board, Economics Working
Paper EPWP #08-03, http://www.conference-board.org/economics
/workingpapers.cfm.
Chenery, H.B. (1960), Patterns of industrial growth, American Economic Review, Vol.
50 (4), pp. 624-654.
Cheung, K. and P. Lin, (2004), Spillover effects of FDI on innovation in China:
evidence from the provincial data, China Economic Review, Vol. 15, pp. 25-14.
Chow, G.C. (1993), Capital formation and economic growth in China, The Quarterly
Journal of Economics, Vol. 108 (3), pp. 809-842.
Chow, G.C. (1994), Understanding China's Economy, World Scientific Publishing Co.
Pte.Ltd.
Chow, G.C. (2004), Economic reform and growth in China, Annals of Economics and
Finance, Vol. 5, pp. 127-152.
Chow, G.C. and K. Li, (2002), China's economic growth: 1952-2010, Economic
Development and Cultural Change, Vol. 51 (1), pp. 247-256.
Coe, D.T. and E. Helpman, (1995), International R&D spillovers, European
Economic Review, Vol. 39, pp. 859-887.
References
198
Coelli, T.J. (1996), A guide to FRONTIER version 4.1: A computer program for
stochastic frontier production and cost function estimation, CEPA Working
Papers, No. 7/96.
Coelli, T.J., D.S.P.Rao, C.J.O'Donnell, and G.E.Battese, (2005), An Introduction to
Efficiency and Productivity Analysis, Second Edition, Springer
Science+Business Media, Inc.
Cohen, W.M. and D.A. Levinthal, (1989), Innovation and learning: The two faces of
R&D, The Economic Journal, Vol. 99, pp. 569-596.
Cohen, W.M. and D.A. Levinthal, (1990), Absorptive capacity: a new perspective on
learning and innovation, Administrative Science Quarterly, Vol. 35 (1), pp.
128-152.
Coughlin, C.C. and E. Segev, (2000), Foreign direct investment in China: a spatial
econometric study, The World Economy, Vol. 23 (1), pp. 1-23.
Démurger, S., J.D. Sachs, W. T. Woo, S. Bao, G. Chang and A. Mellinger, (2002),
Geography, economic policy, and regional development in China, Asian
Economic Papers, Vol. 1 (1), pp. 146-197.
Deng, H. and R. McGuckin, with J. Xu,, Y. Liu and Y. Liu, (2005), The dynamics of
China’s labour market: productivity and jobs in China, Working Paper Series,
the International Centre for the Study of East Asian development, Kitakyushu
Vol. 2005-18.
Denison, E.F. (1967), Why Growth Rates Differ, Brookings.
Dinc, M., K. Haynes and W. Anderson, (1998), Regional Productivity Change in
Canadian Manufacturing: A multifactor/ Shift-Share Approach, Canadian
Journal of Regional Science, Vol. 21 (3), pp. 385-414.
Dosi, G, C. Freeman, R. Nelson, G. Silverberg and L. Soete (eds), (1998), Technical
Change and Economic Theory, London, Pinter.
Dowrick, S. and D. Nguyen, (1989), OECD comparative economic growth 1950-85:
Catch-up and convergence, American Economic Review, Vol. 79 (5), pp. 1010-
1030.
Ertur, C., J. Le Gallo, and C. Baumont, (2006), The European regional convergence
process, 1980-1995: do spatial regimes and spatial dependence matter?
International Regional Science Review, Vol. 29 (1), pp. 3-34.
Erumban, A.A. (2006), Lifetimes of machinery and equipment-Evidence from Dutch
Manufacturing, Groningen Growth and Development Centre, GD-87.
References
199
Esteban, J. (2000), Regional convergence in Europe and the industry mix: A shift-
share analysis, Regional Science and Urban Economics, Vol. 30, pp. 353-364.
European Commission, IMF, OECD, and World Band, (1993), System of National
Accounts, Brussels/ Luxembourg, New York, Paris, and Washington, DC.
Fabricant, S. (1942), Employment in Manufacturing 1899-1993, NEBR, New York.
Fagerberg, J. (2000), Technological progress, structural change and productivity
growth: A comparative study, Structural Change and Economic Dynamics, Vol.
11, pp. 393-411.
Fagerberg, J. and B. Verspagen, (2002), Technology-gaps, innovation-diffusion and
transformation: An evolutionary interpretation, Research Policy, Vol. 31, pp.
1291-1304.
Fagerberg, J. and M. M. Godinho, (2005), Innovation and Catching Up, in J.
Fagerberg, D.C. Mowery, and R.R. Nelson (eds), The Oxford Handbook of
Innovation, Oxford, Oxford University Press, Chapter 19.
Fagerberg, J., B. Verspagen, and M. Caniëls, (1997), Technology, growth and
unemployment across European regions, Regional Studies, Vol. 31 (5), pp.
457-466.
Färe, R., S. Grosskopf, M. Norris and Z. Zhang, (1994), Productivity growth,
technical progress, and efficiency change in industrialized countries, American
Economic Review, Vol. 84 (1), pp. 66-83.
Farrell, M.J. (1957), The measurement of productivity efficiency, Journal of the
Royal Statistical Society, Series A, Vol. 120, Part 3, pp. 253-282.
Feder, G. (1982), On exports and economic growth, Journal of Development
Economics, Vol. 12, pp. 59-73.
Fei, J. C. and G. Rains, (1964), Development of the labour-supply economy: theory
and policy, Irwin, IL, Homewood.
Firoz, N.N., F. Rezvani, T. Ramin, and W. Abdallah, (2003), Investing in China's
economic development zones: A managerial guide, International Journal of
Management, June 2003.
Forbes, K.J. (2000), A reassessment of the relationship between inequality and growth,
American Economic Review, Vol. 90 (4), pp. 869-887.
Freeman, C. and L. Soete, (1990), Fast structural change and slow productivity
change: Some paradoxes in the economics of information technology,
Structural Change and Economic Dynamics, Vol. 1 (2), pp. 225-242.
References
200
Fujita, M. and Thisse, J.F. (1996), Economics of agglomeration, Journal of the
Japanese and International Economies, Vol. 10, pp. 339-378.
Funke, M. and A. Niebuhr, (2005), Regional geographic research and development
spillovers and economic growth: evidence from West Germany, Regional
Studies, Vol. 39, pp. 143-153.
Gao, T. (2004), Regional industrial growth: evidence from Chinese industries,
Regional Science and Urban Economics, Vol. 34, pp. 101-124.
Gerschenkron, A. (1962), Economic Backwardness in Historical Perspective, Harvard
University Press, Cambridge Massachusetts.
Girma, S. (2005), Absorptive capacity and productivity spillovers from FDI: a
threshold regression analysis, Oxford Bulletin of Economics and Statistics, Vol.
67 (3), pp.281-306.
Girma, S. and K. Wakelin, (2001), Regional underdevelopment: Is FDI the solution:
A semiparametric analysis, GEP Workign Paper, 2001/14, University of
Nottingham.
Girma, S. and K. Wakelin, (2002), Are there regional spillovers from FDI in the UK?
In D. Greenaway, R. Upward and K. Wakelin (eds), Globalisation and Labour
Markets, Macmillan: Basingstoke.
Girma, S. and K. Wakelin, (2007), Local productivity spillovers from foreign direct
investment in the U.K. electronics industry, Regional Science and Urban
Economics, Vol. 37, pp. 399-412.
Girma, S., D. Greenaway and K. Wakelin, (2001), Who benefits from foreign direct
investment in the UK? Scottish Journal of Political Economy, Vol. 48 (2), pp.
119-133.
Glass, A. J. and K. Saggi, (1998), International technology transfer and the
technology gap, Journal of Development Economics, Vol. 55, pp. 369-398.
Griffith, R., S. Redding, and J. Van Reenen, (2004), Mapping the two faces of R&D:
productivity growth in a panel of OECD industries, The Review of Economics
and Statistics, Vol. 86 (4), pp. 883-895.
Griliches, Z. (1979), Issues in assessing the contribution of research and development
to productivity growth, The Bell Journal of Economics, Vol. 10, pp. 92-116.
Griliches, Z. (1992), The search for R&D spillovers, The Scandinavian Journal of
Economics, Vol. 94, Supplement, pp. S29-S47.
References
201
Groenewold, N., G. Lee and A. Chen, (2006), Papers in Regional Science, Vol. 86 (1),
pp. 101-122.
Grossman, G.M. and E. Helpman, (1991), Innovation and Growth in the Global
Economy, MIT Press, Cambridge, MA.
Grossmann, V. (2001), Inequality, Economic Growth, and Technological Change,
Physica-Verlag.
Hale, G. and C. Long, (2006), Firm ownership and FDI spillovers in China, Federal
Reserve Bank of San Francisco Working Paper Series, Working Paper 2006-25.
Hale, G. and C. Long, (2007), Are there productivity spillovers from foreign direct
investment in China? Federal Reserve Bank of San Francisco Working Paper
Series.
Harberger, A.C. (1998), A vision of the growth process, American Economic Review,
Vol. 88 (1), pp. 1-32.
Harhoff, D. (2000), R&D spillovers, technological proximity, and productivity
growth-Evidence from German panel data, Schmalenbach Business Review,
Vol. 52 (3), pp. 238-260.
Harrison, B. and J. Kluver, (1989), Reassessing the ‘Massachusetts Miracle’:
Reindustrialization and Balanced growth, or Convergence to
‘Manhattanization’? Environment and Planning A, Vol. 21, pp. 771-801.
Haskel, J.E., S.C. Pereira, and M.J. Slaughter, (2002), Does inward foreign direct
investment boost the productivity of domestic firms? NBER Working Paper
Series, No. 8724.
Haynes, K. and Dinc, M. (1997), Productivity change in manufacturing regions: A
multifactor/ shift-share approach, Growth and Change, Vol. 28, pp. 201-221.
Hirschman, A. (1973), The changing tolerance for income inequality in the course of
economic development, World Development, Vol.1 (12), pp. 29-36.
Holz, C.A. (2005), China's Economic Growth 1978-2025: What We Know Today
about China's Economic Growth Tomorrow, 2 Nov. 2005, at
http://ihome.ust.hk/~socholz.
Holz, C.A. (2006), China’s reform period economic growth: How reliable are Augus
Maddison’s estimates? Review of Income and Wealth, Vol. 52 (1), pp. 85-119.
Holz, C.A. (2006), New capital estimates for China, China Economic Review, Vol. 17,
pp. 142-185.
References
202
Holz, C.A. and Y. Lin, (2001), The 1997-1998 break in industrial statistics-Facts and
appraisal, China Economic Review, Vol. 12, pp. 303-316.
Hsueh, T.T. and Q. Li, (1999), China's National Income, Boulder: Westview Press.
Hu, A.G. (2001), Ownership, government R&D, private R&D and productivity in
Chinese industry, Journal of Comparative Economics, Vol. 29, pp. 136-157.
Hu, A.G.Z. and G.H. Jefferson, (2002), FDI impact and spillover: evidence from
China's electronic and textile industries, The World Economy, Vol. 25, pp.
1063-1076.
Huang, Y. and W.T. Woo, (1998), Free to lose: autonomy and incentives in Chinese
state enterprises, China Economy Papers, 1998/1.
Huang, Y., R. Ren and X. Liu, (2002), Capital stock estimates in Chinese
manufacturing by perpetual inventory approach, China Economic Quarterly,
Vol. 1 (2), pp. 377-396.
Huang, Y. (2004), Ownership Biases and FDI in China: Two Provinces, MIT Sloan
School of Management.
Hulten, C.R and Wykoff, F.C. (1996), Issues in the measurement of economic
depreciation: Introductory remarks, Economic Inquiry, 34 (1), pp. 10-23.
Hulten, C.R. (1990), The measurement of capital, in E. R. Berndt and J. E.Triplett
(eds), Fifty Years of Economic Measurement, National Bureau of Economic
Research Studies in Income and Wealth, Vol. 54, University of Chicago Press.
Jacob, J. (2006), International Technology Spillovers and Manufacturing Performance
in Indonesia, ECIS PhD thesis, Eindhoven.
Jacob, J. and A. Szirmai, (2007), International knowledge spillovers to developing
countries: the case of Indonesia, Review of Development Economics, Vol. 11(3),
pp. 550-565.
Jefferson, G.H., T.G. Rawski, L. Wang and Y. Zheng, (2000), Ownership,
productivity change, and financial performance in Chinese Industry, Journal of
Comparative Economics, 28, pp. 786-813.
Jefferson, G.H., T.G. Rawski, and Y. Zheng, (1992), Growth, efficiency, and
convergence in China's state and collective industry, Economic Development
and Cultural Change, Vol. 40 (2), pp. 239-266.
Jefferson, G.H., T.G. Rawski, and Y. Zheng, (1996), Chinese industrial productivity:
Trends, measurement issues, and recent developments, Journal of Comparative
Economics, Vol. 23, pp. 146-180.
References
203
Jian., T, J.D. Sachs, and A. M. Warner, (1996), Trends in regional inequality in China,
China Economic Review, Vol. 7 (1), pp. 1-21.
Jorgenson, D.W. (1963), Capital theory and investment behavior, American Economic
Review, Vol. 53 (2), pp. 247-259.
Jorgenson, D.W. (1973), The economic theory of replacement and depreciation, in
Econometrics and Economic Theory, W. Sellekaerts (ed.), New York:
Macmillan, pp. 189-221.
Jorgenson, D.W. (1990), Productivity and economic growth, in E. R. Berndt and J.
E.Triplett (eds), Fifty Years of Economic Measurement, National Bureau of
Economic Research Studies in Income and Wealth, Vol. 54, University of
Chicago Press.
Jorgenson, D.W. (1999), New methods for measuring capital, paper prepared for
Canberra Group on Capital Stock Statistics, Washington DC, 8 to 10
November, 1999.
Kanbur, R. and X. Zhang, (2005), Fifty years of regional inequality in China: a
journey through central planning, reform, and openness, Review of
Development Economics, Vol. 9 (1), pp. 87-106.
Keller, W. (2002), Geographic localization of international technology diffusion,
American Economic Review, Vol. 92 (1), pp. 120-142.
Knudsen, D. (2000), Shift-share analysis: further examination of models for the
description of economic change, Socio-Economic Planning Sciences, Vol. 34,
pp. 177-198.
Kumbhakar, S.C. and C.A.K.Lovell, (2000), Stochastic Frontier Analysis, Cambridge
University Press.
Kuznets, S. (1955), Economic growth and income inequality, American Economic
Review (Nashville), Vol. 45 (1), pp. 1–28
Kuznets, S. (1979), Growth and structural shifts, In: W. Galenson, ed., Economic
Growth and Structural change in Taiwan: The Postwar Experience of the
Republic of China, Ithaca, N.Y.: Cornell University Press.
Li, D. and C. Wu, (2002), The colour of the cats, The Economic and Social Review,
Vol. 33 (1), pp. 133-146.
Li, X., X. Liu, and D. Parker, (2001), Foreign direct investment and productivity
spillovers in the Chinese manufacturing sector, Economic Systems, Vol. 25, pp.
305-321.
References
204
Lin, J. Y. (2004), Lessons of China's transition from a planned economy to a market
economy, WSPiZ and TIGER Distinguished Lectures Series n. 16, Warsaw,
December 2004.
Lin, J.Y., F. Cai and Z. Li, (1996), The lessons of China's transition to a market
economy, Cato Journal, Vol. 16 (2), pp. 201-231.
Liu, X., D. Parker, K. Vaidya, and Y. Wei, (2001), The impact of foreign direct
investment on labour productivity in the Chinese electronics industry,
International Business Review, Vol. 10, pp. 421-439.
Liu, Z. (2002), Foreign direct investment and technology spillover: evidence from
China, Journal of Comparative Economics, Vol. 30 (3), pp. 579-602.
López-Bazo, E., E. Vayá and M. Artís, (2004), Regional externalities and growth:
evidence from European regions, Journal of Regional Science, Vol. 44 (1), pp.
43-73.
López-Bazo, E., Vayá, E., Mora, A.J. and J. Suriňach, (1999), Regional economic
dynamics and convergence in the European union, The Annals of Regional
Science, Vol.33, pp. 343-370.
Los, B. and B. Verspagen, (2000), R&D spillovers and productivity: evidence from
U.S. manufacturing microdata, Empirical Economics, Vol. 25, pp. 127-148.
Lucas, R.E. (1988), On the mechanics of economic development, Journal of
Monetary Economics, Vol. 22, pp. 3-42.
Lucas, R.E. (1993), Making a miracle, Econometrica, Vol. 61(2), pp. 251-272.
Madariaga, N. and S. Poncet, (2007), FDI in Chinese cities: Spillovers and impact on
growth, The world Economy, Vol. 30 (5), pp. 837-862.
Maddison, A. (1987), Growth and slowdown in advanced capitalist economies:
Techniques of quantitative assessment, Journal of Economic Literature, Vol.
XXV, pp. 649-698.
Maddison, A. (1998), Chinese Economic Performance in the Long Run, OECD, Paris.
Maddison, A. (2006), Do official statistics exaggerate China's GDP growth? A reply
to Carsten Holz, Review of Income and Wealth, Vol. 52 (1), pp. 121-126.
Maddison, A. and H.W. Wu, (2008), Measuring China’s Economic Performance,
World Economics, Vol. 9 (2), pp. 13-44.
Mankiw, N.G., D. Romer, and D.N. Weil, (1992), A contribution to the empirics of
economic growth, The Quarterly Journal of Economics, Vol. 17 (2), pp. 407-
437.
References
205
Martin, P. (1999), Are European regional policies delivering? European Investment
Bank (EIB), Vol. 4 (2), pp. 10-23.
McGuckin, R. H. and M. Spiegelman, with J. Xu, Y. Liu and Y. Jiang, (2004),
Restructuring China’s industrial sector: productivity and jobs in China, East
Asian Economic Perspectives, Vol.15 (2), pp. 60-79.
Mohnen, P. (1996), R&D externalities and productivity growth, STI Review, Vol. 18,
pp. 39-66.
Montobbio, F. (2002), An evolutionary model of industrial growth and structural
change, Structural Change and Economic Dynamics, Vol. 13, pp. 387-414.
Nelson, R.R. and H. Pack, (1999), The Asian miracle and modern growth theory, The
Economic Journal, Vol. 109, pp. 416-436.
Nishimizu, M. and J.M. Page, (1982), Total factor productivity growth, technological
progress and technical efficiency change: Dimensions of productivity change
in Yugoslavia, 1965-78, The Economic Journal, Vol. 92 (368), pp. 920-936.
OECD, (1993), Methods Used by OECD Countries to Measure Stocks of Fixed
Capital, Paris OECD.
OECD, (2001a), Measuring Capital-OECD Manual, Measurement of Capital Stocks,
Consumption of Fixed Capital and Capital Services, Paris, OECD.
OECD, (2001b), Measuring Productivity-OECD Manual: Measurement of Aggregate
and Industry-level Productivity Growth, Paris, OECD.
Orlando, M.J. (2004), Measuring spillovers from industrial R&D: on the importance
of geographic and technological proximity, Rand Journal of Economics, Vol.
35 (4), pp. 777-786.
Osada, H. (1994), Trade liberalisation and FDI incentives in Indonesia: The impact on
industrial productivity, The Developing Economics, Vol. 32 (4), pp. 479-91.
Paci, R. and F. Pigliaru, (1997), Structural change and convergence: An Italian
regional perspective, Structural Change and Economic Dynamics, Vol. 8, pp.
297-318.
Peneder, M. (2003), Industrial structure and aggregate growth, Structural Change and
Economic Dynamics, Vol. 14, pp. 427-448.
Qian, Y. (2003), How Reform Worked in China, In: D. Rodrik (eds), In Search of
Prosperity: Analytic Narratives on Economic Growth, Princeton University
Press, pp. 297-333.
References
206
Qian, Y. and J. Wu, (1999), China's transition to a market economy: How far across
the river? Paper prepared for the Conference on Policy Reform in China at the
Center for Research on Economic Development and Policy Reform, Stanford
University, November 18-20, 1999.
Quah, D.T. (1996), Convergence empirics across economies with (some) capital
mobility, Journal of Economic Growth, Vol. 1, pp. 95-124.
Raut, L.K. (1995), R&D spillovers and productivity growth: evidence from Indian
private firms, Journal of Development Economics, Vol. 48, pp. 1-23.
Reinsdorf, M. and M. Cover, (2005), Measurement of Capital Stocks, Consumption of
Fixed Capital, and Capital Services, Report on a presentation to the Central
American Ad Hoc Group on National Accounts, Santo Domingo, Dominican
Republic.
Romer, P.M. (1986), Increasing returns and long-run growth, Journal of Political
Economy, Vo. 94 (5), pp. 1002-1037.
Romer, P.M. (1990), Endogenous technological change, The Journal of Political
Economy, Vol. 98 (5), pp. S71-S102.
Schreyer, P., P.Bignon and J. Dupont, (2003), OECD capital services estimates:
Methodology and a first set of results, OECD Statistics Working Paper, 2003/6.
Shorrocks, A. and G. Wan, (2004), Spatial Decomposition of Inequality, WIDER
Discussion Paper No. 2004/01.
Sun, L. and R. Ren, (2007), Estimates of capital input index by industries: The
People's Republic of China (1980-2000), IARIW Conference, Beijing, China.
Syrquin, M. (1984), Resource allocation and productivity growth, in: M. Syrquin, L.
Taylor, and L.E. Westphal (eds), Economic Structure and Performance –
Essays in Honor of Hollis B. Chenery. Orlando, Florida, Academic Press, pp.
75-101.
Syrquin, M. (1988), Patterns of structural change, in: H.B. Chenery and T.N.
Srinivasan (eds), Handbook of Development Economics, Vol.I, Amsterdam,
North Holland, pp. 205-273.
Szirmai, A and R. Ren, (2007), Measuring Labour Productivity in Chinese
Manufacturing: Statistical Problems and Solutions, in B. Pant (eds), The
National Accounts of the People's Republic of China, Problems, Recent
References
207
Developments and the Way Forward, Asian Development Bank, Manilla, pp.
58-113.
Szirmai, A. (2005), The Dynamics of Socio-Economic Development, Cambridge,
Cambridge University Press.
Szirmai, A. (2008), Explaining success and failure in development, Inaugural lecture
in Maastricht University, February 2008. .
Szirmai, A., M.Timmer and R. van der Kamp, (2002), Measuring Embodied
Technological Change in Development. The Core Machinery Approach,
Journal of Development Studies, Vol. 39 (2), pp. 155-177
Szirmai, A., R. Ren and M. Bai, (2005), Chinese manufacturing performance in
comparative perspective, 1980-2002, Yale Economic Growth Center,
Discussion Paper No. 920, July 2005.
Timmer, M.P. (1999), Indonesia's ascent on the technology ladder: Capital stock and
total factor productivity in Indonesian manufacturing, 1975-95, Bulletin of
Indonesian Economic Studies, Vol.35 (1), pp. 75-97.
Timmer, M.P. (2000), The Dynamics of Asian Manufacturing- A Comparative
Perspective, 1963-1993, Cheltenham, Edward Elgar.
Timmer, M.P. and A. Szirmai, (2000), Productivity in Asian manufacturing: The
structural bonus hypothesis examined, Structural Change and Economic
Dynamics, Vol. 11, pp. 371-392.
Toft, G.S. and R.R. Stough, (1986), Transportation employment as a source of
regional economic growth: A shift-share approach, Transportation Research
Board, National Research Council, Washington, D.C.
Tsui, K. (2007), Forces shaping China’s interprovincial inequality, Review of Income
and Wealth, Vol. 53 (1), pp. 60-92.
Triplett, J.E. (1996), Depreciation in production analysis and in income and wealth
accounts: Resolution of and debate, Economic Inquiry, Vol. 34 (1), pp. 93-115.
Triplett, J.E. (1997), Concepts of capital for production accounts and for wealth
accounts: The implications for statistical programs, Paper prepared for
International Conference on Capital Stock Statistics, March 10-14, 1997,
Canberra, Australia.
Triplett, J.E. (1998), A Dictionary of Usage for Capital Measurement Issues,
Brookings Institution, Washington D.C., August, 1998.
References
208
Tsui, K.Y. (1991), China's regional inequality, 1952-1985, Journal of Comparative
Economics, Vol. 15, pp. 1-21.
Wang, L. and A. Szirmai, (2008a), Productivity growth and structural change in
Chinese manufacturing, 1980-2002, Industrial and Corporate Change, Vol. 17
(4), pp. 841–874.
Wang, L. and A. Szirmai, (2008b), Regional capital inputs in Chinese industry and
manufacturing, 1978-2003, Joint working paper of UNU-MERIT and the
Groningen Growth and Development Centre, UNU-MERIT Working Paper.
2008-028.
Verspagen, B. (1991), A new empirical approach to catching up or falling behind,
Structural Change and Economic Dynamics, Vol. 2 (2), pp. 359-380.
Verspagen, B. (2000), Growth and Structural Change: Trends, Patterns and Policy
Options, ECIS Working Paper.
Verspagen, B. (2001), Economic growth and technological change: An evolutionary
interpretation, OECD STI Working Papers, 2001/1
Vial, V. (2006), Total factor productivity growth in Indonesian manufacturing: using
estimates of elasticities of value added with respect to labour and capital,
Bulletin of Indonesian Economic Studies, Vol. 42(3), pp. 357-369.
Ward, M. (1972), The Measurement of Capital, OECD.
Wei, Y. and X. Liu, (2006), Productivity spillovers from R&D, exports and FDI in
China's manufacturing sector, Journal of International Business Studies, Vol.
37, pp. 544-557.
Williamson, J.G. (1965), Regional inequality and the process of national development:
A description of the patterns, Economic Development and Cultural Chang,
Vol. 13, pp. 1-84.
Wölfl, A. (2004), Productivity growth in services industries: Is there a role for
measurement? International Productivity Monitor, Vol. 8, pp. 66-80.
Woo, W.T., W. Hai, Y. Jin, and G. Fan, (1994), How successful has Chinese
enterprises reform been? Pitfalls in opposite biases and focus, Journal of
Comparative Economics, Vol. 18, pp. 410-437.
Wu, H.X. and X. Xu, (2002), Measuring the capital stock in Chinese industry-
Conceptual issues and preliminary results. Paper prepared for the 27th General
Conference of International Association for Research in Income and Wealth.
Stockholm, Sweden, August 2002.
References
209
Wu, H. X. (2000), China's GDP level and growth performance: Alternative estimates
and the implications, Review of Income and Wealth, Vol. 46 (4), pp. 475-499.
Wu, Y. (2000), Is China's economic growth sustainable? A productivity analysis,
China Economic Review, Vol. 11, pp. 278-296.
Wu, Y. (2004), China's Economic Growth: A Miracle with Chinese Characteristics,
Routledge, Curzon.
Xu, L. C. (2000), Control, incentives and competition: The impact of reform on
Chinese state-owned enterprises, Economics of Transition, Vol. 8(1), pp. 151-
173.
Zhang, A. and Y. Zhang, (2001), Impact of ownership and competition on the
productivity of Chinese enterprises, Journal of Comparative Economics, Vol.
29, pp. 327-346.
Zhang, Q. and B. Felmingham, (2002), The role of FDI, exports and spillover effects
in the regional development of China, Journal of Development Studies, Vol. 38
(4), pp. 157-178.
Statistical Sources
Census 1985: National Bureau of Statistics, (1987-88), Office of Leading Group of the
National Industrial Census under the State Council, Peoples Republic of China,
Industrial Census 1985, Vol. I-X, China Statistics Press (in Chinese).
Census 1995: National Bureau of Statistics, Office of the Third Industrial Census
(1997), Vol: The Data of the Third National Industrial Census of the People's
Republic of China in 1995, China Statistics Press.
China Economic Census Yearbook 2004, China Statistics Press.
CIESY: National Bureau of Statistics, Zhongguo Gongye Jingji Tongji Nianjian
(Chinese Industrial Economy Statistics Yearbook), various issues.
CLSY: National Bureau of Statistics, Zhongguo Laodong Tongji Nianjian (China
Labour Statistical Yearbook), various issues.
CSY: National Bureau of Statistics, China Statistical Yearbook, Beijing: China
Statistics Press, various issues.
References
210
DSIFA (Department of Statistics on Investment in Fixed Assets National Bureau of
Statistics of China), (1997), China Statistical Yearbook on Investment in Fixed
Assets, 1950-1995, China Statistics Press.
DSIFA (Department of Statistics on Investment in Fixed Assets National Bureau of
Statistics of China), (2002), Statistics on Investment in Fixed Assets of China,
1950-2000, China Statistics Press. 2002.
Regional Statistical Yearbooks, various issues, various regions, 1980-2005.
IO Table 1995: National Bureau of Statistics, P.R. China, Department of National
Accounts, Input-Output Table of China, 1995, China Statistics Press, 1997.
IO Table 1997: National Bureau of Statistics (1999), P.R. China, Department of
National Accounts, Input-Output Table of China, 1997, China Statistics Press,
1999.
Aggregate and Regional Productivity Growth in
Chinese Industry, 1978-2002
Summary
This book analyses the growth experience in Chinese industry and manufacturing,
with a special emphasis on the decomposition of growth, structural change, regional
divergence and convergence, and technology spillovers. The decomposition analysis
focuses on three dimensions: sectoral, regional and institutional. The book examines
regional productivity differentials and convergence or divergence trend in regional
industry. It includes an analysis of the regional, institutional and technological sources
of growth.
Chapter 2 provides a general review of the literature on regional disparity, structural
change and technological spillovers. In later chapters the insights from this literature
are applied to the analysis of China's regional industrial performance.
In Chapter 3, a summary is provided of the aggregate growth in China since 1978.
This chapter describes the main stages of the reform process and the corresponding
institutional changes. As the largest transition economy in the world, China's reform
has been carried out through step-by-step experimentations. From conventional points
of view, the reform process might be puzzling to some researchers. However, the
growth resulting from the reforms is unmistakable. China's GDP has grown more than
9 per cent per year after 1978 according to the official data provided by China's
National Bureau of Statistics (NBS). China’s institutional change is characterized by a
big drop of the share of state-owned enterprises and an increase of various different
ownership types. Township and village enterprises (TVEs), which started from the
negligible small community-funded enterprises, have shown an impressive growth
and contribution to China's economy. In addition, Chapter 3 also presents a survey of
the developments of technology indicators and education levels in China. Along with
the openness policy, the changes of foreign investment and trade are also discussed.
Summary
212
Chapter 4 tackles data issues and statistical problems at both the national level and
regional levels. Adjustments for value added and labour are made to create consistent
time series under comparable coverage.
In Chapter 5, we provide new estimates of capital inputs in the Chinese economy.
Estimates are made for the total economy (1953-2003), for the industrial sector (1978-
2003) and for the manufacturing sector (1985-2003). The estimates for industry and
manufacturing are broken down into thirty regions. This chapter makes a systematic
attempt to apply SNA concepts to the estimation of Chinese capital inputs, according
to the Perpetual Inventory Method. It makes a clear distinction between capital
services and wealth capital stocks. After a general discussion of theoretical issues in
capital measurement, a detailed analysis of the relevant Chinese statistical concepts
and data is provided.
Chapter 6 focuses on the contribution of structural change to aggregate manufacturing
performance in China. Since the start of the reform period the booming Chinese
economy has experienced rapid structural change. Using shift-and-share techniques,
this chapter examines three types of structural change: changes in the sectoral
structure of production, changes in the regional structural of production and changes
in the ownership structure. Overall productivity growth was slow in the 1980s, but
accelerated dramatically from 1990 onwards. In 1980s, we found evidence of a
structural change bonus, with sectoral shifts contributing 24% to overall productivity
growth. However, when productivity growth accelerated in the 1990s, the contribution
of the shift effect dropped to a mere 3.3%. In contrast to sectoral changes, changes in
the ownership structure in the early 1980s contributed negatively to overall
productivity growth. The contributions of ownership change turned positive after
1985, reaching 23% of productivity growth in the period 1992–1997. Shifts in
ownership explain a substantial part of productivity growth during the productivity
boom. Like shifts in ownership, regional shifts initially contributed negatively to
productivity growth till 1992, and positively thereafter. However, the general
contribution of regional shifts is lower than the contributions of sectoral and
ownership shifts. Contrary to initial expectations, the regional analysis of productivity
trends does not indicate regional divergence.
Summary
213
Chapter 7 explores the extent to which there is regional productivity divergence or
productivity convergence in Chinese manufacturing. Traditional regression methods
are based on the relationships between productivity growth rates and initial
productivity levels. Instead of these methods, we use the stochastic kernel density
approach, which provides a better view of distribution dynamics. Besides the
commonly used variables such as GDP per capita and labour productivity, we use
Data Envelopment Analysis to measure the productive efficiency of manufacturing in
Chinese regions relative to best regional practice. The evolution of regional
productivity performance can thus be compared among 30 Chinese regions. Our
results show that there was substantial regional convergence from 1978 to around
1990. This was followed by a period of modest divergence up till around 2001. After
2001, convergence trends resumed. Whatever indicator was used, the degree of
regional inequality was substantially lower than at the beginning of the reform period.
In Chapter 8, the contribution of technological spillovers in the process of industrial
growth and catching-up in Chinese regions is analyzed. Concerning the sources of
technological spillovers in Chinese regions, we distinguish between the regional level
and the international level. The former refers to R&D inputs in other regions, the
latter concerns international R&D investment which is embodied in foreign direct
investment (FDI). Our analysis covers the impact of spillovers from R&D in other
regions, from FDI in the own region, as well as FDI in other regions. Our empirical
analysis indicates that there are stronger R&D spillover effects than FDI spillover
contributions at the national level. Regional spillovers are an important explanation of
the catching-up in middle regions of China.
Chapter 9 concludes the whole thesis, with a brief discussion of the main results of
our analysis, and some indications for further research.
215
Curriculum Vitae
Lili Wang was born in Hebei, China. She studied technological economics and
management at the Hebei University of Technology where she got her bachelor and
master degrees in 1997 and 2000, respectively. Between 2000 and 2004 she worked as
a lecturer at the Hebei University of Technology. From 2004 till 2008, she worked as
a PhD student at the faculty of technology management, Eindhoven University of
technology. Her research has focused on regional productivity and convergence,
structural change and technological spillovers in China. Since September 2008, she
has joined UNU-MERIT (United Nations University – Maastricht Economic and
social Research and training centre on Innovation and Technology) as a researcher
working on the ObservatoryNano project.
top related