Professional Documents
Culture Documents
Industrial Agglomeration and Development A Survey of Spatial Economic Issues in East Asia and A Statistical Analysis of Chinese Regions
Industrial Agglomeration and Development A Survey of Spatial Economic Issues in East Asia and A Statistical Analysis of Chinese Regions
To cite this article: C. Cindy Fan & Allen J. Scott (2003) Industrial Agglomeration and
Development: A Survey of Spatial Economic Issues in East Asia and a Statistical Analysis of
Chinese Regions, Economic Geography, 79:3, 295-319
Article views: 91
Allen J. Scott
Department of Geography, University of California–Los Angeles,
1255 Bunche Hall, PO Box 951524, Los Angeles, CA 90095-1524
ajscott@ucla.edu
Abstract: In this article, we explore the issue of industrial agglomeration and its
relationship to economic development and growth in the less-developed countries
of East Asia. We present theoretical arguments and secondary empirical evidence
as to why we should have strong expectations about finding a positive relationship
between agglomeration and economic performance. We also review evidence from
the literature on the roles of formal and informal institutions in East Asian regional
economic systems. We then focus specifically on the case of China. We argue that
regional development in China has much in common with regional development
in other East Asian economies, although there are also important contrasts
because of China’s history of socialism and its recent trend toward economic liber-
alization. Through a variety of statistical investigations, we substantiate (in part)
the expected positive relationship between agglomeration and economic perfor-
mance in China. We show that many kinds of manufacturing sectors are character-
ized by a strong positive relationship between spatial agglomeration and produc-
tivity. This phenomenon is especially marked in sectors and regions where
liberalization has proceeded rapidly. We consider the relevance of our comments
about industrial clustering and economic performance for policy formulation in
China and the less-developed countries of East Asia.
Key words: industrial clusters, agglomeration, regional development, new economic
geography, East Asia, China.
This research was supported by the World Bank. We express our thanks to Wenfei Wang for her
able research assistance on this project; to Chase Langford for his cartographic work; and to George
Lin, Henry Yeung, and four anonymous referees for their comments on earlier drafts of the article.
295
#9905—ECONOMIC GEOGRAPHY—VOL. 79 NO. 3—79304-Fan_Scott
296 ECONOMIC GEOGRAPHY JULY 2003
investigations that seek to document and statistical analyses accordingly fall short of a
substantiate the expected positive relation- full-blown test of the theoretical ideas laid
ship between agglomeration and economic out in the first half of the article. Nevertheless,
performance. Our discussion is focused in so far as they go, the analyses are consis-
specifically on an examination of the tent with our main conceptual claims, and
spatial distribution of manufacturing activity the results provide an encouraging basis
in China. for moving forward into new rounds of
Most of the literature on industrial statistical research in the future.
agglomeration and economic performance
has been concerned with empirical situa-
tions in the advanced Western capitalist Locational Agglomeration and
world. In this article, we hope to demon- Regional Industrial
strate not only that the cluster approach is Performance
as useful in the East Asian context as it is
in Western situations, but that it can also The notion of the region as a nexus of crit-
shed important light on critical dilemmas of ical developmental and growth processes has
development that are specific to non- long been familiar to heterodox economists,
Western countries. We emphasize that such as Hirschman (1958), Jacobs (1969),
theory needs to address both the economic Kaldor (1970), Lampard (1955), Myrdal
bases of regional development and socio- (1959), and Perroux (1961), who took exter-
cultural factors, such as formal and informal nalities and increasing returns seriously—in
institutions. In China, in particular, a history part because they explicitly or implicitly
of central planning and the recent transition recognized the importance of the intellec-
further complicate the applicability of tual legacy of Marshall (1890, 1919)—and
economic theories that are based primarily who saw that one major expression of
on Western experiences. In this article, we these phenomena can be found in local-
show that in China, the positive relations ized complexes of economic activity. Since
between industrial agglomeration and the early 1980s, an enormous surge in the
economic performance are generally as literature in economic geography and
hypothesized, but that genuine industrial allied fields has greatly expanded on this
clusters are more characteristic of sectors earlier work. All this literature has repeat-
and spaces that have been most deeply trans- edly suggested that selected regions—espe-
formed by economic reforms and market cially those in which industries are organized
orientation. in transactions-intensive networks—are
In the first half of the article, we lay out capable of exerting powerful push effects on
a brief statement on the relations between national economic development.
industrial agglomeration and economic Many kinds of industrial systems are trans-
performance and consider the ways in which actions intensive both in a static sense (trans-
these relations throw light on the East Asian actions are numerous and finely grained)
case, and vice versa. In the second half, we and in a dynamic sense (transactions are
focus on regional development in China, first subject to constant flux because of rapidly
by giving a short account of how recent polit- changing production arrangements and
ical economic changes have affected the market instabilities). These features are espe-
space-economy and then by engaging in a cially marked today in industries like elec-
number of statistical exercises to demon- tronics, communications equipment, special-
strate that there is a significant link between ized machinery and components, toys,
agglomeration and economic performance watches, garments, furniture, software,
in Chinese regions. Unfortunately, empir- and business services, where many small
ical work of this sort commonly encounters establishments with narrowly defined core
severe constraints resulting from the defi- competencies perform critical functions
ciencies of available industrial data. Our within constantly shifting and information-
#9905—ECONOMIC GEOGRAPHY—VOL. 79 NO. 3—79304-Fan_Scott
VOL. 79 NO. 3 INDUSTRIAL AGGLOMERATION AND DEVELOPMENT 297
intensive systems of interlinkages. Mutual learning and innovation whose cumula-
proximity is often crucial to the success of tive effects greatly reinforce local
the individual establishments that make competitive advantages.
these industries up, partly because it reduces 4. The clustering of many different
the costs of transacting and partly because producers can significantly enhance the
of a series of other factors that intensify local- formation of beneficial business alliances
ized positive externalities. Benefits of and organizations that help to augment
these sorts are identifiable more generally local competitive advantage. Equally,
as agglomeration economies, although in agglomeration promotes the develop-
certain cases, agglomeration diseconomies ment of distinctive business cultures in
can also appear as regions grow in size. particular places, thus facilitating the
Agglomeration effects, in turn, are often tasks of interfirm communication and
categorized as so-called localization understanding.
economies (i.e., efficiency-boosting 5. Significant economies can be obtained
phenomena that come from the clustering when the consumption of infrastructural
of firms in a given sector) and as urbaniza- artifacts is spread out over many indi-
tion economies (i.e., efficiencies that result viduals in any one place. Large localized
from the agglomeration of many different clusters of firms and workers make it
kinds of activities in a given region). In the possible to construct disproportionately
vocabulary of more recent analysts, these dense and rich infrastructures with many
economies are often referred to, respectively, positive effects on local competitive
as Marshall-Arrow-Romer externalities advantage.
and Jacobs externalities (e.g., Beardsell
and Henderson 1999). Regions that have some or all these attrib-
In practice, these types of externalities can utes, combined with dynamic learning capac-
be further broken down into more detailed ities, stand some chance of becoming signif-
categories. The following five points summa- icant articulations of value-adding activity
rize the main issues: and entrepreneurial energy in the new global
economy. They are apt to exhibit strong
1. Interfirm transactions that are small in increasing-returns effects that lead to ever-
scale, variable in content, and subject superior competitive advantages (as
to frequent readjustment usually incur expressed, for example, by rising quality/cost
high costs per unit of distance. The ratios for final products) as they grow. In
mutual proximity of firms in networks practice, given statistical measurement diffi-
made up of transactions like these is an culties, these effects are typically analyzed
important factor in preventing costs from by economists in the form of aggregate
spiraling out of control and reducing the productivity levels per worker (operationally,
risks of any failure to establish prompt value-added per worker), and in the analysis
interfirm contact (Scott 1988). presented here, we pursue the same strategy.
2. Dense local labor markets represent Much of the econometric work in this regard
spatial concentrations of job seekers and has been focused on cases in the more devel-
job vacancies, and high levels of mutual oped countries (see, e.g., Shefer 1973;
proximity make it relatively easy to Kawashima 1975; Sveikauskas 1975; Carlino
acquire, process, and act on information 1979), and only in recent years has there
about relevant opportunities. been a concerted attempt to apply similar
3. Transactional relations also involve flows approaches to regions in less-developed
of certain kinds of business information countries (e.g., Henderson 1986, 1988; Y.
or knowledge spillovers. Untraded inter- Chen 1996; Shukla 1996; Lee and Zang
dependencies of these sorts are all the 1998). We build on this latter literature with
more important because they tend to a detailed multilevel analysis of regional
underpin many small-scale processes of development across China as it moves
#9905—ECONOMIC GEOGRAPHY—VOL. 79 NO. 3—79304-Fan_Scott
298 ECONOMIC GEOGRAPHY JULY 2003
from central planning to an economic order Fujita and Tabuchi 1997), our concern here
that is increasingly influenced by liberal- is with the less-developed parts of East Asia.
ization policies. As such, our effort can be seen as a contri-
At the same time, current theories in bution to knowledge about agglomeration
economic geography pay attention not only in the non-Western world in general, and in
to the economic foundations of industrial low- and middle-income countries in partic-
agglomerations, but also to the important ular (see also Nadvi and Schmitz 1994; and
role played by institutional factors in Scott 2002).
promoting localized growth and develop- In these countries, dense industrial clus-
ment. It is generally acknowledged that these ters derive from both local entrepreneurial
factors tend to be extremely place specific efforts and foreign direct investment. In
and often vary greatly from one location to numerous cases, clusters can be shown to
another (Amin and Thrift 1995; Storper and have their roots in traditional artisanal forms
Scott 1995). In most East Asian countries, of production, as in the gem and jewelry
formal and informal institutions have been industry of Bangkok, the furniture industry
critical in shaping regional economic of the Philippines, the pottery and silk indus-
outcomes. In addition, governments have tries in China, and the shoe industry of Agra
played a notably directive role in assigning in India. At the same time, and especially in
investment to different locations and in the case of East Asia, we need to remain
setting up export-processing zones and other alert to the political context and the actions
local development schemes. China presents of governments at various levels of authority
an especially interesting case in this regard in molding regional patterns of development.
because of its history of centralized economic In East Asia, moreover, as in the economi-
planning. An important research and cally developed parts of the world, large city-
policy question raised by any investigation regions are the locations of many of the most
of industrial districts in China is whether vibrant industrial districts. Singapore, Hong
agglomerations that begin as governmental Kong, Seoul, Beijing, Shanghai, and Kuala
projects are able to develop into genuine Lumpur, for example, harbor many special-
growth centers with a strong endogenous ized industrial districts that draw on dense
dynamic of growth. In our empirical analysis, local supplies of skilled labor, infrastructure,
we show that viable industrial regions have educational and research facilities, and so
made their historical and geographic appear- on. In South Korea, high-technology firms
ance in those parts of China where economic have been shown to prefer the Seoul metro-
liberalization has been most prominent, but politan region because of high levels of
that the record is much less positive in other access to technical labor (Park 1994). In
parts of the country. Informal and trust- addition, foreign investors tend to favor
based relationships, too, are important in metropolitan areas that offer a diversity of
China, all the more so since economic liber- infrastructural and input services, as well
alization has relied heavily on ethnic and as a predictable regulatory environment
kinship ties with Hong Kong, Taiwan, and (Scott 1987; Leung 1993, 1996; Wong and
overseas Chinese. Goldblum 2000). The largest of Asia’s city-
regions sometimes exhibit agglomeration
diseconomies because of congestion,
Geographic Agglomeration in pollution, crime, and so on, although this
the Less-Developed Countries problem is not absolute, in the sense that
of East Asia appropriate policy and planning interven-
tions can invariably clear away some of the
Although the phenomenon of industrial barriers to further growth. In any case, far
agglomeration has been well documented in from seeing large-scale urbanization as an
the advanced economies, including Japan aberration, as some development theorists
(e.g., Nakamura 1985; Patchell 1993; have done in the past (e.g., Lipton 1977),
Kanemoto, Ohkawara, and Susuki 1996; our argument is that it is one of the impor-
#9905—ECONOMIC GEOGRAPHY—VOL. 79 NO. 3—79304-Fan_Scott
VOL. 79 NO. 3 INDUSTRIAL AGGLOMERATION AND DEVELOPMENT 299
tant channels through which agglomera- thrive on the basis of its vibrant transactional
tion economies are achieved and acceler- networks, its low wages, and the skilful polit-
ated development occurs. Smaller urban ical maneuvering of its representatives.
centers, too, can play significant roles in the Rasiah’s (1994) study of the machine-tool
development process, especially in Asia industry in Malaysia underscored the
where many traditional craft industries connection between subcontracting and
outside the main metropolitan areas are localization. A number of studies on
beginning to show signs of modern entre- Indonesia have highlighted positive exter-
preneurial energy and orientation to wider nalities and productive effects of industrial
markets. We now examine more precisely clusters (Poot, Kuyvenhoven, and Jansen
how these relationships are worked out on 1990; Smyth 1992; Sandee, Rietveld,
the ground. Supratikno, and Yowono 1994; Nadvi and
Schmitz 1994; Sandee 1995; Weijland 1999;
Industrial Linkages, Subcontracting, Sandee and Rietveld 2001).
and Flexible Production
Learning and Knowledge
There is much evidence on the impor-
tance of industrial linkages as a factor in Many studies have pointed to the impor-
agglomeration processes in East Asia. For tance of the relation between learning and
the case of South Korea, Park (1991) argued proximity in the cities of developing coun-
that polarized regional economies with a tries in general, and East Asia in particular.
diversity of industries, information, and tech- At the same time, formal learning institu-
nical workers generate innovative firms and tions, such as universities and research infra-
spin-offs, which, in turn, encourage further structures, are located mostly in large urban
growth. Park (1993, 1994) further noted that areas in Asia. In Taiwan, the close spatial
vertical disintegration and the clustering of relations between the Industrial Technology
small plants increase the productivity of Research Institute and firms that produce
labor-intensive industries and that subcon- integrated circuits in the Hsinchu Science-
tracting is an important organizational Based Industrial Park are widely regarded
strategy for lowering wages and deflecting as a major factor in the latter’s success
industrial disputes. Similar arguments have (Chang and Hsu 1998). In a comprehensive
been made for the case of Taiwan, where article, Bell and Albu (1999) argued that
subcontracting networks and cluster-based “knowledge systems” are central to the long-
manufacturing were found to be important term dynamism of industrial clusters and
strategies for increasing the competitiveness stressed that the diffusion of innovation
of firms (Shieh 1990; W.-H. Chen 1999). within clusters, as well as the openness of
Focusing on Taiwan’s footwear industry, clusters to flows of knowledge from outside,
Levy (1991) contended that the transactions- are important in developing countries. Liu
costs hypothesis provides a powerful expla- (1998) concluded that learning capability
nation for the emergence of localized and human capital will determine the
subcontracting networks. durability of Taiwan’s industrial success.
Detailed empirical studies on industrial Concomitantly, the absence of these quali-
linkages in the more peripheral parts of East ties in any region means that barriers to
Asia are relatively scarce. In a study of the development are likely to be rapidly encoun-
semiconductor industry in Southeast Asia, tered, even if the region possesses other
Scott (1987) showed that production units kinds of assets. Sandee (1995) demonstrated
in Manila’s semiconductor complex in the how collaboration in a rural industrial cluster
mid-1980s were clustered close to one in Indonesia fosters the diffusion of inno-
another and were intricately linked to vation and the deepening of technological
minimize transactions costs. Scott (1994) capabilities. Johansson and Nilsson (1997)
demonstrated that the gem and jewelry showed that clustering in export processing
industry of Bangkok has been able to zones in Malaysia stimulates local firms to
#9905—ECONOMIC GEOGRAPHY—VOL. 79 NO. 3—79304-Fan_Scott
300 ECONOMIC GEOGRAPHY JULY 2003
learn from foreign investors how to produce, cated, many proponents of the new
market, sell, and distribute manufactured economic geography have sought to extend
goods on the world market. the theory of agglomeration by emphasizing
the institutional bases of regional economic
development, particularly the role of public
Cluster Development and the
agencies in regulating local market failures
International Division of Labor and other impediments to rapid forward
For the most part, East Asia has been at momentum (e.g., Peck 1992; Amin 1999).
the receiving end of the decentralization of This approach has been greatly influenced
manufacturing production from advanced by institutional and evolutionary economics
industrialized economies to developing and economic sociology, which stress that
countries. For example, until the mid-1970s, economic activity is always socially
the four newly industrializing economies of embedded and therefore context specific
South Korea, Taiwan, Hong Kong, and and path dependent (Granovetter 1985;
Singapore accounted for the bulk of the Hodgson 1988).
offshore assembly work of U.S. electronics
corporations (Scott 1987). As these Governance Structures and Regional
economies developed, branch plants in Development
search of cheap labor moved to yet more
peripheral parts of East Asia, including East Asian governments have had signif-
Malaysia, the Philippines, Indonesia, icant impacts on economic development
Thailand, and China (Tsay 1993; Ho 1994; through the adoption of export-oriented
Park 1994; Tang 1996; Eng 1997; Wong and industrialization policies and the establish-
Goldblum 2000). In the same way, consid- ment of industrial parks and export-
erable subcontracting now occurs from processing zones. Kaohsiung in Taiwan,
North America and Western Europe to East Masan in South Korea, Penang and Johore
Asian producers in such sectors as clothing, in Malaysia, and Shenzhen in China were
furniture, and machinery. It is important among the first generation of government-
to note that large urban centers in East Asia sponsored industrial zones for fostering
are especially attractive as locations for this export-oriented industrialization. Yuan and
kind of manufacturing decentralization. Such Eden (1992), in their comparative study of
development typically draws on the cheap, Taiwan, South Korea, and China, under-
surplus labor in cultural and political settings scored the role of governmental policy in
where workers are willing to tolerate long determining the performance of export-
hours, accelerated rhythms of work, and processing zones. Governments in Taiwan
round-the-clock manufacturing schedules. and Singapore have played particularly crit-
As manufacturing decentralization to East ical roles in fostering technological advance-
Asian cities occurs, moreover, local input ment in these zones (Yuen 1991; Ho 1994;
industries often spring up in the local envi- Xue 1997).
ronment, providing backup services and In many cases, however, government-initi-
strengthening local agglomeration ated development zones have not yet really
economies. taken off, in the sense that they generate
limited agglomeration economies. In
contrast to the Marshallian formulation, in
Institutions and Regulation in which endogenously driven growth and local
East Asian Regional embeddedness sustain agglomerations, these
Development clusters are more often than not “hub and
spoke” or “satellite platform” industrial
Formal and informal institutions play a districts (Markusen, Lee, and DiGiovanna
prominent role in the structuring of East 1999). For example, in South Korea, Pohang
Asian economies. Indeed, as we have indi- is anchored by a steel company, and Ulsan
#9905—ECONOMIC GEOGRAPHY—VOL. 79 NO. 3—79304-Fan_Scott
VOL. 79 NO. 3 INDUSTRIAL AGGLOMERATION AND DEVELOPMENT 301
is anchored by a petrochemical plant and an ality, trust, and cooperation. Ties of prox-
automobile plant, whereas Kumi, imity and association are especially impor-
Changwon, and Ansan are satellite districts tant in promoting learning effects and the
with branch plants whose external linkages diffusion of know-how (Amin and Thrift
are mostly extralocal (Markusen and Park 1995; Sunley 1996; Storper 1997). Face-to-
1993; Park 1996). Park (1993) observed that face contacts, in particular, help to engender
the South Korean government has increased relational assets or untraded interdepen-
support for collaboration between large firms dencies in the form of information, reputa-
and small or medium-sized firms so that tional capital, economically useful sensibil-
satellite clusters can benefit from the forma- ities and forms of habituation, and so on.
tion of local networks and linkages. Economic agents and firms in regions with
Furthermore, Park (1996) showed that as rich traditions of reciprocity benefit from
firms begin to lose the advantages that the sharing of know-how and expertise,
accrue from governmental incentives, they and by doing so, they increase the local stock
tend to pursue strategies such as the of competitive advantages (Cooke and
formation of local linkages, cooperative inter- Morgan 1998). Transactions, therefore, are
firm relations, and contracting-out activities, to be evaluated not only in terms of their
while spin-off activities and new start-ups costs, but also in terms of their qualitative
also tend to increase (e.g., in Taeduck social characteristics (Harrison 1992).
Science Town). Through this process, Park
contended, initial satellite and hub-and-
spoke industrial districts begin to evolve Geographic Agglomerations in
toward systems marked by local networks China
and embeddedness, as illustrated by the case
of Kumi. China has much in common with other
East Asian economies, from the many big
cities scattered across its landscape to the
Informal Bases of the Local Economy prominent roles of governments and social
Trust-based business relationships are of networks in economic life. However, its
great significance in many East Asian development history has also been heavily
cultures and are reflected not only in local shaped by socialist ideology and, more
transactions, but also in the intricate ties recently, by the trend toward economic
based on the Asian diaspora that exist liberalization.
between East Asia and other parts of the China has a mixed economy, one in which
world (Hamilton 1991; Smart 2000; Yeung the government’s role and institutional lega-
and Olds 2000). Since the 1990s, an cies from the former command economy
emerging body of work on East Asia has have had profound effects on the geography
sought to highlight the role of interpersonal of production. Prior to the 1980s, the
relations and ethnic and kinship ties in Chinese government’s blueprint for national
regional economic development. Thus, Olds economic development, with its stress on
(2001) argued for a relational geography that large-scale vertically integrated units of
focuses on networks and flows, linkages, production, was not especially conducive
interdependence, connections, and mutu- to the formation of dynamic industrial
ality. This kind of relational geography is districts rich in positive externalities. The
characteristic of “Chinese capitalism” and of socialist economy was focused on state-
the operations of large Chinese-controlled owned enterprises, domestic investment and
international conglomerates. markets, and a locational rationale that
The importance of interpersonal relations emphasized political goals more than
in East Asia further reinforces the notion economic efficiency. From the 1950s to
that economic behavior is embedded, in part, the early 1970s, the Chinese government
in social networks that are defined by mutu- deliberately discouraged investment in
#9905—ECONOMIC GEOGRAPHY—VOL. 79 NO. 3—79304-Fan_Scott
302 ECONOMIC GEOGRAPHY JULY 2003
coastal areas and promoted the growth of ment depended on parts brought from
inland cities. It also selected remote inland Taiwan. Since the mid-1990s, however,
sites for key sectors, such as cars and iron strong local networks underpinning the
and steel, but discouraged spatial clustering production of printed circuits have emerged,
(Huang 1999; Naughton 1988). Sit and Lu with interrelated factories constituting a
(2000) showed how the geography of China’s growing spatial agglomeration. As a result,
car industry is still strongly influenced by a complete computer can now be assembled
these past locational decisions. This influ- from parts made entirely in factories located
ence is reinforced by regulatory structures within 50 kilometers of one another. Park
that restrict foreign investments in the car (1996) also noted that small firms in satel-
industry to joint ventures with domestic lite industrial districts in China, such as
firms. Shenzhen, are becoming increasingly inte-
Since the economic reforms of the 1980s, grated into local network structures.
however, the Chinese government has In contrast to the socialist period, when
actively pursued alternative policies that are the government had strict control over
directed to economic liberalization. It economic activities, economic liberalization
promoted export-oriented industrialization since the 1980s has boosted the role of the
programs by designating special economic market and has been conducive to industrial
and open zones, most notably in the eastern linkages that respond to market signals.
coastal zone (Fan 1995, 1997). New policy Perhaps the most vivid example is the hosiery
incentives, infrastructural investments, and cluster in Zhejiang province, where more
social relationships that include ties with than 10,000 village households in 120 villages
overseas Chinese (Hsing 1996; Leung 1993) in and near Datang have become China’s
have paved the way for the eastern coastal biggest center of the production of socks and
zone to be integrated quickly with Western stockings and cluster development is now
capitalist production. Fan (2001, 2002) strongly under way (Wang 2001). Each
contended that the government has used enterprise specializes in a particular aspect
socialist-type policies to foster rural-urban of production, including manufacturing,
migration and, by doing so, has engendered machinery, marketing, and services, and this
a migrant labor regime and increased the networked organization is considered to be
supply of cheap rural labor to coastal areas. the primary reason for the success of the
In contrast to industries that are heavily cluster (Guo and Cai 2000). The intricate
influenced by Maoist legacies, the consumer division of labor and face-to-face networking
electronics and garments industries, which in the system enable enterprises to respond
tend to dominate in these areas, are char- quickly to changes in the market and to
acterized by high levels of foreign invest- ensure high levels of efficiency and mini-
ment, export orientation, and functional flex- mization of risks. This system is strongly
ibility. We argue that it is in these spaces, embedded in local social relationships in that
where economic reforms and market orien- interfirm linkages depend on trust and social
tation are the most deeply rooted, that capital. Wang (2001) argued that the close
agglomeration economies appear most force- and specific relations among enterprises, the
fully in China. low barriers to entry and exit, and external
Some government-sponsored industrial economies that are due to interlinkages
areas in China’s open economic zones create a situation that is similar to the Third
initially resembled satellite platforms of Italy model of flexible specialization.
the South Korean type, where many firms Similarly, flexible production systems in the
were tied to overseas markets but had only garment, textile, toy, and footwear/leather
weak local linkages. Tong and Wang’s (2002) industries in south China permit fast
study of the personal computer industry in turnover and quick response to market
Dongguan, Guangdong, for example, changes—key factors in firms’ competi-
showed that during the 1980s, printed-circuit tiveness in the global market (Christerson
assembly plants with foreign capital invest- and Lever-Tracy 1997). Also prominent in
#9905—ECONOMIC GEOGRAPHY—VOL. 79 NO. 3—79304-Fan_Scott
VOL. 79 NO. 3 INDUSTRIAL AGGLOMERATION AND DEVELOPMENT 303
south China are firms in the computer analysis of subcontracting activities in the
industry that have attained to high levels of Pearl River Delta that highlighted the role
competitiveness because agglomeration of cultural and kinship ties between distant
enables them to ensure that their products investors and local producers (see also Hsing
are high quality through rapid access to 1995, 1996, 1998). Christerson and Lever-
information and inputs (Tong and Wang Tracy (1997), who also focused on Fujian
2002). Even many small rural areas in the and the Pearl River Delta, argued that the
Pearl River Delta have grown considerably family and social relationships of overseas
because of subcontracting orders from over- Chinese investors have been of major impor-
seas. Here, as in some other parts of East tance in the growth of the garment industry
Asia, industrialization has blurred the distinc- in south China.
tion between the urban and the rural, giving Y. Chen (1996) made a major econometric
rise to the desakota phenomenon described contribution to the study of industrial
by McGee (1991) and Lin (2001). agglomeration in China. He estimated
Another case is the agglomeration of high- sectoral value added as a function of capital,
technology firms in Zhong’guancun, Beijing, labor, and a regional factor multiplier that
which benefits not only from a local pool consists of five components, including
of skilled labor, but also from the proximity agglomeration. The impact of agglomeration
of a large number of research establish- on productivity was found to be positive and
ments, such as Beijing University and high for the machinery industry but lower
Qinghua University (Wang and Wang 1998). for less technology-intensive industries such
Many firms in Zhong’guancun are spin-offs as food manufacturing. In the next section,
from state-owned enterprises and continue we propose a framework similar to Chen’s
to receive strong support from parent insti- and on this basis seek to push the study of
tutions or universities. Wang (2001) industrial agglomeration in China forward
observed that a club culture has emerged in one more notch. Even though our analysis
the area, which further stimulates interac- was greatly hampered by the lack of data,
tion among entrepreneurs, managers, we have nevertheless been able to bring a
engineers, and professionals and promotes variety of different perspectives and scales
a climate of entrepreneurship, innovation, of analysis to bear on the central problem.
and risk taking. This case suggests that locally
embedded cultures, norms, and social rela- Agglomeration and Industrial
tions are major factors in the dynamism of
agglomerations in China. Productivity in China
Similarly, the strong ethnic and kinship We begin our analysis by probing gener-
ties of China with Hong Kong, Taiwan, ally into the relations between industrial
and overseas Chinese in other countries have location and productivity in China. One way
heavily shaped spatial patterns of foreign of measuring the broad locational charac-
direct investment and have been a crucial teristics of an industry is to compute a
factor in the development of new indus- Herfindahl index (hereafter H-index) to
trial agglomerations in the eastern and assess the industry’s overall level of spatial
southern coastal areas. Labor-intensive agglomeration or dispersal. Thus, for Sector
industrial clusters in Fujian and the Pearl i, let pij be the proportion of total activity
River Delta in Guangdong, for example, are located in the jth region or province. The
strongly dependent on such ties. Much of term pij is defined more explicitly as xij / Xi,
the growth of these regions has been due where xij is the amount of activity in industry
to the relocation of manufacturing produc- i in province j and Xi is the total amount of
tion from Taiwan and Hong Kong and to activity in industry i in China as a whole (i.e.,
subcontract orders from manufacturers of Xi = Σjxij). The H-index for Sector i is then
Chinese origin in the United States (cf. computed as Hi = Σjpij2. When all the activity
Bonacich and Appelbaum 2000). Leung in Sector i is concentrated in one province,
(1993, 1996) stressed the same point in an the index is equal to one; when all activity
#9905—ECONOMIC GEOGRAPHY—VOL. 79 NO. 3—79304-Fan_Scott
304 ECONOMIC GEOGRAPHY JULY 2003
is evenly dispersed, the index converges obtained for all 31 provinces in China,
asymptotically to zero for a sufficiently large although for 10 sectors, data were available
number of provinces; intermediate locational for only 30 provinces, and for 2 sectors, they
patterns are represented by values of the were available for 29 provinces. The quality
index between these two extremes. We of these data leaves much to be desired, but
computed two sets of H-indexes for all two- the results we obtained appear, with certain
digit manufacturing sectors in China (except reservations, to be reasonably reassuring.
armaments, for which the published statis- Table 1 shows H-indexes for this exercise,
tics are heavily censored), using the number ordered from high to low values. The index
of establishments and employment as raw for establishments is usually smaller than the
data inputs. Data for this purpose were taken index for employment, a statistical effect that
from the statistical yearbooks of the Chinese seems to arise because of the occurrence
provinces for 2000. In most instances, the of large establishments in many sectors (and
number of establishments and employment all the more so given the labor-hoarding
figures for two-digit sectors could be propensities of state-owned enterprises in
Table 1
H-Index Values for Two-Digit Industrial Sectors in Chinese Provincesa
Values of H-Index
Table 2
Regression Analysis of H-Indexes for Two-Digit Industries in Chinese Provinces
Values of Regression Statistics
H (Establishments) H (Employment)
Table 3
Regression Analysis of the Relations between Productivity and Locational Structure for
Two-Digit Industries in Chinese Provinces
Values of Regression Statistics
Regression 1 Regression 2
Industrial Clusters in Chinese firms identified as being below the township level,
Counties such as those established by villages. In addition,
firms registered at or above the township level
The data set we explored in this phase of include the bulk, if not all, of state-owned and
the investigation consisted of information share-holding enterprises and joint ventures
for 510,381 firms that are defined as inde- between Chinese and overseas investors.
pendent accounting units registered at or Collective-owned enterprises and private- and
above the township level of authority,1 as individual-owned enterprises are less highly
represented in the data set. As such, the data
set is probably biased toward larger firms. In total,
1
The definitional criteria are specific to the the 510,381 firms in the data set accounted for
Chinese industrial statistics system. Firms that 86.2 percent of all firms at or above the township
are independent accounting units are not level.
financed by or ancillary to any other institu- 2
County-level units include counties, county-
tions. Firms are also identified in terms of the level cities, and urban districts of prefecture-level
administrative rank of the places in which they and provincial-level cities. In this article, we refer
are registered. The data set used here excludes to all county-level units as counties.
#9905—ECONOMIC GEOGRAPHY—VOL. 79 NO. 3—79304-Fan_Scott
310 ECONOMIC GEOGRAPHY JULY 2003
Figure 1. Cluster analysis of establishments in the consumer electronics and garments industries
and the computers, electronic equipment and instruments industries, China, 1995. Figure based on
four-digit data by county from the 1995 Industrial Census provided by the State Statistical Bureau,
People’s Republic of China.
subgroups. The first subgroup is comprised nation, but there is strong prima facie
of consumer electronics and garments evidence, based on our general knowledge
(including cotton knitting) industries, with of how these industries work, that at least
the former breaking further down into some of them are both transactionally inter-
sectors like radios, television sets, cameras, connected and strongly associated with
toys, and clocks; these are among the most one another in geographic terms. In Figures
prominent export-oriented industries in 2 and 3 we map out values of the number of
China. The second subgroup consists of establishments and location quotients for
computers, electronic equipment and instru- establishments in the two subgroups. The
ments industries, including semiconductors agglomerative tendencies of these industries
and medical equipment. Unfortunately, our are clearly evident.
data do not allow us to identify functional The consumer electronics and garments
linkages between the sectors under exami- group is concentrated in the Pearl River
#9905—ECONOMIC GEOGRAPHY—VOL. 79 NO. 3—79304-Fan_Scott—Pg. 311
Number of
Establishments Location Quotient
1001-1500 ≥ 4.50
501—1000 3.00—4.49
251—500 1.50—2.99
1—250
Beijing
Tianjin
JIANGSU
Shanghai
ZHEJIANG
FUJIAN
GUANGDONG
Shenzhen
ta
l
De
v er
P e a rl Ri
0 600 km
0 600 mi
Number of
Establishments Location Quotient
201-500 ≥ 4.50
101—200 3.00—4.49
51—100 1.50—2.99
1—50
Shenyang
Beijing
Tianjin
Jinan
Xian
Hongyuan
Shanghai
Wuhan
Chongqing
Shenzhen
0 600 km
0 600 mi
Number of
Establishments Location Quotient
101-200 ≥ 4.50
Changchun
51—100 3.00—4.49
21—50 1.50—2.99
1—20
Beijing
Tianjin
Shiyan
Wuhan
Chongqing
0 600 km
0 600 mi