Hs i nc hu ( Ta i wa n) a nd Be i j i ng’ s Hi gh t e c h pa r ks c ompa r e d
Jenn hwan Wang Professor
Graduate Institute of Development Studies National Chengchi University, Taiwan
Paper to be presented at the 21stSociety for the Advancement of Socio-Economics (SASE),
Sciences Po
Paris, France, July 16-18, 20091, Introduction
This purpose of this paper is to compare the innovation system in Hsinchu ScienceIndustrialPark (HSIP)in Taiwan and Beijing’shigh-tech industrial park, or Zhongguancun (ZGC), and their respective competitiveness. HSIP was established by Taiwan’sstatein 1980 with theintention to upgradetheeconomy.Afterdecades’ development, HSIP currently has become a successful high-tech information
technology (IT) cluster that attracted many foreign and domestic firms to locate their manufacturing orR&D functionsin it.Beijng’sZGC isarguably the most innovative region in China, or called the Silicon Valley of China (Segal, 2003; Yu, 2008). It was originated from the initial stage of the Chinese economic reform and was formerly established as a high-tech science park in 1988 by the Chinese state. Currently, ZGC has agglomerated most of the innovative Chinese IT firms and many MNCs to locate their R&D centers there. Together with the high concentration of R&D personnel and institutes in the Beijing area, ZGC has become the most important center for
technology learning and innovation.
These two clusters however face enormous challenges ahead. For HSIP, the pressing issue is the threat of ZGC as a latecomer that has the advantages of low cost and massive domestic market. Especially when the global innovation networks (GINs) has emerged since the 1990s (Ernst, 2005a) that has witnessed the increasing
outsourcing from large MNCs of their R&D functions to China and India. This therefore creates the pressure to HSIP as to whether it still has the innovative capability to sustain its competitiveness. For ZGC, the emerging issue is to catch up the technologies as fast as it can to become the innovation center that has its own technologies,orcalled ‘self-reliantinnovation’(
z-zhu-chuang-xin). Therefore, there is
a common issue for both industrial clusters: that is, how to continue to upgrade technologically in orderto sustain theregion’scompetitiveness.This above pressing issue for both HSIP and ZGC is the question that is often asked in the study of regional development: the sustainability of an economic region.
Recently, studies of industrial clusters have paid much attention to the issues of
knowledge learning (Bathelt et al, 2004; Malmberg and Maskell, 2002; Maskell, 2005;
Morisini, 2002), and the global-local relational networks (Amin, 2002) that may enhance regional competitiveness and sustainability. Along the above theoretical perspectives, many existing studies have shown that HSIP is an industrial cluster that has built-in learning mechanisms which sustain the innovative capability of the firms in the cluster (Saxenian and Hsu, 1999; Hu, et al., 2005; Chen, 2003). However,
regarding learning mechanisms in ZGC, the picture is not as clear as those of HSIP.
While some argue that ZGC has become an innovative region that has learned extensively from MNCs (Zhou, 2005; Zhou and Tong, 2003), others maintain that ZGC has rarely established industrial networks that can generate knowledge learning and diffusion. What factors make HSIP different from ZGC? Why are they different in enhancing technology and knowledge learning?
Based on the geographical version of the global production networks approach, or the GPNs, (Henderson, et. al., 2002; Coe, et al., 2004), this paper will argue that the major difference that makes HSIP different from ZGC is due to the fact that HSIP has been well inserted into the global production networks of the ICT industries, with strong supports from local institutions in upgrading the ladder in GPNs; whereas ZGC’smurky picturein technological learning mainly comes from its obvious linkages within GPNs that support it to learn codified knowledge however the local institutions lack of systemic supports for technological upgrading.
2, From catching up to innovation cluster- a GPN approach
A clustercan bevery loosely defined as‘sectoraland spatialconcentrationsof firms’(Schmitz and Nadvi,1999:1503).Recentstudieson industrialclustersin developing countries, though stressing the importance of cost advantage, highlight that learning is essential for a cluster to become competitive due to its crucial role in disseminating knowledge and generating possible innovation (Humphrey and Schmidt, 2002; Giuliani et al, 2005). There are two essential elements that have been stressed in the literature on a cluster to become innovative (Bathelt et al, 2004; Lundvall, 1990;
Camagni, 1991; Stoper, 1995; Malmberg and Maskell, 2002; Maskell, 2005; Morisini, 2002). The first is a thick local institutional infrastructure that supports collective learning. This involves the intertwining of R&D institutes, networking among firms and a shared cultural tradition that facilitate information flow and knowledge
diffusion. The second consists of the extra-local linkages that enable information and knowledge to flow from outside so that local firms not only can learn technologies form local settings but also from outside actors, which can avoid lock-in effect and facilitate knowledge creation.
While most of the researches on innovative clusters are focused on localities of advanced countries, researches on the sustainability of industrial clusters in
developing countries are increasingly brought to the concern from development studies (Humphrey and Schmitz, 1996; Schmitz and Nadvi, 1999; Humphrey and Schmidt, 2002; Giuliani, et al, 2005 ). It is because most of the industrial clusters in developing countries are state creation that uphold their competitiveness in attracting
both local and foreign firms based on low cost. These clusters are in severe
competition and in consequence they are vulnerable to similar competition from other low labor cost countries. An urgent issue of sustainability of the industrial clusters in developing countries is thus on industrial upgrading –meaning ‘to makebetter products, make them more efficiently, or move into more skilled activities (Humphrey and Schmidt, 2002:1017) - which may generate innovative capability of local firms and clusters’competitiveness.
Nevertheless, industrial upgrading for developing countries is not an easy task.
Since the latecomer firms in definition do not have the advanced knowledge in producing better products, they have to learn from firms in advanced countries.
Latecomer firms learn, emulate and assimilate technologies from advanced countries through the channels such as reverse engineering, outsourcing, licensing, joint
ventures, etc. Therefore, the states in developing countries tend to adopt policy on the one hand to build good environment in attracting foreign firms to invest in their industrial clusters, and on the other hand to construct local institutions that are favorable for technology learning. In the process the latecomer firms accumulate technologies and capability in due course, if they have made sufficient effort, and gradually have build up their own innovative ability (Hobday, 1995; Amsden and Tschang, 2003; Amsden and Chu, 2003; Kim, 1997).
It is on this perspective that the geographic perspective on the GPNs approach (Henderson, et. al., 2002; Coe, et. al., 2004) is useful in accounting for the catching up toward innovation issue in clusters in developing countries. In order to better account for the rapidly changing nature of globalization and its impact on production and consumption, the geographic perspective on GPNs propagates that territorial elements have to be added to the lineal concepts of the existing global production networks (Ernst, 2005a,b; Ernst and Kim, 2002) and global commodity chain perspectives (Gereffi, 1994; 1999). Therefore, the geographic
version of GPNs is defined as the
‘ globally organized nexus of interconnected functions and operations by firms and non-firm institutions through which goods and services are produced and distributed’
(Coe, et. al., 2004:471). This GPNs perspective asserts that most of the industrial
goods at the current stage can be globally produced and reintegrated, through which
the production process has cut across national borders via production networks. Firms
in the networks on the one hand are articulated in the production chains and on the
other hand are
deeply influenced by the concrete socio-political contexts within which they are embedded, produced and reproduced. The GPNs approach maintains that this process is especially complex because while the local embeddedness are essentially territorially specific, the production networks themselves are not (Henderson, et al.,2002: 446). The intertwining of the territorial and production networks in GPNs generate coupling and decoupling processes in which regional development is or is not achieved.
As Coe, et. al., (2004) maintain, regional development
depends in a large degree on t he a bi l i t y of t he ‘ s t r a t e gi c c oupl i ng’ be t wee n gl oba l pr oduc t i on ne t wor ks a nd regional assets. It is because the local institutional arrangements that fit the need of GPNs whi l e s i mul t a ne ous l y pr omot e l oca l f i r ms ’ a r t i c ul a t i on i n t he gl oba l pr oduc t i on ne t wor ks t ha t c a n uphol d t he r egi on’ s e c onomi c de ve l opme nt . They a r gue , ‘ r egi ona l assets can become an advantage for regional development only if they fit the strategic ne e ds of gl oba l pr oduc t i on ne t wor ks . The pr oc e s s of ‘ f i t t i ng’ r egi ona l a s s e t s wi t h strategic needs of global production networks requires the presence of appropriate institutional structures that simultaneously promote regional advantages and enhance t he r egi on’ s a r t i c ul a t i on i nt o gl oba l pr oduc t i on ne t wor ks ’ ( Coe , e t a l . , 2004: 474) .
Nevertheless, by inserting into the GPNs does not mean that a local cluster can continue to upgrade easily. The continuing upgrading of a cluster needs global-local linkage and local institutional supports. The road is not an easy one, because the power relation in GPNs is asymmetrically distributed. The more a firm in a region is articulated into global production networks, the more likely it is able to reap the benefits of economies of scale and scope in these networks, the less however it is able to control its own fate (p. 475). In the same vein, if a region attempts to upgrade the technology through which many institutional arrangements have been changed, this may take
the risk of confronting with the interest of the leading firms. The leading firmsin theprocessmay leavethelocation to find othermore‘fitted’locationsfor production.The upgrading of industrial clusters in the developing countries may take advantage of the opening of GPNs in continuing outsourcing its technologies to developing countries (Ernst, 2005a,b), or to make great purposeful efforts to break out the lock-in effect and make industrial upgrading effective (Humphrey, 2002). To the former, since the information revolution, global brain circulation and the new tendency of global vertical disintegration, these may provide new opportunities for latecomer to easily to access to international knowledge sources. To the latter, as Humphrey and Schmidt(2002:1025)argue,‘thegreatertheleap in upgrading,theless likely it is that knowledge acquired in existing linkages suffices. Firms will have to rely to agreaterextenton localand nationalsourcesofinnovation.’Thatis,firms need greater institutional support to engage in technological upgrading as to engage into the GPNs for a higher position.
The building of local institutional structure however is not out of vacuum, it has its institutional roots that have inherited from historical legacy. It is because of the vested interests in the system has been established, people tend to choose paths that are more familiar with to adjust to the new environment. As institutionists argue, since institutions both enable and constrain individual behaviour by defining the incentive framework in which agents make decisions. Most of the time, this inhibits people from choosing more efficient arrangements than the current ones. Institutions therefore are generally not optimal (North, 1990), and continueto affectpeople’s behaviour. North argues that this act mainly results from the fixed cost of the institutional architecture and the increasing returns to adoption. As a result, this creates the path dependence or the lock-in effect, which explains the phenomenon in which existing institutions tend to persist and become the growth trajectories that are followed by different countries.
Based on the above institutionist perspective, we propose that the existing industrial system or institutional arrangements of a late industrializing region will continue to play a major role in its strategy in pursuing for innovation. The existing socially-embedded institutionalarrangementswhich existin aregion’sindustrial system may lead to the creation of both enabling and constraining conditions for the local economy to adjust and to transition toward innovation. Whether a latecomer industrial cluster can transform into an innovative cluster and change its status in the GPNs, as argued above is mediated by its existing institutional arrangements, which is a question that can only be answered empirically.
3, The Hsinchu Science Industrial Park and its semiconductor industry
The successful story of HSIP and its learning capability has been documented by many studies (Saxenian and Hsu, 1999; Hu, et al., 2007; Mathews and Cho, 2000).
The major findings of the above studies includes: the closer relationship between Silicon Valley and HSIP which intensifies the global-local linkages; the closer
network relationship between R&D institutes and local firms, including the Industrial Technology Research Institute (1974) and the adjacent universities, such as Tsinghua and Jiaotong, that provide valuable contribution to technological learning; the closer networking relationship among firms, especially the upstream and downstream firms in the production chains in PC-related and Semiconductor industries; and most of all, the informal network relationships exist in HSIP that help to disseminate information and knowledge (including university alumni, peer in MNCs in the US, and friends in ITRI before working in HSIP). All these elements contribute HSIP as a learning region that resists the possibility of declining when many of the IT firms moved to
design sector in particular and its relationship with the semiconductor industry in generalto illustratehow HSIP’sdevelopmentisrelated to theevolution ofGPNsin the semiconductor industry.
The significance of semiconductor industry in HSIP can be shown in the following description. Although many of Taiwanese IT firms already moved their production facilities to China (the last production line of notebook PC move to China in 2005), the related semiconductor industry in HSIP is still growing over the years.
The importance of semiconductor industry in HSIP can be shown in Table 1 in which the sale ratio of Integrated Circuits (IC or semiconductor) has been in increasing from 50% to 71.6% in 2007. Whereas the computer and peripherals has been in decreasing from 35% to merely 8.3% in the same period. Clearly, the semiconductor industry has become a predominant sector in HSIP.
Table 1. Sales ratio of HSIP by industry, 2007 (source: MIC, 2008) Yea
3.1. The value chain of the semiconductor industry
The development and manufacturing of semiconductors involve three primary activities in the value chain: design, fabrication, and test and assembly. Overall, the worker skill requirement goes all along the value chain from design to assembly (Brown & Linden, 2005: 2). Before the emergence of the pure play foundry, such as Taiwan Semiconductor Manufacturing Company (TSMC) which fabricates ICs for fabless design houses, the standardized feature of the semiconductor companies was to keep all activities in-house. This is also referred to as the IDM (Integrated Device
Manufacture) model and firm, and includes such notable examples as Intel, Motorola and Texas Instruments. However, since the cost of building new fabrication facilities started to rise exponentially in the 1960s, the vertically-integrated organizational model began to disintegrate. The least skillful part, namely, test and assembly, began to break out of the chain. In due course, the test and assembly sector moved to developing countries, especially East Asia, in order to lower production costs. It was under these circumstances that the East Asian countries began to build their
semiconductor industries.
The 1980s saw the emergence of hundreds of fabless semiconductor firms, which designed and marketed semiconductor components and relied on contract IDM manufacturing for their designs. In addition, due to the rising cost of manufacturing facilities, IDMs were also seeking approaches that could control costs and risks. This resulted in the first pure-play foundry (TSMC) emerging in Taiwan in 1987 that specialized in semiconductor manufacturing. The launching of the foundry sector led to the establishment of more fabless design houses and foundries. At this point, the design houses were able to concentrate their activities on innovative design and leave the fabrication aspects to other firms. The value chain of the semiconductor industry thus further disintegrated into segments that could be located in different localities.
Currently, most of the foundries are located in East Asia in which Taiwanese firms have taken over 70% of the world market share (Table 2).
Table 2. World top four foundries in 2006 Firm Countries of
TSMC Taiwan 10.1 50%
UMC Taiwan 3.8 19%
Charter Singapore 1.6 8%
SMIC China 1.5 7%