Top-Down versus Bottom-Up Agglomeration effects among Chinese domestic firms Author: Johanna Gómez Ramírez Candidates of Master of Science in International Economics with Focus on China Department of Economics, Lund University Supervisor: Professor Sonja Opper Submission date: 15th October 2011 Abstract Industrial Parks in China have played a crucial role in attracting foreign investment and in enhancing regional development. However, there is an open question: do industrial parks develop comparable effects as natural grown clusters? Following agglomeration theory that highlighted the economic advantages from proximity between firms, and making use of a rich dataset from the Chinese domestic firms, this paper estimates the relationship between agglomeration effects and the performance of firms. The results show that firms located within industrial parks are more innovative and present better performance; however there is not evidence that linkages between firms have the same positive relation what contradicts the theory. Keywords: Industrial parks, natural grown clusters, agglomeration effects. Table of Contents Chapter 1 Introduction ............................................................................................................ 1 1.1 Research Question ............................................................................................................ 1 1.2 Method .............................................................................................................................. 3 1.3 Limitations ........................................................................................................................ 3 1.4 Structure ........................................................................................................................... 4 Chapter 2 Agglomeration Theory and Hypothesis ................................................................. 4 2.1 Defining Agglomeration Effects ...................................................................................... 4 Chapter 3 Industrial Parks Vs Natural Clusters .................................................................... 12 3.1 Overview of Industrial Parks and Cluster Development ................................................ 12 3.2 Features of Industrial Parks in China ............................................................................. 14 3.2.1 Advantages offered ...................................................................................................... 15 3.2.2 Development constrains .............................................................................................. 17 3.3 Features of natural clusters in China .............................................................................. 19 3.3.1 Gains from clustering .................................................................................................. 20 3.4 Top down versus Bottom-up: The China case................................................................ 22 Chapter 4 Empirical Analysis...............................................................................................23 4.1 Research Design: Modelling Agglomeration Effects.....................................................23 4.2 Data.................................................................................................................................25 4.3 Variables.........................................................................................................................25 4.4 Methodology and Results................................................................................................29 4.4.1Results………………………………………………………………………………...29 5. Conclusions…………………………………………..……………………….................33 References………………………………………………………………………………….35 Appendix 1 Pairwise Correlations ………………………………………………………..39 Chapter 1 Introduction 1.1 Research Question Industrial zones are playing a crucial role in the Chinese economy and research related to this topic is still limited. This study is an effort to understand the effectiveness of industrial zone policy by determining whether firms located within industrial parks enjoy the advantages of agglomeration effects that are found in naturally grown clusters. In China’s regional development, local governments rely on the creation of industrial agglomeration advantages by promoting the construction of industrial parks in order to generate growth and attract investment. However, it is important to analyze whether successful industrial zone experiences can be duplicated; and if these initiatives actually create positive effects on firm's performance, in the same way that natural clustering develops agglomeration advantages over time. Around the world, governments are creating industrial areas with the idea to offer services and special conditions for firms, which aims to transfer and generate growth in lagging regions or potential ones. In recent years, the creation of industrial areas worldwide has increased dramatically; in the USA there are currently around 8800 industrial parks, 1200 in Canada, 200 in the UK and 300 in Germany (Geng, 2009). However, it is important to recognize that some of the most successful industrial parks in developed countries are the result of interactions between pre-existing institutions and firms. Over time, these interactions create clusters that develop agglomeration effects, which at some point are named industrial parks and become priority for regional economic growth. In other words, IP´s in developed countries are the result of an evolutionary process. Silicon Valley in USA is a recognize example; in 1950’s start-up companies found advantages from being located around Stanford University, and dynamics that involve the region, firms and institutions, evolved to make of this area an example of exceptional development.. The success of Silicon Valley has inspired governments around the world to rely on government sponsored industrial parks to try and imitate similar agglomeration advantages and spur economic development. However, it is an open question whether the top-down approaches used to design parks have comparable effects as naturally grown firm clusters. 1 In China, industrial zones are identified as the cornerstone of industrial development (Geng, 2009). The interest from local governments in attracting capital using industrial parks that offer special benefits for firms is considered crucial for regional economic development (Chen, 2005). However, there are also successful cases of clustering in China where local firms and the dynamics from competition have created the conditions for industry growth. In clusters unlike industrial parks, local governments have played a supporting role by providing public goods and promoting clustering. In those cases, first initiatives were caused by local dynamics such as entrepreneurship, history, institutions and location. In this sense, the purpose of this analysis is to look at the effectiveness of the establishment of industrial parks as a government policy that incentivizes firm's growth and innovation; the reasoning behind this is to determine if the government should create and support new industrial areas, rather than simply reinforce and build on established and emerging clusters. The theory of agglomeration covers both, industrial zones and clusters development, highlighting the benefits of localization, economies of scale and spillover effects. The importance of industrial agglomeration, from a broad perspective, is to reduce costs and increase returns not only for the firm but also for the geographical space where the industrial park or the cluster is situated (Krugman, 1991). More specifically, this research will analyze if being located in an industrial park is significant for Chinese domestic firms; and if the actual impact of agglomeration economies reflected on domestic firm´s growth and innovation is due to the localization within clusters and industrial parks. My hypothesis follows the theory of agglomeration; I argue however, that despite the great contribution of industrial zones in attracting foreign capital and upgrading industrial techniques in China (Chen, 2005), these areas are still in the construction phase, due to the fact that the local government’s objective is to attract capital at any cost, which has created intense competition and has lead to an over development of industrial zones. The research is, therefore, an analysis of the industrial park policy in its purpose of creating agglomeration advantages for firms. 2 1.2 Method This study was conducted in two phases. In the initial phase the goal was to build up an understanding of industrial parks in China; this phase involved two face to face interviews with park project developers. These interviews were semi-structured and conversational using open-ended questions that allowed for the development of areas of interest. Despite the limitation in the number of interviews, the sites are very different from each other, which gives a better picture of their current situation and development. With the information from these interviews and a literature review about agglomeration theory, a case study was built using quantitative data from the 2003 World Bank Investment Climate Survey. This survey provides information on 2400 domestic Chinese firms, from which 600 are located within industrial parks. The questionnaire covers different aspects of firm’s activities and its business environment, which made possible the elaboration of measures of agglomeration with the aim of understanding the effect of proximity in firm’s performance. 1.3 Limitations The main limitations identified during the investigation were primarily related to the database. First of all, the survey was done between 1999 and 2003, which may show a different picture from what is the actual situation within domestic firms in China today, so the interpretation of the results should be done carefully. Furthermore, not all regions were considered in the survey and specifically inland regions show a small participation overall (Dollar, 2004). However, as industrial parks are mainly located within medium and big cities, the survey is a good approach to analyze IPs effectiveness. Additionally, there is a high proportion of state owned enterprises compared with other legal ownership forms which could also bias the results. The survey also offers different questions that try to approach the same subject; as a result, the variables used as a proxy for agglomeration can be determined from different perspectives. In this study, I used the ones considered relevant based on a subjective exercise that took into account the theory, but I recognize that the research outcome can 3 vary if instead other questions are used as a proxy. Finally, I should emphasis that the survey offers a good picture of both, firms located within industrial parks and those outside, which fits perfectly with the aim of the study. 1.4 Structure The structure of the paper is as follows. In the first section I placed the features of industrial zones within agglomeration theory, where the main contributions and recent approaches are explained. In this part, I present the research hypothesis which will be explored in the rest of the paper. The second part focuses on industrial parks and cluster development in China; first, by explaining the advantages that are offered to firms and then by identifying its main development constrains. The next section describes the empirical results from the multivariate regression and explains the relation found between agglomeration effects and firm´s performance and innovation. In this section, the variables used to measure clustering are disclosed. The final section considers possible implications of the empirical evidence on policy initiatives, making a critical analysis of the government’s role in the construction of IP´s and the promotion of clusters in China. Chapter 2 Agglomeration Theory and Hypothesis 2.1 Defining Agglomeration Effects There are two central theoretical fields that explain firm localization and its relation with productivity and development; one is agglomeration theory and the other one endowmentdriven localization. The latter states that firms will be located in regions with favourable factor endowments (Deavtz, 2008), for example, an area with abundant input supply or natural resources. However, as firms concentrate and compete for the use of endowments, prices increase, which reduces the benefits from localization. For this reason, in order to justify long term benefits for firms, industry localization has to be determined not only by endowment factors but by other aspects highlighted within agglomeration theory. 4 Agglomeration is considered in economic geography literature as high spatial density of economic activity. The notion of agglomeration could be used to explore different situations like the development of cities, industrial clusters, commercial districts and the imbalance between regions (Masahisa, 2002); in general, this theory is used to analyze economic concentration, due to proximity between economic agents. In this sense, agglomeration can occur at different geographic scales and examples can be taken from almost every location; from the dense shopping streets in Central London, the urban corridor from Liverpool to Milan, or the concentration of economic activity in the city of Shanghai (Lindquist, 2009, p:13). Agglomeration theory analyzes the effect of spatial distance on firm’s performance; distance refers to the space between firms and between firms and consumers. Research in this field has been focused on understanding ´what factors determine and sustain the advantages of proximity within firms´. Theorists in this field have identified economic benefits and general advantages due to the reduced distance between firms, however empirical studies supporting this theory present conflicting results. One of the advantages from agglomeration is external economies of scale that are considered the essence of this theory. Marshall (1920) identified three different external economies within what he called industrial districts. These industrial districts were areas of high spatial concentration of firms within the same sector. External economies are an explanation for benefits that firms find outside their own production process, for example: • Access to a market of workers with specialized skills, • Facility to develop specialized inputs and services, which would otherwise be impossible to obtain if a single firm was the sole producer. • Benefits from technological spillovers, available through the transfer of skills and the sharing of ideas between colleagues and competitors, which leads to the creation of inventions. These three aspects show the benefits from localization arising from firms sharing production factors. In other words, proximity reduces costs because it avoids the difficulties 5 of finding a sufficient market for inputs and labor. The main idea behind economies of scale is the shared access to goods and services (Deutz, 2008). In addition to the externalities found by Marshall, it is considered that agglomeration effects are enhanced by the dynamics of competition. Porter (2007) supports this idea by arguing that within clusters, transactions are more efficient, there is more flexibility in operations, it is easier to start a new business, and it is faster to implement and perceive innovations. Rather than being concerned with a single industry, agglomeration of firms is based on the link between different industries and entities that are considered important to competition. The causes of agglomeration of certain industries within a determined area are also a subject of study. Within these causes, transport cost has been identified as the most important. Krugman (1991) concludes that the nature of the macroeconomic adjustment processes within regions and the character of external economies are due mainly to the reduction in transport costs. He argues that manufacturing firms have an incentive to concentrate near markets and suppliers, which produces “a centripetal tendency to agglomeration”. For Krugman, the combination of scale economies and transport costs are positive externalities that encourage agglomeration. Models of monopolistic competition support the idea that as transport costs decrease there is an increased tendency towards agglomeration. Reducing transport costs increases competition; firms tend to differentiate their products and locate close to a big pool of consumers. It has been also determined that firms locate close to each other to exploit internal economies of production. Multinational companies, for example, would rather produce in one site, and ship their production to other countries, gaining from internal economies of scale (Masahisa, 2002). There are different arguments behind the causes of agglomeration and its relation with firm´s performance, but in general the core theory identifies a positive relation between both. The evolution of this theory has included new analyses that also look at aspects within the business environment at industrial parks and clusters: co-operation between firms, the importance of personal contact, network connections to support innovation within firms and business relationships with institutions, are some of the features studied by the agglomeration theorists in resent years. To understand the contributions of this new 6 approach, Deutz and Gibbs (2008) identify four common features present in agglomeration theory research: 1. External economies of scale. These are benefits outside a firm’s operation, which reduce the average unit cost and enhance productivity. A common example is industry growth. In the presence of proximity effects, firms share access to factors of production and supporting institutions, which is crucial to economic performance; for instance, aspects like: experienced labor, financial and legal services, and public services among others, are usually better established in an industrial location. Firms benefit from sharing services as it gives them access to factors that they cannot produce internally or for what they may not constitute a sufficient market to attract suppliers. 2. Networking. Agglomeration brings benefits to the firm by facilitating face-to-face contact with managers from different organizations. The possibility of establishing close business relations reduces transaction costs, increase the willingness to do business with other firms, and also creates the opportunity for alliances and organizations that enhance competitiveness are more likely to arise. Proximity increases the likelihood of personal contact which allows for exchanging tacit knowledge or knowledge that cannot be written down, for example information from past business experiences. Economic relations are embedded in social networks that depend on the characteristics of the society. In the Chinese case, and in East Asia in general, interpersonal relations are highly important. Networks are based on mutuality, trust and cooperation. In this sense, face to face contact enhances the development of assets like reputational capital, sharing of information and reciprocity; aspects that determine transaction cost and enhance competiveness (Fan, 2003). 3. Policy. One of the main interests for policy makers is to generate economic growth. Two aspects have led policy initiatives to promote proximity between firms. First of all, most successful regional experiences around the world have been based on agglomeration advantages (a well known example is Silicon Valley, USA); additionally, academic literature has recognized the importance of agglomeration formation as a policy initiative that can promote investment in lagging regions or 7 enhance regional capabilities. Policy makers have taken the role of providing opportunities and supporting external economies and network connections between firms; for example when building industrial parks or promoting clustering, the aim is to enhance contact between firms, infrastructure and business services. In the last 30 years, the economic transformation of China from a socialist system towards a social economy market has pushed industrial policy towards the creation of areas that offer certain benefits to the firms with the idea of accelerating investment, innovation and growth. Hypothesis 1. Proximity between firms generates agglomeration effects that benefit firm’s performance. 4. Knowledge spillover effects. Spillovers refer to valuable information that appears to flow within firms. The belief is that proximity enhances the flow of knowledge and makes casual communication inexpensive (Deutz, 2008). Due to the dependence of external economies on the proximity effect, benefits increase as more firms are agglomerated in one location. These externalities generate benefits for business depending on their absorptive capacity; more specifically, technological spillovers rely on firm’s ability to use information and new technologies. Firm proximity increases the possibility of innovation and the emergence of new techniques when enterprises share knowledge, technologies and resources. The activity of exchanging information enables companies to be more flexible and innovative (Walcott, 2003 p.1). So, in general, the idea is that a space with a large number of firms generates dynamism. This is because when firms are close to each other, there is a big chance to contact early adaptors of new technologies, and information flows occur more efficiently. Spillover effects are considered an external factor; however, because of its importance in the literature, I have separated it from the general concept of external economies. To analyze the importance of spillover effects in firm’s performance, there have been used two hypotheses, the first one is called the Marshall-Arrow-Romer proposition, which specifies that information and knowledge are easily exchanged between firms within the same industry, which could be referred to the cluster cases. However, there is evidence that 8 innovation increases when firms interact with actors from other industries (Johansson, 2004); the last argument supports the development of industrial areas that captures firms within different industries. Nevertheless, these arguments have not been proved by empirical studies, and research in this field is still limited (Johansson, 2004), for instance, there may be cases where the value of innovation is reduced due to the rapid flow of information between firms, what produces disadvantages over firm’s performance instead. Within agglomeration theory, one of the areas that captures the most interest is the relation of agglomeration effects and innovation. The reason is that spatial concentration of firms is considered a desirable phenomenon, because reducing the distance between firms increases the likelihood of the transmission of information, which in turn increases the occurrence of innovative activities and decreases the learning curve (Walcott, 2003 p.35). Hypothesis 2. Due to proximity, information flow is more efficient within industrial areas and knowledge spillovers are most likely to occur. Therefore, the innovative capacity for a firm within an industrial park or cluster will be higher compared to outside firms. The literature states that firms benefit by locating in particular areas or districts due to the agglomeration advantages. Being in an area with improved infrastructure, a skilled labor market, financial and legal services, information access and contact with firms with similar needs generates positive effects on performance (Walcott, 2003 p.4), network creation is also improved. In this sense, research in agglomeration theory has identified positive effects for firms due to their localization. 9 Figure 1 Firm´s benefits from Agglomeration Effects Industrial Parks and Clusters External Economies of Scale Share of factors, institutions,markets Networking Reduction in transaction costs Policy incentives Tax exceptions, export licenses, acces to credit Knowledge spillover effect s Increases the possiblity of innovation Agglomeration Effects Agglomeration effects have been evaluated with different models and methods, all of them with the purpose of understanding the benefits of proximity within firms. However, there is an on-going debate about the positive effects of agglomeration on firm performance since empirical studies resulted in contradictory findings. Malmberg and others (2000), studied a sample of Swedish export firms and found that the importance of localization economies (links between firms) in firm performance is not as crucial as theory predicts. The effects of urbanization economies and traditional scale economies are more important. This study recognized the need to reinforce the empirical evidence within economic localization literature, specifically in how and to what extend localization strengthens firms performance. However, looking at how proximity improves innovation, Audretsch (1998) explains why and how geography matters, arguing that globalization has increased the importance of knowledge based economic activity. His argument is that as knowledge is generated and transmitted efficiently by local proximity, economic activities based on knowledge will tend to agglomerate. His study is based on the idea that knowledge is vague and difficult to codify, and that the marginal cost of transferring knowledge increases with distance. 10 In the case of China, there are no studies about the relation between agglomeration and firm´s operation; however, Fan and Scott (2003) found positives effects in region´s economic performance. There are interesting examples of positive cases within certain industries, specifically industries that are essential for the economic transformation process, like consumer electronics for example. When looking at the causes of agglomeration in China, Chen and Lu (2005), identify a positive relation between economic opening and industrial agglomeration. On the other hand, it is considered that there is an over excitement about industrial zone implementation that is not supported by enough empirical evidence. Potter and Watts (2010) argue that Marshall’s agglomeration economies decrease the economic performance of the firms in the manufacturing sector; the research shows that agglomeration economies are positive within the first stages of the industry life cycle. Their analyses are based on evolutionary theories; accordingly, effects from agglomeration economies depend on time. Similar findings are presented in Kukalis (2010), he finds that there is no relationship between agglomeration externalities and a firm’s performance in the semiconductor and pharmaceutical industries. He compares firms within a cluster with outside ones, determining that outside firms can even have higher benefits at the later stages of the industry cycle. He emphasizes that the enthusiasm of politicians and scholars in this field. In this scenery, my research is an effort to understand the effects of proximity and agglomeration in firm´s performance and innovation, evaluating whether localisation within clusters and industrial parks have an actual relation with firm´s operations. Considering that there is no strong empirical evidence within this field and bearing in mind that there is an over excitement in China to use industrial zones as a policy tool to encourage growth and innovation, my empirical model determines to what extend the enthusiasm in this policy shall be revaluated. 11 Chapter 3 Industrial Parks Vs Natural Clusters 3.1 Overview of Industrial Parks and Cluster Development Agglomeration effects occur when firms concentrate within a specific area, sharing benefits from being near each other; proximity within firms is product of a variety of factors like: history, localization, natural resources or government initiative. In recent years, governments have promoted proximity through the construction of industrial parks with the idea of attracting investment and economic growth. On the other hand, governments have supported cluster´s formation as a policy tool to enhance development, due to the advantages from dense networks of firms, powerful externalities and information spillovers (Porter, 2007). Within this context, this chapter analyse the development process of industrial parks and clusters in China, as policies that seek to generate local economic growth, underlying the importance of regional institutional development, when taking advantages of these areas. Around the world, two types of industrial agglomeration can be found. The first one emerges deliberately by government policies to attract foreign investors and certain type of industries within a limited space. An example is the construction of industrial parks, where the main characteristic is the presence of large firms with low inter-firm connections (Long, 2009). The second type is the spontaneously grown clusters; in this case, natural agglomeration forces create conditions that incentivise economic agents to follow commercial initiatives (Yoshino, 2010 p.43). In China, many clusters overlap between spontaneously formed clusters and policy-driven initiatives, and in some cases, it is difficult to make distinctions between them. However, when looking to natural grown clusters, I refer to clusters conformed by small and medium firms, with little or non government support. It is considered that this type of clusters emerges in developing countries to encounter difficult business environment, with weak institutions and high transaction costs (Yoshino, 2010 p. 44). Specifically, proximity and strong linkages between firms increase the likelihood of relying on trust relations instead of contracts and formal market institutions that usually are not efficient in developing countries. For instance, in terms of access to credit and financial resources, Ruan and Zhang (2009) 12 provide empirical evidence from the cashmere sweater cluster in China, showing that regular interaction between buyers and suppliers builds strong relationships, what helps small business to face financial constrains through trade credit among firms. In general, industrial agglomeration in China has increased dramatically with the economic opening process that takes place in the last 30 years (Chen, 2005). In 1978, the Chinese government launched the economic reform that has lead to an unprecedented development in the country. The main characteristics of this reform are its gradualism, experimental approach and most importantly, the government commitment; The Special Economic Zones are an example; everything began when the central government opened the doors of the country to foreign investment and offered special benefits to international companies within delimitated areas. This areas are characterized by having better conditions in terms of infrastructure, labour and institutions, and by a more liberal and safe environment to make business. During the first years of the reform, four special economic zones were established, with the idea of testing a market orientated economy and with this attract foreign investment, technologies, promote exports and generate employment. After the success of the zones, the central government have promoted its development all over the country by giving more autonomy to local governments to capture investment. During this process, different types of economic zones have been created and usually, within these areas industrial parks are located. Different aspects of the Chinese economy have been improved due to the economic zones; contributions in growth of GDP, employment, technology upgrading, exports promotion, attraction of investment, among others, are some of the benefits attributed. In 2008, the GDP of the 54 state level Economic and Technological Development Zones (ETDZs) accounted by 5.1% of China´s GDP (US$ 220.3 billion); the High-tech Industrial Development Zones (HIDZs) generated 9.1% of China total value-added industrial output. And in terms of FDI, the ETDZs attracted 21.1% of China´s total (US$19.5 billion) (China Knowledge Online, 2011). It has been recognized that despite the differences in performance between these zones, their general effect in economic development has been positive. 13 On the other hand, clusters have benefit from the opening up process through knowledge spillovers from foreign companies and new market niches; for example, it is considered that companies like IBM, Acer and Compaq located in the costal areas, have provided initial technologies and skills needed to develop software clusters. In China, clusters have played an important role in promoting local economic growth, enhancing competiveness of industrial sectors within domestic and foreign markers, and in the technological upgrading process. However, it is difficult to measure the contributions of clusters to the economy; Douglas (2008) find that when looking at ten representative clusters in China, the average number of firms is 3.427 per cluster with 160.000 employees and 157 million yuan annual output; the sample includes toys, men´s clothes, children´s clothes, lamps, furniture, among other clusters. 3.2 Features of Industrial Parks in China1 Around the world, governments are using industrial areas with the idea of offering services and special conditions for firms, aiming to transfer and generate growth in lagging regions or potential ones. In resent years, the creation of industrial areas worldwide has increased dramatically. In developing countries, governments have the role to create and support economic growth, in some cases resources are invested to attract certain kind of capital through constructing sites that offer to investors, good deals in terms of transport facilities, infrastructure, low taxes, and good business environment, among others. Industrial parks (IP) in China are considered the cornerstone to attract investment. IPs represent groups of enterprises in the field of production activities and services concentrated within a specified territory, and which share the same infrastructure, roads, transport and public utilities; the function of these structures is to reduce costs and generate economies of scale (Zekovic, 2009). In other words, an industrial park, or area, is a collective location or limited space that belongs to a number of firms from the same or different industrial sectors. In China, there are 11 different types of industrial areas (Table 1), which present attractive opportunities for firms. 1 This section is based on personal interviews to managers of two industrial zones in China and the paper by Geng, 2009; there were limited sources of information to construct this section, industrial park´s websites were also used. 14 Table 1 Industrial zones in China Industrial Parks Industrial Area Industrial Development Zone Economic Development Zone Free Trade Zone Scientific-Industrial Park Software Park Economic & Technological Development Zone Export Processing Zone High-tech Industrial Development Zone Development Area Source: http://industrial-zone.fengpu.com/ For the purpose of this research, the name of industrial parks is given to all the different industrial zones in China. In 2007, there were 1568 industrial parks. It has been recognized there is a boom of industrial parks construction, due mainly to the autonomy of local governments in attracting private investments which has generated competition between regions; IPs are considered the major policy tool to meet targets of local growth. A typical industrial park in China has some areas that are common for firms; the industrial production area, scientific research area, residential area, recreational area and business and services area (Geng, 2009), however these features varies depending on each site, since industrial parks have been designed to cater firm´s needs; to do that, there are different types of parks: technology parks, export zones, software parks, trade zones, among others. In general, industrial parks have better infrastructure and conditions, compared with the terms offered to firms located elsewhere. Within this context, it is important to understand the advantages offered by industrial parks and its development constraints, as these areas are considered an important engines for China´s remarkable growth and its experience could be an example of the positive benefits of agglomeration economies. 3.2.1 Advantages offered Industrial parks offer a bunch of good deals to private investors, within them, the most attractive are the preferential policies; for instance, inexpensive land, tax breaks, custom clearance, duty free import of raw materials for export products and export tax exemptions, are the most common ones (China knowledge, 2011). 15 On the other hand, these sites are considered to have better infrastructure and investment conditions, due to a better planning and support (Geng, 2009); advantages like transport facilities, logistics, real state, infrastructure and qualified labor force, are better established within these areas. In terms of infrastructure, industrial zones offer better quality in public services like electricity, roads, water, telecommunication, drain system, gas, water disposals and in some cases, steam for manufactures; prices for these services are the same for all companies in the area. When looking at websites, there is a common pattern in the information offered; government support and preferential policies are the principal aspects that shows up, after that, location and transport facilities are presented in clear detail, and finally, information about other advantages is presented, for example specific features of the park, the surrounding universities, R&D investments, linkages with other parks or the characteristics of the residential areas. The business environment is also improved in the parks. A common example is the services offer to set up SME and the constant assistance in business operations; the Nordic Industrial Park (NIP) presents a pool of specialist to assist its investors: lawyers, accountants, human resources specialists, IT engineers, among others (NIP website). Due to the characteristics of the economic and bureaucratic system in China, specialized services are offered, especially for foreign investors; among them, the most common one is advisory services for foreign and domestic companies in terms of bureaucratic procedures; the administration office of the park is committed to help dealing with registrations, licenses, tax process, and certifications. For instance, dealing with the High Technological Certification is a common service; in recent years this process has become very important, not only for firms, but also for the industrial zone recognition. Looking at these services and conditions, it can be concluded that industrial parks reduce infrastructure construction costs for single firms, they also offer better access to markets due to transport facilities, and finally, regulation is easier to follow; for example, in environmental issues, the parks as a whole has to deal with all government requirements, and the firms has to comply and follow the environmental friendly system already established by the park, what is less complicated than dealing with these issues individually. 16 3.2.2 Development constrains Despite the attractive benefits that parks offer to firms and its success in attracting capital and new technologies, there are challenges that the central government has to face. The over-development of industrial parks in China is related to the incentives from local governments to attract new investments at any cost, due to the fact that political careers of local bureaucrats are linked to economic growth. Therefore, financial risk and the environmental consequences are not always taken into account when industrial areas are planned. Some of the areas have been created without any thought and are not interesting for investors; therefore, in some cases the real capabilities and potential of the parks are in doubt (Geng, 2009). One of the main limitations for industrial park development is the scarcity of land. First of all, project developers have to face complicated bureaucratic processes to acquire land. The central government determines a land quota every year, which is given to industrial zones at the municipal and county-town level. In 2008, industrial parks accounted by 0.004% of the country's land that attracted 10% of foreign investment (Geng, 2009). In terms of land administration, the government has a bind between their motivation to attract capitals and the importance of land for food production. In some areas, usually far from the coastal region, industrial zones developers break the law in terms of land acquisition. However, land scarcity has helped in controlling industrial park´s construction that has been a difficult issue in the implementation of this policy. In 2002, there were 6.866 industrial parks (state-level, provincial, municipal and county- level), after the program to normalize land usage launched by the central government, the number of parks has been reduced dramatically to 1,568 by the end of 2007 (China Knowledge Online 2011). On the other hand, during first stages of industrial park's development, the initial investments on infrastructure rely on bank loans that are approved by the municipal and central government. The importance on careful planning is crucial to use public resources in a proper way. To deal with the investment goals imposed by different levels of administration, during the first years, parks concentrate their efforts in attracting investment, in most cases, foreign capital and well-known companies that could be signal of credibility for other firms; the process is done through sellers who hunt investment, 17 offering a variety of facilities and good deals. The competition between parks and the urgency to meet the goals has caused the waste of public resources; firms have been located within certain areas without planning future operations or without taking into account the importance of linkages with other firms. It is considered the desirability to concentrate the same or similar industry sectors in specific areas, instead of placing firms randomly what decreases the gains from agglomeration and comparative advantages (Douglas, 2011). Moreover, it is well recognized that institutional features are crucial for firm’s development and growth. In this context, despite the great support and proactive participation from the Chinese government in the promotion of IPs, aspects like trust, planning and pro-business environment are engine for the success of industrial parks and more specifically crucial for attracting investment and generating growth. These aspects are not well developed in all regions in China; Douglas (2011) gives a good example: While Shenzhen was quick in identifying its industrial position and to build a good enabling environment, Zhuhai and Shantou seemed a step behind…Zhuhai actually overbuilt its infrastructure beyond sustainable demand … Its over- sized airport exhausted its initial capital ... Shantou has reached average rates of economic growth, but at various times that growth has been stalled by scandals traced to corruption, customs irregularities, smuggling, and the like. It also suffers from poor social credit and trust. In addition, the urban and zone management is not well planned, and there have been some institutional conflicts…Shenzhen seems to be more innovative in designing many pro-business policies and institutions, perhaps because of its immigrant culture, where investors feel more accepted and have a sense of ownership. In comparison, Zhuhai and Shantou are historic cities with strong local customs and culture, as well as their own languages. Such an environment might sometimes deter foreign investors and innovative approaches. (p.23) In this context, not all parks are successful; some of them offer poor services and exaggerate their capabilities on websites and advertising. With the boom of industrial zones in China, policy makers have left behind concerns about difficulties in duplicating successful experiences that arise due to differences between regions or districts, in terms of 18 local policies, historic contingencies and quality of institutions (Lovering, 1999). The government's interest in attracting firms in certain areas has been dominating, and policy makers have promoted setting up considerable number of industrial parks. Nevertheless, the region or district sometimes does not have the necessary condition to transfer the agglomeration advantages to firms; as a result, the actual impact of the industrial zones policy is in doubt, despite of the general argument, supported by the economic literature, about agglomeration advantages in firm’s performance and local growth (Hengxin, 2009). In this context, Cluster´s support could play a better role in regions where the necessary conditions for attracting investment are not sufficient; since within time, clusters improve institutions and the business environment (Porter, 2007). 3.3 Features of natural clusters in China Clusters are the concentration of firms, suppliers, support services, specialized infrastructure and specialized institutions located in a specific area; usually within a cluster there is a mix of manufacturing and services companies that support similar industries (Porter, 2007). The whole idea and main contributions in this field were done by Porter 1990, in his book The Competitive Advantage of Nations. The cluster idea is based in the agglomeration theory, where efforts and resources from nearby firms, institutions and governments, are shared to improve productivity and business development. In this section, I will focus my attention on natural grown clusters in China, clusters that emerge spontaneously and that develop agglomeration advantages over the time. An example is the eyewear cluster in Danyang City located in Jiangsu Province; the industry started in the 1930s and 1940s when peasants moved to Shanghai and Suzhou to work in eyewear factories. When they went back to their home town and the economic reform process started, the production of different products like lenses, frames and screws expanded within the area. In 1985 there were 23 townships factories and, by 2004, there were 50,000 workers and more than 1,000 factories and firms related to the sector (Ding, 2007). 19 In China, there are certain general characteristics that are common among natural clusters. The first one is the specialization of rural towns in a specific manufactured product, who frequently label themselves as the world’s “sweater city”, “socks city”, “kid’s clothing city”, “footwear capital” and so on (Yeung, 2009), formed by small and medium firms with strong linkages between them often due to family and social relationships in a specific town. The most common features of Chinese clusters are: the high division of labour, low capital barriers to entry, low reliance on external financing and fierce competition (Long, 2010). The benefits from clusters are expanding over the time due to historical, locational, and institutional factors. In general, it is well recognized that clusters improve industrial competitiveness through different channels as product specialization, strong relation between firms that creates business value chains and the stimulation of productivity and innovation (Douglas, 2011). 3.3.1 Gains from clustering It is considered that China’s manufacturing clusters compete basically with low price, cheap materials, labour flexibility and low-cost labour. Criticism about unfair conditions has been extensively done by different institutions around the world, what has pushed the central government to support cluster´s development, and its upgrading process (Wang, 2009). However, most clusters in China are still labour intensive with a low use of capital when comparing to the not clustered firms. These features are in line with the country’s comparative advantage, low capital-labour ratio and high population density, what explains the fast development of some clusters (Long, 2010). Different conditions, due to the characteristics of the Chinese economy, have helped the growth and development of industrial clusters and have brought benefits for firms and industry competitiveness. First of all, the efficient division of labour, within value chains and the constitution of strong production networks, are considered advantages for clusters competitiveness in China. Small and Medium Enterprises (SME) focus their operation in a specific phase of production, in some cases production tasks are developed in family workshops. At first stages of the cluster formation, there are low capital and technological barriers, what facilitates the entry of new firms and also increases the competition (Yeung, 2010). The 20 children´s garment cluster in Zhili Township, located in Zhejiang Province is a good example where the whole town is involved in the production of these goods with more than 6,000 enterprises, about 200,000 workers and account for one-third of the national market (Fleisher, 2009). The reduction of credit constraints is another positive effect from clustering in China; proximity reduces the temptation of dishonesty and makes trade credit possible (Long 2010). Also, when clusters are developed within towns and villages, long term relationships between family and neighbours become institutional substitutes for court enforcement, between lenders and borrowers and between buyers and suppliers (Fleisher, 2009). Moreover, the importance of proximity between firms within the same industrial sector is crucial to compete and capture domestic and foreign markets, since clusters create a collective image for firms that facilitate marketing capabilities and the positioning of products, what become a competitive advantage and a public good for all firms within the cluster (Kang, 2007). In addition, the development of clusters has brought the government support. Infrastructure, institutions and a good business environment have been placed within successful clusters. The construction of specialized markets, logistic business centres, industrial parks, public technology innovation centres, and also preferential legislation like tax reduction, are some of the strategies used. An example is the implementation of regulation that keeps production standards and has become very common among the different clusters, for instance, the Quality Control and Inspection System in the Cashmere Marketplace in Puyuan, Tongxiang. However, this support comes when the cluster has shown its potentiality; usually, at first stages, SMEs base their production process in cheap labour, family capital and old techniques. There are different constraints that affect the long term development of clusters. The main constraint is the difficulties on implementing technological improvements. It is considered that intense price competition, common within Chinese natural grown clusters, leads to the deterioration of quality (Kang, 2007) in this cases competition does not generate efficiency 21 or innovation. Also, difficulties arise from shortages of skilled labor, weak management capabilities and lack of training programs, what limits a step forward on the value chain (Yeung, 2010). Despite those problems, studies have found that firms within natural clusters have a better performance in terms of sales; also, their ability to reach markets is superior than for outsider firms (Mytelka, 2000). It has been recognized that government support, especially in the upgrading process, is crucial for the survival of clusters; because of these the local governments in China have developed different policies to forecast clusters capabilities. There are still problems to overcome in terms of quality, rising wage costs and institutions that enforce market relations; however, there are positive cases where clusters have brought the attention of local governments, and public goods and institutions needed have been placed. In general, the government commitment and its awareness about the importance of clusters are positive signals of future cluster development. 3.4 Top down versus Bottom-up: The China case. In Table 2, there is a detail description of clusters and industrial parks development, which summarizes their advantages and constraints what explained before. From the table, a main difference between bottom-up and top-down approach can be identified. Successful natural grown clusters are based on linkages between firms and strong collective action that allow to face market failures and generate a collective image that helps in capturing markets; on the other hand, policy-driven initiatives forget, in some cases, the importance of linkages between firms, focusing on capturing capital and investment to support the infrastructure and future development of the sites. There are successful cases of industrial parks and clusters; however, there are also concerns for their long term development. Moreover, clusters are established over time, where market forces and natural conditions evolve and create linkages between economics agents, in this cases, the government role is to speed up the process, for example through industrial parks promotion that brings the infrastructure needed to enhance the clusters operations. However, the characteristics of regions and market's needs should be taken into account where industrial parks are constructed, and a careful planning about investors and sector specification must be done. It is considered that a combination of bottom-up and top-down approach is an ideal policy, 22 specially for developing economies, where it is crucial to enhance innovation and accelerate the developing process (Douglas, 2011). Table 2 Top-down vs. Bottom up approach in China. Main Features Origen Industrial Parks- Top-down Clusters – Bottom-up Government initiatives to attract investment in specific areas. Different factors evolve spontaneously and create the necessary linkages between firms, the market and institutions that support and enlarge the development of markets. Difficult to develop, big capital investment, need of government commitment Emerge spontaneously, over time It is composed by SEO, foreign and private, medium It is composed by domestic small and medium firms and large firms Localization In its construction has to be considered the capacities and the quality of institutions of the region. Long history of production and business activities in a particular sector, which allows for tacit knowledge and skills that are transmitted by generations. Mainly in Costal Areas Usually located where there are abound human and natural resources, mostly in towns or major cities. There is proximity between firms, however not from Firms emerge to support a common market and they are the same or related sectors specialized in related products within the same sector. Characteristics Operate within more technology and capital intensive sector Operate in low technology and labor intensive sectors Relies on government support Relies on face to face contact and strong social relations developed between firms; for example, to lower financial constraints. The government's support is received after the cluster shows its potential. Better infrastructure and transport facilities Low entry barriers and capital investment More foreign investment There are successful cases, where the products compete in foreign markets Strong links with the foreign market Source: Douglas (2011); Long (2010) Chapter 4 Empirical Analysis 4.1 Research Design: Modelling Agglomeration Effects This paper asks if there is a positive effect between agglomeration economies and firm´s performance and innovation. Most of the studies have approached the subject using aggregate data and it is considered that there is little evidence to support the importance of agglomeration on the performance of firms (Burger et al., 2008). This study wants to contribute in this field using a micro-level approach. A number of researchers have analyzed this phenomenon using different methods; Andersson and Lööf (2009) study the relation between agglomeration effects and labour 23 productivity; they find a positive effect using the size of regions as a proxy for agglomeration, their main conclusion is that firms located in larger regions become more productive. Malmberg et al. (2000) focus on the relation between scales economies and urbanization economies on firm´s export performance, its results expose a positive relation between them; however, linkages between firms did not appear to have a significant impact. When looking at firm’s survival and employment growth, Bruger et al. (2008) conclude that agglomeration economies have a limited effect that depends on the size of the firm. On the other hand, Konoben (2001) find that a firm’s characteristics determine the relation between agglomeration advantages and firm´s performance, where firm´s size and face to face communication are found to be crucial factors that moderate this relation. When looking at agglomeration effects, advantages can be found from the use of a firmlevel approach. To begin with, the theory is based on the relationship between the external local environment and individual firms; when controlling a firm´s attributes, the effect of these external forces can be estimated. Moreover, a micro analysis allows reducing heterogeneity, when controlling for firm´s features the internal attributes which could affect the performance of firms are controlled (Andersson, 2009). And, by using firm level method, I avoid the common critiques to economic geography studies, in terms of the use of highly aggregated industry units (Malmberg, 2000). The empirical model develop in this study is similar to Malmberg (2009), who looks at the relation between export performance and agglomeration economies, where different measures are constructed to determine the importance of urbanisation, localisation and scale economies. However, in this study, I apply two different proxies for agglomeration: industrial parks and linkages with other firms within a city. The proxies are chosen for two reasons; first, industrial park´s effect is assumed to be external to the firm, where urban and external economies are available for all firms within a limited area. Second, the relation with other firms evaluates the effect of linkages and proximity between firms with similar conditions and technology. Different studies have used firm’s linkages as a proxy for agglomeration (Amiti 2007), and in my analysis, it is also used to describe cluster’s effect. As in other studies, which have used firm-level data, I control by a set of firm´s features that can affect firm´s performance and innovation. 24 4.2 Data The data for this study is from the 2003, World Bank Investment Climate Survey. The dataset contains information from 2400 domestic Chinese firms of which 604 are located within an industrial park. In the survey, there are relevant questions that could help us to determine the relationship between agglomeration effects, firm’s performance and innovation. In general, the survey covers aspects such as labour relations, constrains for the establishment, finance, the firm's performance, infrastructure and services, conflict resolution, government-firms relations, legal environment, crime, capacity and innovation (Dollar, 2004). The survey includes domestic firms which have been randomly chosen in terms of sector and location; more specifically, the firms have been selected from different regions of the country. However, there are some limitations in the database that could affect the results in the econometric analysis. Among these limitations are that firms located within inland regions show a small participation (Dollar, 2004), the survey also covers medium and big firms, giving a limited picture of the whole business environment; in addition there are a high proportion of state owned enterprises in comparison with other forms of legal ownership. However, the dataset is still a good core to analyze the effectiveness of IPs and clusters, as it contains information about both, firms located within and outside industrial parks (For a better understanding of the survey, see Dollar et al. 2004). As previously stated, agglomeration theory highlights the positive benefits from proximity and economies of scale. To analyse these benefits and its relation with firm´s performance, I have estimated a regression model that considers the theoretical background exposed in chapter 2. 4.3 Variables Dependent Variables I study agglomeration advantages on firm’s performance in two domains: sales growth and innovation. In the first model, the firm’s economic performance is determined by the firm’s sales growth in 2002, in order to use the absolute value of sales in other years as a control variable, this method has been applied in previous firm level studies (Opper, 2010) when 25 using the same dataset. Sales growth is considered a measure of firm´s performance, since it captures the firm's survival capacity and its ability to enter new markets. Economic theory determines that the drivers of firm's growth are the firm’s internal factors and the market environment (Bahadir, 2005); therefore in the model I have controlled by firm´s features that makes possible to observe the effect of the market environment. In the second model, innovation is considered as a share of new products introduced by a firm since 1999. The term innovation covers different activities that vary between firms, like the introduction of new products, new processes and patents, among others. It is also considered that the use of yes or not question to measure innovation, allows to categorise firms into innovative or not (Roger, 1998). From the survey, 40 % of the firms have introduced at least one product to the market between 1999 and 2003. Independent Variables In many agglomeration studies it is believed that disentangling concepts such as clusters and industrial districts are fruitful and necessary to study agglomeration effects (Knoben, 2011). In this study I use two proxies to capture both concepts. Industrial Park. A dummy variable is created where I give the value of 1 if a firm is located within an industrial park or the value of zero for a firm which is located outside. The survey asks if the firm is located in an industrial park, a science parks or an export processing zone; 30% of the firms are located within these areas. The variable intends to capture the benefits of urbanization economies and external economies available to all firms placed within a specific location, irrespective of the sector they belong (Burger, 2005). This pattern is the most common within IPs in China, where firms are placed irrespective of their sector of origin, and through the sharing of infrastructure, labour markets and institutions may develop certain advantages. With this variable, I try to evaluate IPs as a policy instrument that tries to enhance firm’s development. Suppliers and Competitors within a City: Two variables were created to measure cluster effect, firm’s relation with suppliers and firm´s relation with competitors. The question used as a proxy is the percentage of competitors and suppliers located within the same city. The variable is considered as a good proxy due to its capacity to capture firm’s links with 26 other firms, since clusters are considered as dense networks of interrelated firms that take place in a specific place or region (Porter, 2007). Most specifically, the variables capture localisation economies that are considered the benefits for firms within the same or similar sector which arise from being situated close to each other. Different studies have used external links with other firms as a measure for agglomeration (Burger, 2005; Amiti, 2007). Amiti (2007), uses industry data to estimate linkages with potential suppliers within the same industry and district. However, the dataset used in this study allows me to look at the actual level of suppliers within the same location, which could be considered as a better approximation. When using competitors, two different aspects are captured: first, the proximity between firms that share related technology, which is considered desirable to generate spillover effects (Amiti, 2007). Second, when city is used as a determinant for localisation, it is made certain that the distance between firms allows to face-to-face contact which in turn cultivates the exchange of knowledge (Burger, 2005). Control Variables The additional economic variables used to control firm’s features have been selected based on firm-level studies: Firm Size: It is measured by the natural logarithm of total assets in 2003. This variable represents the internal economies of scale, since the unit cost of production for firms is a decreasing function of output (Burger, 2005). I also include this variable, to capture highly heterogeneous firms within the dataset. Also, since larger firms are usually more attractive to locate within industrial parks and clusters are characterized by the presence of small and medium firms, it is important to control by this feature. Ownership: I divided the database between private and government owned firms, with the idea of differentiating between agglomeration effects and the form of ownership. In 22% of the cases, the government owns more than 50% of the firm. This control variable has been used by Amiti et al. (2007) to control by firm heterogeneity. Industry: Eight dummy variables have been constructed for industry with the idea of capturing the effect of differences in fixed costs and industry mark-ups (Amiti, 2007). This variable has been included in various agglomeration studies (Andersson, 2011; Amiti, 2007; Knoben, 2001), the reason being that traditional sectors are considered to be more 27 aggregated, therefore when controlling this aspect, it is possible to look at specific effects determined by a firm´s location. Also, industry policies in China are sector specific and can affect directly firm´s performance. Firm's age: The likelihood of being located within an industrial area can be affected by the age of the firm. Also, previous studies have determined that older firms have advantages in innovation as they are more likely to access loans, assets, among others; and, that newer firms are more flexible and fast in this aspect (Kanoben, 2008). The natural logarithm is taken for this variable. City: I have controlled by city level to separate its effect, from the agglomeration advantages. A city can offer better infrastructure and business environment that should be differentiating from the features of industrial parks and clusters. Export license: It is considered that firms that are able to compete abroad have more capabilities since they must meet foreign standards and its innovative capabilities may be affected due to the contact with other markets (Malmberg, 2000). A dummy variable is used, which capture whether the firms has an export license. Manager's education: The performance of the firm is related to the manager’s capacities; managers are in charge of the utilization of internal assets and the operational efficiency of the firm (Bahadir, 2005). From the survey, a dummy variable is created where I give the value of 1 when the manager has completed undergraduate education or higher, and 0 otherwise. Table 3 reports summary statistics of the data used. I have applied logarithm to all continuous variables, as they were skewed to the left since the transformation helps to reduce the impact of outliers. Among the most important features, the size of the firm determined by the firm´s assets shows that the survey represents mainly medium and large firms, whit 25 % with more than 1 million yuan on fix assets. Moreover, the correlation matrix (Appendix 1) shows that the independent variables are not heavily related to each other, with pairwise correlation coefficients from -0.21 to 0.30. Also, there is a weak positive correlation between competitors and suppliers within the same city (0.31) however, to analyse agglomeration effect it is important to look at the relations of firms within a limited space, and those variables may capture this relation. When looking at the relationship between industrial parks and firm´s age, there is a negative correlation (-0.23) 28 that indicates that older firms are slightly less likely to be located within IPs. On the other hand , the matrix shows that an agglomeration´s proxies are independent from each other; industrial parks, and suppliers and competitors within the same city correlates negatively, however, the correlation is rather low (-0.12,-0.11). In terms of innovation, there is a positive correlation with the industrial park variable with a low value (0.20), which is not surprising as firms within these sites have larger investment in R&D, also in some cases these sites are specialised in high-tech sectors. The variance inflation factors also confirmed the low correlation between variables with values around 1, so in this model there is no evidence of multicollinearity. Table 3. Summary statistics Variable Observations Mean SD Min Max Sales Growth 2002 2354 22.4 93.3 -100 897.56 Innovation 2335 0.4 0.49 0 1 Industrial Park 2353 0.25 0.43 0 1 % of Suppliers within a city 2212 26.3 32.2 0 100 % of Competitors within a city 2274 23.3 31.7 0 100 Firm age (Log) 2399 2.42 0.79 1.09 3.97 Value of sales 2001 (Log) 2372 9.08 2.25 0.19 18.86 Total fix assets 2002 (Log) 2395 8.93 2.43 0.18 17.7 Export activity 2277 0.23 0.42 0 1 Government 2399 0.21 0.41 0 1 Dependent Variables Independent Variables Agglomeration Firm´s features Source: Investment Climate Survey 2003 4.4 Methodology and Results The general model that I use for the empirical analysis is expressed as: Yi = βo + β1 Ai + β2 Xi + ci + si + u In the equation, Y is the vector that captures the performance of the firm in terms of innovation and growth. A is defined as the vector for agglomeration effects; X reflects the 29 vector for the control variables at a firm level; ci and si denotes the control variables for city and sector. 4.6.1 Results I run separate regressions to look at sales growth and innovation performance, the results are presented in table 4. The first column gives the estimates for sales growth using OLS; and the second column presents the results when looking at the relation between innovation and agglomeration effects, the estimation is done using the probit method. For sales growth (Model 1), the main finding is that firms located within an industrial park have a positive advantage over outside firms; this means that firms located in industrial parks have higher sales growth, all else equal. In terms of cluster relations, none of the variables seems to have a significant effect in sales performance. The results suggest that the effect of proximity is consistent to hypothesis 1, meaning that agglomeration effects contribute to firm´s performance. From the model, the industrial parks variable has a coefficient 0.20 and it is significant at 5%; if the firm is located within an IP its sales will grow faster than for an outside firm. In the case of China, aspects related with urbanisation economies that are common for all firms, like infrastructure and better transport facilities, have a positive effect on the performance of the firm. However, linkages with other firms do not have the same effect; relations with competitors has a positive sign what is expected from the theory however the relation with suppliers has a negative sign, both measures are not statistically significant. Having a higher percentage of firms related to the same business line in the local environment does not have an effect on sales growth. One of the interviewers highlighted that transport facilities are good enough within the area and that it makes no big difference to having suppliers within short distances or not. 30 Table 4. Agglomeration effects and firm´s performance Sales Growth Innovation Coefficient Coefficient (t-ratio) (z-ratio) Agglomeration Effects 0.20** 0.29*** Industrial Parks (2.03) (3.93) -0.00 -0.00** Suppliers city (-0.09) (-1.98) 0.00 -0.003** Competitors city (0.09) (-3.05) Firm's features -0.16** 0.00 Log firm age (-2.37) (0.22) -0.32 *** 0.18*** Log Sales 2001 (-8.05) (6.83) 0.05 *** -0.003 Log Total assets 2002 (4.67) (-0.11) -0.176 -0.035 Government (4.67) (-0.42) 0.23 ** 0.12 Exports (2.09) (1.44) 0.06 0.25** Education (0.54) (2.84) Industry Yes 0.89** (2.38) 0.36 ** (2.52) 0.38** (2.01) 0.38 ** (2.19) 0.51 * (1.83) -0.15 (-0.26) 0.90 ** (2.51) 0.89 ** (2.25) 1. Garment & Leather 2. Auto parts 3. Electronics 4. Information Technology 5. Chemical Products and Medicine B.P 6. Food Processing 7. Metallurgical Products 8. Business services City Yes *(1) 3.41 (7.22)*** OLS 0.11 1194 Yes 0.242* (1.80) 0.252 ** (2.03) 0.37 *** (2.65) 0.36 *** (2.58) 0.510 (1.41) 0.799 * (1.91) 0.305 ** (1.98) 1.08 ** (2.10) No (2) -1.25 (-2.65)** Probit 0.13 1950 Constant Method R^2 N *p < 0.10 **p <0.05 ***p<0.01 (1) Changsha, Kunming, Nanning, Wenzhou, Wuhan: Non significative (2) Chongqing, Haerbin * Firm's age and size have significant coefficients and present the expected signs, which can support the consistency of the model with the theoretical analysis. Among the control 31 variables, export has the strongest effect on firm´s performance, which supports the general hypothesis that firms which are able to compete in foreign markets have some advantages over firms focus on domestic markets (Malmberg, 2000). Also, industry sectors and city variables are significant and positive, which signals that sectoral composition and localisation are crucial in determining firm’s performance. A probit model (Model 2) has been conducted to look at the relation between firm’s innovation and agglomeration; the results are presented in column 2. The model evaluates whether firm´s innovation status is determined by agglomeration effects and firm´s features. The main conclusion is that being located within an industrial park increases the probability for firms to engage on innovative activities. Relationships with suppliers and competitors within the same city are also significant; however, there is a negative relation but the coefficient is very small in magnitude. The hypothesis 2 is supported with the results, proximity enhances the innovation capabilities of the firms when looking at industrial park's effect, the coefficient has positive sign as expected and is significant at 0.01%. However, it is not clear whether spillover effects, due to face-to-face contact with other firms have a positive significant relation. There is the possibility that within industrial parks, a firm develops strong relations with other firms, but in the model, the importance of the actual relationship with suppliers and competitors is not significant. Among the control variables, the education of the manager has a significant and positive effect on innovation, what is expected from the theory. Moreover, sector composition is relevant to increase the probability of firm’s innovation activities; although city variables do not have the same effect. The above models have an acceptable but not very high R-square and pseudo R-square (0.11 and 0.13 respectively), implying that some of the variation of sales growth and innovation can be explained by the variation of the explanatory variables included in the model. To determine the consistency of the findings, I determine whether industrial parks present advantages for the firms due to firm’s performance. Measures are applied to avoid 32 reverse causality, i.e. whether a firm's performance determines its presence within an industrial park. I applied lagged measures for firm size and the substantial findings remain the same. Also, I observed whether the results are sensitive to the spatial scale; for this, district level variables are used instead of city level as control variable, and interaction with suppliers and competitors is taken at district level; as before, all main findings were confirmed. 5. Conclusions This study presents a simple test for agglomeration economies at firm level. Localisation within industrial parks and the relations between the firm, its suppliers and competitors, are assumed to reflect the effects of agglomeration economies. This paper investigates the relationship between agglomeration and firm’s performance, and provides evidence that being located in industrial parks brings benefits for the firm; however, when looking at the effect of linkages between firms, there is not evidence of a relation with the performance, and a negative effect on innovation. Three questions were studied; first, does proximity create agglomeration effects that benefit the performance of firms? Second, due to proximity, is innovative capacity higher in an industrial park or cluster? Third, do industrial parks, as a top-down approach, generate comparable effects as natural-grown clusters? From the empirical exercise, three aspects are indentified. First of all, agglomeration theory highlights advantages from proximity; the dataset used, gives evidence of a positive relation between firm´s performance and proximity, when looking at industrial park's features. In China, industrial parks are characterised by their superior conditions in terms of infrastructure and facilities, where in most of the cases, firms are located irrespective of the sector they belong. The results can be generalized by saying that proximity through urban economies produces positive effects on the performance of firms. Second, industrial parks generate positive effects on firm’s innovative capacity, while linkages between firms do not have any significant impact. Proximity between firms which belong to similar sectors has negative effect on the innovative capacity, which goes against agglomeration theory. As previously explained, small firms rely on local institutional 33 arrangements and cooperation, however the dataset used in this study includes medium and larger size firms which can give biased results. Finally, as a top down approach, industrial parks offer infrastructure and conditions that benefit firm’s performance and innovation; the positive effects of natural grown clusters are not clear from the model, however, its importance for the Chinese economy is substantially significant for industrial development, meaning that a combination of both methods could enhance the future development of clusters and the competitiveness of the Chinese domestic firms. The analysis in this paper can be extended in future studies. For example, the relation between city features and agglomeration economies could help to determine how the characteristics of the environment shape the effect of the agglomeration economies. The importance of continued research on this field is due to the increase interest of local and industrial policies, in supporting industry agglomeration and proximity between firms. 34 References Andersson, Martin, and Lööf Hans. 2009. Agglomeration and Productivity-evidence from firm-level data. Centre of Excellence for Science and Innovation Studies CESIS. Working Paper 170. Amity, Mary, and Lisa Cameron. 2007. Economic Geography and Wages. The Review of Economics and Statistics. Volume 89(1) p.15-29. Audretsch, David B. 1998. Agglomeration and the location of innovative activity. Oxford review of Economics Policy. Volume 14, No 2, p. 18- 29. Bahadir, Cem, Sundar Bharadwaj, and Michael Parzen. 2005. A Meta-Analysis of the Determinants of Organic Sales Growth. International Journal of Research in Marketing. Available at: http://www.people.fas.harvard.edu/~mparzen/published/parzen34.pdf Burger, Martijn, Otto Raspe, and Frank van Oort. 2008. Agglomeration Economies and Firm Performance: A Mixed Hierarchical and Cross-Classified Model. Available at: http://www.israelrsa.org.il/ppt6/5.1.pdf Chen, Zhao, Jin Yu, and Lu Ming. 2005. Economic Opening and Industrial Agglomeration in China. Industrial Organisation Working Paper 0511012. Available at: http://ideas.repec.org/p/wpa/wuwpio/0511012.html. China Knowledge Online. 2009. China Special Report: Industrial Parks – China Vehicles for Manufacturing. Available at: http://www.chinaknowledge.com. Deutz, Pauline, and David Gibbs. 2008. Industrial Ecology and Regional Development: Eco-Industrial Development as Cluster Policy. Regional Studies. Volume 42:10, p. 13131328. Ding, Ke. 2007. Domestic Market-based Industrial Cluster Development in Modern China. Institute of Developing Economies. Discussion Paper 88. 35 Dollar, David, Shilin Wang, Lixing Colin Xu, and Anqing Shi. 2004. Improving City Competitiveness through the Investment Climate: Ranking 23 Cities. Available at: http://papers.ssrn.com/sol3/papers.cfm?abstract_id=530744. Douglas, Zhihua. 2008. Innovation and Cluster Development in China. Clustering for Competitiveness: Third International Clustering Conference. Istanbul. Turkey. Available at: http://www.clusteringconference.com/html/EN/ _ _ _. 2010. How Do Special Economic Zones and Industrial Clusters Drive China´s Rapid Development? Available at: http://www.worldbank.org/research/2011/03/13844692/ special-economic-zones-industrial-clusters-drive-chinas-rapid development. Fan, C. Cindy, and 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. Volume 79(3), p. 295-319. Geng, Yong, and Hengxin Zhao. 2009. Industrial Park Management in the Chinese Environment. Journal of Cleaner Production. Volume 17, 2009, p. 1289-1294. Johansson, Borje, and John M. Quigley. 2004. Agglomeration and Networks in Spatial Economies. Papers in Regional Sciences. Volume 83. Kang, Yi, and Stefanie Ramirez. 2007. Made In China: Coastal Industrial Clusters and Regional Growth. Issues in Political Economy. Volume 16. Knoben, Joris. 2008. Localized Inter-organizational Linkages, Agglomeration Effects, and the Innovative Performance of Firms. The Annals of Regional Science. Volume 43(3) p. 757-779. Krugman, Paul. 1991. Increasing Returns and Economic Geography. Journal of Political Economy. Volume 99, p. 483-499. Kukalis, Sal. 2009. Agglomeration Economies and Firms Performance: The Case of Industry Clusters. Journal of Management. Volume 36, p.453. 36 Long, Cheryl, and Zhang Xiaobo. 2009. Cluster-Based Industrialization in China: Financing and Performance. International Food Policy Research Institute Working Paper 00937 . Available at: http://www.ifpri.org/sites/default/files/publications/ifpridp00937.pdf Lindqvist, Göran. 2009. Disentangling Clusters, Agglomeration and Proximity Effects. The Economic Research Institute. Stockholm School of Economics. Malmberg, Anders, Bo Malmberg, and Per Lundequist. 2000. Agglomeration and firm performance: economies of scale, localization, and urbanization among Swedish export firms. Environment and Planning. Volume 32, p. 305-321. Marshall, Alfred. 1890. Principles of Economics. Macmillan. Masahisa, Fujita and Jacques-Francois Thisse. 2002. Economics of Agglomeration. Cities, industrial location and regional growth. Cambridge University Press, New York. Mytelka, Lynn, and Fulvia Farinelli. 2000. Local Clusters. Innovation Systems and Sustained Competitiveness. The United Nations University. Discussion Paper 55. Nee, V. and Opper, S. 2010. “Political Capital in a Market Economy.” Social Forces. Volume 88(5) p. 2105-2132. Nordic Industrial Park. Retrieved May 25, 2011. http://www.nip.com.cn/index.php?id =33&lang=en Porter, Michael. 1990. The Competitive Advantages of Nations. New York: Free Press. _ _ _ . 2007. Clusters and Economic Policy: Aligning Public Policy with the New Economics of Competition. ISC White Paper, November 2007. Harvard Business School. Potter, Antony, and H. Doug Watts. 2010. Evolutionary agglomeration theory: increasing returns, diminishing returns,and the industry life cycle. Journal of Economic Geography. Volumen 11(2011), p. 417-455. 37 Ruan, Jianqing, and Xiaobo, Zhang. 2009. Finance and cluster-based industrial development in China. Economic Development and Cultural Change. Volume 58, p. 143– 164. Schroder Friederike, Michael Waibel and Uwe Altrock. 2009. Global Change and China´s Clusters: The Restructuring of Guangzhou´s Textile District. Pacific News. Volume 33 p. 4-8. Shanghai Fengpu Industrial Park Export Processing Zone. Retrieved May 23, 2011. http://industrial-zone.fengpu.com/. Yeung, Yue-man, Joanna Lee and Gordon Kee. 2009. China’s Special Economic Zones at 30. Eurasian Geography and Economics. Volume 50(2) p. 222-240. Yoshino, Yutaka (Ed). 2011. Industrial Clusters and Micro and Small Enterprises in Africa: from Survival to Growth. World Bank. Zekovic, Slavka. 2009. Regional Competitiveness and Territorial Industrial Development in Serbia. SPATIUM International Review. Volume 21 p. 27-38. 38 Appendix 1. Pairwise Correlations Variable 1. Sales Growth 2. Innovation 1 2 3 4 5 6 7 1.000 0.065 0.171 1.000 4. Suppliers C -0.031 -0.109 -0.147 1.000 5. Competitors C -0.025 -0.105 -0.110 0.311 1.000 7. Log total assets. -0.057 0.253 0.116 -0.100 -0.181 1.000 6. Log firm age -0.053 0.001 -0.215 0.033 -0.050 0.253 1.000 0.032 0.183 0.240 -0.134 -0.171 0.377 -0.044 -0.047 -0.014 -0.188 0.044 0.038 0.244 0.004 0.182 0.164 -0.139 -0.112 0.264 8. Export activities 9. Government 10. Education 9 10 1.000 -0.017 3. Industrial Park 8 39 1.000 0.416 -0.085 -0.062 0.156 1.000 0.066 1.000