Cross-border Urban Networks Based on Manufacturing Global Value Chain: A Study of Listed Companies in Western China

Cross-border investment is essential for western China’s globalization. Global value chain (GVC) forms cross-border investment networks between industries in western China and overseas cities. Focusing on GVC, this study uses the social network analysis method, entropy method, multi-index comprehensive evaluation method, and quadratic assignment procedure analysis method to examine the characteristics and influencing factors of the urban networks of research and development (R&D), production, and sales formed as a result of the overseas investments of listed manufacturing companies in western China. Results showed that the three types of investment networks involved multiple industry types and multiple central cities with differentiated diversity and multicentrality. The R&D urban network’s leading sub-industries were the mechanical equipment and instruments, medicine and biological products, and metal and nonmetal industries. The destination cities were mostly those home to educational and scientific research centers. The production urban network’s leading sub-industries were the mechanical equipment, instrument, and food and beverage industries. The destination cities were mostly regional central cities in developing countries. The sales urban network’s leading sub-industries were the mechanical equipment and instrument, metal and nonmetal, and petrochemical and plastics industries. The destination cities were numerous and scattered. In addition, the R&D urban network easily formed specialized clusters, core nodes easily controlled the production urban network, and individual nodes did not easily control the sales urban network. Technological and economic system advantages greatly impacted the three network types. Considering the different influencing factors, this study suggests optimizing the institutional investment environment to narrow the institutional gap, adjusting and optimizing the investment layout to expand overseas markets, and increasing R&D funds to stimulate technological progress and overseas investments in western China.


Introduction
Transnational corporations' global investment has accelerated the movement of population, goods, capital, and technology (Wall and van der Knaap, 2011).As the main node of transnational corporations' global investment network, cities have attracted the attention of inter-national geographers and economists (Zhao et al., 2015).Hymer (1972) predicted that cities would tend to form a hierarchical division of labor corresponding to the vertical division of labor within transnational corporations in a geographical space; this prediction laid the foundation for introducing transnational corporations into world urban research.Cohen (1981) studied the relationship between transnational corporations and world cities' grades in the international division of labor.Later, Friedmann (1986) studied the world cities' grades from the new international division of labor perspective, used the headquarters of transnational corporations as a standard to measure the grades of cities, and proposed the 'world city hypothesis'.Sassen (1991) explained why globalization was accompanied by an increasing concentration of economic activities in New York, London, and Tokyo and emphasized that the agglomeration of producer service companies promoted the formation of overseas urban networks.Castells (1996) proposed the theory of the 'flow space' of cities, believing that the growing global connections of cities was becoming more networked.Since 2000, domestic and foreign scholars represented by the Globalization and World Cities Study Group and Network (GaWC) and Alderson have conducted empirical studies on world cities from the perspective of enterprises, forming the research paradigm of world city network (WCN) by expanding enterprise types and optimizing network research models (Taylor, 2001;Taylor et al., 2002;Alderson et al., 2010;Hanssens et al., 2011;Xue and Zou, 2018;Zou et al., 2019).Based on the context of flow space and network, cities increasingly gain global functions and status from their connections with other cities worldwide.The research paradigm of WCN fits precisely in this context to redefine and shape the role and status of cities, and their relationship (Zhao et al., 2019).
Globalization has promoted the international division of labor and the formation of the GVC system.GVC theory originated from the intra-enterprise value chain proposed by Porter (1985) in the Competitive Advantage, which states that the various activities of enterprises are conducted to maximize value.Meanwhile, Kogut (1985) proposed the concept of a value-added chain to analyze the international strategic advantages of enterprises.Hopkins and Wallerstein (1986) presented the concept of a commodity chain in the world system, while Gereffi and Korzeniewicz (1994) proposed the global commodity chain on this basis, arguing that transnational companies will strengthen their competitiveness in the international market by allocating resources and production links worldwide through commodity chain.Subsequently, Gereffi et al. (2001) continued to study industrial connection, division of labor, and upgrading on a global scale and described the glob-al geography, organizational structure, and network characteristics of GVC.Following the research achievements of Gereffi et al., a relatively systematic research paradigm of GVC gradually formed.It refers to the cross-enterprise network organization that connects the whole process of production, sales and recycling in order to create and realize the value of a certain commodity or service in the world participating in the international division of labor and trade.It includes all organizations involved in production and sales activities and the distribution of value and profit (Zhang, 2004).The GVC theory eventually became widely used in studying transnational corporations' international division of labor and was integrated into the research of overseas urban networks.
With the increasingly prominent international division of labor, emerging markets and new organizational forms, many scholars have conducted in-depth analysis and discussions on GVC.In the context of the deepening international division of labor, especially the intraproduct division of labor, the active integration into cross-border urban networks through GVC has become an inevitable choice for urban economic growth and industrial development in developing countries.This integration also promoted the development of cities in developing countries dominated by the manufacturing industry.As the GVC provides an export-oriented development model driven by trade and competitiveness, countries do not need to establish complete vertical industry to participate in global trade.They can integrate into the international market as long as they have specific capabilities in a certain link of the value chain (Cattaneo et al., 2013).Owing to the development of the global intra-product division of labor, China's manufacturing industry has been able to start from the most elementary assembly processing and then gradually ' embed' into the global industrial chain system with an 'export-oriented' development strategy and penetrate the high end of the industrial and value chains (Yang and Huang, 2013).At the end of the 1990s, especially after China acceded the World Trade Organization, the Chinese manufacturing industry rapidly integrated into the GVC system through processing trade or original equipment manufacturing, relying on relatively complete infrastructure and low factor cost advantages and providing an effective impetus for China's industrialization and modernization (Ding, 2014).To meet the coun-try's balanced development, some scholars have begun to focus on the location transfer of the manufacturing industry in China.The path of location transfer is summarized as 'east coast-North China/Northeast Chinacentral and western China' (Lin and Wang, 2006;Li et al., 2010;Feng et al., 2010).Further research has found that some provinces and cities in western China have better efficiency in connecting the manufacturing industry (Wu, 2017).Even some technology-intensive manufacturing industries in the eastern coastal regions of China have moved directly to the western region across the central region before the labor-intensive manufacturing industries (Guan and Cao, 2016).
With the implementation of 'the development of the western region of China' and 'Belt and One Road Initiative', the connection between the inland areas in western China and overseas markets has been strengthened to a certain extent, especially with the markets in central Asia, west Asia, and Europe.However, owing to the disadvantages of inland areas in terms of economy, talents, technology, and export experience, a large gap still exists between these areas and the developed areas in eastern China (Cao, 2021).Relative to the process of industrial globalization in eastern China, the industrial participation in globalization in western China is still dominated by manufacturing, and the producer service industry remains limited.Meanwhile, the research on outward investment in manufacturing is mostly concentrated in developed countries, regions, and cities, ignoring the less developed regions.Moreover, most research is focused on overseas investment in the manufacturing industry as a whole, hence the lack of studies on subdivided industry types, especially from the perspective of GVC, to understand how manufacturing industry segments participate in the international division of labor and integrate into the global economic market.Under such background, the current study focuses on analyzing the sub-industries of the manufacturing industry in western China.It also constructs the relationship framework of upstream, midstream, and downstream cross-border urban network connections between sub-industries and overseas cities from the GVC perspective.Furthermore, it analyzes the characteristics of three cross-border urban networks, selects the influencing factors according to the eclectic theory of international production, and uses quadratic assignment procedure (QAP) analysis to identify the main influencing factors of the three urban networks.Therefore, on the one hand, this study is conducive to exploring the integration and development of GVC and WCN from a theoretical perspective.On the other hand, it is conducive to understanding the basic laws and internal differences of the overseas investment of China's manufacturing enterprises from a regional perspective.This study helps to rationally allocate resources globally, expand overseas markets, and optimize investment layouts.It also provides a basis for the scientific formulation of such cities' globalization strategies and related policies.Given the particularity of China's Hong Kong Special Administrative Region, Macao Special Administrative Region, and Taiwan Province, in accordance with the basic state policy of 'one country, two systems', they are listed alongside overseas cities in the following exposition and are expressed as 'Hong Kong (China)' in the specific analysis.The expression of this article does not involve 'Macao (China)' and 'Taiwan (China)'.

Relationship framework between GVC and cross-border urban networks
GVC and cross-border urban networks are integrated and developed.Both are typical products under the new international division of labor, new international trade, and international industrial transfer (Shi, 2005;Liu et al., 2021).Specifically, GVCs and cross-border urban networks are redeveloped around enterprises (especially transnational corporations).The GVC is an important way for enterprises to participate in global economic activities.A city is the carrier of enterprises to participate in global economic activities, and urban network development is the change in the spatial organization structure of the carrier (Man et al., 2021).According to the international division of labor, the GVC divides the production activities of enterprises into upstream, midstream, and downstream value chains.Then, different value links choose appropriate cities, and the cities under the guidance of the GVC inevitably cluster in different spaces around the world, thus forming crossborder city networks (Zhou, 2015).Under the GVC system, the flow of production factors becomes more frequent and convenient, creating a win-win platform for developed and developing countries.Relying on its ad-vantages in factor endowment, China has actively participated in the global market and rapidly integrated into the GVC system, promoting the development of the country's manufacturing industry and national economy (Yang, 2017).China's manufacturing industry invests a part of its value chain in the world through the GVC and participates in the global economy in such a manner.
Therefore, the GVC is an important bridge of crossborder urban networks between industries and cities.In this study, upstream, midstream, and downstream crossborder urban networks are formed by investing in the manufacturing sub-industries in western China to overseas cities.Among them, product development, core technology, and material supply belong to the upstream value link; production and assembly comprise the midstream value link; and finished product shipping, sales, and after-sales service comprise the downstream value link.By subdividing the industry types of the listed manufacturing companies' headquarters in western China and dividing the value chain links of the overseas subsidiaries based on business scope, the operation types of the overseas subsidiaries in the upstream, midstream, and downstream links of the value chain can be summarized as follows: the overseas subsidiaries in the upstream, midstream, and downstream links are mainly engaged in high-end, low-end, and middle-end businesses, respectively.Moreover, the overseas subsidiaries of different value chain links form the upstream, midstream.and downstream urban networks through modular development and spatial reorganization.The relationship framework in Fig. 1 shows that the GVC is essential for listed companies to establish subsidiaries abroad.The upstream, midstream, and downstream cross-border urban networks formed through this route can directly reflect the main industry types of overseas investment and show which industry links are mainly invested in overseas cities.

Influencing factors and mechanisms of crossborder urban networks
Understanding the main influencing factors of crossborder urban networks helps choose a more suitable investment path for current national conditions to promote overseas investment and move up to the higher part of the value chain.The enterprise network is the micro foundation of the city network, and the city is the carrier of enterprises.Cross-border city networks naturally form when enterprises make cross-border investments.Therefore, the factors that affect enterprises' cross-border investment fundamentally affect the pattern of urban networks.The existing theories on foreign investment involve more or less the location selection problem.Dunning's eclectic theory of international production, for the first time, clearly proposed that location advantage is the key factor leading to the cross-border investment of enterprises, emphasizing that only when ownership, internalization, and location advantages are met can favorable overseas investment behavior occur (Dunning, 1977).According to the changes in the operating environment of transnational corporations, Dunning introduced institutional elements for these three advantages in his later research (Dunning and Lundan, 2008).The eclectic theory of international production considers the enterprise and the city level, as well as the Fig. 1 Relationship framework of cross-border urban networks between industries and cities formed through GVC home country and host country.It is an important theoretical model for the current study to analyze the influencing factors of cross-border networks (Fig. 2).
Most previous studies on overseas investment have focused on the number of overseas subsidiaries.Although studying overseas investment from the overseas investment scale perspective is meaningful, we convert the scale index of overseas investment into an efficiency index -the overseas investment tendency.The overseas investment tendency refers to the number of overseas subsidiaries of A (domestic city) investing in B (overseas city) as a percentage of all foreign subsidiaries of A. This index can better reflect the actual investment tendency of A to B; it also supplements and expands the existing research on overseas investment from the scale perspective.
According to the eclectic theory of international production, ownership advantage includes intangible assets and economies of scale that are available to cities/enterprises in one country but are unavailable to cities/enterprises in other countries (Dunning and Lundan, 2008).It includes resources and capabilities that can bring monopoly advantages to enterprises, such as production, management, technical, and marketing capacities.The intangible asset advantage of the home country's city can be reflected by the production, management, technical, and sales personnel of the listed company.Higher personnel efficiency leads to more investments.The intangible as-set advantage of the host country's city is mainly through its brand effect, which can be reflected by whether it hosts enterprises among the world's top 500 or whether it is a first-tier city worldwide (World city ranking is based on GaWC which divides world cities into four levels: Alpha (first-tier cities in the world), Beta (second-tier cities in the world), Gamma (third-tier cities in the world), and Sufficiency (fourth-tier cities in the world)).A better brand effect makes it easier to attract investments.
Location advantage includes factor endowment and system advantages (Zhang, 2015).Factor endowment advantages include geographical, market potential, proximity (geographical and relationship proximity), and technological advantages.Cities with geographical advantages have a perfect transportation infrastructure and a developed economy, thus likely to attract investments.Geographical advantages include whether the home country's city is a provincial capital, whether the host country's city is a seaport or an airport city, and whether it is a tax haven (offshore financial center or flag-of-convenience city) (He and Xiao, 2011;Quan, 2014;Zhao et al., 2017;Martinus et al., 2019).Market potential advantage affects the ability of the home country's city to explore the overseas market and the market size of the host country's city.This advantage includes population growth rate, gross domestic product (GDP) growth rate and the proportion of the working popula- tion of the city in the home and host countries (Hu, 2015).Geographical proximity advantage reduces the transportation costs between two cities, as reflected mainly in geographical distance (Xu et al., 2012).Relationship proximity advantage shortens the psychological distance between two cities and reduces investment risks; it includes whether the cities belong to the Chinese culture circle and whether the countries of the two cities have a bilateral investment treaty (Ly et al., 2018;Sun, 2018).Technological advantage improves production efficiency, including the proportion of science and technology expenditure to the home country's city's financial expenditure and the host country's global innovation index (Wang et al., 2018).System advantage affects overseas investment mainly through political and economic systems (Xiang, 2015).The political system is measured mainly by control of corruption, government effectiveness, political stability and absence of violence/terrorism, regulatory quality, rule of law, voice, and accountability.The effective implementation of the political system lowers enterprises' operation risk, guaranteeing efficient market operation.Meanwhile, the economic system affects overseas investment mainly through government spending and fiscal, business, labor, monetary, trade, investment, and financial freedoms.
Internalization advantage helps consolidate the investment strength of the host and home countries.Enterprises internalize ownership advantage by expanding business activities rather than engaging in market transactions with other enterprises or externalization (Yang, 2006).A company's internalization advantage can be realized by establishing a new production and operation organization to maintain and utilize the monopoly position of the ownership advantage of the company and thereby maximize profit.Therefore, the number of overseas subsidiaries of overseas investments is an important measurement index.Meanwhile, the government's internalization advantage affects overseas investment mainly through tariff rates (Hu, 2015).

Research data
In this study, the cities in western China included all prefecture-level cities/regions/autonomous prefectures/ leagues in the 12 provinces/municipalities/autonomous regions, namely, Chongqing, Sichuan, Yunnan, Guizhou, Shaanxi, Gansu, Qinghai, Xinjiang, Ningxia, Guangxi, Inner Mongolia, and Tibet.The overseas cities were identified as destinations for the overseas investment of listed companies in the western cities of China.The data on the listed companies were primarily from the 2010, 2014, and 2018 documents in the China Stock Market & Accounting Research database (https://www.gtarsc.com/).The listed companies' overseas subsidiaries, including their addresses and business scopes, were mainly obtained from their annual reports.Some entities with missing data were identified as country capital, regional capital, or government residence.The data on the western Chinese cities included in this study were mainly obtained from the official website of the Chinese government, the Ministry of Commerce of China, and various statistical yearbooks, while the relevant data on overseas cities were from various channels and databases, such as the official Fortune website and the official GaWC website (Table 1).National data were used instead for some overseas cities for which indicators were unavailable.For the years for which indicator data were unavailable, the data from adjacent years or the average data of previous and subsequent adjacent years were used instead.

Network analysis method
Density reflects the degree of connection among various nodes in the network.The greater the number of connections in the network, the greater the network density.This study used network density to measure the closeness of connections between nodes.It is calculated as follows: where D is the network density, n is the number of nodes in the network, and L is the number of lines.Degree centrality measures the power of a node in the overall network and reflects the node's importance in the network.A network with a specific direction can be divided into out-degree centrality and in-degree centrality.In this study, degree centrality means that if a node is connected to many nodes, it is in a relatively central The Heritage Foundation database Notes: as the political and economic system advantages of the home country's city are replaced by those of the country, the weight can not be calculated.To maintain data level consistency, this study took the average position and has strong control power.It is calculated as follows: where C i is the degree centrality of node i, that is, the sum of the number of all nodes directly connected to node i, and x ij is 0 or 1, indicating whether there is a connection between nodes i and j.
Correlation represents the structural characteristics of the network, such as the concentration of power and information and the susceptibility of the overall structure to the influence of some nodes.This study considered the network highly correlated if the nodes had high connectivity and accessibility.The correlation indicators include connectedness, hierarchy, efficiency, and least upper boundedness.High network connectedness, efficiency, low hierarchy and least upper boundedness imply a high network correlation.
Modularity is an important indicator in community detection.It divides a network into several communities according to node relationships (Newman and Girvan, 2004).It is generally believed that higher modularity corresponds to better community detection.Modularity is calculated as follows: where n c is the number of communities, l s is the number of internal links in community s, and d s is the sum of all node degrees in community s.

Influencing factor analysis model
Based on Dunning's eclectic theory of international production, we selected ten influencing factors: ownership, location, and internalization advantages.The ten influencing factors were all selected from the respective cities of the home and host countries while excluding the advantages of geographical proximity, relational proximity, and company internalization, which can directly represent the relationship between the two countries.
We selected 17 explanatory variables, from which 51 specific indicators were derived (Table 1).We used the entropy method to determine the weights of the 51 specific indicators and utilized the multi-index comprehensive evaluation method to calculate the values of 17 explanatory variables.Before conducting the QAP ana-lysis, we converted the explanatory variables into relational variable data.The conversion method was mainly based on Dunning's international production compromise theory-overseas investment was the home country's thrust and the host country's pull.This study consolidated the explanatory variables of the home and host countries to form relational data on both sides and determine the influencing factor of overseas investment.
Through QAP analysis, we analyzed the correlation and regression relationship between overseas investment tendencies and various influencing factors.We used the entropy method to assign weights to the indicators.The entropy method measures the uncertainty in the state of a system.The calculation is as follows: where e j is the informational entropy of the index, y ij is the standardized value of the indicator, ω j is its weight, m is the number of samples, and n is the number of indicators in the specific indicator level.We used the multi-index comprehensive evaluation method to measure the explanatory variables.Each explanatory variable influences the overseas investment tendency in various aspects.The calculation is as follows: where t is the value of the explanatory variable, a i is the standardized value of indicator i at the specific indicator level, and ω i is the weight of indicator i at the specific indicator level.QAP is a nonparametric test method based on random permutation.QAP analysis includes correlation and regression analyses; correlation analysis examines the correlation between two matrices, while regression analysis examines the regression relationship between one matrix and multiple matrices (Liu et al., 2018).Therefore, this method is suitable for studying the influencing factors of investment matrices.

Analysis of Network Structure Characteristics
The Guidelines for Industry Classification of Listed Companies (https://www.chinanews.com/2001-04-04/26/83146.html) was used to classify the sub-industries of the listed manufacturing companies in western China.In addition, the value chain links of overseas subsidiaries were divided according to the business scope of the companies' overseas subsidiaries.Overseas subsidiaries in the upstream link are mainly involved in high-end businesses, such as investment, R&D, and product design.Overseas subsidiaries in the midstream link mainly operate low-end businesses, such as production and manufacturing.Lastly, overseas subsidiaries in the downstream link are mainly involved in middle-end businesses, including sales, trade, and services.Given the diversity of business scopes, many overseas subsidiaries were classified into multiple business types.

Network structure characteristics
This study examined which overseas cities invested in the upstream, midstream, and downstream businesses of the sub-industries of the companies.We formed upstream, midstream, and downstream urban networks, which were named R&D, production, and sales urban networks, respectively.The three urban networks involved multiple industry types and multiple central cities with differentiated diversity and multicentrality.The specific structural characteristics of each network are discussed below.

Leading sub-industries and investment destinations of R&D urban network
Fig. 3 shows that from 2010 to 2018, the number of subindustries establishing high-end businesses in overseas cities increased from two to eight.The ME&I industry had a relatively strong external connection ability, and the M&BP and M&N-M industries gradually increased their external connection capacity.In 2018, 123 overseas subsidiaries were set up by various sub-industries, accounting for 29.4% of the total overseas subsidiaries.The most popular overseas investment destination cities are education and scientific research centers or cities in developed countries.At the same time, many destination cities were alpha and beta level cities, including Hong Kong of China, Singapore, Perth, Wilmington, Virgin Islands, Trenton, Santiago, Washington, Tokyo, and Amsterdam.
Hong Kong of China is an important investment destination because of its world-renowned universities for scientific research and medical science.Singapore has a world-leading education level, and its scientific research capacity not only leads the Association of Southeast Asian Nations but is also on par with that of China and Japan.Perth, Australia, has five world-renowned universities with outstanding technological innovation capabilities.Wilmington, Trenton, and Washington in the United States have many colleges/universities, and Washington is an important cultural and educational center.Santiago, Chile, has three colleges/universities with strong scientific research capabilities, while Tokyo is Japan's educational and technological center.Amsterdam, Netherlands, is home to many research institutions.Many transnational companies establish R&D bases and product design centers in these cities.

Leading sub-industries and investment destinations of production urban network
Fig. 4 shows that from 2010 to 2018, the number of subindustries that established low-end businesses in overseas cities increased from five to eight.The ME&I industry had the strongest external connection ability, followed by the F&B industry.In 2018, 103 overseas subsidiaries were set up by various sub-industries, accounting for 24.6% of the total overseas subsidiaries.Most overseas investment destination cities are mainly regional centers in developing countries and some industrial centers in developed countries.At the same time, many were alpha-and beta-level cities, including Jakarta, Hanoi, Ho Chi Minh, Kuching, Dhaka, Phnom Penh, Manila, Mumbai, New Delhi, Lansing, Bangkok, Nursultan, Perth, Dusseldorf, Holzminden, Kleinfeld, and Derby.
Jakarta, Ho Chi Minh, Mumbai, Manila, and Lansing are all port cities. Hanoi, Kuching, Dhaka, New Delhi, Bangkok, Phnom Penh, and Nursultan are national or regional capitals.Perth in Western Australia, the center of global resources, attracts many transnational companies to manufacture resource-dependent products.Dusseldorf, Holzminden, Krefeld, and Derby are all important industrial cities in each region.These cities have a rich labor force and a solid production base; hence, many transnational companies operate in these cities to engage in production and manufacturing.

Leading sub-industries and investment destinations of sales urban network
Fig. 5 shows that from 2010 to 2018, the number of subindustries that established mid-end businesses in overseas cities remained at eight.The ME&I industry had a relatively strong external connection ability, while the M&N-M and P&P industries gradually increased their external connection capacity.In 2018, 280 overseas subsidiaries were set up by various sub-industries, accounting for 67% of the total overseas subsidiaries.The most popular overseas investment destination cities are global or regional administrative and economic centers, cit- Fig. 3 Upstream urban network of sub-industries of listed manufacturing companies in western China and destination cities from 2010 to 2018.Inv, investment; R&D, technology research and development; PD, product design; F&B, food and beverage; TC&F, textile, clothing, and fur; W&F, wood and furniture; P&P, petrochemical and plastic; Ele, electronics; M&N-M, metal and nonmetal; ME&I, mechanical equipment and instrument; M&BP, medicine and biological products.The larger the circle, the more overseas subsidiaries there are.Each color distinguishes the network formed by overseas investment in different types of sub-industries.The thicker the line, the more subsidiaries of that type of sub-industry invest overseas ies in developing or Belt and Road countries, and alphaand beta-level cities or cities close to them.Major destination cities include Hong Kong of China, Singapore, Washington DC, Moscow, Sao Paulo, Wilmington, Berne, Dusseldorf, Perth, Jakarta, Tokyo, Hanoi, Kuala Lumpur, New Delhi, Virgin Islands, Ontario, Dubai, Dublin, Kuching, and London.These cities have a high administrative level, a large population, a developed economy, a superior geographical position, and friendly political relations with China.Many transnational companies operate overseas subsidiaries in these cities to conduct sales, trade, and service activities.

Comparative analysis of network structure characteristics
The scale of the three networks increased from 2010 to 2018, with the sales urban network consistently being the largest (Table 2).The number of overseas subsidiar- Fig. 4 Midstream urban network of sub-industries of listed manufacturing companies in western China and destination cities from 2010 to 2018.P&M, production and manufacturing; F&B, food and beverage; TC&F, textile, clothing, and fur; W&F, wood and furniture; P&P, petrochemical and plastic; Ele, electronics; M&N-M, metal and nonmetal; ME&I, mechanical equipment and instrument; M&BP, medicine and biological products.The larger the circle, the more overseas subsidiaries there are.Each color distinguishes the network formed by overseas investment in different types of sub-industries.The thicker the line, the more subsidiaries of that type of subindustry invest overseas ies operating sales, trade, and service businesses was the largest.Meanwhile, the density of the R&D urban network decreased the most, which means that the number of overseas cities connected to the upstream link increased rapidly.The rapid increase in upstream link connections indicated that the overseas investment value chain of the listed manufacturing companies in western China gradually upgraded.
From degree centrality, the three networks showed the characteristics of a multicenter structure from 2010 to 2018.The sales urban network had the highest degree of centrality, indicating that this network had the strongest external connection ability.There were more important sub-industry and city nodes, and the characteristics of the multicenter structure were the most obvious.
From 2010 to 2018, the correlation was always network 3 > network 1 > network 2, indicating that the sales urban network had a relatively strong external connection ability and accessibility.Its power and information were dispersed; thus, individual nodes could not easily control it.By 2018, the production urban network had the largest least upper boundedness, indicating that power and information were concentrated in a few sub-industry and city nodes.The polarization phenomenon and hierarchical structure were evident, and the external connection ability was concentrated in a few nodes; the core nodes easily controlled it.
The three networks' modularity exceeded 0.45 from 2010 to 2018, and all exhibited a community structure, except in 2010 when the scale of the R&D urban network was miniscule while the community structure was unnoticeable.From 2010 to 2018, the modularity of the R&D urban network developed the fastest, increasing from 0.142 to 0.551.The modularity degree of the R&D urban network was the easiest to form specialized clusters.

Influencing factors
Through the QAP correlation analysis, we found a significant correlation between all the influencing factors and the outward investment propensity of the two leading industries in western China's manufacturing industry.On this basis, we further analyzed the degree of influence of each influencing factor in the R&D, production, and sales networks formed by foreign investment in the two leading industries of western China's manufacturing industry through QAP regression (Table 3).The absolute values of the regression coefficients of technological advantage (TA) and economic system advantage (ESA) in the three networks were relatively large, indicating that these factors greatly influenced the manufacturing industry's upstream, midstream, and downstream urban networks.TA had a significant negative influence, while ESA had a significant positive influence.Nevertheless, the three networks still had different key influencing factors.
The factors of the R&D urban network that exerted a relatively strong influence on the overseas investment tendency in 2018 were market potential advantage (MPA), TA, ESA, and the government's internalization advantage (GIA).ESA, MPA, and GIA had a significant positive influence on them.This shows that ESA is an important driving factor for the manufacturing industry in western China, establishing a high value-added business in the host country.The economic systems of the cities of the two countries directly affect the acquisition of sustainable profits from investment in both countries in terms of the flow form and allocation efficiency of production factors.Therefore, the economic system plays a significant role in promoting overseas investment.MPA also plays a significant positive role.The market potential of the cities of the two countries affects the high-tech sector of industrial outbound investment in terms of population and economy.The growth of population, GDP, and working-age population provides professional talents for the high-tech sector, thus promoting and attracting industrial investment, R&D, design, and other activities.In addition, a high GIA implies that the higher tariff rates of the home country's city increase financial revenue, thus improving the economic strength of overseas investment.The higher tariff rate of the host country's city decreases competition.Therefore, a city's higher tariff rate makes it easier to attract investments in new-technology businesses.Meanwhile, TA was found to exert a significant negative influence.This indicates that although the gov-ernments of western China cities spent more money on science and technology, most of it went into scientific research institutions, and listed manufacturing companies received less.Therefore, TA did not play a positive role in promoting overseas investments.In addition, the higher the global innovation index of the host country's city, the more technological competitors; thus, technological innovation also negatively influences overseas investments.TA, political system advantage (PSA), ESA, and GIA strongly influenced overseas investment in 2018 for the production urban network.PSA, ESA, and GIA had a significant positive influence.This indicated that when the manufacturing industry in western China established low value-added businesses in the host country, PSA became a key factor.The political system directly affects the stability and security of the political power of the cities of the two countries, enterprises' access to relevant information, and, consequently, enterprises' business environment and investment decisions.Therefore, foreign investment in the production and manufacturing links of the manufacturing industry in western China tends to fall in the host city with a superior political system.Meanwhile, ESA was found to play a significant positive role.The economic system mainly affects the convenience of the operation of enterprise activities, the way of market entry, and the transaction cost of entering the market from the economic aspect and thus im- pacts the investment behavior of both parties.In addition, the higher the tariff rate of the home country's city, the stronger the economic strength of overseas investments.The higher the tariff rate of the host country's city, the lesser the competition, in this case, it is easier to attract investments in production and manufacturing businesses.As for TA, it had a significant negative influence.Similar to that in the R&D urban network, listed manufacturing companies received less technological investment from the city, and the host city had greater technological competition.
Lastly, the factors of the sales urban network that exerted a greater influence on the overseas investment tendency in 2018 were geographical advantage (GA), TA, PSA, and ESA.Among them, ESA and PSA had a significant positive influence.Institutional factors mainly control the flow form and allocation efficiency of production factors in two countries from the security and convenience of various economic activities.They also ensure the expansion of supply channels and market scale, thus affecting the sales, trade, and service investment behavior of both sides and ultimately influencing the sustainability of foreign investment.The better the system is or, the more similar the cities are, the more conducive they are to the occurrence or attraction of investment.By contrast, TA and GA had a significant negative influence.The influence of TA was the same as that in the R&D and production urban networks because the manufacturing industry in western China did not receive sufficient technological investment, and the competition of the host country's city was large; thus, TA played a negative role.GA mainly influences the location choice of overseas investment from the location conditions.A negative GA means that when the manufacturing companies made overseas investments, the cities were not considered optimal locations.On the contrary, these well-located cities host many competitors, preventing them from becoming a major overseas investment destination for the downstream business of manufacturing companies.

Optimize the institutional investment environment to narrow the institutional gap
The government should pay attention to the optimization of the institutional environment.While pursuing globalization, the government should target investment countries according to the institutional risk preference and reduce their institutional differences.Bilateral investment treaties should be consolidated and developed.Signing bilateral investment treaties can reduce the institutional gap to a certain extent, especially for the host countries that have not signed but whose institutional environment is better than that of China.It can also help enterprises enter these countries and enjoy their good institutional environment.Moreover, it will make the investment between the two countries more convenient.At the same time, the government should attach significant importance to signing free trade agreements with some key strategic countries along 'Belt and One Road'.These key strategic countries, such as India, Indonesia, Turkey, and Kazakhstan, play a vital role in world economic development, and their existence cannot be ignored.If we can cooperate with these countries in a friendly way, the investment of western China in the countries along the line will be greatly improved.
Enterprises should fully understand the differences in institutional environments in different countries and better integrate into the local political and economic environment to reduce the uncertain costs of institutional distance and seek benefits.Therefore, enterprises in western China should pay attention to localization operations when investing abroad, that is, transnational enterprises do not regard themselves as invaders in the host market but as an individuals inherent in the target market.They should quickly integrate into the local environment, actively assume social responsibility, and obtain recognition from local consumers.To this end, enterprises can improve their employment concepts, business strategies, enterprise values, capital elements, and other aspects to establish a more suitable management model for the host country.For example, they may adhere to the localization of employment, only send a small number of directors and senior executives to participate in the management and decision-making of the core business of the investment enterprise, actively conduct public welfare projects, and work for the interests of local society and the people.

Adjust and optimize investment layout to expand overseas market
The government should strengthen the guidance of enterprise investment and adjust and optimize the investment layout.The investment motivation should be clari-fied to be prepared for and improve the probability of investment success.With the accelerated pace of 'going out' of emerging market enterprises, Chinese enterprises, as the largest emerging market, are frequently involved in vicious competition, blind investment, blind decision making, illegal operation, labor disputes, product quality problems, etc.These issues have led to negative international influence among Chinese enterprises.Therefore, the government should guide enterprises to 'go global'.First, it should establish policies and regulations to avoid blind investment and disregard for laws and regulations as well as brands and images.Second, the government should guide enterprises in entering the overseas market through enterprise alliances and group investment, especially the overseas investment of small and medium-sized enterprises, which needs to be protected and supported.
Enterprises invest abroad to occupy overseas markets.When overseas markets are large, enterprises can enjoy lower transportation costs and avoid trade barriers.Therefore, enterprises should choose appropriate foreign investment models and paths to optimize the global layout.Enterprises with monopoly technology advantages should first export, establish brands in overseas markets, and then start greenfield investment.On the one hand, it can reduce production costs; on the other hand, it can conduct localized production and operation of products.For enterprises lacking competitive advantage, if they have a relatively broad home market and their capital is strong enough, and if the market demand and environment of the host country are well understood, these enterprises can acquire the overseas market through mergers and acquisitions.

Increase investment in R&D funds to promote technological progress
The manufacturing industry in western China has long been dominated by imitation with low technology content.A large amount of sustained R&D funding is needed to achieve technological innovation and promote technological progress.First, the government should increase financial support for enterprise technology R&D, establish technology innovation subsidies, and encourage enterprises to carry out technology R&D and innovation.With these approaches, enterprises can reduce costs while engaging heavily in technology R&D.Owing to the risk of technology R&D, enterprise funds tend to increase their production costs.To reduce the risk, an R&D insurance system should be established.Second, the government should improve the technical standard system, proceed from the actual situation of the manufacturing industry, refer to the current international technical standards and technical regulations of developed countries, break industry restrictions, expand the scope of technical regulations and standards to the front and back value chains and related products, and be prepared to adjust technical standards at any time.
As technology R&D usually takes a long time and the results are uncertain, enterprises should go beyond government funding and expand R&D financing channels.Enterprises can choose appropriate financing channels according to their actual conditions and finance through traditional means such as loans, bond issuance, and financial leasing.Owing to the low scientific and technological content of manufacturing products in western China, the concept of technology and product R&D must be established under intensified competition, rising costs, and weak external markets.In terms of raw materials, enterprises should strengthen the research, development, and design of core raw materials; control product quality from the source; and innovate products.In production, enterprises should actively promote independent technological innovation and apply the results to product production.Finally, enterprises should actively make use of international trade, international investment, and the division of labor in the GVC; strengthen cooperation with enterprises in the upstream and downstream links of the value chain; learn from foreign advanced technologies and production processes; improve the ability of technologically independent innovation; and improve the status of the manufacturing industry in western China.

Discussion
With the further deepening role of global integrated production in the urban space, many late-developing cities are becoming linked to cross-border urban networks through the GVC, and the manufacturing GVC plays a vital role in building urban relations in the WCN (Li, 2008;Xue and Huang, 2013).This study investigated the cross-border urban networks (WCN formed by crossborder investment) of manufacturing listed companies in western China through the GVC.The work offered certain implications in relation to previous studies on China's manufacturing cross-border networks (Yang, 2013;Zhu, 2018;Ma, 2019;Yuan, 2021).For example, it constructed a relationship framework of cross-border urban networks between sub-industries and overseas cities from the GVC perspective.This study also enriched the research on cross-border urban networks in China's vast inland areas, including their paths, thus benefiting China's balanced development.In addition, this study divided the manufacturing industry into detailed sub-industries and refines the related investigation.Hence, the results laid a foundation for expanding foreign investment in various industries in western China.
In terms of the relationship framework.Although GVC and WCN are aimed at building a theoretical system on a global scale, the current research, at many levels, is independent and rarely intersects (Zhou, 2015;Wang and Xu, 2017).The relationship framework constructed herein implies that the GVC and cross-border urban networks have integrated development (Fig. 1) and are developed around enterprises (especially multinational corporations).The GVC is an important way for enterprises to participate in global economic activities.Cities are the carriers for enterprises to participate in global economic activities, and the development of urban networking is the change in the spatial organization structure of carriers.Specifically, the GVC divides the production activities of enterprises into upstream, midstream, and downstream value links according to the international division of labor.Then, different value links choose appropriate urban locations, and the GVC, under the guidance of this spatial division of labor, is accompanied by agglomeration in different urban spaces around the world, thus forming a world urban network.The research on integrating GVC and WCN is a frontier, difficult, and emerging interdisciplinary research field (Derudder and Witlox, 2010;Sassen, 2010;Wang and Xu, 2017).The proposed relationship framework also proves that the value distribution process of crossborder urban networks under the functional division system of the value chain has made the GVC theory applicable to transnational corporations' international division of labor.It also shows that this theory has become integrated into the cross-border urban networks dominated by transnational corporations.
In terms of research content and results.First, this study explores the urban network formed by cross-border manufacturing investment in cities in western China (underdeveloped regions) from the perspective of the GVC.This research content is derived from some scholars' doubts and criticisms regarding relevant research defects.An increasing number of scholars believe that existing studies pay too much attention to the role of international metropolises, advanced producer services, and multinational corporate headquarters in the construction of the WCN, thus ignoring the 'Third World Cities' (Robinson, 2002;Wu, 2009;Coe et al., 2010;Xue and Huang, 2013).Second, although the results show that the multicenter network characteristics are similar to previous research on cross-border networks in eastern China (Xu and Li, 2020), the multicenter network characteristics of cross-border networks in western China are new and relatively simple.In terms of overseas investment destination cities, the proportion of investment destination cities in eastern China investing in western Europe and North America is relatively high (Li, 2012), while the overseas investment cities in western China are more inclined toward cities in developing countries.Similar to the research results on advanced producer services (Liu et al., 2021), the R&D urban network will likely form professional clusters.However, the difference is that the density of the R&D urban network in advanced producer services is high, while the density of the R&D urban network in manufacturing is low, indicating that the external investment capacity of the R&D urban network in manufacturing is still weak.Western China has a vast area that eastern China industries can utilize for expansion.However, many enterprises in western China are resource-dependent, and the resources have begun to diminish.Therefore, opening up western China's economy to the outside world is an important challenge for the region.The policy implications proposed in this study are helpful for cities in western China to scientifically choose the realistic globalization path.Although certain limitations exist, this study also reveals the development path of overseas investments in western China.
The limitations of this study include the following: we only selected listed companies as research cases, and this selection has a certain significance.However, the overseas investment of small and medium-sized enterprises is also important for developing western regions.In the future, small and medium-sized enterprises must still be included in the study for comparative analysis.
In addition, the urban globalization network in western China is characterized by a dynamic process.Owing to the limited data for overseas cities, the latest data in this study were only from 2018.Therefore, we could not analyze the development path of outbound investment in cities in western China after the COVID-19 outbreak.Subsequent research can further expand the timeframe and describe the network's dynamic change process by considering the factors brought about by the sudden pandemic.

Conclusions
To explore the overseas investment path of late-developing cities, this study investigated the overseas investment network of the manufacturing industry in western China from the perspective of the GVC.The results of this study showed that in terms of structure, the three urban networks involved multiple industry types and multiple central cities.In the R&D urban network, overseas investment sub-industries were dominated by the ME&I, M&BP, and M&N-M industries.Overseas investment destinations were mostly educational and scientific research center cities.In the production urban network, most overseas investment sub-industries were ME&I and F&B, and overseas investment destinations were mostly regional central cities in developing countries.In the sales urban network, overseas investment sub-industries were mostly ME&I, M&N-M, and P&P; overseas investment destinations were diverse and spread out.In terms of network structure characteristics, the sales urban network had the highest centrality and correlation and could not easily be controlled by individual nodes.The production urban network had the largest least upper boundedness and could easily be controlled and dominated by core nodes.The R&D urban network's modularity also made it easy to form specialized clusters.In terms of influencing factors, TA and ESA had a major influence on the upstream, midstream, and downstream networks.TA had a significant negative influence, while ESA had a significant positive influence.Nevertheless, the three types of urban networks still had different influencing factors.According to the different influencing factors, the government and enterprises can promote overseas investment in western Chinese cities by optimizing the institutional investment environment to narrow the institutional gap, ad-justing and optimizing the investment layout to expand the overseas market and increasing investment in R&D funds to promote technological progress.To improve the sustainability of overseas investment in late-developing cities, extensive theoretical and experimental work should be conducted in the future.

Fig. 2
Fig. 2 Influencing factors and overseas investment network mechanisms

Table 1
Index system of influencing factors of overseas investment tendency

Table 2
Characteristics of urban networks formed by cross-border manufacturing investment in western China Notes: N1 stands for R&D urban network, N2 stands for production urban network, N3 stands for sales urban network

Table 3
Regression coefficients of influencing factors of overseas investment tendency ** and ** represent 1% and 5% significance levels.AA is an asset advantage, GA is a geographical advantage, MPA is a market potential advantage, GNA is geographical proximity advantage, RNA is a relational proximity advantage, TA is a technical advantage, PSA is a political system advantage, ESA is an economic system advantage, CIA is the company's internalization advantage, GIA is government's internalization advantage *