The proximity and dynamics of intercity technology transfers in the Guangdong–Hong Kong–Macau Greater Bay Area: Evidence from patent transfer networks

Abstract

The patent transfer provides an important indication of technology flows and knowledge diffusion across space. Drawing on patent transfer data, we modeled intercity technology transfer networks in the Guangdong–Hong Kong–Macau Greater Bay Area, a city region special for its “one country, two systems” structure, in the periods 2007–2011 and 2012–2016. We then explored the evolutionary characteristics of the networks and further examined the impact of, and interaction between, different forms of proximities in relation to technology transfer over time. Our results show that some kinds of proximities (institutional, cognitive, and social) are able to promote technology transfers, while others (geographical and cultural) do not exert significant impacts. Of the latter category, geographical proximity can, however, indirectly affect technology transfer by acting on the proximity of other dimensions (institutional, cognitive, and social). For instance, cognitive proximity can compensate for the lack of geographical proximity and social proximity frequently accompanies geographical proximity—and both relationships are reinforced over time. In contrast, the interrelatedness of geographical and institutional proximities have transformed from a relation of substitution to complementarity.

Keywords

Technology transfer intercity network patent transactions proximity and dynamics the Greater Bay Area

Introduction

As technology transfers between cities intensify with the rise of the knowledge economy, the spatial dimension of innovation—and in particular, the role that cities play in innovation networks—is changing (Hannigan et al., 2015; Maggioni et al., 2017). “Technology transfer” describes the process whereby, due to the existence of a technology gap or complementarity, a technology supplier transfers technology and relevant rights to another party (the transferee), who then transforms that technology into productivity through its absorption and application (Teece, 1977). Because technology transfer can break resistant geographical boundaries and facilitate the realization of cross-regional flows of innovation elements, transfers have become an important source of external technology for countries, regions, and cities (Ferraro and Iovanella, 2017). Due to the uneven distribution of scientific and technological resources between cities, technology transfer plays a critical role in stimulating knowledge spillovers and redistributing innovative elements (Aghion et al., 1997; Eaton and Kortum, 1996).

Alongside scientific knowledge, knowledge can also encompass types of knowledge that are related to enterprises, including technological knowledge (Karlsson et al., 2006). Patents are commonly perceived as a codified proxy for tacit technological knowledge in general and technology transfer in particular (Buenstorf and Geissler, 2012; Krugman, 1991). The process of a patent transaction includes discovery, disclosure, evaluation, protection, marketing, negotiation, and licensing, as well as development and commercialization. These processes are carried out by individuals, enterprises, or research institutions, and these actors (like the transactions themselves) are embedded in and influenced by local spaces. Departing from a more conventional focus on actors’ performance in patent transfer networks, or network impacts on innovation capacity and productivity at the individual or organizational levels, a growing body of empirical analysis has been conducted at the city and region levels, and a scale of analysis that has yielded new insights into the spatial knowledge diffusion (Bednarz and Broekel, 2019; Liu et al., 2019; Maggioni et al., 2011). In this paper, we examine intercity patent transfer flows in order to explore the patterns and evolution of regional technology diffusion.

Technology does not diffuse frictionlessly, and it is widely recognized that economic activity is not simple “purely economic” but constitutes an instituted process and a socially embedded activity (Amin, 1999; Thrift and Olds, 1996). The patent transfer is a kind of transaction behavior that is formed by micro-entities and based on market rules; through repeated instances of such behavior, a social network based on transaction relations is formed between assignors and assignees. As an inherently sociocultural activity, the patent transfer is dependent on the institutional settings within which it takes place (Martin, 2000). Other factors such as geographical, cognitive, organizational, social, and cultural distances can also potentially play a role, which is often unequal and selective, in relation to technology diffusion across space (Bednarz and Broekel, 2019; Boschma, 2005; Torre and Rallet, 2005). However, despite a great deal of attention that has been paid to multidimensional proximities in spatial knowledge diffusion, the issue of institutional proximity has not yet been sufficiently examined. Most existing studies have rather chosen to focus on the intercity knowledge networks of a city region or a single country, categorizing all city entities as an identical institution. In addition, the multidimensional proximities that are examined in relation to different kinds of intercity knowledge flows are frequently assumed to be static and independent on each other. Notwithstanding an increasing awareness of the interrelatedness of proximities, changes of this interrelatedness remain underreported (Broekel, 2015; Cao et al., 2019; Tel Wal, 2014).

This paper seeks to fill the gap by unfolding the interrelated and dynamic nature of multidimensional proximities within the diffusion of spatial technology. Following the work of Duan et al. (2018, 2019), we used patent transfer as an indication of technology transfer, examining intercity patent transfer networks in the Guangdong–Hong Kong–Macau Greater Bay Area (GBA), China, for the periods of 2007–2011 and 2012–2016. This city region is characterized by the “one country, two systems” structure, making it a fertile ground for exploring the effect of institutional proximity. The GBA is now drawing a great deal of interest as an innovation hotspot in China and throughout the world (Liu et al., 2019; Ma et al., 2020). This marks a departure from the “front shop, back factory” export-oriented growth model that was previously invoked in describing the relationship between Hong Kong and the Pearl River Delta (PRD) (Meyer et al., 2012; Shen, 2008). With the largest number of patent applications in China, the GBA is the most active region for innovation in the country—in 2018, it produced 330,832 innovation patents, accounting for 21% of the national total. Recognizing the region's importance, in 2019, the Chinese Government released the Outline Development Plan for the Guangdong-Hong Kong-Macau Greater Bay Area to promote the GBA as a global center of scientific and technological innovation. Given this development history, the question of how intercity technology transfer networks are structured and evolving in this region, as well as their underlying determinants, is of great importance and further studies are urgently needed. Our study responds to this challenge.

The remainder of this paper is arranged as follows. The next section reviews the literature on multidimensional proximities and their interactions in technology transfers. Section 3 introduces the data and methodologies used in the empirical analysis, after which the results are discussed in the fourth section. Finally, Section 5 outlines our concluding remarks and proposes potential directions for future research.

Literature review

Urban endowment and technology transfer

Technology transfer has become a key focus in innovation geography and urban studies in the context of a globalized knowledge economy (Kratke, 2010; Van Egeraat and Kogler, 2013). The dynamics of technology transfer are complex: transfers are not only sensitive to the age, generality, and historical trade of the patents themselves (Serrano, 2010), but are influenced by the willingness of both transferring parties and transaction costs (Teece, 1977). Moreover, they are also affected by the capacity of third-party technology transaction services such as patent information service platforms. While patent inventors and applicants are the subjects of a patent transfer, these innovation entities and their transactional behaviors are constrained by their local space. The aggregation of patent transfer flows between actors or organizations on the micro scale into spatial flows between cities or regions on the macro scale constitutes an appropriate measure to reflect spatial technology diffusion and spillover (Serrano, 2010) since the data can explicitly visualize the supply–demand relations that characterize technology transfer markets and avoid political bias (Gambardella et al., 2007).

Existing studies addressing the innovation performance of cities or regions have taken into account a wide range of urban or regional characteristics in order to approximate the study region's knowledge or technology endowment. These characteristics include government research and development (R&D) investment, enterprise R&D investment, the level of economic development, innovation ability, administrative hierarchy, human capital, the industrialization level, scientific and technological service ability, regional status, the level of openness on technology transfer, etc. (Wei, 2015; Wei et al., 2011). The patent transfer is a bidirectional behavior: on the one hand, it can be a demand-side led behavior, where the recipient city takes the initiative to search for patents that meet its own interests and development needs in a patent bank or on a patent trading platform; on the other hand, such transfers can constitute a supply-side dominated activity, where a patentee in a specific city searches for the demand side in order to obtain certain commercial benefits. The transaction willingness of the transferor and transferee might be affected by the same factors to different extents. Existing research has, however, had difficulty in distinguishing the influence of various factors on cities that transfer patents in and cities that transfer patents out. A city's technological invention ability and spillover incentives affect outbound patent transfers, just as a city's technological transfer capacity and absorption ability influence patent absorption. Therefore, the attribute characteristics of the origin city, the destination city, and the relationship between them all need to be considered in the analysis of a directed transfer network.

Intercity proximity and technology transfer

Among the various factors that might contribute to technology flows and innovation spillover, the proximity concept has become a powerful framework to explore the mechanisms underlying network formation and evolution. Inspired by the French school of proximity dynamics, geographical proximity is assumed to be the key force for knowledge exchanges (Torre and Rallet, 2005). Geographical proximity is beneficial to face-to-face interactions and builds mutual trust between actors, which is necessary for tacit knowledge to spread and be transformed into codified knowledge (Hagerstrand, 1952; Howells, 2002). In addition to geographical proximity, Boschma (2005) proposed four distinct forms of proximities: cognitive, social, organizational, and institutional. Empirical evidence confirms that these four types of proximities act as significant drivers of intercity knowledge networks in a range of processes, including patent citation (Ter Wal, 2014), scientific collaboration (Gui et al., 2018), joint R&D project (Bednarz and Broekel, 2019), and patent transfer (Duan et al., 2018, 2019; Liu et al., 2019).

Institutions maintain a rich relation to technology transfer. Institutional proximity refers to the extent to which innovation actors are embedded in similar institutional regulations such as habits, routines, established practices, and incentive structures (Hoekman et al., 2009). Innovation actors with the same institutional environment are able to foster basic trust levels and reduce transaction costs by standardizing interactions (Lagendijk and Lorentzen, 2007). Since technology transfer involves various forms of interaction costs including negotiation, contract formation, and information exchange, institutional proximity could have a substantial impact on the technology transfer process (Caviggioli and Ughetto, 2013).

Similar conclusions can be drawn when innovation actors are embedded in the same organization. Organizational proximity refers to the extent to which innovation actors belong to similar organizational environments. This is a control-related dimension, which addresses business routines, hierarchies, and value systems. Relations with higher organizational proximity—e.g. actors being at the same institutes (Balland, 2012) or cities being in the same province (Li and Phelps, 2017)—tend to have lower levels of uncertainty and opportunism, leading to stronger knowledge diffusion. Successful knowledge diffusion also depends on senders’ intentions and their capacity to communicate, as well as recipients’ willingness and capacity to absorb (Cohen and Levinthal, 1990). Cognitive proximity, also called “technological proximity” when discussed in relation to technology transfer (Knoben and Oerlemans, 2006), reflects the degree of overlap between two actors with respect to their technical experiences, communication language, and knowledge bases. Similarities in knowledge backgrounds are required for understanding specialized technical languages, concepts, and methods (Mancusi, 2008); as such, technological proximity can facilitate effective learning and absorption of new knowledge in the process of patent transfer. However, clusters that become too specialized can be detrimental to knowledge spillover, because knowledge bases that become too similar can result in technological lock-in (Boschma, 2005; Nooteboom, 2000). Social proximity—a term which refers to the relational embeddedness of actors in terms of the partnership, kinship, and friendship (Boschma, 2005; Granovetter, 1985)—can also act as an explanatory factor in relation to technology diffusion. Multilevel social relations in other fields can strengthen trust among actors and serve as direct or indirect channels in knowledge diffusion (Feldman and Massard, 2002; Storper and Venables, 2004). Because the spatial patent transfer is built on interpersonal social networks, the transaction willingness of the transferor and the transferee can be quite path dependent.

Nevertheless, some scholars have also argued that geographical proximity per se is neither a necessary nor sufficient condition for interaction and innovation to take place (Boschma, 2005; Malecki and Oinas, 1999). It may, however, indirectly affect network relations by acting on the proximity of other dimensions. A study by Cao et al. (2019) noted that relatedness (either in the form of substitution or complementarity) between geographical and non-geographical proximities can variously depend on the spatiotemporal scales of innovation activities, the characteristics of innovation actors, and the types of innovation networks involved. In addition to being interrelated, the impact of these proximities on network evolution may not be constant. For instance, Balland et al. (2013) stated that geographical and cognitive proximity became more relevant with the development and maturation of particular industries. Furthermore, current proximity structures may influence the future collaboration behaviors of actors (Balland et al., 2015). Broekel (2015) explicitly confirmed the existence of systematic and dynamic interrelatedness between proximities in the evolution of 280 networks of subsidized R&D collaboration in Germany. Building on this work, this paper contributes to the existing literature by revealing how different types of proximities have shaped the structure of intercity technology transfer networks in the GBA and showing how the influence and interrelatedness of these proximities have changed over time.

Data and methodologies

Study area

The GBA is composed of 11 cities and includes 2 special administrative regions, Hong Kong and Macau, and 9 prefecture-level cities in Guangdong Province, previously known as the PRD. Guangzhou is the provincial capital city and Shenzhen is a sub-provincial city (Figure 1). The PRD has close geographical and business relations, as well as a high level of population migration and mobility with Hong Kong and Macau, which makes these cities form a sub-national cross-border urban area (Yang, 2005). Unlike other world-class urban agglomerations, the most typical feature of the GBA is its heterogeneity, which is most reflected in its unique “one country, two systems, three tariff zones” management structure (Liu et al., 2019). Although Hong Kong, Macau, and the PRD belong to the same sovereign state, due to the long-term administrative division barriers, they have different political, legal, and administrative systems, and the flows and distribution of innovative resources between them are uneven, which poses obstacles to cross-regional and intercity collaborative technological innovation activities (Chen and Xie, 2018; Meyer et al., 2012). For these reasons, the GBA is encouraged to promote integrated and coordinated development and facilitate the spatial reconfiguration of resources, especially innovation elements.

Figure 1.

Cities of different levels in the Guangdong–Hong Kong–Macau Greater Bay Area (GBA).

Construction of intercity technology transfer networks

Patents are an important output of technological development, not only as a major indicator of regional innovation capacity, but as an effective vehicle to obtain external knowledge spillover (Ferraro and Iovanella, 2017). Patents are transferable assets, which, by the early 20th century, had made it possible to separate the person who invents patents from the one who commercializes them. This development recognized the fact that someone who is good at coming up with ideas is not necessarily the best person to bring those ideas to market. These records provide information on the inventor–applicant relationship (or origin and destination) of patent transfer flows. In contrast to patent transfer, patent collaboration is to a large extent limited to actors in the same institution due to the exclusivity of intellectual property, thus making collaborative patents count for quite a small proportion of all patent applications (Duan et al., 2018). In this sense, patent transfer data is preferred as a means to depict and explore technology flows and diffusion across regions or cities (Liu et al., 2019; Maggioni et al., 2011).

It is important to acknowledge a number of issues and limitations in relation to the use of patent transfer data. First, the patent transfer may only account for technological knowledge flows that are encodable, commercially exploitable, and legally patentable (Criscuolo and Verspagen, 2008). As such, not all technological knowledge ends up being patented and the propensity to patent differs greatly across industries and technology fields (Bednarz and Broekel, 2019). This somewhat confirms the validity of our focus on cities, as it aggregates transferred patents across specific industries and fields. While patents capture only a portion of the technological knowledge—and correspondingly, patent transfer reflects only a fraction of technology flows and knowledge diffusion across space—it has been argued that these issues are of smaller relevance at the regional level and patent transfer would still be a practical indicator of possible spillover in many cases (Paci and Usai, 2009; Serrano, 2010).

The data on patent transactions used in this study is taken from the China National Intellectual Property Administration (http://www.cnipa.gov.cn/), which offers detailed information in the form of the patent name, type, cooperative patent classification (CPC), transferor, transferee, and the cities in which they reside. Based on this information, the patent transfer flows between 11 cities of the GBA were collected for each year from 2006 to 2016, as 2006 was the year in which a number of innovative development strategies were initiated and the GBA witnessed a substantial improvement in the subsequent 10 years. We removed all intra-city transfer flows, as our interest lies in the intercity technology diffusion. Moreover, we employed a 5-year moving window for the patent transaction counts to control for yearly fluctuations, which is also consistent with the literature, reflecting the fact that most patents lose their economic impact outside of this timeframe (Griliches, 1979). Consequently, we modeled weighted and directed intercity patent transfer networks for the periods 2007–2011 and 2012–2016; these formed the basis for the later analysis.

Specification of model and variables

To explore the characteristics of intercity patent transfer networks, social network analytical indicators such as density of the network and degree centrality of cities are calculated as the following equations:

D = \frac{m}{[n \times (n - 1)]}

(1)

C_{a} = \sum_{i = 1}^{n} R_{a i}

(2)

where D is the network density, m is the actual number of connections between cities in the GBA, n is the number of cities, C_a is the degree centrality of city a, and R_ai is the total patents transferred between city a and city i. Therefore, C_a is the sum of out-degree and in-degree of city a with the former denoting the total patents transferred from city a to all the other cities and the latter denoting the total patents transferred to city a from all the other cities.

Building on the conceptual gravity model, a more general form can be specified, in which patent transfer between cities is hypothesized to be dependent on the urban technology or knowledge endowment (which at the origin generates flows and at the destination receives flows), as well as on the multiple forms of intercity proximities. Since these proximities are neither independent on nor isolated from each other, the interactive terms of geographical and non-geographical proximities should be considered. The specific form of the regression model can be expressed through the following formula:

F_{i j} = α + \sum_{q ϵ Q} β^{q} O_{i t}^{q} + \sum_{r ϵ R} γ^{r} D_{t j}^{r} + \sum_{p ϵ P} θ^{P} R_{i j}^{P} + \sum_{s ϵ S} η^{S} I_{i j}^{S} + υ_{i j, t}

(3)

where

F_{i j}

is the number of patents transferred from city i to j after taking the logarithm;

O_{i t}^{q}

and

D_{j t}^{r}

, respectively, represent the set of variables that reflect the technology endowment of origin city i and destination city j, which examine the scale effects of the characteristic variables in bilateral intercity patent transfer.

R_{i j}^{P}

is the variable set of multidimensional proximities between cities i and j,

I_{i j}^{S}

is the corresponding interactive set between geographical and non-geographical proximities.

β^{q}

γ^{r}

θ^{P}

, and

η^{S}

are coefficients to be estimated. In order to eliminate the effect of heteroscedasticity, the explanatory variables were converted into natural logarithms. In practice, one was added to each of the actual values before the log transformation to deal with cases where the data are zero.

Following Cantner and Graf (2006), Broekel and Boschma (2012), and Broekel et al. (2014), the multiple regression quadratic assignment procedure (MRQAP) was employed to perform the regression analysis. The reason for using this approach reflects two considerations: (i) it incorporates relational variables and considers their inherent interdependence when assessing their statistical relevance (Krackhardt, 1988) and (ii) it allows for categorical data without violating distribution assumptions (Zhang et al., 2020). In the MRQAP, all variables were represented in the form of a 11 × 11-city square matrix with the diagonal value being zero. The statistical calculations were performed in the UCINET software. To estimate regression coefficients, the double-semi-partialling approach was preferred due to its relative robustness in the presence of multicollinearity (Dekker et al., 2003).

Urban technology endowment was represented by total patent applications of origin (PATO) and destination (PATD), the governmental expenditure on education and technology of origin (EETO) and destination (EETD), and the administrative level (ADM) of cities. According to Andersson et al. (2014) and Cao et al. (2018), the top-down administrative system of China is a crucial factor that affects knowledge diffusion, since cities with higher administrative levels are endowed with more resources and better policies. In the GBA, Hong Kong, and Macau are special administrative regions, Guangzhou is the provincial capital, and Shenzhen is a sub-provincial city; together, these cities constitute a high-level administrative community. The categorical variable ADM was designed to reflect this structure; it was set at 2 for transfers where both cities were part of the high-level administrative community, 1 for transfers where one of the cities was from this community, and 0 for other transfers. Other factors such as gross domestic product per capita, foreign direct investment, the number of universities, and the share of second and tertiary industry were also taken into account at the beginning, but eliminated in the regression because of their strong linearity with PAT, EET, and ADM.

In terms of intercity proximities, our proximity framework included geographical, institutional, cultural, cognitive, and social dimensions. Organizational proximity was excluded from the study because when patents are transferred between actors from different organizations (i.e. different universities, institutions, enterprises, etc.), this data is aggregated at the city level, and the original organization (affiliation) information is disregarded. Instead, the “affiliation” of a city would be the province or region where it is situated. Further, since cities in the same province or city region are often subject to the same institutional framework at the macro level, the organizational proximity of two cities is intertwined with their institutional proximity (Broekel, 2015). Geographical proximity (GEO) is specified by the reciprocal of Euclidean distance between the centroids of cities i and j in kilometers standardized by the maximum distance. Institutional proximity (INS) is a dyadic dummy variable with cities both in mainland China or both in Hong Kong and Macau being 1, and 0 otherwise. Cultural proximity (CUL), literally an informal institutional proximity, is also a dyadic dummy variable in which city pairs sharing a dialect are 1, and 0 otherwise, based on the 2010 Atlas of Chinese Dialects (Xiong and Zhang, 2012). According to Jaffe (1986), cognitive proximity (COG) is estimated by the cosine similarity between cities’ technological profiles. For each of the 11 cities in the GBA, we counted the applied patents during 2007–2011 and 2012–2016 in each CPC subclass, distinguished by a three-digit code resulting in an 11 (cities) × 122 (subclasses) matrix. Based on these, the intercity technological relatedness is calculated by cosine similarity, which serves as a cognitive or technological proximity measure. Social proximity (SOC) is specified as the number of patents transferred between cities in the previous year. Therefore, the SOC matrix for 2007–2011 shows the intercity patent transfer network in 2006, and the matrix for 2012–2016 shows the network in 2011.

Following Cao et al. (2019), we tested the interactive effects between geographical proximity and the four kinds of non-geographical proximities—i.e. geographical proximity × institutional proximity (DISINS), geographical proximity × cognitive proximity (DISCOG), geographical proximity × social proximity (DISSOC), geographical proximity × cultural proximity (DISCUL).

Results and discussion

Evolutionary characteristics of intercity technology transfer networks

Figure 2(a) and 2(b) respectively shows the spatial patterns of intercity patent transfer flows in the GBA for the periods 2007–2011 and 2012–2016 (referred to as “2007” and “2012” in the following discussion). The directed flows are weighted proportional to the total patents transferred from one city to another, and the node size is proportional to the degree centrality of a city which is the sum of patents transferred in and out. Several main features can be discerned in the structure and evolution of intercity patent transfer networks. First, it is obvious that the network grew dramatically from 2007 to 2012, with density rising from 0.66 to 0.83 and the average strength of intercity flows increasing from 12.38 to 73.05. In addition, the reciprocity of bidirectional patent transfer flows enhanced considerably between 2007 and 2012, with the symmetry of the overall network ascending from 0.65 to 0.79, making the GBA the leading region for technological inventions in China (Ma et al., 2020).

Figure 2.

Intercity patent transfer networks in the Guangdong–Hong Kong–Macau Greater Bay Area (GBA) during 2007–2011 (a) and 2012–2016 (b).

Second, a large number of patent transactions occurred across different city levels within the GBA. The total number of patents transferred from high-level cities to ordinary cities counted for 43.91% in 2007 and 45.08% in 2012, whereas those transferred in the opposite direction counted for 20.04% in 2007 and 24.36% in 2012, suggesting a greater growth. Although the patents transferred between cities at the same administrative levels grew from 491 to 2456 in number, the proportion declined from 36.05% to 30.57%, implying that a hierarchical discrepancy between cities increasingly contributed to technology diffusion. The technology transfer between a number of prefecture-level cities (e.g. Dongguan, Zhongshan, Foshan) with high-level cities (e.g. Guangzhou and Shenzhen) was shown to be very frequent. The high-level special administrative regions of Hong Kong and Macau were also observed to have established relatively unidirectional technology links to many prefecture-level cities in the PRD, e.g. Dongguan, Zhongshan, Zhuhai, Jiangmen, and Huizhou.

Third, it seems that there was no evident distance decay effect in relation to patent transfer flows in the GBA. As seen in Figure 3, the number of patents transferred reached its peak between cities within distances of 60–70 km and 100–110 km during both time periods. When the distance between cities is larger than 110, distance makes sense. However, the institutional barriers caused by political system differences were discovered to significantly restrict technology transfer between Hong Kong, Macau, and cities in the mainland, i.e. the technological flows between cities in the PRD were much stronger than those between the mainland and Hong Kong and Macau.

Figure 3.

The relationship between the number and distance of transferred patents during 2007–2011 and 2012–2016.

In terms of node status in the network, we first note that Hong Kong and Macau had a significantly lower centrality in the network and also demonstrated an almost unidirectional technical output trend (Table 1). The marginalized status of Hong Kong and Macau in the technology transfer network is not consistent with the dominant position of these cities with respect to their economic development, knowledge bases, and so forth, indicating that there is still a lot of room to rise for Hong Kong's participation and role in the coming round of technological innovation development in the GBA. Another feature of the network is that Hong Kong and Macau were characterized predominantly by patent transfer but less by patent absorption, a situation which, on the one hand, somewhat reflects the relatively fewer industries and subsequently lower demand for patents. On the other hand, due to the fact that the patent transfer between Hong Kong, Macau, and the PRD needs to follow technology export/import regulations and procedures, the different institutional frameworks could have a far-reaching impact. We also note an evident core–periphery structure in the regional technology transfer networks, with Shenzhen and Guangzhou being the two technology centers in patent transfer flows and accounting for half of the total flows within the GBA during study periods. As a result of Shenzhen's excellent enterprise innovation ability and Guangzhou's abundant basic scientific resources, the two cities have jointly promoted the technological innovation and application ability of the entire GBA. Compared with Hong Kong and Macau, prefecture-level cities on the mainland side were found to become more actively integrated into the GBA technology transfer network; Dongguan and Foshan, located close to Shenzhen and Guangzhou, emerged as the new stars of the network in recent years.

Table 1.

Urban degree centrality.

City	2007–2011			2012–2016
City	Out-degree	In-degree	Total degree	Out-degree	In-degree	Total degree
Shenzhen	457	292	749	3297	1515	4812
Guangzhou	263	220	483	1346	1457	2803
Dongguan	141	332	473	907	1540	2447
Hong Kong	178	79	257	369	153	522
Foshan	97	137	234	837	788	1625
Zhongshan	104	89	193	398	694	1092
Zhuhai	54	64	118	217	328	545
Huizhou	30	84	114	371	901	1272
Zhaoqing	4	39	43	221	174	395
Jiangmen	14	24	38	167	358	525
Macau	20	2	22	25	5	32

Proximity and dynamics underlying the intercity technology transfer

To reveal the proximity and dynamics underlying the intercity patent transfer networks, we ran MRQAP for the data for 2007–2011 and 2012–2016. Model (1) only considered variables of urban knowledge endowment and intercity proximities. When the interactive terms of geographical and non-geographical proximities were added, model (2) confirmed a robust result. Table 2 summarizes the regression results, which explain how urban knowledge endowment and intercity proximities (and their interactions) affected technology diffusion in the GBA. A comparison of results for 2007–2011 and 2012–2016 could shed light on the dynamics of these impacts.

Table 2.

MRQAP results for the intercity patent transfer networks during 2007–2011 and 2012–2016.

Variables	2007–2011		2012–2016
Variables	(1)	(2)	(1)	(2)
Intercept	0.000***	0.000***	0.000***	0.000***
ADM	0.164*	0.137*	0.158*	0.121*
PATO	0.253***	0.249***	0.181**	0.152*
PATD	0.222***	0.218***	0.075**	0.082***
EETO	0.214*	0.187*	0.164***	0.141**
EETD	0.356**	0.262**	0.257***	0.174***
Proximities
DIS	0.054	0.127	0.017	0.257
INS	0.211*	0.175***	0.193**	0.131*
COG	0.114**	0.149*	0.149**	0.181*
SOC	0.441***	0.262**	0.545***	0.943***
CUL	0.116	0.035	0.175	0.084
Interactions
DISINS		−0.334**		0.109*
DISCOG		−0.287*		−0.394*
DISSOC		0.242**		0.440**
DISCUL		0.057		0.037
OBS	110	110	110	110
PERMS	2000	2000	2000	2000
R-SQUARE	0.523	0.546	0.606	0.638
ADJ R-SQR	0.491	0.500	0.579	0.602

ADM: administrative level; COG: cognitive proximity; CUL: cultural proximity; EETO: expenditure on education and technology of origin; EETD: expenditure on education and technology of destination; INS: Institutional proximity; MRQAP: multiple regression quadratic assignment procedure; PATO: patent applications of origin; PATD: patent applications of destination; SOC: social proximity.

*p < 0.1, **p < 0.05, ***p < 0.01.

Regarding urban endowment, all the coefficients of variables were found to be statistically positive, notwithstanding some being so at a low significance level. Cities with more patent applications and governmental expenditure on education and technology were shown to have better performance in transferring and absorbing patents, despite a decline in the influence of these kinds of size effects on intercity technology diffusion between 2007 and 2012. The same variable with different coefficients for origin and destination suggests a marginal difference. Our results show that the total number of patent applications played a more significant role in the transferring capacity of origin cities, while governmental expenditure on education and technology made a greater contribution to improving the absorption capacity of destination cities. Administrative effects exerted the strongest influence on intercity patent transfer, as was previously demonstrated by Cao et al. (2018), reflecting the fact cities with higher administrative levels can apply preferential policies to expand their resource bases, thus having an advantage over other cities in terms of developing more intercity linkages. The number of patents transferred between ordinary cities was 171 in 2007 and 1161 in 2012, respectively counting for 12.56% and 14.45% of the total. A slight drop in the ADM coefficients from 2007 to 2012 implies a decentralization trend in technology transfer flows, confirming the aforementioned emergence of Dongguan and Foshan as new technological stars.

Regarding intercity proximities, the impacts of geographical proximity and cultural proximity were not found to be significant. It might seem counterintuitive that distance does not appear to matter at first sight, but this finding holds true when Figure 3 is revisited. This result is to a large extent attributed to the relatively small scale of the GBA. Cities in this region are physically quite close to each other and most patents (86.71% in 2007 and 92.07% in 2012) were transferred between cities within 110 km of each other. Cultural proximity was not found to be a driver of technology transfer in the GBA. One possible explanation for this is that while dialects are a reflection of the informal vernacular culture of daily life, the culture of technology diffusion is more strictly and formally structured (Cao et al., 2019). Institutional, cognitive, and social proximities were shown to be able to facilitate the intercity patent transfer, as their coefficients were significantly positive for both time periods. Our results demonstrate that institutional differences remain a key barrier to technology transfers between cities—as seen in Figure 2, cities on the mainland side of the GBA maintained tighter technology transfer networks. In practice, the patent transfer between mainland cities and Hong Kong and Macau constitutes a form of technology import and export, which must get extra approval from the department in charge of foreign trade under the State Council, before being registered and announced by the Patent Office. For this reason, the trans-institutional patent transfer takes more time and effort. However, institutional barriers appear to have weakened between 2007 and 2012, with the deepening integration of the GBA (Ma et al., 2020). Cognitive proximity is shown to be positively associated with the intensity of intercity patent transfer, indicating that actors and firms in the GBA tended to assimilate new patents that were related to the technological structures of their own. This effect was enhanced over time, as the coefficient increased from 2007 to 2012, signaling that although technology diffusion is dependent on a combination of different and diverse knowledge, there are still certain technology proximity criteria under which actors are able to communicate efficiently (Boschma, 2005). A similar conclusion can be drawn in relation to social proximity—we found that cities in the GBA more frequently tended to transfer patents to those with whom they had already connected in the past, a finding which is consistent with path-dependence thinking that holds that initial conditions often have long-lasting effects (Boschma and Frenken, 2006). This path dependency strengthened between 2007 and 2012, since stronger social cohesion around a relationship reinforces willingness and motivation, as well as smooths communication barriers to sharing knowledge with others (Ter Wal, 2014).

Regarding proximity interactions, despite the insignificance of geographical proximity on its own, interactions with institutional, cognitive, and social proximities are all statistically significant to intercity patent transfer. As was argued by Boschma (2005), geographical proximity itself is neither necessary nor sufficient for intercity technology spillover in the GBA to take place. Instead, it indirectly affects network relations by acting on the proximity of other dimensions. The estimated interaction effect between geographical and institutional proximities was negative in 2007 but positive in 2012, suggesting a relational transformation from substitution to complementarity. In 2007, the intercity patent transfer between Hong Kong, Shenzhen, and Guangzhou was prominent, suggesting that geographical proximity can offset institutional differences (Ponds et al., 2007). In 2012, with the strengthened intercity patent transfer flows between Shenzhen, Guangzhou, Dongguan, and Foshan, where cities sharing the same institutional system were also located close to each other, the relationship between geographical and institutional proximities became complementary. There existed a consistent substitutional relationship between geographical proximity and cognitive proximity, which fits with the theory that geographical proximity substitutes for missing cognitive proximity in knowledge spillover and vice versa (Boschma, 2005). The substitutional effect was increasingly apparent as the coefficient decreased from −0.287 in 2007 to −0.394 in 2012, in line with the findings by Broekel (2015) in relation to German R&D collaboration networks, where decreasing cognitive distances between linked organizations across space tended to correlate with increasing geographical distances. The relationship between geographical proximity and social proximity is thus complementary in nature. Future transfers among cities that have already developed patent transfers in the past are more likely to take place if those cities are spatially close to each other. This complementarity was reinforced during the study period, corroborating the observations in Breschi and Lissoni (2009) and Broekel (2015) who have observed that social contacts are frequently located within the same geographical vicinity and growing social distances in networks accompany increasing geographical distances. Finally, the interaction between geographical proximity and cultural proximity is not statistically significant, so there is neither substitution nor complementarity between them.

Conclusions

Despite the fact that patent transfer can only capture technological knowledge flows that are encodable, exploitable, and patentable (Criscuolo and Verspagen, 2008), it is a practical indicator with great potential to depict possible technology diffusion across cities (Paci and Usai, 2009; Serrano, 2010). Following the work of Duan et al. (2018, 2019) and Liu et al. (2019), and drawing on patent transfer data, we constructed intercity technology transfer networks for the periods of 2007–2011 and 2012–2016 in the GBA, a city region characterized by innovation-oriented development and heterogeneous institutional embeddedness. Departing from conventional examinations of the evolutionary characteristics of the networks, in this study we explored the multidimensional proximity mechanisms underlying technology transfer flows, seeking to reveal the interactive effects of geographical and non-geographical proximities as well as their dynamics over time. The results showed a remarkable expansion and reinforcement of patent transfer flows across cities in the GBA from 2007 to 2012, which can be associated with urban technology endowment and intercity proximities.

In terms of urban technology endowment, while our results confirmed the positive impacts of patent size and expenditure on intercity patent transfer, the extent to which these factors influence the origin and destination was different. Administrative effects on intercity patent transfer turned out to be the strongest, but the effect decreased with the polycentric development of the networks over time. In terms of intercity proximities, our results confirmed the transfer facilitating effects of different forms of proximities (institutional, cognitive, social). While the effects of cognitive and social proximities became stronger, the promoting role of institutional proximity in relation to intercity patent transfer faded to some extent, due to the increasing integration of the GBA. Geographical and cultural proximities nevertheless did not perform as expected, a result which could be ascribed to the quite small scale of the GBA and the relatively formal and structured technology transfer behaviors. In terms of proximity interactions, the relationship between geographical and institutional proximities transformed from substitution to complementarity. There existed substitution between geographical and cognitive proximities but complementarity between geographical and social proximities. Both relationships were consistent and strengthened over time—that is, increasing cognitive proximity between actors across space tended to correlate with increasing geographical distances (i.e., decreasing geographical proximity), whereas decreasing social proximity in networks tended to accompany decreasing geographical proximity (Breschi and Lissoni, 2009; Broekel, 2015).

In the GBA, the impact of institutional differences on technology transfer is the result of an urban geographical division of labor under the long-term influence of two systems in one country. Differences in the internal systems of the GBA allow cities to take advantage of their own systems and assume different regional functions (Yang, 2005). The continuation of the pre-reunification capitalist system in Hong Kong and Macau has strengthened Hong Kong's position as an international financial, shipping, and trade center, and increased Macau's influence as a foreign exchange and international tourist destination. This has enabled Hong Kong and Macau to better perform their functions as important external windows in the GBA. The PRD has implemented the socialist system, and the central government of China has always attached importance to the economic development and transformation of PRD—the reform and opening-up policy. At present, the government is paying increasing attention to policies to support innovation-led development (Li et al., 2020; Wang et al., 2019); this has promoted the development and internal communication and cooperation within the PRD. While higher technological innovation and diffusion capacity are promoted in the PRD, Hong Kong, and Macau show characteristics of one-way output and low network status. Despite that the empirical research of this paper was carried out in relation to a region with a rather special background, the results can, to some extent, be applied to other cases, and even to international technology transfer. GBA is an epitome of the conflictual opposition, but also symbiosis, of socialism and capitalism, and as such it provides a certain reference for the technological innovation cooperation between mainland China and western capitalist countries.

Notwithstanding the merits of this study, there are some shortcomings that need to be acknowledged. First, this paper only considers technology transfer flows that lead to new patents and transactions but ignores technology diffuses from one city to another without leaving a trail in patent data. Future research could employ alternative data sources to reflect technology transfer such as technology transfer contracts, firms’ product portfolio diversification (Bednarz and Broekel, 2019). Second, we only considered intra-GBA patent transactions. The particularity of Hong Kong and Macau lies in the gateway role they have played in connecting the international and domestic spheres. Only focusing on the regional scale may lead to a one-sided understanding of their functions. As technology flows increasingly break through the boundaries of cities, regions, and even countries, it is necessary to pay more attention to the external flows outside the GBA. Finally, while the proximity framework has been widely applied in spatial knowledge diffusion, the factors underpinning the creation of intercity technology transfer are far from exhaustive. Future research could resort to other frameworks such as the market access framework (Donaldson and Hornbeck, 2016; Hanson, 2005) to combine measures in different dimensions.

Footnotes

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work is supported by the National Natural Science Foundation of China (grant No. 41971209 and 41901189) and the National Social Science Fund (grant No. 21&ZD107).

ORCID iD

Haitao Ma

References

Aghion

Howitt

Brant-Collett

, et al. (1997) Endogenous Growth Theory. Cambridge, MA: MIT Press.

Amin

(1999) An institutionalist perspective on regional economic development. International Journal of Urban and Regional Research 23(2): 365–378.

Andersson

Gunessee

Matthiessen

, et al. (2014) The geography of Chinese science. Environment and Planning A 46(12): 2950–2971.

Balland

(2012) Proximity and the evolution of collaboration networks: Evidence from research and development projects within the global navigation satellite system (GNSS) industry. Regional Studies 46(6): 741–756.

Balland

De Vaan

Boschma

(2013) The dynamics of interfirm networks along the industry life cycle: The case of the global video games industry 1987–2007. Journal of Economic Geography 13(5): 741–765.

Balland

Boschma

Frenken

(2015) Proximity and innovation: From statics to dynamics. Regional Studies 49(6): 907–920.

Bednarz

Broekel

(2019) The relationship of policy induced R&D networks and inter-regional knowledge diffusion. Journal of Evolutionary Economics 29(5): 1459–1481.

Boschma

(2005) Proximity and innovation: A critical assessment. Regional Studies 39: 61–74.

Boschma

Frenken

(2006) Why is economic geography not an evolutionary science? Towards an evolutionary economic geography. Journal of Economic Geography 6(3): 273–302.

10.

Breschi

Lissoni

(2009) Mobility of skilled workers and co-invention networks: An anatomy of localized knowledge flows. Journal of Economic Geography 9(4): 439–468.

11.

Broekel

(2015) The co-evolution of proximities-a network level study. Regional Studies 49(6): 921–935.

12.

Broekel

Boschma

(2012) Knowledge networks in the Dutch aviation industry: The proximity paradox. Journal of Economic Geography 12(2): 409–433.

13.

Broekel

Balland

Burger

, et al. (2014) Modeling knowledge networks in economic geography: A discussion of four methods. Annals of Regional Science 53(2): 423–452.

14.

Buenstorf

Geissler

(2012) Not invented here: Technology licensing, knowledge transfer and innovation based on public research. Journal of Evolutionary Economics 22(3): 481–511.

15.

Cantner

Graf

(2006) The network of innovators in Jena: An application of social network analysis. Research Policy 35(4): 463–480.

16.

Cao

Derudder

Peng

(2018) Comparing the physical, functional and knowledge integration of the Yangtze River Delta city-region through the lens of inter-city networks. Cities 82: 119–126.

17.

Cao

Derudder

Peng

(2019) Interaction between different forms of proximity in inter-organizational scientific collaboration: The case of medical sciences research network in the Yangtze River Delta region. Papers in Regional Science 98(5): 1903–1924.

18.

Caviggioli

Ughetto

(2013) The drivers of patent transactions: Corporate views on the market for patents. R&D Management 43(4): 318–332.

19.

Chen

Xie

(2018) How technological proximity affect collaborative innovation? An empirical study of China’s Beijing-Tianjin-Hebei region. Journal of Management Analytics 5: 287–308.

20.

Cohen

Levinthal

(1990) Absorptive capacity: A new perspective on learning and innovation. Administrative Science Quarterly 35: 128–152.

21.

Criscuolo

Verspagen

(2008) Does it matter where patent citations come from? Inventor vs. examiner citations in European patents. Research Policy 37(10): 1892–1908.

22.

Dekker

Krackhardt

Snijders

(2003) Multicollinearity robust QAP for multiple regression. In: Feiock

Schol

(eds) 1st Annual Conference of the North American Association for Computational Social and Organizational Science. Pittsburgh, PA: NAACSOS, pp.22–25.

23.

Donaldson

Hornbeck

(2016) Railroads and American economic growth: A “market access” approach. The Quarterly Journal of Economics 131(2): 799–858.

24.

Duan

Chen

, et al. (2018) Spatial-temporal complexity and growth mechanism of city innovation network in China. Scientia Geographica Sinica 38(11): 1759–1768.

25.

Duan

Zang

Chen

, et al. (2019) Regional integration in the inter-city technology transfer system of the Yangtze River Delta, China. Sustainability 11(10): 2941.

26.

Eaton

Kortum

(1996) Trade in ideas-patenting and productivity in the OECD. Journal of International Economics 40(3–4): 251–278.

27.

Feldman

Massard

(2002) Institutions and Systems in the Geography of Innovation. New York: Kluwer Academic Publishers.

28.

Ferraro

Iovanella

(2017) Technology transfer in innovation networks: An empirical study of the Enterprise Europe Network. International Journal of Engineering Business Management 9: 1–14. DOI: 10.1177/1847979017735748.

29.

Gambardella

Giuri

Luzzi

(2007) The market for patents in Europe. Research Policy 36(8): 1163–1183.

30.

Granovetter

(1985) Economic action and social structure: The problem of embeddedness. American Journal of Sociology 91(3): 481–510.

31.

Griliches

(1979) Issues in assessing the contribution of research and development to productivity growth. The Bell Journal of Economics 10(1): 92–116.

32.

Gui

Liu

(2018) International knowledge flows and the role of proximity growth and change. Growth and Change 49(3): 532–547.

33.

Hagerstrand

(1952) The Propagation of Innovation Waves . London: Royal University of Lund.

34.

Hannigan

Cano-Kollmann

Mudambi

(2015) Thriving innovation amidst manufacturing decline: The Detroit auto cluster and the resilience of local knowledge production. Industrial and Corporate Change 24(3): 613–634.

35.

Hanson

(2005) Market potential, increasing returns and geographic concentration. Journal of International Economics 67(1): 1–24.

36.

Hoekman

Frenken

Van Oort

(2009) The geography of collaborative knowledge production in Europe. Annals of Regional Science 43(3): 721–738.

37.

Howells

(2002) Managing the Chinese environment. Political Studies 50(5): 1029.

38.

Jaffe

(1986) Technological opportunity and spillovers of R&D: Evidence from firm’s patents, profits and market value. American Economic Review 76(5): 984–1001.

39.

Karlsson

Johansson

Stough

(2006) Entrepreneurship and Dynamics in the Knowledge Economy. London: Routledge.

40.

Knoben

Oerlemans

LAG

(2006) Proximity and inter-organizational collaboration: A literature review. International Journal of Management Reviews 8(2): 71–89.

41.

Krackhardt

(1988) Predicting with networks: Nonparametric multiple regression analysis of dyadic data. Social Networks 10(4): 359–381.

42.

Kratke

(2010) Regional knowledge networks: A network analysis approach to the interlinking of knowledge resources. European Urban and Regional Studies 17(1): 83–97.

43.

Krugman

(1991) Increasing returns and economic geography. Journal of Political Economy 99(3): 483–499.

44.

Lagendijk

Lorentzen

(2007) Proximity, knowledge and innovation in peripheral regions. On the intersection between geographical and organizational proximity. European Planning Studies 15(4): 457–466.

45.

Phelps

(2017) Knowledge polycentricity and the evolving Yangtze River Delta megalopolis. Regional Studies 51(7): 1035–1047.

46.

Wei

YHD

Miao

, et al. (2020) Innovation, innovation policies, and regional development in China. Geographical Review 110(4): 505–535.

47.

Liu

Niu

Han

(2019) Spatial dynamics of intercity technology transfer networks in China’s three urban agglomerations: A patent transaction perspective. Sustainability 11(6): 1–24.

48.

Huang

(2020) Proximity and the evolving knowledge polycentricity of megalopolitan science: Evidence from China’s Guangdong-Hong Kong-Macao Greater Bay Area, 1990–2016. Urban Studies 58(12): 2405–2423.

49.

Maggioni

Uberti

Usai

(2011) Treating patents as relational data: Knowledge transfers and spillovers across Italian provinces. Industry and Innovation 18(1): 39–67.

50.

Maggioni

Uberti

Nosvelli

(2017) The “political” geography of research networks: FP6 within a “two speed” ERA. International Regional Science Review 40(4): 337–376.

51.

Malecki

Oinas

(eds) (1999) Making Connections. Technological Learning and Regional Economic Change. Aldershot: Ashgate.

52.

Mancusi

(2008) International spillovers and absorptive capacity: A cross-country cross-sector analysis based on patents and citations. Journal of International Economics 76(2): 155–165.

53.

Martin

(2000) Institutional approaches in economic geography. In: Sheppard

Barnes

(eds) A Companion to Economic Geography. New Jersey: Wiley-Blackwell, pp.77–94.

54.

Meyer

Schiller

Diez

(2012) The localization of electronics manufacturing in the Greater Pearl River Delta, China: Do global implants put down local roots? Applied Geography 32(1): 119–129.

55.

Nooteboom

(2000) Learning and Innovation in Organizations and Economics. Oxford: Oxford University Press.

56.

Paci

Usai

(2009) Knowledge flows across European regions. Annals of Regional Science 43(3): 669–690.

57.

Ponds

Van Oort

Frenken

(2007) The geographical and institutional proximity of research collaboration. Regional Science 86(3): 423–443.

58.

Serrano

(2010) The dynamics of the transfer and renewal of patents. Rand Journal of Economics 41(4): 686–708.

59.

Shen

(2008) Hong Kong under Chinese sovereignty: Economic relations with Mainland China, 1978–2007. Eurasian Geography and Economics 49(3): 326–340.

60.

Storper

Venables

(2004) Buzz: Face-to-face contact and the urban economy. Journal of Economic Geography 4(4): 351–370.

61.

Teece

(1977) Technology-transfer by multinational firms-resource cost of transferring technological know-how. Economic Journal 87(346): 242–261.

62.

Ter Wal

ALJ

(2014) The dynamics of the inventor network in German biotechnology: Geographic proximity versus triadic closure. Journal of Economic Geography 14(3): 589–620.

63.

Thrift

Olds

(1996) Refiguring the economic in economic geography. Progress in Human Geography 20(3): 311–337.

64.

Torre

Rallet

(2005) Proximity and localization. Regional Studies 39: 47–59.

65.

Van Egeraat

Kogler

(2013) Global and regional dynamics in knowledge flows and innovation networks. European Planning Studies 21(9): 1317–1322.

66.

Wang

Wei

YHD

Lin

(2019) How does tolerance affect urban innovative capacities in China? Growth and Change 50(4): 1242–1259.

67.

Wei

YHD

(2015) Network linkages and local embeddedness of foreign ventures in China: The case of Suzhou municipality. Regional Studies 49(2): 287–299.

68.

Wei

YHD

Liefner

Miao

(2011) Network configurations and R&D activities of the ICT industry in Suzhou municipality, China. Geoforum; Journal of Physical, Human, and Regional Geosciences 42(4): 484–495.

69.

Xiong

Zhang

(2012) Language Atlas of China. Beijing: The Commercial Press.

70.

Yang

(2005) Multilevel governance in the cross-boundary region of Hong Kong-Pearl River Delta, China. Environment and Planning A-Economy and Space 37(12): 2147–2168.

71.

Zhang

Derudder

Wang

, et al. (2020) An analysis of the determinants of the multiplex urban networks in the Yangtze River Delta. Tijdschrift voor Economische en Sociale Geografie 111(2): 117–133.