TL;DR
This study investigates how different types of proximity influence the development of knowledge and technological innovation networks within China's Nanjing Metropolitan Area (NMA). By analyzing co-authored publications and patent data, the research reveals that while Nanjing remains a central hub, technological collaborations are increasingly extending beyond provincial borders, driven by cognitive proximity.
What Problem Does This Paper Address?
Urban innovation networks are crucial for regional economic growth, but understanding how these networks evolve and what drives their development is complex. Specifically, this paper addresses the question of how different forms of "proximity" – such as geographical, organizational, social, cognitive, and technological – shape the formation and evolution of both knowledge-based (publications) and technology-based (patents) innovation collaborations within a specific metropolitan region, the Nanjing Metropolitan Area (NMA) in China. The authors aim to fill a gap in research by providing a quantitative analysis of these dynamics, moving beyond theoretical discussions to empirical evidence.
Methods
The study employs a multi-faceted approach:
- Social Network Analysis (SNA): UCINET and Gephi software were used to analyze the structure and dynamics of the innovation networks. Key metrics like network density, average clustering coefficient, and average path length were calculated to understand network connectivity and organization. Degree centrality was used to identify central cities.
- Multiple Regression Quadratic Assignment Procedure (MRQAP): This statistical technique was employed to assess the influence of various proximity factors on the innovation networks. MRQAP is suitable for analyzing relationships between matrices, making it ideal for examining how proximity matrices relate to network matrices.
- Proximity Measurement: The study operationalized five key proximity dimensions:
- Geographical Proximity: Measured by inter-city spatial distance using latitude and longitude data.
- Organizational Proximity: Assessed by whether cities are within the same province (binary: 1 for same province, 0 otherwise).
- Social Proximity: Based on dialectal regions, assigning a score of 1 for cities within the same dialectal region and 0 for different regions.
- Institutional Proximity: Evaluated using the "China Marketization Index" to assess marketization structure congruity.
- Cognitive Proximity: Measured by the disparity in city innovation indices, derived from patent value models.
- Technological Proximity: Assessed by the congruence of patent portfolios across cities.
Datasets
The study utilizes two primary datasets:
- Co-authored Publications: Data was collected from China's Wanfang Database and the Web of Science (WOS) for the period 2013-2020. This dataset includes 36,708 Chinese co-authored publications and 18,319 English co-authored publications between cities in the NMA.
- Collaborative Patents: Data was sourced from the China National Intellectual Property Administration's patent information database for invention patents filed between January 1, 2013, and December 31, 2020. The study focused on 2422 invention patents filed collaboratively among the nine prefecture-level cities within the NMA.
Key Insights and Ideas
- Diverging Evolution Paths: Knowledge innovation networks (KINs) are largely confined within Jiangsu province, while technological innovation networks (TINs) are increasingly expanding to cross-province collaborations.
- "Core-Periphery" Dominance: Both KINs and TINs exhibit a strong "core-periphery" structure, with Nanjing consistently acting as the dominant central city.
- Scale vs. Growth: KINs are larger in terms of city-pair collaborations, but TINs show a faster growth rate, indicating a more dynamic expansion in technological cooperation.
- Technological Spillover Effect: Technological collaborations demonstrate more pronounced spillover effects than knowledge collaborations, suggesting that technology diffusion is more outward-reaching.
- Cognitive Proximity Reigns Supreme: Cognitive proximity emerged as the most significant driver of innovation cooperation, indicating that cities with similar innovation capabilities are more likely to collaborate.
Key Results
- Network Growth: The number of co-authored papers in the NMA increased significantly from 2013 to 2020, with a notable surge in English-language papers indicating growing internationalization. The number of inter-city patent applications also showed an upward trend.
- Network Structure: Network density and average clustering coefficients generally increased for both KIN and TIN, indicating greater interconnectivity. However, TIN metrics remained lower than KIN metrics.
- Proximity Influence:
- Geographical and institutional proximities had insubstantial influence on innovation cooperation.
- Organizational proximity positively influenced collaborations, but its significance decreased over time, suggesting a move beyond provincial ties.
- Social proximity (dialectal diversity) did not negatively impact collaboration, likely due to the widespread use of Standard Mandarin.
- Cognitive proximity was the strongest positive predictor for both knowledge and technological collaboration, especially in the early stages of network development.
- Technological proximity had a positive effect on patent collaboration (similar tech infrastructure) but a negative effect on paper collaboration, suggesting that while similar tech can foster joint development, it might create challenges for broader knowledge exchange.
Limitations and Future Work
- Tacit Knowledge: The study acknowledges that a significant amount of tacit, unrecordable knowledge and technological cooperation is not captured by publication and patent data. Future research should explore ways to measure these aspects.
- Exogenous vs. Endogenous Drivers: The MRQAP model primarily focuses on exogenous proximity factors. Future work could explore endogenous drivers of city networks.
- Dynamic Perspective: The current analysis primarily compares initial and final years. A dynamic perspective exploring the interactions of proximity factors at different stages of network evolution is recommended.
- Scale of Analysis: The paper suggests refining the precision of innovation network research by delving into city and county levels, especially as metropolitan areas become more critical for national strategies.
- Technological Proximity Measurement: The use of patent types as proxies for technological proximity might not fully capture the nuances of collaboration strength.
Summary updated