Knowledge related diversity and its impact through inter-city knowledge spillovers based on community detection algorithm

This article constructs a spatial Durbin model using inter-city knowledge spillover as the spatial weight matrix, calculates knowledge related diversity through community detection algorithm, and discusses the impact of related diversity on innovation and its spatial spillover effects. Research has found that: 1) Knowledge spillovers between Chinese cities are showing a rapid growth trend. The knowledge spillovers between provinces are mainly concentrated between Beijing, Shanghai, Guangdong, Jiangsu, and Zhejiang, but the spillover gap between provinces is gradually narrowing. The inter-city knowledge spillover network exhibits a clear diamond structure, with developed cities playing a pivotal role. The knowledge spillovers within and between the 3 major urban agglomerations (the Beijing-Tianjin-Hebei, the Yangtze River Delta and the Pearl River Delta) are much higher than that among other urban agglomerations, but the differences between urban agglomerations have significantly narrowed. 2) The impact of knowledge related diversity on innovation has significant spatial spillover effects, and the growth of innovation relies more on the influence of knowledge spillovers from other city’s related diversity. 3) Heterogeneity analysis shows that the related diversity of eastern cities has a significant direct impact on innovation without spatial spillover effects, while there are very significant spatial spillover effects between central and western cities, and cross-regional influences also have significant spatial spillover effects. The knowledge related diversity is conducive to the development of breakthrough innovation and exploratory innovation, and this impact has significant spatial spillover effects. 4) Knowledge related diversity exhibits a clear buzz effect within urban agglomerations, driving innovation growth in each city through knowledge spillovers, while the role of pipelines across urban agglomerations is not yet evident. There is a significant buzz effect in the core-periphery urban agglomerations, while the equilibrium urban agglomerations show both buzz and pipeline effects.

本文以城市间知识溢出为空间权重矩阵（spatial weight matrix）构建空间杜宾模型（Spatial Durbin Model），通过社区检测算法（community detection algorithm）计算知识关联多样性，并探讨关联多样性对创新的影响及其空间溢出效应。研究结果表明：1）中国城市间知识溢出呈现快速增长态势。省域间知识溢出主要集中于北京、上海、广东、江苏与浙江之间，但省域间的溢出差距正逐步缩小。城市间知识溢出网络呈现清晰的钻石结构，发达城市发挥着核心枢纽作用。京津冀、长三角、珠三角三大城市群内部及城市群间的知识溢出水平远高于其他城市群，但城市群间的差异已显著收窄。2）知识关联多样性对区域创新的影响具有显著空间溢出效应，区域创新增长更依赖于其他城市关联多样性所引致的知识溢出效应。3）异质性分析显示，东部城市的关联多样性对创新存在显著直接影响，但不具备空间溢出效应；中西部城市间则存在高度显著的空间溢出效应，跨区域影响同样具有显著空间溢出效应。知识关联多样性有利于突破性创新与探索式创新的发展，且该影响存在显著空间溢出效应。4）知识关联多样性在城市群内部呈现显著的本地蜂鸣效应（buzz effect），通过知识溢出带动各城市的创新增长；而城市群间的跨区域管道效应（pipeline effect）尚未显现。核心-边缘型城市群存在显著的本地蜂鸣效应，而均衡型城市群则同时兼具本地蜂鸣效应与跨区域管道效应。