中国城市网络关联与经济增长溢出效应——基于大数据与网络分析方法的研究

深化城市网络空间特征的认识,科学测度经济增长的溢出效应,对推进新时期区域协调发展具有重要意义。本文结合城市网络和网络外部性两种研究视角,基于网络分析和空间计量方法,使用互联网大数据构建信息网络、企业网络、人口网络三种流要素城市网络,对中国336个地级以上行政单元的网络空间异质性、经济增长溢出效应进行实证分析。研究发现：中国城市网络联系具有显著的空间异质性,呈现出“中心-外围”特点,形成全国范围和省域、城市群范围两种尺度下的空间结构,表现出兼具择优选择与地理邻近的复杂特征;网络关联下城市经济增长具有显著的正向和负向溢出效应,反映出网络外部性对区域经济增长的重要影响,对比实体空间溢出效应表现出地理邻近关系下的相似性,以及跨越省级边界的差异性;对于全国范围下的网络关联矩阵,使用对称化和标准化处理更为合适,表明省域和城市群是推进区域经济协调发展的重要载体。

Urban network association and spillover effects of economic growth in China: A study based on big data and network analysis

Understanding the urban network association and measuring the spillover effects of economic growth have great significance in promoting coordinated regional development in the new era. This paper combined the perspective of urban networks and network externalities and took 336 China′s prefecture-level administrative units as research objects. With big data from the Internet, this research constructed three networks based on three kinds of "flow": information search, enterprise organization, and population migration. Based on the network analysis method, this study focused on the spatial heterogeneity of networks and examined the network characteristics, including urban hierarchy structure, node asymmetry, and spatial organization. Then, we utilized spatial econometric models to measure the spillover effects of economic growth caused by network associations and physical spatial correlation. The results show that: (1) China′s urban network based on three kinds of "flow" reveals significant spatial heterogeneity and complex characteristics of both preferential attachment and geographic proximity. Two spatial organization patterns are recognized in China′s urban network. One is on the national scale, and the other is on urban agglomerations and provincial scales. The "center-periphery" structure reveals the core position of central cities, like Beijing, Shanghai, Guangzhou, Shenzhen, and some new first-tier cities. (2) Spillover effects caused by network association could be positive or negative. Spillover effects of network association are essential factors in the economic growth of prefecture-level cities. Compared with physical spatial correlation, spillover effects caused by network association show both similarities under geographic proximity and differences across provincial boundaries. Focusing on the spillover effects across provincial boundaries, foreign capital, industrial fixed assets, and industrial land have overall significant positive effects, and one-period lag industrial indicators have a significant negative effect. (3) According to the extension of the J test and Log-Likelihood index in different models, matrix symmetrization and standardization processing is a more appropriate method in the network analysis on the national scale. The result emphasizes that the spillover effects of the economic growth are bidirectional in the "flow" element connection, which means both sides could benefit from the connection. The spillover effects are more significant on urban agglomerations and provincial scales than on the national scale. Finally, this paper makes several policy recommendations to coordinate regional economic development.