基于空间计量模型的城市网络外部性定量测度

在经济全球化、区域一体化的背景下，城市发展动力由内部功能集聚转向外部关系协调，城市间的关联作用愈发重要，成为“城市的第二本质”。通过文献计量分析，发现空间计量模型逐渐成为城市网络外部性定量测度的主要方法之一。从研究设计角度，空间计量模型作为城市网络外部性的定量建模工具，存在至少两方面的问题：城市网络外部性的非对称效应和城市网络的多尺度问题。本文结合城市网络外部性测度与建模，首先阐述如何规范地解读空间计量模型参数及其与网络外部性测度的对应关系；其次基于蒙特卡洛模拟实验论证非对称城市网络外部性和多尺度城市网络外部性的建模方法。研究表明：(1)忽视城市网络外部性中潜在的非对称效应会显著降低模型参数估计的准确性，本文开发的非对称空间效应模型（Asymmetric Spatial Econometric Model,ASEM）可以准确识别网络外部性的非对称效应，给出更加准确的参数估计；(2)忽略现实存在的多尺度网络效应，只在单一尺度对网络效应建模会造成参数估计失真，而空间多尺度统计模型（Hierarchical Spatial Autoregressive model,HSAR）为多尺度网络外部...

Urban network externality modelled through spatial econometric models

In the context of economic globalization and regional integration, urban development forces have shifted from geographical agglomeration to external network coordination. The connectivity among cities or city network is increasingly becoming “the second essence of cities”. With bibliometric analysis of urban network externality studies, we find that spatial econometric models have been increasingly employed to quantify whether, and to what extent, urban network externality matters in the developmental process of cities or regions. This study first investigated the mathematical properties of spatial econometric models and in particular, the correspondence of model parameters and various measures of urban network externality. More importantly, we highlighted two essential issues of classic spatial econometric models when being applied to model urban network externality: the asymmetric effect and scale effect of urban network externality. Then, we developed an asymmetric spatial economic model (ASEM) to capture the asymmetric effect of urban network externality and, a hierarchical spatial autoregressive model (HSAR) to deal with the scale effect of urban network externality. A series of Monte Carlo simulation experiments were conducted to assess the benefits of these two proposed methodologies. We found that: (1) Ignoring the potential asymmetric effects in the urban network will significantly reduce the accuracy of model parameter estimation. On the other hand, the proposed ASEM can accurately identify the asymmetric effects of network externalities and yield more accurate parameter estimation. (2) Ignoring multi-scale network effects in real world data and modeling the network effects at a single scale will cause serious estimation distortion. The HSAR model instead can provide a reliable quantitative method for modeling the externalities of multi-scale networks.