中国城市网络的凝聚子群及影响因素研究
|
盛科荣(1977-),男,山东日照人,博士,副教授,研究方向为城市地理和区域可持续发展。E-mail: shengkerong@163.com |
收稿日期: 2018-07-06
要求修回日期: 2018-10-16
网络出版日期: 2019-12-02
基金资助
国家自然科学基金项目(41771173)
版权
Cohesive subgroups and underlying factors in the urban network in China
Received date: 2018-07-06
Request revised date: 2018-10-16
Online published: 2019-12-02
Copyright
凝聚子群特征及形成机理的研究是理解城市网络发育规律及其动力机制的重要切入点。利用2016年中国上市公司500强企业总部-分支机构数据,研究了中国城市网络凝聚子群的多维度特征,定量测度了城市间链接关系的影响因素,探索性的分析了凝聚子群的形成机理。结果发现:派系、k-核、lambda集合、核心-边缘方法都表明中国城市网络存在凝聚子群现象,揭示了城市网络链接强度的层级特征;经济规模、政治资源、知识资本是凝聚子群发育的重要影响因素,网络邻近性、地理区位和历史基础也深刻的影响着凝聚子群的空间格局;择优链接和路径依赖是凝聚子群发育的动力机制,城市关键资源和区位优势将进一步转化为城市网络竞争优势。在网络发展环境下,中国政府需要在更大空间尺度上推动城市间合作,并积极应对城市间发展差距趋于扩大的问题。
盛科荣 , 杨雨 , 张红霞 . 中国城市网络的凝聚子群及影响因素研究[J]. 地理研究, 2019 , 38(11) : 2639 -2652 . DOI: 10.11821/dlyj020180729
Cohesive subgroup constitutes a bridge connecting individual cities and urban network. This paper aims to analyze the cohesive subgroups and their mechanisms in the urban network in China. First, data on headquarter and branch locations of China's top 500 public companies in 2016 are subjected to ownership linkage model to approximate the urban network, resulting in a 294×294 valued urban network. Second, four measures of cohesive subgroup analysis, i.e. cliques, k-cores, lambda sets and core-periphery techniques are employed to generalize about the link strengths between cities. Finally, the influencing factors of the cohesive subgroups in the urban network are examined by using quadratic assignment procedure, and the mechanisms are explored under a conceptual framework of urban network growth. Three main findings are concluded. First, the four measures of cliques, k-cores, lambda sets and core-periphery techniques all indicate the presence of cohesive subgroups, revealing the hierarchical structure of link strengths in the urban network in China. The cohesive subgroups are mainly composed of core cities of urban agglomerations, and the cities in the eastern and central regions have more active economic ties compared to the cities in the western region. Second, key resources possessed by cities, such as economic scale, political resources, and knowledge capital, are important factors underlying the formation of cohesive subgroups. Links are more likely to occur between cities with larger economies, richer political resources and more intensive knowledge capital. Temporal distance, geographical location and path dependence also have a profound influence on the spatial pattern of cohesive subgroups. Third, network homophily and path dependence are the dynamic mechanisms underlying the development of cohesive subgroups, and the key resources and location advantages of cities will be further translated into network competitiveness. In the network environment, China's urban governance system and urbanization policies need to be adjusted accordingly. The Chinese government needs to promote network cooperation between cities on a larger spatial scale, and actively respond to the widening economic gap between cities under the network environment.
表1 k-核的分析结果(二值对称数据)Tab. 1 Results of k-cores analysis (symmetric binary data) |
| k-核 | 城市 | 数量 |
|---|---|---|
| 5-核 | {北京,上海,南京,苏州,南通,杭州,宁波,福州,厦门,济南,烟台,武汉,长沙,广州,深圳,重庆} | 16 |
| 4-核 | {天津,大连,嘉兴,青岛,郑州,佛山,惠州,海口,昆明} | 25 |
| 3-核 | {沈阳,长春,合肥,芜湖,马鞍山,潍坊,济宁,珠海,成都,西安,乌鲁木齐} | 36 |
| 2-核 | {石家庄,唐山,廊坊,呼和浩特,鄂尔多斯,哈尔滨,扬州,温州,三明,宜昌,襄阳,揭阳,昌吉州} | 49 |
| 1-核 | {邯郸,邢台,保定,太原,大同,包头,鞍山,本溪,朝阳,伊春,泰州,宿迁,铜陵,龙岩,南昌,鹰潭,赣州,上饶,临沂,商丘,济源,黄石,荆门,湘潭,香港,玉林,绵阳,遵义,攀枝花,曲靖,拉萨,西宁} | 81 |
表2 城市之间最大流量分析结果(Lambda≥7)Tab. 2 Maximum flows between city pairs (Lambda≥7) |
| 北京 | 上海 | 深圳 | 广州 | 武汉 | 重庆 | 南京 | 杭州 | 福州 | 济南 | |
|---|---|---|---|---|---|---|---|---|---|---|
| 北京 | – | 34 | 29 | 13 | 12 | 10 | 9 | 9 | 7 | 7 |
| 上海 | 34 | – | 29 | 13 | 12 | 10 | 9 | 9 | 7 | 7 |
| 深圳 | 29 | 29 | – | 13 | 12 | 10 | 9 | 9 | 7 | 7 |
| 广州 | 13 | 13 | 13 | – | 12 | 10 | 9 | 9 | 7 | 7 |
| 武汉 | 12 | 12 | 12 | 12 | – | 10 | 9 | 9 | 7 | 7 |
| 重庆 | 10 | 10 | 10 | 10 | 10 | – | 9 | 9 | 7 | 7 |
| 南京 | 9 | 9 | 9 | 9 | 9 | 9 | – | 9 | 7 | 7 |
| 杭州 | 9 | 9 | 9 | 9 | 9 | 9 | 9 | – | 7 | 7 |
| 福州 | 7 | 7 | 7 | 7 | 7 | 7 | 7 | 7 | – | 7 |
| 济南 | 7 | 7 | 7 | 7 | 7 | 7 | 7 | 7 | 7 | – |
表3 QAP矩阵回归结果Tab. 3 Analysis results of QAP regression |
| 多值有向网络 | 二值有向网络 | |||||
|---|---|---|---|---|---|---|
| 拟合系数 | P(Large) | P(Small) | 拟合系数 | P(Large) | P(Small) | |
| Intercept | 0.018*** | 0.011*** | ||||
| GDP | 0.079*** | 0.005 | 0.995 | 0.028*** | 0.000 | 1.000 |
| Capital | 0.305*** | 0.001 | 0.999 | 0.076*** | 0.000 | 1.000 |
| Knowledge | 0.203*** | 0.000 | 1.000 | 0.043*** | 0.000 | 1.000 |
| Distance | -0.005*** | 0.996 | 0.004 | -0.002 | 0.724 | 0.276 |
| Passenger | 0.045* | 0.066 | 0.934 | 0.018** | 0.022 | 0.978 |
| Telecom | 0.048** | 0.026 | 0.975 | 0.012*** | 0.007 | 0.993 |
| East | 0.068*** | 0.000 | 1.000 | 0.013** | 0.000 | 1.000 |
| Lag2005 | 2.068*** | 0.000 | 1.000 | 0.913*** | 0.000 | 1.000 |
| R-square | 0.896 | 0.647 | ||||
注:***、**、*分别代表在1%、5%和10%的水平上显著。 |
表4 城市类别-网络关系的假设检验结果Tab. 4 Join-count statistics between city attributes and link matrix |
| 关系代码 | 期望值 | 观测值 | 差量 | P ≥ Diff | P ≤ Diff | |
|---|---|---|---|---|---|---|
| GDP | 1-1 | 1227.974 | 193.000 | -1034.974 | 1.000 | 0.000 |
| 1-2 | 777.511 | 1203.000 | 425.489 | 0.000 | 1.000 | |
| 2-2 | 120.514 | 730.000 | 609.486 | 0.000 | 1.000 | |
| Capital | 1-1 | 1610.930 | 470.000 | -1140.930 | 1.000 | 0.000 |
| 1-2 | 480.453 | 1312.000 | 831.547 | 0.000 | 1.000 | |
| 2-2 | 34.618 | 344.000 | 309.382 | 0.000 | 1.000 | |
| Knowledge | 1-1 | 1513.741 | 470.000 | -1043.741 | 1.000 | 0.000 |
| 1-2 | 561.659 | 1273.000 | 711.341 | 0.000 | 1.000 | |
| 2-2 | 50.600 | 383.000 | 332.400 | 0.000 | 1.000 |
注:关系代码中的1表示属性值低于截断值的城市类别,2表示属性值高于截断值的城市类别;1-1表示类别为1的城市间关系,1-2表示类别为1和2的城市间关系,2-2表示类别为2的城市间关系。 |
表5 地理距离和网络关系的QAP相关分析结果Tab. 5 Results of QAP correlation between geographical distance and network links |
| 皮尔森 相关系数 | 欧式距离 | 汉明 距离 | 简单匹配 系数 | 杰卡德 相似系数 | 古德曼-古鲁斯卡 Gamma值 | 休伯特Gamma值 | |
|---|---|---|---|---|---|---|---|
| Obs Value | -0.032 | 1024.196 | 0.970 | 0.030 | 0.033 | -0.690 | 21181.000 |
| Significa | 0.002 | 0.998 | 1 | 1 | 1 | 0 | 0.998 |
| Average | 0 | 1017.132 | 0.967 | 0.033 | 0.036 | 0.005 | 28383.277 |
| Std Dev | 0.016 | 3.486 | 0 | 0 | 0 | 0.086 | 3539.423 |
| Minimum | -0.039 | 999.227 | 0.965 | 0.031 | 0.035 | -0.344 | 19595.000 |
| Maximum | 0.081 | 1025.744 | 0.969 | 0.035 | 0.037 | 0.363 | 46386.000 |
| P(Large) | 0.998 | 0.002 | 0 | 1 | 1 | 1 | 0.998 |
| P(Small) | 0.002 | 0.998 | 1 | 0 | 0 | 0 | 0.002 |
| [1] |
李小建 . 公司地理论. 北京: 科学出版社, 1999.
[
|
| [2] |
宁越敏, 武前波 . 企业空间组织与城市: 区域发展. 北京: 科学出版社, 2011.
[
|
| [3] |
甄峰, 刘晓霞, 刘慧 . 信息技术影响下的区域城市网络: 城市研究的新方向. 人文地理, 2007,94(2):76-81.
[
|
| [4] |
盛科荣, 张红霞, 佀丹丹 . 基于企业网络视角的城市网络研究进展与展望. 人文地理, 2018,33(2):11-17.
[
|
| [5] |
王成, 王茂军, 柴箐 . 城市网络地位与网络权力的关系: 以中国汽车零部件交易链接网络为例. 地理学报, 2015,70(12):1953-1972.
[
|
| [6] |
盛科荣, 杨雨, 孙威 . 中国城市网络中心性的影响因素及形成机理: 基于上市公司500强企业网络视角. 地理科学进展, 2019,38(2):248-258.
[
|
| [7] |
冷炳荣, 杨永春, 李英杰 , 等. 中国城市经济网络结构空间特征及其复杂性分析. 地理学报, 2011,66(2):199-211.
[
|
| [8] |
赵新正, 李秋平, 芮旸 , 等. 基于财富500强中国企业网络的城市网络空间联系特征. 地理学报, 2019,74(4):694-709.
[
|
| [9] |
赵金丽, 盛彦文, 张璐璐 , 等. 基于细分行业的中国城市群金融网络演化. 地理学报, 2019,74(4):723-736.
[
|
| [10] |
赵渺希, 魏冀明, 吴康 . 京津冀城市群的功能联系及其复杂网络演化. 城市规划学刊, 2014(1):46-52.
[
|
| [11] |
李仙德 . 基于上市公司网络的长三角城市网络空间结构研究. 地理科学进展, 2014,33(12):1587-1600.
[
|
| [12] |
路旭, 马学广, 李贵才 . 基于国际高级生产者服务业布局的珠三角城市网络空间格局研究. 经济地理, 2012,32(4):50-54.
[
|
| [13] |
|
| [14] |
|
| [15] |
|
| [16] |
刘军 . 整体网分析讲义: UCINET软件实用指南. 上海: 格致出版社, 2009.
[
|
| [17] |
|
| [18] |
|
| [19] |
|
| [20] |
|
| [21] |
|
| [22] |
|
| [23] |
|
| [24] |
|
| [25] |
|
| [26] |
姚士谋, 陈振光, 朱英明 . 中国城市群. 合肥: 中国科技大学出版社, 2006.
[
|
| [27] |
|
| [28] |
|
| [29] |
|
| [30] |
|
| [31] |
盛科荣, 张红霞, 佀丹丹 . 中国城市网络中心性的空间格局及影响因素. 地理科学, 2018,38(8):1256-1265.
[
|
| [32] |
|
| [33] |
潘峰华, 夏亚博, 刘作丽 . 区域视角下中国上市企业总部的迁址研究. 地理学报, 2013,68(4):449-463.
[
|
| [34] |
陈伟, 修春亮, 柯文前 , 等. 多元交通流视角下的中国城市网络层级特征. 地理研究, 2015,34(11):2073-2083.
[
|
| [35] |
|
| [36] |
|
| [37] |
刘君德, 靳润成, 周克瑜 . 中国政区地理. 北京: 科学出版社, 1999.
[
|
| [38] |
彭翀, 林樱子, 顾朝林 . 长江中游城市网络结构韧性评估及其优化策略. 地理研究, 2018,37(6):1193-1207.
[
|
| [39] |
|
| [40] |
|
| [41] |
|
/
| 〈 |
|
〉 |