-
Although lamenting the lack of genuinely relational data for cities has become common, rich sources of such data are available in the emerging era of big data (Neal, 2012). Publicly-listed firms, in particular, provide a useful dataset with two major benefits. First, publicly-listed firms are usually large corporations that dominate economic life throughout both the developed and leading developing countries (Chen, 2004). The annual ranking of leading companies released by Forbes magazine is a representative list of achievement-based, publicly-listed firms. Second, each listed company is commonly required to report its economic activities, major shareholders, and financial performance throughout the preceding year. These reports contain plentiful and comprehensive data on capital investment, which is genuinely relational data for cities. The urban network can be built by drawing on the investment linkages from the listed firms to their major shareholders and to their first-tier subsidiaries as well. Specifically, the annual reports provide records on each listed company’s activities and financial performance, in which the principal subsidiaries and branch offices are also revealed to the public. According to the past practice, both the major shareholders’ equity for each listed company and the investment volume from listed firms to their subsidiaries and other corporations are reported in terms of monetary measurement (Daude and Fratzscher, 2008).
In common sense, the capital investment recorded in the annual reports represent the long-term strategies rather than the short-term, speculative capital investment of investors. In the mature securities market, investors may use both capital in a narrow sense (e.g., cash, loans, and bonds) or broad capital (e.g., trademarks and patents) to capture a larger share of the market and reap more revenue. There is no minimum or maximum volume of capital investment, and all the investments recorded in the annual reports can be precisely measured in monetary units.
In China, with the development of the stock market, the proportion of listed companies in the national economy has also been increasing. As of the end of 2015, there were 2494 publicly-listed companies in the Shanghai and Shenzhen Stock Exchanges, which were headquartered in all provinces, municipalities, and autonomous regions of China. The total market value of listed companies reached 52.96 trillion yuan (RMB), which was almost the same as China’s GDP in 2015. China’s stock market is playing an increasing role in allocating capital across regions (Li, 2012; Bian, 2014), and it is a viable enterprise sample that can be used to conduct urban network research.
Despite the above-mentioned benefits, we acknowledge the limitations of the capital investment dataset of 2494 publicly-listed firms. On one hand, those publicly-listed companies in the Shanghai and Shenzhen Stock Exchanges only account for a very limited share of all enterprises in China. The capital investment derived from some China-located but oversea-listed companies (e.g., Alibaba and Tencent) as well as some other unlisted but influential firms (e.g., Huawei) are not included in this work. On the other hand, investment is nothing but only one of the various kinds of capital flows. Other capital flows such as capital transfers and capital transactions are equivalent perspectives for shaping inter-city linkages but not discussed in this research.
All in all, by collecting annual reports from 2494 listed firms, 3123 inter-city linkages from shareholders to listed firms and 15 027 inter-city linkages from listed firms to subsidiaries are screened out. Thus, 18 150 valid inter-city investment linkages comprise the final dataset, involving a total of more than 3 trillion yuan (RMB). Additionally, the 21 644 offices (including all listed firms, shareholders, and subsidiaries) are widely distributed across 356 prefectures and higher-level cities; these cities cover all parts of China (not including Chinese Taiwan, Hong Kong and Macao). Based on this investment dataset, we constructed a 356 by 356 city matrix.
The inter-city connectivity (rab) between two cities a and b is thus given by aggregating intra-firm connections rabj across all firms:
$$ {r}_{ab}={\sum }_{j=1}^{m}{r}_{abj} $$ (1) where rabj denotes the investment volume between cities a and b generated by firm j, and m is the total number of listed firms.
Finally, the total network connectivity (TNC) of city a is generated by summing up the city’s connectivity to all other cities:
$$ {TNC}_{a}={\sum }_{i=1}^{n-1}{r}_{ai}\left(a\ne i\right). $$ (2) For reasons of clarity, the connectivity of each city and dyad will be reported as standard connectivity, which is calculated by the percentages of the maximum connectivity.
-
Based on the same capital investment dataset, we developed two separate urban networks using two different quantitative approaches. The first is an intervention network based on the capital investment volume, which we call the capital-weighted urban network. The second is based on the number of investment linkages, named the unweighted network, which we used as a control network. The variable for intra-firm connections rabj was assigned as the number of investment linkages in the unweighted network, while rabj was assigned as the volume of investments with the monetary unit in the capital-weighted network. In this section, we will try to identify the quantitative biases of China’s urban network by comparing the capital-weighted and unweighted networks.
According to the Jenks Natural Breaks Classification Method (McMaster and McMaster, 2002), the connectivity of cities and city-dyads were divided into five categories individually: high, relatively high, medium, relatively low, and low. We employed the Quadratic Assignment Procedure (QAP) correlation (Krackardt, 1987) varying between 0 and 1 to compare the similarity of the two subnetworks. The larger the correlation coefficient, the higher the network similarity. Fig. 1 presents the overall spatial patterns of two different urban networks based on unweighted and capital-weighted investment linkages; they show noticeable differences with QAP correlation of 0.664. In the network formed by unweighted investment linkages, the Beijing-Tianjin-Hebei Urban Agglomeration (BTH), Yangtze River Delta (YRD), and Pearl River Delta (PRD) have formed divergent network radiation at the national level as three major urban agglomerations in the coastal China areas (Lu and Wan, 2014). The economic core of western China, Chengdu-Chongqing Economic Zone (CCEZ), shows prominent connections with three major urban agglomerations in the coastal areas. Thus, a diamond-shaped spatial pattern, which consists of almost all most-connected cities and strongest dyads, is presented in the territory. As for the capital-weighted network, Beijing, as the single-center feature, is the most prominent spatial pattern. As the command center of the state administration, Beijing has strong connections with other major cities in China. Although cities such as Shanghai, Guangzhou, and Shenzhen also reflect the characteristics of specific central radiation, the intensity and spatial scale of the central radiation are far from being comparable to Beijing.
Figure 1. The spatial patterns of unweighted and capital-weighted urban networks in China (not including Chinese Taiwan, Hong Kong and Macao). a) The unweighted urban network; b) The capital-weighted urban network
Additionally, a geographic concentration of city connectivity can also be observed in both urban networks along the eastern seaboard, which is consistent with our expectation (Zhang and Cai, 2020). Specifically, the average connectivity ratio of western, central, and eastern Chinese cities is 1.0∶1.5∶3.7 in the unweighted network, while the ratio of the capital-weighted network is 1.0∶1.6∶4.5 (Western Chinese cities include cities in Chongqing, Sichuan, Guizhou, Yunnan, Tibet, Shaanxi, Gansu, Qinghai, Ningxia, Xinjiang, Guangxi, and Inner Mongolia; Central Chinese cities consist of cities in Shanxi, Jilin, Heilongjiang, Anhui, Jiangxi, Henan, Hubei, and Hunan; Eastern Chinese cities include cities in Beijing, Tianjin, Hebei, Liaoning, Shanghai, Jiangsu, Zhejiang, Fujian, Shandong, Guangdong, and Hainan). Cities in eastern China have a greater say in national capital policymaking and market control. In line with the persistent east-west divide in socioeconomic development in China, the Gini coefficient of both unweighted and capital-weighted networks also shows a downward trend from west to east. Moreover, this hierarchical structure is becoming increasingly evident from the unweighted urban network to capital-weighted urban network. In particular, the Gini coefficient of the unweighted urban network connectivity among cities in western, central, and eastern China is 0.67, 0.59, and 0.54, while the value of the capital-weighted urban network is 0.74, 0.71, and 0.68. This can be ascribed to the Matthew effect in network science (ital policymaking and market control. In line with the persistent east-west divide in socioeconomic development in China, the Gini coefficient of both unweighted and capital-weighted networks also shows a downward trend from west to east. Moreover, this hierarchical structure is becoming increasingly evident from the unweighted urban network to capital-weighted urban network. In particular, the Gini coefficient of the unweighted urban network connectivity among cities in western, central, and eastern China is 0.67, 0.59, and 0.54, while the value of the capital-weighted urban network is 0.74, 0.71, and 0.68. This can be ascribed to the Matthew effect in network science (Barabási and Albert, 1999) and the cumulative advantages of top Chinese cities in terms of economic capital (Dannefer, 2003).
Table 1 lists the 20 most-connected Chinese cities in two different networks. Beijing, Shanghai, and Shenzhen are the top three cities, whereas Guangzhou is ranked fifth in both networks. Thus, we can verify China’s four tier-one cities’ leading positions in China’s urban network. It is also noteworthy that top cities hold much more dominant positions in the capital-weighted network. Beijing’s primacy index increases from 1.49 of the unweighted network to 2.38 of the capital-weighted network. As the second most-connected city of China’s network, Shanghai’s score is seven times larger than the tenth city in the capital-weighted network, which is in line with the comparison of the Gini coefficient. Furthermore, the different ranking order of China’s top 20 most-connected cities is likely to be a preliminary piece of evidence revealing potential quantitative biases. Ürümqi, as the most strategic place in the northwest border area of China, occupies a relative core position in both networks, which is a rather surprising result. This verifies the ‘regionality’ that exists within the city network (Taylor et al., 2014a), which is to say that a primary city of a large enough region is usually fully connected with other cities from and beyond its region. Additionally, a few cities with rich strategic resources—Taiyuan as the central city in the coal-producing area and Daqing as the main oil-producing area—are ‘punching above their weight’ in the capital-weighted urban network, which indicates an enhanced ‘strategic-ness’ (Taylor et al., 2014b).
Urban networks based on two
quantitative specificationsThe unweighted (control) urban network The capital-weighted (intervention) urban network City Relative connectivity City Relative connectivity Leading Chinese cities Beijing 1.000 Beijing 1.000 Shanghai 0.670 Shanghai 0.421 Shenzhen 0.505 Shenzhen 0.125 Hangzhou 0.278 Tianjin 0.103 Guangzhou 0.266 Guangzhou 0.097 Chengdu 0.223 Taiyuan 0.080 Wuhan 0.167 Chengdu 0.069 Tianjin 0.161 Hangzhou 0.061 Suzhou 0.152 Wuhan 0.059 Changsha 0.152 Ürümqi 0.059 Chongqing 0.125 Changsha 0.055 Nanjing 0.122 Hefei 0.054 Fuzhou 0.117 Chongqing 0.052 Hefei 0.117 Nanjing 0.049 Xi’an 0.095 Daqing 0.048 Ürümqi 0.094 Dalian 0.043 Xiamen 0.092 Xi’an 0.043 Jinan 0.088 Fuzhou 0.039 Nanchang 0.081 Datong 0.038 Shenyang 0.080 Nanchang 0.034 Table 1. Top 20 leading Chinese cities in two networks
Table 2 ranks the 10 strongest city-dyads from the unweighted and capital-weighted networks. In addition to the high-frequency appearance of China’s four tier-one cities in the most-connected city-dyads, what interests us more is the difference between these two sets of city-dyads. The most compelling finding is the prominent links between neighboring cities (for example, Beijing and Tianjin, Guangzhou and Shenzhen, and Shanghai and Suzhou) in the unweighted network. As for the capital-weighted city-dyads, the 10 most-connected city-dyads are all the connections between Beijing and other cities, including outward gateway cities like Shanghai and Shenzhen, regional central cities like Chongqing and Guangzhou, and strategic cities like Ürümqi and Daqing. Given China’s unique institutional contexts (Pan and Xia, 2014), we can assume that the significant role played by the SOEs shapes the dominant position of some top cities in the capital-weighted network.
Urban networks based on two
quantitative specificationsThe unweighted (control) urban network The capital-weighted (intervention) urban network City-dyad Relative connectivity City-dyad Relative connectivity Leading Chinese city-dyads Beijing-Shanghai 1.000 Beijing-Shanghai 1.000 Beijing-Shenzhen 0.667 Beijing-Tianjin 0.157 Beijing-Tianjin 0.470 Beijing-Chongqing 0.139 Shanghai-Shenzhen 0.459 Beijing-Chengdu 0.138 Beijing-Guangzhou 0.339 Beijing-Guangzhou 0.120 Shanghai-Hangzhou 0.333 Beijing-Shenzhen 0.119 Beijing-Chengdu 0.330 Beijing-Ürümqi 0.112 Guangzhou-Shenzhen 0.319 Beijing-Wuhan 0.100 Shanghai-Suzhou 0.302 Beijing-Daqing 0.097 Beijing-Wuhan 0.274 Beijing-Changsha 0.088 Table 2. Top 10 Chinese city-dyads in two different networks
Examination and Interpretation of the Quantitative Validity in China’s Corporate-based Urban Network Analysis
doi: 10.1007/s11769-021-1175-y
- Received Date: 2020-03-10
- Available Online: 2020-08-27
- Publish Date: 2021-01-05
-
Key words:
- corporate-based urban network /
- quantitative validity /
- capital investment flows /
- listed enterprises /
- China
Abstract: As a matter of expediency, most existing corporate-based urban networks can only be quantitatively measured by either counting the number of linkages or calculating the product of estimated service values. However, the impreciseness arising due to the limits of quantitative analysis may prove fatal to studies about non-market economies like China. Employing the capital investment dataset as an example, we build a capital-weighted intervention network as well as an unweighted control network to carry out an examination of the quantitative validity in China’s corporate-based urban network analysis. Both the overall spatial pattern and top city-dyads within the capital-weighted network witness Beijing, as the most dominant city, overshadow the performance of the others, and the unweighted network shows multilateral interactions between China’s top cities including Beijing, Shanghai, Shenzhen, and Guangzhou. To further interpret the noticeable differences, we divide the overall network into two subnetworks, inferred by focusing on state-owned enterprises (SOEs) and private enterprises. The results show that the public and private sectors have separately created vastly different subnetworks in China and that SOEs play a much more significant role in terms of capital. Besides fresh insights into China’s urban network, this study provides a cautionary tale reminding researchers of the essentiality and complexity when making a quantitative distinction between different linkages.
Citation: | ZHANG Ze, TANG Zilai, 2021. Examination and Interpretation of the Quantitative Validity in China’s Corporate-based Urban Network Analysis. Chinese Geographical Science, 31(1): 41−53 doi: 10.1007/s11769-021-1175-y |