Spatial-Temporal Characteristics of Urbanization Efficiency in Coastal Cities of China

Taking the coastal cities in China as study case, this article used Super-SBM DEA and DEA Malmquist index models to analyse the urbanization efficiency, then use the Moran’s I to conclude the spatial cluster characteristics of the urbanization efficiency in coastal cities. The results show that: (1) The urbanization efficiency of coastal cities keeps increasing gradually, with the average value rising from 0.4920 in 2006 to 0.8262 in 2016. The cities with higher values locate mainly in the south of the Yangtze River, especially in Guangdong Province. (2) Technical level is the main reason that restricts the improvement of urbanization efficiency in coastal cities. On the contrary, scale efficiency and pure technology efficiency promote the improvement of urbanization efficiency. (3) There is an obvious positive agglomeration relationship in the urbanization efficiency of coastal cities, but this relationship becomes weaker in recent years. (4) The local Moran’s I of urbanization efficiency in coastal cities shows that: H-H cities are mainly located in the Pearl River Delta and Yangtze River Delta. L-L cities are mostly distributed in the north of the Yangtze River. L-H and H-L cities are scattered in each province and their spatial agglomeration is weaker.


Introduction
Since the reform and opening-up, China's urbanization level has been continuously improving. As the pioneer area of China's economic development, coastal cities have a higher urbanization rate, which has exceeded 60% in 2017. However, with the deepening of the resource exploitation, the resource and environment issues have become increasingly serious. To coordinate the development of resource, environment and urbanization, China has put forward the New-type Urbanization Plan, aiming at improving the urbanization efficiency.
China's urbanization efficiency has attracted scholars' attention since the 1980s, Charnes measured the urban economic efficiency of 28 cities in China in 1984, which verified the feasibility of DEA model in the measurement of urbanization efficiency [1]; Henderson calculated the process efficiency of urbanization, pointing out that inter-regional cooperation is a major factor to improve the efficiency of urbanization [2]. Chinese scholars started the empirical research on the urbanization efficiency since the 1990s, concentrating on the regional differences [3], dynamic evolution process [4] and influencing factors [5]. Existing studies found that: ① There are obvious regional differences in urbanization efficiency in China. The urbanization efficiency in the eastern region is much higher than 7th Annual International Conference on Geo-Spatial Knowledge and Intelligence IOP Conf. Series: Earth and Environmental Science 428 (2020) 012095 IOP Publishing doi: 10.1088/1755-1315/428/1/012095 2 that in the central, western and north-eastern regions [6]. Also, there is obvious spatial correlation between urbanization efficiency in Chinese cities [7]; ② There is significant dynamic change trend of urbanization efficiency in Chinese provinces and cities [8]. Recently, China has put up with relevant policies to balance the development of the eastern and western regions. As a result, the urbanization efficiency growth rate of the central and western regions has been higher than that of the eastern regions in recent years [9]; ③ The optimization of industrial structure, investment promotion, administrative promotion, infrastructure construction and land scale effect are reasons of fast urbanization development and the improvement of development efficiency in China [10]. Also, these factors are reasons to the differences of regional urbanization efficiency; ④Most researchers used the mainstream efficiency measurement methods such as DEA, SFA and their improved models to evaluate the urbanization efficiency. As one strategic area for China's development, coastal urban cities have not yet been specially studied by scholars on the urbanization efficiency issue.
As a priority development area for urbanization construction, coastal cities have comparative advantages in resource adsorption, innovation, transportation and so on, thus making them leaders in China's overall economic and social development. Considering the problems such as greater pressure of resource, ecological problems, higher cost of factors and more intense market competition in coastal cities, it is quite urgent for these cities to improve the urbanization efficiency to optimize urban space use, energy resource utilization and improve the ecological environment. Based on this, we choose the 53 coastal cities (including Shanghai and Tianjin, two province-level municipalities; HK, Macao and Taipei are not included) as the research area, and used the cross-sectional data of 2006, 2011 and 2016 to summarize the spatial evolution and differentiation characteristics of urbanization efficiency, so as to provide a reference basis for local governments to carry out urbanization construction.

Indicators
According to the definition of urbanization efficiency and existing researches, we choose the urban built-up area, urban fixed assets investment and non-agricultural population as the input indicators from the three aspects of land, capital and labour force. Since population urbanization, economic urbanization, social urbanization and ecological urbanization are four basic units of urbanization construction, we choose urbanization rate, non-agricultural output, total consumer goods, green coverage rate of urban built-up areas as output indicators. Data in this paper are collected from China Urban Statistical Yearbook (2007, 2012 and 2017). Partial missing data are interpolated according to the data of adjacent years.
x , jk y are the amount of the i-th input and j-th output, k  is the weight of the k-th input indicator.

DEA-Malmquist Model.
Compared with the traditional DEA model, DEA-Malmquist model is used to measure the dynamic characteristic of the efficiency among two neighboring periods. It was firstly created by Cave, who combined the Malmquist method and the DEA model. Based on the static results, we use the DEA-Malmquist to find the dynamic characteristic of the urbanization efficiency of coastal cities in China. So, the changing index of urbanization efficiency of coastal cities in China can be expressed as follow [12]: VRS implies the variable return of scale. The changing index of the urbanization efficiency is the product of the changes of technical level, pure technical efficiency and scale efficiency, among which the product of pure technical efficiency change and scale efficiency change is technical effect. It means that the urbanization efficiency of coastal cities shows an increase trend when 1 I  , and vice versa.

Spatial Autocorrelation 2.3.1. Global Moran's I. Global
Moran's I was created by Professor Luc to measure regional spatial correlation to evaluate whether the spatial distribution of geographical elements is clustering model, discrete model or random model according to the location and attributes of elements. The formula is as follow [13]: Where I is a global autocorrelation index with a range of values [-1,1]. When I is positive, it represents positive spatial autocorrelation. Conversely, when I is negative, it represents negative spatial autocorrelation. The smaller the difference between absolute value of I and 0 is, the weaker spatial correlation and the stronger spatial randomness are. xi is the variable attribute value of spatial position i, which is the urbanization efficiency in this paper; n is the number of observations, which is the number of coastal cities in this paper; Wij is the spatial weight matrix, that is, the spatial weight of spatial location i and j.

Local
Moran's I. Compared with Global Moran's I, Local Moran's I is used to evaluate the local heterogeneity of spatial autocorrelation. The formula of Local Moran's I is as follow: Where Ii is the local autocorrelation index.

Spatial-temporal Characteristics Analysis of Urbanization Efficiency in Coastal Cities
We

Dynamic Analysis of Urbanization Efficiency in Coastal Cities
We used the DEA-Malmquist model to analyze the dynamic trend of the urbanization efficiency of coastal cities, the results (Figure 2.) show that, among 2011 to 2016, the average value of total factor productivity index in coastal cities was 0.931, indicating that the urbanization efficiency decreased by 6.9%. Specifically, the average value of scale efficiency index was 1.056, increased by 5.6%; the technical level index was only 0.84, decreased by 16%, showing that the decrease of technical level is the main reason for the decrease of urbanization efficiency. There was obvious difference on the dynamic change trend of total factor productivity in coastal cities since the standard deviation was 0.3894. There were 24 cities with an upward trend of total factor productivity, accounting for 45.28% of the total cities. Fangchenggang was the one with the highest increase trend, with an increase index of 163.1%. Specifically, there were 28 cities with technical efficiency index more than 1, 8 cities with technical efficiency index equal to 1, and 17 cities with technical efficiency index less than 1. Among them, Fangchenggang had the highest increase in technical efficiency and Ningbo had the highest decline. As technical efficiency is obtained by multiplying pure technical efficiency and scale efficiency, we took a further analysis of the pure technical efficiency and scale efficiency. The results showed that there were 22 cities whose pure technical efficiency index exceeded 1, 15 cities whose technical efficiency index equaled 1 and 16 cities whose technical efficiency was less than 1, showing similar changing trend to the technical efficiency. The change of the scale efficiency was weaker than the overall technical efficiency, since there were 27 cities whose scale efficiency index was greater than 1, 11 cities whose scale efficiency index was equal to 1, and 15 cities whose scale efficiency was less than 1. Based on upon, we can tell that the change of technical efficiency was more caused by the change of pure technical efficiency level. Since technical level is another important factor in the changing of urbanization efficiency, we took the technical level analysis as well. The results showed that only 9 cities whose technical level index was greater than 1. Fangchenggang was the city with the highest improvement by 27.3%, on the contrary, Hangzhou was the city with the highest technological level reduction by 40.1%.

Spatial Agglomeration Characteristics Analysis of Urbanization Efficiency
To find the spatial agglomeration characteristic of the urbanization efficiency in the coastal cities, we use the GeoDa software to calculate the global and local Moran's I.The results ( Table 1.) showed that the global Moran's I values in all three years were positive, indicating that urbanization efficiency presents a certain trend of agglomeration. In 2006, Moran's I eaquled 0.200, and P value was significantly less than 0.1, indicating that the urbanization efficiency of coastal cities was significantly positively spatial correlated. Dynamicly, Moran's I shows a downward trend, indicating that the agglomeration degree of urbanization efficiency has been weakened. In 2016, not only the Moran's I index declined almostly to 0, but also the P-value was significantly higher than 0.1, so it can be seen that the urbanization efficiency of coastal cities tended to be random in 2016.  ①With more and more cities reaching DEA efficiency, the urbanization efficiency of China's coastal cities has an obvious growth trend, however, it still needs to be further improved. There is significant spatial differences of the urbanization efficiency among coastal cities, since the cities in the south of the Yangtze River are more urbanized than that of northern cities. Among the north region, cities in Shandong Province have a higher urbanization efficiency, and correspondingly, cities in Guangdong Province have a higher urbanizaiton efficiency among all southern coastal cities. ②The DEA-Malmquist index shows that the average of total factor productivity index of urbanization in coastal cities in China is less than 1, implying that the average of total factor productivity has decreased. Decomposition efficiency shows that the decline of technical level is the reason behind the decrease of total factor productivity efficiency. On the contrary, the pure technical efficiency and scale efficiency index are greater than 1, showing an increasing trend, indicating that pure technical efficiency and scale efficiency both improved the total factor productivity efficiency. To conclude, technical level has become the main reason that restricts the improvement of urbanization efficiency.
③The global Moran's I shows that the urbanization efficiency of China's coastal cities are positively agglomerated, but this relationship tends to be quite weaker in 2016; the local Moran's I shows that the spatial characteristics of four types of cities are as follow: L-L-type cities are mostly located in the north of the Yangtze River, H-H type cities are mainly located in the Pearl River Delta and the Yangtze River Delta. The distribution of L-H and H-L cities are widely distributed in coastal cities without obvious agglomeration characteristics.
In this paper, we measured the urbanization efficiency of coastal cities and found that although the urbanization rate of coastal cities has reached a quite high level, optimizing the structure of resource allocation and upgrading the technical level are still important works to improve the urbanization efficiency. Since cosatal cities are the pioneers of China's economic and urbanization development, it can be seen that technical innovation and adjustment of input structure are essential works to improve the urbanization efficiency in China. By Summarizing the temporal and spatial characteristics of urbanization efficiency of coastal cities, we found that there are obvious differences in urbanization efficiency between the north and the south. As coastal cities with similarities in factor endowments, we can see that the difference of regional development policies may be the potential cause of urbanization efficiency. Therefore, how to draw lessons from high-efficiency urbanization cities to achieve overall high efficiency urbanization worth to be further studied.