Ecological Well-Being Performance Evaluation and Spatio-Temporal Evolution Characteristics of Urban Agglomerations in the Yellow River Basin

Abstract

The urban agglomerations in the Yellow River Basin are important carriers for China’s high-quality development. It is an inevitable trend to promote sustainable development and people’s well-being in the urban agglomerations of the Yellow River Basin. A case study of 70 cities of seven urban agglomerations in the Yellow River Basin from 2011 to 2017 is presented. The two-stage super-efficiency network slacks-based measure (Super-NSBM) model considering undesirable outputs is adopted to measure the ecological well-being performance (EWP), and the spatial correlation analysis method is used to analyze the spatio-temporal evolution characteristics of the EWP. The results show that the average EWP of the seven urban agglomerations in the Yellow River Basin was less than 1, showing a trend of firstly rising and then slowly decreasing. The average EWP presents the distribution pattern of the highest in the middle reaches, followed by the upper reaches, and the lowest in the lower reaches of the Yellow River Basin, respectively. The average ecological economic efficiency in the first stage of urban agglomerations of the Yellow River Basin was greater than the average economic well-being efficiency in the second stage. The average EWP in the Yellow River Basin shows local spatial heterogeneity during the study period. Policy measures are proposed to promote the improvement of the EWP of urban agglomerations in the Yellow River Basin. This study can provide reference for the policy formulation of high-quality green development and sustainable construction of urban agglomeration in the Yellow River Basin.

1. Introduction

The Yellow River Basin straddles the three major plates of the east, central, and west region of China, and it is rich in natural resources and densely populated. It is an important ecological shield and northern economic development zone in China and has profound significance for stabilizing national economic development and consolidating ecological security [1]. However, there is a big gap in economic development, serious ecological and water resources problems, and more acute development contradictions of unbalanced and insufficient development in the Yellow River Basin. Meanwhile, affected by the location endowment and other factors, urban agglomerations in the Yellow River Basin need to improve their economic connectivity, division of labor and cooperation quality, and coordinated development level [2]. Therefore, how to steadily promote sustainable economic development and improve people’s well-being has attracted the attention of governments in the Yellow River Basin. Recently, China’s government has attached great importance to the ecological protection and high-quality development of the Yellow River Basin. In October 2021, the outline of the Yellow River Basin ecological protection and high-quality development plan (hereinafter referred to as the outline) issued by the Communist Party of China (CPC) Central Committee and the State Council clearly upgraded the ecological protection and high-quality development of the Yellow River Basin as a major national strategy. With the deepening of the regional collaborative development strategy, the urban agglomerations have gradually become an all-new carrier of economic improvement. In January 2020, the Chinese government made a major deployment, emphasizing that the ecological environment of urban agglomerations in the Yellow River Basin should be improved and the people’s well-being should be enhanced. Therefore, how to realize the coordinated development of social development and residents’ ecological well-being within the constraints of resources and environment has become a major issue facing the sustainable development of the Yellow River Basin at this stage.

The essence of ecological well-being performance (EWP) is to obtain the maximum well-being output with the minimum resource input and environmental cost, which is to some extent an extension of the concept of eco-efficiency. The conception about the EWP can be traced back to Daly’s steady-state economic theory [3,4]. Daly (1974) used the ratio of natural consumption to the well-being level (service/throughput) to measure the level of sustainable development of a country or region at an early stage [5]. In the “empty world,” natural resources are relatively abundant, and the economic system is small compared to the ecosystem. When the “empty world” changes, man-made capital and the economic system expands rapidly, natural resources are scarce, and economic growth and the advancement of people’s well-being are curtailed under the limits of environment and resources; this state of advancement is referred to as the “full world” [6]. Max-Neef suggested that when the economic level grows to a certain level, the well-being level no longer increases with economic growth, but may also show a decreasing trend, making the quality of life and happiness decrease, and the relationship between them is an inverted “U” curve [7]. The strong sustainability research paradigm is based on the view that natural resources are absolutely scarce and cannot be replaced by other alternatives, and the social and economic efficiency is limited by the ecosystem [8,9]. Rees proposed the concept of ecological footprint, which can quantify the natural consumption and the degree of ecological impact, and the EWP has been widely used and recognized [10]. With continuous research, Zhu et al. improved the concept of EWP based on Daly’s achievements [11]. The research on the EWP has gradually shifted from theoretical studies to the quantitative research of the EWP evaluation, spatio-temporal evolution, and influencing factors [12,13]. From the perspective of research subjects, most of them take nation, province, or large cities as samples.

The EWP evaluation methods mainly include the indicator ratio method and the comprehensive evaluation method. (1) The indicator ratio method of the EWP is generally the proportion of people’s well-being degree to natural resource consumption. The human development index (HDI) and other indicators are used to measure social well-being, and ecological footprint (EF) or other indicators are used to measure ecological consumption. For example, Common studied the relationship between environment, economy, and satisfaction [4]. Although the indicator ratio method is simple and easy to understand, it has some drawbacks, such as the inability to calculate the optimal set of ratios and the lack of flexibility in selection. (2) Comprehensive evaluation methods mainly include the data envelopment analysis (DEA) method, stochastic frontier analysis (SFA) method, and regression analysis. DEA models are non-parametric methods, and the existing literature mainly uses the Super-SBM model and Malmquist index. The SFA method is the commonly used parametric method. For instance, Dietz et al. adopted the SFA method to assess the EWP of 135 countries [14]. Furthermore, Altıntaş adopted a heterogeneous panel model to analyze the relationship between per capita ecological footprint and CO₂ emissions of 14 European countries [15]. Coluccia used a DEA approach to assess agroecological efficiency in the Italian region [16]. However, the single stage efficiency measurement conducted by the traditional DEA models evaluates the production process as a “black box”, which cannot effectively evaluate the real efficiency in the process of system operation. The Super-NSBM model can further analyze the direction and path of efficiency improvement in each stage of EWP and open the “black box” of EWP transformation process.

Most studies used the Gini coefficient, coefficient of variation, and spatial autocorrelation methods to explore the spatial heterogeneity and spatial correlation effects of the EWP [17,18,19]. For example, Xu et al. selected the exploratory spatial data analysis (ESDA) method to study the spatio-temporal evolution characteristics of the EWP in 30 Chinese provincial regions from 2005 to 2014 [20,21,22]. Xiao and Ji used the Getis–Ord Gi* index to analyze the spatial characteristics and the evolution of regional patterns of cold hotspots of the EWP in 30 Chinese provinces from 2004 to 2016 [23]. Wang and Luo used the entropy weight method, Super-SBM model, and spatial correlation analysis to measure the regional EWP and spatial correlation of the Beijing–Tianjin–Hebei region from 2005 to 2017 [24]. Storto et al. calculated the eco-efficiency of 116 provincial capitals of Italy in 2011 by using DEA cross-efficiency and Shannon’s entropy method [25]. Bianchi et al. adopted the metafrontier DEA to research the temporal evolution of ecological efficiency in 282 European regions from 2006 to 2014 [26]. Kporsu et al. used the Malmquist–Luenberger productivity index to investigate the environment performance of 104 countries between 1980 and 2016 [27]. Dagiliūtė explored the ecological efficiency level, and the course of unhooking connections of environmental impact from economic growth in Lithuania between 1990 and 2008 [28]. The review of the EWP evaluation indicators is shown in Appendix A.

In summary, the existing studies on the EWP provide a reference for this paper, but there are also certain shortcomings. (1) The existing studies on the EWP mainly focused on the national or provincial scales, and few research studies haves paid attention to the EWP of urban agglomerations in the Yellow River Basin. (2) Few studies explored the spatial distribution, correlation, and agglomeration of the EWP, and the traditional analysis methods ignore the spatial correlation of data. (3) The existing literature mostly studied single-stage efficiency assessment of DEA, measuring the whole ecological well-being transformation procedures as a “black box”, which was unable to identify the validity of sub-stages.

Therefore, the research objectives of this paper are (1) to construct the EWP evaluation index, and to use the Super-NSBM model for measuring the EWP of the urban agglomerations along the Yellow River Basin from 2011 to 2017; (2) to empirically analyze the spatial and temporal evolution of the EWP in the urban agglomerations along the Yellow River Basin through spatial correlation analysis method; and (3) to propose the policy recommendations for the EWP improvement, and to provide scientific decision-making references for promoting high-quality development of urban agglomerations along the Yellow River Basin. This study will help to promote China’s green development and enrich the theoretical research on the EWP.

2. Method

2.1. Two-Stage Super-NSBM Model

Data envelopment analysis (DEA) was first put forward by Charnes in 1978 [29]. It is an effective method to measure the efficiency of decision-making units (DMUs) with a variety of input and output indicators. The Charnes–Cooper–Rhodes (CCR) model and the Banker–Charnes–Cooper (BCC) model are mainly widely used. Tone proposed the slack-based measure (SBM) model, which can reflect the slack variables of input surplus and output deficit [30,31]. However, the traditional DEA model regards the conversion process as a “black box” for single-stage efficiency measurement, which cannot exactly measure the actual efficiency of the transformation system [32]. In order to solve the problem, Tone constructed the super-efficiency network SBM (Super-NSBM) model [33]. Compared with the traditional DEA model, the Super-NSBM model has the advantages of higher scientific and metric accuracy, and it opens the “black box” of the conversion process of urban ecological input and people’s well-being output. Therefore, this paper constructs the urban EWP evaluation index system on the basis of the two-stage Super-NSBM model.

$${\rho }_{se}^{\ast }=\min \frac{{\displaystyle \sum _{k=1}^K{w}^k[1+\frac{1}{m_k}({\displaystyle \sum _{i=1}^{m_k}\frac{s_i^{k^{-}}}{x_{i0}^k})]}}}{{\displaystyle \sum _{k=1}^K{w}^k}[1-\frac{1}{v_{1k}+v_{2k}}({\displaystyle \sum _{r=1}^{v_{1k}}\frac{s_r^{gk}}{y_{r0}^{gk}}+{\displaystyle \sum _{r=1}^{v_{2k}}\frac{s_r^{bk}}{y_{ro}^{bk}})]}}}$$

(1)

The relevant formula is as follows:

$$s.t.\left\{\begin{array}{c}{\displaystyle \sum _{j=1,\ne 0}^n}{x}_{ij}^k{\lambda }_j^k+s_i^{k^{-}}={\theta }^k{x}_{i0}^k,i=1,\cdots ,m_k,k=1,\cdots ,K\\ {\displaystyle \sum _{j=1,\ne 0}^n}{y}_{rj}^{gk}{\lambda }_j^k+s_{}^{gk}={\phi }^k{y}_{r0}^{gk},r=1,\cdots ,s_k,k=1,\cdots ,K\\ {\displaystyle \sum _{j=1,\ne 0}^n}{y}_{rj}^{bk}{\lambda }_j^k-s_{}^{bk}={\delta }^k{y}_{r0}^{bk},r=1,\cdots ,s_k,k=1,\cdots ,K\\ \epsilon \le 1-\frac{1}{v_{1k}+v_{2k}}({\displaystyle \sum _{r=1}^{{\nu }_{1k}}}\frac{s_r^{gk}}{y_{r0}^{gk}}+{\displaystyle \sum _{r=1}^{{\nu }_{2k}}}\frac{s_r^{bk}}{y_{r0}^{bk}})\\ z^{(k,h)}{\lambda }^h=z^{(k,h)}{\lambda }^k,{\displaystyle \sum _{j=1,\ne 0}^N}{\lambda }_j^k={\displaystyle \sum _{k=1}^K}{w}^k=1\\ {\lambda }^k\ge 0,s^{k^{-}}\ge 0,s^{gk}\ge 0,s^{bk}\ge 0,w^k\ge 0\end{array}\right.$$

(2)

In Equations (1) and (2), $m_k$ and $v_k$ denotes the number of inputs and outputs of the $k$ stage, respectively. ${\varphi }_k$ denotes the number of intermediate indicators. $(k,h)$ denotes the connection from $k$ stage to $h$ stage. $x$ represents the input, $y$ represents the output, $z$ represents the intermediate output, ${\lambda }^k$ represents the model weight of the $k$ stage, and ${\omega }_k$ represents the weight of the $k$ stage. $s^{k-}$ denotes the slack variables of input indicators, and $s^{gk}$ and $s^{bk}$ denote the slack variables of desirable and undesirable outputs, respectively. The two-stage Super-NSBM model was selected in this paper, so that $k=2$. Meanwhile, the weights of each stage are set equally since the first stage and second stage are equally important.

2.2. Spatial Autocorrelation Analysis Method

Moran’s I test is most commonly used to study the spatial autocorrelation, including the global spatial autocorrelation and the local spatial autocorrelation [34]. The global spatial autocorrelation primarily reflects the overall trend in the region, while the local spatial autocorrelation mainly reflects the local spatial correlation and heterogeneity [35]. This study used Moran’s I index to research the spatio-temporal evolution patterns of the EWP of urban agglomerations along the Yellow River Basin. The formulation of global spatial autocorrelation is:

$$Moran’s\begin{array}{cc}I=\frac{{\displaystyle \sum _{i=1}^n{\displaystyle \sum _{j=1}^n{\omega }_{ij}(Y_i}-\overline{Y})(Y_j-\overline{Y})}}{S^{{}^2}{\displaystyle \sum _{i=1}^n{\displaystyle \sum _{j=1}^n{\omega }_{ij}}}}& \end{array}$$

(3)

where $S^2=\frac{1}{n}{\displaystyle \sum _{i=1}^n{(Y_i-\overline{Y})}^2}$, $\overline{Y}=\frac{1}{n}{\displaystyle \sum _{i=1}^n{Y}_i}$, $Y_i$ is the observed value of the $i$ city; $n$ denotes the total number of cities; and ${\omega }_{ij}$ denotes the spatial weight matrix. This study uses Rook’s first order adjacency weight. The Moran’s I index is usually tested with the $Z$ score, and the calculation formula is as follows:

$$Z=\frac{\begin{array}{cc}Moran’s& I-E(I)\end{array}}{\sqrt{VAR(I)}}$$

(4)

where $E(I)$ and $Var(I)$ represent expectation and variance, respectively. When $\left|z\right|\ge 1.96$,$p\le 0.05$, the correlation statistic is significant.

The range of Moran’s I values is [−1,1]. When Moran’s I value > 0, it is a positive correlation. When Moran’s I value < 0, it is a negative correlation. When Moran’s I value = 0, there is no correlation [36]. The local Moran’s I is shown as follows:

$$I_i=\frac{(x_i-\overline{x})}{S^2}{\displaystyle \sum _{j=1}^n{\omega }_{ij}}(x_j-\overline{x})$$

(5)

where $I_i$ is the local Moran’s I index, $S^2$ is the sample corrected variance, and other variables are the same as in Equation (3). If $I_i$ is positive, the city and its neighboring cities have similar spatial characteristics, and it is represented as a “High-High” or “Low-Low” type in the local indicators of spatial association (LISA) agglomeration map. If $I_i$ is negative, the city and its neighboring cities do not have similar spatial characteristics, and it is represented as a “High-Low” or “Low-High” clustering on the LISA agglomeration map.

2.3. Overall Research Method

This paper studies the EWP measurement and spatial-temporal evolution of urban agglomerations along the Yellow River Basin. Firstly, an index system for evaluating the EWP of urban agglomerations along the Yellow River Basin is constructed. Secondly, the two-stage Super-NSBM model considering undesirable output is constructed to evaluate the EWP of urban agglomerations in the Yellow River Basin. Based on the above research, the spatial correlation analysis of the EWP is applied to measure the spatio-temporal evolution patterns of the EWP, as shown in Figure 1.

3. Indicators and Data

3.1. Indicator System of the EWP

Combining the characteristics of the two-stage Super-NSBM model with undesirable outputs, this paper constructs the structure of the EWP evaluation index system. As shown in Figure 2, the EWP can be decomposed through the two-stage transformation system, in which the economic development is the intermediate variable. Stage I is represented as ecological economic efficiency, which refers to the efficiency of converting the ecological resource inputs to the economic outputs and undesirable outputs. Stage II is represented as the economic well-being efficiency, which refers to the efficiency of transforming economic inputs into well-being outputs. Based on the concept of the EWP and the following literature [37], the formula for measuring the EWP is presented as follows:

$$EWP=\frac{HDI}{EF}=\frac{GDP}{EF}\times \frac{EF}{GDP}\Rightarrow EWP=\frac{WB}{EI}=\frac{GDP}{EI}\times \frac{WB}{GDP}$$

(6)

In Equation (6), $EWP$ is the ecological well-being performance. $GDP$ reflects the level of economic development. $WB$ represents the comprehensive well-being level, including education, health care, and environmental development levels. $EI$ represents the ecological resource inputs.

3.1.1. Resource Input Indicators

The essential purpose of the EWP is to obtain the maximum well-being output under the constraint of minimum ecological input [36]. Referring to the existing research [38], the input indicators include resource input and capital input indicators. In this paper, water, energy, and land resource consumption are selected to measure the ecological resource input, which are expressed by per capita water consumption, per capita energy consumption, and per capita urban construction land area, respectively. According to the availability of data, this paper selects labor force and fixed assets to represent capital inputs, which are measured by labor force per 10,000 people and per capita fixed assets investment, respectively.

3.1.2. Intermediate Indicators

In the ecological economy transformation system, ecological input and capital input are regarded as input factors. Through the transformation of the system, desirable output and undesirable output are obtained in the first stage. Since economic growth is only an intermediate means, not the ultimate goal, this paper selects the economic development index represented by per capita GDP for measuring the desirable output of intermediate indicators. In addition, in the process of urban economic and social development, certain pollutants are inevitably discharged. Considering the availability of data, wastewater, exhaust gas, and waste are selected as undesirable outputs in this study, which are characterized by per capita wastewater emissions, per capita SO₂ emissions, and per capita soot emissions.

3.1.3. Well-Being Output Indicators

According to the indicators of the HDI proposed by the United Nations Development Programme (UNDP) in 1990, it evaluates well-being output from three aspects: economy, education, and health care development. GDP output in the first stage is used as an intermediate variable, and education and health care development are used as well-being output indicators. Moreover, referring to the research by Hu et al. [39], the environment level is employed to measure the well-being outputs. Therefore, comprehensive well-being mainly includes three dimensions: education level, health care level, and environmental level, which are characterized by the number of students enrolled per 10,000 people [40], the number of doctors per 10,000 people [41], and the per capita green area [39], respectively. The indicators of the EWP are shown in

Table 1. The indicator descriptions of the EWP evaluation.

Stage	Category	Dimension	Secondary Indicators	Unit
Input indicators	Capital input	Workforce	Labor force per 104 persons	Person
		Fixed assets	Per capita fixed assets investment	Yuan
	Resource input	Water consumption	Per capita water consumption	Ton
		Energy consumption	Per capita energy consumption	Kw·h
		Land consumption	Per capita urban construction land area	m2
Intermediate indicators	Undesirable outputs	Wastewater emission	Per capita wastewater	Ton
		Exhaust gas emission	Per capita SO2	kg
		Waste emission	Per capita Soot/dust	kg
	Desirable output	Economic development	Per capita GDP	Yuan
Output indicators	Well-being level	Education level	The number of students enrollment per 104 persons ($y_2^d$)	Person
		Health care level	The number of doctors per 104 persons ($y_3^d$)	Person
		Environmental level	Per capita green area ($y_4^d$)	m2

.

3.2. Data Sources

The Yellow River flows through nine Chinese provinces, with a total length of 5464 km, a drainage area of 752,773 km², and a population of 160 million. The 14th five-year plan stipulates that there are seven urban agglomerations along the Yellow River Basin. As shown in Figure 3, from eastern China to western China, they are the Shandong Peninsula Urban Agglomeration, Central Plains Urban Agglomeration, Shanxi Central Urban Agglomeration, Hubao Eyu Urban Agglomeration, Guanzhong Plain Urban Agglomeration, Ningxia Yellow River Urban Agglomeration, and Lanxi Urban Agglomeration. According to the principle of data availability and continuity, and considering that the composition of these seven urban clusters is slightly overlapped, 70 cities of the seven agglomerations along the Yellow River Basin are selected as sample cities, and the classification of the upper, middle, and lower reaches of the Yellow River Basin is shown in Appendix B.

This paper selects 2011–2017 as the study period, which has experienced the “12th Five-Year Plan”, “13th Five-Year Plan”, “New Urbanization Construction”, and “Comprehensive Deepening Reform”. The data for this research are mainly obtained from the China Urban Statistical Yearbooks, statistical yearbooks of various cities, and social development statistical bulletins. To avoid the error deviation of the measure values caused by the use of total scale data, the per capita consumption of each indicator is taken, namely the total amount of the indicator divided by the number of the urban resident population. To avoid the influence of inflation, the per capita fixed asset investment and per capita GDP are converted to the constant price level in 2011.

4. Results

4.1. Analysis Results of the EWP among the Urban Agglomerations in the Yellow River Basin

4.1.1. Comprehensive Level and Two-Stage Efficiency of the EWP

The two-stage Super-NSBM model with undesirable output was adopted, and Max DEA 8 Ultra software was used to calculate the EWP in this paper. The results can be seen in Appendix C. During the study period, the EWP of the seven urban agglomerations in the Yellow River Basin was 0.865, and most urban agglomerations failed to achieve DEA effectiveness. The seven major urban agglomerations along the Yellow River Basin belong to the upper, middle, and lower reaches of the Yellow River. Figure 4 shows the EWP of the three reaches of the Yellow River Basin from 2011 to 2017. The average EWP in the upper, middle, and lower reaches of the Yellow River Basin was 0.862, 0.895, and 0.861 respectively. The average EWP of the seven urban agglomerations in the Yellow River Basin reflects a distribution pattern of the highest in the middle reaches, followed by the upper and lower reaches, which indicates that there are obvious regional differences among the upper, middle, and lower reaches.

The Super-NSBM model was used to assess the two-stage EWP in the urban agglomerations of the Yellow River Basin between 2011 and 2017, and the results are shown in Appendix C. The overall economic well-being efficiency of the second stage from 2011 to 2017 was significantly lower than the ecological economic efficiency of the first stage. From the perspective of urban agglomeration, there are three different types, and the specific analysis is as follows. There are four urban agglomerations whose economic well-being efficiency in the second stage is greater than the ecological economic efficiency in the first stage, namely Guanzhong, Jinzhong, Lanxi, and Ningxia Urban Agglomerations. These four urban agglomerations are situated in the middle and upper reaches of the Yellow River. There are two urban agglomerations whose efficiency in the first stage is greater than that in the second stage, that is Shandong Peninsula Urban Agglomeration and Hohhot-Baotou-Ordos-Yulin Urban Agglomeration. Moreover, the Central Plains Urban Agglomeration is different from the above two types, and the comparison of EWP in two stages has a significant turning point. The efficiency of the first stage was greater than that of the second stage (2011–2012), and the efficiency of the second stage was greater than that of the first stage (2013–2017).

4.1.2. Temporal Distribution Characteristics of the EWP

As seen in Figure 5, the average EWP value of the seven urban agglomerations first increased from 0.864 in 2011 to 0.920 in 2013, an increase of 6.48% compared with 2011. After that, it decreased slowly, reaching 0.858 in 2017, a decrease of 6.74% compared with 2011. This shows that the EWP of the urban agglomerations is in the adjustment period from 2011 to 2017, and there is a large improvement space and development potential. The EWP of the urban agglomeration of the Yellow River Basin has an obvious trend of change. It can be seen that 2014 was an obvious inflection point for the EWP of urban agglomeration in the Yellow River Basin, which increased first and then decreased.

From 2011 to 2014, except for Guanzhong Plain Urban Agglomeration and Lanxi Urban Agglomeration, the other five urban agglomerations showed an upward trend. Among the urban agglomerations, the EWP of Ningxia Urban Agglomeration along the Yellow River Basin increased from 0.796 in 2011 to 1.013 in 2013, and it increased 27.3%. From 2014 to 2017, except for Lanxi Urban Agglomeration and Central Plains Urban Agglomeration, the EWP of all urban agglomerations showed a significant downward trend. Among them, urban agglomerations in the upper reach, such as Ningxia Urban Agglomeration and Hohhot-Baotou-Ordos-Yulin Urban Agglomeration, showed a significant downward trend. The urban agglomeration in the middle and lower reaches, such as Jinzhong Urban Agglomeration, Guanzhong Plain Urban Agglomeration, and Shandong Peninsula Urban Agglomeration, declined slowly.

4.1.3. Regional Distribution Characteristics of the EWP

The seven major urban agglomerations are situated in the upper, middle, and lower reaches of the Yellow River Basin. Based on the temporal distribution characteristics of the EWP, the year of 2014 is an important turning point in time. China has begun to perform the comprehensively deepen reform, and the ecological civilization has reached a new stage. Therefore, this paper selects the EWP in 2011, 2014, and 2017 and the average EWP value from 2011 to 2017 for analysis, and the spatial pattern of urban EWP is as shown in Figure 6 a–d. According to the research of Li et al. [42], the cities in the seven urban agglomerations are divided into six groups according to their EWP scores: excellent (EWP $\ge$ 1); better (0.8 ≤ EWP < 1); good (0.6 ≤ EWP < 0.8); poor (0.4 ≤ EWP < 0.6); bad (0.2 ≤ EWP < 0.4); and worst (EWP < 0.2).

As shown in Figure 6a, in 2011, the EWPs of Guanzhong Urban Agglomeration, Central Plains Urban Agglomeration, Shandong Peninsula Urban Agglomeration, and other urban agglomerations in the middle and lower reaches were mainly distributed in the excellent group. As shown in Figure 6b, compared with 2011, the number of cities in the excellent groups of Guanzhong and Shandong Peninsula Urban Agglomerations decreased in 2014, while the number of cities in the excellent groups and better groups of middle and upper reaches urban agglomerations such as Jinzhong, Central Plains, and Hohhot-Baotou-Ordos-Yulin Urban Agglomerations increased steadily. As shown in Figure 6c, in 2017, the number of cities in the worst group of Hohhot-Baotou-Ordos-Yulin and Ningxia Urban Agglomerations increased, while the number of cities in the excellent group of Guanzhong and Jinzhong Urban Agglomerations decreased. It can be seen from Figure 6d that from 2011 to 2017, the cities in the worst group were located in the middle and upper reaches, such as the Central Plains, Ningxia, and LanXi Urban Agglomerations.

4.2. The Spatial Correlation Analysis of the EWP

4.2.1. Moran Scatter Plot of the EWP

The spatial autocorrelation analysis method was used to conduct a global autocorrelation analysis of the EWP of the seven major urban agglomerations between 2011 and 2017. The measurement results are shown in

Table 2. Moran’s I statistical values for the EWP in the Yellow River Basin between 2011 and 2017.

Year	Moran’s I	p Value	z Value
2011	−0.180	0.008	−2.096
2012	−0.145	0.034	−1.753
2013	−0.073	0.198	−0.808
2014	−0.027	0.431	−0.229
2015	−0.096	0.140	−1.058
2016	−0.147	0.028	−1.733
2017	0.034	0.243	0.658

. Except that the P value of Moran’s I index was not significant in 2013, 2014, 2015, and 2017, the P value of Moran’s I index was significant at least at the 5% level, with global spatial autocorrelation. The global Moran’s I index in 2011–2016 was negative respectively, showing a negative spatial correlation. This shows that the urban EWP in the Yellow River Basin shows a discrete distribution on the whole, with obvious spatial differences. The Moran’s I index changed from negative to positive in 2017, showing a spatially positive correlation. This result expresses that the EWP level of the urban agglomerations presents a discrete distribution as a whole in the study period. With the Chinese government’s attention to ecological civilization and coordinated development of regional Yellow River Basin [43], the spatial difference of the EWP of the seven urban agglomerations has gradually narrowed and a spatial positive correlation has emerged.

To further investigate the local spatial heterogeneity of the EWP of the urban agglomerations, the local autocorrelation analysis of the EWP was analyzed, as shown in Figure 7. From 2011 to 2016, the cities in the Yellow River Basin were mainly located in the second and fourth quadrants, namely the high-value heterogeneous region (H-L quadrant) and the low-value heterogeneous region (L-H quadrant). This indicates the cities with high EWP levels are surrounded by those cities with low EWP values, and cities with low EWP values are surrounded by those cities with high EWP values, showing a high degree of local spatial heterogeneity. Compared with 2011, the spatial agglomeration and distribution characteristics of the EWP in 2017 were more significant, and the Moran’s I index became larger. Compared with 2014, the spatial agglomeration degree was optimized to a certain extent. This indicates that cities with higher EWP values in the Yellow River Basin are surrounded by cities with higher EWP values, while cities with lower EWP values are surrounded by cities with lower EWP values.

4.2.2. LISA Cluster Analysis of the EWP

Although the Moran scatter plot clearly shows that there is a certain positive or negative spatial correlation between cities in the Yellow River Basin, it does not indicate the degree of spatial autocorrelation between cities. The LISA cluster analysis compensates for this defect, which can measure the degree of similarity (positive correlation) and difference (negative correlation) between the spatial unit attributes and surrounding units. To find out the spatial dependence of the EWP, this study conducted the local spatial autocorrelation test. Due to the negative correlation of the EWP from 2011 to 2016 and the positive correlation in 2017, the LISA cluster analysis in 2011 and 2017 is mainly analyzed to explore the spatial correlation of local areas, as shown in Figure 8a–d.

It can be found that in 2011, Zhongwei and Xinxiang were situated in the H-L quadrant, and the significance of the two cities was at the 1% level, while Baotou, Weinan, and Rizhao were situated in the L-H quadrant. It shows that the difference between these cities and surrounding cities is gradually expanding. In 2017, Wuzhong, Yulin, and Yuncheng were situated in the L-L quadrant, while Yinchuan and Qingyang were situated in the H-L quadrant. The significance of Wuzhong, Qingyang, and Yuncheng passed the 1% level. To sum up, there are remarkable spatial differences in the EWP of cities situated in the seven urban agglomerations, and the spatial dependence and spatial effect change over time. Although a majority of the cities in the urban agglomerations have not reached a significant level, it also shows that the EWP of the cities has an obvious spatial distribution pattern.

5. Discussion

The results in Section 4 have presented the distribution pattern of the highest in the middle reaches, followed by the upper reaches, and the lowest in the lower reaches of the Yellow River Basin, according to the average EWP, respectively. This phenomenon is mainly owing to the huge differences in economic construction and natural environment endowments among the upper, middle, and lower reaches of the Yellow River Basin. The urban agglomeration in the upper reach of the Yellow River Basin has a sparse population and low-quality economic construction. The backward science, education, medical conditions, and social public services make the output level of social well-being relatively low. The urban agglomeration in the middle reach has a good natural environment, solid economic development, strong scientific and educational strength, relatively well social security, and the EWP level is in the leading position. Although the economic development level of the urban agglomeration in the lower reach is relatively high, the resource and environment problems are more prominent due to the large population, and the extensive economic development model needs to be further improved.

This paper found that the EWP of the seven urban agglomerations in the Yellow River Basin during the study period showed three different two-stage characteristics. Specifically, there are four urban agglomerations whose economic well-being efficiency in the second stage is greater than the ecological economic efficiency in the first stage, namely Guanzhong Plain, Jinzhong, Lanxi, and Ningxia Urban Agglomerations. These four urban agglomerations are situated in the upper and middle reaches. The low ecological economic efficiency is the main cause for the low EWP comprehensive standard of these urban agglomerations. The ecological environment of the middle and upper reaches is fragile, and the resources and energy are overexploited [44]. There are two urban agglomerations whose ecological economic efficiency in the first stage is greater than the economic well-being efficiency in the second stage, that is Shandong Peninsula Urban Agglomeration and Hohhot-Baotou-Ordos-Yulin Urban Agglomeration. The two urban agglomerations cannot translate economic growth into social well-being effectively, and they should pay more attention to the improvement of people’s well-being. The Central Plains Urban Agglomeration is different from the above two types, and the comparison of EWP in two stages has a significant turning point. The efficiency of the first stage was greater than that of the second stage (2011–2012), and the efficiency of the second stage was greater than that of the first stage (2013–2017). This may be related to the fact that the economic construction of the Central Plains Urban Agglomeration developed rapidly at the early stage, but neglected environmental protection and people’s well-being, leading to the ecological economic efficiency being greater than the economic well-being efficiency. However, the extensive development model of “high pollution + high consumption+high input” began to show its drawbacks in the later period, and negative impacts such as environmental pollution were concentrated, which caused the economic output to hardly make up for ecological losses. The government began to shut down high pollution and energy consuming industries and paid more attention to the improvement of people’s well-being and environmental protection.

The above study found that the EWPs in urban agglomerations have great differences. In 2011, the urban agglomerations in the middle and lower reaches have an important position and influence because of population and economic agglomeration [45]. Therefore, these urban agglomerations have made long-term progress in natural ecological restoration and the development of people’s well-being. The urban agglomeration in the upper reach is situated in western China, with a sparse population, fragile ecology, and backward economic development and infrastructure. Consistent with this view, Sun et al. found that the ecological and economic coordination of urban agglomeration in the upper reaches was lower than that in the middle and lower reaches [46]. With the economic development of the urban agglomerations in the middle and lower reaches, its EWP has reached a higher level. However, when a certain threshold is reached, the negative impacts such as resource waste and environmental pollution begin to break out. From 2011 to 2017, with the high economic development of the downstream urban agglomeration, a large number of heavy chemical and high-energy industries were transferred from the eastern coastal areas to these cities in central and western China [47,48]. Although it has brought some improvement to the economic level of the urban agglomeration in the upper and middle reaches, environmental pollution and ecological damage have become increasingly serious. The lower economic development, limited investment to solve the pollution problem, and low level of social well-being such as education and medical care in the middle and upper reaches of the Yellow River Basin are also the reasons for the reduction of the EWP.

6. Conclusions and Policy Recommendations

6.1. Conclusions

This paper evaluates the EWP of urban agglomerations in the Yellow River Basin, in which the two-stage Super-NSBM model considering undesirable outputs was used. The spatio-temporal evolution characteristics of the EWP were studied by using the spatial autocorrelation analysis method. The research conclusions of this paper are as follows. (1) From the overall characteristics of the EWP, the EWP of the seven urban agglomerations developed relatively stable and at an overall low level during the research period. The EWP of each urban agglomeration presented a phenomenon of first rising and then slowly declining. The EWP comprehensive level of the urban agglomerations showed a distribution pattern of “the highest in the middle reaches, followed by the upper and lower reaches”. (2) In terms of efficiency with the two-stage model, the economic well-being efficiency of the second stage of the seven urban agglomerations between 2011 and 2017 was significantly lower than the ecological economic efficiency of the first stage, which indicates that the low economic well-being efficiency is the main cause for the low overall level of EWP of the urban agglomerations in the Yellow River Basin. The second-stage economic well-being efficiency of the upper area and midstream urban agglomerations was greater than the first-stage ecological economic efficiency during the study period, while the first-stage ecological economic efficiency of the lower Yellow River urban agglomeration was greater than the second-stage economic well-being efficiency, showing different stage characteristics. (3) The EWP of the seven urban agglomerations shows high local spatial heterogeneity, and the spatial heterogeneity and spatial effect change over time. The Moran’s I index was negative from 2011 to 2016, while it changed from negative to positive in 2017, showing a spatially positive correlation.

Due to the dynamics and complexity of the research questions, this study also has some limitations. The following aspects will be improved and supplemented in the future: (1) In terms of EWP evaluation of urban agglomerations, there are some limitations in data collection, such as air quality-related death rates, subjective well-being, and so on. If these data can be collected in the future, the EWP index system needs to continue to improve. (2) This paper has carried out the evaluation and spatio-temporal evolution research on the EWP of the urban agglomerations in the Yellow River Basin, and the study on the influencing factors of EWP in urban agglomerations can be further carried out.

6.2. Policy Recommendations

First, the Chinese government should take the ecological protection as the guide and build the green development industrial system for the urban agglomerations. First, the government should strengthen the bottom line thinking of ecological protection, implement ecological restoration projects according to the characteristics of each urban agglomeration, and establish a monitoring and evaluation mechanism for the adverse ecological impact of urban construction and industrial development of each urban agglomeration. At the same time, it is required to reduce the proportion of traditional resource-based industries, promote the development of high-tech green technology industry in combination with the advantages of various regions, and actively build the green industrial system.

Second, it is essential to take the development of core cities and urban agglomerations as a breakthrough, bring into play the radiation role of the head urban agglomerations, and develop according to the characteristics of urban agglomerations and local conditions. The seven urban agglomerations should grasp the historical development opportunities and strengthen the radiation and driving ability of urban agglomerations by strengthening the central cities and important cities. On the basis of the characteristics of each urban agglomeration and the transformation efficiency of the two stages, the government should develop according to local conditions, avoid homogenization and vicious competition, and realize the improvement of the EWP.

Third, it is crucial to break through the restrictions of administrative divisions and improve the intercity communication and coordination mechanism. The lack of the concept of regional overall development has seriously hindered the high-quality development of the areas. Therefore, the seven urban agglomerations should break through the administrative division, establish and improve the coordination and communication mechanism between provinces and cities, so that the local governments of the urban agglomerations can communicate and negotiate, strengthen close cooperation, and provide a mechanism guarantee for accelerating the high-quality development of the Yellow River Basin. In the implementation, governments should actively establish and improve the communication and coordination mechanism within the limits prescribed by law and pay attention to the implementation and supervision of responsibilities.