Spatial Pattern and Spillover Effects of the Urban Land Green Use Efficiency for the Lanzhou-Xining Urban Agglomeration of the Yellow River Basin

Abstract

Green development is necessary for building high-quality modern economic systems. Due to limited land resources, ensuring the ecological benefits will not be damaged while developing the economy is necessary, as is improving urban land green use efficiency (ULGUE). Lanzhou-Xining Urban Agglomeration is a crucial urban agglomeration in the upper reaches of the Yellow River Basin. The improvement of ULGUE can promote the integrated management of the region’s land and achieve harmonious development with the social economy. This study adopts a super-efficiency SBM model to quantify the ULGUE of the Lanzhou-Xining urban agglomeration between 2015 and 2020, analysing its spatial and temporal patterns by combining exploratory spatial data analysis. The spatial spillover effect was discussed using the spatial Durbin model. The results showed that the average value of ULGUE in the study period was on the rise and showed positive spatial autocorrelation. The ULGUE of provincial capital cities was higher than non, along with the efficiency of city proper, compared to other counties. Many factors were found to positively influence the ULGUE of the whole study region. Among these were the optimisation of industrial structure, economic development, urbanisation level, and government intervention. Population concentration and environmental regulation had a negative spatial spillover effect.

1. Introduction

Green development is an important part of ecological civilisation. It aims to establish a development model constrained by the capacity of the ecological environment and the carrying capacity of resources. The rational use of natural resources can improve the living conditions of the human social environment and achieve harmonious coexistence between humans and nature [1]. Land resources are important carriers of economic, social, and ecological benefits. Improving the ULGUE is key in solving the contradiction between man and land. With the continuous improvement of social and economic development and the acceleration of industrialisation and urbanisation, the expansion of urban land has led to the squeezing of agricultural production and ecological spaces. Simultaneously, problems such as the chaotic layout of urban land, low land output, and negative impacts on the ecological environment have become increasingly prominent [2]. In this context, promoting ULGUE is critical for the green development of urban agglomerations. Land’s green use plays an important role in resolving the discrepancy between limited land resources and the demand for land due to population growth. Land green use refers to the integration of the green development concept of harmonious coexistence between man and nature in land use. Under the constraints of limited urban ecological environment capacity and land resources, multiple inputs, benefits, and outputs with land elements at their core as well as environmental pollution from land use are considered to achieve the unity of economic, social, and ecological benefits. Therefore, it is essential to improve the ULGUE in urban regions.

Relevant theoretical research on urban land-use efficiency began in the 1920s. The ecological school proposed fan, concentric circle, and multicore urban land expansion models [3,4]. Subsequently, an economic location theory school was formed. Urban land use is believed to be affected by many factors, such as the economic development level, population size, geographical location, and labour market separation [5]. With the deepening of the concept of green development, research on ULGUE has gradually become the focus of current land-use evaluation. Early scholars used single-indicator evaluation methods to consider urban land-use efficiency, such as land-use density or unit land output [6,7,8,9], or used GDP to measure land area [10]. However, this evaluation method did not fully reflect the relationship between multiple inputs and outputs in the process of urban land use in terms of efficiency. Later, some scholars gradually established a comprehensive input–output evaluation index system for urban land-use efficiency. Outputs include both desirable and undesirable outputs [11,12,13]. Scholars have considered environmental pollution factors in the study of ULGUE. For example, environmental outputs such as “three industrial wastes” were added to the efficiency calculation as undesirable outputs [14,15,16]. In terms of research methods for measuring the ULGUE, the SBM model is an improvement over the traditional DEA model. It considers the undesirable output and is the mainstream measurement method of ULGUE [17,18]. As for research on driving factors, some scholars have focused on the impact of certain factors on the ULGUE, such as industrial agglomeration [19], urban agglomeration [20], new urbanisation [21], and their coping strategies. Many factors were found to affect the ULGUE. Scholars mostly use the Tobit model or spatial econometric model to study influencing factors. For example, Chen et al. (2022) analysed the factors that affect the ULGUE of resource-based cities in the Yellow River Basin under carbon emission constraints [22]. Zhang et al. [19,23] used the spatial dynamic Durbin model to discuss the influencing factors and spatial spillover effects of ULGUE in 282 prefecture-level cities in China from 2009 to 2019 [23].

Relevant research results in China and abroad have provided a rich theoretical basis and reference method for this study. However, there are still general shortcomings: previous studies have ignored the consideration of carbon emissions. The latter is actually a key indicator of green development [24]. More studies focus on the impact of relevant factors on the ULGUE of the region but less on the spatial spillover effect; from the perspective of research scope, most of the current studies focus on the national, provincial, and urban agglomeration scales, and research on the ULGUE at the county scale needs to be strengthened. Relevant research on land use in the Lanzhou-Xining urban agglomeration also focuses on resource and environmental carrying capacity and ecological benefits. Therefore, this study takes 39 counties of the Lanzhou-Xining urban agglomeration as the research objects and constructs the input–output evaluation index system of the ULGUE. The SBM model was used to measure the green land-use efficiency of this region from 2015 to 2020. Here, the current situation and problems of land green use in the Lanzhou-Xining urban agglomeration are analysed, and the factors that affect the ULGUE are discussed. The direction and degree of action to take for each influencing factor is identified quantitatively using an econometric model. This study is expected to enrich the research scope of the ULGUE and provide a reference for promoting the healthy and sustainable development of urban agglomerations in the Yellow River Basin as well as regional coordinated development.

2. Materials and Methods

2.1. Study Area

According to the Lanzhou-Xining Urban Agglomeration Development Plan [25], the extent of the Lanzhou-Xining urban agglomeration is centred on Lanzhou, the capital of Gansu Province, and Xining, the capital of Qinghai Province, including 39 counties (districts and cities) in 22 prefectures, with a total area of 9.75 × 10⁴ ${\mathrm{km}}^2$ (Figure 1). The total economic output of the Lanzhou-Xining urban agglomeration is relatively small, and there are many development weaknesses and bottlenecks. The background of natural resources is fragile, and constraints on the ecological environment are becoming increasingly severe. The overall development of the region is backward, and there are large differences within the region itself. From 2015 to 2020, the urbanisation rate of the population in the Lanzhou-Xining urban agglomeration increased from 38.9% to 47.6%. Urban problems are also exacerbated by a growing urban population. The Yellow River Basin is an important ecological barrier and economic zone in China. It faces serious ecological risks. The instability of natural conditions and intensive development have caused excessive pressure on land resources. Lanzhou-Xining urban agglomeration is an important urban agglomeration in the upper reaches of the Yellow River and the “the Belt and Road” economic belt. It maintains homeland and ecological security and maintains the prosperity and stability of northwest China. At present, the urban agglomeration of Lanzhou-Xining urban agglomeration is in a critical period of rapid urbanization and industrial structure adjustment and optimization. Improving ecological carrying capacity is an inevitable requirement for high-quality economic development and green urbanization of urban agglomerations. As one of the spatial carriers and driving mechanisms for ecological protection and high-quality development in the Yellow River Basin, the concept of green development promotes the utilization of land resources and gets rid of the inefficient and extensive development mode. The study of green land utilization in the urban agglomeration of Lanzhou-Xining urban agglomeration has a unique value for the coordinated development of economic, social, and ecological environmental benefits of land utilization. Therefore, it is of great significance to study the land green use of the Lanzhou-Xining urban agglomeration. This will boost the intensive and economical use of land, maximise the economic output per unit area, and unify the ecological benefits of the Lanzhou-Xining urban agglomeration, which is also conducive to narrowing the gap between the east and west.

2.2. Data Sources and Indicator Selection

2.2.1. Data Sources

From 2015 to 2020, economic and social data such as GDP, population, and fixed asset investment of the counties and districts of the Lanzhou-Xining urban agglomeration in the Yellow River Basin were obtained from the following sources: the Statistical Yearbook of China’s County Town Construction, the Statistical Yearbook of China’s Counties, the Gansu Development Yearbook, the Qinghai Statistical Yearbook, the Xining Yearbook, the Lanzhou Statistical Yearbook, the statistical bulletin of national economic and social development of each city (prefecture), and the official website of each city’s statistics bureau. The fixed asset investment amount for all counties and districts except 2018 comes from the Statistical Yearbook of China’s County Town Construction. The 2018 data were obtained from the statistical bulletins of all counties and districts. Some data were missing, and SPSS21.0 was used to supplement the data via the time-series method. The carbon emission data from 2015 to 2018 were estimated with reference to the calculation method of the IPCC, and the carbon emission coefficient was calculated with reference to the IPCC Guidelines for National Greenhouse Gas Inventories. Carbon emission data from 2019 to 2020 were obtained by forecasting and testing the Lasso BP neural network model based on the gross national product, energy structure, energy efficiency, industrial structure, population size of each unit, and carbon emission data from 2015 to 2018 [26]. The annual grid satellite data of PM2.5 concentration, published by the International Geoscience Information Network Centre of Columbia University, was selected as the respective PM2.5 concentration data, which were further extracted with ArcGIS 10.6.

2.2.2. Indicator Selection

The input–output evaluation index system of the ULGUE is constructed according to the connotation of green use of urban land under the dual constraints of carbon emissions and environmental pollution and following the principles of science, systematics, and operability [22,23,27,28]. In terms of input indicators, the built-up area, population, and fixed-asset investment of Lanzhou-Xining were selected to represent the inputs of land, labour, and capital, respectively. The latter are the three most important elements in production activities, and in terms of desirable output, the GDP and green space area of built-up areas are used to represent economic and ecological benefits, respectively. In terms of undesirable output, the double constraints of carbon emissions and environmental pollution are mainly considered. The average concentration of PM2.5 is taken as the undesirable output indicator for environmental pollution. Further details are provided in

Table 1. Index system for measuring the ULGUE.

Layer of Criteria	Layer of Factors	Layer of Indicators	Unit
Input	Land	Urban built-up area	Square kilometre
	Labour	Population size	10 thousand persons
	Capital	Investment in fixed assets	CNY 10 thousand
Desirable output	Economic benefits	GDP	CNY 10 thousand
	Ecological benefit	Green area of the built-up area	hectare
Undesirable output	Negative impact on the environment	Carbon emissions	One hundred thousand tons
		PM 2.5 average concentration	pg/m3

.

2.3. Methodology

2.3.1. Super-Efficiency SBM Model

The DEA model can evaluate the technical efficiency of multi-input and multi-output decision making units (DMUs) without dimensionless data processing. It can determine the index weight in an optimal way, avoids subjective influence, and is widely used in the measurement of urban land-use efficiency. However, this method cannot solve the problem of measuring efficiency with undesirable outputs. The SBM model proposed by Tone [29] is now recognised and is also used to measure urban land-use efficiency including unexpected outputs, compensating for the deficiencies in the input and output relaxation variables. It is more accurate to measure the ULGUE with an unexpected output; therefore, this study used the SBM method to measure the ULGUE. The model is expressed as follows:

$$\begin{gathered} {\rho }^{*}=min\frac{1-\frac{1}{m}{{\displaystyle \sum }}_{i=1}^m\frac{s_i{}^{-}}{x_{ik}}}{1+\frac{1}{p_1+p_2}({{\displaystyle \sum }}_{r=1}^{p_1}\frac{s_h{}^{b-}}{b_{hk}})} \\ s.t.\{\begin{array}{c}\begin{array}{c}{x}_k=X\lambda +s^{-}\\ y_k=Y\lambda -s^{+}\\ b_k=B\lambda +s^{b-}\\ \lambda \ge 0, s^{-}\ge 0, s^{+}\ge 0, s^{b-}\ge 0\end{array}\end{array} \end{gathered}$$

(1)

As shown in Formula (1), each city represents a decision-making unit, AND each decision-making unit has $m$ inputs, $p_1$ desirable output, and $p_2$ undesirable output; $s^{-}$, $s^{+},$ and $s^{b-}$ represent the relaxation variables of input, desirable output, and undesirable output, respectively; $X,Y,$ and $B$ represent the input matrix, desirable output matrix, and undesirable output matrix, respectively; λ represents a weight vector; ${\rho }^{*}$ is the ULGUE value of cities in the study area. The larger the ${\rho }^{*}$ value, the higher the efficiency value. The ULGUE is calculated with the help of MATLAB R2020b.

2.3.2. Exploratory Spatial Data Analysis

Exploratory spatial data analysis (ESDA) aims to determine whether a geographic variable has spatial autocorrelation and the specific degree of correlation, mainly global and local spatial autocorrelation. Global spatial autocorrelation was largely used to discuss the overall spatial correlation characteristics of ULGUE in the study area. This method evaluates the significance by calculating the global Moran’s I, z scores, and p-values [30,31]. The value range of the global Moran’s I is (−1, 1), and its formula is shown in Equation (2). If the Moran’s I is significantly positive at a given level, it indicates significant spatial agglomeration. Moran’s I is usually tested using the Z test. If the Z test is significantly positive, it indicates positive spatial autocorrelation. A significantly negative Z test indicates a negative spatial autocorrelation [32]. The calculation formula is as shown in Equation (3), where E [I] is the mathematical expectation, and VAR [I] is the variance. Moran’s I is calculated with the help of GeoDa.

$$\mathit{Moran}’\mathit{s} \mathit{I}=\frac{{{\displaystyle \sum }}_{i=1}^n{{\displaystyle \sum }}_{j=1}^n\left(x_i-\overline{x}\right)\left(x_j-\overline{x}\right)}{S^2{{\displaystyle \sum }}_{i=1}^n{{\displaystyle \sum }}_{j=1}^n{W}_{ij}}$$

(2)

$$Z=\frac{I-E\left[I\right]}{\sqrt{VAR\left[I\right]}}$$

(3)

2.3.3. Spatial Durbin Model

The spatial Durbin model (SDM) includes the spatial lag of dependent variables and the change in error terms caused by random shocks. Therefore, the SDM [33,34] should be constructed when both the spatial effects of the explained variables and spatial interaction effects of error terms in the model are considered. Compared with the traditional ordinary least squares estimation, the SDM can not only examine the endogenous correlation of dependent variables but can also detect the direct and interactive effects of external factors. As this study not only focuses on the influencing factors of the ULGUE in the Lanzhou-Xining urban agglomeration but also explores the spatial spillover effect of explanatory variables on the ULGUE in adjacent areas, the SDM is used for analysis and measurement [35]. The formula is as follows:

$$Y_{it}=\rho {\displaystyle \sum }_{j=1}^n{W}_{ij}{Y}_{it}+{\displaystyle \sum }_{q=1}^m{\beta }_{iq}{X}_{it}+{\displaystyle \sum }_{j=1}^n{W}_{ij}{X}_{it}\theta +{\mu }_i+{\delta }_t+{\epsilon }_{it}$$

(4)

In the above formula, $Y_{it}$ is the explained variable representing the ULGUE of the $i\mathrm{th}$ city of Lanzhou-Xiningin period $t$. $X_{it}$ is the vector of each influencing factor; $\rho$ is the spatial autoregression coefficient, which indicates the degree to which the local ULGUE is affected by the adjacent areas; $W_{ij}$ is the spatial weight matrix; ${\beta }_{iq}$ is the coefficient of each influencing factor; $\theta$ is the coefficient vector of the spatial lag term, reflecting the extent to which the ULGUE of a city is affected by the factors of neighbouring cities. ${\mu }_i$ is the individual spatial fixing effect, ${\delta }_t$ is the individual time fixing effect, ${\epsilon }_{it}$ is the random interference term, and $\lambda$ is the spatial autocorrelation coefficient. $i$ and $j$ represent the $i\mathrm{th}$ and $j\mathrm{th}$ cities, respectively. In this study, $j$ represents the 39 counties of the Lanzhou-Xining urban agglomeration.

Using the SDM in the regression process, the commonly used spatial weight matrices are the spatial adjacency matrix (i.e., 0–1 matrix), geographic distance matrix, economic distance matrix, and economic geographic distance matrix. In this study, a spatial adjacency weight matrix was used. The influence of the independent variable on the dependent variable in the spatial econometric model is divided into direct, indirect, and total effects by using the spatial partial differential decomposition method. The direct effect represents the impact of the relevant driving factors on the ULGUE of local cities. The indirect effect refers to the impact of relevant driving factors on the ULGUE of neighbouring cities, namely the spatial spillover effect. The total effect reflects the average impact of the relevant driving factors on the overall ULGUE.

The results, including LM test and Hausman test, were all calculated through STATA16.0, and further model selection was made.

3. Results

3.1. Analysis of Spatial Pattern of the ULGUE

According to the calculation method of the super-efficiency SBM model, the ULGUE of counties and districts in the Lanzhou-Xining urban agglomeration in the Yellow River Basin from 2015 to 2020 were calculated. To compare the ULGUE of the Lanzhou-Xining urban agglomeration in 2015 and 2020, based on the current value of the ULGUE, the natural breakpoint method was used to divide it into four levels: low, lower, high, and higher, and visual expression was made [36].

The overall ULGUE of Lanzhou-Xining urban agglomeration is fluctuating and rising, and the high-value areas are gradually increasing. In 2015, the high-value areas were mainly distributed in Lanzhou and Xining. In 2015, the GDP of Lanzhou and Xining accounted for 45% of the total GDP of Lanzhou-Xining urban agglomeration, with a large population and a compact industrial layout, making their ULGUE level advantageous in the entire urban agglomeration. The low-value areas are mainly the counties of the Lanzhou-Xining urban agglomeration with a low economic development level. It is difficult for such counties to integrate into the regional economic development centre, and the overall level of ULGUE is low.

In 2020, compared with the distribution areas in 2015, to the high-value areas were added Pingchuan District of Baiyin and Ledu, Minhe, Linxia, and other areas around Lanzhou (Figure 2b), which benefit from the spatial spillover effect of neighbouring high-value counties. It is worth noting that Lintao, Guinan, and Dongxiang are also high-value areas, which may be due to the low intensity, scale, and resource input of their own land use; the weakening of the ecological environment cost of land burden; and the reasonable allocation of input and output. On the whole, from 2015 to 2020, the ULGUE of the Lanzhou-Xining urban agglomeration showed a trend of improvement, with distinct levels and typical distribution. The ULGUE of provincial capital cities was higher than that of non-provincial capital cities, and the efficiency value of municipal districts was significantly higher than that of counties under municipal jurisdiction.

We calculated the univariate global Moran’s I and its significance for the Lanzhou-Xining urban agglomeration ULGUE in the Yellow River Basin from 2015 to 2020. The results show that Moran’s I in the world is positive and passes the significance test (

Table 2. Global Moran’s I of the ULGUE in Lanzhou-Xining urban agglomeration.

	Moran’s I Value	Z Score	p-Value
2015	0.4807 ***	4.6868	0.01
2016	0.2786 **	2.8101	0.02
2017	0.3947 ***	4.0263	0.01
2018	0.0990 *	1.4638	0.08
2019	0.215 ***	2.6430	0.01
2020	0.142 *	1.4946	0.09

Notes: This study used the queen’s contiguity weight matrix. *** p ≤ 0.01, ** p ≤ 0.05, and * p ≤ 0.1.

), showing a positive spatial autocorrelation on the whole. That is to say, the high-value area of ULGUE is surrounded by the same high-value area with spatial correlation.

3.2. Analysis of Influencing Factors

3.2.1. Identification of Influencing Factors

With reference to relevant research results, indicators such as population size, optimisation degree of industrial structure, economic development level, environmental regulation, urbanisation level, and government intervention were selected as the driving factors for the ULGUE of the Lanzhou-Xining urban agglomeration.

Table 3. Variables of driving factors of the ULGUE in Lanzhou-Xining urban agglomeration.

Driving Factors	Index	Unit	Code
Population agglomeration	Population density	10 thousand persons/km2	density
Industrial structure	Ratio of output value of tertiary and secondary industries	—	is
Economic development	Per capita GDP	CNY 10 thousand	pgdp
Environmental regulation	Sewage treatment rate	%	er
Urbanization level	Ratio of urban population to total population	%	ur
Government intervention	Fiscal expenditures	CNY 10 thousand	gov

shows the specific connotations of each indicator.

• Population agglomeration: A high level of population agglomeration will correspondingly improve the concentration level of the labour force, capital, etc., which will bring economies of scale. It will simultaneously promote the development of urban land, which is conducive to the improvement of ULGUE. In this study, population density is selected to represent the level of population agglomeration;
• Industrial structure: The optimisation of the industrial structure can greatly reduce environmental pollution, optimise the allocation of land resources, and promote the intensive use of land. This study uses the ratio of the gross output value of tertiary and secondary industries as the evaluation index for the optimisation of industrial structures [37];
• Economic development: The economic development level reflects the number of input factors per land area unit. The higher the level of economic development, the higher the financial support for intensive land use and thus the higher the ULGUE. This study uses GDP per capita to characterise the level of economic development [38];
• Environmental regulation: Environmental regulation is a type of restriction measure taken by government departments on social and economic activities to reduce pollutant emissions. The most direct impact is that the government can improve the ULGUE by formulating measures to control pollution. In this study, the level of environmental regulation was measured using the sewage treatment rate [32];
• Urbanization level: The urbanisation level is represented by the population urbanisation rate, which is calculated as the ratio of the permanent urban population to the total population. The process of urbanisation refers to the continuous improvement of urban infrastructure, the adjustment of land-use structure, and the accelerated expansion of land construction, leading to problems such as the abuse of land resources and the uncertainty of land-use structure, thus reducing the level of land green use;
• Government intervention: The government has a certain decision-making power and credibility, and reasonable government intervention can improve the ULGUE. However, there is a competitive relationship between local governments. For example, to develop the economy, ecological losses are ignored, foreign capital is introduced blindly, and land is sold at will in pursuit of economic benefits. This study selected financial expenditures to characterise the level of government intervention [39].

3.2.2. Empirical Results Analysis of Influencing Factors

Through the overall Moran’s I spatial correlation analysis of the ULGUE in the study area, the Lanzhou-Xining urban agglomeration was found to have a certain spatial correlation. The following section presents an empirical analysis of the influencing factors and spillover effects on the ULGUE of the Lanzhou-Xining urban agglomeration through the spatial Durbin model. To further determine whether random or fixed effects should be selected, the Hausman test [40] was conducted on variables with a statistical value of 14.56. The significance test at the 5% level showed that the spatial Durbin model with fixed effects should be selected. The LR test was used to further determine which fixed effects should be selected [41]. In the original hypothesis, the LR statistics of individual time double fixed effects, which can be reduced to a single fixed effect, were 14.07 and 174.75, respectively. The p-values failed the test and passed the significance test at the 1% level, indicating that the spatial Durbin model of spatial fixed effects should be selected. The regression results are presented in

Table 4. Regression results of spatial Durbin model for factors influencing the ULGUE of Lanzhou-Xining urban agglomeration in the Yellow River Basin.

Variable Name	Coefficient	Std. Error	Z Score	p-Value
density	0.138 **	0.057	2.43	0.015
is	0.022 ***	0.008	2.68	0.007
pgdp	0.043 ***	0.014	3.18	0.001
er	0.345 ***	0.109	3.16	0.002
ur	−0.031	0.066	−0.47	0.635
gov	−0.014	0.431	−0.31	0.754
r	0.171 *	0.090	1.89	0.059
sigma2_e	0.028 ***	0.003	10.78	0.000

Note: This result uses the spatial weight matrix of spatial adjacency. *** p ≤ 0.01, ** p ≤ 0.05, and * p ≤ 0.1.

.

The coefficient of the spatial lag term is positive and significant at the 10% level, indicating that there is a spatial correlation between the ULGUE of the Lanzhou-Xining urban agglomeration and that the improvement of the ULGUE of a county may have a positive impact on the ULGUE of adjacent areas. From the regression results, the coefficient of the spatial lag term of the explained variable is significantly less than 0. This indicates that the influence of the independent variable on the dependent variable cannot be explained simply by the regression coefficient [42], and its marginal effect cannot be directly reflected. Therefore, the influence of the independent variable on the ULGUE is decomposed into direct, indirect, and total effects by the partial differential decomposition method. Further details are provided in

Table 5. The results of the partial differential regression.

Variable	Total Effect	Direct Effect	Indirect Effect
Density	0.035	0.136 **	−0.101
	(0.232)	(0.060)	(0.205)
Is	0.040	0.023 ***	0.018
	(0.027)	(0.008)	(0.025)
Pgdp	0.096 **	0.046 ***	0.050
	(0.039)	(0.013)	(0.035)
Er	−0.089	0.330 ***	−0.419 **
	(0.246)	(0.107)	(0.207)
Ur	0.124	−0.027	0.150
	(0.130)	(0.064)	(0.113)
Gov	0.183 ***	−0.006	0.188 **
	(0.061)	(0.041)	(0.076)

Note: *** p ≤ 0.01, ** p ≤ 0.05, and * p ≤ 0.1. The brackets are the standard errors.

.

• Population agglomeration (density): The total effect coefficient of the population concentration level is 0.035, indicating that population concentration has a positive impact on the ULGUE. The direct effect coefficient is 0.136, passing the significance test at the 5% level, which shows that the population agglomeration of the Lanzhou-Xining urban agglomeration can improve the ULGUE of the region. Additionally, the increase in population density will result in the agglomeration effect, which will promote industrial agglomeration and bring about economies of scale. The increase in population can promote land development. For example, Lanzhou and Xining, as the central cities of the Lanzhou-Xining urban agglomeration, have attracted more capital, talent, technology, and other resources. The spillover effect coefficient is −0.101, which means the unbalanced regional development will cause the population to flow from economically underdeveloped areas to economically developed areas. The siphon effect of Lanzhou and Xining has expanded the difference between the core city and the surrounding areas, reducing the development potential of adjacent areas. Simultaneously, low-end industries are transferred to adjacent areas. This increases the load-bearing pressure. As ecological space shrinks, it is difficult to improve ULGUE;
• Industrial structure optimisation (is): The total effect coefficient of the industrial structure optimisation level is 0.040, but it did not pass the significance level test, which indicates that the industrial structure optimisation of each county in the Lanzhou-Xining urban agglomeration has a potentially positive impact on the ULGUE. The direct effect coefficient is 0.023, passing the significance test at the 1% level. The optimisation of the industrial structure can promote the optimal allocation of land resources and the intensive and economical use of land. The spillover effect coefficient is 0.018. Compared to the eastern coastal areas, there are more resource-based cities in the western region at this stage. However, the industrial structure in most regions remains low. The secondary industry still occupies the main position, and its development and transformation are slow. This situation weakens the role of industrial structure in driving ULGUE;
• Economic development level (pgdp): All effect coefficients of the economic development level are positive. The total effect coefficient is 0.096, passing the 5% significance level test. The direct effect coefficient is 0.046, passing the 1% significance level test. The spillover effect coefficient is 0.050. This shows that the improvement in the economic development level will improve the overall ULGUE of counties and districts in the Lanzhou-Xining urban agglomeration. Thus, the direct effect is more obvious than the spillover effect. To some extent, economic development promotes industrial agglomeration and optimisation of the industrial structure. This could promote a compact land layout. The leading industry has gradually changed from a secondary industry to a service industry. This can reduce environmental pollution. Better economic capacity can also promote government expenditure on various services, such as environmental protection, governance, and land-use planning, promoting the improvement of the ULGUE in the region. At the same time, the improvement in the economic development level will stimulate the absorption and conversion of resources in neighbouring regions through the trickle-down effect and drive the improvement of the ULGUE in neighbouring regions;
• Environmental regulation (er): The total effect coefficient of environmental regulation was −0.089. This shows that environmental regulation has a potential negative impact on the ULGUE of the Lanzhou-Xining urban agglomeration, but it is not significant. The direct effect coefficient was 0.330, passing the 1% significance level test. The higher the level of environmental regulation in this region, the higher the level of land green use. The reason may be that the concept of the Lanzhou-Xining urban agglomeration sacrificing the ecological environment for improving the economic development level is still deeply rooted at this stage. The improvement of the environmental regulation level will force highly polluting enterprises to transform, optimise, and enhance the technology of pollution control. The spillover effect coefficient is −0.419 and passed the significance level test of 5%, indicating that the improvement in the environmental regulation level in this region will have a negative impact on the surrounding areas. A possible reason is that the improvement in the environmental regulation level in this region causes the transfer of pollution-intensive industries to adjacent areas;
• Urbanisation level (ur): The total effect coefficient of urbanisation was 0.124. The direct effect coefficient was −0.027. The spillover effect coefficient is 0.150. All of these failed the significance level test. This shows that the improvement in urbanisation level cannot directly lead to an improvement in ULGUE. The influx of a large number of people will cause the expansion of urban construction land. The spatial expansion over a short period of time will lead to uncertainty in the land layout and occupy the ecological space. Land resources cannot be used in an orderly manner because of wastage or idleness. The ecological environment of Lanzhou-Xining urban agglomeration is fragile. Rapid urbanisation will cause environmental pollution and increase disordered land use. Adaptability to the population size and industrial development of the urban agglomeration is not high, which reduces ULGUE. The improvement in urbanisation level has not yet played a statistically significant role in ULGUE in adjacent areas;
• Government intervention (gov): The total effect coefficient of government intervention on ULGUE was 0.183, passing the 1% significance level test. The direct effect coefficient is −0.006, and government intervention may have a negative impact on itself. The green use of land is significantly affected by the will of the government, rules, and regulations. There is a competitive relationship between the government departments. Some governments may promote their own economic development by introducing land investment and other means, resulting in the disorderly use of land and a waste of resources. Disordered competition had a negative impact on ULGUE. The spillover effect coefficient is 0.188, which passes the 5% significance level test. This may be because the Lanzhou-Xining urban agglomeration covers a large area. To some extent, the development planning and land resource allocation of each county and district still depend on government regulations.

3.2.3. Robustness Test

This study shows the regression results based on the spatial adjacency weight matrix. The spatial econometric model is highly sensitive to the spatial weight matrix. Therefore, the spatial weight matrix of geographical distance, economic distance, and economic geographical distance are selected for the robustness test.

Table 6. The results of the robustness test.

Variable	Spatial Weight Matrix of Geographical Distance	Spatial Weight Matrix of Economic Distance	Spatial Weight Matrix of Economic Geographical Distance
density	0.129 **	0.106 *	0.139 **
	(2.19)	(1.82)	(2.44)
is	0.023 ***	0.027 ***	0.024 ***
	(2.62)	(3.19)	(2.82)
pgdp	0.048 ***	0.038 ***	0.049 ***
	(3.37)	(2.75)	(3.47)
er	0.347 ***	0.323 ***	0.325 ***
	(3.12)	(2.90)	(2.97)
ur	−0.036	−0.006	−0.019
	(−0.55)	(−0.09)	(−0.27)
gov	−0.005	−0.049 *	−0.052
	(−0.13)	(−1.76)	(−0.92)

Note: ***, **, and * are significant at the 1%, 5%, and 10% levels, respectively, and z-statistics are shown in brackets.

presents the regression results. The regression results show that because there are many variables involved in the model, the significance level of the estimated coefficients of the individual variables in the results of the three spatial weight matrices has changed. The other results are consistent with those above. Thus, the conclusion is similar to that above. It can be considered that the regression results in this study are robust.

4. Discussion and Policy Implications

4.1. Discussion

Compared with the traditional measurement of urban land-use efficiency [43], this study uses the super-efficiency SBM model and adds carbon emissions and other undesirable outputs to consider the impact of the ecological environment, which can make up for the deficiencies in input relaxation variables and output relaxation variables so as to focus on ecological protection and reduce pollutant emissions in the process of land use.

This study focuses on green use of the land and its spatial spillover effects in counties with urban agglomeration. It is helpful to understand the small-scale urban land green use development model. The research results show that at present, the difference of land green use efficiency in Lanzhou-Xining urban agglomeration is obvious, the imbalance of regional development is prominent, and the provincial capital cities Lanzhou and Xining have high ULGUE, which has not yet fully formed a level-by-level driving role. Although the role of radiation and driving is limited, there is still a certain space spillover effect. For example, the optimization of industrial structure and the improvement of economic development level can not only promote the green use of local land but also drive the development of surrounding areas. This is consistent with the research conclusions of some scholars [19]. The reason may be that according to the First Law of Geography [44], these cities are closer in spatial location. In the future, we should fully release the positive spillover effect of the core cities of the urban agglomeration, avoid the low-lying cities around the cities with high efficiency from becoming environmental pollution shelters in the process of green land-use development, and promote the balanced development of regional integration.

There are still many deficiencies in the study. The efficiency of land green use is the result of the comprehensive effect of multiple factors. This paper only selects some driving factors when discussing the mechanism and focuses on the impact of a single factor on land green use without involving the interaction between various factors.

Furthermore, the analysis of ULGUE in this study remains at the macro level, and a more micro perspective has not been explored. For example, land can be classified according to the land-use type. Some scholars divided the cultivated land into five different types of functional areas and studied the cultivated land-use efficiency of Zhejiang Province [45]. It can be further discussed on this basis in the future.

4.2. Policy Implications

According to the research results, the following countermeasures are proposed:

(1)

Promote the green transformation of the industry: The optimization of industrial structure and the improvement of economic development level will have a positive impact on the green land-use efficiency of the Lanzhou-Xining urban agglomeration. Therefore, industrial development must aim at saving resources and protecting the environment, promoting industrial upgrading and innovation, accelerating the development from labour-intensive industries to knowledge- and technology-intensive industries, and increasing support for green development, production, and technology research and development of enterprises as well as raising enterprise pollution standards and environmental protection access threshold;

(2)

Advocate ecological environmental friendliness: The economic development mode of high energy consumption and high pollution in cities is the main reason for the increase of carbon emissions, air pollution, and water pollution from urban land use. Therefore, we should build a land carbon emission trading market, accelerate the transformation of the economic development model, and explore a low-carbon circular economy;

(3)

Regulate urban expansion: The results of the spatial econometric model show that although urbanization is conducive to the improvement of ULGUE, population gathering has a negative spillover effect on adjacent areas. Based on this, the future direction of the urbanization of the Lanzhou-Xining urban agglomeration should be to avoid the impact of siphon effect, strive to promote the implementation of local urbanization, optimize the allocation of urban land resources, and encourage the development of low land-consumption industries;

(4)

Take differentiated management measures according to local conditions: For the high-value area of the Lanzhou-Xining urban agglomeration with ULGUE, the radiation and driving ability should be strengthened to maintain the advantage of land resource allocation. For the low-value area, the development gap between regions should be squarely faced, and the coordinated development with surrounding counties and districts should be strengthened.

5. Conclusions

In this study, we quantified the ULGUE of 39 counties of the Lanzhou-Xining urban agglomeration from 2015 to 2020 based on the super-efficiency SBM model and considered the undesirable output of land green use. Exploratory spatial data analysis and other methods were used to analyse spatial differentiation characteristics. The spatial Durbin model was also used to analyse its influencing factors. The main conclusions are as follows:

(1)

During the study period, the average ULGUE of the Lanzhou-Xining urban agglomeration in the Yellow River Basin showed an overall upward trend. High-value areas increased and are developing well. However, there is still much room for improvement. The ULGUE presents an obvious spatial positive autocorrelation. The ULGUE of provincial capital cities is higher than non-capital cities, and the efficiency of the city was significantly higher than that of the counties. The high-value areas are mainly distributed among the municipal districts of prefecture-level cities, and the scope of the municipal districts gradually expanded during the study period. Low-value areas were mainly distributed in the north-eastern and southern counties of the urban agglomeration. Overall, the ULGUE in the west was higher than that in the east;

(2)

The results of the spatial Durbin model showed that an increase in the population concentration level can promote the ULGUE in this region. However, it produces a negative spillover effect on the adjacent areas. There was a positive correlation between the level of industrial structure optimisation and the ULGUE in the Yellow River Basin. The improvement in the economic development level will have a positive impact on the ULGUE, which will have a direct and significant impact on itself. A strong economic foundation is the premise for promoting new urbanisation and land-use renovation. There is a potential negative correlation between environmental regulation and the ULGUE. The improvement in urbanisation level has a positive impact on the land green use of the whole Lanzhou-Xining urban agglomeration. The steady progress of urbanisation will promote an increase in land output benefits and improve the overall ULGUE. Government intervention may have a negative impact on itself but a positive spillover effect on neighbouring areas.