Spatiotemporal Evolution and Driving Mechanisms of Urban Ecological Asset Utilization Efficiency from a “Technology-Scale-Structure” Perspective

Abstract

This study focuses on Hohhot (the capital city of Inner Mongolia Autonomous Region, northern China), a representative arid-semi-arid town in northern China. Against the backdrop of concurrent rapid urbanization and ecological constraints, it undertakes a systematic investigation into the spatiotemporal evolution and driving mechanisms of ecological asset utilization efficiency, aiming to furnish scientific evidence for sustainable development in ecologically fragile urban areas. Employing a “technology-scale-structure” analytical framework and constructing an “input-output-benefit” evaluation system, this research integrates the super-efficiency slack-based measure (SBM) model with spatial analysis methodologies to conduct multidimensional assessments of ecological asset utilization efficiency across all administrative districts and counties from 2000 to 2020. Empirical results demonstrate an overall upward trajectory in Hohhot’s ecological asset utilization efficiency, with comprehensive efficiency increasing from 1.132 in 2000 to 1.397 in 2020. However, pure technical efficiency and scale efficiency exhibit significant asynchrony, reflecting inherent tensions between technological advancement and scale expansion. Spatially, efficiency distribution manifests substantial spatial clustering and heterogeneity, with identified hotspots demonstrating temporal migration patterns. Peripheral counties exhibit distinct “technological isolation” phenomena and diseconomies of scale. Mechanism analysis reveals that industrial structure optimization constitutes the primary driver of efficiency enhancement, while the catalytic effects of economic development and governmental investment exhibit diminishing marginal returns. Urbanization maintains a moderate influence, transitioning from extensive spatial expansion toward intensive functional upgrading. This study recommends a synergistic enhancement of ecological asset utilization efficiency through strategic pathways, including the following: First, advancing green industrial transformation. Second, establishing regional technology-sharing platforms. Third, implementing systematic ecological compensation mechanisms. Fourth, adopting spatially differentiated governance approaches. These measures are projected to foster coordinated environmental and economic development. This research provides theoretical underpinnings and policy implications for urban ecological asset management in arid and semi-arid regions globally.

1. Introduction

Urban ecological assets represent the core components of urban ecosystems, providing essential material foundations and service functions for socioeconomic development and human well-being. Their utilization efficiency directly influences urban sustainability [1]. Recent accelerated urbanization has driven the rapid expansion of urban construction land, a prominent feature of China’s land-use change [2]. Such expansion frequently occurs without orderly regulation, readily undermining ecosystem stability and service capacity [3]. China’s economic development has entered a “new normal” phase, characterized by the “triple overlap” of (1) economic growth transition, (2) structural adjustment pains, (3) digestion of previous stimulus policies [4]. Within this context, resource and environmental constraints intensify significantly. Persistent reliance on traditional development pathways and entrenched conceptual paradigms necessitates urgent improvement in development quality and resource utilization efficiency [5].

Particularly in the arid and semi-arid regions of northern China, urban development encounters heightened ecological constraints. Taking Hohhot as a representative case, the region faces severe water scarcity and fragile ecosystems, where grassland degradation and soil erosion pose prominent environmental challenges. Concurrently, rapid urbanization progressively compresses ecological spaces, intensifying resource supply-demand imbalances and imposing significant constraints on regional ecological security and sustainable development. Consequently, research into the utilization efficiency of urban ecological assets within ecologically fragile regions possesses significant theoretical and practical urgency.

Research on the efficiency of urban ecological asset utilization constitutes a focal area within disciplines such as ecological economics, urban ecology, and sustainable development. This research aims to quantify the transformation efficiency between the supply capacity of urban ecosystems (ecological assets) and the actual services delivered for human well-being (utilization benefits). Its core objective is to identify strategies for maximizing socioeconomic benefits while minimizing ecological resource consumption and environmental impact, thereby propelling cities towards sustainable development [6,7,8].The conceptualisation of ‘ecological assets’ initially emphasized natural capital stocks (e.g., forest area, water resources, biodiversity) [9,10,11]. In recent decades, the focus has progressively shifted towards the dynamic flow and multidimensional value of ecosystem services. These services encompass provisioning services (e.g., food, water), regulating services (e.g., climate regulation, pollution purification), supporting services (e.g., soil formation), and cultural services (e.g., recreational tourism) [12,13]. This conceptual evolution underscores that the core value of ecological assets resides not merely in their static presence but fundamentally in their capacity to generate and deliver ecosystem services continuously.

Domestic research demonstrates strong policy-driven characteristics, being closely aligned with national strategies such as ecological civilization development, the dual carbon goals, new urbanization, and high-quality development [14,15,16]. Methodologically, energy-value analysis, ecological footprint assessment, material flow analysis, and geospatial technologies (including GIS, remote sensing, and satellite imagery) have been widely employed to map the spatial distribution of ecological assets and evaluate the alignment between ecosystem service supply and demand [17,18,19,20]. However, existing research predominantly focuses on humid eastern regions or basin scales, with limited systematic investigations conducted in arid and semi-arid urban areas, particularly the city of Hohhot. While studies within Inner Mongolia or the Yellow River basin have explored land-use changes, ecosystem service valuation, spatio-temporal evolution patterns, and driving factors, these investigations primarily emphasize ecological security assessments or single-factor analyses [21,22,23]. Within Hohhot or comparable arid/semi-arid cities, three principal limitations persist: firstly, the absence of long-term dynamic efficiency evaluations; secondly, insufficient multi-scale analysis of driving mechanisms; and thirdly, inadequate depth in characterizing spatial heterogeneity and exploring its underlying mechanisms.

In contrast, international research prioritizes advancing theoretical frameworks and integrating interdisciplinary approaches. Applications of foundational theories—including ecosystem service assessment (MA, TEEB frameworks), urban metabolism theory, ecological economics, and sustainability science—demonstrate considerable maturity [24,25,26]. Furthermore, international scholars are pioneering the integration of diverse models (e.g., coupling ecosystem service assessments with economic models and land-use simulations), alongside leveraging big data analytics, artificial intelligence-driven forecasting, and multi-scenario simulation techniques [27,28]. Concurrently, research scales are becoming increasingly refined, with a growing emphasis on the efficiency of ecological asset allocation and environmental equity considerations at the neighbourhood and community levels [29,30,31,32].

In summary, recent domestic and international research has increasingly emphasized the need for systematic, comprehensive, and spatially explicit analyses, thereby fostering stronger interdisciplinary integration. Notable advances have been made, particularly in examining the interactions between ecological assets and socioeconomic factors. Nevertheless, ecological asset efficiency assessment continues to face persistent challenges, including inconsistent indicator systems and limited cross-study comparability. Moreover, investigations into the dynamic mechanisms governing utilization efficiency remain insufficiently developed.

To address these gaps, this study develops an analytical framework tailored to the ecological characteristics of Hohhot, integrating the “technology–scale–structure” perspective. We apply a super-efficiency SBM–DEA model to evaluate ecological asset utilization efficiency from 2000 to 2020. We employ global and local spatial autocorrelation techniques to characterize spatiotemporal patterns and identify key driving factors and their relative contributions using grey correlation analysis.

This work contributes by closing critical gaps in long-term assessment, multidimensional mechanism exploration, and spatial heterogeneity analysis in arid and semi-arid urban contexts. It provides a scientific basis for targeted ecological asset management, coordinated eco-economic development, and the formulation of sustainable regional policies.

The structure of this paper is as follows (Figure 1): The Section 2 outlines the study area, data sources, and analytical framework, including the super-efficiency SBM–DEA model, spatial autocorrelation methods, and grey relational analysis. Results detail the temporal dynamics and spatial differentiation of ecological asset utilization efficiency, elucidate its driving mechanisms, and highlight implications for ecological governance and policy. Discussion and Conclusions summarize the main findings, reflect on limitations, and propose directions for future research.

The key contributions of this study are threefold: First, it provides a two-decade, county-level dynamic assessment and spatially explicit analysis of ecological asset utilization efficiency in an arid–semi-arid urban setting. Second, it establishes an integrated framework that couples the “technology–scale–structure” dimensions with spatial statistics and correlation analysis to identify multidimensional drivers systematically. Third, it offers differentiated, evidence-based policy recommendations, supplying actionable insights for ecological asset management and sustainable urban development in environmentally fragile regions.

2. Materials and Methods

2.1. Overview of the Study Area

Hohhot City (Figure 2) is situated in the western Inner Mongolia Plateau at the eastern terminus of the Yin Mountains. The region’s topography is defined by the Daqingshan Mountains in the north, the Manhanshan Mountains in the southeast, and the Tumochuan Plain in the south, with elevations ranging from 875 to 2313 m and a general slope from northeast to southwest. The area has a temperate continental monsoon climate, characterized by mean annual temperatures of 2–6.7 °C and precipitation of 280–412 mm, increasing from north to south. Evaporation is highest from June to August, accentuating the region’s pronounced aridity and water scarcity. Hydrologically, the area belongs to the Yellow River basin; its main rivers, including the Dahan and Hun, display strong seasonality and limited self-purification capacity. Hasu Lake, the largest freshwater body in the region (32 km²), constitutes a critical habitat for migratory birds.

In terms of ecological assets, Hohhot demonstrates both considerable natural endowments and marked vulnerabilities. Agricultural land dominates the land-use structure, accounting for 78.79% (with arable land comprising 43.31%), followed by woodland (23.68%, largely shrubland), grassland (18.54%, primarily natural grassland), and wetlands (1.03%, concentrated in the Hasu Wetland). Forest coverage is 21.4%, with a growing stock of 55.50 m³/ha, reflecting relatively low vegetation carbon sequestration efficiency. Water scarcity is severe: per capita water availability is only 32.84 m³, well below the regional average. Groundwater is heavily overexploited, covering 1031.84 km² with an annual overdraft of 38.92 million m³. Mineral resources are dominated by niobium–tantalum and construction materials, mainly extracted using extensive methods with inadequate ecological restoration. Significant ecological challenges include soil erosion over 8207.63 km² (47.65% of the total area), driven by both wind and water processes; persistently poor water quality in the Dahan and Xiaohan Rivers (below Class V, primarily due to ammonia nitrogen and total phosphorus pollution); uneven distribution of urban green space, with tree-lined road coverage at only 67.15% compared to the national standard of 85%; and high ecological water consumption.

Socio-economically, Hohhot’s GDP reached CNY 280.07 billion in 2020, with a tertiary-dominated industrial structure. The economy is led by the energy and dairy sectors, with energy consumption by large-scale industrial enterprises reaching 13.792 million tonnes of standard coal—well above the regional average. The urbanization rate stands at 70.5%. However, ecological efficiency is constrained by three interrelated challenges. First, regional disparities in pure technical efficiency persist, as peripheral counties (e.g., Tuoketuo) continue to lag in technological adoption. Second, construction land has expanded by 112% over the past two decades (from 547 km² to 1162 km²), encroaching on ecological space and resulting in a decline in scale efficiency with an inverted V-shaped trajectory. Third, although industrial restructuring has become the primary driver of efficiency, energy-intensive industries still account for a substantial share, underscoring the urgent need for a more profound transition toward green development.

This regional overview provides a coherent framework that links ecological endowments, utilization conflicts, and socio-economic drivers, thereby laying the groundwork for subsequent dynamic assessments of ecological asset utilization efficiency and mechanism analysis.

2.2. Data Source

To ensure a comprehensive examination of the study area, this research draws on three main categories of data. (1) Socio-economic data were obtained from the Hohhot Statistical Yearbooks (2000, 2010, 2020), the Inner Mongolia Statistical Yearbook, the Hohhot Hydrology and Environment Bulletin, the Hohhot Environmental Bulletin, the Hohhot Agricultural Economic Statistical Ledger, and Hohhot Territorial Spatial Planning. (2) Geospatial data include land-use maps, elevation, slope, and aspect data for 2000, 2010, and 2020, derived from a Digital Elevation Model (DEM) provided by the Environmental Science Data Platform, Chinese Academy of Sciences (Environmental Science Data Platform Website: https://www.resdc.cn/). (3) Survey data were collected through multiple field investigations in Hohhot, supplemented by questionnaire surveys with local stakeholders and ecological experts familiar with the region.

To ensure data accuracy and reliability, all collected multivariate datasets underwent systematic pre-processing and quality control, comprising three key steps:

(1)

Missing values: Isolated gaps in socio-economic data were corrected using time-series linear interpolation or averaging with adjacent counties/districts to preserve continuity. For geospatial datasets, gaps were addressed through cross-validation with remotely sensed image interpretation; records with >1% missing values were excluded.

(2)

Outliers: Boxplot analysis was applied to detect anomalies across all indicators, with final assessments guided by Hohhot’s local context. Values identified as statistical or entry errors were replaced with the corresponding annual mean or median.

(3)

Standardization: To address differences in units, scales, and indicator orientations, all input–output variables and driving factors were normalized to the [0, 1] range using Min–Max normalization prior to DEA modelling and grey relational analysis. This ensured comparability and eliminated unit-dependent distortions. The normalization formula is the following:

$$R_{ij}^{\prime }={\displaystyle \frac{R_{ij}-R_{j\min }}{R_{j\max }-R_{j\min }}} \text{(Positive indicator)}$$

$$R_{ij}^{\prime }={\displaystyle \frac{R_{\max }-R_{ij}}{R_{j\max }-R_{j\min }}} \text{(Reverse indicator)}$$

In this formula, $R_{ij}^{\prime }$ denotes the normalized value of indicator $j$ for City $i$; $R_{ij}$ represents the observed value of indicator $j$ for City $i$; and $R_{j\max }$ and $R_{j\min }$ correspond to the maximum and minimum values of indicator $j$, respectively.

(4)

Data Reliability Assurance: Socio-economic data were primarily sourced from the Hohhot Statistical Yearbook, the Inner Mongolia Statistical Yearbook, and official bulletins and planning documents issued by government agencies, ensuring both authority and credibility. Geospatial datasets were derived from a Digital Elevation Model (DEM) and land-use remote sensing interpretations provided by the Computer Network Information Centre of the Chinese Academy of Sciences, with resolution and accuracy sufficient for mesoscale regional analysis. Critical datasets, including ecological land-use types and ecological asset valuations, were further validated using field surveys and expert interviews, enhancing the overall authenticity and reliability of the data.

2.3. Analytical Framework

(1)

Selection of Evaluation Indicators for Urban Ecological Asset Utilization Efficiency

Drawing on ecological efficiency theory, urban ecological asset utilization efficiency is defined as the system-level transformation efficiency achieved through the combined deployment of ecological resources (e.g., land, water, and energy) and economic production factors (capital and labour). This efficiency reflects the synergistic optimization of both socio-economic outputs and ecological–environmental benefits. Its core attributes include the following: systemicity (coordinating the triadic relationship among resources, environment, and economy), dynamism (capturing spatiotemporal variations in factor allocation), and sustainability (incorporating ecological carrying capacity constraints). Accordingly, this study selects indicators that strike a balance between representativeness and data accessibility and constructs a Data Envelopment Analysis (DEA) framework encompassing both input and output dimensions. Input indicators include ecological resource inputs (e.g., ecological land area) and economic factor inputs (total fixed-asset investment, non-agricultural labour force). In contrast, output indicators cover economic outputs (e.g., value-added of secondary and tertiary industries, fiscal revenue) and ecological benefits (ecological asset value). This framework thus captures the dual objectives of economic growth and ecological conservation.

(2)

Spatiotemporal Evolution Pattern Analysis Methodology

To examine the spatiotemporal dynamics of ecological asset utilization efficiency in Hohhot between 2000 and 2020, this study employs exploratory spatial data analysis (ESDA), integrating both global and local spatial autocorrelation metrics. The global Moran’s I index quantifies overall spatial correlation, whereas the local Gi index* identifies spatial clusters, including hotspots and coldspots. This methodological approach minimizes subjective bias and provides an objective depiction of agglomeration patterns and temporal evolution trends.

(3)

Identification of Driving Factors and Mechanism Analysis

Multiple interacting factors, including urbanization, industrial structure, government regulation, and economic development, shape urban ecological asset utilization efficiency. Five key driving indicators are selected (

Table 1. Determinants of the utilization efficiency of urban ecological assets in Hohhot.

Independent Variable	Definition	Index Calculation	Type of Argument
Economic development level	GDP per capita	Total GDP/Total population	Continuous variable
	Per capita fiscal revenue	Total fiscal revenue/Area of urban ecological land	Continuous variable
Industrial structure level	The proportion of the output value of the secondary and tertiary industries	The output value of the secondary and tertiary industries/the total GDP	Continuous variable
Macro-regulation by the government	Fixed-asset investment per area	Total fixed-asset investment/urban ecological land area	Continuous variable
Urbanization level	The urbanization rate	Total non-agricultural population/Total population	Continuous variable

): GDP per capita, share of secondary and tertiary industry output, fixed-asset investment per unit land area, fiscal revenue per unit land area, and urbanization rate. Grey relational analysis is then applied to assess the correlations between these factors and ecological efficiency across Hohhot’s nine districts and counties for 2000, 2010, and 2020, thereby elucidating the principal driving mechanisms and their spatiotemporal evolution characteristics.

2.4. Methods

(1)

Super-SBM Model Based on Data Envelopment Analysis [33]

This study adopts a nonparametric Data Envelopment Analysis (DEA) framework to evaluate the efficiency of urban ecological asset utilization. In particular, the super-efficiency Slack-Based Measure (Super-SBM) model is applied to calculate relative efficiency scores—comprehensive efficiency, pure technical efficiency, and scale efficiency—for each decision-making unit over the period 2000–2020.

The Super-SBM model effectively overcomes biases introduced by traditional radial and angular approaches, as it explicitly accounts for slack variables that capture both input redundancies and output shortfalls. To evaluate ecological asset efficiency across j regions, each inter-provincial region is treated as a distinct decision-making unit (DMU), with X₀ and Y₀ representing input and output variables, respectively. Each DMU is characterized by m input variables and r output variables: X_jm denotes the total quantity of the m input variables in region j, and Y_jr represents the total quantity of the r output variables across j regions. The weighting variable λ_j links all efficient points to construct the efficiency frontier, enabling the assessment of economies of scale across regions. The model is formally specified as follows:

$$\min \rho ={\displaystyle \frac{1+\frac{1}{m}{\displaystyle \sum _{i=1}^m{s}_i^{-}/x_{ik}}}{1-\frac{1}{s}{\displaystyle \sum _{r=1}^s{s}_i^{+}/y_{rk}}}}$$

When $\rho$ > 1 or when $\rho$ = 1 and $s^{+}=s^{-}=0$, the ecological efficiency of the city reaches a strong state; when ρ = 1 and $s^{+}\ne s^{-}\ne 0$, the ecological efficiency of the city reaches a weak state; when ρ < 1, the ecological efficiency of the city is in an ineffective state. The ecological efficiency of the city ρ is equal to the product of pure technological efficiency (PTE) and scale efficiency (SE):

$$\rho =PTE\times SE$$

This model rests on three key assumptions:

• Variable Returns to Scale (VRS): Decision-making units (DMUs) are assumed not to operate at an optimal scale, and scale effects influence their efficiency. This reflects the reality of Hohhot’s districts and counties, where developmental stages vary and pronounced scale differences exist. Accordingly, a VRS-based model is more appropriate.
• Input Orientation: Given the stringent ecological constraints in the study area—particularly regarding water resources and ecological land-use—the model assumes minimisation of inputs for a given output level. This orientation captures the potential for conserving ecological resources while maintaining current production levels.
• Non-radial and Non-angular Specification: Unlike traditional DEA models, the SBM framework discards radial and angular assumptions by directly incorporating slack variables for input surpluses and output shortfalls. This formulation provides a more accurate representation of efficiency, particularly in contexts where undesirable outputs are present.

Based on these assumptions, this study applies the super-efficiency Slack-Based Measure (Super-SBM) model, implemented in MaxDEA, to estimate the comprehensive efficiency, pure technical efficiency, and scale efficiency of urban ecological asset utilization in Hohhot (

Table 2. Ecological asset utilization efficiency and related explanations.

Ecological Asset Utilization Efficiency	Conceptual Connotation
Comprehensive efficiency	Eco-efficiency represents a holistic metric for evaluating the resource allocation and utilization efficiency of decision-making units. Mathematically, eco-efficiency can be expressed as the product of pure technical efficiency multiplied by scale efficiency.
Pure technical efficiency	Pure technical efficiency refers to the productive efficiency of a decision-making unit under managerial and technological constraints. In this study, it specifically denotes the capability to achieve maximal economic output while minimizing undesirable environmental outputs, given the resource inputs of each research unit.
Scale efficiency	Scale efficiency refers to production efficiency influenced by factors including the scale of decision-making units. In this study, this concept specifically denotes a city’s capacity to maximize output through resource allocation and collaboration with other administrative units.

). Spatial patterns of efficiency are subsequently examined using the spatial analysis module of ArcGIS 10.4.

(2)

Exploratory spatial analyses [34]

a.

The global Moran’s I index quantifies the degree of spatial autocorrelation, capturing both the similarity of neighbouring units and the extent to which adjacent regions exhibit comparable attribute values. It provides a statistical measure of spatial clustering or dispersion across the study area. The global Moran’s I is calculated using the following formula:

$$I={\displaystyle \frac{n{\displaystyle {\sum }_{i=1}^n{\displaystyle {\sum }_{j=1}^n{W}_{ij}\left(X_i-\overline{X}\right)\left(X_j-\overline{X}\right)}}}{S^2{\displaystyle {\sum }_{i=1}^n{\displaystyle {\sum }_{j=1}^n{W}_{ij}}}}}$$

In the formula, $n$ represents the total number of spatial units in the study area; $X_i$ and $X_j$ denote the observed values of units $i$ and, $j$ respectively; $W_{ij}$ is the element of the spatial weight matrix, assigned 1 for adjacent units and 0 otherwise; S² corresponds to the variance of the observed values; and $\overline{X}$ represents their mean.

b.

Hotspot Analysis $G_i^{*}$ Index is used to identify clusters of high (hotspots) and low (coldspots) values across spatial regions, thereby capturing patterns of local spatial autocorrelation. Its formula is given as follows:

$$G_i^{*}\left(d\right)={\displaystyle \sum _{i=1}^n{W}_{ij}\left(d\right)X_j}/{\displaystyle \sum _{i=1}^n{X}_j}$$

In the formula, $W_{ij}$ denotes the spatial weight matrix, where adjacent units are assigned a value of 1 and non-adjacent units 0. A significantly positive $G_i^{*}$ indicates that the values surrounding unit i are relatively high, forming a hotspot, whereas a significantly negative $G_i^{*}$ implies that the values around unit i are relatively low, forming a coldspot.

In this study, exploratory spatial analysis employs the Queen’s adjacency rule to construct the spatial weight matrix. Two districts are considered spatially adjacent (assigned a value of 1) if they share a common boundary or vertex; otherwise, they are non-adjacent (assigned a value of 0). This approach is justified by the spatial configuration of Hohhot’s districts and counties, which form contiguous geographical units where ecological and economic interactions—such as resource flows, pollution dispersion, and technology spillovers—are most likely to occur between directly neighbouring units. Compared with distance-threshold-based weight matrices, the adjacency rule more accurately captures the effects of geographical proximity while avoiding subjective bias from arbitrarily defined distance thresholds.

(3)

Grey correlation analysis model [35]

Grey Relational Analysis (GRA), rooted in grey system theory, provides a quantitative approach for characterizing system dynamics. It evaluates the degree of association between factors by measuring their geometric similarity. The method is robust to variations in sample size, computationally straightforward, and widely applicable. The procedure consists of the following steps:

a.

The analysis begins by defining the reference sequence X_ij and the comparison sequence X_0j (i = 1, 2, 3,…, m; j = 1, 2, 3, …, n).

b.

Apply the initialization method to normalize both the reference and comparison sequences using the following formula:

$$X_{ij}^{\prime }={\displaystyle \frac{X_{ij}}{X_i}}$$

c.

Calculate the grey correlation coefficients:

$$\delta ={\displaystyle \frac{1}{n}}{\displaystyle \sum _{j=1}^n\frac{\stackrel{\min }{i} \stackrel{\min }{j}\left|X_{0j}^{\prime }-X_{ij}^{\prime }\right|+\mu \stackrel{\max }{i} \stackrel{\max }{j}\left|X_{0j}^{\prime }-X_{ij}^{\prime }\right|}{\left|X_{0j}^{\prime }-X_{ij}^{\prime }\right|+\mu \stackrel{\max }{i} \stackrel{\max }{j}\left|X_{0j}^{\prime }-X_{ij}^{\prime }\right|}}$$

In the formula, $\stackrel{\min }{i} \stackrel{\min }{j}\left|X_{0j}^{\prime }-X_{ij}^{\prime }\right|$ and $\stackrel{\max }{i} \stackrel{\max }{j}\left|X_{0j}^{\prime }-X_{ij}^{\prime }\right|$ denote the minimum and maximum range values, respectively, and μ is the resolution coefficient, conventionally set to 0.5.

In this study, the exploratory spatial analysis employed the Queen’s adjacency rule to construct the spatial weight matrix. Two districts are considered spatially adjacent (assigned a value of 1) if they share a common boundary or vertex; otherwise, they are non-adjacent (assigned a value of 0). This approach is justified by the spatial configuration of Hohhot’s districts and counties, which form contiguous geographical units where ecological and economic interactions—such as resource flows, pollution dispersion, and technology spillovers—are most likely to occur between directly neighbouring units. Compared with distance-threshold-based weight matrices, the adjacency rule more accurately captures the effects of geographical proximity while avoiding subjective bias from arbitrarily defined distance thresholds.

3. Results

3.1. Evaluation Results of Urban Ecological Asset Utilization Efficiency in Hohhot City

This study applies a super-efficiency SBM model to evaluate comprehensive efficiency, pure technical efficiency, and scale efficiency of ecological asset utilization across Hohhot’s districts and counties in 2000, 2010, and 2020, with results reported in

Table 3. Hohhot ecological asset utilization efficiency (2000–2020).

Region Name	The Year 2000			The Year 2010			The Year 2020
	Combined Efficiency	Pure Technical Efficiency	Scale Efficiency	Combined Efficiency	Pure Technical Efficiency	Scale Efficiency	Combined Efficiency	Pure Technical Efficiency	Scale Efficiency
Horinger County	1.195	6.096	0.196	1.280	1.971	0.650	1.122	1.847	0.607
Hui Min District	1.223	7.072	0.173	1.137	1.284	0.886	2.376	3.018	0.787
Qingshuihe County	1.268	1.391	0.912	2.577	2.671	0.965	1.558	2.295	0.679
Saihan District	1.060	1.140	0.929	1.302	1.305	0.998	1.084	1.679	0.646
Tuomute Zuo Banner	1.117	1.279	0.873	1.369	2.616	0.523	1.495	7.930	0.189
Tuoketuo County	1.235	1.242	0.994	1.686	1.711	0.985	0.562	0.935	0.601
Wuchuan County	1.029	1.309	0.786	1.209	1.319	0.917	1.445	1.484	0.974
Xincheng District	1.325	1.241	1.068	1.291	1.158	1.115	1.872	3.692	0.507
Yuquan District	0.738	1.162	0.635	0.861	1.087	0.791	1.058	1.266	0.835
Mean value of Hohhot	1.132	2.437	0.730	1.412	1.680	0.870	1.397	2.683	0.647

. Between 2000 and 2020, ecological asset utilization efficiency exhibited distinct evolutionary patterns: a steady increase in overall efficiency, a V-shaped recovery in pure technical efficiency, and an inverted V-shaped decline in scale efficiency. These trends highlight the asynchronous development of different efficiency dimensions under the constraints of limited ecological resources.

(1)

Comprehensive Efficiency: Steady Improvement with Marked Spatial Variations

Between 2000 and 2020, Hohhot’s comprehensive urban ecological asset utilization efficiency exhibited an overall upward trend, with the citywide average increasing from 1.132 to 1.39, representing a 23.4% gain. Efficiency peaked at 1.4124 in 2010 before slightly declining, yet remained above initial levels, indicating enhanced ecological–economic synergy and resilience. Spatially, significant divergence persists and appears to be widening, as Xincheng District, Qingshuihe County, and Huimin District consistently maintain high efficiency, reflecting strong ecological endowments and synergistic advantages in economic, technological, and scale factors. In typical years, roughly 88.9% of districts and counties achieved DEA efficiency, primarily including Helingeer County, Huimin District, Qingshuihe County, Saihan District, Tumed Left Banner, Tuoketuo County, Wuchuan County, and Xincheng District. Conversely, Yuquan District exhibits persistent inefficiency, while Tuoketuo County shows periodic inefficiency, dropping to 0.562 in 2020. These patterns reveal structural constraints, including rapid urbanization and industrial agglomeration, which squeeze ecological land-use and lead to misalignments between resource allocation and ecological carrying capacity. High-efficiency zones correlate with superior ecological endowments and factor synergies, whereas low-efficiency zones follow a chain of “extensive scale expansion → ecological displacement → efficiency decline.” Enhancing efficiency in underperforming areas, therefore, requires replacing scale-driven expansion with factor efficiency and structural optimization. Implementing strict ecological land protection, promoting green technologies, and controlling industrial intensity can break the “expansion–squeeze–inefficiency” feedback loop, narrowing inter-county efficiency gaps while maintaining an upward trajectory.

(2)

Pure Technical Efficiency: V-Shaped Recovery Reflects Technological and Managerial Gains

Pure technical efficiency follows a pronounced V-shaped trajectory. From 2000 to 2010, it declined by approximately 31.1%, suggesting that urban expansion and management practices during this period failed to convert inputs effectively, thereby suppressing ecological asset utilization efficiency. Between 2010 and 2020, efficiency rebounded sharply by 59.7%, with the 2020 average rising to 2.683, exceeding the 2000 level of 2.439. This reflects substantial improvements in technology application and management optimization, enhancing resource conversion efficiency. Early declines were closely linked to extensive urban expansion and lagging ecological management.

In contrast, the subsequent recovery benefited from sustained ecological governance, accelerated adoption of green technologies, and progressive management reforms. Notably, Tuoketuo County’s pure technical efficiency remained below 1, illustrating an “island effect” in peripheral technology absorption and application, highlighting persistent spatial barriers to interregional technology diffusion. Addressing these challenges requires cross-county technology-sharing platforms, differentiated technology support, a “technical efficiency–ecological compensation” linkage, alignment of industrial green transformation with technological capacity, spatial optimization of technology investment, and strengthened monitoring and talent development. Such measures can transition pure technical efficiency from “isolated breakthroughs” to “holistic coordination,” supporting the sustainable enhancement of ecological asset utilization.

(3)

Scale Efficiency: Inverted V-Shaped Decline Indicates Diseconomies of Scale

Time-series analysis reveals an inverted V-shaped trajectory for scale efficiency. After peaking at 0.87 in 2010, scale efficiency declined to 0.647 by 2020, a 25.6% reduction. Newly developed urban districts exhibited scale efficiency above 1 in 2000 and 2010, indicating an optimal ecological–economic scale. By 2020, however, most districts and counties recorded scale efficiency below 0.65, reflecting increasing diseconomies of scale. This pattern closely parallels Hohhot’s 112% expansion of construction land over two decades, indicating that rapid urban growth failed to yield proportional ecological–economic efficiency gains. Instead, excessive development led to ecological fragmentation and exceeded the carrying capacity of urban ecosystems.

3.2. Analysis of the Spatial and Temporal Variability of the Utilization Efficiency of Urban Ecological Assets in Hohhot

(1)

Spatial autocorrelation analysis (Global Moran’s I)

The spatial distribution of ecological asset utilization efficiency in Hohhot exhibits pronounced agglomeration, shaped by the interplay of policy interventions, resource endowments, and regional economic development stages.

The global Moran’s I index for comprehensive efficiency steadily increased from 0.049 in 2000 to 0.320 in 2020, reflecting a strengthening positive spatial correlation and increasingly distinct clustering of high- and low-efficiency zones (

Table 4. Global Moran’s I Index for the utilization efficiency of ecological assets in Hohho, 2000, 2010, and 2020.

Indicator	Comprehensive Efficiency			Pure technical Efficiency			Scale Efficiency
	2000	2010	2020	2000	2010	2020	2000	2010	2020
Moran’s I	0.049	0.289	0.320	0.166	0.518	0.117	0.210	0.286	0.159
Z	3.28	16.27	9.84	5.29	15.69	3.84	6.58	8.80	5.07
Variance	0.00034	0.00034	0.00113	0.00113	0.00114	0.00113	0.00114	0.00114	0.00114

). This pattern mirrors the trends observed in other rapidly urbanizing regions of China, such as the Yangtze River Delta and Pearl River Delta. Early efficiency disparities were driven by variations in resource allocation and pilot policy programmes, while subsequent infrastructure improvements and regional integration reinforced spatial interconnections. As a representative of ecologically fragile northern China, Hohhot’s intensifying efficiency clustering is closely associated with initiatives such as the “Ecological Barrier Construction” and the Western Development Strategy. The notable post-2010 rise in Moran’s I (0.289 → 0.320) likely reflects coordinated development of the Hohhot–Baotou–Ordos urban cluster and implementation of ecological compensation mechanisms, which, while optimizing cross-regional resources, may have amplified core–periphery polarization.

For pure technical efficiency, the global Moran’s I follows an inverted U-shaped trajectory, progressing through intensification, peak, and decline phases (Table 4). This evolution reflects how technological policies and infrastructure upgrades initially amplified spatial unevenness and the phased nature of technology spillovers. Efficiency peaked in 2010, coinciding with the expansion of Hohhot’s High-Tech Development Zone and the concentration of innovative city initiatives, which reinforced core area advantages (e.g., Xincheng and Saihan Districts) and generated technological agglomeration hotspots. By 2020, however, pure technical efficiency declined markedly, indicating that, while technological diffusion disrupted prior agglomeration structures, peripheral counties (e.g., Tuoketuo and Wuchuan) remained unable to assimilate core technologies. Constraints, including talent shortages, inadequate infrastructure, and weak innovation ecosystems, produced a “technological island” effect, demonstrating that ecological fragility and developmental lags exacerbate barriers to technology diffusion.

Scale efficiency exhibits an inverted V-shaped trajectory (Table 4), characterized by rapid enhancement, a peak, and subsequent decline, highlighting its sensitivity to policy interventions while revealing sustainability limits. The 2010 peak corresponded to urban construction expansion policies, particularly the “new district development + land finance” model, which concentrated capital in central areas and temporarily boosted scale efficiency. By 2020, however, scale efficiency had fallen below initial levels, reflecting that extensive urban expansion had exceeded the ecological carrying capacity. Emerging challenges—including the over-extraction of water (averaging 38.92 million m³ annually) and grassland degradation affecting 47.65% of the total area—have intensified diseconomies of scale, constraining sustainable urban ecological development.

(2)

Hotspot analysis (Getis-Ord Gi*)

Based on local spatial autocorrelation analysis using the Gi* index, Hohhot’s urban ecological asset utilization efficiency exhibited pronounced hotspot and coldspot dynamics from 2000 to 2020. These patterns, however, reflect more than a simple “strong south–weak north” gradient or northward hotspot migration; they emerge from the combined influence of regional economic development models, policy orientations, and resource endowments.

The evolution of pure technical efficiency hotspots illustrates both technology diffusion and a persistent “island effect” (Figure 3). In 2000, hotspots were primarily concentrated in Helingeer County and southern Wuchuan County, whereas coldspots clustered in central urban areas, including the Xincheng and Saihan Districts. This distribution corresponded to Hohhot’s prevailing “peripheral resources–core consumption” development model: peripheral regions benefited from stronger ecological foundations and lower input intensity, generating high-efficiency zones. In contrast, core urban areas exhibited lower efficiency due to rapid urban expansion and extensive industrial activity. By 2010, hotspots shifted southward to Qingshuihe County and Tumed Left Banner, and coldspots expanded into the central and northern regions. This transition closely aligned with regional and national policies—particularly the “ecological resettlement and grassland restoration” programme implemented after 2008—which facilitated the adoption of technology and ecological restoration in southern areas, producing new hotspots. By 2020, however, hotspots had contracted substantially, with only Tumed Left Banner maintaining a stable high-efficiency zone. The dissolution of other hotspots indicates that, despite improvements in technological efficiency, diffusion remained limited, producing a pronounced “technological island effect.” Compared to eastern coastal cities, Hohhot continues to struggle in sustaining technological efficiency clustering, highlighting the absence of scalable mechanisms for technology dissemination.

The evolution of scale efficiency hotspots in Hohhot reveals alternating patterns of economies of scale and diseconomies of scale (Figure 4). In 2000, hotspots were concentrated in Xincheng and Saihan Districts, while coldspots clustered in Helingeer County, producing a “core advantage–peripheral inefficiency” pattern. By 2010, hotspots expanded further within Xincheng and Saihan Districts, reflecting the impacts of large-scale infrastructure development and new district expansion policies. By 2020, however, hotspots shifted northward to Wuchuan County, and core urban areas experienced weakened or vanished scale efficiency. This trajectory suggests that initial expansion generated economies of scale; however, as urban land-use exceeded ecological carrying capacity, scale efficiency declined sharply, resulting in diseconomies of scale. Hohhot thus exemplifies an “expansion trap,” where physical growth fails to enhance efficiency and instead precipitates decline due to constrained ecological space and structural industrial imbalances.

The evolution of comprehensive efficiency hotspots in Hohhot demonstrates both synergistic effects and pronounced regional differentiation (Figure 5). Between 2000 and 2010, efficiency exhibited a stable “strong south–weak north” pattern, with the southern region gaining synergistic advantages through industrial upgrading and ecological restoration. By 2020, however, hotspots shifted markedly northward: former central coldspots became new high-efficiency zones, while southern advantages diminished. This spatial rebalancing reflects the dynamic interplay between ecological and economic factors. Synergistic effects in the south weakened due to overexploitation and population concentration. In contrast, northern regions benefited from policy interventions, such as the Grain-for-Green Program and poverty alleviation initiatives, as well as infrastructure improvements. The transition from a “single-core-driven” to a “multi-point balanced” development pattern aligns with theoretical expectations for ecological compensation and regionally balanced development in arid northern cities.

3.3. Analysis of Driving Factors for Changes in the Pattern of Utilization Efficiency of Hohhot’s Urban Ecological Assets

Analysis of the spatiotemporal patterns of ecological asset utilization efficiency in Hohhot indicates that the complex interplay of multiple factors shapes its evolution. Prior studies suggest that urbanization directly affects the efficiency of ecological space allocation by altering land-use structures and landscape patterns. Economic development determines the intensity of factor inputs, thereby constraining the utilization of ecological assets. Optimization of industrial structure can enhance ecological efficiency by improving resource use and reducing environmental pressures. Meanwhile, market mechanisms and government regulation jointly influence resource allocation patterns. Additionally, China’s demographic transition and the strategic shift in land policy from “incremental expansion” to “stock optimisation” provide a macro-level context for interpreting changes in ecological efficiency.

To identify key drivers, this study synthesizes existing research to construct a driver system centred on four core dimensions: economic development (per capita GDP, fiscal revenue per unit of land area), industrial structure (share of secondary and tertiary industries), government regulation (fixed-asset investment per unit of land area), and urbanization (urbanization rate). Grey relational analysis was applied to quantify the strength of each factor’s association with ecological efficiency (

Table 5. Grey correlation degrees between driving factors and urban ecological asset utilization efficiency in 2000, 2010, and 2020.

Influence Factor	Driving Force Indicator	The Year 2000		The Year 2010		The Year 2020
		Grey Relational Degree	Associated Level	Grey Relational Degree	Associated Level	Grey Relational Degree	Associated Level
Economic development level	Per capita GDP	0.786	Strong	0.799	Strong	0.667	Medium
	Per capita fiscal revenue	0.781	Strong	0.767	Strong	0.617	Medium
Industrial structure level	The proportion of the second and third output values	0.884	Strong	0.889	Strong	0.788	Strong
Governmental macroeconomic regulation and control	Fixed-asset investment per area	0.605	Medium	0.712	Medium	0.553	Medium
Urbanization level	Urbanization rate	0.600	Medium	0.680	Medium	0.660	Medium

Note: Grey relational coefficients are classified into three levels: weak [0, 0.35], moderate (0.35, 0.75), and strong (0.75, 1.00).

). Results indicate that all factors exhibited correlation coefficients above 0.5 between 2000 and 2020, confirming the sustained influence of these variables and supporting the validity of the proposed driver model.

Temporally, the driving strength of each factor exhibits an overall declining trend (2000 > 2010 > 2020), reflecting a structural shift in the mechanisms influencing ecological efficiency as development stages and policy contexts evolve. Industrial structure consistently emerges as the strongest driver, indicating that increasing the share of secondary and, particularly, tertiary industries exert a stable positive effect on ecological efficiency. This finding aligns with Hohhot’s emphasis on green and high-end industrial transformation and underscores the general role of industrial upgrading in promoting ecological efficiency.

In contrast, the influence of economic development has weakened significantly: correlations for per capita GDP and fiscal revenue per unit area declined from 0.786 and 0.781 to 0.667 and 0.617, respectively. This suggests that traditional scale-driven economic growth is gradually giving way to high-quality development, a shift reinforced by post-2010 restrictions on energy-intensive industries under China’s ecological civilization initiatives and dual-carbon goals. Similarly, government regulation has shown a limited and diminishing impact, indicating that reliance on investment expansion alone cannot sustainably enhance ecological efficiency and may even exacerbate diseconomies of scale. Urbanization exhibits an “initial rise–subsequent stabilization” pattern, remaining an important driver. However, its mode has shifted from extensive expansion to intensive upgrading, consistent with Hohhot’s compact urban development and green infrastructure strategies.

In summary, Hohhot’s ecological asset utilization efficiency has transitioned from a primarily “scale-driven” model to a synergistic “technology–scale–structure” paradigm. Future policy priorities should focus on industrial green transformation, optimization of investment structures, and enhancement of urbanization quality to achieve high-level ecological–economic coordination. To further investigate the spatial heterogeneity and evolution of these drivers, and to fully capture their spatiotemporal dynamics, the analyses below were conducted.

(1)

Influence of government economic regulation on the utilization efficiency of urban ecological assets

Government economic regulations, reflected by the intensity of fixed-asset investment per unit of land, mediates ecological asset management. Grey relational analysis indicates that this factor maintains a moderate correlation with ecological efficiency. Although it experienced a temporary increase in 2010, driven by the Western Development Strategy and ecological compensation policies, its influence declined below initial levels by 2020, reflecting limited driving effectiveness and diminishing marginal returns.

The spatial distribution of fixed-asset investment per unit area evolved from a “northwest high–southeast low” pattern to a centralized cluster (Figure 6). In 2000, high-value zones were concentrated in ecologically fragile areas such as Wuchuan County and Tumed Left Banner, reflecting their reliance on government ecological infrastructure. By 2010, high-value areas expanded to central districts, including Huimin and Xincheng, consistent with capital concentration under the “land finance” and new district development models. By 2020, high-value zones had contracted again to central regions, such as Helingeer County and Huimin District, indicating a spatial shift in policy-responsive sensitive areas. Notably, regions with lower ecological efficiency, such as Tuoketuo County and Yuquan District, exhibited higher correlations with government investment, highlighting both the dependence on external funding and the urgent need for ecological restoration in underdeveloped areas. This also suggests that single-source capital investment, without complementary technical and managerial measures, struggles to sustain improvements in ecological efficiency.

Government investment impacts ecological efficiency via dual pathways. Targeted investments in ecological infrastructure—for example, restoration projects in Qingshuihe County—directly enhance resource-intensive utilization. Conversely, extensive infrastructure expansion, such as road construction in new districts or industrial park development, can exacerbate ecological fragmentation and trigger diseconomies of scale. These findings suggest that regulatory strategies should shift from “scale expansion” to “structural optimisation,” prioritizing green infrastructure and ecological technology deployment. Such an approach can mitigate diminishing returns while fostering synergistic gains in ecological benefits and capital allocation efficiency.

(2)

Impact of economic development level on urban ecological asset utilization efficiency

Economic development, measured by per capita GDP and fiscal revenue per unit of land, exhibits a general weakening in its driving influence on ecological efficiency, reflecting Hohhot’s gradual transition from a scale-oriented to a quality-oriented growth model. The correlation between per capita GDP and ecological efficiency has declined from strong to moderate. In contrast, the correlation with fiscal revenue per unit of land has decreased even more substantially, indicating the reduced impact of economic output per unit area on ecological outcomes.

High-value zones for per capita GDP have shifted spatially from resource-dependent counties, such as Tumed Left Banner and Tuoketuo County in 2000, to service-oriented core districts, including Huimin and Xincheng, by 2020 (Figure 7). This evolution mirrors Hohhot’s industrial upgrading trajectory: early economic growth relied heavily on resource extraction and energy-intensive industries (e.g., Tuoketuo’s thermal power cluster), which simultaneously expanded the economy and degraded ecological assets. In contrast, central urban districts have progressively decoupled economic growth from ecological impact through the development of service sectors such as finance and innovation. High-value areas of fiscal revenue per unit of land consistently align with ecological efficiency hotspots (Figure 8). For example, in 2020, Huimin and Saihan Districts—generating high fiscal revenues—also corresponded to high-efficiency ecological zones, confirming that economically concentrated areas with robust fiscal capacity can effectively support ecological management.

Nevertheless, the overall decline in economic driver strength indicates that traditional GDP-focused growth models are insufficient to sustain improvements in ecological efficiency. In ecologically sensitive regions, such as Wuchuan and Qingshuihe Counties, continued reliance on resource extraction or energy-intensive industries risks exceeding ecological carrying capacities. Future development should prioritize a quality-driven economic pathway, leveraging green industrial innovation and ecological compensation mechanisms to achieve synergistic economic and ecological progress.

(3)

Influence of industrial structure on the utilization efficiency of urban ecological assets

Industrial structure, measured by the share of secondary and tertiary industry value-added in GDP, emerged as the primary driver of urban ecological asset utilization efficiency, exhibiting the most stable and significant correlation from 2000 to 2020. Its influence shifted over time from an early “inverse suppression” effect to a consistent “positive promotion,” confirming that industrial upgrading and optimization actively enhance ecological efficiency.

High-value zones of industrial structure–ecological efficiency correlation demonstrate pronounced spatial clustering. In 2000, these zones were scattered across regions such as Wuchuan County and Xincheng District; by 2020, they had consolidated into extensive clusters encompassing Huimin District, Yuquan District, Tuoketuo County, and others (Figure 9), largely overlapping with ecological efficiency hotspots. This pattern aligns with Hohhot’s “industry in the south, tourism in the north” strategy: southern regions, including Tuoketuo and Helingeer Counties, improved ecological efficiency through industrial greening initiatives—such as wastewater recycling and clean production technologies—while northern areas, including Wuchuan and Qingshuihe Counties, enhanced ecological asset value via grassland eco-tourism and low-carbon agriculture.

Mechanistically, industrial structure enhances ecological efficiency through three main pathways: first, technological progress improves the utilization efficiency of ecological assets; second, capital and technology substitute for natural resource inputs, reducing ecological depletion; third, industrial agglomeration fosters circular economic development, optimizing the spatial allocation of ecological assets. Moving forward, Hohhot should further promote industrial greening and expand the ecological services sector, prioritizing the development of “ecology-plus” integrated industries and reinforcing the synergistic integration of industrial and ecological policies to achieve coordinated ecological–economic development.

(4)

The Impact of Urbanization Levels on the Utilization Efficiency of Urban Ecological Assets

The correlation between urbanization levels and ecological asset utilization efficiency has consistently remained moderate, following a pattern of “initial increase followed by stabilization.” This reflects a transition in urbanization from extensive expansion toward quality-oriented development. The correlation peaked in 2010 (0.680), coinciding with a rapid phase of urbanization during which population concentration and land expansion for construction exerted significant pressure on ecological resources. By 2020, the correlation had declined slightly to 0.660, but remained above the 2000 levels, indicating that urbanization had entered a quality-driven phase, with compact urban construction and ecological restoration measures beginning to yield tangible benefits.

Spatially, high-value zones of the urbanization–ecological efficiency correlation evolved from northern regions to the south and finally to central areas (Figure 10). In 2010, low-value clusters emerged in the northeast, reflecting regional disparities during the period of rapid urban growth. By 2020, high-value zones were concentrated in central Hohhot, including Tuoketuo County, Helingeer County, Tumed Left Banner, Huimin District, and Xincheng District, forming a “high center–low north and south” spatial pattern closely aligned with ecological efficiency hotspots and coldspots. This evolution mirrors Hohhot’s urbanization trajectory: early northern counties relied on ecological resettlement and village construction to drive urbanization but lacked industrial support, resulting in a co-occurrence of population growth and ecological degradation. In contrast, central urban areas later achieved synergistic development of urbanization and ecological efficiency through urban renewal and the deployment of green infrastructure.

Mechanistically, the initial urbanization phase involved extensive expansion that encroached on ecological space, producing significant diseconomies of scale. Subsequent stages effectively reversed efficiency losses by increasing development density, promoting mixed land-use, and implementing ecological restoration initiatives. Moving forward, the “compact city” development model should be maintained, with strict control over new construction land, to deepen coordination between urbanization and ecological conservation.

4. Discussion

Framed within the “technology-scale-structure” perspective, this study systematically examines the spatiotemporal evolution and driving mechanisms of urban ecological asset utilization efficiency in Hohhot between 2000 and 2020. Overall, comprehensive efficiency exhibited a sustained upward trend. Pure technical efficiency followed a V-shaped trajectory, initially declining before recovering, whereas scale efficiency displayed an inverted V-shaped pattern, rising before subsequently declining. Spatially, efficiency was markedly clustered and mobile, forming a distinct hotspot–coldspot structure. This “asynchronous-heterogeneous” evolution was particularly pronounced in arid and semi-arid regions characterized by limited ecological carrying capacity and strong resource constraints. Peripheral counties concurrently exhibited “technological enclaves” and scale diseconomies, reflecting the combined effects of barriers to technology diffusion and uncoordinated spatial expansion. Industrial structure optimization emerged as the most stable and decisive driver of efficiency gains.

In contrast, the marginal effects of economic growth and government investment diminished over time, while urbanization gradually shifted from extensive expansion toward intensive upgrading. These dynamics are consistent with the observed expansion of construction land from 547 km² to 1162 km² (a 112% increase) and the persistent dominance of energy-intensive industries, providing a coherent explanation for the inverted V-shaped evolution of scale efficiency under dual resource and spatial constraints. Meanwhile, the V-shaped recovery of pure technical efficiency closely corresponds to staged improvements in technology and management practices.

Based on these findings, the synergy mechanism of “technology-scale-structure” can be summarized as follows: technological progress acts as an endogenous driver, enhancing resource output efficiency; scale constraints, if overlooked during rapid construction, create a trade-off between short-term output gains and long-term efficiency losses; and structural upgrading sustains efficiency improvements by limiting energy-intensive industries and fostering green, service-oriented development. This mechanism not only systematically accounts for Hohhot’s efficiency trajectory over the past two decades, but also provides a transferable framework for efficiency governance in arid and semi-arid urban clusters.

At the policy level, relying solely on expanded ecological investment is insufficient to sustain long-term efficiency gains. A multidimensional, systemic strategy is therefore required. First, accelerate the transformation of green industries to reduce path dependence on energy-intensive sectors. Second, establish cross-regional platforms for technology sharing and innovation diffusion to overcome the “technological isolation” of peripheral counties. Third, optimize fiscal expenditure by prioritizing investment in green infrastructure and ecological restoration. Fourth, implement spatially differentiated governance: enforce scale controls and ecological redlines in high-efficiency hotspots, while emphasizing technology empowerment and ecological compensation in low-efficiency coldspots to foster regional coordination and balanced efficiency improvements.

Despite rigorous methodological controls, several limitations of this study warrant careful consideration. First, the input-oriented super-efficiency SBM model has limited capacity to account for non-desired outputs. Second, grey correlation analysis is constrained in capturing nonlinear interactions and lagged effects among multiple factors. Third, the county-level scale may obscure micro-level variations at the neighbourhood or community level, as well as potential environmental equity issues. Fourth, the findings are most directly applicable to arid and semi-arid cities; further validation is required for their application to other regions or highly industrialized urban contexts.

Future research could address these limitations by adopting network DEA or non-dominated sorting SBM models to differentiate dual outputs of economic gains versus environmental losses; integrating dynamic panel models or machine learning approaches to improve causal inference and predictive accuracy; leveraging remote sensing and social media data for more fine-grained neighbourhood-level analyses; and employing system dynamics or multi-agent simulations to assess the long-term impacts of policies under dual-carbon targets.

5. Conclusions

This study integrates the super-efficiency SBM model, global and local spatial autocorrelation analysis, and grey relational analysis to construct a multi-method evaluation framework. It systematically examines the spatiotemporal evolution and driving mechanisms of Hohhot’s urban ecological asset utilization efficiency from 2000 to 2020, spanning both county and temporal scales. Key findings are as follows. First, comprehensive efficiency steadily increased from 1.132 in 2000 to 1.397 in 2020, yet exhibited pronounced spatial heterogeneity. Central urban areas and regions with favourable ecological endowments demonstrated higher efficiency, whereas peripheral and rapidly expanding zones formed efficiency depressions. Second, pure technical efficiency followed a recovery trajectory of initial decline followed by improvement, while scale efficiency displayed a long-term inverted V-shaped pattern, indicating that relying solely on scale expansion is insufficient for achieving sustainable efficiency gains; therefore, careful regulation of construction intensity and land expansion is essential. Third, industrial structure optimization emerged as the most stable and dominant driver. The marginal effects of economic growth and government intervention declined over time, while urbanization gradually shifted from extensive outward expansion to functionally driven, intensive growth. Fourth, differentiated governance strategies proved effective: high-efficiency hotspots should prioritize scale controls and ecological redlines, whereas low-efficiency coldspots should focus on technological enhancement and diffusion, complemented by ecological compensation and fiscal optimization to achieve synergistic development of efficiency and ecological security.

This study makes three primary contributions. First, it develops a replicable, multidimensional “efficiency–spatial–mechanism” evaluation framework by integrating super-efficiency SBM, spatial hotspot analysis, and grey relational analysis, tailored explicitly to arid and semi-arid regions. Second, it addresses a critical research gap by documenting the evolution of county-level ecological asset efficiency over a two-decade period in northern arid regions. Third, it proposes and validates a “technology–scale–structure” synergistic mechanism, clarifying the interactive effects among technological diffusion, scale constraints, and structural upgrading, thereby providing theoretical foundations and policy guidance for enhancing urban ecological efficiency.

Based on these findings, several policy implications are suggested to promote the high-quality development of regional ecological assets in Hohhot, as well as in other arid and semi-arid regions and similar contexts in developing countries. First, restructure green industrial systems by curbing energy-intensive sectors and promoting service- and ecology-oriented economies. Second, establish cross-city and cross-basin platforms for technology sharing and standardization to accelerate the diffusion of green technologies in peripheral areas. Third, optimize fiscal and investment structures to prioritize green infrastructure, ecological restoration, and innovative governance capacity. Fourth, implement spatial zoning and differentiated controls to limit the expansion of disorderly construction while promoting stock renewal and intensive development. Fifth, establish regional ecological compensation and collaborative governance mechanisms to internalize environmental externalities and foster sustainable cooperation across watersheds and regions.

In summary, in ecologically fragile regions with stringent resource constraints, enhancing urban ecological asset utilization efficiency requires structural upgrading as the core driver, technological diffusion as the key lever, and scale constraints as the boundary to development. This integrated approach advances the principles of “fewer but better, intensive and efficient, and ecologically secure” urban development.