The Impact of Land Use Efficiency on County-Level PM2.5: Evidence from 1125 Counties in China

Abstract

With the severe harm caused by high concentrations of PM2.5 pollution to human health and the environment, effectively reducing county-level PM2.5 concentrations has become an urgent environmental issue. This paper explores the potential of land use efficiency in mitigating county-level PM2.5 pollution, proposing that improving land use efficiency could be an effective approach to reducing PM2.5 emissions. Based on balanced panel data from 1125 counties in China from 2000 to 2021, this paper employs a two-way fixed effects model to analyze the impact of land use efficiency on county-level PM2.5 concentrations and further examines the mechanism effects of urbanization and industrialization. The main findings are as follows: (1) Improvements in land use efficiency significantly reduce county-level PM2.5 concentrations. This conclusion remains robust after a series of robustness tests. (2) Enhanced land use efficiency indirectly reduces PM2.5 concentrations by promoting urbanization and industrialization processes. (3) The suppressive effect of land use efficiency on PM2.5 is more pronounced in non-resource-based cities and non-traditional industrial cities. (4) As population density increases, the suppressive effect of land use efficiency on PM2.5 gradually weakens. (5) In regions with higher levels of economic development, the impact of land use efficiency on PM2.5 concentrations follows an inverted U-shaped curve. This study provides theoretical support and policy recommendations for optimizing land resource allocation, promoting urbanization and industrialization, and formulating targeted environmental policies.

1. Introduction

With the rapid development of global industrialization and urbanization, air pollution has become increasingly severe, particularly with fine particulate matter (PM2.5) posing a serious threat to human health and the environment. PM2.5 is not only closely associated with health issues such as respiratory and cardiovascular diseases but also has profound impacts on climate change and ecosystem stability [1]. In China, although the government has implemented various measures to improve air quality in recent years, the problem of PM2.5 pollution remains critical. This is especially evident at the county level, where significant spatial disparities in PM2.5 concentrations exist due to differences in economic development, industrial structure, and environmental governance capacity [2].

Against this background, effectively reducing county-level PM2.5 concentrations has emerged as a critical issue in both academic research and policy formulation, particularly given the profound impacts of air pollution on public health, economic sustainability, and environmental resilience. The extensive literature has consistently demonstrated a close relationship between land use patterns and air quality [3]. However, existing research has predominantly focused on the influence of specific land use types (e.g., residential, commercial, or industrial) on air pollution, with relatively limited attention paid to the direct link between LUE and PM2.5 concentrations. As a key indicator of rational land resource allocation, LUE reflects not only economic intensification but also the optimization of urban spatial structure, industrial layout, and transportation infrastructure [4]. Improving LUE has the potential to significantly reduce PM2.5 pollution by fostering compact urban development, shortening commuting distances, optimizing industrial distributions, and developing efficient transportation networks. Despite its substantial policy implications, empirical studies explicitly examining how and to what extent LUE directly impacts PM2.5 levels—and under what conditions this relationship may vary—remain insufficient. Moreover, it remains unclear whether factors such as urbanization level, industrialization degree, and regional economic development moderate the impact of LUE on PM2.5 pollution. Clarifying this relationship is especially important for policymakers, as it provides context-specific evidence for designing targeted urban planning, environmental regulation, and land management policies. Thus, elucidating the role of land use efficiency in improving air quality can enable policymakers to implement more effective measures that simultaneously achieve economic growth and reduce air pollution, ultimately promoting healthier communities and sustainable regional development.

Based on this, this paper takes 1125 counties in China as the research sample and uses balanced panel data from 2000 to 2021 to explore the impact of land use efficiency on county-level PM2.5 concentrations and its underlying mechanisms. The marginal contributions of this study are as follows: (1) It is the first to systematically analyze the impact of land use efficiency on PM2.5 concentrations, addressing a gap in existing research. (2) By introducing the mediating effects of urbanization and industrialization, this paper reveals the internal mechanisms through which land use efficiency affects PM2.5 concentrations. (3) It further examines the moderating roles of factors such as resource-based cities, old industrial cities, and population density, providing a theoretical basis for the formulation of differentiated environmental policies. This study not only deepens the understanding of the relationship between land use efficiency and air pollution but also offers important theoretical support and practical guidance for optimizing land resource allocation, promoting urbanization and industrialization, and developing targeted environmental policies.

The remainder of this paper is organized as follows: Section 2 provides a literature review; Section 3 presents the research hypotheses; Section 4 outlines the research design; Section 5 discusses the empirical results; Section 6 conducts further analysis, including heterogeneity and threshold effect analyses; and the final section summarizes the empirical findings and offers policy recommendations.

2. Literature Review

2.1. Land Use Efficiency

Land use efficiency is a key indicator for evaluating the effectiveness of land resource utilization, and its optimization is crucial for enhancing the sustainable development of cities and regions. According to existing research, the average land use efficiency of urban agglomerations in China is relatively low [5], with significant regional disparities. An assessment of 115 resource-based cities found notable spatial heterogeneity in land use efficiency improvements [6]. Additionally, the land use efficiency of the service sector exhibits an inverted U-shaped growth pattern, with lower efficiency in developed areas and higher efficiency in less developed areas, indicating a complex relationship between economic development and land use efficiency [7].

At the regional level, land use efficiency in the Yangtze River Economic Zone has steadily improved between 2010 and 2022, although the efficiency gap between regions has widened [8]. Meanwhile, the land use efficiency of cities in the West Taiwan Strait Economic Zone showed an overall upward trend from 2011 to 2020, with coastal areas generally outperforming inland regions [9]. At the county level, studies revealed a heavily right-skewed distribution of agricultural land use efficiency, with many counties performing below the national average. Past research also uncovered a robust U-shaped relationship between non-agricultural employment and changes in agricultural land use efficiency, indicating that the substitution effect of capital and technology on rural labor has strengthened over time [10].

Nationally, the average land use efficiency is low, at only 0.288, with a declining trend from the eastern to the central and western regions. Provinces that experience land inflows tend to have higher land use efficiency than those experiencing land outflows, reflecting the severity of China’s agricultural land scarcity issue [11]. Furthermore, population size and structure dividends significantly inhibit urban land use efficiency, while human capital and technological dividends initially suppress it but ultimately promote improvements [12].

Additional studies point out that the development of digital infrastructure has significantly boosted land use efficiency, particularly in enhancing the efficient allocation of urban resources. The implementation of market-oriented policies for industrial land has also played a positive role in improving urban land use efficiency [13]. These findings suggest that the enhancement of land use efficiency is influenced not only by economic development levels but also by policies, technology, and infrastructure development.

2.2. Land Use Efficiency and PM2.5

The existing literature has extensively explored the relationship between land use patterns and air pollution. Some studies have indicated that improvements in industrial land use efficiency may influence PM2.5 pollution, with this effect exhibiting significant spatial heterogeneity across different regions [14]. Additionally, research has found that the expansion of industrial land can significantly increase PM2.5 concentrations in surrounding areas [15], further demonstrating the complex mechanisms by which land use patterns affect air quality. However, despite these valuable insights into the link between land use patterns and air quality, most existing studies have primarily focused on the impacts of different land use types (such as residential, commercial, or industrial) on air pollution. There remains a relative scarcity of empirical research that directly examines the relationship between LUE—a key indicator of rational land resource allocation and economic intensification—and PM2.5 concentrations. Given that LUE has the potential to improve air quality through the optimization of urban spatial structure, industrial layout, and transportation networks, this research gap urgently needs to be addressed. Therefore, the present study aims to systematically investigate the direct impact of land use efficiency on PM2.5 concentrations, along with the underlying mechanisms, in order to fill the existing research gap.

3. Research Hypothesis

3.1. Land Use Efficiency Can Reduce the PM2.5 Concentration

Based on the theoretical framework of compact city theory and new economic geography, efficient land use can substantially mitigate PM2.5 emissions through multiple interconnected mechanisms. First, from a transportation perspective, inefficient land use—characterized by urban sprawl—often results in longer commuting distances and higher traffic volumes, which in turn lead to increased vehicle emissions [16]. In contrast, compact cities employ mixed-use planning that seamlessly integrates residential, commercial, and industrial functions, thereby significantly shortening travel distances and reducing reliance on fossil fuel-powered vehicles. This approach not only decreases transportation-related pollution but also promotes sustainable urban mobility [17]. Second, efficient land use contributes to energy efficiency improvements. In densely developed urban areas, shared infrastructure—such as centralized heating systems and robust public transportation networks—can operate more efficiently on a per-unit area basis. This higher efficiency not only conserves energy but also results in lower PM2.5 emissions, particularly those associated with the combustion of fossil fuels like coal or natural gas [18]. Furthermore, scientific and rational land planning facilitates the spatial segregation of pollution sources from residential areas. By strategically designating industrial zones away from residential and commercial districts and by incorporating natural buffers—such as green spaces and water bodies—cities can effectively limit the spread of pollutants. This spatial separation helps to reduce residents’ direct exposure to harmful PM2.5 emissions, thereby enhancing overall urban air quality.

Based on these considerations, this paper proposes Hypothesis 1:

Hypothesis 1 (H1):

Improvements in land use efficiency can directly reduce PM2.5 concentrations.

3.2. Mechanism Analysis

There is a bidirectional interaction between urbanization and improvements in land use efficiency [19]. On the one hand, urbanization promotes the intensive use of land, driving the concentration of population and economic activities in urban core areas. On the other hand, efficient land use provides the spatial foundation for upgrading urban infrastructure and applying green technologies [20]. According to the Environmental Kuznets Curve (EKC) theory, pollution may worsen in the early stages of urbanization due to industrial expansion. However, as land use efficiency improves, cities transition to a development model led by service industries and technological innovation, thereby reducing the share of pollution-intensive industries [21]. Therefore, this paper hypothesizes that the urbanization process mediates the relationship between land use efficiency and PM2.5 concentrations.

Hypothesis 2 (H2):

Land use efficiency indirectly reduces PM2.5 concentrations by promoting urbanization.

The theory of Industrial Ecology emphasizes that the intensive allocation of land resources can pressure enterprises to adopt cleaner production technologies, driving a shift in industrial structure from high-pollution, high-energy consumption industries to high-value-added, low-emission sectors [22]. Specifically, improvements in land use efficiency may influence the industrialization pathway through the following mechanisms: (1) raising the entry threshold for industrial land, thereby encouraging enterprises to adopt energy-saving and emission-reducing technologies to secure land resources [23]; (2) promoting the centralized layout of industrial parks to achieve coordinated pollution control and resource recycling [24]; and (3) unlocking the economic value of land, thereby providing financial support for green technological innovation [25]. Therefore, this paper hypothesizes that industrial upgrading is an important mediating variable in the relationship between land use efficiency and PM2.5 concentrations.

Hypothesis 3 (H3):

Land use efficiency indirectly reduces PM2.5 concentrations by driving industrial upgrading.

3.3. Threshold Effect Analysis

The congestion effect theory suggests that, when population density exceeds a certain threshold, improvements in land use efficiency may have diminishing environmental benefits due to factors such as traffic congestion and surging energy demand [26]. For instance, in high-density areas, although land use rates are higher, the combined effect of vehicle emissions and construction dust may exacerbate local PM2.5 pollution [27]. Therefore, this paper hypothesizes that population density negatively moderates the relationship between land use efficiency and PM2.5 concentrations, meaning that, as population density increases, the emission reduction effect of land use efficiency gradually weakens.

Hypothesis 4 (H4):

Population density weakens the inhibitory effect of land use efficiency on PM2.5 concentrations.

Based on an extended Environmental Kuznets Curve (EKC), this paper posits that, in the early stages of economic development, improvements in land use efficiency may primarily serve economic growth objectives, potentially worsening environmental quality. However, once the economy surpasses a certain threshold, technological progress and heightened environmental awareness will drive a decoupling of land use efficiency from PM2.5 concentrations [28]. For example, high-income regions may leverage land-based fiscal policies to support green infrastructure investment, whereas low-income regions may prioritize economic growth at the expense of environmental externalities [29]. Therefore, this paper hypothesizes that the level of economic development exerts a non-linear moderating effect on the relationship between land use efficiency and PM2.5 concentrations, following an inverted U-shaped curve.

Hypothesis 5 (H5):

Economic development level has an inverted U-shaped threshold effect on the relationship between land use efficiency and PM2.5 concentrations.

4. Research Design

4.1. Model

4.1.1. Two Way Fixed-Effect Model

The two-way fixed effects model is a commonly used panel data regression method. By simultaneously controlling for individual fixed effects and time fixed effects, it effectively eliminates time-invariant individual characteristics and time-related factors affecting all individuals, thereby improving the accuracy and robustness of the model’s estimates. The main advantages of this model are as follows: first, it fully leverages the dual dimensions of cross-sectional and time series data inherent in panel datasets; second, by controlling for omitted variables, it helps to reveal the causal relationships between variables more accurately; and, finally, when dealing with inherent differences and dynamic changes across regions or individuals, the two-way fixed effects model is better equipped to explain complex spatiotemporal data structures, providing a powerful tool for empirical research. Referring to the existing literature [30,31,32], this paper employs a two-way fixed effects model to evaluate the relationship between land use efficiency and PM2.5 concentrations. The specific model is set up as follows:

$$PM{2.5}_{it}={\beta }_0+{\beta }_{1}LU{E}_{it}+\lambda {\displaystyle \sum Control{s}_{it}}+{\nu }_i+{\tau }_t+{\epsilon }_{it}$$

(1)

“PM2.5 represents the standard deviation of PM2.5 concentrations. LUE denotes land use efficiency. Controls include all control variables in this study. $\nu$, $\tau$, and $\epsilon$ represent county fixed effects, time fixed effects, and the random error term, respectively”.

4.1.2. Mechanistic Effect Model

This paper primarily examines the mediating effects of urbanization and industrialization in the relationship between land use efficiency (LUE) and PM2.5 concentrations and thus adopts a two-step mediation model. The core structure of the model remains based on a two-way fixed effects framework, with modifications to the dependent variable to suit the analysis of mediating effects. This model is mainly used to evaluate how urbanization and industrialization mediate the relationship between LUE and PM2.5 concentrations after controlling for individual and time fixed effects, thereby revealing the underlying transmission mechanisms. To further investigate the specific pathways through which LUE indirectly impacts PM2.5 concentrations, this paper constructs the following model, referencing Dell (2010) [33].

$$M_{it}={\beta }_0+{\beta }_{1}Treate{d}_i\times Pos{t}_{it}+\lambda {{\displaystyle \sum Control{s}_{it}+{\nu }_i+{\tau }_t+\epsilon }}_{it}$$

(2)

In model (2), M represents the mechanism variables, primarily urbanization and industrialization.

4.1.3. Threshold Effect Model

To further explore the impact of population density and economic development level on the relationship between land use efficiency and PM2.5 concentrations and to examine whether these factors exhibit different effects at certain threshold levels, this paper employs a threshold effect model for analysis. The threshold effect model helps identify non-linear relationships between variables, particularly when factors such as population density and economic development reach specific threshold values, potentially causing significant changes in their influence on the relationship between land use efficiency and PM2.5 concentrations. By setting different threshold values, the model allows for an analysis of how these factors affect PM2.5 concentrations at various stages of development, enabling a more accurate evaluation and interpretation of environmental impact differences across regions and development stages.

To investigate the threshold effects of population density and economic development level, this paper applies a threshold effect model [34]. Based on the existing literature [35,36], a double-threshold effect model is constructed:

$$\begin{array}{l}\mathrm{PM}{2.5}_{it}=\eta +{\eta }_{11}LU{E}_{it}I(q_{it}<\gamma )+{\eta }_{12}LU{E}_{it}I(\gamma \le q_{it}<\mu )\\ +{\eta }_{13}LU{E}_{it}I(\mu \le q_{it})+{\eta }_2{\displaystyle \sum control{s}_{it}^{\prime }}+u_i+{\epsilon }_{it}\end{array}$$

(3)

$q_{it}$ is the threshold variable, specifically the population density and economic development level. $\gamma$ and $\mu$ is the threshold to be estimated. The individual effect is represented by the parameter u_i.

4.2. Variable Description and Source of the Data

4.2.1. Explained Variable

PM2.5: This paper uses the standard deviation of PM2.5 concentrations as a proxy variable to better capture the magnitude of air quality fluctuations and their spatial and temporal variations. PM2.5 concentration is a key indicator of air pollution levels, but single-point PM2.5 data often fail to fully reflect fluctuations in air quality. PM2.5 concentration can vary across regions and time periods due to multiple factors, such as seasonal changes, meteorological conditions, and the distribution of pollution sources. By introducing the standard deviation of PM2.5 concentrations, this approach more accurately reveals the extent and instability of PM2.5 fluctuations, thereby providing a better reflection of air quality variability across different regions and time points.

4.2.2. Explanatory Variable

Land Use Efficiency: Referring to the existing literature [37], this paper uses the ratio of the combined output value of the secondary and tertiary industries to the land area of the administrative region as a proxy variable for land use efficiency.

4.2.3. Control Variable

Control Variables: Referring to the existing literature [38,39,40], this paper selects the following control variables: fiscal revenue and expenditure level, savings, welfare, education, healthcare, and informatization level. Detailed information on each variable is provided in

Table 1. Variable definitions.

	Variable	Abbreviation	Definition	Data Sources
Dependent Variable	PM2.5	PM2.5	Standard deviation of PM2.5 concentrations	Atmospheric Composition Analysis Group
Independent Variable	Land Use Efficiency	LUE	Ratio of the combined output value of the secondary and tertiary industries to the land area of the administrative region	China County Statistical Yearbook
Control Variables	Fiscal Revenue and Expenditure Level	budget	General fiscal budget revenue and expenditure divided by regional GDP
	Savings	sav	Balance of saving deposits of urban and rural residents divided by GDP
	Welfare	welfare	Number of beds in social welfare institutions divided by the registered population
	Education	student	Number of secondary school students enrolled divided by the registered population
	Healthcare	hos	Number of hospital beds divided by the registered population
	Informatization Level	imf	Number of fixed telephone users divided by the registered population

.

4.2.4. Sample Selection and Data Source

This study selects the period from 2000 to 2021 primarily because this timeframe covers a critical phase in China characterized by rapid urbanization, industrial restructuring, and the comprehensive implementation of environmental policies, effectively reflecting the long-term dynamic trends of how land use efficiency impacts PM2.5. In addition, this study encompasses 1125 county-level units spread across a wide range of regions, including areas in the eastern, central, and western parts of China that represent various stages of economic development and income levels. This broad coverage adequately reflects the diversity and spatial imbalances of county-level development in China. Furthermore, when selecting the land use efficiency indicator, we carefully considered the specific characteristics of urban planning and land management practices at the county level in China. We use the ratio of the combined output of the secondary and tertiary industries to the administrative land area to represent land use efficiency—a measure that comprehensively reflects the actual land planning and management practices of county-level governments and possesses strong practical relevance.

During the data processing stage, the following steps were taken to ensure data completeness and reliability: first, all samples with missing data were excluded to prevent potential bias or errors caused by data gaps. Second, samples that did not span the entire research period were removed, retaining only those with a complete time series, resulting in a balanced panel dataset.

The data used in this study were sourced from multiple datasets. PM2.5 data were obtained from the Atmospheric Composition Analysis Group, which integrates aerosol optical depth data from various satellites (such as MODIS, VIIRS, MISR, and SeaWiFS) along with multiple retrieval methods (Dark Target, Deep Blue, MAIAC) and GEOS-Chem simulation data [41,42]. Data on land use efficiency, control variables, mechanism variables, and threshold variables were collected from the China County Statistical Yearbook.

Table 2. Descriptive statistics.

Variable	N	Mean	SD	Min	Max	Units
PM2.5	24,750	44.67	18.74	1.504	130.4	10 μg/m3
LUE	24,750	0.955	2.142	0	50.61	10 million per square kilometer
budget	24,750	0.237	0.170	0.010	3.959	%
sav	24,750	0.704	0.350	0	5.649	%
welfare	24,750	0.225	0.226	0	4.776	beds per 100 people
student	24,750	0.533	0.177	0.015	2.069	per 10 people
hos	24,750	0.296	0.183	0	2.157	beds per 100 people
imf	24,750	0.129	0.103	0	2.203	households per 10 people

presents the descriptive statistics of the main variables in this study. According to the statistical results, the average value of PM2.5 is 44.67, with a maximum value reaching 130.4, indicating significant fluctuations in PM2.5 concentrations over the sample period. This suggests that some regions may face severe air pollution problems. The high maximum value reflects that environmental governance or control measures may still be insufficient in certain areas, necessitating the enhanced regulation and management of high-pollution regions.

The average value of LUE is 0.955, indicating relatively low overall land use efficiency. While land use efficiency is reasonable in many areas, the fact that the average is close to 1 also implies that the potential of land resources has not been fully utilized. Low efficiency may be closely related to factors such as irrational land development, resource allocation, and industrial structure, suggesting that there is still room for further optimization in land use management.

5. Empirical Results

5.1. Benchmark Regression Analysis

Table 3. Results of the benchmark regression analysis.

	(1)	(2)	(3)	(4)	(5)	(6)	(7)
	PM2.5	PM2.5	PM2.5	PM2.5	PM2.5	PM2.5	PM2.5
LUE	−0.044 ***	−0.046 ***	−0.051 ***	−0.053 ***	−0.054 ***	−0.052 ***	−0.052 ***
	(0.008)	(0.008)	(0.009)	(0.009)	(0.010)	(0.010)	(0.010)
budget		−0.173 **	0.101	0.114 *	0.193 ***	0.167 ***	0.166 ***
		(0.083)	(0.062)	(0.062)	(0.063)	(0.064)	(0.064)
sav			−0.258 ***	−0.260 ***	−0.256 ***	−0.256 ***	−0.256***
			(0.049)	(0.050)	(0.048)	(0.048)	(0.048)
welfare				0.137 ***	0.142 ***	0.158 ***	0.159 ***
				(0.039)	(0.038)	(0.039)	(0.039)
student					−0.557 ***	−0.504 ***	−0.504 ***
					(0.053)	(0.058)	(0.058)
hos						−0.241 ***	−0.242 ***
						(0.069)	(0.070)
imf							0.023
							(0.096)
_cons	4.508 ***	4.552 ***	4.673 ***	4.643 ***	4.917 ***	4.961 ***	4.958 ***
	(0.007)	(0.022)	(0.036)	(0.035)	(0.048)	(0.048)	(0.049)
County	YES	YES	YES	YES	YES	YES	YES
Year	YES	YES	YES	YES	YES	YES	YES
Cluster	County	County	County	County	County	County	County
N	24,750	24,750	24,750	24,750	24,750	24,750	24,750
R2	0.935	0.935	0.936	0.936	0.937	0.937	0.937

Standard errors in parentheses. * p < 0.1, ** p < 0.05, *** p < 0.01.

reports the results of the stepwise regression analysis using model (1). Column (1) presents the regression results without control variables, where the coefficient of LUE is −0.44 and statistically significant at the 1% level. This result indicates that improvements in land use efficiency have a significant negative effect on county-level PM2.5 concentrations, meaning that higher LUE effectively reduces PM2.5 pollution in the air. As control variables are gradually added, the coefficient and standard error of LUE do not change significantly, suggesting that the control variables have a minimal impact on the relationship between land use efficiency and PM2.5 concentrations. This further confirms the stable role of LUE in reducing PM2.5 concentrations. Column (8) reports the regression results after including all control variables, where the coefficient of LUE decreases to −0.052. This implies that a 1% increase in LUE would lead to a 0.052% reduction in PM2.5 concentrations. This result further supports the negative relationship between land use efficiency and air quality, highlighting the potential of improving LUE to enhance air quality. In summary, these findings provide strong support for Hypothesis 1, demonstrating that improving land use efficiency can effectively reduce PM2.5 concentrations.

5.2. Robustness Test

5.2.1. Replace Variables

This paper uses the minimum and maximum values of PM2.5 as proxy variables for PM2.5 concentrations. The regression results are presented in columns (1) and (2) of

Table 4. Robustness test.

	(1)	(2)	(3)	(4)	(5)	(6)	(7)
	Min PM2.5	Max PM2.5	PM2.5	PM2.5	PM2.5	PM2.5	PM2.5
LUE	−0.543 ***	−0.497 ***		−0.034 ***	−0.044 ***	−0.051 ***	−0.052 ***
	(0.105)	(0.093)		(0.005)	(0.008)	(0.012)	(0.018)
L.LUE			−0.050 ***
			(0.010)
budget	2.029 ***	1.295 *	0.364 ***	0.178 ***	0.125 **	0.173 *	0.166
	(0.591)	(0.750)	(0.118)	(0.055)	(0.060)	(0.093)	(0.141)
sav	−2.369 ***	−2.712 ***	−0.356 ***	−0.130 ***	−0.216 ***	−0.258 **	−0.256
	(0.453)	(0.513)	(0.079)	(0.036)	(0.043)	(0.104)	(0.209)
welfare	1.041 ***	1.998 ***	0.233 ***	0.133 ***	0.128 ***	0.159 **	0.159
	(0.346)	(0.444)	(0.059)	(0.034)	(0.036)	(0.070)	(0.108)
student	−3.564 ***	−6.148 ***	−0.501 ***	−0.371 ***	−0.502 ***	−0.504 ***	−0.504 **
	(0.558)	(0.625)	(0.075)	(0.048)	(0.054)	(0.114)	(0.231)
hos	−1.736 ***	−3.201 ***	−0.375 ***	0.070	−0.282 ***	−0.239 **	−0.242
	(0.658)	(0.779)	(0.112)	(0.063)	(0.065)	(0.100)	(0.188)
imf	0.335	0.224	0.123	−0.130 *	0.004	0.024	0.023
	(0.900)	(1.086)	(0.126)	(0.076)	(0.089)	(0.145)	(0.188)
_cons	42.069 ***	56.864 ***	5.034 ***	4.852 ***	5.013 ***	4.961 ***	4.958 ***
	(0.475)	(0.510)	(0.068)	(0.038)	(0.044)	(0.112)	(0.216)
County	YES	YES	YES	YES	YES	YES	YES
Year	YES	YES	YES	YES	YES	YES	YES
Cluster	County	County	County	County	County	City	Province
N	24,750	24,750	12,144	22,500	23,625	24,596	24,750
R2	0.936	0.931	0.939	0.948	0.941	0.937	0.937

Standard errors in parentheses. * p < 0.1, ** p < 0.05, *** p < 0.01.

. Using these two different proxy variables, the coefficients of LUE (land use efficiency) are −0.543 and −0.497, respectively, both statistically significant at the 1% level. This indicates that the negative relationship between land use efficiency and PM2.5 concentrations remains robust under different variable settings, further validating the main conclusions of this study.

5.2.2. A Lag Phase

This paper applies a one-period lag of land use efficiency (L.LUE) to account for potential lag effects that might influence the current results. The regression results are shown in column (3) of Table 4. Under the lagged variable specification, the coefficient of L.LUE is −0.050 and remains statistically significant. This indicates that, even after controlling for lag effects, the negative relationship between land use efficiency and PM2.5 concentrations still holds, further confirming the robustness of the main conclusions in this study.

5.2.3. Shorten the Sample Interval

Considering that the COVID-19 pandemic may have significantly influenced the relationship between land use efficiency (LUE) and PM2.5 concentrations, we excluded data from 2019 to 2021 to eliminate potential external disturbances during the pandemic period. Moreover, since the impact of COVID-19 diminished after 2021, only data from 2020 were removed in this study. Column (4) and (5) of Table 4 report the regression results after excluding these data. The results show that the coefficient of LUE is −0.034 and −0.044 and remains statistically significant. This indicates that, even after removing the effects of the pandemic, the negative impact of land use efficiency on PM2.5 concentrations persists and is consistent with the main conclusions, with no significant changes observed.

5.2.4. Change the Cluster Standard Error

This paper uses county-level clustered standard errors for regression analysis. To test the robustness of the results, city-level and province-level clustered standard errors were also applied. Columns (6) and (7) of Table 4 present the regression results under city-level and province-level clustered standard errors, respectively. The results show that the coefficients of LUE are −0.051 and −0.052, both statistically significant at the 1% level. These findings indicate that, even when changing the clustering method, the negative impact of land use efficiency (LUE) on PM2.5 concentrations remains significant, with minimal variation in the coefficient values. This further validates the robustness of the main conclusions. By testing clustered standard errors at different levels (county, city, and province), this study ensures that the regression results are not influenced by the choice of clustering method, enhancing the reliability of the research findings.

5.3. Endurance Test

5.3.1. SYS-GMM

Column (1) of

Table 5. Endogeneity test results.

	(1)	(2)	(3)
	SYS-GMM	2SLS
	PM2.5	LUE	PM2.5
L.PM2.5	−0.023 ***
	(0.002)
LUE	−0.027 ***		0.535 ***
	(0.004)		(0.057)
iv		−0.445 ***
		(0.035)
budget	−0.193	−1.326 ***	−3.657 ***
	(0.203)	(0.103)	(0.203)
sav	−0.138 *	−0.925 ***	1.576 ***
	(0.071)	(0.052)	(0.065)
welfare	0.127 *	1.456 ***	0.182 *
	(0.071)	(0.105)	(0.110)
student	−0.233 ***	0.330 ***	1.678 ***
	(0.084)	(0.082)	(0.088)
hos	−0.638 ***	1.508 ***	−2.424 ***
	(0.117)	(0.157)	(0.126)
imf	0.175	6.660 ***	−4.272 ***
	(0.226)	(0.393)	(0.432)
_cons	4.963 ***	−0.241 ***	3.229 ***
	(0.092)	(0.057)	(0.067)
AR(1)	0.001
AR(2)	0.342
Hansen test	25.34
Kleibergen–Paap rk LM statistic		368.915 ***
Cragg–Donald Wald F statistic		859.608(16.38)
Kleibergen–Paap rk Wald F statistic		163.459(16.38)
N		24,750
R2		0.200

Standard errors in parentheses. * p < 0.1, *** p < 0.01.

reports the regression results using the System Generalized Method of Moments (SYS-GMM) model for PM2.5 concentrations. The regression results show that the lagged value of PM2.5 (L.PM2.5) has a significant negative impact on current PM2.5 concentrations, with a coefficient of −0.023, statistically significant at the 1% level. This indicates that the effect of past PM2.5 concentrations on current concentrations is relatively small and gradually diminishes, suggesting a certain degree of self-recovery in PM2.5 levels. Moreover, LUE also shows a significant negative effect on PM2.5 concentrations, with a coefficient of −0.027, significant at the 1% level. This finding implies that improvements in land use efficiency can effectively reduce PM2.5 concentrations. The diagnostic tests for the model indicate the following: the p-value for AR(1) is 0.001 and statistically significant, confirming the presence of first-order autocorrelation in the regression model. However, the p-value for AR(2) is 0.342 and not statistically significant, indicating that there is no evidence of second-order autocorrelation, suggesting good model performance in this regard. The Hansen test statistic is 25.34, indicating that the instrument variables are valid, and the model specification is appropriate. In summary, the regression results from the SYS-GMM model confirm that land use efficiency has a significant negative impact on PM2.5 concentrations. Additionally, the model is well specified with no significant autocorrelation issues.

5.3.2. SLS

This paper uses the product of administrative land area and time as an instrumental variable for land use efficiency (LUE). The rationale for selecting this instrument is that the product of land area and time reflects exogenous changes in land use efficiency, which are correlated with variations in PM2.5 concentrations but not influenced by reverse causality. Therefore, this instrumental variable effectively addresses the endogeneity problem and ensures the credibility of the regression results. Columns (2) and (3) of Table 5 present the regression results using the instrumental variable method. In these models, both the coefficient of LUE and the instrumental variable (IV) are statistically significant, indicating that land use efficiency has a significant impact on PM2.5 concentrations. This finding aligns with the research hypothesis of this paper. Specifically, the Kleibergen–Paap rk LM statistic is 368.915, which is statistically significant at the 1% level, indicating that the instrumental variable is valid and effectively addresses the endogeneity issue. Additionally, the Kleibergen–Paap rk Wald F statistic is 163.459, well above the critical value of 16.38, further confirming the strength and validity of the instrumental variable, thereby ensuring the robustness of the regression model. Therefore, the selected instrumental variable meets the requirements and reliably supports the empirical findings of this study.

5.4. Mechanism Analysis Results

5.4.1. Urbanization Process

Referring to the existing literature, this paper uses the ratio of urban population to total population as a proxy variable for the urbanization process (CP) to explore the indirect impact pathway of land use efficiency (LUE) on PM2.5 concentrations. The regression results in column (1) of

Table 6. Results of the mechanistic analysis.

	(1)	(2)	(3)
	CP	IL1	IL2
LUE	0.009 ***	0.179 ***	0.824 **
	(0.003)	(0.018)	(0.373)
budget	−0.014	−0.112 ***	0.272
	(0.021)	(0.033)	(4.727)
sav	0.025 **	−0.077 ***	−7.801 ***
	(0.010)	(0.023)	(2.519)
welfare	0.037 ***	0.050 **	−18.296 ***
	(0.011)	(0.024)	(2.744)
student	−0.014	−0.051 **	−68.577 ***
	(0.009)	(0.023)	(7.611)
hos	0.086 ***	0.111 *	−132.488 ***
	(0.023)	(0.063)	(12.007)
imf	0.060 ***	−0.055	25.958 ***
	(0.018)	(0.059)	(5.604)
_cons	0.149 ***	0.101 **	85.537 ***
	(0.010)	(0.041)	(6.949)
Sobel Z	−9.691 ***	−3.251 ***	−2.432 **
Bootstrap Times	1000	1000	1000
Confidence Interval	[−0.0332, −0.0176]	[−0.0414, −0.0152]	[−0.0079, −0.0053]
County	YES	YES	YES
Year	YES	YES	YES
Cluster	County	County	County
N	16,698	19,789	24,710
R2	0.835	0.884	0.264

Standard errors in parentheses. * p < 0.1, ** p < 0.05, *** p < 0.01.

show that the coefficient of LUE is significantly positive, indicating that improvements in land use efficiency promote urbanization, which in turn affects changes in PM2.5 concentrations. Further analysis shows that the Sobel Z-value is −9.691, and the bootstrap confidence interval does not include zero, indicating that the urbanization process plays a significant mediating role between land use efficiency and PM2.5 concentrations. This suggests that enhancing land use efficiency not only directly reduces PM2.5 concentrations but also indirectly improves environmental quality by promoting urbanization. Therefore, Hypothesis 2 is validated, demonstrating that improvements in land use efficiency indirectly lead to a reduction in PM2.5 concentrations through the promotion of urbanization.

5.4.2. Industrialization Process

This paper measures the industrialization process from two perspectives: output value and the number of enterprises. Specifically, the total output value of large-scale industries (IL1) and the number of large-scale industrial enterprises (IL2) are used as proxy variables for industrialization. Columns (2) and (3) of Table 6 present the regression results for these two aspects. The analysis shows that the coefficient of LUE (land use efficiency) is significantly positive in both models, indicating that improvements in land use efficiency effectively promote industrialization, whether through increasing industrial output or the number of industrial enterprises.

Moreover, the Sobel Z-values are significant, and the bootstrap confidence intervals do not include zero, further confirming the mediating role of the industrialization process in the relationship between land use efficiency and PM2.5 concentrations. In other words, industrialization plays a significant indirect role in this relationship. Enhancing land use efficiency not only directly reduces PM2.5 concentrations but also indirectly improves environmental quality by driving industrial development. Therefore, Hypothesis 3 is validated, demonstrating that improvements in land use efficiency promote industrialization, which in turn indirectly reduces PM2.5 concentrations through industrial development.

6. Further Analysis

6.1. Analysis of Heterogeneity

6.1.1. Resource Endowment

According to the National Plan for Sustainable Development of Resource-Based Cities (2013–2020) issued by the State Council in 2013, resource-based cities (RBCs) are defined as cities whose dominant industries are based on the extraction and processing of natural resources, such as minerals and forests, and which are highly dependent on resource exploitation (https://www.gov.cn/zwgk/2013-12/03/content_2540070.htm, accessed on 13 March 2025). The economic development, industrial structure, and employment patterns of these cities significantly rely on local resource extraction, processing, and utilization. In contrast, non-resource-based cities (NRBCs) are characterized by more diversified economic structures and lower dependence on the extraction of local natural resources. Following this classification standard, this paper divides the samples into resource-based cities and non-resource-based cities for separate regression analyses. The regression results in columns (1) and (2) of

Table 7. Results of the heterogeneity analysis.

	(1)	(2)	(3)	(4)	(5)	(6)
	Resource-based	Non-resource type		Old industry	Non-old industry
	PM2.5	PM2.5	PM2.5	PM2.5	PM2.5	PM2.5
LUE	−0.121 ***	−0.045 ***	−0.049 ***	−0.100 ***	−0.050 ***	−0.045 ***
	(0.027)	(0.009)	(0.009)	(0.024)	(0.009)	(0.008)
LUE_re			−0.054 ***
			(0.018)
LUE_oib						−0.095 ***
						(0.022)
budget	0.023	0.262 ***	0.154 **	0.133	0.139 **	0.139 **
	(0.103)	(0.075)	(0.065)	(0.165)	(0.066)	(0.065)
sav	−0.239 ***	−0.300 ***	−0.265 ***	−0.077	−0.313 ***	−0.266 ***
	(0.078)	(0.061)	(0.049)	(0.087)	(0.058)	(0.048)
welfare	0.248 ***	0.130 ***	0.159 ***	0.215 ***	0.119 ***	0.161 ***
	(0.070)	(0.046)	(0.039)	(0.068)	(0.045)	(0.039)
student	−0.393 ***	−0.602 ***	−0.510 ***	−0.787 ***	−0.343 ***	−0.518 ***
	(0.082)	(0.077)	(0.058)	(0.092)	(0.072)	(0.057)
hos	−0.246 **	−0.235 ***	−0.241 ***	−0.629 ***	−0.137 *	−0.240 ***
	(0.110)	(0.090)	(0.070)	(0.143)	(0.072)	(0.068)
imf	0.127	−0.088	0.011	0.315**	−0.033	0.017
	(0.180)	(0.109)	(0.096)	(0.146)	(0.113)	(0.095)
_cons	4.850 ***	5.080 ***	4.982 ***	5.640 ***	4.637 ***	4.996 ***
	(0.077)	(0.061)	(0.049)	(0.101)	(0.054)	(0.048)
County	YES	YES	YES	YES	YES	YES
Year	YES	YES	YES	YES	YES	YES
Cluster	County	County	County	County	County	County
N	9042	15,708	24,750	7942	16,808	24,750
R2	0.938	0.938	0.937	0.933	0.938	0.937

Standard errors in parentheses. * p < 0.1, ** p < 0.05, *** p < 0.01.

indicate that LUE (land use efficiency) significantly reduces PM2.5 concentrations in both resource-based and non-resource-based cities. Specifically, improvements in land use efficiency positively impact air quality in both types of cities.

However, the coefficient size alone is not sufficient to fully assess the heterogeneity effects between different city types. To further explore these differences, this paper constructs an interaction term (LUE_re) to investigate the differentiated effects of land use efficiency between resource-based and non-resource-based cities. The construction process is as follows: first, a dummy variable (re) is created, assigning a value of 1 to resource-based cities and 0 to non-resource-based cities. This dummy variable is then multiplied by land use efficiency (LUE) to obtain the interaction term LUE_re. The regression results in column (3) of Table 7 show that the coefficient of LUE_re is significantly negative at the 1% significance level. This indicates that, compared to resource-based cities, non-resource-based cities achieve more effective PM2.5 reductions through improvements in land use efficiency. In other words, the emission reduction effect of land use efficiency is more pronounced in non-resource-based cities. This difference may be attributed to the reliance of resource-based cities on resource extraction and heavy industries, which makes environmental governance and transformation more challenging. In contrast, non-resource-based cities may find it easier to achieve sustainable ecological improvements through land use efficiency enhancements.

6.1.2. Industrial Foundation

Referring to the National Plan for the Adjustment and Transformation of Old Industrial Bases (2013–2022) issued by the National Development and Reform Commission in 2013, this paper divides the sample into old industrial cities and non-old industrial cities. The plan provides detailed coverage of 120 old industrial cities across 27 provinces, including 95 prefecture-level cities and 25 municipalities, sub-provincial cities, and provincial capitals. Based on this background, separate regression analyses were conducted for both types of cities, with results presented in columns (4) to (6) of Table 7. The regression results in columns (4) and (5) show that LUE (land use efficiency) significantly reduces PM2.5 concentrations in both old and non-old industrial cities. This indicates that improving land use efficiency positively contributes to air quality improvement in both city types. To further examine whether the impact of land use efficiency on PM2.5 concentrations varies between the two city types, this paper constructs an interaction term (LUE_oib) to test for heterogeneity. Specifically, a dummy variable (oib) is created, assigning a value of 1 to old industrial cities and 0 to non-old industrial cities. This dummy variable is then multiplied by LUE to obtain the interaction term LUE_oib.

The regression results in column (6) show that the coefficient of LUE_oib is −0.095 and statistically significant at the 1% level. This indicates that, compared to old industrial cities, non-old industrial cities experience a more pronounced reduction in PM2.5 concentrations through improvements in land use efficiency. This difference may be attributed to variations in economic structure, industrial base, and the implementation of environmental policies between the two city types. Old industrial cities often have a concentration of heavily polluting industries and face significant challenges in transformation and upgrading, which may limit the effectiveness of land use efficiency improvements on PM2.5 reduction. In contrast, non-old industrial cities tend to have more diversified industrial structures, allowing land use efficiency improvements to more directly drive environmental quality enhancement.

6.2. Threshold Effect Analysis Results

This paper employs a threshold effect model to test the threshold effects of population density and economic development level.

Table 8. Threshold estimates.

	pop			pgdp
Model	Threshold	Lower	Upper	Threshold	Lower	Upper
Th-1	0.0758	0.0753	0.0761	0.2314	0.2245	0.2399
Th-21	0.0755	0.0752	0.0758	0.2314	0.2245	0.2386
Th-22	0.0557	0.0554	0.0560	0.0399	0.0393	0.0403
Th-3	0.0350	0.0344	0.0352	0.1758	0.1744	0.1767

Note: Th represents the threshold value.

reports the results of the threshold estimation. From Table 8, the threshold values can be obtained, allowing the division of the threshold variables into multiple intervals.

Table 9. Test of the threshold effect.

	Threshold	Fstat	Prob	Crit10	Crit5	Crit1
pop	Single	194.38	0.000	34.403	40.604	48.563
	Double	62.84	0.0133	31.201	39.964	64.895
	Triple	21.03	0.6967	68.515	85.958	109.681
pgdp	Single	533.94	0.0000	233.769	241.976	285.166
	Double	280.77	0.0000	161.462	170.864	188.805
	Triple	129.41	0.1100	131.020	150.179	196.611

reports the results of the threshold effect tests. In the threshold effect test for population density, the probability values (Prob) for the first and second thresholds are 0.000 and 0.0133, respectively, both of which are statistically significant. This indicates that there is a significant threshold effect of population density on the relationship between land use efficiency and PM2.5 concentrations. However, the probability value for the third threshold is 0.6967, which is not statistically significant, suggesting that a triple-threshold effect does not exist in this model. Therefore, a double-threshold effect model is selected to further analyze the threshold effect of population density. In the threshold effect test for the economic development level, the probability values for both the first and second thresholds are 0.000, indicating statistically significant effects. This demonstrates that the economic development level also exerts a significant threshold effect on the relationship between land use efficiency and PM2.5 concentrations. In contrast, the probability value for the third threshold is 0.1100, which is not statistically significant, suggesting that the triple-threshold effect is not applicable to the economic development level test. Consequently, a double-threshold effect model is also used to further analyze the threshold effect of economic development level.

6.2.1. Density of Population

This paper uses the ratio of population to administrative land area as a proxy variable for population density. The results of the double-threshold effect test are presented in column (1) of

Table 10. Results of the threshold effect test.

	(1)		(2)
	pop		pgdp
_cat#c.LUE	PM2.5	_cat#c.LUE	PM2.5
pop < 0.0557	−0.220 ***	pgdp < 0.0399	−3.779 ***
	(0.011)		(0.254)
0.0557 < pop < 0.0755	−0.128 ***	0.0399 < pgdp < 0.2314	0.280 ***
	(0.008)		(0.019)
0.0755 < pop	−0.058 ***	0.2314 < pgdp	−0.054 ***
	(0.004)		(0.004)
Controls	YES	Controls	YES
N	24,750	N	24,750
R2	0.3586	R2	0.3727

Standard errors in parentheses. *** p < 0.01.

. Specifically, the impact of LUE (land use efficiency) on PM2.5 concentrations varies significantly across different population density intervals. First, when population density is less than 0.0557, the coefficient of LUE is −0.220 and statistically significant at the 1% level, indicating that, in this interval, improvements in land use efficiency can significantly reduce PM2.5 concentrations. Next, when population density ranges between 0.0557 and 0.0755, the coefficient of LUE is −0.128, also significant at the 1% level. This suggests that, while the effect of land use efficiency on PM2.5 concentrations weakens as population density increases, it remains significant. Finally, when population density exceeds 0.0755, the coefficient of LUE decreases to −0.058, though it remains statistically significant at the 1% level. This indicates that, with further increases in population density, the effect of land use efficiency on PM2.5 concentrations continues to weaken but still maintains a significant negative impact. In summary, there is a clear double-threshold effect of population density on the relationship between land use efficiency and PM2.5 concentrations. In low-density intervals, the impact of land use efficiency on PM2.5 reduction is most pronounced. However, as population density increases, the effect gradually diminishes.

6.2.2. Economic Development Level

This paper uses the ratio of regional GDP to population as a proxy variable for economic development level. The results of the double-threshold effect test are presented in column (2) of Table 10. The impact of LUE on PM2.5 concentrations exhibits significant differences across different economic development intervals.

First, when the economic development level is below 0.0399, the coefficient of LUE is −3.779 and statistically significant at the 1% level. This indicates that, in regions with low economic development, improvements in land use efficiency have a strong negative impact on PM2.5 concentrations, significantly reducing air pollution. Next, when the economic development level ranges between 0.0399 and 0.2314, the coefficient of LUE is 0.280 and statistically significant at the 1% level. This suggests that, in this interval, the effect of land use efficiency on PM2.5 concentrations turns positive, meaning that improvements in land use efficiency may lead to an increase in PM2.5 concentrations. Finally, when the economic development level exceeds 0.2314, the coefficient of LUE decreases to −0.054 and remains statistically significant at the 1% level. This indicates that, in economically developed regions, improvements in land use efficiency again significantly reduce PM2.5 concentrations, although the effect is weaker.

In summary, there is a clear double-threshold effect of economic development level on the relationship between land use efficiency and PM2.5 concentrations. In regions with low economic development, the impact of land use efficiency on PM2.5 concentrations is most significant and negative. In regions with moderate economic development, the effect turns positive. However, in regions with high economic development, the effect reverts to negative, though relatively weaker.

6.3. Discussion

In this section, we discuss some limitations identified in this paper.

(1)

In Eastern China, despite the relatively high level of LUE, PM2.5 pollution remains a significant issue. Empirical findings in this study indicate that improvements in land use efficiency can significantly reduce PM2.5 concentrations, clearly demonstrating a negative correlation between the two. However, Eastern China is characterized by intensive land use, high levels of urbanization, and advanced economic development, which have led to mature land use structures and urban spatial layouts. Consequently, there is limited potential for further improvements in LUE within these areas. This implies that continuing to increase land use efficiency may face diminishing returns, where environmental benefits decrease progressively as land use efficiency approaches saturation. Furthermore, PM2.5 pollution in Eastern China is influenced by multiple interconnected factors, including industrial structures, transportation systems, and energy consumption patterns. Solely focusing on enhancing land use efficiency may thus be insufficient to fundamentally alleviate air pollution. Therefore, to effectively reduce PM2.5 concentrations, Eastern China should prioritize comprehensive and multi-dimensional governance strategies, including industrial transformation, transportation system optimization, and green infrastructure development, alongside exploiting the remaining potential of land use efficiency improvements.

(2)

Hypothesis H2 assumes that urbanization serves as a mediating mechanism in the relationship between LUE and PM2.5 pollution. Nevertheless, during urbanization, different land use types (e.g., residential, commercial, and industrial) may contribute distinctively to PM2.5 concentrations. For example, areas dominated by residential and commercial land use may experience PM2.5 pollution primarily from motor vehicle emissions and volatile organic compounds produced by domestic energy consumption, whereas areas dominated by industrial land use may be affected mainly by heavy pollution industries such as metallurgy and chemical manufacturing. Consequently, using a generalized indicator of urbanization to analyze how LUE affects PM2.5 concentrations through urbanization processes might not accurately capture the inherent heterogeneity among different land use types. This represents a limitation of the current study. Additionally, the role of urbanization can vary considerably depending on regional contexts, development stages, and industrial structures; therefore, oversimplifying or generalizing urbanization’s effects could obscure underlying mechanisms and limit the applicability of policy recommendations. Future research should thus further refine land use classification and spatial configuration within urbanization analyses to more accurately identify how varying urbanization patterns and structures differentially impact PM2.5 pollution, leading to more targeted and practical policy implications.

(3)

Land use efficiency, as a key indicator of rational resource allocation and economic intensification, has profound implications for policymaking. By improving land use efficiency, not only can urban spatial structures be optimized and compact city development promoted, but residents’ commuting distances can also be shortened, leading to reduced traffic congestion and lower vehicle emissions, which in turn help to decrease the release of air pollutants such as PM2.5. Moreover, efficient land use enhances the operational efficiency of public infrastructure and energy systems, further reducing energy consumption and its associated environmental burden. At the policy level, governments can utilize land use efficiency as a critical indicator to formulate more precise urban planning and land management measures, promote industrial transformation and the development of green infrastructure, and ultimately achieve a win–win scenario for both economic growth and environmental protection.

7. Conclusions and Policy Recommendations

7.1. Conclusions

The main conclusions of this study are as follows: First, improvements in land use efficiency significantly reduce county-level PM2.5 concentrations, and this effect remains robust after a series of sensitivity tests, indicating its long-term effectiveness in improving air quality. Second, land use efficiency not only directly lowers PM2.5 concentrations but also indirectly reduces them by promoting urbanization and industrialization, highlighting its close relationship with regional development. Efficient land planning and use optimize urban functional layouts, reduce pollution sources, and improve air quality. Third, the inhibitory effect of land use efficiency on PM2.5 concentrations is more pronounced in non-resource-based and non-old industrial cities, likely due to differences in economic structure and industry distribution. Fourth, as population density increases, the effect weakens, possibly due to factors such as increased development pressure, traffic, and infrastructure overload. Finally, in regions with higher economic development, the relationship between land use efficiency and PM2.5 concentrations follows an inverted U-shaped curve, suggesting that, while improvements initially reduce pollution, further economic growth and intensified urbanization and industrialization may introduce counteracting effects. These findings emphasize the importance of accounting for non-linear dynamics in policymaking.

7.2. Policy Recommendations

Based on this study’s findings, the following policy recommendations are proposed regarding the relationship between land use efficiency and PM2.5 concentrations:

Targeted Urban Redevelopment to Improve LUE: Our empirical findings demonstrate that improvements in LUE significantly reduce county-level PM2.5 concentrations, with robust long-term effects. Therefore, local governments should prioritize urban redevelopment policies that enhance LUE, particularly in county-level areas and urban clusters. In non-resource-based and non-old industrial cities—where the inhibitory effect of LUE on PM2.5 is more pronounced—zoning reforms and targeted investments in urban infrastructure can help optimize spatial layouts and reduce pollution sources.

Address Challenges of High Population Density: Our research indicates that the positive impact of LUE on reducing PM2.5 diminishes as population density increases. In densely populated areas, the effective management of urban growth is essential. Policies should focus on strengthening public transportation networks, promoting green building practices, and enhancing energy efficiency to mitigate the environmental pressures associated with over-concentration.

Stage-Based Economic and Environmental Regulation: The inverted U-shaped relationship observed between LUE and PM2.5 concentrations in highly developed regions suggests that rapid economic growth may eventually counteract the benefits of improved LUE. Policymakers should therefore adopt regulatory strategies tailored to different stages of economic development. In the early stages of growth, the focus should be on integrating clean technologies and environmental controls to manage pollutant emissions. In contrast, in highly developed regions, stricter environmental regulations should be implemented alongside measures to further optimize land use efficiency, ensuring sustained improvements in air quality.