The Impact of Distorted Land Supply Structures on Green Economic Growth in Chinese Cities: The Moderating Role of Housing Prices

Abstract

This study investigates the impact mechanism of distorted land supply structures on green economic efficiency in Chinese cities, with a particular focus on the mediating and moderating role of the real estate market. Innovatively, the study constructs a comprehensive index to measure land supply structure distortion and employs spatial econometric methods for empirical analysis using panel data from 285 prefecture-level and above cities in China from 2010 to 2022. The findings reveal that: (1) distortions in land supply structure significantly hinder the improvement of urban green economic efficiency (GEE); (2) this inhibitory effect exhibits a significant spatial spillover effect; (3) housing prices play a notable mediating and moderating role in the relationship between land supply structure distortion and green economic efficiency; (4) the impact mechanisms demonstrate significant regional heterogeneity. These findings offer important policy implications for optimizing urban land supply structures and promoting green economic development.

1. Background

As China transitions from a phase of rapid economic growth to an era of high-quality development, the constraints on resources and the environment have become increasingly stringent, and the concept of green development has gained deeper traction. Under the strategic goals of “carbon peaking and carbon neutrality,” urban economic development urgently requires transformation and upgrading. However, under the land supply system characterized by local government monopoly over land supply and reliance on “land finance,” a dual-track system is implemented for the transfer methods and pricing of industrial land versus commercial/residential land. To attract investment and promote economic development, local governments often transfer industrial land at low or even zero prices through negotiated agreements or at reserve prices via the “bidding, auction, and listing” process. In contrast, commercial and residential land must be publicly auctioned through the same process, with the highest bidder acquiring the land. Revenues from commercial and residential land auctions constitute an important source of local government fiscal income. As a result, industrial land prices are deliberately suppressed, while commercial and residential land prices are continuously driven up during competitive bidding. This “scissor effect” (the term “scissor effect” refers to a widening gap between two related prices, returns, or growth rates, creating a divergence that visually resembles an opening pair of scissors) in urban land prices leads to structural distortions in land supply, becoming a key constraint on the development of urban green economies. Coordinating the optimization of land supply structures with green economic development presents several challenges: (1) low-cost industrial land supply has led to the overexpansion of resource-intensive industries, exacerbating environmental pollution; (2) insufficient supply of commercial and residential land has driven up housing prices, crowding out innovation investments and hindering industrial transformation; (3) imbalances in regional land supply structures have intensified developmental disparities, constraining coordinated regional development.

Against this backdrop, this study focuses on the mechanisms through which land supply structure affects urban green economic efficiency, with particular emphasis on the moderating role of housing prices. By doing so, the study not only broadens the research horizons of land economics and green development theories but also provides theoretical guidance and policy insights for optimizing land supply structures and fostering urban green development. The findings offer significant theoretical and practical implications for advancing high-quality economic development and accelerating ecological civilization construction in China.

2. Literature Review

The impact of distorted land supply structures on urban green economic efficiency spans multiple research fields. This study reviews relevant literature from three perspectives: land supply structure, green economic efficiency, and the real estate market, to clarify existing findings and identify research gaps.

Existing studies explain the causes of land supply structure distortion from two main perspectives: institutional constraints and economic incentives. Institutional constraints on policy have been widely examined in the literature [1,2]. Immergut demonstrates a close historical and theoretical connection between institutional theory and policy analysis, highlighting that institutions shape policies, while policies, in turn, influence how people perceive and understand institutions [1]. In the Chinese context, Tao, Lu [3] pointed out that the tax-sharing reform and GDP-oriented performance assessment serve as the institutional roots of local governments’ distorted land supply behavior. Empirical evidence indicates that the coexistence of low-priced industrial land supply and high-priced commercial and residential land transfers, which often referred to as a “dual-track system”, significantly influences local fiscal revenues and corporate investment decisions [4,5].

In terms of manifestation, prior studies have systematically measured the price scissors effect and quantitative imbalance in land supply [6,7], and have further quantified the structural characteristics of different land transfer modes [8]. The impacts of land supply structure distortion have been confirmed mainly in two dimensions: economic growth and resource–environmental outcomes. Some studies find that the low-cost allocation of industrial land promotes industrialization but leads to resource misallocation [9], whereas insufficient supply of residential and commercial land tends to raise housing prices and reduce household welfare [10].

The theory of green economic efficiency has evolved from traditional production efficiency to environmental efficiency and, more recently, to green total factor productivity (GTFP). Earlier frameworks have incorporated undesirable outputs into efficiency evaluation [11,12], while subsequent studies have refined the comprehensive evaluation systems under resource and environmental constraints [13]. The literature generally identifies institutional factors, industrial structure, marketization, and technological innovation as the main determinants of green economic efficiency. Empirical analyses from multiple perspectives, such as environmental regulation and marketization [14,15], industrial structure and openness [16,17], and technological innovation [18], have further substantiated these relationships.

The interaction between the housing market and the land market has also been widely documented [19,20,21]. Potepan highlights that housing prices, rents, and urban land prices constitute three closely interrelated submarkets, each influencing the others [22]. Similarly, Fik et al. emphasize that housing prices are strongly affected by the land market, particularly due to locational factors, as land scarcity and accessibility significantly shape residential values [23]. These studies collectively underscore the importance of considering both land market dynamics and housing price fluctuations when analyzing urban development patterns. In the Chinese context, existing research reveals that land supply influences housing prices through cost transmission mechanisms [24], while housing price fluctuations, in turn, shape land market expectations [25]. Furthermore, the transmission mechanisms linking housing prices to economic development operate through multiple channels. Evidence supports the wealth effect [26], the credit and monetary policy transmission channel [27], and the broader effects of housing prices on economic structure, consumption patterns, innovation activities, and environmental governance [28,29,30].

In conclusion, existing studies have made substantial contributions to understanding the institutional roots of land supply structure distortion, the mechanisms through which it affects green economic efficiency, and the transmission channels of the real estate market. However, the interrelationships among these three dimensions remain insufficiently explored. Specifically, it is still unclear how land supply structure distortion influences green economic development, how the real estate market mediates this relationship, and whether these effects vary across regions due to spatial heterogeneity.

Accordingly, this paper focuses on three main aspects. First, it empirically examines the impact of land supply structure distortion on green economic efficiency. Second, it incorporates housing prices to investigate their mediating and moderating roles in the relationship between land supply structure distortion and green economic efficiency. Third, it tests for spatial correlation, regional heterogeneity, and endogeneity issues. The study seeks to make an innovative contribution by constructing a comprehensive index of land supply structure distortion, systematically revealing its influence mechanisms on green economic efficiency and the mediating and moderating effects of housing prices, while applying spatial econometric models and instrumental variable methods to address spatial dependence and endogeneity problems.

3. Theoretical Analysis and Research Hypotheses

This study constructs an analytical framework to systematically explore the mechanisms through which land supply structure distortion affects urban green economic efficiency (GEE), and the moderating effect of housing prices. Based on this, corresponding research hypotheses are proposed.

3.1. Theoretical Mechanism of Land Supply Structure Distortion Affecting Green Economic Efficiency

3.1.1. Factor Allocation Effect

New institutional economics suggests that institutional distortions induce factor misallocation and reduce resource allocation efficiency [1]. From the perspective of land use theory, land is a scarce production factor that should be allocated according to the principle of “highest and best use,” with market-based rents guiding its distribution across alternative uses. The provision of industrial land at artificially low prices disrupts rent formation and weakens the land rent gradient, limiting the reallocation of land toward service sectors with higher marginal returns. This distortion directly lowers land-use efficiency and indirectly intensifies the misallocation of labor and capital through industrial linkages and factor substitution, thereby constraining industrial upgrading at a broader scale. Within China’s land system, distortions in the land supply structure affect factor allocation efficiency through several channels. Low-priced industrial land reduces production costs and encourages overinvestment in resource-intensive industries [31], while insufficient supplies of commercial and residential land restrict service-sector development and hinder structural upgrading [32]. Distorted land allocation further amplifies the misallocation of labor and capital through inter-industry linkages [33], undermining overall resource efficiency and aggravating environmental pollution.

3.1.2. Innovation Suppression Effect

Endogenous growth theory emphasizes the institutional environment as a key determinant of innovation activity [34]. Distortions in the land supply structure can suppress innovation through several channels. Artificially low industrial land prices reduce firms’ incentives for technological upgrading by lowering the cost of traditional production factors. Insufficient supplies of commercial and residential land increase housing costs and crowd out resources that could otherwise be allocated to innovation investment. Imbalances in land supply also reshape the spatial distribution of innovative activities, reducing innovation efficiency and weakening knowledge spillovers.

3.1.3. Environmental Governance Effect

Environmental economics theory points out that institutional arrangements profoundly affect the effectiveness of environmental governance [35]. Land supply structure distortion affects environmental governance through the following mechanisms: The over-supply of industrial land encourages the expansion of pollution-intensive enterprises, thereby intensifying environmental pressures. Meanwhile, the insufficient supply of other land types alters the functional layout of urban space, reducing the efficiency of environmental governance and public service provision. In addition, the imbalance in land supply structure disrupts the spatial allocation of environmental infrastructure, weakening overall environmental management capacity.

Based on the above theoretical analysis, the first research hypothesis is proposed:

H1. 

Land supply structure distortion significantly inhibits the improvement of urban green economic efficiency.

3.2. The Moderating and Mediating Role of Housing Prices

3.2.1. Mediating Effect Analysis

Based on price transmission theory and market equilibrium theory, land supply structure distortion affects green economic efficiency by shaping the dynamics of the real estate market through supply constraints [36], market expectations, resource allocation, and innovation investment. In the relationship between land supply structure distortion and green economic efficiency, the mechanism through which housing prices exert their influence is primarily reflected in their asset attributes. Specifically, limited land supply reduces market elasticity, exacerbating imbalances, driving up housing prices, and distorting real estate investment structures. Institutional land policies shape market expectations, fueling speculative behavior and reinforcing price surges, which in turn influence investment decisions and capital distribution [37]. Rising housing prices further alter market equilibrium by redirecting resources toward real estate at the expense of industrial upgrading, leading to a pattern of de-industrialization. Additionally, higher housing costs suppress innovation by increasing R&D expenses, limiting talent acquisition, and weakening regional innovation capacity, ultimately constraining green economic efficiency. Based on the above theoretical analysis, the second research hypothesis is proposed:

H2. 

Housing prices play a significant mediating role in the process through which land supply structure distortion affects green economic efficiency.

3.2.2. Moderating Effect Analysis

According to institutional complementarity theory and system synergy theory, institutional factors do not operate in isolation but exhibit mutual reinforcement or constraint, creating a combined effect on market outcomes. In the context of the housing market, these interactions shape how land supply policies translate into housing price dynamics and, in turn, how housing prices influence broader urban development. Specifically, housing prices moderate the effects of land supply structure distortion through resource allocation, innovation, and environmental governance. Rising housing prices intensify resource misallocation by diverting financial and land resources away from productive or green industries toward speculative real estate activities, exacerbating market imbalances and further distorting investment distribution. Additionally, higher housing costs suppress innovation by increasing financial pressures on R&D and limiting technological advancements. Moreover, escalating housing prices amplify the challenges of environmental governance by constraining fiscal space for green infrastructure and increasing the marginal cost of environmental projects, thereby worsening the negative impact of land supply structure distortion on sustainable development. Based on the above analysis, the third research hypothesis is proposed:

H3. 

The higher the housing price level, the stronger the inhibitory effect of land supply structure distortion on green economic efficiency.

3.3. Spatial Correlation Effect Analysis

There is significant spatial dependence and heterogeneity in economic activities between regions [38]. The effects of land supply structure distortion may produce spatial spillover through the following mechanisms:

3.3.1. Policy Spillover Mechanism

Economic activities exhibit significant spatial dependence and heterogeneity across regions, leading to spatial spillover effects from land supply structure distortion. Based on local government competition theory and policy diffusion theory, regional land policies are interconnected through strategic interactions, demonstration effects, and policy coordination. Local governments engage in competitive policy adjustments to attract investment, expand fiscal revenue, and optimize industrial structures, while also imitating successful policies from pioneering regions. Additionally, shared regional development goals encourage policy coordination, reinforcing spatial spillover effects and shaping land supply strategies across different areas.

3.3.2. Market Spillover Mechanism

Drawing upon spatial equilibrium and factor flow theories, land supply structural variations generate significant spatial spillovers across regional markets. Price and demand dynamics in real estate propagate spatially because land adjustments in core cities trigger demonstration effects that drive up costs in peripheral areas. These structural shifts additionally reconfigure industrial layouts, where efficient allocation attracts premium clusters while land utilization inefficiencies prompt industrial relocation. Moreover, land policies catalyze the formation of factor networks by regulating the mobility of capital, labor, and technology. Such adjustments facilitate regional resource allocation, often pushing production factors from hubs limited by land scarcity toward surrounding territories.

3.3.3. Environmental Spillover Mechanism

Environmental issues exhibit strong spatial spillover effects, making the environmental impact of land supply structure distortion regionally interconnected. Changes in land supply can influence the spatial distribution of high-polluting enterprises, leading to cross-regional pollution diffusion. Additionally, excessive land development disrupts ecological balance by affecting water cycles and biodiversity, with consequences extending beyond local boundaries. Given the transboundary nature of environmental challenges, effective governance requires interregional collaboration, where local governments coordinate policies, share resources, and jointly manage issues like air pollution and water conservation to enhance environmental sustainability.

Based on the above analysis, this study proposes the fourth research hypothesis:

H4. 

Distorted land supply structures have significant spatial spillover effects on green economic efficiency.

4. Research Design

Based on the theoretical analysis framework, this study employs multiple econometric methods to empirically test the mechanisms through which land supply structure distortion affects urban green economic efficiency and the moderating effects of housing prices.

4.1. Model Specification

This study constructs the baseline regression model, spatial econometric model, mediation effect model, and moderating effect model to test the research hypotheses proposed earlier. The mediation effect model captures the transmission pathway through which land supply structure distortion influences green economic efficiency via housing prices as a key intermediary variable, thereby explaining how institutional distortions exert indirect effects. In contrast, the moderation effect model reveals how different levels of housing prices strengthen or weaken this relationship, reflecting the institutional complementarity and system synergy between land and real estate systems. Furthermore, integrating these approaches with the Spatial Durbin Model (SDM) provides a more comprehensive understanding of how land supply structure distortion in one region influences green economic efficiency in neighboring regions, as the SDM captures both direct and spillover effects.

4.1.1. Baseline Regression Model

To test Hypothesis H1, which posits that land supply structure distortion affects urban green economic efficiency, the following baseline model is constructed:

$${GEE}_{it}=\alpha +{\beta }_1{LDI}_{it}+{\sum }_{k=2}^K {\beta }_k{Controls}_{it}^k+{\mu }_i+{\lambda }_t+{\epsilon }_{it}$$

(1)

In this model, ${GEE}_{it}$ represents the green economic efficiency of the city $i$ in period $t$, ${LDI}_{it}$ represents the land supply structure distortion index, ${Controls}_{it}^k$ represents the k-th control variable, ${\mu }_i$ and ${\lambda }_t$ represent the city and time fixed effects, respectively, ${\epsilon }_{it}$ is the random disturbance term.

4.1.2. Spatial Econometric Model

Considering spatial dependence, in order to test Hypothesis H4, the Spatial Durbin Model (SDM) is constructed as follows:

$${GEE}_{it}=\alpha +\rho W{GEE}_{it}+\dot{{\beta }_1}{LDI}_{it}+\theta W\cdot {LDI}_{it}+{\sum }_{k=2}^K \dot{{\beta }_k} {Controls}_{it}^k+{\mu }_i+{\lambda }_t+{\epsilon }_{it}$$

(2)

where $W$ represents the spatial weight matrix, $\rho$ is the spatial autocorrelation coefficient, and $\theta$ is the spatial interaction effect coefficient. At the same time, the SAR model and SEM model are constructed for robustness checks.

4.1.3. Mediation Effect Model

In the context of land and real estate policy analysis, the impact of land supply distortions on green economic efficiency (GEE) is rarely direct. Instead, it typically operates through intermediate mechanisms, especially housing price. The mediation effect model allows us to disentangle this indirect pathway by decomposing the total effect into direct and indirect components. Therefore, to test Hypothesis H2, the stepwise regression method proposed by Baron and Kenny [39] is used to construct the mediation effect model:

$${HP}_{it}={\gamma }_0+{\gamma }_1{LDI}_{it}+{\sum }_{k=2}^K {\gamma }_k{Controls}_{it}^k+{\mu }_i+{\lambda }_t+{\eta }_{1it}$$

(3)

$${GEE}_{it}=\alpha +\ddot{{\beta }_1}{LDI}_{it}+\ddot{{\beta }_2}{HP}_{it}+{\sum }_{k=3}^K \ddot{{\beta }_k}{Controls}_{it}^k+{\mu }_i+{\lambda }_t+{\eta }_{2it}$$

(4)

In this model, ${HP}_{it}$ represents the housing price level indicator. According to the mediation effect testing procedure, the following conditions must be met: ${\gamma }_1$ should be significant, and both $\ddot{{\beta }_1}$ and $\ddot{{\beta }_2}$ should also be significant.

4.1.4. Moderating Effect Model

At the same time, housing prices also act as a moderator that shapes the strength and direction of policy impacts. In regions with excessively high housing prices, the negative effects of distorted land supply may be amplified, while in areas with relatively stable prices, the impact may be less pronounced. The moderating effect model thus helps reveal the heterogeneity of policy outcomes under different market conditions. Therefore, to test Hypothesis H3, a model with interaction terms is constructed as follows:

$${GEE}_{it}=\alpha +\stackrel{⃛}{{\beta }_1}{LDI}_{it}+\stackrel{⃛}{{\beta }_2}{HP}_{it}+\stackrel{⃛}{{\beta }_3}\left({LDI}_{it}\times {HP}_{it}\right)+{\sum }_{k=4}^K \stackrel{⃛}{{\beta }_k}{Controls}_{it}^k+{\mu }_i+{\lambda }_t+{\eta }_{2it}$$

(5)

According to Hypothesis H3, it is expected that the interaction term coefficient $\stackrel{⃛}{{\beta }_3}$ will be significantly negative.

4.2. Variable Definition and Measurement

4.2.1. Dependent Variable: Green Economic Efficiency (GEE)

In this study, the green economic efficiency (GEE) is measured using the super-efficiency DEA Malmquist model. Capital, labor, energy, land, and innovation inputs are treated as input factors, while real GDP and environmental benefits are considered as desirable outputs. Industrial wastewater, sulfur dioxide, and industrial smoke emissions are included as undesirable outputs, with their negative externalities internally constrained through penalty mechanisms. GEE thus reflects not only the relative level of green efficiency in the current period but also its dynamic evolution over time, capturing both static efficiency performance and intertemporal changes driven by technological progress and factor allocation adjustments. The model is constructed as follows:

$$GE{E}_{it}={\displaystyle \frac{{\sum }_{m=1}^M u_m{Y}_{mit}}{{\sum }_{n=1}^N v_n{X}_{nit}}}\cdot {\displaystyle \frac{1}{1+{\sum }_{j=1}^J w_j{B}_{jit}}}$$

(6)

In this model, the input factors include capital input (K), labor input (L), energy input (E), land input (T), and innovation input (R). The desired outputs include real GDP (Y) and environmental benefit indicators (ENV). The undesirable outputs include industrial wastewater (WP), sulfur dioxide (SP), and industrial dust (DP) emissions.

4.2.2. Core Explanatory Variable: Land Supply Structure Distortion Index (LDI)

Based on the multidimensional characteristics of the land supply structure, a comprehensive distortion index is constructed:

$$LD{I}_{it}={\sum }_{j=1}^3 {\alpha }_j{Z}_{jit}={\alpha }_{1}IL{R}_{it}+{\alpha }_{2}PL{G}_{it}+{\alpha }_{3}PL{D}_{it}$$

(7)

where $Z_j$ represents respectively:

$IL{R}_{it}$: The proportion of industrial, mining, and warehouse land in the total state-owned construction land supply.

$PL{G}_{it}$: The price ratio between industrial land and commercial residential land.

$PL{D}_{it}$: The proportion of land allocated through agreements in the total land transfer area.

The weight coefficients ${\alpha }_j$ are determined through principal component analysis (PCA):

$${\alpha }_j={\displaystyle \frac{{\lambda }_1{e}_{j1}}{{\sum }_{j=1}^3 {\lambda }_1{e}_{j1}}}, j=\mathrm{1,2},3$$

(8)

where ${\lambda }_1$ is the eigenvalue of the first principal component, and $e_{j1}$ is the corresponding eigenvector. The adoption of the principal component analysis (PCA) method in this study is based on several considerations. First, PCA ensures a high degree of objectivity. By extracting common factors to explain the variance structure of the original variables, it minimizes the potential subjectivity that may arise from manually assigning weights. Second, PCA facilitates information integration. Given the potential correlations among the three sub-indicators—for example, land-use structure and price distortion often exhibit interdependence—direct weighting may result in multicollinearity, whereas PCA reduces dimensionality while enhancing the explanatory power of the composite index. Third, PCA demonstrates strong economic rationality. Our calculation shows that the first principal component typically captures the common feature of land system distortion, with a high eigenvalue contribution rate, indicating that the PCA-based composite index provides a statistically sound representation of the overall distortion level.

4.2.3. Mediator Variable: Real Estate Market Characteristics (HP)

The comprehensive index for the real estate market is constructed using the entropy method:

$$H{P}_{it}={\sum }_{k=1}^3 w_k{X}_{kit}^{\mathrm{*}}$$

(9)

where the variables represent:

$X_{1it}^{\mathrm{*}}$: Standardized Housing Price Index (RPI)

$X_{2it}^{\mathrm{*}}$: Standardized Commercial Real Estate Price Index (CPI)

$X_{3it}^{\mathrm{*}}$: Standardized Housing Price Volatility (HPV)

The weights $w_k$ are calculated based on information entropy:

$$w_k={\displaystyle \frac{1-e_k}{{\sum }_{k=1}^3 \left(1-e_k\right)}}, e_k=-{\displaystyle \frac{1}{\ln n}}{\sum }_{i=1}^n p_{ik}\ln p_{ik}$$

(10)

4.2.4. Control Variables

Based on existing studies and theoretical considerations, a set of control variables was included to account for city-level heterogeneity. These variables are grouped into three categories, economic development characteristics (variables capturing the level of economic development, financial development, and openness to foreign investment), urban development characteristics (variables reflecting population agglomeration, industrial structure), and innovation and environmental characteristics (variables representing the innovation environment). The detailed definitions and measurement methods of all control variables are presented in the descriptive statistics table below (

Table 1. Descriptive Statistics of Key Variables.

Variable	Definitions	N	Mean	Std. Dev.	Min	Max
GEE	Green economic efficiency	3705	0.642	0.158	0.285	0.926
LDI	Land supply structure distortion index	3705	0.384	0.146	0.125	0.832
RPI	Standardized Housing Price Index	3705	1.058	0.124	0.862	1.465
EL	Economic development level (the natural logarithm of GDP per capita)	3705	10.856	0.685	8.956	12.458
FIN	Financial development level (the balance of foreign and domestic currency loans from financial institutions at the end of the year/GDP)	3705	1.246	0.458	0.485	2.865
FDI	Foreign investment level (the natural logarithm of the actual utilized foreign capital)	3705	9.856	1.246	5.865	12.985
PA	Population agglomeration degree (the natural logarithm of population density)	3705	6.485	0.958	4.258	8.956
IND	Industrial structure (the ratio of the value-added of the tertiary industry to GDP)	3705	0.485	0.086	0.256	0.685
INNO	Innovation environment (the number of invention patents granted per 10,000 people)	3705	8.562	12.458	0.125	85.652

).

4.3. Data Sources and Descriptive Statistics

4.3.1. Data and Sample Selection

This study covers a research sample of 285 prefecture-level and above cities in China from 2010 to 2022. The main data sources include the China Statistical Yearbook, regional statistical yearbooks, the CEIC statistical database, and the National Bureau of Statistics of China. The selection criteria were as follows: ensuring data availability and continuity, excluding cities with administrative boundary adjustments, and removing samples with missing key variables. The final sample accounts for 97.27% of China’s 293 prefecture-level cities.

4.3.2. Descriptive Statistics

Table 1 reports the descriptive statistics of the key variables. Overall, the average value of the urban green economic efficiency (GEE) is 0.642, with a standard deviation of 0.158. The minimum and maximum values are 0.285 and 0.926, respectively, indicating significant differences in the green economic efficiency across cities in China. The average value of the land supply structure distortion index (LDI) is 0.384, with a standard deviation of 0.146, suggesting that the phenomenon of land supply structure distortion is widespread and varies in degree.

5. Empirical Result Analysis

5.1. Benchmark Regression Results

Table 2. The impact of land supply structure distortion on green economic efficiency.

Variable	The Dependent Variable: Green Economic Efficiency (GEE)
	(1)	(2)	(3)
LDI	−0.326 ***	−0.315 ***	−0.308 ***
	(0.042)	(0.040)	(0.038)
EL	0.245 ***	0.238 ***
	(0.035)	(0.033)
FIN	0.156 ***	0.148 ***
	(0.028)	(0.026)
Control Variables	No	Yes	Yes
City Fixed Effects	Yes	Yes	Yes
Time Fixed Effects	Yes	Yes	Yes
Obs.	3705	3705	3705
R2	0.385	0.426	0.458

Note: The values in parentheses represent clustered robust standard errors; *** indicates significance at the 1% level.

presents the benchmark regression results. Column (1) includes only the core explanatory variables, Column (2) adds control variables, and Column (3) further controls for fixed effects. It can be seen after controlling for other factors, the coefficient of the Land Supply Structure Distortion Index (LDI) is consistently significantly negative and significant at the 1% level. Specifically, each standard deviation increase in LDI leads to a 0.308 standard deviation decrease in green economic efficiency. This result robustly supports Hypothesis H1, indicating that land supply structure distortion significantly inhibits the improvement of urban green economic efficiency. The negative effect of distorted land supply structures on urban green economic efficiency indicates that the current land supply system has become a key constraint on green transformation. From a policy perspective, efforts should focus on optimizing land supply structures, enhancing regional coordination, and promoting industrial upgrading and innovation to foster the development of the green economy, ultimately achieving a win-win outcome for both the economy and the environment.

5.2. Endogeneity Treatment Results

We conduct the Instrumental Variable (IV) validity test and present the results of both the IV tests and the 2SLS estimation. Here, we select terrain ruggedness index (TRI) as the main instrumental variable:

$$TR{I}_i=\sqrt{{\displaystyle \frac{\left(\mathrm{m}\mathrm{a}\mathrm{x}\right(ele{v}_i)-\mathrm{m}\mathrm{i}\mathrm{n}(ele{v}_i){)}^2}{A_i}}}$$

(11)

${TRI}_i$ represents the terrain ruggedness of different cities, satisfying both relevance and exogeneity requirements for an instrumental variable (IV) for land supply structure distortion. On the one hand, TRI directly affects the expandability and development cost of urban construction land, thereby imposing an objective constraint on local governments’ land supply structures. On the other hand, as a natural geographic condition, TRI is highly exogenous in its formation and does not directly influence green economic efficiency. Thus, it satisfies the fundamental requirements of instrument relevance and exogeneity. Nevertheless, this instrument also has certain limitations. First, although terrain conditions exert long-term constraints on land supply patterns, their influence may indirectly affect green economic efficiency through channels such as population density and industrial layout, making it challenging to fully rule out potential exogeneity risks. Second, the spatial heterogeneity of terrain ruggedness is pronounced, which may cause the explanatory power of the IV to vary across regions and thus affect the robustness of the estimation.

Table 3. The results of IV test.

	Statistic	p-Value
First-stage F	23.56	0.000
Cragg Donald Wald F	26.85	-
Stock–Yogo critical values	19.93	-
Sargan Test	1.246	0.238

reports the outcomes of the instrumental variable tests. The first-stage F-statistic is 23.56, well above the conventional threshold of 10, thereby rejecting the null hypothesis of weak instruments. The Sargan test yields a p-value of 0.238, indicating that we cannot reject the null hypothesis of instrument exogeneity. This suggests that the instrumental variables used in this study are valid.

Table 4. 2SLS Estimation Results.

Variable	First Stage LDI	Second Stage GEE
TRI	0.285 ***	-
	(0.042)
LDI(IV)	-	−0.412 ***
		(0.056)
EL	−0.142 ***	0.256 ***
	(0.035)	(0.045)
FIN	−0.126 ***	0.185 ***
	(0.032)	(0.038)
FDI	−0.108 ***	0.165 ***
	(0.028)	(0.036)
Control Variables	Yes	Yes
City Fixed Effects	Yes	Yes
Time Fixed Effects	Yes	Yes
Observations	3705	3705
R-squared	0.385	0.426

Note: The values in parentheses represent clustered robust standard errors; *** indicates significance at the 1% level.

presents the 2SLS estimation results. After addressing endogeneity, the suppressive effect of land supply structure distortion on green economic efficiency becomes more pronounced: the coefficient increases from −0.308 in the baseline regression to −0.412. This implies that the baseline regression likely underestimated the negative impact of land supply structure distortion.

5.3. Mediation and Moderation Effect Tests

5.3.1. Mediation Effect Analysis

Table 5. Mediation Effect Test Results.

Steps	Dependent Variable	Independent Variable	Coefficient	Standard Error	95% Confidence Interval
Step 1	GEE	LDI	−0.326 ***	0.042	[−0.408, −0.244]
Step 2	RPI	LDI	0.285 ***	0.038	[0.210, 0.360]
Step 3	GEE	LDI	−0.285 ***	0.040	[−0.363, −0.207]
		RPI	−0.196 ***	0.035	[−0.265, −0.127]

Sobel Test: Z = −4.856 ***. Proportion of Mediation Effect: 23.4%. Note: *** indicates significance at the 1% level.

presents the stepwise test results for the mediation effect of housing prices. The results show that land supply structure distortion (LDI) has a significant negative direct effect on green economic efficiency (GEE) (−0.326 ***). Meanwhile, LDI exerts a significant positive impact on housing prices (RPI) (0.285 ***). When RPI is included as a mediating variable, the absolute value of the coefficient of LDI on GEE decreases but remains statistically significant. The Sobel test confirms the mediation effect at the 1% significance level, with the indirect effect accounting for 23.4% of the total effect. These findings support Hypothesis H2, indicating that housing prices play a significant partial mediating role in the impact of land supply structure distortion on green economic efficiency. Nearly one-quarter of the negative effect of land supply structure distortion on green economic efficiency operates through mechanisms such as pushing up housing prices, crowding out innovation investment, and distorting resource allocation.

5.3.2. Moderation Effect Analysis

Table 6. Moderation Effect Test Results.

Variable	The Dependent Variable: Green Economic Efficiency (GEE)
	(1)	(2)	(3)
LDI	−0.326 ***	−0.318 ***	−0.312 ***
	(0.042)	(0.040)	(0.039)
RPI	−0.196 ***	−0.188 ***	−0.182 ***
	(0.035)	(0.033)	(0.032)
LDI × RPI	−0.167 ***	−0.162 ***	−0.158 ***
	(0.028)	(0.026)	(0.025)
Control Variables	No	Yes	Yes
City Fixed Effects	Yes	Yes	Yes
Time Fixed Effects	Yes	Yes	Yes
Observations	3705	3705	3705
R-squared	0.425	0.456	0.485

Note: The values in parentheses represent clustered robust standard errors; *** indicates significance at the 1% level.

presents the test results for the moderation effect of housing prices. It can be seen that the coefficient of the interaction term is significantly negative (−0.167 ***), indicating that the higher the housing prices, the stronger the suppressive effect of land supply structure distortion on green economic efficiency, which supports Hypothesis H3. For each one standard deviation increase in housing prices, the negative impact of land supply structure distortion on green economic efficiency intensifies by approximately 16%. This result reveals a negative synergistic effect between land system distortion and housing price levels: in regions with relatively low housing prices, the adverse effect of land supply structure distortion is limited, whereas in regions with high housing prices, the same level of distortion leads to more severe efficiency losses.

5.4. Spatial Effect Test

5.4.1. Spatial Autocorrelation Test

Table 7. Spatial Autocorrelation Test Results.

Test Type	Statistic	p-Value
Global Moran’s I	0.326	0.000
LM lag	56.85	0.000
Robust LM lag	48.26	0.000
LM error	42.58	0.000
Robust LM error	35.96	0.000

reports the results of the spatial autocorrelation test. The Global Moran’s I index is 0.326 and highly significant, indicating a significant positive spatial autocorrelation in urban green economic efficiency, with a spatial pattern of “high–high” and “low–low” clustering. Both the LM lag and LM error test statistics are highly significant; however, the Robust LM lag statistic (48.26) is greater than the Robust LM error statistic (35.96). Combined with theoretical analysis, this supports the use of the Spatial Durbin Model (SDM), suggesting that green economic efficiency exhibits not only spatial lag effects but also spatial interaction effects of the core explanatory variables.

5.4.2. Spatial Econometric Analysis Results

The estimation results of the spatial econometric models are reported in

Table 8. The estimation results of the spatial econometric models.

Variable	The Dependent Variable: Green Economic Efficiency (GEE)
	SDM	SAR	SEM
LDI	−0.285 ***	−0.276 ***	−0.268 ***
	(0.038)	(0.036)	(0.035)
W × LDI	−0.156 ***	−0.148 ***	−0.142 ***
	(0.026)	(0.025)	(0.024)
ρ/λ	0.285 ***	0.276 ***	0.268 ***
	(0.038)	(0.036)	(0.035)
Direct Effect	−0.298 ***	−0.285 ***	−0.275 ***
	(0.040)	(0.038)	(0.037)
Indirect Effect	−0.168 ***	−0.162 ***	−0.156 ***
	(0.028)	(0.027)	(0.026)
Total Effect	−0.466 ***	−0.447 ***	−0.431 ***
	(0.056)	(0.054)	(0.052)
Control Variables	Yes	Yes	Yes
Fixed Effects	Yes	Yes	Yes
Log-likelihood	2856.5	2845.8	2838.6
AIC	−5685.0	−5663.6	−5649.2

Note: 1. The values in parentheses are the robust standard errors for spatial heterogeneity; 2. *** represents significance at the 1% level; 3. AIC stands for the Akaike Information Criterion.

. To ensure the robustness of the results, the Spatial Durbin Model (SDM), Spatial Autoregressive Model (SAR), and Spatial Error Model (SEM) were simultaneously estimated. Results show that the spatial autocorrelation coefficients (ρ/λ) are significantly positive in all models, confirming the presence of spatial correlation; the spatial lag term (W × LDI) is significantly negative, indicating that land supply structure distortion has a negative spatial spillover effect. Based on the AIC criterion, the SDM model provides the best fit. This supports Hypothesis H4, i.e., that land supply structure distortion has a significant spatial spillover effect on green economic efficiency.

5.5. Heterogeneity Analysis

Table 9. The results of the heterogeneity test.

Variable	Regional Grouping			Urban Level Grouping
	East	Central	West	First-Tier Cities	Second-Tier Cities	Third- and Fourth-Tier Cities
LDI	−0.468 ***	−0.312 ***	−0.186 ***	−0.526 ***	−0.384 ***	−0.245 ***
	(0.058)	(0.045)	(0.036)	(0.065)	(0.048)	(0.038)
RPI	−0.245 ***	−0.186 ***	−0.142 ***	−0.285 ***	−0.212 ***	−0.168 ***
	(0.042)	(0.035)	(0.028)	(0.045)	(0.036)	(0.032)
LDI × RPI	−0.195 ***	−0.156 ***	−0.108 ***	−0.225 ***	−0.176 ***	−0.132 ***
	(0.032)	(0.028)	(0.022)	(0.038)	(0.032)	(0.026)
Control Variables	Yes	Yes	Yes	Yes	Yes	Yes
Fixed Effects	Yes	Yes	Yes	Yes	Yes	Yes
Observations	1245	1230	1230	145	860	2700
R-squared	0.485	0.442	0.398	0.526	0.468	0.412

Note: The values in parentheses represent clustered robust standard errors; *** indicates significance at the 1% level.

presents the results of the heterogeneity tests based on both regional and urban-level groupings. As shown, the impact of land supply structure distortion on green economic efficiency varies notably across regions. Specifically, the effect is most pronounced in the eastern region (−0.468 ***), followed by a moderate effect in the central region (−0.312 ***), while the western region exhibits a relatively weaker impact (−0.186 ***). Moreover, when cities are grouped according to their administrative level, a similar gradient pattern emerges. The negative effect is strongest in first-tier cities (−0.526 ***), somewhat smaller in second-tier cities (−0.384 ***), and weakest in third- and fourth-tier cities (−0.245 ***). Overall, these findings suggest that the adverse influence of land supply structure distortion on green economic efficiency is more severe in economically developed and higher-tier cities, likely due to their greater dependence on land-based fiscal revenues and higher housing market sensitivity.

5.6. Robustness Check

5.6.1. Robustness of Alternative Key Variables

To ensure the reliability of the research findings,

Table 10. Robustness test with alternative key variables.

Test Type	Equal-Weight LDI	Directional Distance Function (GEE)	Winsorization
Coefficient	−0.286 ***	−0.295 ***	−0.295 ***
Standard Error	(0.039)	(0.040)	(0.037)
Observations	3705	3705	3705
R-squared	0.442	0.435	0.455

Note: The values in parentheses represent clustered robust standard errors; *** indicates significance at the 1% level.

conducts robustness checks by reconstructing the core variables using different methods. The results show that when the land supply structure distortion index (LDI) is reconstructed using the equal-weight method, the coefficient is −0.286, which is largely consistent with the baseline regression results. When green economic efficiency (GEE) is recalculated using the directional distance function, the coefficient is −0.295, also supporting the original conclusion. After applying 1% winsorization to the sample to exclude the influence of outliers, the coefficient remains −0.295, indicating that extreme values are not the main driver of the results. The R² values across all tests remain within the range of 0.435–0.455, suggesting stable explanatory power of the model. These results indicate that regardless of the indicator construction method used, the negative effect of land supply structure distortion on green economic efficiency remains robust.

5.6.2. Robustness of Estimation Methods

Table 11. Robustness Tests with Alternative Estimation Methods.

Estimation Method	System GMM	Quantile Regression (25%)	Quantile Regression (50%)	Quantile Regression (75%)
Coefficient	−0.324 ***	−0.278 ***	−0.305 ***	−0.341 ***
Standard Error	(0.055)	(0.038)	(0.041)	(0.046)
95% CI	[−0.432, −0.216]	[−0.352, −0.204]	[−0.385, −0.225]	[−0.431, −0.251]
AR(2) Test	0.185	-	-	-
Hansen Test	0.312	-	-	-
Pseudo R2	-	0.312	0.308	0.305

Note: The values in parentheses represent clustered robust standard errors; *** indicates significance at the 1% level.

validates the robustness of the baseline regression results using different estimation methods. The System GMM estimation shows that the coefficient of land supply structure distortion is −0.324, with the AR(2) test p-value of 0.185 and the Hansen test p-value of 0.312, indicating no evidence of second-order serial correlation or over-identification. The quantile regression reveals pronounced heterogeneity: for low-efficiency cities (25th quantile), the coefficient is −0.278; for medium-efficiency cities (50th quantile), it is −0.305; and for high-efficiency cities (75th quantile), it is −0.341. This increasing pattern indicates that land supply structure distortion exerts a more severe negative impact on cities with higher green economic efficiency. These results are highly consistent with the previous spatial econometric analysis, further verifying the presence of spatial dependence.

6. Research Conclusions and Policy Recommendations

6.1. Main Research Conclusions

This study, based on panel data from 285 cities at or above the prefecture level in China between 2010 and 2022, systematically examines the impact mechanism of land supply structure distortion on urban green economic efficiency and the moderating role of housing prices. The main conclusions are as follows:

6.1.1. Overall Effects

(1)

Significant Suppressive Effect

Empirical results show that land supply structure distortion significantly suppresses the improvement of urban green economic efficiency. Previously, several scholars have emphasized the importance of green economic efficiency and have considered resource and environmental constraints [40]. However, this study proposes a new theoretical framework that directly incorporates distortions in the land supply structure as a factor influencing green economic efficiency, thereby revealing the critical role of imbalances in land supply structure in shaping green economic efficiency. Specifically, a one standard deviation increase in the land supply structure distortion index leads to a 0.308 standard deviation decrease in green economic efficiency. After controlling for endogeneity, this suppressive effect is further strengthened to −0.412 standard deviations. This result remains robust across various robustness tests.

(2)

Spatial Spillover Effect

Spatial econometric analysis shows significant spatial correlation: the spatial autocorrelation coefficient is 0.285 (p < 0.01). The direct effect is 0.298, and the indirect effect is 0.168, with a total effect of 0.466, indicating that spatial correlation amplifies the negative impact.

(3)

Transmission Mechanism

Regarding the transmission mechanisms, this study analyzes the mediating and moderating role of housing prices in the relationship between land supply structure distortions and urban green economic efficiency, offering a novel perspective for understanding the link between land supply and green economic performance. While previous research, such as by Kong et al., has examined the impact of housing prices on economic growth [41], this study is the first to systematically explore how housing prices function as a moderator in the pathway from distorted land supply structures to green economic efficiency. The mediation and moderation effect tests show that approximately 23.4% of the impact is transmitted through housing prices. Each one standard deviation increase in housing prices enhances the suppressive effect by 16.7%. The negative impact of land supply structure distortion is more significant when housing prices are high.

6.1.2. Heterogeneity Features

(1)

Regional Heterogeneity

Different regions exhibit significant differences: the eastern region has the most significant impact (−0.468), followed by the central region (−0.312), and the western region has a relatively weaker impact (−0.186). Regional differences are closely related to economic development levels and industrial structures.

(2)

City Level Heterogeneity

Different levels of cities show systematic differences: first-tier cities have the largest impact (−0.526), followed by second-tier cities (−0.384), and third- and fourth-tier cities show relatively smaller impacts (−0.245). The differences primarily arise from the development level of the real estate market and industrial agglomeration characteristics.

6.2. Policy Recommendations

Based on the research findings, several policy recommendations are proposed.

First, urban governments should establish a green-oriented differentiated pricing mechanism for industrial land and a land adjustment mechanism linked to housing price fluctuations. By increasing land costs for high-energy-consuming and highly polluting industries, cities can encourage low-carbon industrial transformation. Meanwhile, improving the allocation efficiency and market stability of commercial and residential land supply can help optimize the supply–demand structure and mitigate financial risks.

Furthermore, it is essential to develop cross-regional land policy coordination and ecological compensation mechanisms to promote the diffusion of green technologies, foster industrial collaboration, break down administrative barriers, optimize resource allocation, and enhance regional ecological efficiency and sustainability. A regionally differentiated strategy should also be implemented: the eastern region should focus on industrial upgrading and intensive land use; the central region should balance green transformation with economic growth; and the western region should adhere to an ecology-first principle, exploring pathways for realizing the value of ecological products.

Finally, policy tools should be dynamically adjusted according to the stage of urban development to improve both the precision and effectiveness of policy implementation.