LRM refers to the inefficient allocation of land across industrial sectors and functional uses, often stemming from government intervention deviating from market-oriented mechanisms (Zhou et al., 2023). In China, such distortions are reflected not only in the quantity or spatial distribution of land supply but also deeply embedded in land pricing structures and property rights systems (Huang et al., 2025). A prominent manifestation of this phenomenon is that local governments, driven by short-term fiscal revenue and economic growth targets, allocate land at artificially low prices to high-energy-consuming industries (An, 2024; Chen and Yuan, 2025), effectively reducing land use costs for carbon-intensive sectors (Gao et al., 2022). This practice distorts urban spatial development patterns, constrains the emergence of low-carbon industries, and undermines the competitiveness of green and innovative enterprises in the land market, thereby limiting their expansion and contribution to carbon reduction (Jiang et al., 2022; Wu et al., 2023). Such an institutional bias exacerbates the feedback loop between resource misallocation, industrial path dependence, and carbon lock-in, locking cities into a high-emission growth trajectory (Cheng Y. et al., 2025). Furthermore, existing literature suggests that the concentration of land allocation in low-productivity sectors also leads to structural inefficiencies, reflected in declines in total factor productivity (TFP). This misallocation reinforces traditional high-pollution industrial structures, further entrenching rigid emission patterns and reducing the overall carbon emission efficiency (Gao and He, 2024). For instance, Zhou et al. (2022), based on panel data from Chinese cities, found that higher degrees of land misallocation were associated with poorer urban carbon performance. Summarily, we argue that LRM has become a significant structural barrier to improving the efficiency of urban carbon emissions. Hence, we propose the following hypotheses:

H1

LRM hinders improvements in UCEE.

In the context of carbon neutrality, the optimization and upgrading of industrial structures are widely recognized as key pathways for improving the efficiency of urban carbon emissions. The core lies in transforming urban economies from traditional high-carbon, resource-intensive industries toward green, low-carbon, and high-value-added emerging sectors (Zhou et al., 2022), ultimately targeting the development of industries characterized by elevated value capture, advanced technological embeddedness, and eco-efficient production paradigms (Ran et al., 2023). This transformation not only helps reduce carbon intensity and improve energy structure but also enhances green productivity at the urban level through technological advancement (Chen and Yuan, 2025). However, as a typical institutional distortion, land–resource misallocation may serve as a suppressive mechanism during this transition. The misallocation compresses the accessibility of land for high-tech industries and modern services, hampers spatial agglomeration and capital-deepening of green industries, and thus weakens the advancement of industrial upgrading (Xie et al., 2022; Cheng G. et al., 2025). Furthermore, resource dependence impedes the advancement and upgrading of industrial structure (Wang et al., 2019). The suboptimal allocation of land factors may reinforce existing high-carbon industrial structures through “path dependence,” resulting in a “lock-in effect” preventing the transition to more advanced industrial forms and undermines carbon efficiency (Cheng G. et al., 2025). Therefore, we argue that LRM may suppress improvements in UCEE by obstructing ISU. Accordingly, we propose the following hypothesis:

H2

LRM inhibits the improvement of UCEE by impeding ISU.

Under the guidance of the “dual carbon” goals, GTI is widely regarded as a critical lever to enhance UCEE. Its essence lies in decoupling economic growth from carbon emissions by improving resource utilization efficiency and environmental performance through technological progress and institutional innovation (Deng et al., 2019; Nan et al., 2022). The Green Technological Innovation encompasses not only clean production, energy-saving technologies, and renewable energy, but also green transformation in institutions, management, and business models, forming the core of urban green development capacity that aligns with the long-term sustainability paradigm (Li and Liao, 2020). As land is an essential input for green innovation, its allocation directly influences the formation of the green technological capacity of a city. The existence of LRM may hinder the GTI, thereby weakening UCEE. An imbalanced allocation of industrial land increases the carbon intensity per unit of economic output and limits the potential for large-scale adoption of clean technologies (Chu et al., 2019). For example, when land is preferentially allocated to polluting and energy-intensive industries, green firms face higher land costs, weaker infrastructure support, and insufficient innovation network connections, ultimately suppressing investments in green R&D and the diffusion of sustainable technologies (Du and Li, 2021). Contrarily, green industries and service sectors depend on high-quality land and a supportive ecological environment; however, under a distorted land allocation regime, such sectors are often marginalized in urban space, weakening the capacity for sustainable development of the city (Cheng G. et al., 2025). Based on this logic, we propose the following hypothesis.

H3

LRM inhibits the improvement of UCEE by hindering the development of GTI.

We used panel data from 274 prefecture-level cities in China over a 13-year period from 2010 to 2022, comprising 3,562 observations, to investigate the relationship between LRM and carbon emissions efficiency, as well as its underlying mechanisms. Observations with missing values (city-year pairs) were excluded to ensure consistency in the alignment of all variables, and 3,004 valid observations were ultimately retained for the benchmark regression. All data were analyzed using the Stata software. To address heteroscedasticity, robust standard errors were used in all regressions. To mitigate the influence of outliers, all continuous variables are winsorized at the top and bottom 1%.

The data for each indicator were sourced as follows: LRM data were obtained from the China Urban Construction Statistical Yearbook and carbon emission efficiency data were derived from the CEADS database. Data on ISU, population density, human capital, financial development level, foreign direct investment, and per-capita fiscal expenditure were collected from statistical yearbooks, bulletins, and statistical bureaus at various government levels in China. GTI data are obtained from the CNRDS database. The environmental regulation intensity was compiled from government work reports published on official government websites.

Following the methodology of Ma et al. (2025), we constructed a two-way fixed effects model in a benchmark regression to capture the inhibitory effect of LRM on carbon emissions efficiency across cities and over time, thereby testing Hypothesis 1.

Baseline regression model: U C E E c t = α c + α t + α 0 + β 1 L R M c t + γ X c t + ε c t (6)

In the model, c, t, α _c and α _t , α ₀ , ε _ct , UCEE _ct , and γ X _ct denote the city, year, capture city and year fixed effects, constant term, error term, carbon emission efficiency, and a vector of control variables, respectively.

To better understand the characteristics of the research sample and obtain an overview of the panel data, we conducted a descriptive statistical analysis of the main variables. The analysis was performed using the Stata software, and the results are presented in Table 1.

The dependent variable, UCEE, has a mean value of 0.150, a median of 0.302, and a standard deviation of 0.115, indicating a certain degree of variation in carbon efficiency, which provides a sound foundation for further empirical investigation. The independent variable, LRM, shows a mean of 18.64 and a median of 19.00, suggesting that the overall level of the variable is concentrated around 19. Additionally, the descriptive statistics of the other variables indicated that all variables fell within a reasonable range, confirming the appropriateness of the sample selection in this study.

Table 2 presents the results of correlation analyses. At the preliminary level, LRM appeared to be negatively correlated with UCEE, although the relationship was not statistically significant, indicating the need for further investigation. Additionally, most p-values in the correlation test were <0.01, suggesting that the variables exhibited strong correlations at the 1% significance level.

Multicollinearity refers to a strong correlation among the explanatory variables in a regression analysis, which may result in unstable coefficient estimates and reduced statistical significance. The Variance Inflation Factor (VIF) test is commonly used to detect multicollinearity. Here, we applied the VIF test to examine the multicollinearity among the explanatory variables. As shown in Table 3, all the VIF values were below the critical threshold of 10, indicating that multicollinearity was not a concern. Therefore, multicollinearity does not threaten the validity of the empirical results.

The baseline regression employed a two-way fixed effects model and adopted a stepwise regression approach. The results of Equation 6 are presented in Table 4. In the first step, the control variables were excluded. The estimation results show that the coefficient of the variable LRM (β₁) is negative and statistically significant at the 1% level. In the second step, after including the control variables, the coefficient of LRM (β₁) remains significantly negative at the 1% level (coefficient = -0.001, p < 0.01). Therefore, Hypothesis 1 is supported, confirming that LRM significantly reduces UCEE.

Given the potential over-identification issues associated with the traditional three-step approach to causal mechanism identification, we adopt a revised two-step empirical strategy inspired by Zhou et al. (2023) and Qing et al. (2024). This approach was designed to empirically test the mechanism by which LRM affects UCEE. The first step involved empirically testing the effect of LRM on the two mediating variables of ISU and GTI. In the second step, rather than conducting an additional regression analysis, we rely on existing authoritative literature and logical inferences to verify the established correlations between the mediating variables and carbon emissions efficiency. This allowed us to infer the mediating role of ISU and GTI in the relationship between LRM and UCEE.

The corresponding models are presented in Equations 7, 8: I S U c t = α c + α t + α 0 + β 1 L R M c t + γ X c t + ε c t (7) G T I c t = α c + α t + α 0 + β 1 L R M c t + γ X c t + ε c t (8)

Table 7 presents the results. Column (1) shows that the coefficient of LRM on ISU is negative and statistically significant at the 5% level (coefficient = −0.002, p < 0.05), indicating that LRM inhibits ISU. Column (2) shows that the coefficient of LRM on GTI is also negative and significant at the 5% level (coefficient = −0.005, p < 0.05), suggesting that LRM impedes the development of GTI.

Regarding the correlation between ISU, GTI, and UCEE, experts and scholars provided substantial evidence. Studies showed that the advancement and upgrading of industrial structures can, to a certain extent, suppress carbon emissions in neighboring regions and significantly improve urban carbon emission efficiency (Deng et al., 2023; Cheng Y. et al., 2025). Miao et al. (2017) and Liao et al. (2024) empirically validated the relationship between Green Technological Innovation and urban carbon emission efficiency. Their findings indicate that green innovation plays a key role in promoting low-carbon transformation and has a significantly positive impact on improving UCEE, a view widely acknowledged in academic circles.

Summarily, LRM impedes ISU, thereby suppressing improvements in UCEE; similarly, it hinders the advancement of GTI, which in turn limits the enhancement of UCEE. These two causal pathways are well substantiated, thus confirming hypotheses 2, 3.

Based on panel data from Chinese cities, we constructed a two-way fixed effects model to systematically assess the impact of LRM on UCEE. Furthermore, it empirically examines two mediating mechanisms, ISU and GTI, through which misallocation exerts indirect effects. The main findings were as follows:

First, LRM significantly hinders the improvement of UCEE, confirming that such institutional distortions lead to higher carbon emissions per unit output. This finding is consistent with that of Zhou et al. (2022), who showed that a higher degree of misallocation between industrial and commercial land is correlated with lower carbon efficiency. Similarly, Cheng Y. et al. (2025) quantitatively estimated that a 1% increase in the land misallocation index results in an average increase of 0.502% in urban carbon emissions. This implies that the oversupply of low-cost industrial land by the local governments aimed at rapid development promotes the expansion of energy-intensive industries, reduces energy efficiency, and substantially suppresses carbon emission efficiency (Han and Huang, 2022). Generally, the recent studies overwhelmingly confirms the significant negative effect of LRM on carbon efficiency. However, not all studies reached consistent conclusions. For instance, Chen and Yuan (2025) argued that the negative effect of land misallocation on carbon efficiency is significantly weakened in regions where land marketization reforms progressed. This finding suggests that in economically advanced regions with sound market mechanisms, the inhibitory effect of land misallocation is relatively weak. Therefore, differences in regional development stages and policy environments explain the variations in empirical results across studies.

Second, the mechanism analysis indicates that LRM reduces carbon emission efficiency by obstructing the transition of the industrial structure from high-carbon to low-carbon sectors, thereby inhibiting industrial upgrading. Numerous studies support the notion that barriers to industrial transformation are a critical transmission channel through which land misallocation affects carbon efficiency (Zhou et al., 2022; Cheng Y. et al., 2025). Zhou et al. (2022) explicitly stated that land misallocation delays industrial upgrading, increases the proportion of highly polluting and energy-intensive industries, and suppresses carbon efficiency improvements. Cheng Y. et al. (2025) further demonstrated through a mediation analysis that approximately 16.3% of the total effect of land misallocation is transmitted via changes in industrial structure. This provides additional evidence that land misallocation indirectly hinders the shift from high-carbon to low-carbon industries, thus lowering the CEs. However, Cheng Y. et al. (2025) also highlighted that industrial upgrading accounts for only a small part of the total effect, suggesting the existence of other influential mechanisms beyond the scope of structural transformation.

Third, our study shows that LRM suppresses GTI, further weakening UCEE. Several scholars argue that the distorted allocation of land factors restricts the concentration and input of green innovation, thereby indirectly reducing carbon efficiency (Han and Huang, 2022). Han and Huang (2022) provided empirical evidence that land misallocation significantly inhibits green innovation activities and weakens economic agglomeration, leading to increased emissions. Xu et al. (2025) find that land misallocation has a direct and significant negative impact on urban green innovation capacity, operating through structural, scale, and spatial agglomeration. Similarly, Chen and Yuan (2025) showed that misallocated land leads to insufficient investment in green R&D and delays in technological advancement, thus reducing the driving force for emission reduction. However, other studies offer different perspectives. Zhou et al. (2023), through mechanism analysis, empirically identify economic agglomeration, industrial structure, and urbanization as joint mediators of the impact of land misallocation on energy efficiency, without highlighting green innovation as a channel. This may be due to the differences in the measurements of energy efficiency versus carbon efficiency. Overall, the majority of the literature supports the pathway of “land misallocation → inhibited green innovation → reduced carbon efficiency,” though a few studies show divergent conclusions due to differences in research focus or the selection of control variables. This highlights the importance of considering multiple research perspectives and carefully selecting appropriate covariates.

Fourth, regarding heterogeneity, existing studies generally agree that land resource misallocation affects urban carbon emission efficiency, and also emphasize its heterogeneous effects based on city location (An, 2024), population size (Gao et al., 2023), and resource endowment (Zhou et al., 2022), which are consistent with those of the heterogeneity dimensions examined here. Our findings show that the inhibitory effect of land misallocation on carbon efficiency is significant only in the central and western regions (An, 2024), small and medium-sized cities (Gao et al., 2023), and non-resource-based cities (Zhou et al., 2022), aligning with that of most scholarly conclusions. However, Zhou et al. (2022) report that the effect is also significant in resource-based cities, which they attribute to the lagged effects of land policies—i.e., the land misallocation in the earlier period may affect current carbon efficiency. This does not contradict our findings as the lag effect may explain short-term inconsistencies in resource-based cities, whereas our analysis focused on the average effect over time.

This study contributes to the literature in both theoretical and methodological dimensions.

From a theoretical perspective, first, it supplements the institutional explanation of land-resource allocation from an environmental economics perspective. It explicitly highlights that land misallocation is not merely an issue of allocation efficiency but also one with far-reaching environmental consequences. Second, it constructs a dual-mechanism mediation framework of “Land Resource Misallocation–Industrial Structure Upgrading/Green Technological Innovation–Urban Carbon Emission Efficiency,” addressing the limitations of earlier studies regarding the identification of transmission pathways (Gao and He, 2024; Cheng Y. et al., 2025).

From a methodological perspective, by introducing the interaction between the average terrain slope and the annual economic growth target of a city as an instrumental variable, we enhance the credibility of causal inferences regarding LRM and enrich methodological applications in environmental economics to tackle endogeneity problems.

From a policy perspective, our findings offer the following recommendations for promoting green development and optimizing land systems in China.

(1) Accelerate market-oriented reforms in land resource allocation, and break away from the administratively driven logic of land supply—especially by avoiding the preferential allocation of low-cost land to high-emission industries, to curb misallocation at its source.

This study has some limitations.

First, although a comprehensive panel dataset was constructed and a two-way fixed-effects model was employed to explore the linear impact of LRM on UCEE, the spatial dependence and regional spillover effects of land misallocation were not fully addressed. Future research could incorporate spatial econometric models or multiscale geographically weighted regression (MGWR) models and update the dataset with post-2023 observations to further investigate the spatial transmission mechanisms and nonlinear characteristics of the influence of land misallocation on carbon efficiency across cities.

Second, in terms of mechanism analysis, future studies may consider introducing structural equation modeling (SEM) to enhance causal inference in mediating effect analysis. Furthermore, as current measurement indicators rely largely on static city-level data, future work could integrate remote sensing imagery, enterprise-level carbon emission records, and natural language processing (NLP) techniques to depict land-use behavior, green innovation activity, and policy enforcement intensity more dynamically and precisely at the micro level.

Third, given the diverse development stages and functional roles of Chinese cities, subsequent research could incorporate heterogeneity in urbanization levels and examine the distinct characteristics of tourism-oriented cities. Such efforts would provide more scientific and theoretical support and empirical evidence for achieving coordinated regional emission reduction and the modernization of land resource governance capacity.