Abstract
This study examines the impact of China’s Low-Carbon City Pilot (LCCP) policy on urban land use efficiency (LUE), a pivotal yet underexplored aspect of sustainable urban development within the framework of the United Nations Sustainable Development Goals (SDGs). Using panel data from 281 prefecture-level cities between 2005 and 2021, the study employs a staggered Difference-in-Differences (DID) approach to estimate the policy’s effect. The results of the study showed that LCCP could significantly increase LUE, with results remaining robust across various alternative estimation strategies, including Propensity Score Matching Difference-in-Differences (PSM-DID), Difference-in-Differences with Multiple Time Periods (CSDID), and synthetic control (SC) methods. Mechanism analysis shows that these improvements may be driven through two channels: the advancement of green innovation and the upgrading of industrial structures. Additionally, according to the results of the heterogeneity analysis, policy’s effect is more significant in regions with higher marketization levels, stronger green finance capacity, and elevated administrative ranks. By extending the analysis of land governance beyond traditional environmental indicators and identifying the specific mechanisms and contextual factors, this study not only contributes new empirical insights into the spatial dynamics of China’s dual-carbon transition but also clarifies previously mixed findings by uncovering the mechanisms and contextual factors through which the LCCP enhances LUE.
1 Introduction
As urbanization continues to advance, Chinese cities are increasingly confronted with severe resource and environmental constraints. The long-standing extensive land use model has intensified ecological pressures and impeded the pursuit of high-quality urban development and transformation. In response, promoting green and low-carbon transitions has become essential to align with the demands of China’s new development stage (Bai et al., 2014). Within the global context of sustainability—particularly the United Nations Sustainable Development Goals (SDGs)—enhancing land use efficiency (LUE) has emerged as a critical benchmark for assessing urban sustainability. LUE is not only tied to effective resource allocation and reduced carbon emissions but also serves as an indicator of governmental modernization in governance and developmental outlook (Cao et al., 2022). Consequently, revitalizing existing land resources and optimizing land-use structures have become vital policy approaches for facilitating green urban transformation and encouraging compact, sustainable development.
In 2010, China’s National Development and Reform Commission launched the Low-Carbon City Pilot (LCCP) program, which initially involved 36 cities across five provinces and autonomous regions. The program was expanded in 2012 and 2017. Aimed at controlling carbon emission intensity, the LCCP encourages local governments to adopt low-carbon strategies for development, including improvements in energy structures, industrial upgrading, the promotion of green buildings, and more efficient land use. In addition to reducing carbon emissions, the initiative also strives to improve the efficiency of green factor allocation by granting local authorities’ greater autonomy and policy flexibility. As a result, the LCCP offers a quasi-natural experimental framework for evaluating how local governance and institutional arrangements influence LUE (Yu and Zhang, 2021).
While previous studies have extensively explored the environmental and economic impacts of LCCP—including carbon reduction performance, green technological innovation, energy structure optimization, and industrial upgrading—empirical research on the program’s influence on LUE remains scarce and inconclusive (Pan et al., 2022; Zeng et al., 2023). From a policy design standpoint, the LCCP promotes land-saving development through spatial restructuring, increased land-use intensity, and constraints on inefficient land consumption. However, existing findings on its actual impact on land use efficiency are mixed: some studies highlight positive outcomes through improved spatial planning and resource allocation (Liu, 2024), while others point to potential inefficiencies due to rigid land-use quotas or local implementation distortions (Liu J. et al., 2022). As such, whether the policy can substantially reshape local land governance and foster more efficient, compact urban development remains an open empirical question.
To address this research gap, this study draws on panel data from 281 prefecture-level cities between 2005 and 2021, leveraging the LCCP as a quasi-natural experiment to rigorously evaluate its effect on LUE. Theoretically, it identifies two key transmission channels—green innovation and industrial structure upgrading—through which low-carbon policies enhance LUE by fostering technological progress and optimizing spatial resource allocation. Empirically, a staggered Difference-in-Differences (DID) framework is employed, supplemented with robustness checks including PSM-DID, Callaway–Sant’Anna DID, and synthetic control methods. A dual-channel mediation model is developed to validate the mechanisms, while heterogeneity analyses further explore how institutional and market conditions shape policy effectiveness. This study makes the following marginal contributions. First, this study shifts the analytical focus to the impact of low-carbon policies on land use efficiency, thereby expanding the evaluative paradigm of environmental policy and offering new empirical insights into the logic of urban land governance amid China’s green transition. Second, it addresses the limitations of prior literature, which often provided one-sided or fragmented assessments, by systematically identifying the underlying mechanisms—green innovation and industrial structure—as mediating channels (Liu X. et al., 2022; Liu et al., 2024). A dual-path mediation framework is constructed, supplemented with robustness checks and extensive heterogeneity analyses across institutional and market contexts, allowing for a more nuanced understanding of how and under what conditions the LCCP affects LUE. Compared to earlier studies, the empirical strategy adopted in this paper is more rigorous, integrating multiple identification techniques to resolve conflicting conclusions and strengthen causal inference (Liu J. et al., 2022). Third, at the policy level, the findings provide empirical support for advancing more intensive and sustainable land use in alignment with green development goals, while offering region-specific governance recommendations.
2 Literature review and research hypothesis
2.1 Assessing land use efficiency
LUE is a multidimensional concept, and the choice of evaluation indicators often varies based on specific research objectives and analytical perspectives (Ma et al., 2023). In alignment with the SDGs, several studies have highlighted the ratio of urban construction land consumption to population growth as a critical metric. This indicator is commonly employed to assess the balance between urban expansion and population-carrying capacity, particularly in cross-national comparisons of urban development (Estoque et al., 2021).
From the perspective of economic efficiency, some studies have focused on economic output per unit of land, such as GDP or industrial value added, to measure the economic intensity of land use. These indicators aim to assess how effectively land resources are used for economic production, although they may overlook ecological externalities (Herzig et al., 2018). With growing environmental concerns, an increasing body of research has incorporated negative environmental externalities into LUE assessments. These approaches consider not only desirable outputs but also undesirable ones, such as pollutant and carbon emissions, thus accounting for the ecological impacts of land-use activities (Yu et al., 2019). This framework is widely applied in studies on green and ecological efficiency, making it particularly relevant for evaluating land-use sustainability in the context of China’s “dual carbon” goals and ecological civilization agenda.
In terms of methodology, Ma et al. (2023) reviewed 208 relevant studies published between 2005 and 2022 and found that Data Envelopment Analysis (DEA) and its extended models are the most widely adopted methods. The non-parametric nature of DEA allows it to handle multiple inputs, desirable outputs, and undesirable outputs simultaneously, making it ideal for constructing an integrated efficiency framework based on the “input-output–undesirable output” structure (Koroso et al., 2021). Furthermore, Stochastic Frontier Analysis (SFA) is frequently used to estimate efficiency frontiers because it accounts for both random noise and inefficiency. Although SFA is computationally more complex, it offers the advantage of modeling statistical error terms more effectively (Liu et al., 2020; Ferreira and Féres, 2020).
At the macro level, simple ratio indicators, such as land consumption per unit of GDP or output density, are commonly used for interregional comparisons due to their simplicity and ease of comparison. However, these indicators are sensitive to the selection of the numerator and denominator and often fail to systematically characterize input structures (Herzig et al., 2018). Additionally, some studies use composite index methods—such as the entropy method, analytic hierarchy process, and TOPSIS—to construct weighted indicator systems that provide overall efficiency scores. These methods are particularly useful when detailed input-output data are unavailable because they allow for a combination of qualitative and quantitative assessments. However, they often involve a higher degree of subjectivity when assigning weights (Zhang et al., 2020).
2.2 Low carbon policies and land use efficiency
As a major institutional innovation in China’s efforts toward green development and ecological civilization, the LCCP program is designed to improve the efficiency of resource allocation, especially in energy and land use (Liu X. et al., 2022). In the realm of land governance, the LCCP emphasizes spatial control, land-use regulation, and urban development model transformation. It seeks to shift cities away from extensive, land-consuming expansion toward more compact and efficient urban forms. On the one hand, pilot cities are required to impose stricter construction land quotas and increase land-use intensity, thus promoting higher output from limited land resources (Lan et al., 2023). Meanwhile, the policy promotes the reuse of underutilized land and the redevelopment of inefficient areas to enhance overall land use efficiency (Niu et al., 2023).
In addition, key performance indicators of the LCCP—such as carbon emission intensity, resource consumption, and building density—are integrated into official evaluation systems. These indicators impose environmental constraints on land development, effectively shifting urban land-use governance from an expansion-driven model to an efficiency-driven paradigm (Wang et al., 2023). In summary, the LCCP places a dual focus on “low-carbon and intensive” land governance by combining regulatory pressure, incentive mechanisms, and updated governance principles. This approach is expected to improve both the economic output density and ecological efficiency of urban land use. Building on this theoretical foundation, the study puts forward the following hypothesis:
H1LCCP significantly enhances the urban land use efficiency.The LCCP can enhance LUE not only through direct regulation but also indirectly by transforming the urban industrial structure. In particular, the implementation of carbon intensity control targets and energy consumption caps under the LCCP framework incentivizes local authorities to guide industries away from energy-intensive and highly polluting sectors, and toward low-carbon, high-tech, and high value-added industries (Zhao F. et al., 2023; Li et al., 2025). This green-oriented industrial shift supports a transition from a land-extensive, secondary-sector-dominated structure to one that emphasizes compact, service-oriented, and manufacturing-intensive agglomerations, thereby boosting land productivity and spatial carrying capacity (Liu J. et al., 2022).A more optimized industrial structure also reduces reliance on newly developed land and alleviates pressure for spatial expansion by promoting denser, higher-output land use within existing urban footprints (Chang et al., 2023). Furthermore, numerous LCCP cities align industrial spatial planning with urban spatial consolidation strategies—such as “production–city integration” and “compact urban form”—to strengthen multifunctional land use and facilitate brownfield redevelopment (Pan and Song, 2017; Qiao et al., 2025). These dynamics indicate that the LCCP indirectly fosters more intensive and efficient land use by advancing industrial restructuring. Based on this transmission mechanism, the following hypothesis is proposed:
H2LCCP improves urban land use efficiency by promoting industrial structure upgrading.In addition to direct regulatory constraints and industrial structure upgrading, the LCCP enhances LUE by stimulating green innovation. On one side, it promotes low-carbon development through support for science and technology policies, facilitating the development, deployment, and spread of green technologies (Pan et al., 2022; Zou et al., 2022). The strategy, centered on innovation, improves resource allocation efficiency and pushes the technological frontiers of green production.Enhanced green innovation also, serves as a key transmission channel for improving LUE. First, it increases land productivity by improving energy efficiency and promoting the emergence of new green industries that offer higher output per unit of land and more diversified functional uses (Tang et al., 2018). Technologies such as green buildings, clean energy systems, and smart infrastructure reduce the amount of land required to support industrial and service functions, alleviating spatial constraints without expanding the land supply. Second, green innovation facilitates technological substitution, reducing the reliance of traditional high-emission industries on extensive land use. This enables the release and repurposing of idle or underutilized land resources, thereby improving their reallocation efficiency (Sun and Chen, 2023). Moreover, the widespread adoption of digital and intelligent urban governance systems strengthens land-use planning, spatial coordination, and long-term improvements in LUE (Liu et al., 2025). Taken together, the LCCP fosters a shift in land governance from a resource- and expansion-driven approach to a technology- and efficiency-oriented model through green innovation. Based on this mechanism, the following hypothesis is proposed:
H3The LCCP enhances urban LUE by strengthening green innovation capacity.Existing studies have shown that the institutional environment and capacity for financial resource allocation are key moderating factors in the success of green development policies. Notably, the degree of marketization—which indicates how extensively local governments use market-based mechanisms for efficient resource distribution—significantly affects policy outcomes. A higher level of marketization tends to minimize distortions caused by excessive government intervention, improve resource allocation efficiency, and enhance institutional adaptability. Taken together, these elements heighten the capacity of green policies to refine resource allocation and intensify land utilization (Cui and Cao, 2023).In addition, the advancement of green finance acts as a crucial financial support mechanism for achieving green transformation goals. Within the realm of land use, instruments such as green credit, green bonds, and ESG evaluations help channel capital toward land-efficient, high-performance, and sustainable development projects, thereby strengthening the positive impact of low-carbon policies on LUE (Zhao X. et al., 2023; Lee and He, 2024). Consequently, both the institutional environment and the infrastructure supporting green finance influence not only how effectively low-carbon policies are implemented but also how significantly they contribute to enhancing LUE. Based on this understanding, the following hypothesis is proposed:
H4The effect of LCCP on LUE varies with the degree of marketization and the stage of financial development.Despite the growing body of research on the relationship between LCCP and land use efficiency, existing findings remain fragmented and at times contradictory. Such inconsistencies largely stem from differences in research design, particularly in sample coverage (e.g., inclusion of the third batch of pilots), methodological approaches (single-factor versus total-factor efficiency), and spatial scope (eastern versus inland regions). Consequently, it remains unclear whether the observed variations in policy effects reflect true regional heterogeneity or arise from methodological discrepancies. This study reconciles fragmented findings by adopting a dynamic multi-period DID design, refining LUE measurement to include resource constraints, and uncovering both direct and chained mediation pathways linking LCCP to land use efficiency.To clarify the logic of moderating mechanisms, Figure 1 presents the extended theoretical framework of this study. The LCCP improves land use efficiency (LUE) through two primary transmission channels: industrial structure optimization and green innovation. The policy encourages the exit of low-efficiency, high-emission industries and supports the development of high-value-added and technology-intensive sectors, while simultaneously fostering innovation in green R&D and patent transformation. Together, these mechanisms form a dual pathway through which LCCP drives more intensive and sustainable land use.Moreover, the strength of these effects depends on the institutional and financial environment. A higher level of market readiness (measured by the marketization index) enhances the responsiveness of local governments and firms to policy incentives. Greater financial depth (proxied by green finance) facilitates green innovation by reducing financing constraints for low-carbon technologies, consistent with the Financial Deepening Theory (Townsend and Ueda, 2006). Meanwhile, stronger institutional capacity, reflected by the administrative hierarchy, improves the ability of local governments to implement policy directives effectively, in line with Resource Dependency Theory (Hillman et al., 2009). Finally, regional heterogeneity in resource endowments and industrial bases may further shape the extent to which LCCP translates into land use efficiency gains.
FIGURE 1
3 Variable construction
3.1 Dependent variable: land use efficiency
Drawing on existing research on LUE indicator selection (Yu and Zhang, 2021), this study employs the SBM - Undesirable model to construct indicators based on inputs, outputs, and undesirable outputs, with specific variables detailed in Table 1.
TABLE 1
| Category | Selection basis | Data preprocess |
|---|---|---|
| Inputs | Land | Land for construction area |
| Labor | Total number of employees | |
| Capital | Fixed-asset investment | |
| Resource | Industrial power consumption | |
| Desirable outputs | Economy | GDP per unit of land |
| Society | Disposable income per capita | |
| Ecology | Urban greening ratio | |
| Undesirable outputs | Environment Pollution | Industrial wastewater emissions |
| SO2 emissions |
Selection of land use efficiency indicators.
The data are sourced from the China City Statistical Yearbook, China Land and Resources Statistical Yearbook, China Statistical Yearbook of Science and Technology, China Energy Statistical Yearbook, and various prefecture-level local statistical yearbooks.
Charnes et al. (1978) developed Data Envelopment Analysis (DEA) as a technique to evaluate the relative efficiency of decision-making units operating with multiple inputs and outputs. However, traditional DEA models face limitations in accounting for undesirable outputs and slack variables. To overcome these issues, Tone (2002) developed the SBM model, which accommodates undesirable outputs, including its super-efficiency extension. This study adopts the super-efficiency SBM-Undesirable model, assuming constant returns to scale, to measure land use efficiency. The specific model is shown in Equations 1, 2:
Suppose there are decision making units (DMU) and number of inputs, desirable outputs and undesirable outputs for each DMU is respectively. We can have following calculation:
Where is value of LUE. denote input, desirable output, and undesirable output at t period for each respectively. is slack variable of input, desirable output, undesirable output respectively. is weighting vector.
3.2 Core independent variable: low-carbon city pilot
The main explanatory variable is LCCP participation, reflecting the policy’s influence on LUE. Launched in 2010, the LCCP sought to chart a localized path toward green, low-carbon development by lowering carbon-emission intensity, reshaping the energy mix, advancing green-building standards, and fostering industrial transformation. The initial cohort encompassed 36 cities across five provinces, with subsequent rounds of pilots added in 2012 and 2017. Using the official lists released by the NDRC, a binary indicator is constructed: the variable equals 1 for city in year if the city had joined any LCCP batch on or before year , and 0 otherwise. This coding yields a staggered-treatment framework well suited for policy-impact evaluation.
3.3 Mediating variables
To investigate how the LCCP policy affects LUE, this study empirically examines green innovation as a mediating factor. Green innovation capability is proxied by the count of green patent applications. Following the methodology of Amore and Bennedsen (2016), patent data were retrieved from the China National Intellectual Property Administration. Patent entries related to green technologies were identified using the IPC Green Inventory Classification, and only invention patents and utility models were retained to better reflect green innovation output. To minimize the effect of publication delays, data from the policy implementation year and the following year were excluded when constructing the green patent variable. Additionally, industrial upgrading was incorporated as a mediating factor, measured by the proportion of the secondary and tertiary sectors in total GDP, using relevant data obtained from the China City Statistical Yearbook.
3.4 Control variables
Following existing literature on the determinants of LUE (Liu et al., 2021), this study includes the following control variables: (1) the logarithm of GDP per capita (lnGDP), capturing a city’s level of economic development; (2) municipal construction investment (MCI), reflecting the intensity of public infrastructure investment; (3) fiscal surplus (FIS), measured as the difference between fiscal revenue and expenditure, indicating the degree of fiscal capacity (FIS); and (4) urban registered population, representing population density (POPU). All data were obtained from the China City Statistical Yearbook.
3.5 Descriptive statistics of the data
This study utilizes panel data spanning 2005 to 2021 from 281 prefecture-level cities across China. Before proceeding with the empirical analysis, a preliminary descriptive statistical examination of all selected variables was carried out to establish a foundational understanding and to guide the specification and estimation of the econometric models. Table 2 summarizes the descriptive statistics for the primary variables.
TABLE 2
| Variable | Mean | Max | Min | Skewness | Kurtosis | Sd |
|---|---|---|---|---|---|---|
| LUE | 0.685 | 1 | 0.249 | −0.754 | 26.54 | 0.0370 |
| LCCP | 0.103 | 1 | 0 | 2.612 | 7.824 | 0.304 |
| lnGDP | 5.668 | 642.2 | −7.624 | 54.58 | 3,479 | 9.972 |
| POPU | 4.216 | 15.56 | 0.172 | 0.905 | 3.582 | 2.505 |
| FIS | −1.311 | 6.078 | −19.20 | −2.556 | 20.43 | 1.247 |
| MCI | 11.668 | 16.468 | 5.736 | 0.0503 | 3.037 | 1.568 |
| IS | 0.931 | 1 | 0.395 | −2.249 | 11.72 | 0.0670 |
| GI | 355.7 | 10,679 | 0 | 13.23 | 265.9 | 2,123 |
| MI | 10.63 | 20.23 | 2.717 | 0.084 | 2.485 | 3.035 |
| GF | 0.303 | 0.650 | 0.0540 | −0.305 | 2.706 | 0.104 |
Descriptive statistics of variables.
4 Spatial-temporal dynamics of core variables
4.1 Land use efficiency
To further illustrate the dynamic evolution of LUE, this study maps its spatial distribution across China’s prefecture-level cities for the years 2005, 2010, 2015, and 2021 (Figure 2), using a consistent classification scheme and color palette that divides LUE into eight levels for easier cross-sectional comparison.1
FIGURE 2
From a temporal perspective, LUE in China has exhibited a consistent upward trajectory, with the national average increasing from 0.586 in 2005 to 0.724 in 2021. This improvement is likely associated with the progress of new-type urbanization, the optimization of land-use structures, and the increasing focus on green development principles. Additionally, the rise in LUE may be driven by enhanced regulatory measures aimed at promoting land-saving and more intensive land use at the policy level.
In terms of spatial patterns, LUE followed a different dynamic compared to the general economic disparity, where development is more pronounced in the east and south than in the west and north. In 2005, significant spatial disparities were observed, with high efficiency concentrated in coastal regions and relatively low performance in central areas. However, by 2015, most prefecture-level cities across the country had shifted to a medium-efficiency category, displaying a trend toward moderate-to-high efficiency and narrowing regional differences. This change reflects the initial success of regional coordination policies and the gradual intensification of land-use. Yet, by 2021, signs of spatial polarization reappeared, with clusters of high-efficiency cities and persistent low-efficiency areas, indicating that unbalanced regional progress in LUE remains a concern that requires further policy focus.
4.2 Low carbon city pilots
Figure 3 illustrates the spatial distribution of the LCCP policy across the study sample. According to the implementation batches classified by the NDRC, the first batch of LCCP cities was launched in 2010, followed by an additional 28 cities in 2012, and a third batch in 2017, further expanding the pilot coverage. In the figure, different-colored circles represent pilot cities initiated in different years, while red dots indicate non-pilot cities that were not included in any batch. Although data from western regions are somewhat limited, pilot cities are clearly concentrated in eastern and central China, especially in the middle and lower Yangtze River basin. In contrast, coverage is relatively sparse in the central-western and remote regions. Due to data availability constraints, some western cities were excluded from the sample, and the map reflects only the valid prefecture-level data used in this study. The spatial layout shown in the figure highlights the regional implementation pattern of the policy and offers a geographic basis for later assessments of its effectiveness and heterogeneity.
FIGURE 3
5 Empirical analysis
5.1 Baseline regression
To evaluate the effect of the LCCP policy on urban land use efficiency, this study applies a multi-period difference-in-differences (DID) method to build the baseline regression model. This approach accommodates the varying timing of policy adoption across cities, offering a more accurate reflection of its dynamic implementation over time. Additionally, it helps mitigate the estimation bias that can arise from traditional DID models, which typically assume uniform treatment timing across treated units. The specific regression model is as follows:
Where represents the level of land use efficiency for city at year ; represents a dummy variable for whether the city is a pilot of LCCP (1 for pilot cities from the policy year onwards, 0 otherwise); represents about control variables, including logarized gdp, industrial structure, population density, and fiscal adequacy; and represent the prefecture-level city fixed effect and year fixed effect respectively to control for unobservable time-invariant urban characteristics and common shocks at the national level; is random error term. Moreover, during the time scope we use, there is another policy called pilot zones for eco-environment conservation. To control the influence of similar policy, a dummy variable is also added. The baseline regression result is in Table 3:
TABLE 3
| Variables | (1) | (2) | (3) |
|---|---|---|---|
| LUE | LUE | LUE | |
| LCCP | 0.0092*** | 0.0083*** | 0.0071** |
| (3.38) | (3.99) | (2.02) | |
| lnGDP | | −0.0002 | −0.0002 |
| | (-1.49) | (-1.24) | |
| MCI | | −0.0116*** | −0.0143*** |
| | (-3.30) | (-2.94) | |
| POPU | | 0.0031 | 0.0028 |
| | (1.21) | (1.12) | |
| FIS | | −0.0003 | 0.0001 |
| | (-0.21) | (0.04) | |
| ECP | | | 0.0040* |
| | | (1.72) | |
| Constant | 0.6645*** | 0.7088*** | 0.6967*** |
| (1,241.16) | (32.78) | (32.32) | |
| Observations | 4,777 | 4,777 | 4,777 |
| R-squared | 0.107 | 0.197 | 0.199 |
| City fixed | Yes | Yes | Yes |
| Year fixed | Yes | Yes | Yes |
The role of low-carbon city pilots in improving land use efficiency.
Robust t-statistics in parentheses: ***p < 0.01, **p < 0.05, *p < 0.1.
Table 3 presents the baseline regression results on the impact of the LCCP policy on LUE. Column (1) includes only city and year-fixed effects. The findings reveal that the LCCP has a statistically significant positive effect on LUE at the 1% level, suggesting that the pilot policy improves the output efficiency of urban land. Column (2) includes additional controls for lnGDP, MCI, POPU, and FIS. The estimated effect of the LCCP remains robust, although the coefficient is slightly reduced; it remains significant at the 5% level. Column (3) adds further policy controls and the effect of the LCCP remains significant. Moreover, the significantly positive coefficient for the Ecological Civilization Pilot suggests that these demonstration zones also play a role in enhancing land use efficiency.
5.2 Parallel trend test
The parallel trends assumption is a crucial identification condition for the multi-period DID model. To verify its validity, this study adopts an event study framework to examine the policy’s dynamic impacts before and after implementation. Figure 4 displays the estimated coefficients along with their 95% confidence intervals over the relative time periods surrounding the launch of the LCCP policy. Prior to the policy’s implementation, the coefficients were statistically insignificant and hovered around zero, suggesting that both the treatment and control groups followed comparable trends in land use efficiency, thus confirming the parallel trends assumption. After the policy was enacted, a consistent upward trend in the coefficients was observed, and statistical significance was attained, implying that improvements in LUE were brought about by the LCCP policy, with effects that appeared to be sustained over time. Parallel trend tests with different time windows are shown in the (Supplementary Figure A1–A3).
FIGURE 4
Existing research has identified two major limitations of the traditional event study approach when testing the parallel trend assumption. First, limited statistical power may lead to a failure in detecting significant differences even when the assumption is violated—this issue is especially pronounced in studies with small sample sizes. Second, using the pre-trend test as a prerequisite for causal inference can result in selective inference problems, which may undermine the accuracy of treatment effect estimates (Roth et al., 2023). To address these concerns and improve the reliability of the parallel trends test, this study applies the robust diagnostic method developed by Borusyak et al. (2024), which separates testing of pre-treatment trends from estimation of post-treatment effects. This approach helps avoid bias that can arise from conditioning causal inference on pre-treatment test outcomes.
By estimating treatment effects within the pre-policy time window and presenting the dynamic effects for each pre-treatment period separately, more robust evidence is provided to support the parallel trend assumption. As illustrated in Figure 5, the treatment and control groups exhibit similar trends in the dependent variable during the 5 years leading up to the policy implementation, providing a robust basis for identifying effects using the multi-period DID model.
FIGURE 5
To further reduce the dependence on the strict parallel trend assumption inherent in conventional DID estimation, this study employs the Honest DID approach developed by Rambachan and Roth (2023) to assess the sensitivity of treatment effects to potential deviations from pre-treatment trends. This methodology is grounded in the principle of partial identification, and constructs robust confidence intervals for policy impacts without requiring exact parallel trends before treatment. In particular, the Honest DID framework introduces a bounding parameter, Mbar, which sets the maximum allowable deviation from parallel trends. By placing a restriction on pre-treatment trend deviations using Mbar, the method defines an identification region for post-treatment outcomes. If the robust confidence interval under a specific Mbar excludes zero, the treatment effect is considered robust to violations of the parallel trend assumption.
In the analysis, the approach was followed by setting Mbar to 0.5 times the standard error, as suggested by Rambachan and Roth (2023), and sensitivity tests were carried out on the estimated treatment effects for the policy year and the 3 years that followed. Figure 6 illustrates the changes in robust confidence intervals under various values of Mbar. It is shown that for the implementation year (top-left), the first year after (top-right), and the second year after (bottom-left), the estimated intervals exclude zero across multiple Mbar values, indicating that the results are not substantially affected by reasonable deviations from the assumption of parallel trends. While the confidence interval for the third-year post-implementation (bottom-right) includes zero, the estimate is still near the lower limit, and the overall direction of the treatment effect remains consistent. Thus, further support for the robustness of the policy’s positive impact on LUE is provided by the sensitivity analysis.
FIGURE 6
5.3 Robustness check
5.3.1 Placebo test
To further verify the credibility of the DID estimates, a placebo test was carried out by randomly assigning treatment groups to cities, while retaining the actual timing of policy implementation. This process was repeated 500 times, and the resulting coefficients from the placebo regressions were collected. Figure 7 displays the distribution and kernel density of these placebo estimates. The coefficients are concentrated within the interval [-0.005, 0.005], centered around zero, and approximately normally distributed. Furthermore, the placebo estimates are clearly distinct from the original DID estimate of 0.0063. These findings indicate that the observed treatment effect is unlikely to result from unobserved confounding factors, thereby supporting the robustness of the original results.
FIGURE 7
5.3.2 PSM - DID
In the former analysis, cities were divided into treatment and control groups according to their involvement in the LCCP program, and a multi-period DID model was employed to estimate the policy’s effect. The findings indicates that the implementation of the LCCP strategy significantly enhances LUE. However, because the selection of pilot cities was likely non-random and influenced by pre-existing economic and technological conditions, treated cities may systematically differ from their non-treated counterparts—particularly in terms of infrastructure, economic development, FIS, or innovation potential—potentially increasing their likelihood of being selected for the pilot program.
To address the potential selection bias arising from non-random assignment of cities to the LCC pilot program, this study employs the PSM-DID approach. A matched control group is constructed based on cities with similar pre-treatment characteristics, after which the treatment effects are re-estimated using the staggered DID framework. As illustrated in Figure 8, after matching, the standardized biases of covariates were significantly reduced, demonstrating better balance and comparability between the treatment and control groups, which enhances the credibility of causal inference. Table 4 presents the PSM-DID estimation results across five matching algorithms: nearest-neighbor matching, k-nearest-neighbor matching, caliper matching, radius matching, and kernel matching, as reported in Columns (1) through (5), respectively. The consistently significant positive coefficients across all matching methods confirm that the beneficial impact of the LCCP on LUE remains robust even after adjusting for sample selection bias.
FIGURE 8
TABLE 4
| Variables | LUE | ||||
|---|---|---|---|---|---|
| (1) | (2) | (3) | (4) | (5) | |
| LCCP | 0.0052** | 0.0051** | 0.0054*** | 0.0044* | 0.0038* |
| (2.10) | (2.03) | (2.77) | (1.84) | (1.89) | |
| Constant | 0.6343*** | 0.5836*** | 0.5962*** | 0.6084*** | 0.6013*** |
| (79.19) | (23.34) | (19.91) | (49.29) | (40.85) | |
| Control | Yes | Yes | Yes | Yes | Yes |
| Observations | 3,372 | 1,567 | 2,462 | 3,170 | 2,166 |
| R-sq | 0.1092 | 0.1062 | 0.1324 | 0.1173 | 0.1900 |
| City fixed | Yes | Yes | Yes | Yes | Yes |
| Year fixed | Yes | Yes | Yes | Yes | Yes |
PSM - DID regression results.
Robust t-statistics in parentheses: ***p < 0.01, **p < 0.05, *p < 0.1.
5.3.3 Synthetic DID
In causal inference practice, the DID approach relies heavily on the assumption of parallel trends. However, if systematic differences persist between the treatment and control groups before the policy takes effect, visually similar pre-treatment trends may still fail to guarantee truly parallel paths over time, potentially leading to identification bias. Additionally, the selection of control groups often involves subjective judgment, which can affect the robustness of the estimates. To address these concerns, recent studies have increasingly employed the Synthetic Control (SC) method to enhance the construction of counterfactuals. Nonetheless, SC also has limitations, especially when the treated unit differs substantially from the donor pool, making it challenging to create a well-fitted SC.
To overcome the identification limitations inherent in both the DID and synthetic control methods, Arkhangelsky et al. (2021) introduced the synthetic difference-in-differences (SDID) estimator, which integrates the advantages of the two approaches. SDID mitigates reliance on the strict parallel trends assumption by integrating pre-treatment trend matching with weighted SC construction. Furthermore, the proposed method incorporates unit fixed effects, which is particularly well-suited for settings with heterogeneous treated units. Accordingly, the SDID method was adopted in this study as a robustness check to complement the main regression results.
The SDID estimator and its extended dynamic form were applied to evaluate the impact of the LCCP on improvements in LUE. For the static average treatment effect, a coefficient of 0.00546 was estimated, which was found to be statistically significant at the 5% level and aligned with earlier findings. Regarding dynamic effects, a trend comparison between treated units and their synthetic counterparts across different policy implementation years is illustrated in Figure 9. The timing of LCCP implementation (2010, 2012, 2017) is indicated by vertical dashed lines; the treatment group is represented by a red line, the SC group by a blue line, and the distribution of synthetic weights by green shaded areas. The figure shows that before the policy was implemented, the treatment group and the synthetic control group followed comparable trends, supporting the parallel trends assumption. After the policy was introduced, the treatment group displayed a markedly higher outcome trajectory than the synthetic control, suggesting a pronounced positive policy impact. Additionally, the concentration of lambda weights in the pre-treatment period indicated that the SC group was credibly constructed, thereby reinforcing the robustness and reliability of the estimated treatment effects.
FIGURE 9
5.3.4 Additional robustness tests
To further confirm the robustness of the findings, this study employed the Callaway and Sant’Anna (2021) Difference-in-Differences with Multiple Time Periods (CSDID) estimator. This method accommodates staggered treatment adoption across groups and enables dynamic tracking of treatment effects, effectively addressing issues such as negative weighting and treatment effect heterogeneity, which are common in traditional two-way fixed-effects models. As shown in Column (1) of Table 5, the coefficient for the LCCP policy is significantly positive at the 5% level, aligning with the main regression results. This finding reinforces the conclusion that the LCCP policy effectively enhances land use efficiency, confirming the robustness of the core results.
TABLE 5
| Variables | LUE | LUE2 | LUE3 | LUE | LUE |
|---|---|---|---|---|---|
| (1) | (2) | (3) | (4) | (5) | |
| LCCP | 0.00687** | 0.00478** | 0.00692*** | 0.00510** | 0.00682** |
| (2.06) | (2.01) | (2.67) | (2.15) | (2.35) | |
| Constant | | 0.8329*** | 0.8291*** | 0.8031*** | 0.8458*** |
| | (5.09) | (4.87) | (3.44) | (3.05) | |
| Control | Yes | Yes | Yes | Yes | Yes |
| Observations | 4,777 | 4,777 | 4,777 | 4,777 | 4,777 |
| R-sq | | 0.5431 | 0.5632 | 0.6771 | 0.6977 |
| City fixed | Yes | Yes | Yes | Yes | Yes |
| Year fixed | Yes | Yes | Yes | Yes | Yes |
| Combined fixed | No | No | No | Yes | No |
Additional robustness tests.
Robust t-statistics in parentheses: ***p < 0.01, **p < 0.05, *p < 0.1.
In this study, land use efficiency (LUE), the dependent variable, is first assessed using the super-efficiency SBM-Undesirable model, assuming constant returns to scale. To test the robustness of the results to alternative efficiency measurement methods, LUE is re-estimated using a different DEA-based SBM-Undesirable model, producing an alternative outcome variable denoted as LUE2. The regressions are then re-run using LUE2 as the dependent variable. As shown in Column (2) of Table 5, the positive effect of the LCCP policy on LUE remains statistically significant, confirming that the main findings are robust to changes in the definition of the outcome variable. To more comprehensively capture non-industrial resource constraints, we further re-estimate LUE by adding total water consumption as an additional input variable in the super-SBM-Undesirable model. The inclusion of water consumption reflects broader resource-use efficiency beyond industrial energy inputs. As reported in Column (3) of Table 5, the estimated coefficient of the LCCP policy remains significantly positive at the 1% level, indicating that the core conclusion is robust to the inclusion of water-resource utilization.
While former regressions include city and year fixed effects, there is still a risk that unobserved time-varying factors at the provincial level may distort the estimates. To mitigate this concern, the baseline model is extended by adding province-by-year fixed effects to capture potential regional heterogeneity. The revised regression results in Column (4) of Table 5 indicate that the LCCP policy coefficient remains significantly positive, suggesting that the baseline conclusions hold even after accounting for province-level time-varying factors.
Lastly, to evaluate the impact of potential outliers, all explanatory variables—except for the LCCP indicator—are winsorized at the 1st and 99th percentiles, and the model is re-estimated based on this adjusted sample. As reported in Column (5) of Table 5, the coefficient of LCCP remains significantly positive, providing additional evidence that the main findings are robust to the presence of extreme values.
6 Mechanism tests
6.1 Mediation analysis
The mediation analysis employs the causal mediation framework proposed by Imai et al. (2010), which is based on the counterfactual methodology. This method offers significant advantages because it separates the definition of mediation effects from the model specification, thereby allowing for nonparametric identification under relatively weak assumptions. Additionally, it includes a built-in sensitivity analysis module that systematically assesses the robustness of the identifying assumptions, providing a flexible and reliable approach to estimating the mediation effect.
Table 6 presents the results of the mediation analysis. Panel A shows that the LCCP policy significantly promotes industrial structure upgrading, which subsequently has a positive and statistically significant impact on land use efficiency. After accounting for the mediator, the direct effect of LCCP remains statistically significant, implying that the industrial structure serves as a partial mediator in the relationship between policy and LUE. Additionally, the left panel of Figure 10 displays the sensitivity curve for the Average Causal Mediation Effect (ACME), illustrating how it responds to different levels of deviation from the sequential ignorability assumption. The confidence intervals remain consistently above zero across the sensitivity parameter range of −0.5 and 0.5, indicating that the mediation effect is relatively robust to potential deviations from the identifying assumption.
TABLE 6
| Variables | LUE | IS | LUE |
|---|---|---|---|
| Panel A: The mediating role of the industrial structure in enhancing LUE | |||
| LCCP | 0.0083*** | 0.0355*** | 0.0077*** |
| (3.99) | (10.94) | (4.28) | |
| IS | | | 0.0298*** |
| | | 3.77 | |
| Control | Yes | Yes | Yes |
| Constant | 0.7088*** | 0.9323*** | 0.6352*** |
| (32.78) | (25.83) | (32.57) | |
| City fixed | Yes | Yes | Yes |
| Year fixed | Yes | Yes | Yes |
| Variables | LUE | GI | LUE |
|---|---|---|---|
| Panel B: The mediating role of green innovation in enhancing LUE | |||
| LCCP | 0.0083*** | 5.2926*** | 0.0058*** |
| (3.99) | (6.82) | (3.17) | |
| GI | | | 2.8343*** |
| | | (4.28) | |
| Control | Yes | Yes | Yes |
| Constant | 0.7088*** | −0.1481*** | 0.6356*** |
| (32.78) | (19.18) | (33.96) | |
| City fixed | Yes | Yes | Yes |
| Year fixed | Yes | Yes | Yes |
| Pathways | Coefficient | Bootstrap SE | P-value |
|---|---|---|---|
| Panel C: The mediating role of green innovation in enhancing LUE | |||
| GI → IS | 5.29 | 6.811 | 0.00 |
| IS → LUE | 0.51 | 0.188 | 0.07 |
| LCCP → GI → IS → LUE | 0.0005 | 0.000 | 0.03 |
| Total effect (LCCP → LUE) | 0.0102 | 0.002 | 0.02 |
Mechanisms through which the LCCP influences LUE.
Robust t-statistics in parentheses: ***p < 0.01, **p < 0.05, *p < 0.1.
FIGURE 10
Panel B demonstrates that the LCCP policy significantly boosts green innovation, which subsequently contributes positively and significantly to improvements in land use efficiency. The direct effect of LCCP remains significant even after accounting for green innovation as a mediator, indicating that green innovation constitutes an additional and meaningful transmission channel through which policy influences LUE. As illustrated in the right panel of Figure 10, the ACME curve shows a slight downward trend; however, the confidence interval remains above zero across the full range of sensitivity parameters from −0.5 to 0.5, thereby reinforcing the robustness of the mediation effect.
Overall, these results indicate that the LCCP policy improves LUE by both advancing the industrial structure and strengthening green innovation capacity.
Beyond the single mediation mechanisms presented above, we further examine whether green innovation and industrial structure upgrading operate as a sequential mediation chain. Specifically, we estimate a serial mediation model (LCCP → GI → IS → LUE) using a bootstrap approach with 1,000 replications clustered at the city level.
Panel C further investigates whether green innovation and industrial structure upgrading jointly operate as a sequential (serial) mediation chain. The results reveal that LCCP first enhances green innovation (β = 5.29, p < 0.01), which subsequently drives industrial upgrading (β = 0.51, p < 0.1), ultimately improving land use efficiency. The estimated chained mediation effect (LCCP → GI → IS → LUE) is positive and statistically significant (β = 0.0005, p = 0.03), while the total indirect impact remains moderate in magnitude. This evidence supports the view that the policy’s influence on land use efficiency occurs not only through direct and single-stage mediation channels but also through an innovation-driven structural transformation process, consistent with the theoretical expectation of cumulative effects between technological advancement and structural optimization.
Taken together, these results provide additional evidence supporting the theoretical expectation that the LCCP promotes land use efficiency via an innovation-structure pathway.
6.2 Heterogeneity analysis
To investigate the heterogeneous effects of the LCCP policy on LUE, subgroup regression analyses were conducted across two dimensions: regional distribution and administrative hierarchy. The corresponding results, which highlight variations in policy impacts across different regions and levels of city administration, are presented in Table 7.
TABLE 7
| Variables | (1) | (2) | (3) | (4) | (5) |
|---|---|---|---|---|---|
| LUE | LUE | LUE | LUE | LUE | |
| LCCP | 0.0114*** | 0.0034 | 0.0096*** | 0.0153*** | 0.0059* |
| (3.12) | (0.87) | (2.77) | (2.85) | (1.68) | |
| Control | Yes | Yes | Yes | Yes | Yes |
| Constant | 0.6335*** | 0.6226*** | 0.5538*** | 0.0563*** | 0.6234 |
| (22.70) | (50.26) | (15.77) | (30.18) | (37.25) | |
| City fixed | Yes | Yes | Yes | Yes | Yes |
| Year fixed | Yes | Yes | Yes | Yes | Yes |
| Observations | 1887 | 1700 | 1,190 | 544 | 4,233 |
Heterogeneous effects of LCCP across regions and administrative levels.
Robust t-statistics are reported in parentheses: ***p < 0.01, **p < 0.05, *p < 0.1. Using the median as the dividing criterion, columns (1), (2), and (3) present the regression results for subgroups in the eastern, central, and northeastern, and western regions, respectively. Column (4) reports estimation for provincial capitals and sub-provincial cities, while column (5) includes results for all remaining prefecture-level cities.
First, using regional classifications, Columns (1) through (3) represent the eastern, central, and northeastern plus western regions, respectively. In the eastern region, the LCCP coefficient is positive and highly significant at the 1% level, suggesting a substantial positive effect of the pilot policy. Conversely, the estimated coefficient for the central and northeastern regions is 0.0034 and statistically insignificant, implying potential implementation bottlenecks or constraints in factor allocation that may hinder effective resource optimization. In contrast, a significantly positive effect is observed in the western region, implying that, it is supported by national policies and driven by untapped development potential, the LCCP policy has also contributed to improvements in LUE in this area.
Second, Columns (4) and (5) categorize cities by administrative level. The findings indicate that the LCCP has the most pronounced impact in sub-provincial and provincial capital cities, with a coefficient of 0.0153 significant at the 1% level. By comparison, the effect in regular prefecture-level cities is weaker, reflected by a smaller coefficient of 0.0059, which is only marginally significant at the 10% level. This indicates that sub-provincial and provincial capital cities—due to their greater administrative resources, stronger policy implementation capacity, and higher fiscal autonomy—are better equipped to enforce the pilot policy and achieve substantial improvements in LUE.
In conclusion, the LCCP exhibits a stronger positive effect on LUE in the eastern and western regions, as well as in sub-provincial cities, whereas its influence appears comparatively weaker in the central and northeastern areas and in standard prefecture-level cities. These findings further confirm the significant moderating roles that regional development conditions and administrative hierarchy play in shaping the policy’s effectiveness.
To further explore the heterogeneous effects of the LCCP policy under varying institutional and financial environments, this study conducts subgroup and interaction-term regressions from two perspectives: market mechanisms and financial support. The objective is to assess whether the policy’s marginal effect on land use efficiency varies notably under different contextual settings. The corresponding results are shown in Table 8.
TABLE 8
| Variables | (1) | (2) | (3) | (4) | (5) | (6) |
|---|---|---|---|---|---|---|
| LUE | LUE | LUE | LUE | LUE | LUE | |
| LCCP | 0.0064** | −0.0004 | 0.0094* | 0.0076*** | 0.0059** | 0.0024 |
| (2.37) | (-0.16) | (1.45) | (2.99) | (1.98) | (0.92) | |
| LCCP × MI | | | 0.0027** | | | |
| | | (2.14) | | | | |
| LCCP × GF | | | | | | 0.0129*** |
| | | | | | (4.21) | |
| Control | Yes | Yes | Yes | Yes | Yes | Yes |
| Constant | 0.6151*** | 0.6644*** | 0.6293*** | | | |
| (49.54) | (61.68) | (60.94) | | | | |
| City fixed | Yes | Yes | Yes | Yes | Yes | Yes |
| Year fixed | Yes | Yes | Yes | Yes | Yes | Yes |
| Observations | 2,388 | 2,389 | 4,777 | 2,388 | 2,389 | 4,777 |
Heterogeneous effects of LCCP in different institutional and financial environments.
Robust t-statistics in parentheses: ***p < 0.01, **p < 0.05, *p < 0.1. Using the sample median as the dividing threshold, columns (1) and (4) presents the regression results for subgroups with high levels of marketization and green finance, respectively, while columns (2) and (5) correspond to subgroups with low levels of marketization and green finance. Columns (3) and (6) display the results for the full sample as a benchmark for comparison.
First, from the perspective of marketization, a city-level marketization index is constructed by adjusting provincial-level indicators from China’s Marketization Index Report to reflect prefecture-level characteristics. The sample is divided into high and low marketization subgroups to estimate the policy’s effects separately. The results indicate that the LCCP policy significantly improves LUE in highly marketized regions, whereas the effect is statistically insignificant—and even negative—in less marketized areas. Moreover, the regression incorporating the interaction term (LCCP × MI) yields a significantly positive coefficient, indicating that greater marketization enhances the policy’s effectiveness. This finding implies that well-functioning market mechanisms improve resource allocation and responsiveness, enabling more efficient and intensive land-use, and thereby amplifying the positive impact of the LCCP policy. Second, to assess the moderating role of green finance, the sample is divided by a green finance index—measured by the share of interest expenses from non-energy-intensive sectors. Subgroup and interaction regressions reveal that LCCP significantly improves LUE in regions with stronger green financial systems, with the LCCP × GF coefficient estimated at 0.0129 and statistically significant. This suggests that green finance enhances the policy’s effectiveness by easing financing constraints for green transformation.
In summary, the institutional environment and financial support are jointly recognized as critical external conditions influencing the effectiveness of the LCCP policy. A high level of marketization provides a solid foundation for efficient policy transmission and resource allocation, while well-developed green financial systems offer the necessary financial backing and incentives for green transformation. Together, these factors enhance the capacity of the policy to promote more intensive and efficient land use, underscoring the importance of contextual heterogeneity in determining policy outcomes.
7 Conclusion
This study offers strong empirical support for the conclusion that the LCCP policy markedly improves urban land use efficiency. By applying a range of quasi-experimental methods—including PSM-DID, SDID, and CSDID—the analysis confirms that the positive impact of the LCCP remains consistent across various identification strategies and after controlling for potential confounders. Mechanism analyses further indicate that improvement in industrial structure and green innovation function as key transmission channels through which policy operates. Moreover, the policy demonstrates stronger effects in China’s eastern and western regions, and in sub-provincial and capital cities, underscoring the importance of regional institutional and financial environments in shaping policy effectiveness. Overall, the findings indicate that the LCCP is not merely a symbolic policy initiative, but a substantive driver of green-oriented spatial governance in China’s transitional development context. Drawing from these findings, the study puts forward the following policy suggestions:
Promote Green-Oriented Land Governance. Integrate land-saving, efficiency, and ecological criteria into performance evaluations and land approval processes to shift focus from expansion to stock optimization. Grant local governments more flexibility in urban renewal and industrial land consolidation to support sustainable redevelopment.
Enhance Fiscal and Property Incentives for Green Innovation. Promote green patent commercialization and improve innovation efficiency. Use targeted tax and credit policies to steer traditional industries toward sustainable, high-value development. Support green industrial clusters to boost land productivity and align land use with industrial upgrading.
Tailor Strategies to Regional Conditions. Eastern cities should strengthen innovation–land coordination and reduce dependence on heavy industry through cross-regional industrial collaboration; central and western cities should receive greater fiscal support and capacity-building to enhance green governance; prefecture-level cities should establish talent exchange mechanisms with sub-provincial cities to improve policy implementation capacity.
Improve Coordination to Support Lower-tier Cities. Institutionalize vertical support via fiscal transfers, talent exchange, and data sharing. Encourage inter-city collaboration networks between sub-provincial and ordinary prefecture-level cities to facilitate policy diffusion and knowledge transfer. Build cross-sectoral and regional platforms to enhance coordination and optimize spatial resource allocation.
Leverage Green Finance and Market Tools to Enhance Policy Impact. Green finance complements government efforts by providing market-based incentives. Policymakers should expand green credit and bonds, improve investment standards, and align land pricing with ecological costs to boost resource efficiency. In regions with higher marketization, the focus should be on refining financial instruments to internalize environmental costs, whereas in low-marketization regions, priority should be given to improving institutional and fiscal capacity to ensure policy implementation.
8 Conclusion
In addition to the general positive effect of the LCCP, several findings warrant further reflection:
The stronger effects observed in the western region indicate that national policy support can partially offset historical underdevelopment, enabling resource-constrained areas to improve LUE through targeted green transition strategies.
The limited or insignificant effects in central China may reflect policy implementation bottlenecks caused by rigid industrial structures, lower market responsiveness, or weaker fiscal incentives, highlighting the challenges of the so-called “middle-zone trap” in green governance.
Striking differences between sub-provincial and ordinary cities reveal a clear policy responsiveness gap driven by administrative rank, highlighting how decentralization and institutional autonomy can amplify or constrain policy effects—a critical factor in reforming hierarchical governance frameworks.
The moderating roles of marketization and green finance further demonstrate that low-carbon land governance is inseparable from broader systemic reforms, with Policies functioning more effectively when embedded in mature market and financial ecosystems.
These insights reinforce the study’s theoretical contribution that LUE is not merely a technical matter of planning or investment but a complex governance outcome shaped by policy design, institutional capacity, and financial instruments.
This study has several limitations. First, due to data constraints, some prefecture-level cities—especially in western China—were excluded, potentially biasing spatial generalizations. Second, although two key mediators were identified, other plausible mechanisms such as digital infrastructure and institutional accountability remain underexplored. Third, the analysis centers on average treatment effects and did not address dynamic spatial spillovers or land market distortions. Moreover, the potential policy externalities of the LCCP, such as compliance cost burdens, short-term crowding-out of small enterprises, and overinvestment in green infrastructure, should also be recognized. These side effects may temporarily hinder local economic vitality, suggesting the need for more balanced and adaptive policy design in future implementations.
Looking ahead, future research could further explore the integration of digital ecology and dual-carbon coordination as emerging governance frontiers. The convergence of digital infrastructure, AI-driven monitoring, and carbon management systems may provide new pathways for enhancing land use efficiency while supporting low-carbon transitions. Combining spatial econometric models with digital governance indicators could help capture the interplay between technological innovation and ecological regulation. Expanding the scope beyond China would also provide valuable comparative insights into green urban transformation in other developing economies.
Statements
Data availability statement
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.
Author contributions
TZ: Conceptualization, Data curation, Formal Analysis, Methodology, Software, Visualization, Writing – original draft. LD: Conceptualization, Data curation, Funding acquisition, Project administration, Supervision, Validation, Writing – review and editing.
Funding
The authors declare that financial support was received for the research and/or publication of this article. This work was supported by Philosophy and Social Science Foundation of China (24FKSB033).
Conflict of interest
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
Generative AI statement
The authors declare that Generative AI was used in the creation of this manuscript. The author(s) verify and take full responsibility for the use of generative AI in the preparation of this manuscript. Generative AI was used for embellishing language.
Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.
Publisher’s note
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
Supplementary material
The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/10.3389/fenvs.2025.1676840/full#supplementary-material
Footnotes
1.^Based on data accessibility, land use efficiency in some western provinces cannot be presented. At the same time, it is also not measured for municipalities (Beijing, Tianjin, Chongqing, and Shanghai) because of differences in their administrative and development levels.
2.^There are no control variables on the left, and control variables on the right.
3.^There are no control variables on the left, and control variables on the right.
4.^The top left shows the year the policy was implemented; the top right indicates 1 year after the policy was implemented; the bottom left indicates 2 years after the policy was implemented; the bottom right indicates 3 years after the policy was implemented.
References
1
AmoreM. D.BennedsenM. (2016). Corporate governance and green innovation. J. Environ. Econ. Manag.75, 54–72. 10.1016/j.jeem.2015.11.003
2
ArkhangelskyD.AtheyS.HirshbergD. A.ImbensG. W.WagerS. (2021). Synthetic difference-in-differences. Am. Eco. Rev.111 (12), 4088–4118. 10.1257/aer.20190159
3
BaiX.ShiP.LiuY. (2014). Society: realizing China's urban dream. Nature509 (7499), 158–160. 10.1038/509158a
4
BorusyakK.JaravelX.SpiessJ. (2024). Revisiting event-study designs: robust and efficient estimation. Rev. Econ. Stud.91 (6), 3253–3285. 10.1093/restud/rdae007
5
CallawayB.Sant’AnnaP. H. (2021). Difference-in-differences with multiple time periods. J. Econometrics225 (2), 200–230. 10.1016/j.jeconom.2020.12.001
6
CaoM.ChangL.MaS.ZhaoZ.WuK.HuX.et al (2022). Multi-scenario simulation of land use for sustainable development goals. IEEE J. Sel. Top. Appl. Earth Observations Remote Sens.15, 2119–2127. 10.1109/jstars.2022.3152904
7
ChangJ.WangW.LiuJ. (2023). Industrial upgrading and its influence on green land use efficiency. Sci. Rep.13 (1), 2813. 10.1038/s41598-023-29928-8
8
CharnesA.CooperW. W.RhodesE. (1978). Measuring the efficiency of decision making units. Eur. J. Oper. Res.2 (6), 429–444. 10.1016/0377-2217(78)90138-8
9
CuiH.CaoY. (2023). How can market-oriented environmental regulation improve urban energy efficiency? evidence from quasi-experiment in China’s SO2 trading emissions system. Energy278, 127660. 10.1016/j.energy.2023.127660
10
EstoqueR. C.OobaM.TogawaT.HijiokaY.MurayamaY. (2021). Monitoring global land-use efficiency in the context of the UN 2030 agenda for sustainable development. Habitat Int.115, 102403. 10.1016/j.habitatint.2021.102403
11
FerreiraM. D. P.FéresJ. G. (2020). Farm size and Land use efficiency in the Brazilian Amazon. Land Use Policy99, 104901. 10.1016/j.landusepol.2020.104901
12
HerzigA.NguyenT. T.AusseilA. G. E.MaharjanG. R.DymondJ. R.ArnholdS.et al (2018). Assessing resource-use efficiency of land use. Environ. Modelling and Software107, 34–49. 10.1016/j.envsoft.2018.05.005
13
HillmanA. J.WithersM. C.CollinsB. J. (2009). Resource dependence theory: a review. J. Manag.35 (6), 1404–1427. 10.1177/0149206309343469
14
ImaiK.KeeleL.TingleyD. (2010). A general approach to causal mediation analysis. Psychol. Methods15 (4), 309–334. 10.1037/a0020761
15
KorosoN. H.LengoiboniM.ZevenbergenJ. A. (2021). Urbanization and urban land use efficiency: evidence from regional and addis ababa satellite cities, Ethiopia. Habitat Int.117, 102437. 10.1016/j.habitatint.2021.102437
16
LanJ.LiQ.ZhengY.LiuZ. (2023). The impact of the low-carbon city pilots programme on industrial land transfer by local governments in China. Econ. Analysis Policy77, 824–842. 10.1016/j.eap.2023.01.003
17
LeeC.-C.HeZ.-W. (2024). The impact of green finance policy on land ecological security: city-level evidence from China. Sustain. Cities Soc.105, 105347. 10.1016/j.scs.2024.105347
18
LiS.WangY.XuX. (2025). Can low-carbon city pilot policy improve urban energy-environmental efficiency? Evidence from China. Energy Rep.13, 2933–2945. 10.1016/j.egyr.2025.02.020
19
LiuL. (2024). Green innovation, firm performance, and risk mitigation: evidence from the USA. Environ. Dev. Sustain.26 (9), 24009–24030. 10.1007/s10668-023-03632-z
20
LiuS.XiaoW.LiL.YeY.SongX. (2020). Urban land use efficiency and improvement potential in China: a stochastic frontier analysis. Land Use Policy99, 105046. 10.1016/j.landusepol.2020.105046
21
LiuJ.HouX.WangZ.ShenY. (2021). Study the effect of industrial structure optimization on urban land-use efficiency in China. Land Use Policy105, 105390. 10.1016/j.landusepol.2021.105390
22
LiuJ.XuK.JinD.WuC.ZhangY. (2024). Urban low-carbon governance and ecological efficiency: new evidence from prefecture-level cities in China. Front. Ecol. Evol.12, 1442755. 10.3389/fevo.2024.1442755
23
LiuC.HuangH.YangJ. (2025). Can land system innovation promote the improvement of green land use efficiency in urban land—evidence from china’s pilot reform of the approval System for urban construction land. Land14 (4), 791. 10.3390/land14040791
24
Liu JJ.FengH.WangK. (2022). The low-carbon city pilot policy and urban land use efficiency: a policy assessment from China. Land11 (5), 604. 10.3390/land11050604
25
Liu XX.LiY.ChenX.LiuJ. (2022). Evaluation of low carbon city pilot policy effect on carbon abatement in China: an empirical evidence based on time-varying DID model. Cities123, 103582. 10.1016/j.cities.2022.103582
26
MaY.ZhengM.ZhengX.HuangY.XuF.WangX.et al (2023). Land use efficiency assessment under sustainable development goals: a systematic review. Land12 (4), 894. 10.3390/land12040894
27
NiuS.LuoX.YangT.LinG.LiC. (2023). Does the low-carbon city pilot policy improve the urban land green use efficiency? Investigation based on Multi-Period difference-in-differences Model. Int. J. Environ. Res. Public Health20 (3), 2704. 10.3390/ijerph20032704
28
PanM.SongH. (2017). Transformation and upgrading of old industrial zones on collective land: empirical study on revitalization in nanshan. Habitat Int.65, 1–12. 10.1016/j.habitatint.2017.04.014
29
PanA.ZhangW.ShiX.DaiL. (2022). Climate policy and low-carbon innovation: evidence from low-carbon city pilots in China. Energy Econ.112, 106129. 10.1016/j.eneco.2022.106129
30
QiaoR.ZhaoZ.WuT.ZhouS.AoX.YangT.et al (2025). Unveiling the nonlinear drivers of urban land resources on carbon emissions: the mediating role of industrial upgrading and technological innovation. Resour. Conservation Recycl.212, 108000. 10.1016/j.resconrec.2024.108000
31
RambachanA.RothJ. (2023). A more credible approach to parallel trends. Rev. Econ. Stud.90 (5), 2555–2591. 10.1093/restud/rdad018
32
RothJ.Sant’AnnaP. H.BilinskiA.PoeJ. (2023). What’s trending in difference-in-differences? A synthesis of the recent econometrics literature. J. Econ.235 (2), 2218–2244. 10.1016/j.jeconom.2023.03.008
33
SunH.ChenJ. (2023). The road to green innovation in agriculture: the impact of green agriculture demonstration zone on corporate green innovation. Environ. Sci. Pollut. Res.30 (57), 120340–120354. 10.1007/s11356-023-30707-1
34
TangM.WalshG.LernerD.FitzaM. A.LiQ. (2018). Green innovation, managerial concern and firm performance: an empirical study. Business Strategy and the Environment27 (1), 39–51. 10.1002/bse.1981
35
ToneK. (2002). A slacks-based measure of super-efficiency in data envelopment analysis. Eur. J. Oper. Res.143 (1), 32–41. 10.1016/S0377-2217(01)00324-1
36
TownsendR. M.UedaK. (2006). Financial deepening, inequality, and growth: a model-based quantitative evaluation. Rev. Econ. Stud.73 (1), 251–293. 10.1111/j.1467-937x.2006.00376.x
37
WangL.ShaoJ.MaY. (2023). Does China's low-carbon city pilot policy improve energy efficiency?Energy283, 129048. 10.1016/j.energy.2023.129048
38
YuY.ZhangN. (2021). Low-carbon city pilot and carbon emission efficiency: Quasi-experimental evidence from China. Energy Econ.96, 105125. 10.1016/j.eneco.2021.105125
39
ZengS.JinG.TanK.LiuX. (2023). Can low-carbon city construction reduce carbon intensity? Empirical evidence from low-carbon city pilot policy in China. J. Environ. Manag.332, 117363. 10.1016/j.jenvman.2023.117363
40
ZhangL.ZhangL.XuY.ZhouP.YehC. H. (2020). Evaluating urban land use efficiency with interacting criteria: an empirical study of cities in Jiangsu China. Land Use Policy90, 104292. 10.1016/j.landusepol.2019.104292
41
Zhao FF.HuZ.ZhaoX. (2023). Does innovative city construction improve urban carbon unlocking efficiency? Evidence from China. Sustain. Cities Soc.92, 104494. 10.1016/j.scs.2023.104494
42
Zhao XX.ZengS.KeX.JiangS. (2023). The impact of green credit on energy efficiency from a green innovation perspective: empirical evidence from China based on a spatial durbin model. Energy Strategy Rev.50, 101211. 10.1016/j.esr.2023.101211
43
ZouC.HuangY.WuS.HuS. (2022). Does “low-carbon city” accelerate urban innovation? Evidence from China. Sustain. Cities Soc.83, 103954. 10.1016/j.scs.2022.103954
Summary
Keywords
low-carbon city pilot, land use efficiency, green innovation, green finance, environmental policy
Citation
Zhao T and Duan L (2026) Evaluating the impact of low-carbon city policies on urban land use efficiency: evidence from China’s green transition. Front. Environ. Sci. 13:1696599. doi: 10.3389/fenvs.2025.1696599
Received
01 September 2025
Revised
09 November 2025
Accepted
27 November 2025
Published
06 March 2026
Volume
13 - 2025
Edited by
Gianluca Pappaccogli, University of Salento, Italy
Reviewed by
Wei Wang, Henan University, China
Yongshan Du, China University of Petroleum Beijing, China
Updates
Copyright
© 2026 Zhao and Duan.
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
*Correspondence: Longlong Duan, dll2007@scu.edu.cn
Disclaimer
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher.