Data-Driven Green Transformation: How Public Data Openness Fuels Urban Land Use Eco-Efficiency in Chinese Cities

Abstract

Urban land use eco-efficiency (ULUEE) encapsulates the equilibrium between economic gains and environmental sustainability. The improvement of ULUEE has emerged as a critical measure in addressing climate change and achieving dual-carbon objectives. This paper examines the potential of public data in enhancing ULUEE, focusing on public data openness (PDO), using a sample of 294 prefecture-level cities spanning from 2014 to 2022. The findings indicate that PDO has a significant positive impact on ULUEE, a result that remains robust through various sensitivity tests. Further analysis reveals that PDO fosters urban innovation, stimulates industrial agglomeration, optimizes urban industrial structures, and further enhances ULUEE through innovation effects, agglomeration effects, and structural effects. A heterogeneity analysis shows that this positive effect is more pronounced in regions with higher financial development levels and in the economically advanced eastern regions, suggesting that the ecological benefits derived from PDO are contingent upon a solid economic foundation. Additionally, the effect is more substantial in regions with weaker digital infrastructure and suboptimal environmental regulation, implying that public data can compensate for deficiencies in urban digital infrastructure and environmental governance, thereby contributing to improvements in ULUEE. This paper broadens the existing literature on the ecological value of public data, uncovers the potential of PDO in promoting ULUEE, and offers a practical framework for leveraging PDO to facilitate urban green transformation and ecological advancement.

1. Introduction

In recent years, nations worldwide have been actively advancing green and low-carbon transitions to address the pressing challenges of climate change and environmental degradation. The inherent demand of the industrialization process for land resources has driven a trend of disorderly urban expansion, which has exerted considerable pressure on the ecological environment and resulted in a suboptimal state of land use efficiency. While maintaining a focus on economic growth, there is an increasing emphasis on the efficient utilization of resources and environmental protection [1]. According to the International Energy Agency (IEA), China remains the world’s largest carbon emitter, contributing approximately one-quarter of global carbon dioxide emissions—substantially surpassing the combined emissions of the United States (14%) and the European Union (10%) [2]. As the world’s largest developing country and a major carbon emitter, China bears an important international responsibility in the global climate change response In this context, environmentally sustainable land use is recognized as an effective approach to achieve ‘dual-carbon’ goals [3]. This evidence underscores that adopting an environmentally sustainable model of land expansion is a crucial pathway for China to mitigate environmental pressures and advance green and low-carbon development, which holds substantial significance in achieving the dual-carbon goals and fostering the construction of ecological civilization on a global scale. With the rapid advancement of information technology, data have become a key factor of production, which have also significantly reshaped the operational modes and economic structures across various industries worldwide and further influence urban land use eco-efficiency (ULUEE). However, existing studies have neglected the potential of data elements, especially public data elements, in the optimization of ULUEE.

Low ULUEE is progressively undermining the foundation of economic growth, impeding the transformation and upgrading of the economic structure, and presenting a significant challenge to the sustainable development of Chinese cities. There is a consistent upward trend in the average total carbon emissions of prefecture-level cities in China, surpassing 35 million tons annually for three consecutive years. Furthermore, China’s waste treatment and recycling system remains in the developmental stage, with a significant portion of waste yet to be fully recycled and repurposed. Since reaching a peak value of 169.6646 million tons in 2019, solid waste generation has continued its upward trajectory, reaching 149.6677 million tons in 2022. The specific trends, as illustrated in Figure 1, underscore the critical challenges associated with ULUEE in China. Based on the above situation, it is urgent to improve ULUEE of China. ULUEE is influenced by a multitude of factors. However, academic research has predominantly examined ULUEE from the conventional perspective, such as research on innovation effects [4], agglomeration effects [5], and industrial effects [6]. However, the underlying common driving factor of data, which exists behind these influencing factors, has not been fully studied. Especially with the background of the digital economy, data elements have become a crucial factor of production, playing a vital role in promoting innovation, industrial agglomeration, and optimizing industrial structure [7]. However, the existing literature overlooks the role of the data factor on ULUEE.

With the development of the digital economy, data have emerged as a critical catalyst for industrial transformation [8]. In this transformative process, the extensive application of data has facilitated the digital and intelligent transformation of industries [9], enhanced the efficiency [10] and precision of resource allocation [11], and thereby promoted the high-quality development of the economy. Given that public data exhibit greater non-exclusivity compared to general data, their potential to deliver green value is correspondingly stronger, as they enhance resource utilization efficiency, reduce waste, and contribute to the achievement of sustainable development. In practical applications, public data openness (PDO) inherently manifests as two distinct and mutually exclusive states: “open” and “not open”. They function like a switch. Data are either in an open state, accessible and usable by the public and enterprises, or in a closed state, unavailable to external parties. According to a McKinsey report, public data constitute approximately 80% of all data elements and provide foundational support for economic development, social governance, and policy formulation. However, their full potential remains underdeveloped and underutilized [12]. On the one hand, PDO has considerable economic value. The effective utilization of PDO can accurately forecast energy consumption, material flow, and environmental changes, thereby enabling enterprises and governments to adopt more scientifically informed and environmentally sustainable decision-making [13]. On the other hand, PDO holds substantial ecological value. By fostering scientific decision-making, stimulating technological innovation, promoting social participation, and enhancing policy tools, PDO plays a pivotal role in improving ULUEE [14]. However, There is currently a lack of empirical evidence regarding how the PDO influences ULUEE, and the underlying mechanisms remain unclear. This paper clarifies the influencing mechanism and action path of PDO on ULUEE, which not only helps to expand the research perspective of PDO’s green value but also provides a decision-making basis for government departments to formulate precise policies to promote PDO green transformation.

Thus, this paper analyzes a sample of 294 prefecture-level cities using data from 2014 to 2022, examining the impact of PDO on ULUEE. This paper adopts a perspective centered on PDO and investigates the mechanisms through which it influences ULUEE, focusing on three key channels: innovation effects, agglomeration effects, and structural effects. This paper also explores the potential of PDO in enhancing ULUEE. The findings indicate that PDO has significantly contributed to the improvement of ULUEE, with results remaining robust after a series of sensitivity tests. Further analysis reveals that PDO enhances urban innovation levels, promotes industrial agglomeration, optimizes industrial structure, and, ultimately, improves ULUEE through the innovation, agglomeration, and structural effects. Heterogeneity analysis suggests that the impact of PDO is more pronounced in regions with higher levels of financial development and in eastern regions characterized by more advanced economic development. This indicates that the ecological benefits derived from PDO are contingent upon a certain economic foundation. Moreover, the positive effect is particularly evident in regions with underdeveloped digital infrastructure and weak environmental regulation, suggesting that public data can help compensate for deficiencies in urban digital infrastructure and environmental governance, thereby improving ULUEE. This paper expands the existing literature on the ecological value of public data, highlighting their potential to enhance ULUEE. It offers practical pathways for utilizing PDO to support urban green transformation and ecological improvement.

The research contributions of this paper mainly include the following aspects.

Firstly, this paper broadens the scope of research on the influencing factors of ULUEE. On the one hand, prior studies have primarily concentrated on the impact of traditional factors on ULUEE, such as policy guidance, land planning, and resource management [15,16], and neglected the role of the government in market construction, particularly in addressing market failures and mitigating information asymmetry. This paper seeks to explore this dimension. On the other hand, within the background of the digital economy, data have emerged as a critical production factor in market construction, playing a pivotal role in addressing market failures and alleviating information asymmetry problems [17]. PDO, as an important resource of the government, not only provides more precise decision-making support for land management but also optimizes the allocation and utilization of land resources through data-driven approaches, thereby enhancing the ecological benefits of land. By fostering public data sharing and openness, the government can more effectively monitor land use changes, identify potential ecological risks, and engage the community more broadly in land management and ecological protection efforts. However, existing research on ULUEE rarely addresses the impact of data as a production factor, particularly in terms of PDO. From this perspective, this paper elucidates the promoting effect of PDO on the ULUEE, thereby further enriching the research on the driving factors of ULUEE.

Secondly, this paper undertakes a more profound exploration of the ecological benefits derived from PDO, elucidating its potential to enhance ULUEE from the perspective of ecological contributions. Current academic research predominantly focuses on the relationship between PDO and economic growth as well as technological progress [18,19], and neglect the green value of PDO. Moreover, the research approaches employed in existing empirical studies often lack clarity and coherence in their methodologies. The academic community is increasingly focusing on the role of governments [20]; hence, this paper explores the driving factors of ULUEE from the perspective of PDO to enrich the related literature. To a certain extent, it addresses the gap in academic research on the correlation between PDO and green transformation, while also uncovering the significant potential of public data in facilitating the improvement of ULUEE. In addition, much of the existing literature remains confined to the case study level [21], lacking robust empirical evidence, and the mechanism through which PDO affects ULUEE, along with its heterogeneous effects, remains unclear [22]. And, this focus has not been integrated with the concept of PDO, overlooking the government’s role in promoting ULUEE and its critical significance in advancing green transformation.

Thirdly, this paper uncovers the green value of urban digital transformation from the perspective of PDO, thereby expanding the existing literature on the economic effects of urban digital transformation. Existing studies have revealed that GIS has established itself as an essential tool in urban planning and development, offering capabilities such as data integration, visualization, and analysis [23]. Despite its significant potential, the implementation of GIS is accompanied by challenges, including issues related to data quality, high operational costs, and demand for specialized training. Public data constitute a critical element in urban digital transformation [24], which encompass a wide range of information, including government administrative data, public sector information, and data generated by private entities under government contracts (like GIS data) [25]. By combining geographic information systems with advanced technologies such as the Internet of Things and artificial intelligence, PDO provides opportunities to address these challenges and create more resilient and inclusive urban spaces [26]. On this basis, GIS can further enhance its capacity to support sustainable urban development, improve the quality of life in rapidly urbanizing environments, and address the complexities of modern urban challenges. However, few studies have analyzed the source power of urban digital transformation and the release path of its green value from the perspective of PDO. Thus, from the perspective of PDO, this paper expands the ecological value of urban digital transformation from the three mechanisms of innovation effects, agglomeration effects, and structural effects and provides empirical support for the application of PDO in ULUEE.

The structure of this paper is organized as follows. The second section presents the theoretical analysis and hypotheses, outlining the promotional effects and mechanisms through which PDO influences ULUEE. The third section details the research model design, along with the sources and construction methods of the variables. The fourth section presents the results of the empirical analysis, which includes correlation analysis, descriptive statistics, basic regression, and robustness tests. The fifth section offers further analysis, discussing in detail, from an empirical perspective, the mechanisms through which PDO affects ULUEE. Finally, the sixth section provides the research conclusions and policy recommendations.

2. Theoretical Analysis and Hypotheses

2.1. Public Data Openness Improves Urban Land Use Eco-Efficiency Through Technological Innovation Effects

As a crucial tool for promoting industrial clusters, PDO can provide significant technical support for enterprise collaboration and resource integration by establishing an open and shared data resource platform. This platform facilitates collaboration and resource integration among enterprises, enabling stakeholders to better consolidate decentralized resources and enhance interaction and cooperation. Through data sharing, enterprises can gain a clearer understanding of the development potential of surrounding land and conduct in-depth analyses of possible market demand. This, in turn, helps avoid disorderly over-exploitation and resource idleness, ensuring a more rational allocation of land resources and improving their long-term sustainability. In the land use decision-making process, data sharing enhances transparency, reduces uncertainty, and provides a stronger data-driven foundation for decisions. This promotes the more scientific and refined utilization of land resources.

PDO also plays a crucial role in supporting the development of regional digital infrastructure. By leveraging open data, various technical platforms can rapidly access, analyze, and process information, providing a solid decision-making foundation for enterprises and fostering the creation of innovative products and services. At the level of land use, digital infrastructure not only enables the accurate monitoring of land use status but also identifies potential development and transformation opportunities by analyzing multidimensional data related to land space, function, and environment. This facilitates the more intensive and sustainable use of land resources. By opening data related to transportation, employment, environment, and population distribution, the government can utilize high-quality data to drive scientific planning, enhance urbanization levels, promote economic development, and improve transportation infrastructure [27]. Furthermore, these data enable a more precise understanding of urban development conditions and optimize the ecological layout of urban land [28]. PDO provides a wealth of data resources for green technology innovation and reduces the costs for enterprises and research institutions to obtain key information. By opening up environmental and related production data, enterprises and research institutions can conduct green technology research and development more efficiently, promote interdisciplinary and cross-sectoral cooperation, and thereby accelerate the development of green technology. From the perspective of the index composition of ULUEE, environmental benefit is a crucial output indicator of land use eco-efficiency. Additionally, green transformation, as an economic consequence of reduced land ecological destruction, is an important manifestation of optimized land resource utilization. It is evident that regional green transformation is an integral component of ULUEE. In the context of limited urban land resources, in-depth analysis of public data allows the government to allocate land resources more efficiently, reduce the negative environmental impacts of land development, and improve ULUEE. Additionally, through accurate data, the government can monitor urban development trends, strategically plan transportation infrastructure, and implement environmental protection measures, thereby strengthening the sustainable development capacity of society as a whole. Based on the above, this paper proposes the following:

H1. 

PDO promotes technological innovation, thereby improving ULUEE.

2.2. Public Data Openness Improves Urban Land Use Eco-Efficiency Through Industrial Agglomeration Effects

Industrial clusters can significantly enhance the ULUEE through synergistic effects and the optimal allocation of resources. These clusters leverage the concentration of geographical space and the optimization of factor flows, creating substantial economies of scale and network externalities. With the aid of PDO, enterprises can more effectively identify potential development opportunities or transformation spaces, thereby improving the intensive use of land through precise analysis of multidimensional data, such as spatial structure, functional requirements, and environmental changes. Moreover, under the conditions of PDO, the sharing and circulation of information expand market reach beyond local interests, facilitating more efficient communication and decision-making between enterprises and governments. This transparent flow of information not only reduces competitive friction between enterprises and fosters trust but also accelerates collaborative industrial innovation within a region. By utilizing real-time data, enterprises in industrial clusters can gain a better understanding of each other’s needs and development trends, thereby cooperating in production, research and development, logistics, and other operations. This collaborative synergy minimizes blind competition and the wastage of land resources, ultimately promoting the more efficient allocation of land resources [29]. This transparent flow of information mitigates the friction that enterprises may face in collaboration, enhances trust among market participants, and subsequently fosters collaborative innovation within the region. On the one hand, industrial clusters minimize redundant construction and inefficient land use by facilitating collaboration and resource sharing among workers in both production and daily life. Through cluster collaboration, the spatial arrangement and environmental value of land can be optimized, which reduces the excessive consumption of natural resources and enhances ecological service functions, thereby improving the intensive utilization efficiency of land [30]. On the other hand, as enterprises within the cluster develop in a coordinated manner, the comparative advantages between different economies and industries within the cluster gradually emerge, thus improving the overall efficiency of resource allocation in the region. Driven by agglomeration effects, land resources in the area are more fully developed and efficiently utilized, leading to enhanced ecological and economic benefits. As a result, land resources are both maximally developed and efficiently utilized, optimizing both ecological and economic outcomes [31]. Based on the above, this paper proposes the following:

H2. 

PDO promotes industrial agglomeration, thereby improving ULUEE.

2.3. Public Data Openness Improves Urban Land Use Eco-Efficiency Through Industrial Structure Effects

In traditional industrial systems, inefficient resource allocation often results in the excessive expansion of certain industries, while other sectors with significant potential remain underdeveloped, leading to resource waste and inefficient land use. PDO addresses this issue by providing real-time, accurate information to market participants. This facilitates the effective channeling of capital, technology, and talent toward industries with higher economic benefits and greater development potential. Such a process not only promotes the rapid circulation of underutilized resources but also accelerates the adjustment of IS, enabling the regional economy to transition smoothly from low-value-added to high-value-added industries [32], thus promoting the rationalization of industries.

At the same time, PDO provides robust data support for the development of emerging industries. Traditionally, enterprises faced significant barriers to entry when venturing into new fields [33]. However, in the digital era, PDO offers valuable decision-making insights and innovation opportunities, thereby reducing the costs and risks associated with research and development, as well as market entry into emerging sectors [34]. Enterprises can leverage open public data to more accurately identify market demand, adjust product positioning, and make more informed decisions during the product development phase, based on extensive data analysis [35]. PDO significantly diminishes market uncertainty, offering emerging industries greater opportunities for development and advancing industries to higher levels of growth and sophistication.

For urban development, PDO—particularly the sharing of data related to land use, urban development, and traffic flow—enables decision-makers to plan land use with greater precision. This helps prevent the over-occupation of land resources by inefficient industries and encourages the rational use of land by more efficient industries [36]. By analyzing public data such as the distribution of industrial clusters, population mobility trends, and changes in traffic flow, urban managers can identify optimal land use patterns. This enables the allocation of limited land resources to industries with higher production efficiency and lower environmental impact, thus alleviating the scarcity of urban land resources. Additionally, this promotes the sustainable development of the regional economy and helps mitigate the environmental burdens associated with unchecked urban expansion.

Based on this, this paper proposes the following:

H3. 

PDO optimizes IS, thereby improving ULUEE.

The theoretical framework of this paper is the following (Figure 2).

3. Methods

3.1. Models

3.1.1. Baseline Regression Model

This paper begins by examining the relationship between PDO and ULUEE. In contrast to the first-order difference method, this paper utilizes the difference-in-differences approach for the analysis. This methodological choice is motivated by the presence of additional factors that may influence ULUEE both prior to and following the adoption of the PDO platform by a city. Furthermore, other policies implemented concurrently may also impact ULUEE in cities that are not designated as PDO pilot sites. Collectively, these factors can compromise the accuracy of the conclusions. The first-order difference method, due to its inability to account for these confounding variables, may yield overestimated results. Consequently, the role of the PDO platform should be rigorously examined through the application of the more robust difference-in-differences approach. Using the implementation of the PDO platform pilot policy as a quasi-natural experiment, and drawing on the study [37], a multiple-period difference-in-differences method is used to explain the relationship between these two variables. The following two-way fixed-effects econometric model is constructed:

$${\mathrm{U}\mathrm{L}\mathrm{U}\mathrm{E}\mathrm{E}}_{\mathrm{i}\mathrm{t}}={\alpha }_0+{\alpha }_1{PDO}_{it}+{\alpha }_i{controls}_{it}+{\mu }_i+{\upsilon }_t+{\epsilon }_{it}$$

(1)

In the model, ${ULUEE}_{it}$ represents the ULUEE of city i in year t. ${\mathrm{P}\mathrm{D}\mathrm{O}}_{\mathrm{i}\mathrm{t}}$ represents whether city i is a pilot city for the PDO platform in year t. ${\alpha }_0$ represents the intercept term. ${controls}_{it}$ represents the group of control variables. ${\mathsf{\mu }}_{\mathrm{i}}$ and ${\mathsf{\upsilon }}_{\mathrm{t}}$ represent city fixed effects and time fixed effects, respectively. ${\mathsf{\epsilon }}_{\mathrm{i}\mathrm{t}}$ is the random disturbance term. ${\alpha }_1$ represents the average effect of PDO on ULUEE. If ${\alpha }_1$> 0, it indicates that PDO can improve ULUEE; if ${\alpha }_1$ < 0, it indicates that PDO reduces ULUEE.

3.1.2. Mediating Effect Regression Model

PDO leverages big data technology for analysis, decision-making, and digital management, with technological innovation and industrial agglomeration serving as mediating factors to enhance ULUEE. To test this hypothesis, this paper further develops a mediating effect model to investigate whether technological innovation, industrial agglomeration, and the optimization of industrial structure play a mediating role in the influence of PDO on ULUEE.

Drawing on the research of Judd and David (1981) [38], a mediating regression model is used for linear regression analysis, and the model is constructed as follows:

$$Y=(c^{\prime }+ab)X+{\tau }_{1}b+{\tau }_2$$

(2)

${\mathrm{c}}^{\prime }$ is the direct effect of the explanatory variable on the explained variable, $\mathrm{a}\mathrm{b}$ is the mediating effect of the explanatory variable on the explained variable, and ${\mathsf{\tau }}_1, {\mathsf{\tau }}_2$ are the residual term.

We select ULUEE as the dependent variable and PDO as the independent variable and construct the model as follows:

$${ULUEE}_{it}={\beta }_0+{\beta }_1{PDO}_{it}+{\beta }_2{col}_{it}+{\tau }_{it}$$

(3)

$${media}_{it}={\gamma }_0+{\gamma }_1{PDO}_{it}+{\gamma \beta }_2{col}_{it}+{\tau }_{it}$$

(4)

${ULUEE}_{\mathrm{i}\mathrm{t}}$ represents the ability of a city to maximize economic and social benefits from land use while minimizing environmental impacts on the city-period, ${media}_{\mathrm{i}\mathrm{t}}$ denotes the proxy variables for technological innovation, industrial agglomeration, or IS in a city-period, and ${\mathrm{c}\mathrm{o}\mathrm{l}}_{\mathrm{i}\mathrm{t}}$ represents the control variables for the city-period values. The coefficients to be estimated are ${\mathsf{\beta }}_{\mathrm{i}},{\mathsf{\gamma }}_{\mathrm{i}}$, and ${\mathsf{ϵ}}_{\mathrm{i}},$ and ${\mathsf{\tau }}_{\mathrm{i}\mathrm{t}}$ represents the random disturbance term.

3.2. Variables

3.2.1. Explained Variable

This paper employs the SBM-Undesirable model to measure ULUEE. The SBM-Undesirable model is an extension of the traditional DEA model, designed to effectively address efficiency measurement challenges involving slack variables and undesirable outputs. Compared with the traditional DEA model, which cannot effectively deal with the unexpected output, SBM-Undesirable treats the unexpected output as an independent variable rather than forcing it into input or adjusted through mathematical transformation, which is more in line with the actual production logic (the unexpected output is a by-product of the production process, rather than an input). In addition, the efficiency value of SBM-Undesirable is determined by the relaxation of the input, expected output, and undesired output, avoiding the problem of the efficiency overestimated by traditional DEA by ignoring the relaxation. In comparison to the SBM-Undesirable model, the latter is inadequate in addressing unexpected outputs effectively. Conversely, the SBM-Undesirable model considers unexpected outputs as autonomous variable, eschewing the practice of forcing them into being inputs or rectifying them via mathematical alterations. This approach aligns more closely with the authentic production rationale, wherein unexpected outputs are construed as byproducts rather than inputs of the production process. The efficacy metric of the SBM-Undesirable model is ascertained by the alleviation of inputs, anticipated outputs, and unforeseen outputs, thereby precluding the issue of efficiency overvaluation that arises from overlooking alleviation in the traditional DEA model. In line with existing academic research, when discrepancies arise between results under two different technology assumptions, the results obtained under the variable returns to scale (VRS) assumption should be prioritized [39]. Therefore, the SBM-Undesirable model measure based on variable returns to scale (VRS) is used with the following rationale: Assuming that there are n decision evaluation units in the urban land use process, all with m input indicators, ${\mathrm{x}}_{\mathrm{i}0}$, a desired output indicators, ${\mathrm{y}}_{\mathrm{r}0}^{\mathrm{e}}$, and b non-desired output indicators, ${\mathrm{y}}_{\mathrm{h}0}^{\mathrm{n}}$, the matrices X, ${\mathrm{Y}}^{\mathrm{e}}$, and ${\mathrm{Y}}^{\mathrm{n}}$ can be defined as (${\mathrm{x}}_1,{\mathrm{x}}_2,$…${\mathrm{x}}_{\mathrm{n}}$) ϵ${\mathrm{R}}^{\mathrm{m}\times \mathrm{n}}$, (${\mathrm{y}}_1{,\mathrm{y}}_2,$…${\mathrm{y}}_{\mathrm{n}}$) ϵ${\mathrm{R}}^{\mathrm{b}\times \mathrm{n}}$. Also, assuming that X, ${\mathrm{Y}}^{\mathrm{e}}$, and ${\mathrm{Y}}^{\mathrm{n}}$ are all greater than zero, the production possibility set can be defined as $\mathrm{p}=\left\{\left(\mathrm{x},{\mathrm{y}}^{\mathrm{e}},{\mathrm{y}}^{\mathrm{n}}\right)\left|\mathrm{x}\ge \mathrm{X}\mathsf{\lambda }\right.,{\mathrm{y}}^{\mathrm{e}}\le {\mathrm{Y}}^{\mathrm{e}}\mathsf{\lambda },{\mathrm{y}}^{\mathrm{n}}\ge {\mathrm{Y}}^{\mathrm{n}}\mathsf{\lambda },\mathsf{\lambda }\ge 0\right\}$. The SBM-Undesirable model expression is

$$\begin{gathered} \rho *=min{\displaystyle \frac{1-\frac{1}{m}\sum _{i=1}^m\frac{D_i^{-}}{x_{i0}}}{1+\frac{1}{a+b}\left(\sum _{r=1}^a\frac{D_r^e}{y_{r0}^e}+\sum _{h=1}^b\frac{D_h^n}{y_{h0}^n}\right)}} \\ s,t,x_0=X\lambda +D^{-},y_0^e=Y^e\lambda -D^e,y_0^n=Y^n\lambda +D^n \\ {D}^{-}\ge 0,D^e\ge 0,D^n\ge 0,\lambda \ge 0 \end{gathered}$$

In the formula, ρ* represents the value of the ULUEE of each city in the study area, and ρ* is between 0 and 1. ρ* = 1 represents that the evaluation unit is effective; ρ* < 1 indicates that there is an efficiency loss in the evaluation unit and that there is a room for the optimization and improvement of the inputs and outputs; m, a, and b represent the number of inputs, desired outputs, and non-desired outputs, respectively; ${\mathrm{D}}^{-}{, \mathrm{D}}^{\mathrm{e}}$, and ${\mathrm{D}}^{\mathrm{n}}$ represent the redundancy of the inputs, the insufficiency of the desired outputs, and the excess of the non-desired outputs, respectively; ${\mathrm{x}}_{\mathrm{i}0}, {\mathrm{y}}_{\mathrm{r}0}^{\mathrm{e}}$ and ${\mathrm{y}}_{\mathrm{h}0}^{\mathrm{n}}$ represent the corresponding input–output matrix of each evaluation unit; and λ represents the weight vector. In this paper, the index system of ULUEE is constructed from the two dimensions of the input and output, as shown in

Table 1. Indicator system for ULUEE.

Indicator Type	Indicator Explanation		Indicator Composition
Input indicators	Land		Urban construction area (km2).
	Capital		Total social fixed asset investment (in ten thousand CNY).
	Labor		Number of employees in secondary and tertiary industries (in ten thousand people).
Output indicators	Economic benefits	Expected output	Value added in secondary and tertiary industries (in ten thousand CNY).
	Social benefits		Total urban population (in ten thousand people).
	Environmental benefits	Undesirable output	PM2.5 concentration.
			Carbon emissions.

and Figure 3.

(1)

Input Indicators. In accordance with the production function, land factors are incorporated into the production function model. At the input level, this paper primarily selects three factors: land, capital, and labor [40]. Specifically, the land indicator is represented by the urban construction area, capital is represented by the total social fixed asset investment, and labor is represented by the number of employees in the secondary and tertiary industries.

(2)

Output Indicators. These indicators are selected from three dimensions: the economy, society, and the environment. The economic indicator is represented by the value-added output of the secondary and tertiary industries; the social indicator is represented by the total population of the city; and the environmental indicators are represented by PM2.5 concentration and urban carbon emissions. Among these, the environmental indicators are considered undesirable outputs.

3.2.2. Explanatory Variable

This paper uses a sample of 294 prefecture-level cities, of which 221 cities are designated as pilot cities for the PDO platform, forming the treatment group. The remaining prefecture-level cities, which are not designated as pilot cities, constitute the control group, as illustrated in Figure 4. Following the methodology outlined by Hong et al. [41], a dummy variable, PDO, is constructed based on the year each city becomes a pilot for the PDO platform. If a city sample becomes a pilot city during the observation period (i.e., the treatment group) and the observation time is subsequent to the year it is selected, the PDO variable is assigned a value of 1; otherwise, it is assigned a value of 0.

There are primarily four distinct views on PDO: the bureaucratic perspective, the political perspective, the technical perspective, and the economic perspective [42]. Regarding the concept of PDO, most existing international directives on the topic are broadly generic. Given the open nature of public data, it is impossible to achieve absolute uniformity in the degree of PDO across different regions. The launch of China’s public data platforms serves as a quasi-natural experiment, providing a benchmark for measuring the extent of PDO [43,44]. This paper utilizes the launch of public data opening platforms promoted by local governments as an exogenous policy shock. Empirical analysis was conducted employing the multi-time-point difference-in-differences method to assess the impact of PDO.

3.2.3. Control Variables

Referring to Wang et al. (2021) [45], Yu and Zhou (2023) [46], to ensure the accuracy and reliability of the research results, this paper controls for various factors that may affect the outcomes, with the definitions of the main variables shown in

Table 2. The definitions of key variables.

	Variable Name		Variable Symbol	Calculation Method
Explained variable	Urban land use eco-efficiency		ULUEE	Calculated using SBM-Undesirable model.
Explanatory variable	Public data openness		PDO	If the city sample is a pilot city for public data development during the observation period (i.e., the treatment group), and the observation time is after the selected year, the PDO variable takes the value of 1; otherwise, it takes the value of 0.
Control variables	Input for scientific research		ISR	Local financial science expenditure/local financial budget expenditure.
	Informatization level		IL	Regional per capita postal business volume.
	Industrial structure		IS	Ln (tertiary industry GDP/regional GDP).
	Economic development level		EDL	Ln (per capita GDP).
	Educational input		EI	Local financial education expenses/local financial budget expenditure.
Mechanistic variables	Technological innovation	Total patent licensing	TPL	Ln (total regional patent licensing).
		Invention patent grants	IPGs	Ln (regional invention patent grants).
		Utility model patents	UMPs	Ln (regional utility model patent grants).
	Cluster of industries	Population agglomeration	PA	Population density.
		Economic agglomeration	EA	Employment density (labor force per unit land area).
		Industrial agglomeration	IA	Location entropy.
	Industrial structure	Rationalization of industrial structure	RIS	Theil index of industrial structure.
		Upgrading of industrial structure	UIS	Coefficient of industrial structure level.

. This paper takes into account the interference of comprehensive characteristic factors of potential urban development from key dimensions such as technological innovation, resource allocation, economic structure, level of development, and human capital. Specifically, investment in scientific research (ISR) directly affects the technological level and ULUEE; the level of informatization (IL) optimizes decision-making and the management efficiency of land use; industrial structure (IS) determines the economic efficiency and ecological benefits of land use; the level of economic development (EDL) influences the input and output of land use; and educational investment (EI) indirectly affects the ecological efficiency of land use by enhancing public awareness and professional capabilities.

3.2.4. Mechanistic Variables

This paper constructs mediating variables from three perspectives: innovation effects, agglomeration effects, and structural effects. Total patent licensing (TPL), invention patent grants (IPGs), and utility model patents (UMPs) are used as proxies for innovation effects. Regional patent authorizations serve as a measure of regional innovation capacity, providing a more accurate reflection of technological innovation levels.

Population agglomeration (PA), economic agglomeration (EA), and industrial agglomeration (IA) are selected as proxies for agglomeration effects. The degree of PA is measured by regional population density [47], while employment density is used to gauge the level of EA [48]. The industry location entropy of each city is used as an indicator for IA [49], the location entropy index method is generally used to calculate the number of employees or output value, urban land use is dominated by the secondary and tertiary industries, and industrial, commercial, and public service land constitute the core functions, so the secondary and tertiary industries are used for measurement. The specific formula is the following:

$${AG}_{ij}={\displaystyle \frac{E_{ij}/\sum _j{E}_{ij}}{\sum _i{E}_{ij}/\sum _i\sum _j{E}_{ij}}}$$

${AG}_{ij}$ indicates the industrial agglomeration level of industry j in city i, $E_{ij}$ represents the number of employees in city I in industry j, $\sum _j{E}_{ij}$ represents the number of employees in all industries in city i, $\sum _i{E}_{ij}$ represents the number of employees in all cities in industry j, and $\sum _i\sum _j{E}_{ij}$ indicates the total number of employees in China.

For structural effects, this paper selects two proxies: the level of industrial structure rationalization (RIS) and the level of industrial structure upgrading (AIS). From a dynamic perspective, IS upgrading is viewed as comprising two stages: rationalization and upgrading. To measure RIS, the paper utilizes the Theil index of IS [50], while AIS is measured using the IS hierarchy coefficient [51].

3.3. Sample and Descriptive Statistics

This paper selected panel data from 294 prefecture-level and above cities in China from 2014 to 2022 as the research sample. The data sources included the “China Urban Construction Statistical Yearbook”, “China City Statistical Yearbook”, “China Statistical Yearbook”, as well as statistical yearbooks and databases such as the Columbia University Social and Economic Data and Applications Center, Wind Database, and CSMAR Database. For some missing data, linear interpolation was used to fill the gaps, resulting in 2646 sample data points. The linear interpolation method is particularly suitable for scenarios where the number of missing data points is limited and the overall data trend remains relatively stable [52]. Considering the sample characteristics of this paper, this paper used a linear interpolation method combining interpolation and extrapolation to complete the data.

Figure 5 illustrates ULUEE across cities in China. With the ongoing development of China’s economy, there is an overall gradual upward trend in ULUEE. However, the regional disparities in ULUEE have widened, particularly in economically developed areas, where there is a tendency to prioritize development at the expense of ecological benefits.

Table 3. Descriptive statistics.

Variable	Obs.	Mean	Std. Dev.	Min.	Max.
ULUEE	2646	0.5881	0.1760	0.1514	1.0000
PDO	2646	0.3458	0.4757	0.0000	1.0000
ISR	2646	0.0225	0.0722	−0.0998	1.0000
IS	2646	3.7955	0.2589	2.9857	6.2559
EDL	2646	10.8559	0.5450	9.0724	12.4565
EI	2646	0.1713	0.0388	−0.0021	0.3046
IL	2646	2557.9442	3033.3297	41.0000	15,700.0000
TPL	2631	6.0058	1.7462	0.0000	11.2806
IPGs	2625	5.3035	1.8523	0.0000	11.3865
UMPs	2639	7.2959	1.6633	0.0000	11.9827
PA	2538	5.7596	0.8881	2.9084	7.2127
EA	2538	3.5207	1.1521	0.4974	6.5644
IA	2538	1.8826	3.1199	−3.0278	25.4857
RIS	2547	0.3026	0.2993	−1.8416	2.6765
UIS	2547	6.6500	0.3616	4.3192	7.9467

presents the observed values, mean, standard deviation, and both the maximum and minimum values for each variable. It is evident that the mean ULUEE is 0.5881, with a maximum value of 1 and a minimum value of 0.1514. The standard deviation is 0.176, indicating considerable variation and volatility within the sample values. Other statistical measures also reveal similar patterns.

4. Results

4.1. Baseline Results

Firstly, this paper investigates the direct impact of PDO on ULUEE. In Column (1) of

Table 4. Baseline regression results.

Variable	(1)	(2)
	ULUEE	ULUEE
PDO	0.011 *	0.019 ***
	(1.749)	(2.668)
ISR		0.040 ***
		(3.980)
IS		0.100 **
		(2.453)
EDL		−0.045 ***
		(−2.914)
EI		−0.182
		(−1.512)
IL		0.000 ***
		(4.432)
Year FE	YES	YES
City FE	YES	YES
N	2646	2646
R2	0.767	0.766

Note: t values are in parentheses; ***, **, and * represent 1%, 5%, and 10% significance levels.

, the regression results without control variables are presented, while Column (2) shows the regression results with control variables included. The findings indicate that the coefficient of PDO is significantly positive in both cases, suggesting that the policy of PDO has a substantial positive effect on enhancing ULUEE.

4.2. Robustness Test

4.2.1. Parallel Trend Test

In this paper, a parallel trend test was carried out by the Event Study Approach (ESA) [53,54] and the model was set up as follows:

$${ULUEE}_{it}=\alpha +\sum _{k=-10}^5{\beta }_k{D}_{i,t_0+k}+\gamma X_{it}+{\mu }_t+{\nu }_i+{\epsilon }_{it}$$

where $D_{i,t_0+k}$ is a dummy variable indicating the kth year after the implementation of the PDO platform pilot. Corresponding to the baseline regression model, this paper takes the year before the implementation of the PDO platform pilot as the baseline year, and the ${\beta }_k$ coefficient is the variable on which the parallel trend test focuses.

Using the year prior to the policy implementation as the base period, the regression results obtained from the above equation are presented in Figure 6. The regression coefficients prior to the policy implementation are not statistically significant, indicating no notable differences between the experimental and control groups during this period. This result confirms the validity of the parallel trend assumption. Following the policy implementation, the regression coefficients become statistically significant, and the dynamic effects exhibit an upward trend. This suggests that as PDO accumulates, the value of public data progressively increases, leading to a sustained improvement in ULUEE. Therefore, the baseline regression is valid.

4.2.2. Placebo Test

This paper used a placebo test to verify further whether the impact of PDO on ULUEE is driven by other unobservable factors [55], as shown in Figure 7. A dummy group was constructed for 500 repeated sampling regressions, and the regression coefficients approximately follow a normal distribution with a mean of 0. The coefficient of 0.019 from the baseline regression is positioned far on the right tail of the fake regression coefficient distribution, making it a low-probability event in the placebo test. Based on this, it can be concluded that the benchmark results are not driven by unobservable factors at the city and time levels, and the placebo test was passed.

4.2.3. PSM-DID Test

This paper attempted to further mitigate the potential endogeneity problem by using propensity score matching, and the inclusion of appropriate covariates helped to better balance the characteristics of the sample, which in turn improved the accuracy of the PSM-DID test [56]. In this paper, a series of covariates were added to the full set of covariates present in the baseline regression, taking into account the accuracy of the matching results. The scale expansion of social consumption levels (SCLs) exacerbates land supply pressures and generates negative ecological externalities [57], the greening coverage ratio (GCR) of built-up areas can buffer the impact of human activities on eco-efficiency [58], and trade volatility (TD) can trigger eco-efficiency losses through chain transmission [59], so this paper added them to the regression results of the test.

The balance test results presented in

Table 5. Balance test hypothesis.

Variables	Sample	Mean Value		Standard Error (%)	Absolute Value of Standard Error Reduced (%)	t-Test
		Treatment Group	Control Group			t Value	p Value
ISR	unmatched	0.0258	0.0266	−1.100	−170.3	−0.190	0.848
	matched	0.0255	0.0233	3		0.880	0.378
IS	unmatched	3.860	3.784	29.50	95.10	5.490	0
	matched	3.854	3.850	1.500		0.280	0.776
EDL	unmatched	11.11	10.84	55.10	99.50	10.80	0
	matched	11.10	11.10	0.300		0.0500	0.963
EI	unmatched	0.183	0.166	46.40	96.70	8.870	0
	matched	0.183	0.182	1.500		0.280	0.783
IL	unmatched	3131	2236	27.50	73.70	5.980	0
	matched	3006	2770	7.200		1.220	0.221
TD	unmatched	0.223	0.231	−1.900	−106.2	−0.360	0.722
	matched	0.220	0.205	3.900		0.840	0.403
SCL	unmatched	0.401	0.410	−8.800	99.30	−1.730	0.0830
	matched	0.401	0.401	−0.100		−0.0100	0.991
GCR	unmatched	42.77	40.57	48.50	98.40	9.090	0
	matched	42.65	42.61	0.800		0.160	0.870

demonstrate satisfactory covariate equilibrium, with standardized differences below the 10% threshold across all matched variables. The post-matching t-statistics fail to reject the null hypothesis of no systematic intergroup differences (p > 0.05), confirming a successful reduction in selection bias. The matching procedure achieved significant improvements across two key dimensions: (1) a substantial reduction in mean differences between treatment and control cohorts; (2) elevated absolute percentage reductions in standardized bias metrics. These outcomes collectively indicate enhanced distributional parity across matched covariates, verifying the PSM methodology’s robust balancing capacity. The empirical evidence thereby substantiates the technique’s effectiveness in mitigating confounding through systematic intergroup alignment, ultimately strengthening causal inference validity.

Figure 8 presents the covariate balance assessment through standardized mean difference visualizations. Pre-matching analysis revealed substantial systematic disparities across multiple variables, reflecting marked pretreatment heterogeneity between intervention and comparison cohorts (absolute deviations > 10%). This initial imbalance substantiated significant selection bias risks within the observational sample. Post-matching implementation demonstrated notable equilibrium improvements, with standardized deviations reduced to near-zero magnitudes (median < 5%), confirming the PSM methodology’s efficacy in achieving distributional parity across matched covariates. This empirical alignment between treatment and control groups substantially mitigated systematic selection bias through improved distributional alignment, thereby enhancing causal inference validity through systematic intergroup comparability.

Figure 9 further validates methodological rigor through propensity score distribution analysis. The extensive overlap in propensity score common support domains (intersection range: 0.15–0.85) indicates sufficient comparative analytical capacity between matched cohorts. The broad common value range ensures robust counterfactual estimation through like-to-like comparisons, fulfilling the conditional independence assumption prerequisite for quasi-experimental designs. This comprehensive propensity score overlap (85%+ matched pairs within common support) provides empirical validation of the matching procedure’s internal validity while establishing a statistically coherent foundation for subsequent treatment effect estimation.

In this paper, regression analysis of the matched results was carried out. The results are shown in column (1) of

Table 6. After matching results.

Variable	(1) 1:1 Nearest-Neighbor Matching	(2) Kernel Matching	(3) Caliper Matching
	ULUEE	ULUEE	ULUEE
PDO	0.032 **	0.017 **	0.016 **
	(2.253)	(2.230)	(2.147)
ISR	0.207	0.091 **	0.094 **
	(0.515)	(2.141)	(2.178)
IS	−0.026	−0.039 **	−0.040 **
	(−0.759)	(−2.328)	(−2.361)
EDL	0.052 **	0.045 ***	0.046 ***
	(2.350)	(4.201)	(4.196)
EI	−0.138	−0.108	−0.112
	(−0.431)	(−0.725)	(−0.740)
IL	0.000 ***	0.000 ***	0.000 ***
	(2.704)	(2.969)	(2.959)
TD	−0.056 **	−0.030 ***	−0.030 ***
	(−2.080)	(−2.702)	(−2.669)
SCL	0.031	0.074 *	0.077 *
	(0.352)	(1.882)	(1.898)
GCR	0.002	0.001	0.001
	(0.959)	(1.554)	(1.534)
_cons	−0.033	0.134	0.133
	(−0.108)	(0.915)	(0.892)
Year FE	YES	YES	YES
City FE	YES	YES	YES
N	698	1817	1787
R2	0.789	0.782	0.778

Note: t values are in parentheses; ***, **, and * represent 1%, 5%, and 10% significance levels.

, indicating that the PDO coefficient is significantly positive, which is consistent with the benchmark regression results, and the PSM test was passed.

Then, this paper also carried out kernel matching through the introduction of the kernel function, weighting all the control group individuals so that each individual in the treatment group could obtain information from all the individuals in the control group, avoiding the problem of information loss that may have been caused by the strict one-to-one matching, and being able to more fully sample the data to improve the flexibility and accuracy of the matching so that the matched samples were closer to each other in the distribution of covariates in the overall situation, thus better estimating the processing effects [60]. Finally, this paper further conducted caliper matching with a scale of 0.01, which strictly limited the difference in propensity scores between the matched samples by setting a caliper threshold, ensured that the matched samples had a high degree of similarity in propensity scores, further enhanced the comparability between the treatment group and the control group, effectively reduced the matching bias, and made the matched samples more representative of the potential causal relationship [61].

The regression results after kernel matching and caliper matching are shown in columns (2) and (3) in Table 6, which indicate that PDO has a significant positive effect on ULUEE, further enhancing the reliability of the baseline regression results.

4.2.4. Exclusion of Relevant Competitive Policies

There were several policy shocks similar to the introduction of the PDO platform within the sample period, which may potentially confound the regression results in this paper. To address this, three policies analogous to PDO were selected for testing, with the results presented in columns (1), (2), (3), and (4) of

Table 7. The hypothesis test excluding contemporaneous terms and the lagged explanatory variable results.

Variable	(1)	(2)	(3)	(4)	(5)
	ULUEE	ULUEE	ULUEE	ULUEE	ULUEE
PDO	0.019 ***	0.019 ***	0.021 ***	0.022 ***
	(2.668)	(2.725)	(3.002)	(3.078)
L.PDO					0.022 ***
					(3.090)
SMART	0.000			0.000
	(.)			(.)
INFO		0.000		0.000
		(.)		(.)
BIGDATA			−0.039 ***	−0.038 ***
			(−3.887)	(−3.731)
Controls	YES	YES	YES	YES	YES
Year FE	YES	YES	YES	YES	YES
City FE	YES	YES	YES	YES	YES
N	2097	2078	2088	2069	1864
R2	0.766	0.769	0.769	0.773	0.783

Note: t values are in parentheses; ***, **, and * represent 1%, 5%, and 10% significance levels.

.

Firstly, since both the PDO platform and smart city construction (SMART) serve as key tools for providing new types of urban digital public services, this paper included both variables in the regression model. As shown in Column (1) of Table 7, the coefficient for PDO is significantly positive, while the coefficient for SMART is not statistically significant. This indicates that the positive effect of PDO on ULUEE is not influenced by the smart city construction policy.

Secondly, since both the PDO platform and the national pilot of information benefits to the people (INFO) aim to optimize public resource allocation through informatization, this paper included the national pilot of information benefits to the people alongside the PDO platform in the regression model. As shown in Column (2) of Table 7, the coefficient for PDO remains significantly positive, while the coefficient for INFO is not statistically significant. This suggests that the impact of PDO on ULUEE is not influenced by the national pilot policy on information benefits to the people.

Thirdly, since both the PDO platform and the national big data comprehensive experimental zone (BIGDATA) are follow-up initiatives to implement the State Council’s “Action Plan for Promoting Big Data Development”, this paper included the national big data comprehensive experimental zone alongside the PDO platform in the regression model. As shown in Column (2) of Table 7, the coefficient for PDO remains significantly positive, while the coefficient for BIGDATA is significantly negative. This suggests that the national big data comprehensive experimental zone policy has a suppressive effect on ULUEE. Furthermore, the conclusions of this paper remain unaffected by this policy shock.

Finally, this paper includes all the policies and the launch of the PDO platform in the same regression for testing. The results in Column (4) of Table 7 show that the coefficient of PDO is significantly positive, while the other three coefficients are either not significant or significantly negative. This suggests that the policy shock of the PDO platform launch is unique and relatively exogenous, and the conclusions drawn in this paper are not influenced by the other confounding policy factors.

4.2.5. Lag Period Test

To address potential robustness issues arising from improper sample period selection, this paper conducted an empirical analysis by lagging the explanatory variable by one period (L.PDO) [62]. The regression results are presented in Column (5) of Table 7. The coefficient of PDO is significantly positive at the 1% significance level, which is consistent with the results from the baseline regression.

5. Further Discussion

5.1. Mechanism Analysis

Through the baseline regression and robustness tests conducted earlier, it was confirmed that PDO can enhance ULUEE. This paper further investigated the mechanisms through which PDO improves ULUEE from three perspectives: innovation effects, agglomeration effects, and structural effects. The innovation effects of PDO on ULUEE were tested using three variables: TPL, IPGs and UMPs. The agglomeration effects of PDO were examined from three dimensions: PA, EA, and IA. The structural effects of PDO on ULUEE were assessed from two perspectives: the level of IS rationalization and the level of IS upgrading.

5.1.1. Innovation Effects

This paper first examines the innovation effects of PDO through urban patent authorizations.

Table 8. Innovation effects test results.

Variable	(1)	(2)	(3)
	TPL	IPGs	UMPs
PDO	0.031 *	0.072 ***	0.096 ***
	(1.826)	(2.898)	(5.021)
Controls	YES	YES	YES
Year FE	YES	YES	YES
City FE	YES	YES	YES
N	2094	2088	2094
R2	0.984	0.973	0.979

Note: t values are in parentheses; ***, **, and * represent 1%, 5%, and 10% significance levels.

presents the regression results for these innovation effects. Columns (1), (2), and (3) display the regression results for TPL, IPGs, and UMPs, respectively. The results indicate that the coefficient for PDO is significantly positive in all cases, suggesting that PDO fosters regional innovation levels through innovation effects, thereby enhancing ULUEE. Consequently, hypothesis H1 is supported.

5.1.2. Agglomeration Effects

Table 9. Agglomeration effects test results.

Variable	(1)	(2)	(3)
	PA	EA	IA
PDO	0.011 ***	0.069 ***	0.169 **
	(2.775)	(4.650)	(2.499)
Controls	YES	YES	YES
Year FE	YES	YES	YES
City FE	YES	YES	YES
N	2088	2088	2088
R2	0.997	0.973	0.943

Note: t values are in parentheses; ***, **, and * represent 1%, 5%, and 10% significance levels.

presents the regression results for testing the agglomeration effects, with IA serving as a mediating variable. Column (1) shows the results with PA as the mediating variable, Column (2) with EA, and Column (3) with IA. From Columns (1), (2), and (3), it can be observed that the regression coefficients for PDO are positive at various significance levels, confirming that PDO enhances ULUEE by fostering urban IA. Therefore, hypothesis H2 is supported.

5.1.3. Structure Effects

From a dynamic perspective, IS upgrading includes two stages: IS rationalization and upgrading. In this paper, the Theil index (RIS) of urban industry is selected to measure the rationalization level of urban industry, and the specific calculation formula is as follows:

$${\mathrm{R}\mathrm{I}\mathrm{S}}_{\mathrm{i},\mathrm{t}}={\sum }_{\mathrm{m}=1}^3{\mathrm{y}}_{\mathrm{i},\mathrm{m},\mathrm{t}}\ln \left({\mathrm{y}}_{\mathrm{i},\mathrm{m},\mathrm{t}}/{\mathrm{l}}_{\mathrm{i},\mathrm{m},\mathrm{t}}\right),\mathrm{m}=\mathrm{1,2},3$$

In the above, ${\mathrm{y}}_{\mathrm{i},\mathrm{m},\mathrm{t}}$ represents the proportion of the industry m in the region i to the GDP in the period t, and ${\mathrm{l}}_{\mathrm{i},\mathrm{m},\mathrm{t}}$ represents the proportion of the industry m in the region i to the total employment in the period t. The lower the RIS value is, the higher the rationalization level of urban IS is.

The urban industrial structure hierarchy coefficient (UIS) is selected to measure the level of IS upgrading. The specific calculation formula is the following:

$${\mathrm{U}\mathrm{I}\mathrm{S}}_{\mathrm{i}.\mathrm{t}}={\sum }_{\mathrm{m}=1}^3{\mathrm{y}}_{\mathrm{i},\mathrm{m},\mathrm{t}}\times \mathrm{l}{\mathrm{p}}_{\mathrm{i},\mathrm{m},\mathrm{t}},\mathrm{m}=\mathrm{1,2},3$$

In the formula, ${\mathrm{y}}_{\mathrm{i},\mathrm{m},\mathrm{t}}$ represents the proportion of the industry m in the region i to the GDP in the period t. The index reflects the evolution of the proportion of the three major industries in China from the dominant position of the primary industry to the dominant position of the secondary industry and the tertiary industry. $\mathrm{l}{\mathrm{p}}_{\mathrm{i},\mathrm{m},\mathrm{t}}$ represents the labor productivity of the industry m in the region i in the period t, and the calculation formula is

$$\mathrm{l}{\mathrm{p}}_{\mathrm{i},\mathrm{m},\mathrm{t}}={\mathrm{Y}}_{\mathrm{i},\mathrm{m},\mathrm{t}}/{\mathrm{L}}_{\mathrm{i},\mathrm{m},\mathrm{t}}$$

In the above, ${\mathrm{Y}}_{\mathrm{i},\mathrm{m},\mathrm{t}}$ represents the added value in the period t of the industry m in the region i, and ${\mathrm{L}}_{\mathrm{i},\mathrm{m},\mathrm{t}}$ represents the employment in the period t of the industry m in the region i. The larger the UIS value is, the higher the level of urban IS upgrading is, and the regression test was carried out, respectively.

The regression results are presented in

Table 10. Structure effects test results.

Variable	(1)	(2)
	RIS	UIS
PDO	−0.031 *	0.056 ***
	(−1.787)	(4.239)
Controls	YES	YES
Year FE	YES	YES
City FE	YES	YES
N	2088	2088
R2	0.523	0.811

Note: t values are in parentheses; ***, **, and * represent 1%, 5%, and 10% significance levels.

. It can be observed that the coefficient of PDO in Column (1) is significantly negative, while the coefficient of PDO in Column (2) is significantly positive. This suggests that PDO promotes IS transformation and upgrading by enhancing both the rationalization and upgrading levels of urban industries, thereby improving ULUEE. Therefore, hypothesis H3 is supported.

5.2. Heterogeneity Analysis

Through the previous analysis and verification, it was demonstrated that PDO effectively enhances ULUEE, with urban technological innovation and IA serving as mediating factors. Furthermore, IS transformation and upgrading exhibit a moderating effect. This paper further investigates whether PDO has varying impacts on the improvement of ULUEE, depending on differences in urban financial development, geographical location, digital infrastructure levels, and environmental regulations.

5.2.1. Differences in the Level of Financial Development

This paper used the ratio of year-end financial deposits and loans to regional GDP as a measure of urban financial development. The cities were divided into two groups based on the median level of financial development: one group with lower financial development and the other with higher financial development. Regression tests were conducted separately for these two groups, and the results are presented in

Table 11. Financial development level heterogeneity test results.

Variable	(1)	(2)
	ULUEE	ULUEE
PDO	−0.005	0.029 ***
	(−0.417)	(2.799)
Controls	YES	YES
Year FE	YES	YES
City FE	YES	YES
N	910	1100
R2	0.725	0.826
Difference test of regression coefficient between groups	Chi2(1) = 6.82
	Prob > Chi2 = 0.0001

Note: t values are in parentheses; ***, **, and * represent 1%, 5%, and 10% significance levels.

. Column (1) displays the results for the group with lower financial development, where the PDO coefficient is not statistically significant. In contrast, Column (2) presents the results for the group with higher financial development, where the PDO coefficient is significantly positive. This suggests that in cities with higher levels of financial development, PDO has a more pronounced positive effect on improving ULUEE [63]. Cities with higher financial development tend to have stronger incentives for industrial optimization [64] and greater levels of financial technology innovation, contributing to a more mature financial sector [65]. These factors enhance the impact of IA, which in turn optimizes the land use structure, facilitating improvements in ULUEE. Furthermore, a joint significance test for the interaction terms between financial development level and PDO was conducted. The results show that, at a 1% confidence level, there are significant differences in the regression coefficients between the two sub-samples.

5.2.2. Differences in Urban Geographical Location

This paper divided cities into three groups based on their geographic location, with eastern, central, and western regions, and conducted separate regression tests. The regression results are presented in Columns (1), (2), and (3) of

Table 12. Urban geographical location heterogeneity test results.

Variable	(1)	(2)	(3)
	ULUEE	ULUEE	ULUEE
PDO	0.019 *	0.002	−0.007
	(1.786)	(0.158)	(−0.462)
Controls	YES	YES	YES
Year FE	YES	YES	YES
City FE	YES	YES	YES
N	864	747	486
R2	0.812	0.728	0.692
Difference test of regression coefficient between groups	Chi2(1) = 6.66
	Prob > Chi2 = 0.0002

Note: t values are in parentheses; ***, **, and * represent 1%, 5%, and 10% significance levels.

. In Column (1), the PDO coefficient is significantly positive, while in the other two columns, it is not statistically significant. This indicates that PDO significantly enhances ULUEE in eastern cities. A joint significance test for the interaction terms between geographic location groupings and PDO reveals significant differences in the regression coefficients between the sub-samples, at least at the 1% confidence level. Eastern cities, with a strong economic foundation and high levels of IA, benefit substantially from PDO, which further reinforces the agglomeration effect. Additionally, ULUEE in the eastern regions is generally higher than in the central and western regions, demonstrating a clear “high–high” agglomeration pattern. PDO, by providing more accurate market information and improving resource allocation, optimizes the input–output ratio of land use, thereby further enhancing ULUEE. Furthermore, the eastern regions possess significant advantages in technology and innovation capacity, and PDO integrates more effectively with these technological tools to boost ULUEE [66]. Therefore, the eastern regions possess stronger capabilities in leveraging data elements, enabling them to optimize land use planning and management through data-driven approaches. Furthermore, the joint significance test for the interaction terms between the groupings and PDO reveals significant differences in the regression coefficients between the sub-samples, at least at the 1% confidence level.

5.2.3. Differences in Digital Infrastructure Levels

This paper measured the level of urban digital infrastructure construction using the frequency of related terms in local government reports, categorizing cities into two groups based on the median: one with lower levels of digital infrastructure construction and one with higher levels. The regression results are presented in

Table 13. Digital infrastructure construction heterogeneity test results.

Variable	(1)	(2)
	ULUEE	ULUEE
PDO	0.025 **	0.007
	(2.408)	(0.624)
Controls	YES	YES
Year FE	YES	YES
City FE	YES	YES
N	1024	1025
R2	0.788	0.780
Difference test of regression coefficient between groups	Chi2(1) = 2.86
	Prob > Chi2 = 0.0359

Note: t values are in parentheses; ***, **, and * represent 1%, 5%, and 10% significance levels.

. Column (1) shows the results for the lower digital infrastructure construction group, where the PDO coefficient is significantly positive. Column (2) presents the results for the higher digital infrastructure construction group, where the PDO coefficient is not statistically significant. This suggests that in regions with lower levels of digital infrastructure, PDO significantly enhances ULUEE. A joint significance test for the interaction terms between urban digital infrastructure levels and PDO revealed significant differences in the regression coefficients between the sub-samples, at least at the 5% confidence level. Regions with lower levels of digital infrastructure often face information asymmetry issues, leading to inefficient land resource allocation [67]. Digital infrastructure, as a fundamental supply-side condition, plays a crucial role in economic development. PDO can provide multidimensional data, such as those on land markets and planning policies, acting as a substitute for digital infrastructure. This helps local governments and market participants make more precise decisions, thereby improving ULUEE. Consequently, in cities with poorer digital infrastructure, the openness of public data can better mobilize scarce resources on the supply side, thereby stimulating ULUEE to a greater extent.

5.2.4. Differences in the Degree of Environmental Regulation

This paper measured the degree of regional environmental regulation from two perspectives. First, the degree of environmental regulation was assessed based on the frequency of environmental protection-related terms in local government reports, with grouping performed according to the median. The regression results are presented in Columns (1) and (2) of

Table 14. Environmental regulation degree heterogeneity test results.

Variable	(1)	(2)	(3)	(4)
	ULUEE	ULUEE	ULUEE	ULUEE
PDO	0.035 ***	0.007	0.064 ***	0.004
	(3.443)	(0.687)	(5.541)	(0.430)
Controls	YES	YES	YES	YES
Year FE	YES	YES	YES	YES
City FE	YES	YES	YES	YES
N	1024	1041	924	1126
R2	0.808	0.786	0.784	0.757
Difference test of regression coefficient between groups	Chi2(1) = 4.25		Chi2(1) = 17.46
	Prob > Chi2 = 0.0053		Prob > Chi2 = 0.0000

Note: t values are in parentheses; ***, **, and * represent 1%, 5%, and 10% significance levels.

. Second, the degree of environmental regulation was measured using the local government’s annual fiscal expenditure on environmental protection, with grouping also based on the median. The regression results for this approach are shown in Columns (3) and (4) of Table 14. According to the regression results, the coefficients for PDO in Columns (1) and (3) are significantly positive, while the PDO coefficients in Columns (2) and (4) are not statistically significant. This suggests that in regions with lower levels of environmental regulation, PDO significantly enhances ULUEE. A joint significance test for the interaction terms between the two groupings of urban environmental regulation levels and PDO revealed significant differences in the regression coefficients between the sub-samples, at least at the 1% confidence level. The possible explanation for these heterogeneous results is that in regions with higher levels of environmental regulation, the efficiency of land resource allocation is already relatively high. In contrast, in regions with lower levels of environmental regulation, PDO can improve land resource allocation efficiency by enhancing the government’s regulatory capacity over land use through big data. This compensates for the deficiencies in environmental regulation, thereby significantly improving ULUEE.

6. Conclusions and Recommendations

6.1. Conclusions

Under the dual-carbon goal, countries are actively promoting green and low-carbon transformation, emphasizing both economic growth and the efficient utilization of resources while prioritizing environmental protection. In particular, achieving a balance between economic development and environmental sustainability in land resource management has become a critical issue. Using a sample of 294 prefecture-level cities from 2014 to 2022, this paper examines the ecological value release potential of public data in the context of land use and explores its realization pathways from the perspective of PDO. The results indicate that PDO significantly improves ULUEE, primarily through innovation effects, agglomeration effects, and structural effects. PDO contributes to the development of technological platforms, providing enterprises with more comprehensive data support and decision-making foundations [68], thereby driving the creation of innovative products and services. Additionally, PDO facilitates the formation and expansion of industrial clusters. Within these clusters, enterprises can share data related to land markets, supply and demand information, and policy planning, reducing information asymmetry and enhancing resource allocation efficiency. Furthermore, industrial clusters leverage synergy effects and optimized factor flows to generate economies of scale and network externalities [69], thereby fostering the more efficient utilization and sustainable development of land resources.

PDO also effectively guides industrial structure adjustment, promotes the efficient circulation of resources, and facilitates the rational development of industries. By providing real-time and accurate information to market participants, public data enable the flow of capital, technology, and talent toward industries with high economic benefits and significant development potential. This promotes the transformation of industries from low-value-added to high-value-added sectors, supports the growth of emerging industries, reduces entry barriers and risks, and accelerates the process of industrial upgrading. Furthermore, the release of the ecological value of PDO is contingent upon a robust urban economic foundation. The presence of necessary conditions in technology, industry, innovation, finance, and governance allows public data to function more efficiently, thereby enhancing ULUEE. By providing high-quality data resources, fostering technology application and innovation, improving scientific decision-making capabilities, and promoting public participation, PDO not only addresses technological and governance deficiencies but also offers a new pathway and potential for enhancing ULUEE. In a broader sense, this paper provides new theoretical support and practical insights for understanding the management and optimal allocation of land ecological resources in the digital era. However, it does not fully explore the specific implementation differences and effects of PDO in various countries, nor does it consider the potential impact of data privacy and security concerns on the improvement of ULUEE. Furthermore, the SBM-Undesirable model imposes stringent requirements on the completeness and precision of input and output data. In assessments of ULUEE, the presence of incomplete data or statistical biases in metrics related to undesirable outputs—such as ecological harm or pollution emissions—can result in inflated or deflated efficiency estimates. Additionally, this model overlooks critical factors like the spatial interdependence and ecological spillover effects associated with land use efficiency, thereby introducing certain constraints to its applicability.

6.2. Policy Recommendations

Further PDO should be promoted. The government should establish multi-dimensional and unified data openness and management platforms, strengthen data standardization, and ensure the interconnection and interoperability of data across industries, departments, and regions. This will foster the creation of an efficient data-sharing ecosystem. A comprehensive legal framework for PDO should be introduced to regulate the collection, storage, sharing, and use of data, ensuring the effective protection of sensitive information, such as personal privacy and commercial secrets, while facilitating the circulation of data. This approach will help mitigate the risks associated with data openness. Additionally, the government should encourage the involvement of social capital and market forces in the development and application of public data. Expanding the coverage and accessibility of public data will provide strong data support for innovation effects, agglomeration effects, and structural effects, thereby contributing to the sustainable development of the economy.

The release of the ecological value of public data should be accelerated. On the one hand, with regard to the accessibility of public data, the government should ensure that public data are more standardized, simplify the access process for businesses and the public, promote widespread data sharing and efficient utilization, and encourage enterprises and research institutions to leverage public data for innovative applications. On the other hand, in terms of the conversion rate of public data, the government should promote the integration of public data with advanced technologies, such as artificial intelligence and big data, to enhance their application value in land resource management and environmental protection. By optimizing cross-departmental cooperation mechanisms and fostering data collaboration among government agencies, a collective effort can be formed to improve both the efficiency and the ecological value of public data utilization.

Using big data and artificial intelligence technology, environmental data can be deeply analyzed and predicted. The government should further integrate multi-dimensional data, including air quality monitoring, pollution source emissions, meteorological changes, and ecological environment quality. By combining historical data with real-time monitoring data for comprehensive analysis, the government can achieve more efficient pollution prevention and environmental governance. Additionally, concerning the reuse of environmental data, the government should, on the one hand, strengthen the real-time release of pollution source emissions, emission data, and ecological monitoring results. This will ensure the standardization and accessibility of data, enabling the public, media, and social organizations to effectively supervise enterprise behavior and fostering greater attention and participation from all sectors of society in environmental issues. On the other hand, the government should encourage enterprises and local governments to utilize big data, artificial intelligence, and other advanced technologies for the in-depth analysis and early warning of environmental data, thereby improving the accuracy and timeliness of environmental governance.

The government’s redistributive and overall planning roles should be given full priority. Collaboratively, the government should cooperate with social organizations to design inclusive data toolkits and, in combination with government subsidies, guide enterprises to enhance the efficiency of public data utilization. Simultaneously, a policy incentive mechanism linking ecological compensation with data application should be established. Furthermore, the government should leverage its redistributive role to promote the transfer of experience and technical assistance among regions through inter-city data alliances, thereby providing data empowerment to technologically disadvantaged areas. Vertically, the central government should establish an open data platform featuring low access thresholds, systematically integrate public data such as those on ecological land, pollution monitoring, and traffic flow in a phased manner, and utilize visualization tools to streamline data analysis. Upon this foundation, local governments should fully harness the green potential of public data. Specifically, the central government can reduce technological barriers for disadvantaged regions by implementing mobile-friendly interfaces, establishing offline data service stations, and organizing community workshops. Meanwhile, local governments should plan the ecological construction of urban land in alignment with local infrastructure conditions.