The Impact of Information Consumption Pilot Policy on Urban Land Green Use Efficiency: An Empirical Study from China

Abstract

Information consumption has been reshaping the modes of human living and production, and driving the transformation of production and trade activities traditionally dependent on land resources, thus influencing urban land green use efficiency (ULGUE). Based on the panel data of 281 prefecture-level cities in China from 2011 to 2023, this study employs the national Information Consumption Pilot Policy (ICPP) as a quasi-natural experiment and utilizes a double machine learning model to assess the ICPP’s impacts on ULGUE. According to the results of the causal mediating effect analysis, the ICPP has improved ULGUE through three mediating mechanisms: expanding the scale of digital transactions, nurturing future industrial developments, and promoting green consumption behaviors. Moreover, in light of the results of the heterogeneity analysis, the ICPP’s impacts on ULGUE vary significantly. Such variation can primarily be attributed to differences in urban resource endowments, disparities in transportation infrastructure development, and variations in geographical location. Specifically, the ICPP has produced more prominent impacts on enhancing land green use efficiency in resource-based cities, cities with high-speed rail access, and coastal cities. Therefore, the government should proactively establish an urban information consumption environment, enhance the role of digital transactions, strategize future industrial developments, encourage green consumption behaviors, and differentiate local policies to effectively promote the continuous improvement of ULGUE.

1. Introduction

According to the World Urban Report published by the United Nations Human Settlements Programme (UN-Habitat) in 2024, global urban green space decreased by 5.6% between 1990 and 2020. Climate change and extensive urbanization are the two primary drivers behind the inefficient use of land and the degradation of the ecological environment [1,2]. The 2030 Agenda for Sustainable Development serves as a foundational framework for cities’ initiatives to implement the Sustainable Development Goals (SDGs) and tackle climate change [3]. For developing countries, such as China, India, and Brazil, the accelerated urban expansion has not only continuously increased the demand for land resources but also exacerbated issues such as arable land loss and ecological degradation [4,5]. Meanwhile, the traditional extensive land use model results in high energy consumption and pollution emissions, which conflicts with the “Dual Carbon Goals” [6,7,8]. From 2006 to 2019, China’s urban construction land area expanded from 3,176,600 hectares to 5,830,800 hectares, representing an increase of 83.55% with an average annual growth rate of 5.66%, whereas the proportion of green space increased marginally from 9.9% to 11.43%. There remains a significant distance to achieving Sustainable Development Goal 11 (SDG 11) by 2030, which focuses on sustainable cities and communities. In contrast, several developed countries have implemented a range of policy measures to enhance land use efficiency while promoting sustainable development. For instance, the United States enacted the Superfund Act to remediate industrially contaminated lands, thereby enhancing land reuse value while mitigating environmental pollution [9]. Germany established the Rhine-Ruhr metropolitan region and integrated industrial and ecological spaces through cross-administrative planning, hence achieving a 30% improvement in land use efficiency [10]. By comparison, to ensure China’s modernization development, how to improve urban land use efficiency while ensuring sustainable development remains a critical priority.

Faced with the challenge of balancing land use and sustainable development, the Chinese government has implemented a number of targeted measures. The Outline of the 14th Five-Year Plan (2021–2025) explicitly emphasizes the need to “enhance the efficiency of natural resource utilization and promote green, sustainable land management”. In line with this, Chinese governments at all levels have been prioritizing incorporating ecological conservation, industrial growth, and urban land use planning to prevent uncontrolled expansion [11]. Such moves closely align with the connotations of urban land green use efficiency (ULGUE), which denotes the optimal ratio between input factors (such as land and other resources) and land use output (encompassing economic, social, and ecological environment dimensions) within the land use system under specific production technology conditions [12]. Unlike prior studies on land use efficiency, which mostly focused on single-dimensional assessments [13], ULGUE highlights the integration of the three subsystems—“economy, society, and ecological environment” [14]—hence embedding the concept of green development throughout the process of land use to achieve the harmonization of economic, social, and ecological benefits [15]. Clearly, enhancing ULGUE is crucial for optimizing territorial space development patterns while achieving sustainable development, making it a focal point of academic research at the intersection of land economics and environmental economics.

With the rapid development of cloud computing, the Internet of Things, and other digital services, microeconomic activities including production, consumption, and distribution have been undergoing gradual transformation [16]. In this context, information consumption has permeated human life at an unprecedented pace [17]. Scholars have increasingly recognized information consumption as a critical approach to enhancing environmental quality [18,19] and driving urban green development [20]. More importantly, among other effects, information consumption plays a crucial role in transforming transactional modes, reshaping consumer perceptions, and supervising enterprise production processes, thereby triggering significant changes in people’s social and economic behaviors [21,22,23]. These transformations have important implications for urban land planning and sustainable development, which in turn influence ULGUE. In order to promote information consumption, China’s central government, the State Council, deployed information consumption for the first time in 2013. Meanwhile, the Ministry of Industry and Information Technology (MIIT) approved the first batch of 68 cities, counties, and districts as the pilot cities for information consumption in the same year. In 2015, MIIT continued to select the second batch of 36 national information consumption pilot cities, counties, and districts to promote the construction of information consumption pilot areas through project support and other ways to explore new models and new paths for the development of a digital economy. Figure 1 shows the geographical distribution of two rounds of information consumption pilot regions. In March 2021, China’s Outline of the 14th Five-Year Plan (2021–2025) emphasized the active cultivation of new types of consumption such as information consumption, which has been regarded as an important starting point to enable the transformation and upgrading of traditional industries and the birth of new industries, new business forms, and new development models.

However, no existing research has explicitly examined the causal relationship between the Information Consumption Pilot Policy (ICPP) and ULGUE, and their mediating mechanisms remain to be explored. To address such gaps, this study employs the panel data from 281 prefecture-level cities in China spanning from 2011 to 2023, aiming to investigate, for the first time, the impact mechanism through which the ICPP influences ULGUE. The potential contributions of this study are as follows. First, regarding research focus, current studies have mostly centered on the impacts of environmental policies (low-carbon city pilots, environmental protection supervision) or technical factors (digital technology, green technology) on ULGUE. By contrast, this study, for the first time, incorporates the ICPP, derived from digitalization, into the analytical framework. By measuring ULGUE in recent years, it reveals causal relationships and analyzes the heterogeneity in the effects of the ICPP on ULGUE across different dimensions, such as resource endowment, transportation infrastructure, and geographical location. This finding provides useful references for the governments to formulate relevant land policies. Second, from a research perspective, this study integrates two national strategies of “Digital China” and the “Dual Carbon Goals”, proposing the theoretical framework of “information consumption enabling green land use”. It examines the influence pathways of ULGUE through the lenses of transactional modes, industrial development, and consumption behavior. Third, concerning research methods, since the introduction of the ICPP, its economic and environmental effects have garnered significant attention from many scholars, who primarily assessed using traditional methods like multi-phase difference-in-differences (DID), generalized DID, and PSM-DID. In this study, we apply double machine learning to evaluate the policy’s effectiveness, leveraging its methodological advantages such as more flexible robustness testing and semi-parametric estimation to enhance the scientific rigor and accuracy of our research findings.

2. Literature Review

2.1. Research on Information Consumption and the ICPP

Early relevant studies primarily centered on theoretical discussions, which encompassed the connotation of information consumption [21], development conditions [24], and realistic logic [25]. In the subsequent mid-stage, empirical investigations into information consumption began to emerge. Despite the absence of an internationally standardized definition or statistical framework for the concept of information consumption, some scholars adopted ICT (information and communication technology) as a research focus, examining its effects on the income levels of developing countries as well as its influence on the economic growth of Russia and other Eastern European nations [26,27]. Gradually, ICT has been increasingly integrated into the field of environmental economic research. Based on the environmental Kuznets curve hypothesis, existing studies have demonstrated that ICT exhibits an inverted U-shaped relationship with carbon dioxide emissions when using global countries as research samples [28]. Additionally, other scholars, focusing on less developed nations such as Tunisia, argued that ICT can mitigate carbon emissions by enhancing total factor productivity [29]. Following the implementation of the ICPP, many Chinese scholars have recently conducted evaluations of the policy effects related to information consumption pilots. This policy exhibits notable economic impacts. At the macrolevel, the ICPP enhances urban innovation capabilities from both demand and supply perspectives [30], facilitates the optimization of industrial structures [31], and drives high-quality economic development via three primary channels: promoting employment, upgrading industries, and fostering innovation [32]. At the microlevel, the ICPP is considered a reflection of digital transformation within enterprises, which increases employees’ labor income share [33]. Additionally, the ICPP also demonstrates environmental benefits by reducing urban carbon emissions through promoting green technological advancements [18] and production factor agglomeration [19].

2.2. Research on ULGUE

ULGUE’s relevant research originated from the study of land use efficiency. Early scholars adopted a single index evaluation method to consider urban land use efficiency and measured the gross product of urban unit construction land area [34]. In order to overcome the limitation of single index measurement, some scholars used two-stage DEA to measure urban land use efficiency [13]. Subsequent studies have predominantly concentrated on countries such as Ethiopia, Vietnam, China, and those in Europe, examining the influencing factors of land use efficiency from multiple perspectives, including urban landscape patterns, land-leasing policies, population dynamics, and economic structural transformations [35,36,37,38]. With the deepening understanding of sustainability, the concept of ULGUE has emerged. Some scholars have started to incorporate factors such as environmental pollution and carbon emissions as undesirable outputs in the assessment of land use efficiency [39], aligning with the concept of ULGUE presented in this study [40]. On the basis of defining the ULGUE concept, clarifying its measurement method, and enhancing dynamic evolution analysis [41,42], numerous scholars have initiated discussions on the antecedents influencing ULGUE. At the policy level, the impact of existing policies related to carbon emission trading, open government data systems, the establishment of national big data pilot zones, and innovative city pilot policy on ULGUE has been evaluated [43,44,45,46]. At the factor level, existing studies have examined the promoting effect of the digital economy on ULGUE through three mechanisms: scale effect, technical effect, and structural effect [47]. Additionally, some studies proposed an inverted U-shaped relationship between urban expansion and ULGUE [48]. Notably, certain research focuses on urban agglomerations in China as samples, arguing that government intervention moderates the ULGUE gap among cities within these agglomerations, thereby narrowing it [49].

2.3. Literature Review Summary

Existing studies offer crucial theoretical foundations for understanding the relationship between the ICPP and ULGUE. However, there remain gaps in the following three areas that warrant further investigation. First of all, the existing literature still lacks a comprehensive and systematic analysis of the environmental implications of the ICPP, particularly their potential contributions to the green transformation of land use. While the economic impacts of the ICPP have been extensively validated, limited research has delved into how the ICPP reshapes microeconomic behavior and thereby influences the sustainability of urban land use. As a core driver of digital transformation, the ICPP may facilitate the transition of land use from “extensive expansion” to “intensive efficiency enhancement” via mechanisms such as technology spillover, resource allocation optimization, and green demand backforcing. However, these mechanisms remain undertheorized and empirically unverified. Secondly, concerning the drivers of ULGUE, the majority of existing studies emphasize the direct effects of traditional policy instruments or technological factors, yet there is a notable deficiency in addressing digitally derived policies. The ICPP exhibits the dual characteristics of “demand-side stimulation” and “supply-side reform”, potentially influencing the land system via industrial digitization, green consumption, and spatial governance coordination. Nevertheless, the extant literature has not adequately elucidated the causal mechanisms, boundary conditions, and heterogeneous impacts of such policies on ULGUE, leading to a paucity of precise theoretical guidance for practical policy implementation. Finally, there are obvious limitations in the empirical methods. Existing studies mainly rely on traditional econometric models to evaluate policy effects, but these methods are not robust enough to deal with high-dimensional data, transmission mechanisms, and nonlinear relationships in prefecture-level cities. To sum up, the existing research gap echoes the marginal contribution of this research, and it is urgent to enrich the assessment of the effect of the ICPP and the antecedents of ULGUE through theoretical demonstration and empirical analysis.

3. Theoretical Analysis and Research Hypothesis

3.1. Direct Impact of the ICPP on ULGUE

According to Feder’s research [50], the two-sector model, which includes the information product sector and the non-information product sector, is integrated into the extended Solow growth model. This theoretical framework explores the synergistic pathway for economic growth and reducing pollution per unit of land area, thereby enabling an assessment of the impact of the ICPP on ULGUE [51]. It is assumed that the production sectors involved in economic activities consist of information production sectors and non-information production sectors, which are responsible for the production of information goods and non-information goods, respectively. The consumer’s utility function U follows the Cobb–Douglas functional form, expressed as:

$$U=Q_I^{\xi }{Q}_R^{\phi }$$

(1)

In Equation (1), the consumption Q_I of information product I and the consumption Q_R of non-information product R depend on residents’ income, and the preference coefficients of consumers for information product I and non-information product R are $\xi$ and $\phi$, respectively, $\xi +\phi =1$.

It is assumed that information goods may be used as an input in the production of non-information goods; that is, there is an externality from the information goods sector to the non-information goods sector. According to Chen’s practice [52], pollution emissions are used as inputs similar to resource factors [53]. The output of the non-information sector product is set as a production function of the capital input, labor input, resource input of the non-information sector, and the output of the information goods sector. The output of the information product sector is set as the production function of capital input, labor input, and resource input, and the technological progress is assumed to be Harrod-neutral, so we obtain:

$$Y_I=Y\left(K_I,A{L}_I,C_I\right)$$

(2)

$$Y_R=Y\left(K_R,A{L}_R,C_R,I\right)$$

(3)

$$K=K_I+K_R, L=L_I+L_R, C=C_I+C_R$$

(4)

where Y_I, A, Y_R, K, and L are the output and technology level of the information product sector, and the output, capital input, and labor input of the non-information product sector, respectively; K_I, L_I, and C_I represent the capital, labor, and natural resources invested in the information product sector, respectively; and K_R, L_R, and C_R represent the capital, labor, and natural resources invested in the non-information product sector, respectively. The introduction of I in the non-information product sector means that the products of the information product sector can be used to increase the output of the non-information product sector, and the products of the information product sector are included in the production function of the non-information product sector as input factors. It is further assumed that the capital, labor, and resources invested in the information product sector are a proportional function of the total input of each type:

$$K_R=\omega K, K_I=\left(1-\omega \right)K$$

(5)

$$L_R=\rho L, L_I=\left(1-\rho \right)L$$

(6)

$$C_R=\theta C, C_I=\left(1-\theta \right)C$$

(7)

where ω, ρ, θ respectively, represent the proportion of various factors invested in the non-information product sector. It is assumed that the information product market is clear; that is, the supply and demand of information products are balanced, and the consumption of information products and information services is equal to the output of the information product sector. The specific production function is as follows:

$$Y_R=K_R^{\alpha }{I}^{\beta }{C}_R^{\gamma }{\left(A{L}_R\right)}^{1-\alpha -\beta -\gamma };0<\alpha +\beta +\gamma <1$$

(8)

$$I=Y_I={\left(K-K_R\right)}^{\lambda }{C}_I^{\eta }{\left[A\left(L-L_R\right)\right]}^{1-\lambda -\eta };0<\lambda +\eta <1$$

(9)

When the economy reaches steady state, it can be obtained that $K=sY-\delta K$. Where s is the saving rate and δ is the depreciation rate. The expressions of the economic growth rates of the two sectors are as follows:

$$g_R=\alpha k_R+\beta I+\gamma C_R+\left(1-\alpha -\beta -\gamma \right)\left(n_R+a\right)$$

(10)

$$g_I=\lambda k_I+\eta C_I+\left(1-\lambda -\eta \right)\left(n_I+a\right)$$

(11)

Under steady-state conditions, if the economy is on a balanced growth path, output Y and capital stock K both grow at the same rate; that is, g = k. The above two equations can be deduced as follows:

$$C_R={\displaystyle \frac{\left(1-\alpha \right)g_R+\left(\alpha +\beta +\gamma -1\right)\left(n_R+a\right)-\beta I}{\gamma }}$$

(12)

$$C_R={\displaystyle \frac{\left(1-\lambda \right)g_I+\left(\lambda +\eta -1\right)\left(n_I+a\right)}{\eta }}$$

(13)

In Equations (10) and (11), g, k, i, c, a, and n are the economic growth rate, capital growth rate, information factor growth rate, natural resource growth rate, technological progress rate, and labor growth rate, respectively. α, β, and γ are the output elasticity coefficients of capital, information factors, and natural resources in the non-information product sector, respectively. λ and η are the output elasticity coefficients of capital and natural resources in the information goods sector, respectively. According to Equation (12), the pollution emission per unit land area of the non-information product sector is negatively correlated with the input of information product factors. It can be seen from Equation (13) that the pollution emission of the information product sector is negatively correlated with the information product factor input. Divide Equations (12) and (13) by C_R and C_I respectively, and after sorting, we can obtain:

$$c{p}_R={\displaystyle \frac{\gamma +\frac{\left(1-\alpha -\beta -\gamma \right)\left(n_R+a\right)+\beta i}{c_R}}{1-\alpha }}$$

(14)

$$c{p}_I={\displaystyle \frac{\eta +\frac{\left(1-\eta -\lambda \right)\left(n_I+a\right)}{c_I}}{1-\lambda }}$$

(15)

$$\mu =f\left(\omega ,\rho ,\theta \right)$$

(16)

$$cp=\left(1-\mu \right)c{p}_R+\mu c{p}_I$$

(17)

In Equation (14), cp is the ratio of economic growth to pollution emission growth per unit land area, reflecting the change in economic growth brought by the increase in pollution emission per unit area. As the value of cp increases, the green land use efficiency of the sector also increases accordingly. In Equation (17), μ is a function of the proportion of input of various factors in the information product sector, and is the proportion of total production factors allocated to information products. The overall green use efficiency of urban land is closely related to μ. Equations (14) and (17) show that with the increase in factor allocation to the information product sector, the green utilization efficiency of urban land in the non-information product sector is improved, and the increase in the input proportion of the information product sector will further improve the overall green utilization efficiency of urban land. The growth of information consumption demand is bound to expand the supply of information products, increase the allocation of resources in the information product sector and improve the green utilization efficiency of land in the non-information product sector, so as to enhance the overall green utilization efficiency of urban land. To sum up, the expansion of information consumption increases the factor input of the information product sector and adjusts the factor allocation between the information sector and the non-information sector, which helps to improve ULGUE. Therefore, the following research hypothesis is proposed:

H1: 

The ICPP can enhance ULGUE.

3.2. Analysis of the Influencing Mechanism of the ICPP on ULGUE

3.2.1. Expand the Scale of Digital Transactions

Through institutional incentives and technical empowerment, the ICPP significantly expands the scale of urban digital transactions. Based on transaction cost theory, the ICPP facilitates the migration of traditional offline transactions to online platforms by reducing information asymmetry and transaction friction [54]. In a policy-driven context, prefecture-level cities can accelerate the construction of digital infrastructure [32], which directly stimulates the growth of transaction scales in e-commerce, online services and other sectors [55]. Specifically, first, in line with the logic of “the ability to pay determines land use” in rent competition theory, the expansion of digital transaction scales induces a structural substitution in land demand. In other words, the increased scale of digital transactions reduces reliance on physical commercial spaces, hence replacing traditional brick-and-mortar shops with online retail and thereby decreasing the demand for commercial land [56]. Second, in accordance with technology diffusion theory, the expansion of digital transaction scales accelerates the digital transformation of land management. As a result, the big data footprint generated by digital exchanges offers decision-making support for smart city planning [57], by monitoring land use efficiency and optimizing land planning and layout via the Internet of Things [58]. Third, the scale of digital transactions drives the intensive transformation of industrial land. According to the theory of economies of scale and scope, digital transactions promote the coordination of the industrial chains by encompassing intelligent manufacturing and cloud storage, while facilitating the transition of industrial land use from extensive to high-density and multi-functional [59]. Last but not least, online trading platforms enable most manufacturing enterprises to enhance capacity utilization rates and increase output value per unit of industrial land through real-time data element sharing [60]. Along the transmission logic of “policy incentive-market response-resource reallocation” of the above mechanisms, this study proposes the following hypothesis:

H2: 

The ICPP enhances ULGUE by expanding the scale of digital transactions.

3.2.2. Nurture Future Industrial Developments

As an institutional tool, the ICPP plays a pivotal role in reconstructing the institutional environment for industrial development. On the one hand, by leveraging innovation ecosystem theory, the pilot policy enhances resource allocation efficiency and offers targeted financial support to guide enterprises in pilot cities toward digital transformation and green transformation [61]. In this way, it breaks the path dependence of traditional industries, and redirects capital and labor toward knowledge-intensive and environmentally friendly sectors [62], thus establishing a solid foundation for future industrial growth. On the other hand, this policy promotes information consumption scenarios such as telemedicine and online education. Drawing on Hicks’ induced innovation hypothesis, such new consumption scenarios redirect technological change [63]. As consumers’ preference for information services grows, enterprises will be induced to allocate R&D resources toward green and digital technology sectors with higher demand elasticity [64], thereby establishing the technical foundations necessary for future industrial development. As a result, the future industrial development can adhere to the principles of green growth and spatial economics, and its technical characteristics and organizational patterns can significantly influence land resource utilization. Specifically, first, the future industrial agglomeration in the city necessitates a specific spatial carrier, and its production processes are frequently integrated into the circular economy model [65]. According to the industrial symbiosis theory, the exchange network of energy and intermediate products among enterprises within the industry can reduce the land requirement per unit of output while simultaneously enhancing green value to achieve closed-loop utilization of land resources [66]. Second, based on the new economic geography theory, the cluster development of future industries will reshape the urban spatial structure via the “center-periphery” model. Information consumption–driven industrial clusters are more likely to concentrate in knowledge-intensive areas, whereas traditional manufacturing clusters tend to be located in energy-intensive areas near transportation hubs [67]. The spatial differentiation of urban industries influences ULGUE through the land competitive rent mechanism. As high value-added industries face higher land rents, they tend to drive functional upgrades and ecological improvements in inefficient lands such as old industrial zones [68], which ultimately leads to the synergistic enhancement of both land economic value and green value. Third, based on the institutional evolution theory, the concept of sustainable development embedded in future industrial frameworks can reshape the public’s understanding of the land’s green values. As clean energy enterprises utilize their factories’ ecological landscapes as brand capital for promotion, the public gradually forms a spatial cognitive preference for “production-ecology” symbiosis [69]. This transformation in public cognition has deeply penetrated the land management system, prompting land planning departments to move beyond merely pursuing real estate indicator equilibrium, but instead to prioritize balancing the dual functions of land-supporting economic development while preserving ecological integrity [70]. Based on this, this study proposes the following hypothesis:

H3: 

The ICPP enhances ULGUE by nurturing future industrial development.

3.2.3. Promote Green Consumption Behaviors

As a pivotal institutional design for promoting the coordinated development of digitalization and greening, the ICPP plays a crucial role not only in expanding the scale of digital transactions and optimizing the industrial structure but also in reshaping the behavioral values and consumption pursuits of social entities via demand-side reforms [71]. The information consumption scenarios can facilitate consumers’ information processing while increasingly directing their attention toward the attribute characteristics of green products [72]. Through such mechanisms as consumption transparency and the cultivation of green preferences, the ICPP can guide the transformation of residents’ consumption behaviors toward greater environmental friendliness [73]. On the one hand, based on the environmental behavior theory, individuals’ consumption choices are influenced not only by price signals but also by their level of awareness regarding environmental externalities. The ICPP can assist cities in enhancing the green product certification system and mandatory disclosure mechanisms, hence reducing the costs for consumers to identify green goods. This will also guide consumers to make green consumption decisions driven by either self-interest or altruism [74]. On the other hand, drawing from the “boost effect” in behavioral economics, the ICPP has given rise to new forms of consumption, such as the sharing economy and green e-commerce [75]. These innovations will subtly steer the public toward adopting a green lifestyle, which in turn creates market demand pressure for sustainable land use [76] and strengthens the public’s perception of the scarcity and ecological value of land resources [74]. With the rising consciousness in the public’s green consumption behaviors, the green transformation of urban land use will be accelerated in the following ways: first, based on the revealed preference theory, consumers demonstrate their implicit preferences for ecological attributes through green consumption behaviors [77], thereby prompting the land market to reassess the ecological value of locations. This demand-side pull influences the land supply side, leading landowners to proactively optimize land use structures [78]. Second, according to the social learning theory, the dissemination of green consumption culture will reshape group behavior patterns, thus creating informal constraints that support green land use practices [79]. The heightened environmental awareness among residents can also foster collective recognition of eco-friendly land use, reduce the costs associated with managing and protecting public green spaces, and indirectly enhance ULGUE [44]. Finally, drawing from the institutional change theory, the deepening of green consumption ideals will lead residents to prioritize ecological values, and shift unit land output from a “growth-first” to an “ecology-first” paradigm. This shift will channel the land management system to adapt to new developmental demands [80], thus ensuring closer alignment with future ecological carrying capacities and human activity intensities. Based on this, this study proposes the following hypothesis:

H4: 

The ICPP enhances ULGUE by promoting public green consumption behavior.

The theoretical mechanism of this research is illustrated in Figure 2. The figure illustrates the primary research objects and theoretical mechanisms of this study. Additionally, it presents the hypotheses associated with each mechanism as well as the methods employed for empirical analysis.

4. Materials and Methods

4.1. Model Setting

4.1.1. Benchmark Regression Model

At present, traditional models such as difference-in-differences are frequently employed in policy effect assessment studies. However, these models suffer from issues such as model specification bias and the constraints of linear hypothesis assumptions. Particularly, when the parallel trend assumption is difficult to satisfy, biased estimates may arise. Moreover, traditional models struggle to address formal uncertainties of confounding factors, the curse of dimensionality, regularization bias, and excessive focus on “consistency”. To compensate for the limitations of conventional econometric models, this study draws inspiration from Chernozhukov et al. to construct a double machine learning model [81]. Double machine learning, combined with the doubly robust estimation method, employs Neyman orthogonal estimating equations for cross-fitting estimation. This approach overcomes many stringent specific assumptions inherent in traditional causal inference models and constitutes a consistent, asymptotically normal, and efficient semi-parametric causal estimation technique. Below is the partial linear model of double machine learning established in this study:

$$ULGU{E}_{it}={\theta }_{0}ICP{P}_{it}+g\left(X_{it}\right)+U_{it}$$

(18)

$$E\left(\left.U_{it}\right|ICP{P}_{it},X_{it}\right)=0$$

(19)

where i denotes the city; t denotes the year; ULGUE_it, as the dependent variable, represents the ULGUE of prefecture-level city i in year t; ICPP_it serves as the independent variable, a policy dummy for the information consumption pilot policy, which equals 1 after the pilot implementation and 0 otherwise; θ₀ is the treatment coefficient in this study; X_it is a set of high-dimensional control variables, requiring the use of a machine learning algorithm to estimate the specific functional form g(X_it); and U_it is the error term, with a conditional mean of zero.

To accelerate the convergence speed and ensure that the treatment coefficient estimators remain unbiased in small sample scenarios, an auxiliary regression is constructed as follows:

$$ICP{P}_{it}=m\left(X_{it}\right)+V_{it}$$

(20)

$$E\left(V_{it}\left|X_{it}\right.\right)=0$$

(21)

Among them, m(X_it) represents the regression function of the treatment variable with respect to the high-dimensional control variables. It is also essential to employ machine learning algorithms to estimate its specific form. V_it denotes the error term, which is independent of U_it and has a conditional mean of zero.

4.1.2. Causal Mediating Effect Model

In order to explore the specific mechanism of the ICPP affecting ULGUE, a mediating effect test is needed. Due to the problems of ignoring endogeneity and relying on strict linear hypothesis preconditions in the traditional three-step model of mediating effect, relying only on traditional exogenous hypothesis conditions is not enough to fully reveal the complete causal recognition mechanism behind it. Therefore, this study constructs a more general recognition method: a causal mediating effect model based on potential results and a counterfactual framework [82]. Different from the traditional three-step mediating effect model, causal mediating analysis aims to decompose the causal effects of independent variable on the dependent variable into indirect effects that pass through the mediator variable and direct effects that do not pass through the mediator variable. The effect decomposition is shown in Equation (22):

$$\Delta =\theta \left(1\right)+\delta \left(0\right)=\theta \left(0\right)+\delta \left(1\right)$$

(22)

The total effect ∆ can be decomposed into the sum of the direct effect θ(1) of the treatment group and the indirect effect δ(0) of the control group, and the sum of the direct effect θ(0) of the control group and the indirect effect δ(1) of the treatment group.

4.2. Variable Selection and Data Source

4.2.1. Dependent Variable

Urban Land Green Use Efficiency (ULGUE). As outlined by Ma [41] and Li et al. [42], ULGUE is assessed using a super-efficient Slacks-Based Measure (SBM) model that incorporates undesirable outputs. This model enables a detailed analysis of how inputs (land, capital, labor) are transformed into expected outputs (economic, social, ecological benefits) while minimizing unexpected outputs (pollutant emissions, carbon emissions). The input–output index system for ULGUE is presented in

Table 1. Measuring indicators of ULGUE.

Indicator Type	Indicator Name	Indicator Connotation	Units	N	Mean	sd	Min	Max	Reference
Input	Land	Area of land used for urban construction	Km2	3653	147.666	173.098	15	895	[41]
	Captial	Total investment in fixed assets	CNY 100 million	3653	1870	1850	48.919	9440	[43,44,45]
	Labor	Employees in secondary and tertiary industries	10,000 person	3653	53.527	60.992	4.750	326.737	[43,44,45]
Expected output	Economic benefits	Value added of secondary and tertiary industries	CNY 100 million	3653	2528.25	3352.536	97.300	18,790.2	[43]
	Social benefits	Per capita disposable income of urban residents	CNY	3653	31,622.61	10,750.74	11,691.1	66,068	[45]
	Ecological benefits	Green covered area as of completed area	%	3653	40.327	6.095	14.33	61.58	[43,44,45]
Unexpected output	Pollutant emissions	Based on entropy method, industrial sulfur dioxide emission, industrial comprehensive calculation of industrial wastewater discharge and industrial smoke and dust discharge	—	3653	0.074	0.071	0.002	0.424	[40,43]
	Carbon emissions	Total CO2 emissions	10,000 t	3653	3380	3050	196.855	15,700	[45]

. The resulting efficiency score reflects each city’s ability to optimize its inputs to enhance expected outputs and reduce unexpected outputs, thereby offering a robust and comprehensive evaluation framework for ULGUE.

Figure 3 shows the spatial distribution of ULGUE for the sample cities in 2011, 2015, 2019, and 2023. In recent years, the overall level of ULGUE in China has shown a steady year-on-year improvement. From a spatial perspective, ULGUE in the eastern coastal regions is significantly higher than that in the western inland areas, exhibiting a pronounced trend of spatial agglomeration. This finding aligns with the results reported in existing literature [8,45]. At the same time, this study reveals that ULGUE is predominantly concentrated in urban agglomerations. Specifically, cities in the Yangtze River Delta, the Beijing-Tianjin-Hebei urban agglomeration, and the Sichuan-Chongqing metropolitan area exhibit relatively higher levels of ULGUE.

4.2.2. Independent Variable

Information Consumption Pilot Policy (ICPP). The years 2013 and 2015 are respectively designated as the time points when the first and second batches of pilot cities began to be influenced by the policy. For a given city, if the pilot policy was implemented after the specified time point, the ICPP value is assigned as 1; otherwise, the ICPP value remains 0 [18,19,20].

4.2.3. Mediator Variables

Digital Transaction Scale (lnExp). With reference to the practice of Fang et al. [83], this study selects the number of express delivery businesses in prefecture-level cities multiplied by the ratio of China’s online retail sales to the number of express delivery businesses to represent the digital transaction scale of each city, and carries out logarithm processing on it. It can be understood as the average amount of online retail goods contained in each express.

Future Industrial Development Level (lnNewl). According to the Catalogue of Industrial Strategic Emerging Industries (2023), A-share listed enterprises in the new generation of information technology, high-end equipment manufacturing, new materials, biology, new energy vehicles, new energy, energy conservation and environmental protection, aerospace, marine equipment, and other industries are classified as future industries. With reference to the research of Wei et al., this study quantifies the artificial intelligence stock of enterprises classified under the future industry at the city level, and carries out logarithm processing on it [84].

The Green Consumption Index (GC). This index is constructed by drawing on the research of Yu et al. [85]. The GC index comprehensively evaluates green consumption in terms of consumer products, behaviors, and outcomes, encompassing key contemporary consumption domains such as clothing, food, housing, usage, and transportation. The entropy weight method is employed to quantify the index.

4.2.4. Control Variables

Control variables account for other common economic, social, technological, industrial, and other factors that may influence ULGUE. At the macroeconomic level, this study selects per capita GDP [46] and financial development level [44] as control variables to represent overall urban economic development, market vitality, and financial market maturity. Specifically, the financial development level is calculated as the ratio of year-end financial institutions’ deposit and loan balances to regional gross product. In terms of social public services, technological progress, and infrastructure construction, this study chooses per capita road area (square meters) [8], internet penetration rate [44], and technological progress [42] as the control variables to reflect spatial accessibility and technological innovation in urban social activities. The internet penetration rate is measured by the number of broadband internet access users per 100 people (households), while technological progress is quantified by the number of invention patents granted at the prefecture-level city. At the population structure level, population size is selected as a control variable to reflect urban labor supply potential [46], measured by the city’s annual average population (in 10,000 persons). Regarding openness, this study adopts the FDI level as a control variable [19], using the ratio of actual foreign investment to GDP to measure globalization participation and technology spillover effects. Finally, logarithmic transformations are applied to the annual average population, per capita GDP, and technological progress.

4.3. Data Sources

In this study, we utilize panel data from 281 prefecture-level cities in China spanning the period of 2011 to 2023 as research samples. A rigorous data cleaning process is conducted, wherein a substantial number of missing observed values are excluded, and a limited number of discontinuous missing data points are imputed using the linear interpolation method. The ULGUE index data and control variable data are sourced from the China Statistical Yearbook, the China City Statistical Yearbook, and the National Economic and Social Development Statistical Bulletins of the respective prefecture-level cities. For indicators involving nominal prices, adjustments have been made using the GDP deflator to express them in terms of comparable prices with a base year of 2003. Data on the size of digital transactions are obtained from the China E-commerce Market Data Monitoring Report and the China Statistical Yearbook. Future industrial development level data are compiled from the China City Statistical Yearbook and the artificial intelligence enterprise database. Green consumption index data are gathered from local government offices, administration bureaus of prefecture-level cities, government service data repositories, industry and information technology bureaus, market supervision bureaus, and consumer associations at the prefecture-level city level. This study conducted empirical analysis using StataMP 18 and R 4.4.1. The descriptive statistics for the aforementioned variables are presented in

Table 2. Summary statistics.

Variable	Symbol	Obs	Mean	Std.Dev	Min	Max
Dependent variable	ULGUE	3653	0.552	0.275	0.015	1.235
Independent variable	ICPP	3653	0.235	0.424	0	1
Control Variables	lnRgdp	3653	10.819	0.561	9.507	12.101
	Fin	3653	2.628	1.182	1.037	6.898
	UI	3653	18.893	7.782	4.863	44.401
	Inter	3653	2.750	1.850	0.308	9.559
	lnTec	3653	5.533	1.773	1.792	10.032
	lnPop	3653	5.900	0.678	3.850	7.255
	FDI	3653	0.016	0.021	−0.029	0.112
Mediator variables	lnExp	3653	14.177	1.688	10.712	18.549
	lnNewl	3653	5.327	1.850	1.792	10.271
	GC	3653	0.533	0.166	0.121	0.887

.

5. Empirical Analysis

5.1. The Baseline Regression Results

Table 3. Model regression results.

Variable	Random Forest (1)	Random Forest (2)	Lasso (3)	Xgboost (4)	Enet (5)
ICPP	0.223 *** (8.952)	0.227 *** (9.378)	0.232 *** (10.934)	0.209 *** (15.741)	0.148 *** (6.437)
Constant	−0.001 (−0.348)	−0.002 (−0.372)	−0.001 (−0.154)	−0.001 (−0.220)	0.003 (0.655)
Control	YES	YES	YES	YES	YES
Control_Squ	NO	YES	YES	YES	YES
City FE	YES	YES	YES	YES	YES
Year FE	YES	YES	YES	YES	YES
N	3653	3653	3653	3653	3653

Note: *** indicates the statistical significance at 1% level.

presents the baseline regression results of this study. In column (1), a double machine learning model is employed with a sample segmentation ratio of 1:4. The random forest algorithm is utilized to examine the impact of the ICPP on ULGUE. As depicted in column (1), after controlling for city fixed effects, time fixed effects, and the primary term of the control variable, the regression coefficient remains significantly positive at the 1% level. This finding supports H1 of this study, which states that the ICPP significantly enhances ULGUE. On the basis of column (1), this study further controls for the quadratic terms of other control variables. As depicted in column (2), the regression coefficient remains significantly positive, with little change in its value. Additionally, this study employs Lasso regression, gradient boosting, and elastic net algorithms based on the partial linear model for regression analysis. The specific regression results are presented in columns (3), (4), and (5) of Table 3. The findings indicate that regardless of the algorithm selected, the regression coefficient of the ICPP is positively significant at the 1% level. This further confirms H1, which states that the ICPP significantly contributed to the improvement of ULGUE. The implementation of the ICPP facilitates factor input in the information product sector and optimizes the allocation of resources between the information and non-information sectors, thereby enhancing ULGUE. Simultaneously, this finding aligns with prior research conclusions that the ICPP contributes to improving the urban ecological environment [18,19].

5.2. Endogeneity Test

Because cities with high green land use efficiency are often economically advanced, characterized by robust technological innovation and effective environmental governance, they tend to proactively compete for or attract national policy resources (e.g., ICPP). This process can lead to the formation of a “Matthew effect”, where advantages accumulate disproportionately. As a result, regression analysis may be prone to encountering endogeneity issues. Accordingly, in this study, based on Chernozhukov et al. [81], a partial linear instrumental variable model of double machine learning was constructed, with specific settings as follows:

$$ULGU{E}_{it}={\alpha }_{0}ICP{P}_{it}+g\left(X_{it}\right)+U_{it}$$

(23)

$$Instrumen{t}_{it}=m\left(X_{it}\right)+V_{it}$$

(24)

where Instrument_it is the instrumental variable for ICPP_it. Here, this study follows Golin et al. [86] to construct an interaction term by multiplying the number of fixed telephones per 100 people in a city with the year dummy variable. This construction satisfies the assumptions of externality and relevance for the instrumental variable. The regression results obtained after resetting the double machine learning model are presented in column (1) of

Table 4. Robustness test results.

Variable	Endogeneity Test (1)	Refine Urban Areas (2)	Interactive Fixed Effect (3)	Exclusion Concurrent Policies (4)
ICPP	1.173 *** (2.741)	0.228 *** (10.163)	0.230 *** (10.716)	0.205 *** (8.536)
Constant	—	−0.003 (−0.685)	0.025 *** (5.920)	−0.002 (−0.470)
Control	YES	YES	YES	YES
Control_Squ	YES	YES	YES	YES
City FE	YES	YES	YES	YES
Year FE	YES	YES	YES	YES
N	3653	3601	3653	3653

Note: *** indicates the statistical significance at 1% level.

. Clearly, the endogeneity issue does not alter the conclusion that the ICPP promotes ULGUE; it only modifies the magnitude of the policy effect to some extent. This sufficiently demonstrates the robustness of the original conclusion. In addition, to address the endogeneity issue, this study employs an event study approach to further examine the parallel trend before the policy implementation and the dynamic effects following the policy. Furthermore, a placebo test was conducted, as illustrated in Figure A1 and Figure A2 in the Appendix A.

5.3. Robustness Test

First, to avoid potential biases in the benchmark regression results caused by the unique administrative status, policy resource endowment, and city size of municipalities directly under the central government, this study excludes the samples of Beijing, Shanghai, Tianjin, and Chongqing and re-conducts the regression analysis. The results are presented in column (2) of Table 4. The positive effect of the ICPP on ULGUE remains significant, and the direction of the coefficient aligns with that of the baseline regression, thereby confirming the robustness of the research conclusion and H1.

Second, given that provincial governments serve as the core administrative level within the national governance system, prefecture-level cities under their jurisdiction often exhibit spatial convergence in terms of institutional environment, geographical conditions, and social capital. To account for the time-varying heterogeneity of provincial administrative units, the model incorporates province-time-interactive fixed effects. The specific regression results are presented in column (3) of Table 4. Based on these results, after considering the correlation among various urban characteristics within the same province, the impact of the ICPP on urban ULGUE remains significantly positive at the 1% level. Thus, the H1 is further confirmed through this test.

Third, when assessing the impact of the ICPP on ULGUE, potential interference from concurrent policies must be considered. To ensure the accuracy of the policy effect estimation, this study controls for other similar policies implemented during the sample period. After 2014, the “Broadband China” strategy was successively launched, and policies related to the ICPP include the construction of a “Smart City” and the establishment of the “National Big Data Comprehensive Pilot Zone”, which were implemented in 2013 and 2015, respectively. In this study, policy dummy variables for “Smart City”, “Broadband China”, and “National Big Data Comprehensive Pilot Zone” are incorporated into the regression analysis. The specific regression results are presented in column (4) of Table 4. After accounting for the influence of these three concurrent policies, the significance of the ICPP remains unchanged, thereby providing robust support for the conclusions of this study.

5.4. Robustness Test of Transformation Model

In order to mitigate the potential influence of bias in the double machine learning model setting on the conclusions, this study further examines the robustness of the findings from the perspective of model transformation.

Firstly, the sample segmentation ratio of the double machine learning model was adjusted from the original 1:4 to 1:2 and 1:7 to investigate the possible impact of varying sample segmentation ratios on the conclusions. As shown in columns (1)–(2) of

Table 5. Robustness test results for the transformation model.

Variable	Change the Sample Segmentation Ratio		Replacement Model		Interactive Model (5)
	Kfolds = 3 (1)	Kfolds = 8 (2)	SVM (3)	Neural Network (4)
ICPP	0.217 *** (10.038)	0.218 *** (8.836)	0.187 *** (18.254)	0.231 *** (10.829)	0.219 *** (39.795)
Constant	−0.002 (−0.385)	−0.002 (−0.435)	0.023 *** (4.967)	−0.001 (−0.155)	—
Control	YES	YES	YES	YES	YES
Control_Squ	YES	YES	YES	YES	YES
City FE	YES	YES	YES	YES	YES
Year FE	YES	YES	YES	YES	YES
N	3653	3653	3653	3653	3653

Note: *** indicates the statistical significance at 1% level.

, the effect of the ICPP on ULGUE remains significantly positive even after altering the sample segmentation ratio, thereby confirming the robustness of the baseline regression results.

Secondly, we incorporate the machine learning algorithms. Building upon the random forest algorithm, as well as the Lasso regression, gradient boosting, and elastic net methods previously employed for benchmark regression predictions, we further introduce the neural network and support vector machine (SVM) prediction algorithms for regression analysis. The regression results are presented in columns (3)–(4) of Table 5. After integrating the neural network and SVM algorithms, the impact of the ICPP on ULGUE remains significantly positive, thereby reinforcing the robustness of the baseline regression results.

Thirdly, a partial linear model is constructed based on double machine learning in the benchmark regression analysis, and the specification of the model form involves a certain degree of subjectivity. Subsequently, this study employs double machine learning to develop a more generalized interactive model, aiming to investigate the influence of model specification on the conclusions of this study. The adjustments in both the main regression and auxiliary regression for analysis are presented as follows:

$$ULGU{E}_{it}=g\left(ICP{P}_{it},X_{it}\right)+U_{it}$$

(25)

$$ICP{P}_{it}=m\left(X_{it}\right)+V_{it}$$

(26)

Finally, the average estimated coefficients of the independent variable obtained through the interactive double machine learning model are as follows:

$$\theta =E\left[g\left(ICP{P}_{it}=1,X_{it}\right)-g\left(ICP{P}_{it}=0,X_{it}\right)\right]$$

(27)

The regression results obtained after resetting the double machine learning model are presented in column (5) of Table 5. Evidently, transitioning the model from a partial linear model to an interactive double machine learning model does not alter the conclusion that the ICPP enhances ULGUE. Instead, it only modifies the magnitude of the policy effect to some extent, thereby further demonstrating the robustness of the baseline regression conclusion.

Fourth, to address estimation bias caused by unbalanced data distribution and limited temporal variability in causal inference models, this study adopts a generalized random forest approach. Unlike conventional causal inference models that concentrate on estimating the average treatment effect, the generalized random forest model enables the assessment of individual-level heterogeneous treatment effects within the sample. The individual average treatment coefficient is calculated as follows:

$$\widehat{\tau }\left(\cdot \right)=\arg {\min }_{\tau }\left\{{\displaystyle {\sum }_{i=1}^n\left[ULGU{E}_i-{\widehat{m}}^{\left(-i\right)}\left(X_i\right)\right]}-\tau \left(X_i\right){\left[ICP{P}_i-{\widehat{e}}^{\left(-i\right)}\left(X_i\right)\right]}^2+{\varpi }_n\left[\tau \left(\cdot \right)\right]\right\}$$

(28)

where $\tau \left(\cdot \right)$ represents the average individual treatment coefficient given the confusion variable X_i; ${\widehat{m}}^{\left(-i\right)}\left(X_i\right)$ indicates the predicted value of the out-of-bag influence effect on ULGUE given the confusion variable X_i; ${\widehat{e}}^{\left(-i\right)}\left(X_i\right)$ reflects the probability of individual acceptance of treatment under the condition of the given confusion variable X_i in out-of-bag samples; and $\varpi \left[\tau \left(\cdot \right)\right]$ stands for the regularization term. On the basis of screening for the importance of confounding variables, Figure 4 illustrates the distribution of treatment effects of the ICPP on ULGUE when the number of trees is set to 500, 2000, 4000, and 8000, respectively. It is evident that there are significant variations in the average treatment effect across individuals. While a small proportion of individuals exhibit treatment effects approaching zero, the average treatment effect for the majority of individuals primarily clusters around 0.21–0.23. This suggests that the ICPP has a consistently positive impact on enhancing ULGUE in most cities.

5.5. Mediating Effect Test

The aforementioned results demonstrate that the ICPP can significantly enhance ULGUE. This study aims to uncover the specific mediating mechanisms through which these policies improve ULGUE. Based on the preceding theoretical assumptions, the expansion of digital transactions facilitated by the ICPP, the promotion of future industrial development, and the encouragement of green consumption behavior constitute three plausible pathways explaining the increase in ULGUE. Therefore, in this section, the causal mediating effect of double machine learning was analyzed by drawing on the research of Farbmacher et al. [82]. Additionally, the mediating mechanism of the ICPP to ULGUE was tested using the random forest method. The specific test results are presented in

Table 6. Regression results of the mediating effect.

Variable	Total Effect (1)	Treatment Group Direct Effect (2)	Control Group Direct Effect (3)	Treatment Group Indirect Effect (4)	Control Group Indirect Effect (5)
lnExp	0.228 ***	0.223 ***	0.210 ***	0.018 ***	0.004 ***
lnNewI	0.220 ***	0.220 ***	0.202 ***	0.018 ***	−0.001
GC	0.219 ***	0.220 ***	0.202 ***	0.019 ***	−0.000

Note: *** indicates the statistical significance at 1% level.

. It is evident that the total effect under different mediating paths remains significantly positive at the 1% level, thereby reinforcing the conclusion that the ICPP positively enhances ULGUE.

Specifically, first, the ICPP is tested to influence the mediating mechanism of ULGUE by expanding the scale of digital transactions. As shown in the second row of Table 6, the indirect effect of the digital transaction scale for both the treatment group and the control group is significantly positive at the 1% level, while the direct effect of both groups is also positive and has passed the significance test. This indicates that the ICPP can effectively enhance ULGUE by enlarging the scale of digital transactions, thereby supporting H2. The results show that the ICPP significantly reduces the dependence of traditional offline transactions on physical commercial space by promoting digital transaction scenarios, and enhances the green and intensive use of land [56]. According to the indirect effect regression results, the scale expansion of digital transactions in pilot cities of information consumption is greater. Although other cities are not pilot for information consumption, ULGUE can still be improved by expanding the scale of digital transactions, because under the era of digital transformation, online digital transactions have become an inevitable trend in most cities, and the green use of land has been improved.

Second, the ICPP influences the mediating mechanism of ULGUE by promoting future industrial development. As indicated in the third row of Table 6, both the treatment group and the control group exhibit a significantly positive direct effect on the future industrial level at the 1% significance level. Meanwhile, the indirect effect of the treatment group is positive, whereas the indirect effect of the control group is not significant. These results demonstrate that the ICPP has effectively enhanced ULGUE by facilitating future industrial development. The H3 is confirmed. The results indicate that the ICPP alters the allocation of land resources from a “space container” to a “value carrier” via institutional incentives. Currently, as the land use in some traditional industrial parks in China is gradually transitioning into experimental zones for future industrial innovation, despite no expansion in land scale, the economic output per unit of land and ecological benefits have simultaneously improved through the introduction of knowledge-intensive industries [87]. Based on the indirect effect results of future industrial development, the information consumption pilot cities exhibit a more complete system for future industrial development and a more pronounced transformation effect regarding land green use. In contrast, the future industrial development in non-information consumption pilot cities remains at a weaker stage, with less noticeable effects in promoting ULGUE.

Third, the ICPP is examined for its impact on the mediating mechanism of ULGUE by promoting green consumption. As shown in the fourth row of Table 6, the direct effect of green consumption is significantly positive at the 1% level in both the treatment and control groups. However, the indirect effect is positive in the treatment group but insignificant in the control group. This indicates that the ICPP can enhance ULGUE and support H4 by promoting green consumption. The aforementioned results indicate that the ICPP provides clearer product information to the public, accelerates the promotion of green consumption, and fundamentally reshapes the ecological value cognition of the “human–land” relationship [74]. When consumption behavior is imbued with the attribute of ecological responsibility, commercial land planning transitions from prioritizing the maximization of economic benefits to aligning with ecological capacity. Based on the indirect effect regression results of green consumption behavior, the level of green consumption practice in information consumption pilot cities is higher, making it easier to drive the green transformation of urban land use. Conversely, in non-information consumption pilot cities, the transmission function of product information attributes is relatively weak, leading to limited public attention to the ecological attributes of products and green consumption, which consequently has no significant impact on enhancing ULGUE.

In addition, techniques such as cross-fitting and regularization, which are based on double machine learning, can mitigate missing variable bias but may increase the risk of overfitting. Therefore, in addition to employing double machine learning for causal mediating inference, this study integrated the multi-time point difference method with the two-step mediating effect approach proposed by Jiang et al. to provide supplementary validation of the mediating mechanism underlying the impact of the ICPP on ULGUE [88]. As shown in

Table A1. Two-step test of mediating effect results.

Variable	lnExp (1)	lnNewI (2)	GC (3)
ICPP	0.222 *** (4.037)	0.241 *** (4.246)	0.047 *** (5.860)
Constant	0.006 (0.732)	0.010 (1.222)	−0.003 ** (−2.235)
Control	YES	YES	YES
Control_Squ	YES	YES	YES
City FE	YES	YES	YES
Year FE	YES	YES	YES
N	3653	3653	3653

Note: ** and *** indicate the statistical significance at 5% and 1% level.

of the Appendix A, the regression results robustly confirm the existence of the three mediating mechanisms: digital transaction scale, future industrial development, and green consumption behavior.

5.6. Analysis of Heterogeneity

5.6.1. Resource Endowment Heterogeneity

Resource-based cities often exhibit a tendency to depend on a singular and rigid economic structure during their evolutionary process. Their industrial development has traditionally been characterized by high capital input and extensive processing pathways [89]. This resource-oriented industrial structure results in the inefficient allocation of capital and labor factors through the siphon effect in factor markets, significantly crowding out industrial technological upgrading and land green transformation, ultimately leading to the classic “resource curse” phenomenon. To examine the heterogeneity of the ICPP in ULGUE cities with varying resource endowments, this study divides the sample into resource-based cities and non-resource–based cities according to the National Sustainable Development Plan for Resource-Based Cities (2013–2020). Grouped regression analysis is then conducted. Furthermore, prefecture-level and higher-level cities that only partially involve county jurisdictions are excluded from the classification of resource-based cities. The specific regression results are presented in column (1) of

Table 7. Heterogeneity test results.

Variable	Resource-Based Cities	Non-Resource–Based Cities	HSR Cities	Non-HSR Cities	Inland Cities	Coastal Cities
	(1)		(2)		(3)
ICPP	0.249 *** (8.673)	0.154 *** (3.842)	0.235 *** (8.004)	0.138 *** (6.591)	0.246 *** (5.592)	0.185 *** (6.306)
Constant	−0.004 (−0.811)	−0.004 (−0.582)	−0.001 (−0.255)	−0.002 (−0.184)	−0.003 (−0.396)	−0.002 (−0.454)
Control	YES	YES	YES	YES	YES	YES
Control_Squ	YES	YES	YES	YES	YES	YES
City FE	YES	YES	YES	YES	YES	YES
Year FE	YES	YES	YES	YES	YES	YES
N	1456	2197	2349	1304	1105	2548

Note: *** indicates the statistical significance at 1% level.

. After dividing the sample, the ICPP significantly improves ULGUE at the 1% level, regardless of whether the cities are resource based or non-resource based. A further comparison of the regression coefficients indicates that the impact of the ICPP on improving ULGUE is stronger in resource-based cities than in non-resource–based cities. This research conclusion aligns with Qian et al. [90]. The reason lies in the fact that non-resource–based cities typically exhibit a relatively high intensity of original environmental regulations. In these cities, the government places greater emphasis on promoting green urban land use modes. Consequently, the ICPP primarily enhances ULGUE by reinforcing the management and control processes, but its marginal effect is limited. By contrast, resource-based cities frequently encounter issues such as the ineffectiveness of environmental regulations and misallocation of land resources. The digital transaction scenarios fostered by the ICPP, along with the transparency of product information, have compelled enterprises to adopt low-carbon production practices and contribute to green urban construction. This has effectively broken the inertial cycle of “resource gains reinvested in traditional production capacity” within the production process per unit of land, thereby triggering Schumpeterian “creative destruction”, which proves more effective for improving ULGUE.

5.6.2. Transport Infrastructure Heterogeneity

High-speed rail (HSR) serves as a quintessential example of China’s transportation infrastructure and plays a pivotal role in enhancing the efficiency of urban resource allocation. HSR facilitates regional economic integration, fosters the concentration of financial and technological resources in core cities, accelerates the development of manufacturing or specialized industries in small- and medium-sized cities, and reduces the occurrence of redundant land use and inefficient industrial parks [91]. In the context of digitization, the impact of the ICPP on ULGUE may vary depending on the transport infrastructure. Therefore, based on the construction and commissioning information of high-speed railways published by China Railway Corporation, this study manually sorted out the opening time nodes of high-speed railways in various Chinese cities, and divided the samples according to whether HSR was open. The specific regression results are presented in column (2) of Table 7. After dividing the sample, regardless of whether a city has introduced HSR or not, the ICPP significantly enhances ULGUE at the 1% level. By further comparing the magnitude of the regression coefficients, it is evident that the ICPP exerts a stronger positive impact on ULGUE in cities with HSR, consistent with Zhang et al.’s conclusion that informatization fosters inclusive green growth in cities [92]. This may be attributed to the fact that the HSR improves urban accessibility, mitigates information asymmetry, and enables capital guided by the ICPP to more precisely target investments in land green transformation. In addition, cities with HSR can improve the intensive utilization rate of land through the synergistic mechanism of traffic diversion and information consumption. However, due to the limitation of logistics efficiency, the improvement effect of green land use in cities without HSR is weak.

5.6.3. Geographical Location Heterogeneity

The ICPP encompasses both coastal and inland cities. Given the distinct differences between these two types of cities in terms of land resources and economic advantages, this study categorizes the analysis into two groups: inland cities and coastal cities. The regression results are presented in column (3) of Table 7. The promotion effect of the ICPP on ULGUE is significant at the 1% level in both inland and coastal cities. Further comparison of regression coefficients reveals that the ICPP has a more pronounced positive impact on ULGUE in coastal cities compared to inland cities. This conclusion aligns with the findings of Feng et al. [93]. This conclusion indicates that coastal regions typically possess more advanced network infrastructure, which can effectively support the application of digital technologies required by the ICPP. This, in turn, plays a crucial role in intensifying commercial land use and mitigating land pollution. In contrast, the level of informatization in inland cities is relatively low, particularly in northwest China. This deficiency results in delays in the implementation of the ICPP, especially in areas such as technology promotion and data sharing, thereby constraining the progress of land green transformation.

6. Discussion and Conclusions

6.1. Discussion

First, from a global perspective, information consumption has been extensively adopted in many countries such as China, the United States, and those in Western Europe as an effective means to stimulate public consumption and achieve economic development objectives. In recent years, with increasing global attention to dual carbon issues, the ICPP’s function for environmental improvement has become increasingly prominent across different regions. Therefore, the evaluation of the effectiveness of the ICPP emerges as a significant topic in environmental studies. However, the current research predominantly focuses on urban carbon emission reduction and green innovation [18,19,51], without addressing the promotion of ULGUE. To fill the gap, this study innovatively establishes a causal connection between the ICPP and ULGUE, thereby providing a valuable complement to the existing research. To this end, we analyze the changes occurring within the coupling system of “economy–society–ecological environment”, from the perspective of emerging consumption patterns.

Second, in the context of China, at present, the mediating mechanism of the current ICPP regarding its impacts on economic development and environmental quality primarily emphasizes the innovation effect [18,19,32,51], industrial structure transformation [18,19,30,31,32], and government environmental attention [19,44]. Despite the significance of these mechanisms, the role of digital transactions, which serves as the most direct driver of the ICPP, has been frequently overlooked. Moreover, in the process of Chinese-style modernization, future industrial development has been designated as the top priority for the period 2026—2030. While the transformation of industrial structure has been frequently discussed, the future industrial development driven by the ICPP plays a more critical role in shaping the country’s green development strategy. Meanwhile, although governmental environmental supervision can serve as an effective environmental regulation to reflect its impacts on ULGUE, cities are the most important carriers of public production and social life. The green consumption behavior of the social public can more intuitively reflect the intrinsic impact of the ICPP on ULGUE. The three mechanisms identified in this study—namely, digital transaction scale, future industrial development, and green consumption behavior—not only expand the transmission mechanisms of the current ICPP to ULGUE and other ecological and environmental factors, but also deepen our understanding of their vital interrelationships.

Third, like most previous land studies, this study also categorizes China into specific regions based on their distinct geographical characteristics [44,46,68,93] and investigates the varying impacts of the ICPP on ULGUE across different geographical locations. However, in reality, the ICPP and the theoretical mechanism of ULGUE indicate that the disparities among cities with differing resource endowments and transportation infrastructure development are significantly greater than their differences attributable to geographical location [90]. Therefore, when examining the impacts of the ICPP on ULGUE, conducting heterogeneity analysis among cities with varying resource endowment types and transportation infrastructure development levels holds greater practical relevance and significance.

Fourth, the research perspective is uniquely positioned. Although numerous studies have explored the policy effects of ULGUE, prior investigations have predominantly focused on specific dimensions. For instance, open government data policies have been examined from the perspective of signal transmission [44], while the National Big Data Comprehensive Pilot Zone has been analyzed through the lens of industrial transformation [45]. Several research conclusions have also assessed the role of smart city pilot policies, with a focus on investment and infrastructure development as key entry points [94]. In contrast, this study centers on the transformation of consumption patterns, offering a novel and distinctive perspective to prior policy research while reinforcing the connection between people’s livelihoods and ULGUE. This study not only establishes a theoretical foundation for other similar policy research within the context of the digital economy era, but also offers valuable insights and practical recommendations for the development of smart cities in light of emerging consumption patterns.

6.2. Conclusions

In this study, we utilize the panel data of 281 prefecture-level cities in China from 2011 to 2023 to measure ULGUE using the super-efficiency SBM model with unexpected outputs. Additionally, a double machine learning model is employed to assess the impact of the ICPP on ULGUE and to elucidate its specific mediating mechanism. The following conclusions are drawn.

First, this study finds that the ICPP has a significant improvement effect on ULGUE, and this conclusion is still valid under various robustness tests. Second, according to the results of the mechanism analysis, the ICPP can improve ULGUE by expanding the scale of digital transactions, nurturing future industrial development and promoting green consumption behavior. Third, the impact of the ICPP on ULGUE shows different effects among cities with different resource endowments, cities with different levels of transport infrastructure construction, and cities in different geographical locations. Specifically, the ICPP has a stronger effect on improving the green utilization efficiency of land in resource-based cities, cities with high-speed rail, and coastal cities.

6.3. Recommendations

Based on the aforementioned findings, this study proposes the following decision-making recommendations for relevant government departments in China.

According to the benchmark regression analysis in this study, we suggest that the governments should systematically summarize the experiences gained from the ICPP, establish replicable case models, and appropriately extend the scope and depth of the pilot programs. Furthermore, efforts should be made to construct urban 5G networks and other information infrastructure, fully leveraging the advantages of information consumption to facilitate the transition of traditional urban consumption patterns toward informatization, thereby enhancing consumers’ willingness to engage in information-related consumption. In terms of incentives, a dynamic evaluation mechanism should be established to link the ICPP with land spatial planning. The improvement index of green land use efficiency should be incorporated into the annual evaluation system for pilot cities. Provincial natural resources departments can establish special funds for enhancing green land efficiency, rewarding areas within pilot cities that demonstrate a significant increase in the ecological value per unit of land with construction land quotas. This would foster a virtuous cycle of “policy incentive-efficiency enhancement-resource feedback”. In terms of supervision, the ecological and environmental departments can jointly establish a “policy effect tracking and evaluation system” with their Big Data Development Centers, and implement a dual-track supervision system combining quarterly remote sensing monitoring and annual field verification for those information consumption pilot cities. In this regard, we can establish a standard system for green land efficiency among the pilot cities, and regularly release rankings for improving green land efficiency in those cities.

Moreover, according to the results of the mediating analysis in this study, we suggest that, first, in line with the digital transaction scale mechanism, the government should establish a platform for “digital transaction efficiency-green land conversion” in densely populated urban agglomerations, such as the Yangtze River Delta and Pearl River Delta regions, under pilot policies promoting information consumption. This would enable those cities actively engaged in digital transactions to collaborate with neighboring key cities focusing on ecological and environmental protection, while utilizing economized industrial lands to jointly construct peripheral ecological corridors. The above measure will generate cross-regional improvements in land use green efficiency. Subsequently, in accordance with the future industrial development mechanism, the green utilization of land will require local governments to plan spatial layouts from the perspective of their future industrial structures. In information consumption pilot cities, priority zones for fostering future industries should be clearly demarcated. For enterprise clusters aligned with the new business orientation, a two-way commitment system for land demand and ecological performance should be implemented. Industrial upgrading enhances land efficiency, thereby enabling ecological compensation. Finally, based on the green consumption mechanism, to encourage the public to adopt green consumption behaviors, the government should establish a low-carbon lifestyle certification system in each community, by leveraging an information technology platform. Green behavior data would be integrated into community assessments, and those certified communities would undergo infrastructure upgrades to provide enhanced greening services.

Lastly, according to the results of heterogeneity analysis in this study, we suggest that, firstly, considering the heterogeneity of urban resource endowments, during the processes of resource development, processing and transportation, and transformation and development, resource-based cities should not only focus on the economic benefits derived from land use, but also take into account the ecological environment costs associated with such activities. Governments should enhance land use policies by establishing stricter environmental protection standards, so as to regulate emissions and pollution behaviors in land use practices, and promote the comprehensive utilization efficiency of land resources. Additionally, through the implementation of the ICPP, big data technology can be integrated to establish a land use information management system. In contrast, non-resource–based cities should continue to prioritize the development of digital and sharing economies, thereby reducing their reliance on land resources and achieving a balance between economic growth and environmental sustainability. Secondly, in light of the heterogeneity in urban transport infrastructure developments, it is essential to leverage the railway networks to enhance intercity economic and technical cooperation as well as land policy coordination, thereby promoting the overall improvement of regional green land use efficiency. Simultaneously, by integrating information consumption with railway transportation, knowledge spillovers and technology diffusion can be facilitated from regional core cities to peripheral cities, and from high-speed rail–connected cities to those without such connections, thus more effectively meeting the technological needs of relatively underdeveloped cities. Finally, in view of the heterogeneity of urban geographical locations, coastal areas should focus on strengthening the transformation of old and new growth drivers as well as the upgrading of industrial structures by leveraging information consumption as an emerging demand-side consumption model. Meanwhile, inland cities should fully utilize their comparative and late-mover advantages to pursue differentiated development strategies. In the process of absorbing the transfer of certain enterprises from coastal regions, inland cities should, in pursuit of their own sustainable economic growth, strictly regulate land use and enforce stringent environmental entry thresholds to ensure green economic developments.

6.4. Limitation and Future Directions

There are certain limitations in this study, which also serve as potential directions for future research. Firstly, concerning the model, this study establishes a static policy effect evaluation using double machine learning. However, the release of policy dividends requires adherence to temporal sequence rules. Future research could thus develop a dynamic panel model to examine the lagged and cumulative effects of the ICPP on ULGUE. Additionally, future studies should further explore the knowledge spillover effects between pilot and non-pilot cities for information consumption, potentially employing a spatial difference-in-differences (SDID) model to analyze the spatial diffusion pathways of such policies.

Secondly, with respect to the research methodology, this study examines the impact of a single policy on ULGUE. However, during the process of Chinese-style modernization, multiple parallel policies, such as the low-carbon pilot policy and the Broadband China policy, were implemented between 2011 and 2023. Although this study has rigorously tested and addressed the interference of similar parallel policies to ensure the robustness of experimental results, it is of greater practical significance to further investigate the synergistic effects of different policy combinations rather than merely eliminating the influence of parallel policies. To this end, a Computable General Equilibrium (CGE) model can be constructed in future research to explore the impact of policy scenario combinations on ULGUE.

Finally, regarding the data, due to data limitations and challenges in data retrieval, this study incorporated data from 281 prefecture-level cities to evaluate the effects of the ICPP, rather than including all prefecture-level cities individually in the model. In future research, the sample of prefecture-level cities could be further expanded, or the research focus could be refined to county-level administrative regions for more specific analysis.

Furthermore, regarding ULGUE’s desired output variable selection, this study employs green-covered area as a representation of ULGUE’s expected output, which aligns with certain current common practices. However, future research could explore incorporating the value of urban landscapes, given its relevance to the sustainable development requirements for both urban areas and human settlements. Additionally, it is necessary to explore whether the findings of this study can be generalized to other developing countries facing issues of land pollution and rapid industrialization, such as Brazil and India, thereby providing valuable insights for global land green use decision-making.