The Ecological Value Release Effect of Data Elements: Evidence from the Launch of Public Data Open Platforms

Abstract

This study examines the impact of public data openness on environmental welfare performance using a quasi-natural experimental approach based on the establishment of prefecture-level city public data openness platforms. Our findings reveal that public data openness significantly improves urban environmental welfare performance. Furthermore, heterogeneity analysis highlights that public data openness can play a more positive role in cities in eastern China, cities with greater fiscal autonomy, and cities where local governments place greater emphasis on environmental protection. Mechanism analysis demonstrates that public data openness enhances environmental welfare performance through stricter environmental regulatory constraints, industrial structure upgrading, increased public participation and supervision, and advancements in innovation and entrepreneurship. Extensive analysis shows that public data openness within a spatial framework can significantly enhance environmental welfare performance in the region. However, this process will generate a triple “siphon effect” that inhibits improvements in urban environmental welfare performance in surrounding areas. Additionally, this effect exhibits a certain degree of geographical attenuation influenced by economic interdependence, with an attenuation boundary of 1000 km. This study injects internet and big data thinking into ecological civilization construction, endowing it with new models, new scenarios, and new momentum, and providing a brand-new approach to sustainable development practices.

1. Introduction

The data-centric digital transformation is continuously driving the upgrading and iteration of traditional governance methods in countries around the world [1]. In the digital age, data are a strategic and fundamental resource that is essential to creating new kinds of productivity. In the current intricate and dynamic global environment, data elements, with their advantages such as non-competitiveness, replicability, economies of scale, and minimal costs for multiple uses, are fundamentally revolutionizing conventional production inputs, propelling the emergence of a data-driven trajectory toward high-quality growth [2,3].

The “Three-Year Action Plan for ‘Data Factor ×’ (2024–2026)” (hereinafter referred to as the “Action Plan”) was released in January 2024 by the National Data Administration and 16 other departments to maximize the multiplier effect of data factors to support social and economic growth. China has a wealth of data resources. The National Data Resources Survey Report (2023) states that China produced 32.85 zettabytes (ZB) of data in 2023, a 22.44% increase from the previous year. As a crucial part of data resources, public data encompasses not only government governance-related data produced by government departments at all levels in the course of carrying out their public duties and offering public services but also data produced by micro-entities like businesses with substantial public interests in market activities, which consistently play a leading, exemplary, and catalytic role in directing the development and use of data [4,5].

Based on three key considerations, i.e., national strategic needs, technological empowerment potential, and economic transformation logic, the “Action Plan” has selected 12 key industries and fields, intending to create typical demonstration scenarios for data element applications. Among them, green and low-carbon development focused on ecological and environmental construction and is explicitly identified as an important component of this series of application scenarios. Data elements, with their characteristics of low-cost reuse and cross-temporal and spatial allocation, can effectively promote the research and application of green and low-carbon technologies, facilitate the digitalization, intelligentization, and green transformation of traditional industries, and provide new impetus for low-carbon development. Therefore, deeply understanding the important strategic significance of public data openness in fostering and developing new productive forces and fully tapping into the green potential of data to fully unleash the ecological value of public data openness are urgent issues that need to be addressed.

The research presented in this paper aligns most closely with the literature on open public data. At the macro-regional level, the primary impact effects of public data openness include enhancements in urban innovation capability [6,7], economic impact [8], and public trust [9]. At the micro-enterprise level, public data openness unleashes its data dividends primarily through factors such as enterprise total factor productivity [10], enterprise digital transformation [11], enterprise productivity [12], enterprise innovation levels [13], and effective enterprise investment [14]. Existing research on the opening of public data has primarily focused on exploring its economic effects, with only a few studies examining the green effects of public data opening, such as carbon reduction capacity [15], green transition efficiency [16], and new quality productive forces [17]. However, research revealing its environmental value from an ecological welfare perspective remains relatively scarce.

In May 2018, General Secretary Xi Jinping pointed out at the National Conference on Ecological Civilization that “economic development is for the people’s livelihood, and protecting the ecological environment is also for the people’s livelihood”, highlighting the important role of building a good ecological environment and improving people’s well-being.

This paper adopts an ecological welfare perspective to explore the green value and potential of public data openness, aiming to deeply uncover the ecological value release effects of public data openness and its underlying mechanisms. It seeks to establish a set of policies that are theoretically sound, practically applicable, and systematically specific, to contribute to institutional designs that can fully unlock the ecological value of public data openness.

This paper first logically deduces the potential impact of public data openness on environmental welfare performance. Subsequently, it employs a Hybrid-Network Data Envelopment Analysis (Hybrid-Network-DEA) model to accurately measure environmental welfare performance. Building on this, this paper takes the integration of public data openness and environmental welfare performance as its entry point, treating the phased rollout of municipal government public data openness platforms as a quasi-natural experiment. It empirically examines the impact and mechanisms of public data openness on environmental welfare performance to explain the ecological value release effects of public data openness.

Specifically, this paper aims to expand on the following aspects: First, in terms of research perspective, this paper innovatively analyzes the impact of public data openness on environmental welfare performance based on a quasi-natural experimental scenario created by the launch of municipal government data openness platforms, providing new insights for better understanding and evaluating the environmental effects triggered by the launch of public data openness platforms. Second, in terms of theoretical mechanisms, this paper reveals the influence mechanisms of public data openness on environmental welfare performance from the perspectives of environmental regulatory constraints, industrial structure upgrading, public participation and supervision, and innovation and entrepreneurship incentives, thereby expanding the boundaries of research on environmental welfare performance. Third, in terms of research methods and indicator construction, this paper utilizes the Hybrid-Network-DEA model to construct environmental welfare performance measurement indicators that better align with the concept of “harmonious coexistence between humans and nature”. It also explores the heterogeneous effects of public data openness on environmental welfare performance from multiple dimensions, including geographical factors, fiscal autonomy, and environmental concern. Additionally, it further examines the spatial spillover benefits of public data openness on environmental welfare performance and the potential spatial decay boundaries. The findings of this study contribute to exploring modernization pathways for digital ecological civilization, aiming to provide new insights for policies seeking the compatibility of stable growth, environmental protection, and enhanced well-being, thereby offering theoretical support and decision-making references for achieving the goal of modernizing environmental governance.

2. Research Hypotheses

2.1. Policy Background

Data resources, recognized as the fifth major production factor alongside capital, labor, land, and technology, have emerged as a crucial element in developing new forms of productivity due to their unique multidimensional attributes such as reproducibility, non-consumability, and real-time dynamics [18]. In 2023, nearly 2.9% of China’s total data production was preserved. Among this, approximately 40% of the data remained unused for a year, and inadequate data processing capabilities resulted in a significant underestimation of data value and difficulties in mining and reusing [19]. In response, the Chinese government has attached great importance to public data openness. In 2012, China launched government public data openness platforms in Shanghai, Beijing, and Zhanjiang for the first time. In August 2015, the “Notice of the State Council on Printing and Distributing the Action Plan for Promoting the Development of Big Data” emphasized the need to lead with the open sharing of government data to activate dormant data. In January, June, and December of 2022, the State Council issued the “14th Five-Year Plan for the Development of the Digital Economy”, the “Guiding Opinions on Strengthening the Construction of a Digital Government”, and the “Opinions of the CPC Central Committee and the State Council on Establishing a Fundamental Data System to Better Leverage the Role of Data as a Factor of Production”, respectively, further highlighting the importance of “promoting the aggregation, integration, sharing, exchange, and developmental application of public data”. As of September 2024, 219 local governments in cities had launched public data openness platforms (Figure 1), accounting for 65.8% of the total. China’s public data openness and sharing have begun to yield results, gradually forming a scale advantage in data production.

2.2. Research Hypotheses

Public data openness can alleviate information asymmetry between government and enterprises, as well as between government and the public, effectively reducing external costs such as search, negotiation, supervision, and governance inherent in environmental regulation processes. Simultaneously, by empowering internal government organizations and departments through data elements, it facilitates cross-departmental flow of environmental information and the collection and aggregation of relevant data. This reduces internal communication barriers within government, helping local governments align their internal governance mechanisms with external digital development environments. Consequently, it meets residents’ environmental needs and safeguards the quality of urban living environments. Based on this, Hypothesis 1 is proposed.

Hypothesis 1:

Public data openness possesses an ecological value release effect.

Public data openness provides a data platform for environmental regulation. Combined with digital technologies such as artificial intelligence and big data analysis, local governments can gradually change their previous “experience-based” governance decision-making model and develop a scientific decision-making process based on “data.” In the past, under conditions of weak Internet penetration, severe information asymmetry between government and enterprises, and restrictions on administrative resources, it was difficult for local governments to obtain real pollution data from enterprises. As a result, environmental regulation was insufficient, ineffective, and did not satisfy residents. Additionally, influenced by the decentralized system, local governments may choose to ignore corporate pollution behaviors to gain an advantage in economic development, and in severe cases, corruption such as “government-business collusion” and “rent-seeking” may occur [20], leading to the distortion and failure of environmental policies. The opening of public data can effectively alleviate the aforementioned issues, significantly reducing the negative effects caused by information asymmetry. Based on this, this paper proposes Hypothesis 2.

Hypothesis 2:

Public data openness releases ecological value through the regulatory constraint effect of environmental regulations.

Data elements are a non-physical form of resource that can reduce reliance on energy inputs by leveraging their low natural resource consumption, minimal pollution emissions, and scalable economic benefits and environmentally friendly characteristics, thereby driving industrial structure transformation toward high-value-added sectors [21]. Within industries, the flow of data elements among enterprises enhances their capabilities in collecting, obtaining, integrating, and analyzing various types of data information. After internalizing information and technology, enterprises can achieve optimal technical solutions through rapid search, autonomous learning, and comprehensive mastery to support process management, thereby controlling energy consumption and detecting pollution throughout the entire product lifecycle. Between industries, data openness breaks down barriers to integration, eliminating spatial and temporal constraints on inter-industry integration. This enables the gradual penetration and integration of the primary and secondary industries into the tertiary industry. The reallocation of production factors allows industries to connect closely through horizontal and vertical exchange and feedback, blurring industry boundaries and forming an interconnected, advanced industrial structure, which is conducive to improving environmental welfare performance. Based on this, this paper proposes Hypothesis 3.

Hypothesis 3:

Public data openness releases ecological value through industrial structure upgrading effects.

Data-driven information disrupts traditional information dissemination channels, enhances public engagement in urban environmental governance, and provides a practical pathway to establish a multi-stakeholder governance framework for environmental management. On one hand, by leveraging public data open platforms, the public is granted the right to express diverse demands and participate in governance, thereby lowering the barriers and costs of participation in environmental governance. This facilitates the public’s ability to organize and conduct environmental supervision and feedback activities effectively. Especially when facing corporate pollution, the public can communicate, report, protest, or file complaints with local governments promptly, effectively reducing “rent-seeking” and “government-business collusion” between local governments and enterprises [22]. On the other hand, public participation exerts a deterrent effect on enterprises. Negative environmental practices by enterprises may trigger a series of chain reactions, such as investor withdrawal, consumer boycotts, and government-mandated shutdowns, forcing enterprises to proactively engage in substantive energy conservation and emissions reduction efforts to maintain their social image [23]. Based on this, this paper proposes Hypothesis 4.

Hypothesis 4:

Public data openness releases ecological value through the supervisory effect of public participation.

From an innovation perspective, data openness diversifies the integration of upstream and downstream enterprises in the industrial chain, forming a digital ecosystem network. This network provides knowledge and creativity to enterprises within the industrial chain, which continuously engage in self-technological innovation through mechanisms such as “knowledge spillover” and “innovation iteration”, thereby unlocking the ecological value of technological innovation. The close integration of production, sales, and innovation expands the production possibility frontier and technological innovation frontier of enterprises [24], thereby providing sustained momentum for improving environmental welfare. From an entrepreneurial perspective, public data openness platforms can effectively integrate market resources and establish a fair and competitive market environment. The disclosure of large amounts of government data, such as planning information, work reports, and fiscal budgets and final accounts, enables enterprises to understand and supervise local government decisions, enhancing the openness, transparency, and fairness of local government operations and improving the business environment. This not only attracts “green” enterprises to continuously enter the market to maintain competitive vitality but also forces “brown” enterprises to exit more quickly, achieving the dual goals of “stable growth” and “improved environment” [25]. Therefore, this paper proposes Hypothesis 5.

Hypothesis 5:

Public data openness releases ecological value through innovation and entrepreneurship incentives.

3. Research Design

3.1. Identification Strategy

Public data openness is an exogenous event for urban environmental welfare performance, providing an opportunity for this paper to treat it as a quasi-natural experiment. Considering the inconsistent timing of the public data openness by various prefectural-level governments, this paper establishes a multi-period difference-in-differences (DID) model to estimate the policy effect of government public data openness on environmental welfare performance, i.e., to test Hypothesis 1. Meanwhile, to explore the net effect of the policy impact, this paper also controls for individual differences and time differences in the research objects before and after the policy implementation. Because of this, this paper takes the prefectural-level governments that have launched public data openness platforms as the experimental group and those that have not yet launched such platforms as the control group, constructing a multi-period DID model based on two-way fixed effects. The specific model is as follows:

$$Ew{p}_{it}={\alpha }_0+{\alpha }_{1}DI{D}_{it}+{\alpha }_2{X}_{it}+{\lambda }_i+{\mu }_t+{\epsilon }_{it}$$

(1)

Ewp_it represents the environmental welfare performance of the i prefecture-level city during period t, DID_it denotes the multi-period difference-in-differences term, which is a dummy variable for the openness of the city government’s public data platform, with α₁ being its estimated coefficient. For cities that have established a public data openness platform, the difference-in-differences term is assigned a value of 1, while cities that have not established such a platform are assigned a value of 0; X_it and α₂ represent the set of control variables affecting environmental welfare performance and their respective estimated coefficients; α₀, λ_i, and μ_t are the constant term, individual fixed effects, and year fixed effects, respectively; ε_it is the random error term

3.2. Variables

3.2.1. Dependent Variable: Environmental Welfare Performance

At present, China is at an important juncture of ecological civilization system reform, and the practice of ecological environment governance has gradually shifted from key rectification to systematic governance, from passive response to active action, and from global environmental governance participants to leaders. Based on the background, the public is increasingly concerned about the impact of economic and social development on their health and quality of life in the process of green empowerment of new quality productivity. In view of this, based on the traditional welfare economics analysis, this study adds the residents’ health factors closely related to environmental pollution to the monetizable factor aggregation, constructs a monetary health analysis framework, and uses the Hybrid-Network-DEA model to quantify environmental welfare performance.

This paper adopts a comprehensive perspective encompassing economic development, ecological environment, and public welfare, dividing the input–output process into two major nodes: macro-level development performance and micro-level development performance. Among these, the input factors for Node 1 of macro-level development performance are capital factors, labor factors, and energy factors [26,27,28]; micro-level welfare performance’s input factors are the output factors from Node 1, while also incorporating healthcare conditions and environmental governance expenditures into the input factor set of Node 2 [29]. The output factors include expected outputs [30] and unexpected outputs [31,32]. Furthermore, we define the comprehensive efficiency of the two nodes as the dependent variable of environmental welfare performance in this paper. The specific measurement framework is shown in Figure 2.

Specifically, the model for measuring environmental welfare performance using Hybrid-Network-DEA is as follows:

$$\left\{\begin{array}{l}Min({\delta }_1+{\delta }_2)\\ {\displaystyle {\sum }_{j=1,\ne 0}^n{\lambda }_j^1{s}_j^{1,1}\le {\delta }_1{s}_0^{1,1}}, {\displaystyle {\sum }_{j=1,\ne 0}^n{\lambda }_j^1{s}_j^{1,1}\le {\delta }_1{s}_0^{1,2},} {\displaystyle {\sum }_{j=1,\ne 0}^n{\lambda }_j^1{s}_j^{1,3}\le {\delta }_1{s}_0^{1,3}}, {\displaystyle {\sum }_{j=1,\ne 0}^n{\lambda }_j^1{s}_j^{2,1}\ge {\delta }_2{s}_0^{2,1},}\\ {\displaystyle {\sum }_{j=1,\ne 0}^n{\lambda }_j^1{s}_j^{2,2}\le {\delta }_2{s}_0^{2,2}}, {\displaystyle {\sum }_{j=1,\ne 0}^n{\lambda }_j^1{s}_j^{2,3}\le {\delta }_2{s}_0^{2,3},} {\displaystyle {\sum }_{j=1,\ne 0}^n{\lambda }_j^1{s}_j^{2,4}\le {\delta }_2{s}_0^{2,4}}, {\displaystyle {\sum }_{j=1,\ne 0}^n{\lambda }_j^1{s}_j^{2,5}\le {\delta }_2{s}_0^{2,5},}\\ {\lambda }_j^1\ge 0, j=1,\cdots ,n,j\ne 0\end{array}\right.$$

(2)

$$\left\{\begin{array}{l}{\displaystyle {\sum }_{j=1,\ne 0}^n{\lambda }_j^2{s}_j^{2,1}\le {\delta }_2{s}_0^{2,1}}, {\displaystyle {\sum }_{j=1,\ne 0}^n{\lambda }_j^2{s}_j^{2,2}\le {\delta }_2{s}_0^{2,2},} {\displaystyle {\sum }_{j=1,\ne 0}^n{\lambda }_j^2{s}_j^{2,3}\le {\delta }_2{s}_0^{2,3}}, {\displaystyle {\sum }_{j=1,\ne 0}^n{\lambda }_j^2{s}_j^{2,4}\le {\delta }_2{s}_0^{2,4},}\\ {\displaystyle {\sum }_{j=1,\ne 0}^n{\lambda }_j^2{s}_j^{2,5}\le {\delta }_2{s}^{2,5}}, {\displaystyle {\sum }_{j=1,\ne 0}^n{\lambda }_j^2{s}_j^{2,6}\le {\delta }_2{s}_0^{2,6},} {\displaystyle {\sum }_{j=1,\ne 0}^n{\lambda }_j^2{s}_j^{2,7}\le {\delta }_2{s}_0^{2,7}}, {\displaystyle {\sum }_{j=1,\ne 0}^n{\lambda }_j^2{s}_j^{3,1}\ge {\delta }_2{s}_0^{3,1},}\\ {\displaystyle {\sum }_{j=1,\ne 0}^n{\lambda }_j^2{s}_j^{3,2}\ge {\delta }_2{s}_0^{3,2}}, {\displaystyle {\sum }_{j=1,\ne 0}^n{\lambda }_j^2{s}_j^{3,3}\ge {\delta }_2{s}_0^{3,3},} {\displaystyle {\sum }_{j=1,\ne 0}^n{\lambda }_j^2{s}_j^{2,7}\ge {\delta }_2{s}_0^{3,4}}, {\displaystyle {\sum }_{j=1,\ne 0}^n{\lambda }_j^2{s}_j^{3,5}\ge {\delta }_2{s}_0^{3,5},}\\ {\displaystyle {\sum }_{j=1,\ne 0}^n{\lambda }_j^2{s}_j^{3,6}\le s_0^{3,6}, {\lambda }_j^2\ge 0, j=1,\cdots ,n,j\ne 0}\end{array}\right.$$

(3)

$${\delta }^{*}=Min{\displaystyle \frac{1-\frac{1}{p}({\displaystyle {\sum }_{i=1}^{p_1}(Z_i^{z^{-}}/x_{io}^z)+p_2(1-a))}}{1+\frac{1}{q}({\displaystyle {\sum }_{q=1}^{q_1}(Z_q^{zg}/y_{ro}^{zg})+p_2(1-a))}}}, Z_i^{z^{-}}\ge 0, Z_q^{zg}\ge 0, 0\le a\le 1$$

(4)

Equations (2) and (3) are the constraints for solving the efficiency values of Node 1 and Node 2, respectively, and Equation (4) is the efficiency value of a single node obtained through calculation. p = p₁ + p₂, where p₁ and p₂ represent the number of types of divisible and indivisible input factors, respectively; similarly, q = q₁ + q₂, where q₁ and q₂ represent the number of categories of separable desired output factors and non-separable undesired output factors, respectively; x^z_io and y^zg_ro are the input factors and expected output factors with separable attributes, respectively; Z^z−_i and Z^zg_q represent the slack of separable input factors and separable expected output factors, respectively; λ_j is the weight coefficient of the decision-making unit; δ₁ and δ₂ are the efficiency values of Node 1 and Node 2 obtained by solving the problem, respectively, and δ^* is the comprehensive efficiency value of the two stages. In addition, this article adopts the assumption of constant returns to scale when calculating efficiency values.

3.2.2. Independent Variable

As an important manifestation of local government data openness, the launch of local government public data openness platforms provides an ideal quasi-natural experimental setting for assessing the environmental effects of data openness in this study. Referring to the “China Local Government Data Openness Report (City)” published by the DMG Digital and Mobile Governance Laboratory at Fudan University, the first launch dates of public data platforms for a total of 173 cities as of the end of 2021 were obtained. These dates serve as virtual variables indicating whether a city established a public data openness platform in a given year. Specifically, cities that launched (including those already online) a public data openness platform in that year are assigned a value of 1, while others are assigned a value of 0.

3.2.3. Control Variables

To mitigate the bias in regression results caused by omitted variables, this study controls for other variables that significantly impact government environmental governance. These include the degree of openness to foreign investment (Fdi), economic development level (GDP), technological innovation level (Pat), industrial structure (Ins), population size (Tpop), and fiscal autonomy (Financial_F).

3.3. Data Sources

The time window for the aforementioned variable-related data is 2008–2021, primarily sourced from the China Urban Statistical Yearbook, China Statistical Yearbook, China Environmental Yearbook, the China Energy Statistical Yearbook, provincial and municipal statistical yearbooks, the database of the National Development Research Institute at Peking University and the Longxin Data Research Institute, the China Local Government Data Openness Report (Cities) published by the DMG Digital and Mobile Governance Laboratory at Fudan University, the Baidu Data Platform, government work reports from various regions, and relevant policy documents from the State Council. Additionally, to address missing data in some prefecture-level cities, we employed smoothing index methods and linear fitting techniques to fill in the gaps. We employed smoothing index methods and linear fitting techniques to, respectively, fill in the gaps in data with time-dependent characteristics and data with city-dependent characteristics. Considering that monetary data indicators may be influenced by inflation factors, this paper deflated all monetary indicators using the price index for constant prices with 2000 as the base year. Additionally, non-ratio data were log-transformed to mitigate the impact of heteroskedasticity.

3.4. Descriptive Analysis

Table 1. Descriptive statistics.

Symbol	Definition	Observations	Mean	Std	Min	Max	Unit
Ewp	Estimated Results	3990	0.744	0.147	0.007	1	——
DID	Policy dummy variable	3990	0.110	0.313	0	1	——
GDP	GDP	3990	1758.987	2241.287	60.207	29,273.87	Hundred million yuan
Ins	Value-added of the secondary industry/GDP	3990	0.458	0.110	0.107	0.898	——
Fdi	Foreign direct investment/GDP	3990	1.528	3.303	0	73.541	——
Pat	Patent grants	3990	5741.429	15,565.35	1	279,177	Number of items
Financial	General budget expenditure at the local fiscal level	3990	283.087	439.083	6.547	5714.606	Hundred million yuan
Tpop	Urban registered population at the end of the year	3990	473.76	529.87	18.59	9747.1	Ten thousand people

reports the descriptive statistics of the main variables in this study. The average value of Ewp is 0.744, with a median of 0.738. As shown in Figure 3, the kernel density plot of Ewp indicates that, during the sample period, the environmental welfare performance of Chinese cities follows a positively skewed distribution. Over time, the distribution has shifted from a “tall and thin” shape to a “short and fat” shape, suggesting an increase in the disparity among cities. The average value of DID is 0.110, indicating that during 2008–2021, nearly 11% of the samples in this study belonged to the treatment group. This suggests that public data openness in China is still in its initial stages, and there is no significant skewness in the gap between the levels of public data platform openness among cities.

4. Empirical Analysis

4.1. Baseline Regression Results

Table 2. The role of public data openness in promoting urban environmental welfare performance.

	(1)	(2)
	Ewp	Ewp
DID	0.017 **	0.024 ***
	(2.36)	(3.32)
lnGDP		−0.040 **
		(−2.50)
Ins		0.027
		(0.64)
FDI		−0.001 ***
		(−2.87)
lnPat		0.002
		(0.35)
lnFinancial		−0.045 ***
		(−3.32)
lnTpop		−0.038 *
		(−1.89)
City Fixed Effects	Yes	Yes
Year Fixed Effects	Yes	Yes
N	3990	3990
r2	0.095	0.173

z statistics in parentheses;* p < 0.10, ** p < 0.05, *** p < 0.01.

presents the empirical test results of the impact of public data openness on urban environmental welfare performance. This study employs a stepwise regression strategy for the baseline regression. Column (1) of

Table 3. Estimation results with the introduction of periodic provincial policy changes.

	(1)	(2)
	Ewp	Ewp
DID	0.023 ***	0.030 ***
	(3.20)	(4.32)
Control Variables	No	Yes
Province Fixed Effects × Annual Time Trend	Yes	Yes
City Fixed Effects	Yes	Yes
Year Fixed Effects	Yes	Yes
N	3990	3990
r2	0.076	0.155

z statistics in parentheses; *** p < 0.01.

reports the regression results with only the explanatory variable difference-in-differences term and fixed effects included, where the coefficient of DID is 0.017 and significant at the 5% statistical level, indicating that public data openness significantly improves urban environmental welfare performance. Column (2) further controls for factors such as economic development, industrial structure, openness to foreign investment, innovation capacity, fiscal freedom, and total population based on column (1). The estimated coefficient of DID is 0.024 and is significant at the 1% statistical level. From an economic significance perspective, after controlling for the potential impacts of economic, social, resource endowment, and unobservable factors, the influence of public data openness on urban environmental welfare performance further increases. The openness of public data platforms enhances urban environmental welfare performance by 2.4%. On the one hand, compared to the current scarcity of technological factors, the data elements published by open data platforms are non-scarce public goods characterized by non-rivalry and replicability. Different economic entities, including local governments, businesses, and the public, can access and utilize this data information, thereby triggering changes in local governments’ environmental governance behaviors and institutional transformations. On the other hand, the opening of public data has injected new vitality into environmental information resources that were previously dormant. The production, dissemination, and processing of environmental-related information have undergone significant transformations. The opening and sharing of public data have broken down information barriers and high access thresholds between different entities involved in environmental governance by linking them together, thereby optimizing environmental governance processes, reducing residents’ administrative costs, and enhancing urban residents’ environmental welfare performance. Hypothesis 1 of this paper has received empirical support.

4.2. Premise Assumption Test

4.2.1. Parallel Trend Assumption Test

An important premise assumption for using the difference-in-differences method to analyze the causal relationship between public data openness and urban environmental welfare performance is that the treatment group and control group have the same time trend, i.e., passing the parallel trend assumption test. This study adopts the event study approach to conduct a parallel trend assumption test to verify that public data openness enhances urban residents’ environmental welfare performance rather than being affected by other unobservable city characteristics. Specifically, this study uses interaction terms between the year dummy variables within the sample event window (−δ, +τ) and the treatment group dummy variable as dummies to replace the DID term in the baseline regression model. Then, regression is performed to examine the dynamic time trend of public data openness. The regression model is set up as follows:

$$Ew{p}_{it}={\alpha }_0+{\displaystyle \sum _{\delta =-1}^{\delta =-2}{\alpha }_{\delta }pre}+\alpha cru+{\displaystyle \sum _{\tau =1}^{\tau =8}{\alpha }_{\tau }post}+\beta X_{it}+{\lambda }_i+{\mu }_t+{\epsilon }_{it}$$

(5)

In Equation (5), pre and post represent the two counterfactual dummy variables before and after the event. The settings of other variables are the same as those in the baseline model. As shown in Figure 4, to more intuitively examine whether the baseline model meets the parallel trend assumption, we compare the changes in environmental welfare performance before and after the pilot policy implementation. It can be seen that neither α_δ nor α_τ rejects the null hypothesis, satisfying the parallel trend assumption. Thus, using the multi-period difference-in-differences method to assess the impact of public data openness on environmental welfare performance is reasonable and effective. Furthermore, in terms of coefficients, the public data openness platform did not have a significant impact on environmental welfare performance in the year it was launched. One year after the policy was implemented, the environmental welfare performance of the treatment group was significantly higher than that of the control group, and ecological value was gradually released.

4.2.2. Placebo Test

This study further establishes the existence of such influence with a placebo test to confirm if public data availability influences urban inhabitants’ environmental welfare performance more than other unobservable elements. In particular, this study builds pseudo-strategy dummy variables, re-estimates the baseline model, and randomly samples 500 and 1000 times on the dual dimensions of cities and the launch of public data openness platforms. Figure 5 display the kernel density distribution plots of the estimated coefficients and p-values for the pseudo-strategy dummy variables. The kernel density of the predicted coefficients for both sampling techniques is roughly a normal distribution with a mean of zero, according to the findings of the two sampling regressions. Most of the estimated coefficient p-values are greater than 0.10 and not significant at the 10% level, and all estimated coefficients do not exceed the baseline regression coefficient of 0.024. This indicates that the placebo test is passed, suggesting that the conclusion that public data openness promotes environmental welfare performance is not coincidental and demonstrates the robustness of the baseline regression results.

4.3. Endogeneity Test

Although the empirical regression results mentioned above validate that the opening of public data has a certain promotional effect on improving environmental welfare performance, this result is likely to be influenced by endogeneity issues. On the one hand, in cities with higher levels of environmental welfare performance, governments are more enthusiastic about promoting the opening of public data. Through the smooth flow of data and information resources across regions, levels, and departments, and the efficient coordination of business operations, they improve the environmental welfare performance of urban residents, creating a bidirectional causal relationship. On the other hand, the omission of unobservable variables may also lead to endogeneity issues, thereby introducing biased effects on the benchmark regression results. Therefore, this paper adopts two strategies for endogeneity testing: incorporating cyclical provincial policy changes and using the instrumental variables method.

4.3.1. Introduction of Periodic Provincial Policy Changes

Certain periodic and regional features of China’s top-down administrative management system mean that local governments create phased administrative goals for their respective jurisdictions based on the central government’s policy guidelines, modify them to fit local circumstances, and carry out pertinent plans [33]. Provincial policy changes have a significant impact on the environmental welfare performance and public data transparency status of cities in the same province. Omitting this factor could lead to biased estimation in the baseline regression. Therefore, this study introduces the interaction term between provincial fixed effects and annual time trends to control for periodic provincial policy changes. The estimation results are reported in columns (1)–(2) of Table 3. The results indicate that, whether or not periodic provincial policy changes are introduced, public data openness significantly enhances environmental welfare performance at the 1% significance level, further proving the robustness of the baseline regression results.

4.3.2. Instrumental Variable

We select appropriate instrumental variables and use Two-Stage Least Squares (2SLS) to re-estimate the baseline regression model to mitigate potential bidirectional association mechanisms in the model as much as possible.

Firstly, from the vantage point of geographical economics, cities within a province are influenced vertically by provincial policies and horizontally by those in neighboring cities within the same province. An increase in the number of cities within a province that possess public data openness platforms can catalyze cities that have yet to implement such openness. However, the impact of public data openness in other cities within the province on the environmental welfare performance of a specific city remains relatively modest. Consequently, we have chosen the proportion of cities within a province that have implemented public data openness as one of the instrumental variables for the endogeneity testing in our baseline model. Secondly, the implementation of public data openness necessitates the backing of digital infrastructure. Notably, in its nascent stages, internet development was facilitated through the installation of fixed telephone lines. Given that our sample period covers the years from 2008 to 2021, we have intentionally set the research period for our instrumental variable to 1994 to 2007, thereby guaranteeing no temporal overlap and preserving a necessary level of exogeneity. As such, we adopt the number of fixed telephone lines per hundred people in each city from 1994 to 2007 as the second instrumental variable in our baseline model. Additionally, we incorporate the first-order lag of the explained variable, environmental welfare performance (Ewp), as the third instrumental variable in our baseline model.

Table 4. Estimation results of instrumental variable method.

	(1)	(2)	(3)	(4)	(5)	(6)
	DID	Ewp	DID	Ewp	DID	Ewp
DID		0.170 ***		0.325 ***		4.585 ***
		(5.86)		(5.98)		(4.57)
IV_rate	0.194 ***
	(20.01)
IV_Phone			0.001 ***
			(10.31)
IV_L.Ewp					0.134 ***
					(4.57)
Control Variables	Yes	Yes	Yes	Yes	Yes	Yes
City Fixed Effects	Yes	Yes	Yes	Yes	Yes	Yes
Year Fixed Effects	Yes	Yes	Yes	Yes	Yes	Yes
F value	400.37		106.21		20.84
LM value (p)	336.19		55.56		20.75
N	3990	3990	3990	3990	3990	3990

z statistics in parentheses; *** p < 0.01.

presents the endogeneity test results for these various instrumental variables. The first-stage F-statistics for the three instrumental variables are 400.37, 106.21, and 20.84, respectively, with p-values all at 0.000; the LM test results are 336.19, 55.56, and 20.75, respectively, with p-values all at 0.000. This indicates that there are no weak instrumental variables or over-identification issues in the baseline model, and the selection of these three instrumental variables is reasonable and valid. From the results, after ensuring the rationality and validity of the instrumental variables, columns (1), (3), and (5) represent the first-stage regression results for the three instrumental variables. The coefficients of IV_rate, IV_Phone, and IV_L.Ewp are positive at the 1% significance level, consistent with theoretical expectations. Columns (2), (4), and (6) show the second-stage regression results for the three instrumental variables. The regression coefficients for DID are all positive at the 1% significance level. This indicates that, after alleviating endogeneity issues, the baseline regression results of this study remain robust.

4.4. Excluding Competitive Hypotheses

When assessing the impact of public data openness on environmental welfare performance, the results will inevitably be influenced and disrupted by other concurrent policies implemented during the same period, thereby obscuring the evaluation outcomes of this study regarding the implementation effects of the public data openness platform. Therefore, this paper further collected four policies whose implementation periods overlapped with the implementation period of the public data openness platform discussed in this paper, including Smart City Pilot Programs (Smart_city), Big Data Comprehensive Pilot Zones (Big_data), Low-Carbon Pilot Cities (Low_carbon), and Carbon Emission Trading Rights (Carbon_Trading). The results of the analysis are presented in

Table 5. Exclude other confounding policy shocks.

	(1)	(2)	(3)	(4)	(5)
	Ewp	Ewp	Ewp	Ewp	Ewp
DID	0.023 ***	0.024 ***	0.023 ***	0.022 ***	0.021 ***
	(3.24)	(3.40)	(3.27)	(3.27)	(3.18)
Smart_city	0.016				0.015
	(1.29)				(1.02)
Big_data		−0.009			−0.011
		(−0.98)			(−1.11)
Low_carbon			0.009		0.006
			(1.08)		(0.82)
Carbon_Trading				0.013	0.015
				(0.89)	(1.02)
Control Variables	Yes	Yes	Yes	Yes	Yes
City Fixed Effects	Yes	Yes	Yes	Yes	Yes
Year Fixed Effects	Yes	Yes	Yes	Yes	Yes
N	3990	3990	3990	3990	3990
r2	0.174	0.173	0.173	0.173	0.175

z statistics in parentheses; *** p < 0.01.

.

The coefficients of the difference-in-differences (DID) in columns (1) to (4) are still significantly positive at the 1% significance level, while the estimated coefficients of Smart_city, Big_data, Low_carbon, and Carbon_Trading are not significant. After incorporating all four policies and the launch of public data openness platforms, the regression results in column (5) also exhibit the same characteristics, and the estimated coefficient of DID is only slightly affected, decreasing from 0.024 in the baseline regression to 0.021. This indicates that the selected shock of the launch of public data openness platforms in this paper has relative exogeneity, and the research conclusions obtained are not affected by other confounding policy factors.

4.5. Other Robustness Checks

This paper undertakes a comprehensive re-verification of the robustness of the research conclusions from six distinct perspectives.

4.5.1. PSM-DID

When the national government approves the launch of public data openness platforms, it may take into account factors such as geographical regions, economic development, and human resource endowments, giving priority to implementing public data openness in economically developed areas and regions with a high concentration of technical talent. In light of this, this paper employs propensity score matching (PSM) to select samples similar to the treatment group cities in various aspects. To ensure the robustness of the estimation results, this paper processes the original sample using nearest-neighbor matching and then performs difference-in-differences regression based on the matched sample to rule out potential sample selection bias issues. As shown in

Table 6. Estimation results excluding other confounding policy shocks.

	(1)	(2)	(3)	(4)	(5)	(6)	(7)
	Ewp	Ewp	Ewp	Ewp	Ewp	Ewp	Ewp
DID	0.023 ***	0.008 ***	0.019 **	0.023 ***	0.023 ***	0.024 ***	0.018 ***
	(3.29)	(3.94)	(2.91)	(3.30)	(3.30)	(3.21)	(2.92)
Edu					−0.230 ***
					(−3.18)
Control Variables	Yes	Yes	Yes	Yes	Yes	Yes	Yes
City Fixed Effects	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Year Fixed Effects	Yes	Yes	Yes	Yes	Yes	Yes	Yes
N	3984	3990	3990	3990	3990	3990	3990
r2	0.172	0.260	0.101	0.173	0.176	0.180	0.165

z statistics in parentheses; ** p < 0.05, *** p < 0.01.

(1), the benchmark regression results remain valid.

4.5.2. Changing the Identification Strategy

In this group of tests, the variable selection and measurement methods for the explained variables in the aforementioned benchmark model were replaced. On the one hand, the indicator system was constructed using the entropy method to replace the Hybrid-Network-DEA model mentioned earlier for calculating environmental welfare. On the other hand, this paper replaced the input and output indicators for the environmental welfare performance of Node 2. The corresponding results are shown in columns (2) and (3) of Table 6. The DID coefficient results indicate that the research conclusions obtained from the benchmark regression have not changed due to the alteration in the environmental welfare performance measurement method.

4.5.3. Controlling for Time Trends

One key concern regarding the conclusions of this study is whether the causal relationship captured herein is an inevitable result of time trends. To address this issue, a time trend term is added in column (4) of Table 6. It can be seen that there is no significant difference in the DID coefficient and significance level compared to the baseline model, indicating that the baseline model specification is reasonable.

4.5.4. Adding Control Variables

Education expenditure drives the cross-regional flow of factors, promoting regional technological innovation and the accumulation of human capital. This paper uses education expenditure as one of the additional control variables and reports the results in Table 6, column (5). The results indicate that the model remains robust after adding control variables, and the improvement in residents’ education level sets higher standards for their urban living environment.

4.5.5. Excluding the Impact of Outliers

To exclude the interference of outliers, this paper trims the environmental welfare performance data by 1% and 5% at both ends. The regression results in columns (6) and (7) of Table 6 show that the DID coefficient estimates remain significantly positive, indicating that the sample data have little impact on the baseline model regression results after removing outliers, and the research conclusions of this paper still hold.

4.5.6. Excluding the Impact of Urban Factors

Finally, this paper introduces a series of interaction terms between urban factors and time trends in the baseline model to ensure the reliability of the research conclusions. As shown in columns (1)–(5) of

Table 7. Estimation results excluding the impact of urban factors.

	(1)	(2)	(3)	(4)	(5)
	Ewp	Ewp	Ewp	Ewp	Ewp
DID	0.016 ***	0.015 **	0.018 ***	0.015 **	0.015 **
	(2.67)	(2.43)	(2.92)	(2.51)	(2.46)
Control Variables	Yes	Yes	Yes	Yes	Yes
City Fixed Effects	Yes	Yes	Yes	Yes	Yes
Year Fixed Effects	Yes	Yes	Yes	Yes	Yes
N	3990	3990	3990	3990	3990
r2	0.174	0.172	0.165	0.169	0.169

z statistics in parentheses; ** p < 0.05, *** p < 0.01.

, after excluding the impact of provincial capitals, sub-provincial cities, cities along the Yangtze River Economic Belt, and municipalities directly under the central government, public data openness still has an enhancing effect on environmental welfare performance.

4.6. Heterogeneity Analysis

4.6.1. Geographical Factors

This paper divides the sample into eastern and central-western regions. As shown in columns (1) and (2) of

Table 8. Heterogeneity analysis.

	(1)	(2)	(3)	(4)	(5)	(6)
	East	C&West	H_FF	L_FF	H_ER	L_ER
DID	0.028 **	0.018	0.030 ***	0.003	0.009	0.023 ***
	(2.97)	(1.62)	(3.54)	(0.28)	(0.95)	(2.62)
Control Variables	Yes	Yes	Yes	Yes	Yes	Yes
City Fixed Effects	Yes	Yes	Yes	Yes	Yes	Yes
Year Fixed Effects	Yes	Yes	Yes	Yes	Yes	Yes
N	1400	2590	1993	1997	1992	1998
r2	0.254	0.159	0.283	0.150	0.206	0.190

z statistics in parentheses; ** p < 0.05, *** p < 0.01.

and Figure 5, the opening of public data has a significant positive impact on environmental welfare performance in the eastern region only. Compared to the central-western regions, the eastern region has a higher level of economic development, abundant human capital, economic foundations, digital infrastructure, and technological resources, which provide the material basis and development space for the rapid implementation of public data opening. According to the “2021 China Local Government Data Openness Report”, 78% of the top 50 cities in the national urban forest openness index in 2021 were eastern cities. However, central and western China are constrained by geographical, resource, technological, and trade conditions, making it difficult for them to fully leverage the information barrier-breaking effects and factor mobility multiplier effects of public data openness implementation. As a result, their impact on improving environmental welfare performance is not significant.

4.6.2. Fiscal Autonomy

From the perspective of participation constraints, the level of government public service provision is largely limited by local fiscal autonomy. In regions with higher fiscal autonomy, it is easier for local governments to leverage their informational advantage in resource allocation, which may also lead to higher benefits in public service provision. This paper divides cities into high-autonomy cities and low-autonomy cities based on the median of local government fiscal autonomy during the sample period. As shown in columns (3) and (4) of Table 8 and Figure 6, the public data openness only significantly enhances welfare performance in high-fiscal-autonomy cities. The tax-sharing reform not only grants local governments fiscal autonomy but also assigns them the primary responsibility for providing public goods within their jurisdictions. To better fulfill their expenditure responsibilities for public services, cities with high fiscal autonomy place greater emphasis on economic development quality characterized by improved resource utilization efficiency, green consumption transformation, and optimized industrial allocation. They are more likely to leverage public data platforms to foster a favorable landscape where leading high-quality enterprises accelerate resource aggregation, online and offline vitality of new consumption models is enhanced, and industrial-end data exploration forms structural advantages in the industrial chain and sustainable development trends. This contributes to improving urban functions and residents’ living environments, thereby enhancing environmental welfare performance.

4.6.3. Environmental Concern

This paper organized environmental-related terms from local government work reports to create a vocabulary set of 27 environmental keywords, including environmental pollution, energy conservation and emission reduction, CO₂, SO₂, environmental protection, air quality, low-carbon, green, etc. The amount of emphasis local governments placed on environmental preservation efforts was gauged by the percentage of these keywords in the overall word count. Cities were classified as high- or low-concern based on the median frequency of environmental phrases over the survey period. Cities with lower environmental concern are more likely to rely on public data openness to improve their environmental welfare performance, according to the results in Figure 6, columns (5) and (6). One possible reason is that cities with lower levels of environmental awareness often lack sufficient environmental investment and effective environmental governance measures, resulting in more prominent environmental pollution and ecological damage issues. Public data openness can help local governments better understand environmental quality conditions and pollution source distribution by collecting relevant datasets, thereby breaking through the constraints of insufficient funding and technology. In contrast, cities with higher levels of environmental awareness may already have well-established environmental governance systems, leading to lower marginal effects from public data openness.

5. Extension Analysis

5.1. Mechanism Analysis

In the analysis of impact mechanisms, this paper primarily explores four transmission pathways, including government environmental regulation, industrial structure upgrading, public participation and supervision, and innovation and entrepreneurship incentives. Therefore, concerning the research by Lin and Xie (2023) [34], this paper constructs the following analytical model for impact mechanisms.

$$M_{it}={\alpha }_0+{\alpha }_{1}DI{D}_{it}+{\alpha }_2{X}_{it}+{\lambda }_i+{\mu }_t+{\epsilon }_{it}$$

(6)

M_it represents the mechanism variable, and the meanings of the other variables remain consistent with those in the baseline regression model presented earlier. To a certain extent, urban air quality can reflect the effectiveness of local government environmental regulations. This paper selects urban PM2.5 concentration as the metric for environmental regulation [31]. With reference to Wang et al. (2024) [35], the measurement method of industrial structure advancement is as follows:

$$\left\{\begin{array}{l}{S}_0=(S_{10},S_{20},S_{30})\\ S_1=(1,0,0)\\ S_2=(0,1,0)\\ S_3=(0,0,1)\\ {\theta }_m=\mathrm{arc} \cos \left({\displaystyle \frac{{\displaystyle \sum _{i=1}^3(x_{im}·x_{i0})}}{{\displaystyle \sum _{i=1}^3{(x_{im}^2)}^{1/2}·{\displaystyle \sum _{i=1}^3{(x_{i0}^2)}^{1/2}}}}}\right) m=1,2,3\\ Ins\_Advanced={\displaystyle \sum _{k=1}^3{\displaystyle \sum _{m=1}^k{\theta }_m}}\end{array}\right.$$

(7)

For the quantification of public participation, this paper combines data from the Baidu Index platform and uses web crawling keywords “environmental pollution” to characterize the extent of public attention and discussion on environmental governance [36].

In

Table 9. Mechanism analysis.

	(1)	(2)	(3)	(4)
	Environmental Regulation Constraint	Industrial Structure Upgrading	Public Participation Supervision	Innovation and Entrepreneurship Driven
DID	−0.093 **	0.092 ***	0.043 ***	0.276 ***
	(−2.49)	(5.87)	(3.38)	(7.37)
Control Variables	Yes	Yes	Yes	Yes
City Fixed Effects	Yes	Yes	Yes	Yes
Year Fixed Effects	Yes	Yes	Yes	Yes
N	3990	3990	3990	3990
r2	0.834	0.309	0.190	0.850

z statistics in parentheses; ** p < 0.05, *** p < 0.01.

, public data openness can alleviate information friction between enterprises and the government, increasing the government’s sensitivity to polluting enterprise projects. By integrating the environmental planning and policy implementation of both central and local governments, it enhances the local governments’ comprehensive governance capabilities, including pre-planning, mid-control, and post-management. This combines a “bottom-up” and “top-down” approach to formulating environmental policies, thereby improving environmental welfare performance through the effect of environmental regulation. Hypothesis 2 is supported.

The results in column (2) indicate that the DID estimated coefficient is 0.092 and is significant at the 1% significance level, meaning that a 1% increase in the degree of public data openness can be achieved through a 9.2% industrial structure upgrade effect, thereby improving environmental welfare performance. Public data openness unleashes the benefits of data elements, strengthens coordination and collaboration between industries, and enhances the efficiency of factor allocation [37]. At the same time, enterprises obtain market information, analyze development prospects, and seize investment opportunities through public data openness, accelerating the transition from labor-intensive low-end industries to high-tech industries. Therefore, industrial structure adjustment is an important link between data elements and ecological protection [38]. Hypothesis 3 is validated.

The results in column (3) indicate that the openness of public data can improve environmental welfare performance through the effect of public participation and supervision. Leveraging the public data openness platform to promote two-way interaction between the government and the public, and using external public supervision to force local governments to implement effective measures, enhances environmental welfare performance. Hypothesis 4 is confirmed.

As shown in column (4), the estimated coefficient of DID is 0.276, which is significant at the 1% level. On one hand, public data openness helps enterprises, research institutions, and other innovative entities broaden their channels for knowledge and information sources. By combining this with relevant artificial intelligence algorithms and data analysis techniques, they can fully explore and effectively utilize the potential value of public data. The important innovation elements formed can also combine with data elements to shape new combinations and innovation models, fully leveraging the multiplier effect on environmental protection efforts. On the other hand, the public data openness platform discloses scientific research results, industry reports, and other data resources, providing certain information support for entrepreneurs. This encourages more entrepreneurs to actively engage in social production activities, enhancing urban entrepreneurial activity. Guided by concepts such as “integrating technological innovation resources, leading the development of strategic emerging industries and future industries, accelerating the formation of new quality productive forces”, and “timely applying technological innovation achievements to specific industries and industrial chains, transforming and upgrading traditional industries, nurturing and growing emerging industries, and planning for future industries”, entrepreneurs often tend to choose high-growth, high-tech but low-energy-consuming and low-polluting industries. By focusing on both “stabilizing growth” and “optimizing the environment”, they improve environmental welfare performance. Hypothesis 5 is confirmed.

5.2. Spatial Effect Analysis

5.2.1. Spatial Spillover Effect Test

Taking into account spatial dependence, this paper constructs a spatial weight matrix and adopts a spatial difference-in-differences model based on the SAR model to test the spatial correlation effect of public data openness on improving environmental welfare performance. The specific model is as follows:

$$Ew{p}_{it}={\alpha }_0+{\alpha }_{1}DI{D}_{it}+\rho W\_Ew{p}_{it}+{\alpha }_2{X}_{it}+{\lambda }_i+{\mu }_t+{\epsilon }_{it}$$

(8)

ρ represents the spatial coefficient to be estimated; W is the spatial weight matrix. Considering that the explanatory power of ordinary “0–1” spatial adjacency matrices is relatively weak in large samples, and according to “The First Law of Physics”, points that are closer to each other are more similar, while points that are farther apart are less correlated; this paper constructs a reverse distance matrix by calculating the centroid distance between two cities to test for spatial spillover effects, i.e., $W=\left\{\begin{array}{ll}1/d_{ij}\hfill & \hfill i \ne j\\ 0\hfill & \hfill i = j \end{array}\right.$; d_ij represents the geographical distance between region i and region j. The remaining variables are consistent with those previously mentioned.

The empirical results are shown in column (1) of

Table 10. Spatial spillover effect test.

	(1)	(2)	(3)	(4)
	E_wp	Direct Effect	Indirect Effect	Total Effect
DID	0.022 ***	0.022 ***	−0.015 ***	0.007 ***
	(3.788)	(3.725)	(−3.690)	(3.594)
W_Ewp	−2.041 ***
	(−9.649)
Control Variables	Yes	Yes	Yes	Yes
City Fixed Effects	Yes	Yes	Yes	Yes
Year Fixed Effects	Yes	Yes	Yes	Yes
N	3990
r2	0.125

z statistics in parentheses; *** p < 0.01.

. Taking into account the spatial externalities of regional environmental welfare performance, the estimated coefficient value of public data openness on environmental welfare performance remains positive and is significant at the 1% level; compared to the baseline regression results, the estimated coefficient of the policy dummy variable has decreased, indicating that the effect of public data openness on improving environmental welfare performance is overestimated when spatial spillover effects are not considered. The estimated coefficient of the spatial lag term W_Ewp for environmental welfare performance is negative, suggesting that environmental welfare performance has a certain cross-boundary effect, and an increase in local environmental welfare performance levels leads to a decrease in neighboring regions’ environmental welfare performance levels. Under the premise of limited green resources, there is a certain competition for resources between local and neighboring areas, exacerbating the “beggar-my-neighbor” phenomenon.

To accurately measure the spatial spillover effect of public data openness on environmental welfare performance, this paper further calculates the direct, indirect, and total effects of public data openness on environmental welfare performance. The results are reported in columns (2)–(4). This study finds that public data openness can directly improve local environmental welfare performance but reduces the environmental welfare performance of neighboring regions. Regions with better public data openness easily attract factor resources to cluster spatially. With the support of a favorable market environment, high-quality talent flows in, high-tech applications accelerate breakthroughs, entrepreneurial vitality increases, and resource allocation efficiency is optimized. This process may produce a “siphon effect”, leading to a decrease in the positive effects of the public data openness platform environment in surrounding urban areas.

5.2.2. Spatial Decay Boundaries

In light of the decomposition results of the spatial spillover effects mentioned above, this paper constructs a spatial adjacency matrix based on a base period of 400 km and step distances of 100 km and conducts continuous regression to determine the spatial decay boundaries of the spillover effects of public data openness on environmental welfare performance within different spatial distance ranges. As shown in Figure 7, the coefficient estimates of public data openness are significantly negative within a 1000 km range and pass at least the 10% significance test. The negative impact of public data openness on the environmental welfare performance of neighboring regions mainly exists within 1000 km. The reason may be that the opening of public data is more likely to trigger regional resource reallocation effects. When central cities form information hubs based on data opening policies, they generate a triple “siphon effect” in surrounding areas by reducing factor matching costs and enhancing innovation premium capabilities. This mechanism involves the flow of talent to high-value-added industries, the allocation of risk capital to data-intensive areas, and the transfer of pollution-intensive industries to hinterland regions [39]. For example, the “Belt and Road” central cities have a “siphon effect” on surrounding areas [40]. Spatial constraints stem from economic geography laws, with a 1000 km radius covering the critical point for high-speed rail commuting costs within three hours and the maximum radius for regional industrial chain collaboration (as revealed by empirical studies in the Yangtze River Delta, where the core–hinterland linkage range is 800–1000 km). Beyond this distance, independent economic subsystems form (such as the Beijing–Tianjin–Hebei radiation system), and the differentiated evolution of environmental welfare is the explicit outcome of the aforementioned mechanisms.

6. Conclusions and Implications

This paper takes the gradual launch of urban public data platforms as its research context and employs a multi-period difference-in-differences model to empirically study the impact of the launch of municipal-level public data platforms on environmental welfare performance. The study finds, firstly, that access to urban public data helps improve environmental welfare performance, and this conclusion remains valid even after controlling for endogeneity issues and conducting a series of robustness tests. Secondly, public data openness plays a more positive role in cities in eastern China, cities with higher local government fiscal freedom, and cities where local governments place greater emphasis on environmental protection work. Thirdly, mechanism analysis results show that the promoting effect of public data openness on environmental welfare performance can be achieved through four paths: environmental regulation constraints, industrial structure upgrading, public participation supervision, and innovation and entrepreneurship incentives. Finally, public data openness can significantly improve local environmental welfare performance, but this process may produce a “siphon effect”, leading to reduced environmental welfare performance in surrounding urban areas, and this effect has certain geographic decay characteristics, with a spatial decay boundary of 1000 km.

6.1. Policy Implications

Based on the above conclusions, the following policy implications can be drawn:

First, deepen and advance the reform of the public data openness strategy to fully tap into the potential of public data as a new source of productive capacity. At the foundational level, enhance the accuracy, comprehensiveness, and security of environmental data collection, establish unified data standards and norms, and prevent data misuse or symbolic openness without substantive content. At the support layer, streamline the entire process of environmental governance data “monitoring-collection-integration-analysis-publication-sharing-prediction” and utilize advanced technologies such as big data and artificial intelligence to achieve scientific, accurate, and reasonable environmental decision-making and management. At the application layer, promote the innovative development of ecological and environmental data integration and application, actively promote and guide cooperation and communication among innovative entities such as universities, enterprises, and research institutions, establish an ecological and environmental data integration and application innovation alliance, and gather new momentum for the development of green productivity.

Next, explore multi-dimensional pathways to enhance environmental welfare performance. First, strengthen environmental regulatory systems based on the principle of rule-of-law governance. Local governments should establish regulatory guidelines targeting key areas and critical stages of environmental governance, clearly defining procedures for the collection, use, and sharing of environmental data throughout the entire process. Measures such as data encryption and access controls should be implemented to prevent the leakage, theft, or tampering of critical data. Sensitive and non-sensitive information should be categorized to ensure the security of data throughout the entire process, from collection to application. Second, guide the synergistic interaction of data elements with other production factors to empower industrial transformation and upgrading. Data, with its unique advantages, can rapidly and reasonably integrate into various stages of production processes, distribution mechanisms, circulation channels, consumption links, and social service management. By combining different production factors and collaborating with various entities to leverage multiplier effects, it can achieve a leap in productivity and deep industrial transformation. Third, encourage the public to actively participate in environmental governance and respond promptly to public opinion. Leverage the advantages of data and information dissemination to expand the depth and breadth of public participation in environmental governance, fully utilize the supervisory role of the public, and truly achieve the goal of “serving the people”. Fourth, guide enterprises to utilize public data for innovation and entrepreneurship to cultivate digital productivity. Increase the intensity of government fiscal subsidies for enterprises’ digital innovation and entrepreneurship activities. Through fiscal incentives, leverage the leading role of local governments to provide safeguards for high-tech enterprises with the willingness and ideas, but constrained by information acquisition costs to enter the market.

Finally, establish data sharing and collaboration mechanisms to construct a regional urban environmental governance matrix. Urban environmental welfare performance exhibits significant negative spatial spillover effects and diminishing marginal characteristics, which necessitate that cities establish cross-administrative region data governance alliances and ecological compensation mechanisms to build regional coordinated development pathways, create cross-regional environmental governance communities, and gradually internalize spatial negative externalities within the community to avoid welfare imbalances caused by resource concentration.

6.2. Research Limitations

This study has the following limitations. The public data platform openness policy was simplified into a binary variable, failing to capture the heterogeneity among cities in terms of the depth, scope, and quality of data openness. Although constrained by data availability, this approach may have weakened the differentiated mechanisms of policy effects. Future research should incorporate indicators such as the frequency of platform dataset downloads to deepen the analysis. Although benchmark regression and multiple robustness tests support the conclusion that public data openness has an ecological value release effect, our findings may still be influenced by unobserved confounding factors. Subsequent research will further explore this to enhance the precision of the conclusions.