Spatial–Temporal Development Trends and Influencing Factors of Government Environmental Information Disclosure: Empirical Evidence Based on China’s Provincial Panel Data

Abstract

Government environmental information disclosure (GEID) plays an important role in promoting the digital transformation of environmental governance, leading the concept of sustainable development, enhancing public oversight capacity, and promoting democratic decision-making governance. Using provincial panel data from China spanning from 2009 to 2021, we conducted spatial data exploratory analysis and used the dynamic spatial panel model to investigate the spatial–temporal development trends and influencing factors of GEID. The results show that (1) GEID in China exhibits significant spatial agglomeration characteristics, with an “H-H” (High-High aggregation type) agglomeration characteristic observed in three national strategic development regions: Yangtze River Delta, southeast coastal areas, and Beijing–Tianjin–Hebei region. (2) The spillover effect from the southeast coastal provinces gradually radiates to the northwest, resulting in an overall westward movement of GEID. (3) GEID exhibits a significant path-dependency feature in the temporal dimension and a “peer effect” in the spatial dimension. (4) Population size has the greatest impact on GEID. Population size, public participation, and the industrial and transportation sectors positively influence GEID improvement at the local level. However, they generate negative spillover effects to neighbouring provinces. Environmental status and the size of the Real Estate sector have no significant effect. Therefore, China should strengthen regional cooperation, narrow regional disparities, cultivate new quality productive forces, establish a government-led proactive disclosure mechanism under public supervision, and improve the level of GEID at the national level.

1. Introduction

Government environmental information disclosure (GEID) is an emerging environmental management system that promotes the digital transformation of ecological and environmental governance. It represents the “third wave” of environmental regulation following command-and-control and market-incentive environmental regulations. GEID refers to a system in which the government voluntarily discloses environmental information it holds to the public within the scope and requirements stipulated by law and procedures. This practice is conducive to fostering a new governance system characterised by “plural subject co-governance and sharing” and is a key factor in enhancing environmental governance capabilities [1,2]. In recent years, GEID has gained traction and widespread adoption due to its advantages in mitigating information asymmetry and protecting the public’s environmental rights to know, participate, and supervise [3,4]. Developed countries such as Europe, the United States, and Japan have established comprehensive GEID systems, including legal guarantee systems and specific implementation mechanisms. Their practices have demonstrated that GEID has a positive impact on promoting environmental governance and reducing pollution [5]. In contrast, China, as the largest developing country in terms of economic scale, has long been considered to have poor environmental information [6]. China lags behind developed countries in terms of GEID due to constraints within the political system, significant regional development disparities, weak environmental infrastructure, and insufficient public participation. Prevalent issues include non-standardised content, unclear information classification, and a significant Matthew effect between different regions [7,8]. Therefore, studying the spatial–temporal development trends and influencing factors of GEID in China is of significant importance for developing countries to enhance their environmental management capabilities and establish diversified environmental governance systems.

GEID is a complex social issue involving environmental protection, economic development, population size, technological level, and political systems. Scholars have conducted research on the factors influencing GEID from different perspectives. Based on the institutional perspective, Chen et al. [9] concluded that there is no significant correlation between the social–political background and GEID through an experimental questionnaire survey. Based on the policy perspective, Yang et al. [10] empirically verified that increasing GEID can effectively enhance environmental governance performance. Based on the technological progress perspective, Zhang and Huang [11] demonstrated through quasi-natural experiments that green innovation significantly promotes GEID. According to stakeholder theory, current research mainly focuses on the relationship between single influencing factors of enterprises, the public, the government, and GEID. At the enterprise level, the literature demonstrates that factors such as enterprise size, industry type, and enterprise location can significantly impact the level of GEID [12,13]. At the public level, public participation is a key driver of GEID. Empirical studies have shown that regions with higher education levels tend to express more environmental demands, which in turn promotes the improvement of the level of GEID [14]. At the government level, local governments are more motivated to disclose environmental information when there is a higher degree of fiscal freedom, investment in environmental governance, and economic development [15,16]. However, current research on the influencing factors of GEID has primarily relied on single perspectives and case studies of individual entities. It is important to note that GEID is influenced by multiple factors working together, and no single factor can act independently of others. Additionally, there are significant spatial differences in GEID that exist due to regional development disparities. However, few scholars have studied the spatial effects and spatial–temporal development trends of GEID from a dynamic perspective.

Therefore, this study, which is based on the improved STIRPAT model by York et al. [17], is better suited than previous research to comprehensively understand the underlying mechanisms of GEID in theory and explore the relationships among the diverse stakeholders involved. Additionally, this study will combine ArcGIS 10.8 software to deeply explore the multiple factors influencing GEID from a dynamic spatial perspective. In summary, this study comprehensively examines the spatio-temporal evolution of GEID, utilising both Moran’s I index and spatial Standard Deviation Ellipse analysis. Building upon the STIRPAT theoretical framework, we systematically identified and decomposed the key factors that potentially influence GEID. To further elucidate the mechanisms through which these factors exert their impact, we employed both the spatial Durbin model (SDM) and the dynamic spatial Durbin model (DSDM). The framework is presented in Figure 1. Compared with the existing literature, this paper mainly contributes to the following three aspects: Firstly, it studies the spatial agglomeration level, form, pattern, and development trends of GEID in China from 2009 to 2021 at the provincial scale and visualises the results with ArcGIS for the first time. Secondly, this study improves the STIRPAT model by incorporating additional factors such as population, wealth, technological level, policy focus, public participation, environmental factors, and industrial development that may affect GEID. This is theoretically conducive to an overall understanding of the internal mechanisms of GEID, discovering the relationships among diverse stakeholders involved, and enriching the research content of GEID. Thirdly, this paper conducted an empirical study on the impact of multiple factors on GEID using both static and dynamic spatial Durbin models and decomposed the effects into direct and indirect effects to investigate the influence of various factors on the GEID from a spatial spillover perspective. This research has significant practical implications for optimising environmental information disclosure policies, engaging the public in environmental governance, and establishing service-oriented government.

2. Methods and Data

2.1. Methods

2.1.1. STIRPAT Model Improvement and Variable Design

The IPAT model was proposed by the renowned American demographer Ehrlich et al. in 1971 to measure the relationship between environmental impact (I) and population (P), affluence (A), and technology (T) [18]. This model overcomes the shortcomings of single-factor determinism, providing greater explanatory and persuasive power. Researchers have widely applied the model to study the impact of multiple factors on the environment. It can be represented as

$$I=PAT$$

(1)

Based on the IPAT model, York et al. [17] overcame the limitation that the traditional IPAT model could not measure the influence of various variables by introducing differential elastic and random error, obtaining the STIRPAT model as follows:

$$I=\alpha P^b{A}^c{T}^d\epsilon$$

(2)

where α is the equation coefficient; b, c, and d are parameters of population, affluence, and technology, respectively; and ε is the random error.

The STIRPAT model enables the decomposition of various influencing factors for adapting to environmental impact studies under different scenarios. This paper uses the GEID level as the explained variable. Based on the framework of the STIRPAT model and relevant research, the explanatory variables in this paper include the following:

(1)

Population size. Previous studies have shown that population has a significant impact on the environment, which mainly includes three perspectives. The first perspective suggests that population growth has a disastrous effect on the environment [19]. The second perspective suggests that population growth promotes technological advancement, resulting in a neutral or even positive environmental effect [20]. This paper aligns with the third perspective, which considers population density as one of the explanatory variables. It posits that population is not the dominant factor influencing GEID but plays a role with other factors like affluence and technology [21].

(2)

Affluence. Relevant studies indicate that economic disparities are a significant factor leading to regional imbalances in China’s GEID. The level of regional economic development determines the general revenue of local governments, thereby influencing the level of GEID [5]. This paper uses general budget revenue of the local governments as a measure of affluence to investigate its impact on GEID.

(3)

Technical innovation. Technological innovation can enhance the efficiency of information acquisition and processing, enabling the local governments to disclose environmental information more promptly and accurately [22]. However, there is limited research on the spatial effect of technical innovation on GEID. To address this gap, this paper uses the quantity of green patent grants as a proxy indicator for technical innovation, drawing on the study by Du et al. [23].

(4)

Public participation. Public participation is a crucial factor in driving GEID. Relevant studies show that in places with higher income and education levels, the public has stronger requirements for GEID [24]. Therefore, this paper uses the average number of students enrolled in high school per 100,000 people as the measure of public participation.

(5)

Environmental regulation. The political performance evaluations of local officials are green-oriented, leading to strengthened government efforts to enhance environmental protection through regulations. This has resulted in an improvement in GEID [25]. Therefore, this paper uses the environmental attention index to measure the intensity of government environmental regulations and investigate their impact on GEID.

(6)

Environmental status. Signalling theory suggests that governments with better environmental performance are more inclined to disclose more information [26]. Therefore, this paper follows the existing literature and uses SO₂ concentration as the indicator of environmental status [27].

(7)

Industrial structure. Excessive representation of high-emission industries will increase the local government’s environmental pressure, thus inhibiting the initiative of GEID. Considering China’s industrial development, this paper chooses industrial added value as a share of GDP, transportation industry added value as a share of GDP, and Real Estate added value as a share of GDP as indicators of industrial structure, which have long been regarded as high-emission industries and pillar industries in China [28].

In conclusion, the specific measurement indicators for all variables are presented in

Table 1. Variable declaration and definition.

Variable	Symbol	Definition	Unit
Government environmental information disclosure	GEID	Local government environmental information disclosure level	-
Population size	P	Population density	Person/km2
Affluence	A	General budget revenue of local governments	100 million RMB
Technical innovation	T	Number of green patents granted	Pcs
Public participation	PU	Average number of students enrolled in high school per 100,000 people	Person
Environmental regulation	ER	Environmental attention, number of environmental words in government documents	Pcs
Environmental status	ES	Total SO2 emissions	Tonne
Industry	IND	Industrial added value as a share of GDP	%
Transportation industry	TI	Transportation industry added value as a share of GDP	%
Real Estate	RE	Real Estate added value as a share of GDP	%

. Based on the original STIRPAT model, this paper extends the model as follows:

$$GEID=\alpha P^{{\beta }_1}{A}^{{\beta }_2}{T}^{{\beta }_3}P{U}^{{\beta }_4}E{R}^{{\beta }_5}E{P}^{{\beta }_6}IN{D}^{{\beta }_7}T{I}^{{\beta }_8}R{E}^{{\beta }_9}\epsilon$$

(3)

where α is the constant term; β_n is the estimated coefficient for the corresponding explanatory variables; and ε is the random error term. Take the natural logarithm of both sides for linearisation, and the result is as follows:

$$\begin{array}{cc}\hfill \ln GEID& =\alpha +{\beta }_1\left(\ln P\right)+{\beta }_2\left(\ln A\right)+{\beta }_3\left(\ln T\right)+{\beta }_4\left(\ln PU\right)+{\beta }_5\left(\ln ER\right)+{\beta }_6\left(\ln EP\right)\hfill \\ & +{\beta }_7\left(\ln IND\right)+{\beta }_8\left(\ln TI\right)+{\beta }_9\left(\ln RE\right)+\epsilon \hfill \end{array}$$

(4)

2.1.2. Spatial Correlation Test

Based on the fundamental principles of spatial economics, this paper explores the spatial interactions of GEID among different regions using the spatial dimension’s interactions and the temporal dimension’s correlations. To achieve this, we employ both the global Moran’s I index and the local Moran’s I index. The formula is presented below:

$$\mathrm{Moran}\prime \mathrm{s} I={\displaystyle \frac{n{\displaystyle \sum _{i=1}^n{\displaystyle \sum _{j=1}^n{W}_{ij}\left(Y_i-\overline{Y}\right)}}\left(Y_j-\overline{Y}\right)}{{\displaystyle \sum _{i=1}^n{\displaystyle \sum _{j=1}^n{W}_{ij}{\left(Y_i-\overline{Y}\right)}^2}}}}$$

(5)

where n is the total number of regions; Y_i, Y_j represent the coupling coordination degree between different regions; $\overline{Y}$ is the average value; W_ij is the spatial weight matrix; and Moran’s I∈[−1,1].

2.1.3. Standard Deviation Ellipse Analysis

This paper uses the SDE method to investigate the spatial centrality, spatial morphology, and spatial–temporal development trend of GEID from a global and dynamic perspective. The mean centre of the ellipse indicates the relative position of GEID in two-dimensional space, while the changing trajectory of the mean centre can illustrate the overall spatial movement trend of GEID. The azimuth reflects the main trend direction of its distribution. The calculation formula is as follows [29]:

$${\overline{x}}_w={\displaystyle \frac{{\displaystyle \sum _{i=1}^n{w}_i{x}_i}}{{\displaystyle \sum _{i=1}^n{w}_i}}}; {\overline{y}}_w={\displaystyle \frac{{\displaystyle \sum _{i=1}^n{w}_i{y}_i}}{{\displaystyle \sum _{i=1}^n{y}_i}}}$$

(6)

$$\tan \theta ={\displaystyle \frac{\left({\displaystyle \sum _{i=1}^n{w^2}_i{{\tilde{x}}_i}^2-}{\displaystyle \sum _{i=1}^n{w^2}_i{{\tilde{y}}_i}^2}\right)+\sqrt{{\left({\displaystyle \sum _{i=1}^n{w^2}_i{{\tilde{x}}_i}^2-}{\displaystyle \sum _{i=1}^n{w^2}_i{{\tilde{y}}_i}^2}\right)}^2+4{\displaystyle \sum _{i=1}^n{w^2}_i{\tilde{x}}_i{\tilde{y}}_i}}}{2{\displaystyle \sum _{i=1}^n{w^2}_i{\tilde{x}}_i{\tilde{y}}_i}}}$$

(7)

where ${\overline{x}}_w$ is the mean centre; (x_i,y_i) is the spatial location of the research object; w_i is the weight; (${\overline{x}}_w$,${\overline{y}}_w$) is the weighted mean centre; θ is the azimuth; and ${\tilde{x}}_i$, ${\tilde{y}}_i$ represent the coordinate deviation from the location of the research object to the mean centre.

2.1.4. Spatial Panel Model

Based on the dynamic nature of GEID, the level of GEID is affected by both current and past factors, and there exists a competitive effect in environmental policies among neighbouring provinces, which is prone to path dependence and Matthew effect [30]. Therefore, based on the STIRPAT model, this paper employs a dynamic spatial Durbin model to investigate the spatial–temporal characteristics and spillover effects of the influencing factors of inter-provincial GEID. The model as follows [31]:

$$\begin{array}{cc}\hfill \ln GEI{D}_{it}& =\alpha +\chi \ln GEI{D}_{it-1}+\rho W\times \ln GEI{D}_{it}+\gamma W\times \ln GEI{D}_{it-1}+{\beta }_1\ln P_{it}\hfill \\ & +{\beta }_2\ln A_{it}+{\beta }_3\ln T_{it}+{\beta }_4\ln P{U}_{it}+{\beta }_5\ln E{R}_{it}+{\beta }_6\ln E{P}_{it}\hfill \\ & +{\beta }_7\ln IN{D}_{it}+{\beta }_8\ln T{I}_{it}+{\beta }_9\ln R{E}_{it}+{\theta }_1{\displaystyle \sum _{j=1}^N\left(W_{ij}\ln P_{ijt}\right)}\hfill \\ & +{\theta }_2{\displaystyle \sum _{j=1}^N\left(W_{ij}\ln A_{ijt}\right)}+{\theta }_3{\displaystyle \sum _{j=1}^N\left(W_{ij}\ln T_{ijt}\right)}+{\theta }_4{\displaystyle \sum _{j=1}^N\left(W_{ij}\ln P{U}_{ijt}\right)}\hfill \\ & +{\theta }_5{\displaystyle \sum _{j=1}^N\left(W_{ij}\ln E{R}_{ijt}\right)}+{\theta }_6{\displaystyle \sum _{j=1}^N\left(W_{ij}\ln E{P}_{ijt}\right)}+{\theta }_7{\displaystyle \sum _{j=1}^N\left(W_{ij}\ln IN{D}_{ijt}\right)}\hfill \\ & +{\theta }_8{\displaystyle \sum _{j=1}^N\left(W_{ij}\ln T{I}_{ijt}\right)}+{\theta }_9{\displaystyle \sum _{j=1}^N\left(W_{ij}\ln R{E}_{ijt}\right)}+{\mu }_i+{\upsilon }_t+{\epsilon }_{it}\hfill \end{array}$$

(8)

where GEID_it is the level of GEID of province i in year t; GEID_it−1 denotes its lagged value by one period; χ is the time lag coefficient of GEID; ρ is the spatial lag coefficient of GEID; γ is the spatial–temporal lag coefficient of GEID; θ_n is the spatial lag coefficient of explanatory variables; μ_i is the individual effect; υ_t is the time effect; ε_it is the random disturbance term; and W is the spatial weight matrix. In the selection of the spatial weight matrix W, this paper primarily focuses on the geographical adjacency matrix, supplemented by the economic distance matrix and geographical distance matrix for spatial econometric analysis.

2.2. Data Sources

The explained variable, GEID, in this paper is derived from the PITI score jointly published by the Institute of Public and Environmental Affairs (IPE) and the Natural Resources Defence Council (NRDC) from 2009 to 2021. This comprehensive index evaluates the stringency of environmental pollution information disclosure for over 100 major cities nationwide in terms of (1) records of enterprise violations, (2) results of “enforcement campaigns” against polluting enterprises, (3) clean production audit information, (4) enterprise environmental performance ratings, (5) disposition of verified petitions and complaints, (6) environmental impact assessment (EIA) reports and project completion approvals, (7) discharge fee data, and (8) response to public information requests. The PITI index is by far the best proxy for urban GEID in China and has been widely used in prior studies [32,33]. The GEID values of each province are averaged based on the PITI scores of the sample cities in the respective province.

The explanatory variables in this paper, population size (P), affluence (A), public participation (PU), industry (IND), transportation industry (TI), and Real Estate (RE), are sourced from the statistical yearbooks from 2010 to 2022. Technical innovation (T) and environmental regulation (ER) are obtained from the Chinese Research Data Services Platform (CNRDS); environmental status (ES) data are sourced from the China Environmental Statistical Yearbook. To mitigate the limitations imposed by data gaps on a national-scale analysis of GEID levels, this study employed interpolation methods to fill in individual discontinuous and missing data points. Nevertheless, significant data absence persists for Hainan, Tibet, Hong Kong, Taiwan, and Macau, leading to their exclusion from the present investigation. This exclusion may inherently constrain the comprehensiveness of our nationwide discussion on GEID levels.

3. Results

3.1. Spatial Correlation Test and Agglomeration Characteristics of GEID

3.1.1. Global Spatial Correlation Analysis

To verify the spatial correlation and temporal development trends of inter-provincial GEID accurately and intuitively, we used the geographical adjacency matrix to calculate the global Moran’s I of GEID from 2009 to 2021, and the results are shown in

Table 2. Global Moran’s I of 29 investigated provinces from 2009 to 2021.

Year	Moran’s I	p-Value	Z-Value
2009	0.299	0.006	2.735
2010	0.358	0.001	3.242
2011	0.282	0.009	2.595
2012	0.302	0.006	2.759
2013	0.553	0.000	4.910
2014	0.422	0.000	3.753
2015	0.356	0.001	3.269
2016	0.507	0.000	4.446
2017	0.537	0.000	4.685
2018	0.480	0.000	4.173
2019	0.425	0.000	3.801
2020	0.501	0.000	4.507
2021	0.585	0.000	5.190

.

The calculation results show that the Moran’s I value for all study years is greater than 0 and passes the significance test at the 1% level with Z-values exceeding 2.58. The analysis suggests a significant positive spatial correlation among the GEID of Chinese provinces. It implies that the GEID of provincial governments are not isolated and randomly distributed but are influenced by neighbouring provinces, demonstrating a clear agglomeration distribution pattern spatially. This positive spatial dependency among provinces also exhibits periodic fluctuations, showing an overall increasing trend of fluctuation. This indicates that the spatial dependency among provinces in terms of GEID is gradually strengthening, with leading provinces gradually demonstrating their exemplary role.

3.1.2. Local Spatial Correlation Analysis

Global spatial correlation analysis revealed significant agglomeration of GEID from 2009 to 2021. To further investigate, Moran scatter diagrams were plotted for the years 2009 to 2021 to examine the local spatial characteristics of GEID. As shown in Figure 2, the spatial aggregation characteristics of GEID in 29 provinces are mainly “H-H” (High-High aggregation type) and “L-L” (Low-Low aggregation type). The proportion of provinces belonging to these two types increased from 55% in 2009 to 86% in 2021, indicating the spatial spillover effect was significant and the scale of spatial agglomeration gradually expanded.

From 2009 to 2021, the exemplary influence exerted by neighbouring provinces exerted a salutary effect on the enhancement of GEID, causing Tianjin and Henan to transition from “L-H” to “H-H”. Meanwhile, the lower GEID in their own provinces prompted Hunan and Guizhou to transition from “L-H” to “L-L”; The radiating effect of higher GEID in their own provinces facilitated Hebei and Beijing to transition from “H-L” to “H-H”. The ripple effect of lower GEID in neighbouring provinces prompted Yunnan and Heilongjiang to transition from “H-L” to “L-L”.

3.1.3. Spatial Agglomeration Characteristic Analysis

Moran scatter plots do not account for the significance of agglomeration levels; therefore, LISA agglomeration maps were created using ArcGIS software to visually analyse the agglomeration characteristics. As Figure 3 shows, at a significance level of 10%, the agglomeration characteristics of GEID evolved from four types of spatial characteristics to only “H-H” and “L-L”, with the spatial positive correlation agglomeration features being further emphasised. In 2009, the “H-H” included Zhejiang, Jiangsu, and Shanghai. In 2013, the “H-H” expanded to include Anhui, Beijing, and Tianjin, covering all cities in the Yangtze River Delta. In 2017, the “H-H” extended northward, forming a agglomeration radiation belt from south to north in the southeastern coastal region, including Fujian, Zhejiang, Shanghai, Jiangsu, Shandong, Tianjin, and Beijing. In 2021, the “H-H” formed the agglomeration features of Beijing, Tianjin, and Hebei.

The changes in the “H-H” from 2009 to 2021 are highly consistent with adjustments in China’s political and economic policies. During this period, the Chinese government successively formulated major national development strategies such as the YD Integrated Development (strategy for integrated development of the Yangtze River Delta), coastal development strategies, and BTH coordinated development (Beijing–Tianjin–Hebei coordinated development). These strategies were crucial in the regional economic development, industrial transformation, and technological innovation. Consequently, GEID was also influenced by policies, which led to the formation of the Yangtze River Delta agglomeration belt, southeast coastal agglomeration belt, and Beijing–Tianjin–Hebei agglomeration belt.

Despite the emergence of “H-H” spatial agglomeration patterns in regions such as Beijing–Tianjin–Hebei and the southeast coast between 2009 and 2021, a majority of areas in China have yet to exhibit significant spatial agglomeration, with some provinces in the northwest even demonstrating “L-L” agglomeration. These findings underscore substantial spatial disparities in GEID across China’s provinces, with the southeast coastal regions exhibiting notably higher GEID levels compared to their central and western counterparts. This phenomenon is largely attributable to the imbalanced economic development between the eastern and western regions, a consequence of policy preferences favouring coastal areas.

3.2. Spatial–Temporal Trends in the Agglomeration of GEID

This paper quantitatively explains the spatial centrality, spatial morphology, and spatial–temporal development trend of GEID based on the spatial location and structure using the SDE (Standard Deviation Ellipse) method from a global, dynamic spatial–temporal perspective. The mean centre represents the relative position of GEID in two-dimensional space, while the azimuth represents the main trend direction of its distribution.

The SDE, mean centre, and migration path of GEID from 2009 to 2021 were plotted using ArcGIS software [34]. The results in Figure 4 show that the overall GEID in China shows a tendency to move upward in the northwest direction; the area of the ellipse is gradually expanding, indicating that the level of GEID outside the ellipse is outpacing that within, suggesting that provinces beyond the traditional high-performing clusters are actively catching up. This trend underscores the potential for regional spillover effects, where policy innovations and best practices adopted by leading provinces may inspire and accelerate improvements in neighbouring regions. For future regional planning, policymakers should capitalise on this momentum by fostering inter-regional cooperation and knowledge-sharing mechanisms to accelerate the diffusion of successful GEID practices.

In terms of the location of the mean centre, the mean centre of the GEID ellipse shows a general northwest movement, starting in Luohe City in Henan Province, passing through Xuchang City and Pingdingshan City, and finally moving near the junction of Pingdingshan City, Zhengzhou City, and Luohe City, with a total movement of 133 km. This shift in the mean centre signifies a growing recognition among central and western provinces of the strategic importance of GEID. This shift highlights the need for targeted interventions and policies aimed at supporting these regions in their efforts to enhance disclosure and transparency. Future regional planning should prioritise investments in infrastructure, human capital, and technology to build the capacity of central and western provinces to effectively implement and sustain GEID improvements.

In terms of distribution shape and orientation, the semi-major axis of the GEID ellipse is gradually shortening while the semi-major axis is lengthening, resulting in an increasing elliptical-shaped index approaching a perfect circle. This indicates that the level of GEID has become more balanced in the radial direction. The ellipse spreads in the northeast–southwest direction, with the azimuth increasing annually (from 36.68° to 44.61°), suggesting an increasing agglomeration effect in the northeast–southwest direction of GEID, causing the clockwise rotation of the distribution.

Furthermore, the increasing influence of leading provinces underscores the role of policy leadership in driving national-level improvements in GEID. Policymakers in high-performing regions should be encouraged to share their successes, challenges, and lessons learned with their counterparts in other regions. Collaborative initiatives and partnerships aimed at promoting the exchange of best practices and fostering a culture of continuous improvement can significantly contribute to the overall advancement of GEID in China. By fostering inter-regional cooperation, prioritising targeted investments, and leveraging the expertise of policy leaders, China can harness these trends to drive inclusive and sustainable improvements in GEID nationwide.

3.3. Analysis of Factors Influencing the Level of GEID

3.3.1. Spatial Panel Estimation Results

Before estimating model parameters, model selection diagnostics are required. This paper sequentially conducts Lagrange Multiplier (LM) test, Likelihood Ratio (LR) test, Hausman test, and Wald test on Formula (8) to determine the specific forms of static and dynamic spatial econometric models. Firstly, by performing OLS estimation on a model without spatial effects, the Lagrange Multiplier (LM) and its robust statistic (R-LM) are obtained. If the majority of the results are significant, the spatial Durbin model (SDM) is selected. Secondly, the spatial Durbin model is subjected to a Hausman test to determine the presence of fixed effects or random effects in the model. Thirdly, the Likelihood Ratio (LR) test is used to examine the fixed-effect type of the model. If all results are significant, it indicates a spatial–temporal double-fixed Durbin model. Fourthly, Wald or LR test is conducted on the spatial Durbin model to determine whether the model will weaken to an SAR or SEM model, and due to the robustness of the results, this paper adopts the two methods of Wald and LR to conduct the test.

The test results (

Table 3. Model applicability and model selection test results.

W		Wa		We		Wd
Test		Statistical Value	p-Value	Statistical Value	p-Value	Statistical Value	p-Value
LM-error	Moran’s I	10.241	0.000	10.109	0.000	18.808	0.000
	LM	97.067	0.000	95.408	0.000	289.839	0.000
	Robust LM	1.178	0.278	0.260	0.610	140.239	0.000
LM-lag	LM	117.989	0.000	129.200	0.000	165.738	0.000
	Robust LM	22.100	0.000	34.053	0.000	16.228	0.000
Hausman		45.170	0.000	34.610	0.000	75.720	0.000
LR-ind		97.150	0.000	78.820	0.000	66.080	0.000
LR-time		147.090	0.000	147.390	0.000	146.610	0.000
LR-SEM		31.440	0.000	47.570	0.000	19.940	0.018
LR-SLM		36.460	0.000	39.560	0.000	28.620	0.001
Wald-SEM		27.080	0.001	12.460	0.000	24.550	0.002
Wald-SLM		32.630	0.000	21.460	0.000	31.250	0.000

Notes: W_a is geographical adjacency matrix; W_e is economic distance matrix; and W_d is geographical distance matrix.

) show that at the 1% significance level, only one result of robust LM is not significant when the geographical adjacency matrix and the economic distance matrix are used as spatial weight matrices. For the geographical distance matrix, all test results are highly significant. Therefore, from the test results, it can be seen that all three weight matrices are suitable for spatial econometric estimation of spatial–temporal double-fixed static and dynamic spatial Durbin models.

In this paper, we adopt the bias-corrected quasi-maximum likelihood estimation (BC-QML) for parameter estimation, focusing on the results of spatial–temporal double-fixed static and dynamic spatial Durbin model of W_a (geographical adjacency matrix), as shown in

Table 4. Estimated results of factors affecting GEID.

Model	Wa		We		Wd
Variable	SDM	DSDM	SDM	DSDM	SDM	DSDM
lnGEIDt−1		0.478 ***		0.413 ***		0.508 ***
(χ)		(10.34)		(9.19)		(10.67)
W×lnGEID	0.141 **	0.124 *	0.208 **	0.156 *	0.287 **	1.271 ***
(ρ)	(1.99)	(1.67)	(2.37)	(1.69)	(2.16)	(3.80)
W×lnGEIDt−1		0.004		0.178		1.499 **
(γ)		(0.05)		(1.41)		(2.42)
lnP	−1.774 ***	−0.890 **	−1.373 ***	−1.300 ***	−1.222 ***	−1.108 ***
	(−4.31)	(−2.05)	(−3.78)	(−3.51)	(−3.05)	(−2.75)
lnA	0.131	0.136 ***	0.024	0.144 ***	0.253 **	0.105 *
	(1.33)	(2.69)	(0.24)	(2.64)	(2.28)	(1.91)
lnT	0.098 *	0.077	0.090	0.069	0.151 **	0.113 *
	(1.79)	(1.35)	(1.62)	(1.22)	(2.52)	(1.85)
lnPU	0.285 ***	0.241	−0.033	0.169	0.265 ***	0.133
	(3.17)	(1.30)	(−0.36)	(1.01)	(2.75)	(0.80)
lnER	0.164 **	0.016	0.125 *	0.003	0.163 **	0.099
	(2.53)	(0.17)	(1.94)	(0.04)	(2.29)	(0.99)
lnES	0.036	0.044	−0.007	−0.115	−0.038	0.063
	(1.21)	(0.51)	(−0.20)	(−1.31)	(−1.17)	(0.71)
lnIND	0.560 ***	−0.001	0.555 ***	−0.009	0.602 ***	−0.024
	(2.93)	(−0.02)	(3.11)	(−0.33)	(3.38)	(−0.83)
lnTI	0.257 **	0.258 **	0.120	0.141	0.283 **	0.130
	(2.36)	(2.51)	(1.07)	(1.37)	(2.34)	(1.20)
lnRE	0.062	−0.004	−0.018	−0.073	0.080	−0.015
	(0.72)	(−0.05)	(−0.19)	(−0.84)	(0.82)	(−0.17)
W×lnP	2.259 ***	1.367	−0.553	−1.107	1.010	−0.657
	(2.76)	(1.57)	(−0.57)	(−1.04)	(0.14)	(−0.09)
W×lnA	−0.862 ***	0.082	−0.872 ***	0.245	3.007	−1.313
	(−3.56)	(0.70)	(−3.13)	(1.44)	(1.45)	(−1.07)
W×lnT	0.141	−0.188	0.292 *	0.010	−0.356	0.856
	(1.19)	(−1.55)	(1.72)	(0.07)	(−0.32)	(0.81)
W×lnPU	−0.292*	0.050	0.659 ***	−0.457	0.317	−1.081
	(−1.75)	(0.12)	(2.69)	(−1.03)	(0.26)	(−0.25)
W×lnER	−0.188	−0.640 ***	−0.046	−0.324	0.827	2.035
	(−1.36)	(−2.75)	(−0.27)	(−1.22)	(0.67)	(1.09)
W×lnES	0.020	−0.070	0.175 *	0.181	−1.057 **	1.138
	(0.35)	(−0.43)	(1.90)	(0.75)	(−2.50)	(0.96)
W×lnIND	−1.055 **	−0.093 *	−0.131	0.060	2.438	−0.449
	(−2.45)	(−1.73)	(−0.28)	(0.72)	(0.52)	(−1.06)
W×lnTI	−0.862 ***	0.042	−0.253	−0.706 **	2.576	−2.069
	(−2.95)	(0.16)	(−0.78)	(−2.23)	(0.78)	(−0.70)
W×lnRE	−0.130	0.304	−0.051	0.130	−0.255	−0.861
	(−0.65)	(1.60)	(−0.24)	(0.67)	(−0.13)	(−0.51)
City control	YES	YES	YES	YES	YES	YES
Year control	YES	YES	YES	YES	YES	YES
Obs	377	348	377	348	377	348

Notes: *, **, and *** indicate significance on levels of 10%, 5%, and 1%, respectively, and the corresponding Z-value is in the parentheses (); χ indicates the time lag coefficient of GEID; ρ represents the spatial lag coefficient of GEID; and γ represents the spatial–temporal lag coefficient of GEID.

.

(1)

Analysis of the spatial–temporal effects of GEID. In terms of the temporal dimension, the time lag coefficient χ of GEID is significantly positive at the 1% level under the three spatial weight matrices, which are all significantly positive at the 1% level, which fully explains the existence of path dependence characteristics. This manifests that the level of GEID in the current period will have a positive promotional effect on the next period. In terms of the spatial dimension, some provinces located in “H-L” or “L-H” quadrants in Moran scatter diagram exhibit seemingly insignificant aggregation, but the results of the spatial Durbin model show that the spatial lag coefficients ρ of the three spatial weighting matrices under the static and dynamic models are all significantly positive, which firmly verifies that the level of GEID has a significant spatial agglomeration feature overall. This manifests the “H-H” agglomeration effect generated by the radiation-driven effect, where higher levels of GEID in neighbouring provinces drive up the level of GEID in the local province, and the “L-L” agglomeration effect generated by the trough-impacting effect, where lower levels of GEID in neighbouring provinces affect the level of GEID in the local province. In terms of spatial–temporal dimension, only the spatial–temporal lag coefficient γ of the geographical distance matrix is significantly positive, indicating that the spatial–temporal double effect of GEID is not significant. This implies that the GEID of the previous period in neighbouring provinces does not significantly affect the current GEID in the local province.

(2)

Analysis of factors influencing GEID. The estimation results of the spatial Durbin model (SDM) and the dynamic spatial Durbin model (DSDM) under the W_a matrix indicate that among the nine influencing factors, population size (P), affluence (A), technical innovation (T), public participation (PU), environmental regulation (ER), industry (IND), and transportation industry (TI) significantly affect GEID. The coefficient of population size (P) is negative, indicating that higher population density leads to lower levels of GEID in the local province. The remaining factors all have a positive effect on GEID in the local province. Except for technical innovation (T), the spatial spillover effect of the other six factors is significant. The spillover effect of population size (P) is positive, indicating that an increase in population density in neighbouring provinces promotes GEID in the local province. The spillover effect of the other five factors is negative.

3.3.2. Effect Decomposition of Influencing Factors

Elhorst et al. [35] pointed out that when global effects are included in the model specification, the point estimates of the spatial econometric model do not represent the marginal effects of the explanatory variables. Therefore, to compare and analyse the differences in the effects of each explanatory variable and their spatial spillover effects, it is necessary to further calculate the direct and indirect effects of each explanatory variable based on the point estimates of the model.

Accordingly, by employing the partial differential method [36], this paper conducted an effect decomposition of the estimation results from the static spatial Durbin model under the W_a matrix. This allowed us to discern the direct, indirect, and total effects of various variables on GEID. The direct effect captures the influence of explanatory variables on the local dependent variable; whereas, the indirect effect reflects their indirect impact on the dependent variable in neighbouring regions. The total effect, which is the sum of the direct and indirect effects, provides a comprehensive view of the overall influence. The detailed decomposition results are presented in

Table 5. Direct and indirect effects of SDM.

Variable	Direct Effect	Indirect Effect	Total Effect
lnP	−1.688 ***	2.295 **	0.607
	(−4.08)	(2.47)	(0.64)
lnA	0.097	−0.974 ***	−0.877 ***
	(1.01)	(−3.48)	(−2.88)
lnT	0.109 **	0.184	0.293 *
	(2.07)	(1.27)	(1.82)
lnPU	0.274 ***	−0.289 *	−0.015
	(3.20)	(−1.67)	(−0.08)
lnER	0.159 **	−0.178	−0.019
	(2.51)	(−1.09)	(−0.10)
lnES	0.038	0.024	0.062
	(1.27)	(0.37)	(0.90)
lnIND	0.521 ***	−1.099 **	−0.578
	(2.62)	(−2.09)	(−0.89)
lnTI	0.227 **	−0.911 ***	−0.684 *
	(1.97)	(−2.65)	(−1.72)
lnRE	0.065	−0.139	−0.074
	(0.77)	(−0.60)	(−0.29)

Notes: *, **, and *** indicate significance on level of 10%, 5%, and 1%, respectively, and the corresponding Z-value is in the parentheses ().

.

(1)

From the perspective of population factors, the direct effect of population size (P) is significantly positive, while the indirect effect is significantly negative. The estimated coefficient is the largest among the nine influencing factors, indicating that population size has the greatest impact on GEID. The increase in population size will lead to a decrease in the GEID of the local province, and the increase in the population size of neighbouring provinces will promote the improvement of the GEID of the local government. Therefore, the role of population factors in the development of GEID in China is significant.

(2)

From the economic perspective, the direct effect of affluence (A) is positive but not significant, suggesting that economic development helps to improve GEID in the local province, but this effect is not statistically significant. The indirect effect is significantly negative, indicating that economic development in neighbouring provinces reduces the level of GEID in the local province. The “siphon effect” from economically developed neighbouring provinces may be the main reason for this phenomenon. A large number of talents and funds flowed out of the province into the neighbouring provinces, while the high-emission industries correspondingly moved from the neighbouring provinces to the province. As a result, the economic and environmental performances of the local province lag behind the neighbouring provinces, and the government is not sufficiently motivated to disclose environmental information, resulting in a lower level of GEID.

(3)

From the technical innovation perspective, the direct and indirect effects of technical innovation (T) are both positive, but the indirect effect does not pass the significance test. This indicates that progress in technological innovation, especially the improvement of green innovation capabilities, can improve the province’s level of clean production, reduce environmental pollution problems, and strengthen the government’s willingness to disclose environmental information.

(4)

From the public participation perspective, the direct effect of public participation (PU) is significantly positive, indicating that the improvement of GEID depends on extensive public participation. The higher the level of concern among residents in the local province, the greater the public pressure on the local province to improve GEID. However, the indirect effect is significantly negative, suggesting that high levels of public participation in neighbouring provinces reduce the level of GEID in the local province. This phenomenon may be due to the environmental risk aversion psychology of the local province. As the level of public concern about the environment increases in neighbouring provinces, the level of GEID also increases in neighbouring provinces. At this point, the local province faces both public pressure and comparative pressure on environmental performance from neighbouring provinces, leading to a tendency to reduce environmental information disclosure and consequently lower levels of GEID.

(5)

From the environmental regulation perspective, the direct effect of environmental regulation (ER) is significantly positive, and the indirect effect is not significant. It fully indicates that the higher the government’s attention to local environmental problems, the higher the level of GEID.

(6)

From the environmental regulation perspective, both the direct and indirect effects of environmental status (ES) are not significant, indicating that the level of environmental pollution is not a major factor influencing the level of GEID.

(7)

From the industry perspective, the direct and indirect effects of Real Estate (RE) are both insignificant, indicating that although Real Estate (RE) has long been regarded as a pillar industry in China, it has no significant effect on the level of GEID. The direct effect of industry (IND) and transportation industry (TI) is significantly positive, and the indirect effect is significantly negative, indicating that industry and transportation industry will cause the level of GEID of the local province to improve, and the development of industry and transportation industry in neighbouring provinces will reduce the level of GEID of the local province.

In summary, by decomposing the total effect into direct and indirect effects, this study not only provides an in-depth analysis of the mechanisms through which various explanatory variables influence the level of GEID but also robustly validates the stability of the core explanatory variables in their effects on GEID, thereby ensuring the scientific rigour and reliability of the research conclusions. Among the nine influencing factors studied in this paper, the factors with the greatest impact on the level of GEID in the local province are population size, industry, public participation, transportation industry, environmental regulation, and technical innovation. The factors with the greatest impact on the level of GEID in neighbouring provinces are population size, industry, affluence, transportation industry, and public participation (PU). Environmental status and Real Estate have no significant effect on GEID.

4. Conclusions and Suggestions

This paper uses spatial data exploration analysis and static and dynamic spatial Durbin models to systematically analyse the spatial relationships, agglomeration characteristics, spatial–temporal development trends, and influencing factors of GEID in China from a dual perspective of time and space. The study found that (1) from 2009 to 2021, the agglomeration characteristics of GEID in China were significantly pronounced and showed a fluctuating improvement trend, mainly manifested in “H-H” (High-High aggregation type) and “L-L” (Low-Low aggregation type) types. Coastal provinces and the Beijing–Tianjin–Hebei region show higher levels of GEID. The “High-High” aggregation characteristic shifted from the Yangtze River Delta northward to the southeastern coastal areas and is currently concentrated in the Beijing–Tianjin–Hebei region. Its development path is closely related to major national regional development strategies such as the Yangtze River Delta integrated development, the coastal development strategies, and the Beijing–Tianjin–Hebei coordinated development. (2) Overall, GEID in China shows a trend of moving towards the northwest, with the area of the ellipse gradually expanding. The exemplary role of the leading provinces is driving the gradual improvement of the overall level of GEID in China. At the same time, the agglomeration effect in the “Northeast-Southwest” direction continues to strengthen, with spillover effects from the southeastern coastal provinces gradually radiating to the northwest, raising the level of GEID in surrounding provinces. (3) GEID shows significant path-dependency characteristics in the temporal dimension and significant “peer effects” in the spatial dimension. (4) Factors in the decomposition of influences have different effects on the local province and neighbouring provinces. Among them, population size has the greatest impact on the level of GEID. Specifically, an expansion in the local province’s population base tends to negatively correlate with its GEID level, suggesting that a growing population may impose constraints on resources and attention allocated towards environmental transparency initiatives. Conversely, an increase in the population size of neighbouring provinces fosters a positive influence on the local government’s GEID performance. This can be attributed to the heightened environmental consciousness and scrutiny from a larger neighbouring populace, which can create an impetus for the local government to enhance its disclosure practices. Additionally, the positive spillover effects from neighbouring provinces, such as the adoption of advanced disclosure policies and successful implementation strategies, can serve as valuable benchmarks for the local government to emulate and improve upon. (5) Population size, public participation, industry, and transportation industry promote the improvement of the local province’s GEID but generate negative spillover effects on neighbouring provinces. Environmental status and Real Estate have no significant impact.

Based on the above conclusions, the following suggestions are put forward: (1) enhance top-level policy coordination and regional integration: recognising the profound spatial implications of GEID, it is imperative to fortify strategic planning at the highest levels. This involves fostering inter-provincial policy synergy and collaborative development, instituting cross-departmental and cross-regional disclosure frameworks that facilitate information sharing and joint decision making. Furthermore, emphasising the exemplary role of advanced provinces can serve as a catalyst for nationwide progress, encouraging the adoption of best practices and fostering a culture of continuous improvement; (2) promote targeted policy initiatives to bridge regional disparities: the implementation of tailored regional development policies holds the key to elevating GEID levels across the board. For the northwestern region, China should embark on macro-level strategies that prioritise the development of industries and information infrastructure, thereby fostering a conducive environment for growth. Integrating the western region into national development blueprints and frameworks can further amplify its GEID potential, ensuring that all regions share equitably in the country’s economic progress; (3) establish a comprehensive and proactive disclosure system for governmental and public oversight: bolstering transparency and accountability by linking GEID performance to government evaluation criteria is crucial. This will incentivise proactive disclosure and enhance public trust. Additionally, broadening avenues for public participation in oversight activities, such as through online platforms and citizen feedback mechanisms, can strengthen the monitoring role of citizens, fostering a more inclusive and responsive governance ecosystem; and (4) nurture new productive forces, accelerate technological innovation, and drive industrial upgrading: recognising that technological advancements and industrial evolution are pivotal drivers of GEID enhancement, governments must tailor their strategies to local contexts. This includes fostering the development of emerging industries, supporting research and development, and promoting the adoption of cutting-edge technologies. Additionally, upgrading the modern industrial system to incorporate these advancements can unlock new avenues for growth, ensuring that local economies remain competitive and dynamic in the face of global challenges.