Data for this study were sourced from publicly available statistical yearbooks, including the China Statistical Yearbook (2012–2022), China Forestry Statistical Yearbook, and China Forestry and Grassland Statistical Yearbook, covering 30 provinces in China. Tibet, Hong Kong, Macau, and Taiwan were excluded due to missing or inconsistent data, which could affect the reliability of the analysis. The selected sources were chosen based on their comprehensiveness and consistency over time, ensuring reliable and comparable data. In addition, remote sensing data were incorporated from the Resource and Environmental Science and Data Center of the Chinese Academy of Sciences and the National Meteorological Data Center, providing a more granular view of ecological changes across regions. Remote sensing data for green economy products at the provincial and municipal levels were sourced from the Resource and Environmental Science and Data Center, National Tibetan Plateau Science Data Center, National Meteorological Data Center, Geospatial Data Cloud, and the National Earth System Science Data Center. These data are crucial for studying forest ecosystem product environmental changes, ecosystem services, and resource utilization efficiency. Soil data were derived from the World Soil Database (HWSD), jointly constructed by the Food and Agriculture Organization (FAO) and the International Institute for Applied Systems Analysis (IIASA). The soil data for China, available at a 1:1,000,000 scale, cover the period from 2012 to 2022, with missing values supplemented using interpolation methods. Descriptive statistics for each variable are presented in Table 4.

FEPVRE serves as the basis for studying regional efficiency disparities across China. The analysis of efficiency development trends highlights the specific differences and evolution of FEPVRE in the country’s four major regions. The results indicate a fluctuating trend in FEPVRE development across regions. Significant fluctuations from 2012 to 2020 reflect the early stages of China’s green transformation and ecological civilization strategy. Policymakers should address these regional disparities, especially in the western and northeastern regions, where FEPVRE lags behind. The measurement results are presented in Table 5, and the efficiency evolution trend is shown in Figure 3.

From 2012 to 2022, the national average FEPVRE showed a fluctuating trend with significant regional disparities. The Eastern region outperformed the Western and Northeastern regions due to higher investments in green technology and ecological development. Policymakers should prioritize digital infrastructure and green innovation to enhance efficiency in lagging regions, especially the Western and Northeastern areas. In 2016, the FEPVRE peaked at 30.81%, indicating highly efficient utilization of forest ecosystem products in China. However, between 2016 and 2020, a decline occurred due to pressures from China’s “dual carbon goals.” Regions heavily reliant on traditional industries faced challenges aligning with the green transformation. Policymakers should support these areas, fostering a balanced transition to a low-carbon economy with targeted policies and incentives. A noticeable convergence trend emerged, with the Western region’s FEPVRE gradually approaching that of the Central region, while the Northeastern region remained relatively low and grew slowly. The Eastern region’s higher FEPVRE values were driven by investments in ecological construction and green industry innovation. The Central region showed larger fluctuations due to challenges in economic transitions, such as significant population movements and industrialization. The Western region, particularly provinces like Guizhou, Guangxi, and Sichuan, displayed fluctuating growth linked to abundant natural resources and successful green development strategies. The Northeast region, with low FEPVRE values and minimal improvement, struggled with imbalanced resource development due to structural adjustments in industrialization.

Regional disparities between the East and West are the primary drivers of differences in FEPVRE across China. This study employs a modified gravity model to measure the degree of association in FEPVRE among the 30 provinces and cities, aiming to identify the intensity of regional connections. Figures 4(a-k) illustrates the FEPVRE association from 2012 to 2022. The analysis reveals that the Eastern region is closely linked with the Central region, indicating potential for inter-regional cooperation in green development. Conversely, the Western and Northeastern regions show weaker connections, presenting opportunities for enhanced regional cooperation. Policymakers could foster partnerships to improve FEPVRE in less developed regions by promoting knowledge and resource sharing. Using the natural breakpoints classification method, regions are divided into four levels based on cooperation strength. The Western region has fewer intensive connections, resulting in lower cooperation in FEPVRE compared to the Eastern and Central regions. Over time, high-level cooperation has been concentrated in the Southeastern region, with Guangdong leading the way. In contrast, the Western region lacks inter-regional cooperation or collaborative cities. The FEPVRE cooperation network in China demonstrates a “Matthew Effect,” where provinces and cities with higher cooperation intensities cluster together, reinforcing the unequal distribution of FEPVRE development.

Since the southeastern region significantly influences regional differences in FEPVRE across China, this study employs SKDE to explore the spatial distribution dynamics of FEPVRE and assess its temporal and spatial evolution. The static spatial kernel density and contour map of FEPVRE (Figures 5(a,b)) reveal that the overall efficiency values of the forest ecosystem product value realization show a rightward shift and concentration trend, with expanded coverage. This indicates an overall improvement in FEPVRE across China, with a more concentrated distribution. The performance of different provinces has become more consistent, and the range of distribution has widened. The curve polarization trend suggests a dominant peak, indicating that regional differences in FEPVRE are expected to persist and even widen during the sample period. The contour map reveals peaks aligned parallel to the y-axis, located at the upper left and lower right of the 45-degree diagonal, suggesting significant regional differences, with some provinces exhibiting relatively low FEPVRE. The spatial distribution of FEPVRE is concentrated in the Eastern and Southeastern regions, while the Western and Northeastern regions lag behind. To address these disparities, the government can invest in regional capacity building and promote technologies in lagging areas, which will help reduce spatial disparities and support more balanced green development.

From the spatial dynamic kernel density and contour map of FEPVRE across China’s four major regions (Figures 6(a-e)), it is evident that severe polarization exists in both time and space, particularly in terms of spatial spillover effects. In the Eastern region, the peaks are aligned parallel to the y-axis, with one section of the primary peak located just below the 45-degree diagonal and the other section above it. This suggests a polarization trend: after a 3-year period, some provinces experience high FEPVRE, while others show a decline, reflecting instability in the region’s development. In the Western region, the peaks are above the 45-degree diagonal, indicating a positive shift in FEPVRE, suggesting that these areas have significant potential for growth, especially during the green transformation process. In the Central region, the peaks are below the 45-degree diagonal, indicating that after 3 years, FEPVRE tends to stabilize and concentrate, with little room for further growth. This may be due to the full development of forest resources and a slower pace of green transformation. In the Northeastern region, there are two segments: one with a large peak above the 45-degree diagonal, indicating improvement in FEPVRE in some provinces after 3 years, and the other segment, a trough, lying below the diagonal, where provinces exhibit lower FEPVRE. For these provinces, strengthening ecological protection policies, promoting green development, and improving forest resource management are essential.

Due to the significant dual-level development of FEPVRE in China, Markov chain analysis is employed to analyze the dynamic transition trends of FEPVRE. The comprehensive scores during the sample period are categorized into four types: Low (I), Low-Medium (II), Medium-High (III), and High (IV). The results are presented in Table 6.

The analysis in Table 7 reveals that FEPVRE dynamics exhibit both stability and dynamism. The values on the main diagonal are generally higher than those on the off-diagonal, with the maximum value being 0.790 and the minimum 0.547. The probability of FEPVRE remaining unchanged for most provinces is at least 55%, indicating high stability across the majority of provinces. For FEPVRE transitions, provinces in Category I (inefficient) have a transition probability of 0.789 to remain in their status, while provinces in Category IV (highly efficient) have a transition probability of 0.747, indicating strong stability in both low-efficiency and high-efficiency provinces. However, provinces in intermediate categories (II and III) face greater challenges in terms of transition. Specifically, Category I provinces (low-efficiency) have a 0.125 probability of moving up to Category II, reflecting the difficulty in improving efficiency. In contrast, Category III provinces (moderately high-efficiency) have a 0.227 probability of advancing, suggesting significant potential for improvement. These provinces are in the “sweet spot,” with the capability and potential to further enhance efficiency. Category II provinces (moderately low-efficiency) face a 0.125 probability of downgrading to Category I, highlighting the risk of significant decline. Even Category IV provinces (high-efficiency) face a 0.190 probability of transitioning downward, indicating downward pressure on even the most efficient regions.

The Markov chain analysis results, presented in Table 6, show that the transition probabilities for spatial lag types I and IV are 0.790 and 0.747, respectively. For spatial lag type I, the probability of maintaining the original state is significantly higher than the probabilities of upward or downward transitions. Similarly, in spatial lag type IV (high-efficiency provinces), the probability of maintaining the original state is also high, indicating the higher stability of ecological efficiency in these provinces. Both low-efficiency and high-efficiency provinces tend to maintain their existing ecological efficiency levels, especially when the efficiency gap with neighboring provinces is small or stable. In contrast, spatial lag types II and III show more instability, with most off-diagonal elements non-zero. This suggests that provinces in these types have a higher potential for transitioning to either better or worse efficiency levels. Notably, in spatial lag types III and IV, the off-diagonal elements are 0.100 and 0.250, respectively, indicating that the efficiency of these provinces is more susceptible to external shocks and the transition processes of neighboring provinces. The differences in upward and downward transition probabilities under different lag conditions are significant. In spatial lag type III, the probability of a type I province transitioning upward to type II is 0.140, while a type III province has a 0.227 probability of moving upward to type IV, highlighting the significant improvement potential for provinces in type III. Conversely, in spatial lag type IV, the probability of a type II province transitioning upward to type III is 0.250, while the downward transition probability for high-efficiency provinces in type IV is 0.190, demonstrating that high-efficiency provinces show strong resistance to decline, maintaining relatively stable efficiency. These findings indicate that ecological efficiency transitions under different lag states are strongly influenced by the current efficiency status of a province. Provinces in type II and III are at a critical point for upward transitions. If they can leverage external support, such as neighboring provinces’ experiences and technological advancements, there is potential for overcoming existing bottlenecks. On the other hand, provinces in type IV exhibit strong stability and low transition risks.

Overall, the Markov chain analysis of FEPVRE suggests that most provinces maintain their efficiency levels, but provinces in the intermediate efficiency range have significant upward mobility potential. Future research should explore the factors driving these transitions and identify the elements contributing to upward mobility. Policymakers can use these insights to support provinces with improvement potential through targeted interventions.

The analysis of the FEPVRE Gini coefficient (GT) from 2012 to 2022 reveals significant regional disparities across China, reflecting the dynamic interaction of political, economic, and social factors that influence forest ecosystem product value realization efficiency. The following sections provide an overview of national trends and a detailed regional analysis, comparing the development trends in the Eastern, Central, Western, and Northeastern regions. Nationally, the Gini coefficient fluctuated, starting at 0.592 in 2012 and decreasing steadily to 0.417 by 2015, signaling a narrowing of regional disparities. However, from 2015 to 2021, the coefficient rose to 0.572, indicating an increase in the regional gap. In 2022, it dropped to 0.485, reflecting some improvement, particularly in the central and western regions, due to policy interventions. In the Eastern region, the Gini coefficient continuously improved, declining from 0.217 in 2012 to 0.148 in 2022, driven by investments in green technology and infrastructure. In contrast, the Western region experienced significant fluctuations, rising from 0.494 in 2012 to 0.574 in 2022. This reflects challenges like underdeveloped infrastructure, slower economic growth, and limited technological innovation. The Central region had moderate fluctuations, with its Gini coefficient moving from 0.244 in 2012 to 0.283 in 2022, showing gradual progress despite challenges in infrastructure and industrial development. The Northeastern region exhibited large fluctuations, with the Gini coefficient peaking at 0.644 in 2013, then dropping to 0.310 in 2014, and stabilizing at 0.428 by 2022, highlighting struggles with industrial restructuring and the slow transition to green industries. In conclusion, while the Eastern region has made significant progress, other regions like the Western and Northeastern areas continue to face substantial challenges in improving FEPVRE. This underscores the need for targeted policies and investments in green technologies and infrastructure to foster more balanced and sustainable development across the country, as shown in Figure 7.

The decomposition of the Gini coefficient for FEPVRE highlights the key sources of regional disparities. The value of Gw has remained relatively stable, starting at 0.136 in 2012, peaking at 0.138 in 2021, and decreasing to 0.120 in 2022. This suggests that regional disparities have fluctuated only slightly during the study period. The stability of Gw indicates that, despite minor changes in regional balance, the efficiency of forest ecosystem product utilization within regions has generally remained consistent. Contributing factors could include policies promoting more equitable resource distribution and improvements in ecological protection measures, although localized disparities still exist. Gnb shows a fluctuating pattern, starting at 0.394 in 2012 and gradually decreasing to 0.244 by 2022, with a peak of 0.357 in 2021. The decreasing trend of Gnb after 2020 indicates some success in efforts to reduce regional disparities in FEPVRE, particularly in improving the performance of less efficient regions. Policies supporting the digital economy, ecological development, and technological advancements have positively impacted low-efficiency regions, helping reduce the gap. However, the sharp rise in Gnb in 2021 suggests that regional disparities intensified, possibly due to uneven technological adoption or industrial development. Gt fluctuated during the period, starting at 0.0628 in 2012, rising to 0.139 in 2019, and gradually decreasing to 0.121 by 2022. Gt peaked at 0.139 in 2019, followed by a decline in 2020 and 2021. While Gt reflects disparities between high-efficiency and low-efficiency regions, its changes more directly reflect structural differences, particularly in resource distribution and technology adoption. The fluctuations in Gt highlight the gap between leading and lagging regions, while Gnb emphasizes the fundamental disparities between regions and the improvements driven by policy interventions, as shown in Figure 8.

The within-group Gini changes from 2012 to 2022 (Table 8) reveal fluctuating trends in FEPVRE across China, reflecting the changing nature of internal disparities within regions at different stages. In the eastern region, the within-group Gini coefficient gradually decreased from 2012 to 2015, indicating a narrowing of internal disparities and an improvement in internal balance. However, from 2016 to 2021, the Gini coefficient began to rise, reflecting fluctuating internal disparities due to unbalanced ecological protection and resource management. The central region showed relatively stable changes in the within-group Gini but experienced a slight upward fluctuation from 2016 to 2022, suggesting that internal disparities in this region have slightly intensified. This can be attributed to uneven economic development in some provinces, widening internal disparities. The western region exhibited large fluctuations in the within-group Gini coefficient, initially decreasing and then increasing. This pattern indicates that internal differences in FEPVRE in the western region narrowed at first, but expanded after the disruptions caused by the 2019 pandemic, highlighting significant internal disparities. The northeastern region showed large fluctuations from 2012 to 2015, with a decrease and peak in 2018. This indicates intensifying internal disparities, and while the internal FEPVRE improved by 2022, it remained in a state of significant disparity, suggesting a lag in the development resources required. Overall, the Gini analysis reveals growing regional disparities in FEPVRE, with widening gaps between the Eastern, Western, and Northeastern regions. Policymakers should focus on closing these gaps through green policies, equitable investment distribution, and enhanced cross-regional collaboration to promote more balanced and sustainable development.

As inter-group differences are the primary source of regional disparities, this section analyzes the trend in inter-group Gini across regions to reflect dynamic disparities in FEPVRE in China. The results, shown in Table 9, reveal that from 2012 to 2022, the inter-group Gini between the Eastern region and the Central, Western, and Northeastern regions exhibited an overall downward trend. In contrast, the inter-group Gini between the Central and Western, Central and Northeastern, and Western and Northeastern regions showed an increasing trend, indicating significant and uneven disparities. From 2012 to 2015, the inter-group Gini between the Eastern and Central regions continued to decrease, with slight rebounds in 2016 and 2021 before decreasing again, signaling a gradual narrowing of the gap between these regions. The inter-group Gini between the Eastern and Western regions also showed a decreasing trend from 2012 to 2022, though the gap between the Central and Western regions remained larger. Meanwhile, the inter-group Gini between the Eastern and Northeastern regions was higher, reflecting greater differences in FEPVRE between these regions. However, their regional development remained more stable than that between the Eastern and Western regions. From 2013 to 2018, the inter-group Gini between the Central and Western regions showed a fluctuating downward trend, followed by a rapid increase until 2022. Despite this, the inter-group Gini between the Central and Western regions was much lower than that between the Eastern and Western regions, indicating that the Eastern region’s FEPVRE is higher than those of the Central and Western regions, with a larger gap. The inter-group Gini between the Central and Northeastern regions showed slow, fluctuating growth, while the inter-group Gini between the Western and Northeastern regions was higher, suggesting that the gap between the Central and Northeastern regions was smaller than the gap between the Western and Northeastern regions.

The spatial autocorrelation test was conducted using Moran’s I index, which yielded a value of 0.600, indicating a positive spatial effect. This result is statistically significant at the 1% level, suggesting that from 2012 to 2022, there is a positive spatial effect on the FEPVRE. Therefore, spatial variables are introduced for further analysis, as shown in Table 12.

To further investigate the influencing factors of the spatiotemporal FEPVRE in various provinces and municipalities of China, three spatial weight matrices are constructed: the spatial adjacency weight matrix, the digital finance weight matrix, and the economic geography weight matrix.

The spatial distance weight matrix is calculated using the latitude and longitude of different provinces, measuring the square of the geographical distance between provinces. d i j 2 represents the square of the geographical distance between two provinces. The Formula 20 for the spatial distance weight matrix is expressed as: W i j = 1 D ¯ i − D ¯ j i ≠ j (20)

The spatial adjacency weight matrix is the reciprocal of the square of the geographical distance between the central cities of different provinces in China. This matrix is used to determine the adjacency relationship between regions. The distance between region i and region j is denoted by   d i j . The Formula 21 for the spatial adjacency weight matrix is expressed as: W i j =   1     d ij < d   0     d ij ≥ d (21)

The digital economy spatial weight matrix, by analyzing the spatial interactions between digital economy and geographical factors on FEPVRE, helps to understand how the digital economy drives regional green transformation through spatial spillover effects. This matrix provides a quantitative tool for analyzing the mutual influences between different regions, particularly in the context of significant differences in digital economy levels, revealing the spatial disparities in green economic transformation. In turn, it supports the formulation of more targeted policies, promoting the coordinated and sustainable development of green economy across regions in China. The Digital Economy Spatial Weight Matrix W i j is obtained by the product of the digital economy composite index and the square of the geographical distance. Each element W i j in the matrix represents the mutual influence intensity between region i and region j. The digital economy composite index D E I is calculated through the entropy method, using multiple digital economy indicators such as the scale of mobile phone facilities, the length of long-distance optical fiber cables, the number of web pages, the number of domain names, information transmission capacity, the number of legal entities in software and information technology services, the number of domestic patent applications, and digital inclusive finance. Geographical distance d i j is a key factor in measuring the spatial relationships between different regions. In this paper, the geographical distance is calculated based on the latitude and longitude distances between provinces and cities. The Digital Economy Spatial Weight Matrix is obtained by the product of the digital economy composite index and the square of the geographical distance. The calculation Formula 22 is as follows: W i j =   D E I i · D E I j d i j 2   i ≠ j   0   i = j (22)

The SDM, SAR, and SEM are three commonly used spatial econometric models. For the purpose of model selection and assessing spatial dependence, spatial distance weight matrix, the spatial adjacency weight matrix and the digital economy spatial weight matrix were employed to properly assess the spatial dependence in the model selection, and the LM error and LR are performed. These tests are crucial in diagnosing whether spatial dependence exists and determining which form of dependence—error or lag—is more significant. The LM and the LR p-values for all three matrices are below the 10%, 5% and 1% significance levels, indicating the presence of spatial dependence and confirming that the models pass the significance test. In addition, Wald tests are conducted to further refine the model selection. These tests evaluate the adequacy of the selected models by testing the null hypothesis regarding the restrictions on the coefficients. The Wald test results for all three spatial weight matrices show strong significance, providing additional confirmation of the robustness of the chosen models. This combination of LM, LR, and Wald tests ensures the appropriate selection of the most suitable spatial econometric model for analyzing spatial dependence in the context of FEPVRE. As shown in Table 13.

In the robustness test of spatial effects in Table 13, both the SEM and the SAR were applied to test spatial effects under different spatial weight matrices. For the Spatial Distance Weight Matrix, both the Lagrange Multiplier (LM) test for spatial error and the LM test for spatial lag are significant at the 1% level, indicating the presence of both spatial error and spatial lag effects. This suggests that a combination of the SEM and SAR models would be appropriate. Given the presence of both effects, the SDM, which combines features of both the SEM and SAR models, is selected as the most suitable model for this matrix. Additionally, the LR test for SAR and LR test for SEM are significant at the 5% level, confirming that the SDM cannot degrade into either the SAR or SEM models. This result is consistent with the Wald test, further validating the use of the SDM for the Spatial Distance Weight Matrix. Contrast, the Spatial Adjacency Weight Matrix and the Digital Economy Geographic Weight Matrix are best modeled by the SAR, but the reasoning behind the model selection differs. For the Spatial Adjacency Weight Matrix, the LM test for spatial error is significant at the 10% level, while the LM test for spatial lag is highly significant at the 1% level. These results indicate the presence of both spatial error and spatial lag effects, making it suitable to apply both SEM and SAR. Since both models are relevant, the SDM, which combines features of both, is the best choice. However, the LR test for SAR and SEM for the Spatial Adjacency Weight Matrix are not significant at the 10% level, supporting the null hypothesis that the SDM can be simplified to either SAR or SEM. This aligns with the Wald test results, confirming that either model could be chosen. Since the LM test for SAR yields a more rational result, the SAR model is considered the most appropriate for this matrix. For the Digital Economy Geographic Weight Matrix, the LM test for the Lagrange multiplier is not significant at the 5% level (p > 0.05), while the LM test for spatial lag is highly significant at the 1% level (p < 0.01), indicating that the spatial lag effect is dominant. Therefore, the SAR model is more suitable for this matrix. Furthermore, the LR test for SAR and SEM for the Digital Economy Geographic Weight Matrix are both insignificant at the 10% level, supporting the hypothesis that the SDM can be simplified to the SAR model. This is consistent with the results from the Wald test for SAR and SEM, which also suggest that the SAR model is the most appropriate choice.

Using Stata 15 software, the random effects model, time fixed effects model, and double fixed effects model were estimated, incorporating the three spatial weight matrices into the spatial econometric models. The calculation results are shown in Table 14.

The Hausman test results confirm that the fixed effects model is the most suitable for all three spatial weight matrices—spatial distance, spatial adjacency, and digital economy geographic matrices. With a p-value of 0.000, the fixed effects model is preferred over the random effects model due to significant differences in coefficients and standard errors. The time fixed effects model is better suited for capturing temporal variations in FEPVRE, effectively controlling for time-specific factors, while the two-way fixed effects model, though capable of considering both regional and time effects, is more complex and less precise. In the Spatial Distance Weight Matrix’s double fixed effects model, the LogL is −94.252, indicating a good fit. The coefficient for λ is −0.048, statistically significant at the 5% level, and the R-squared value is 0.073, suggesting reasonable reliability for this model. The coefficients for the TOE variables show that both T and E have negative coefficients with respect to FEPVRE, indicating that digital technology and the digital environment exhibit negative spillover effects. In contrast, O has a positive coefficient of 0.951, suggesting a positive spillover effect of the digital industry on FEPVRE. This suggests that weak technology adoption and environmental adaptability hinder the benefits from technological and environmental improvements, potentially causing negative spillover effects due to the unequal distribution of digital infrastructure, technology access, and policy support. On the other hand, digital industries, especially green technology innovation, positively impact green transformation. When developed in certain regions, digital industries not only enhance local green transformation but also promote it in neighboring areas through spillover effects. Therefore, policies should focus on supporting digital industries in less efficient regions to improve overall green economic performance. The maximum LogL values for the Spatial Adjacency Weight Matrix and the Digital Economy Geographic Weight Matrix are −111.685 and −97.652, respectively, with λ values of 0.370 and −0.091, and σ² values of 0.219 and 0.236. The p-values are significant at the 1% level, indicating that the SAR model fits the data well. In both matrices, the effects of digital technology and the digital industry on FEPVRE exhibit positive spatial spillover effects, while the impact of the digital environment shows negative spatial spillover effects. These negative effects tend to weaken over time. This suggests that digital technology and industries generate positive spillover effects on FEPVRE, driving green transformation in neighboring regions through technological diffusion and industrial chain collaboration. Conversely, the digital environment causes negative spillover effects, mainly due to the environmental burdens and resource distribution imbalances associated with digitalization. Over time, policy adjustments and technological advancements help mitigate these negative effects, promoting a more sustainable green transformation.

To explore the impact of T, O, and E on FEPVRE across different regions, China’s provinces are categorized into four regions: Eastern, Central, Western, and Northeastern. The analysis reveals significant regional disparities in FEPVRE, with the Eastern region exhibiting higher efficiency levels compared to the Western and Northeastern regions. These disparities reflect the complex interaction of technological, organizational, and environmental factors, as supported by ecological economics and Green Growth theories. The calculation results show that the SDM model for the Spatial Distance Weight Matrix has relatively high LogL and R-square values for all four regions, indicating a good model fit. Similarly, the Spatial Adjacency Weight Matrix and Digital Economy Geographic Weight Matrix also show higher LogL and R-square values, suggesting a good fit for the SAR model. The spatial autoregressive coefficients are significant at the 10%、5%, and 1% levels, further validating the appropriateness of these models, as shown in Table 15.

In the Spatial Distance Weight Matrix, T and E exhibit negative spillover effects on FEPVRE in the Eastern and Western regions, while showing positive spillover effects in the Central and Northeastern regions. O only has a negative spillover effect in the Central region, with positive spillover effects in all other regions. The Eastern and Western regions face challenges in digital green transformation due to technology adoption gaps, inadequate infrastructure, and regional resistance, which hinder FEPVRE. In contrast, the Central and Northeastern regions benefit from better integration of digital technologies and sustainable policies, leading to positive spillover effects and stronger FEPVRE. Digital industry shows a negative spillover effect only in the Central region, possibly due to industrial overcapacity or lack of green innovation. The Central region should focus on fostering green digital innovation, while other regions can scale up digital industry growth with a sustainability focus. In the Spatial Adjacency Weight Matrix, T generates negative spillover effects on FEPVRE in the Eastern and Western regions, while generating positive spillover effects in the Central and Northeastern regions. This is primarily due to uneven technology diffusion and industrial structure differences, with lower technology adoption rates in some areas of the Eastern region, limiting the positive impacts of digital technology. In the Western region, infrastructure deficiencies and delays in technological innovation hinder effective diffusion, restricting the green transformation of surrounding areas. In contrast, O shows positive spillover effects on FEPVRE across all regions, promoting green industry development and enhancing forest ecosystem product value, particularly in the Central, Eastern, Northeastern, and Western regions. However, E produces negative spillover effects across all regions, mainly because digitalization has exacerbated resource over-exploitation and ecological pressure, especially in digital infrastructure construction. Uneven distribution of environmental resources and a lack of sustainable management have intensified environmental burdens, hindering improvements in FEPVRE.

In the Digital Economy Geographic Weight Matrix, O exhibits negative spillover effects in the Central region, while showing positive effects in other regions. Conversely, T has a positive spillover effect in the Central region but negative spillover effects in other regions. This highlights that regional disparities in digital development and green transformation are influenced by varying spillover effects. In the Central region, where digital infrastructure and technology adoption lag, negative spillovers reflect barriers such as resource inequality, technological gaps, and lack of green innovation. In contrast, the Eastern and Western regions benefit from positive spillovers, indicating better integration of digital solutions aligned with green transformation goals. The negative spillover effect of the digital environment across regions suggests challenges in leveraging environmental improvements, often hindered by insufficient ecological policies and uneven resource distribution. Therefore, regions with negative spillovers should focus on improving digital infrastructure, adopting green technologies, and implementing region-specific innovations to enhance FEPVRE.

To verify the reliability and stability of the conclusions regarding the impact of T, O, and E on FEPVRE, robustness tests are conducted using multiple spatial weight matrices. The LM test, Wald test, Hausman test, and LR test are employed to evaluate the stability and robustness of the spatial econometric models (Zheng et al., 2025). These tests assess whether the results remain consistent across different spatial structures and model specifications.

First, the LM and LR test results are compared using the spatial distance weight matrix, spatial adjacency weight matrix, and digital economy geographic matrix. The p-values for these tests are all below 5%、10% or 1%, confirming that the Spatial Distance Weight Matrix is best modeled by the SDM model, while the Spatial Adjacency Weight Matrix and Digital Economy Geographic Weight Matrix are best modeled by the SAR model, effectively capturing spatial dependencies in all three matrices. The LM test diagnoses spatial dependence due to error correlations, while the LR test assesses lag dependence across regions. Both tests help identify the nature of spatial relationships, thereby refining the choice of the spatial econometric model. Specifically, for the Spatial Distance Matrix, the LM test for both Spatial Error and Spatial Lag shows p-values below 1%. For the Spatial Adjacency Matrix, the LR test results for both SEM and SAR are significant at the 5% level, consistent with the Wald test conclusions, rejecting the possibility of the SDM degenerating into the SEM or SAR models. The Digital Economy Geographic Matrix shows that the LM test for Spatial Lag is significant below the 1% level, and both the LR and Wald tests support the SDM degenerating into the SAR model, further confirming the applicability of the SAR model for all regions. Although the LM test for the Spatial Adjacency Weight Matrix initially suggests the SDM model, the LR and Wald test results are not significant at the 10% level, supporting the selection of the SEM and SAR models. Next, the Hausman test is conducted to compare the fixed effects model and the random effects model. The statistics for the Spatial Distance Weight Matrix and the Spatial Adjacency Weight Matrix are positive, while the statistics for the Digital Economy Geographic Weight Matrix are negative. The prob>chi2 values for the Spatial Distance Matrix is 1.000, for the Spatial Adjacency Matrix is 0.975, and for the Digital Economy Geographic Matrix is 0.982. These results do not significantly support the null hypothesis, indicating that the random effects model is not appropriate.

1. There are significant disparities in FEPVRE across regions in China, particularly between the eastern and central/western regions. To promote a comprehensive regional green transformation, cross-provincial and cross-regional cooperation should be enhanced. This can be achieved through information sharing, technical support, and experience exchange. By learning from the successful experiences of more efficient regions, less efficient areas can improve their FEPVRE. In this context, leadership coaching can play a key role in fostering collaboration and enhancing team-level knowledge creation, ultimately supporting better environmental performance. Leadership coaching positively influences knowledge creation within teams, leading to improved environmental outcomes (Khan et al., 2024). By promoting such knowledge exchange, regions with lower FEPVRE can enhance their efficiency. Additionally, local regions should strengthen policy coordination between the eastern and central/western regions, increasing investment in green development technologies and financial support to accelerate the green transformation in the central and western regions. This approach will help narrow the regional disparities further.

The study has limitations, such as the exclusion of factors like the institutional environment and governance performance, which could provide a more comprehensive understanding of FEPVRE. Additionally, the focus on China limits the applicability of the findings to other regions with different ecological, economic, and technological contexts. Despite these limitations, the study makes significant contributions by integrating digital technologies into the analysis of FEPVRE, particularly in low-efficiency regions, and exploring the role of regional policies. This approach differentiates it from existing studies, which often overlook digital innovation and policy aspects. By combining digital, ecological, and policy factors, the study offers a more holistic perspective. Future research should explore the role of digital technologies in improving FEPVRE, the integration of ecological protection with green economic development, and the long-term dynamics of FEPVRE, considering policy, technology, and market factors. Cross-country comparisons would further enhance FEPVRE policies in China, advancing both research and policymaking in the field.