Government Environmental Protection Expenditure and Regional Green Innovation: The Moderating Role of R&D Element Flow in China

Abstract

Local governments assume the crucial responsibility of advancing regional environmental regulation and protection and fostering green innovation in development. This paper takes the provincial-level data from 2007 to 2018 in China, and investigates how government environmental protection expenditure (GEPE) influences regional green innovation. Also, a gravity model is constructed to figure out R&D element flow, and the moderating mechanisms of the flow of R&D personnel and R&D capital are further examined. The empirical evidence shows that GEPE significantly promotes regional green innovation (coefficient = 0.185, p < 0.01), with robustness confirmed through lagged effect tests, indicating sustained positive impact. Mechanism analysis indicates that R&D personnel flow significantly strengthens the positive effect of GEPE on regional green innovation (interaction coefficient = 0.016, p < 0.01), while the moderating effect of R&D capital flow is statistically insignificant. The spatial Durbin model further confirms that the impact of GEPE on green innovation has a spatial spillover effect in neighboring regions. Additionally, excessive environmental decentralization suppresses the positive influence of GEPE on regional green innovation. These findings provide empirical evidence for local governments to promote regional green innovation through fiscal expenditures. It emphasizes the necessity of giving full play to the guiding and “leveraging” role of government environmental governance expenditure while fostering a synergistic effect between government environmental protection expenditure and the free flow of R&D elements, ultimately promoting coordinated green development in regions.

1. Introduction

The large-scale utilization of fossil fuels since the Industrial Revolution has significantly advanced human development while simultaneously triggering systemic global crises, including resource depletion, climate change, and geopolitical conflicts [1,2,3,4]. The escalating global energy crisis highlights the critical interconnection between energy security and climate policy, establishing this relationship as a primary focus of international attention. The planetary boundaries framework, as proposed by Rockström et al. [5,6], establishes that human activities have transgressed the safe operating space of the Earth system in crucial areas, including climate change, biodiversity loss, and so on [7]. This evidence confirms that the traditional industrial model driven by fossil fuels presents a substantial challenge to global sustainable development. This reality has thus cemented global consensus on the urgent need to expedite energy transitions and embrace Industry 5.0, which emphasizes sustainability, resilience, and human-centricity [8,9]. Industry 5.0 explicitly states that the fundamental objectives of future industries must align with the United Nations Sustainable Development Goals (SDGs). Green innovation, however, is an essential pathway to achieve these goals.

In response, Chinese government has implemented a strategic initiative to promote green development and accelerate ecological civilization construction [10], while reforming its environmental governance system to strengthen local governments’ environmental accountability [11]. Meanwhile, this was further institutionalized through the Reform Plan for the Division of Central—Local Fiscal Responsibilities in Ecological and Environmental Protection in 2020, which explicitly enhanced local governments’ expenditure obligations in ecological governance. Currently, China’s environmental governance is heavily dependent on governmental regulation, particularly in local governments’ revenue and expenditure operations at every level [12,13]. According to the China Statistical Yearbook, since 2007, China has incorporated environmental expenditure into fiscal budget management, the scale of GEPE experienced significant growth, expanding from RMB 96.12 billion to RMB 521.38 billion in 2022, representing an average annual growth rate of approximately 13%. Notably, in 2022, local environmental protection expenditure was 29 times that of the central government.

GEPE, being an essential measure for local governments in environmental governance, has a direct influence on green innovation within their administrative regions. Some literature indicates that such expenditure can drive green technological advancement by reducing uncertainty in research and development activities and optimizing innovation factors’ allocation [14,15,16]. Specifically, GEPE directly acts on green innovation projects. Through means like financial subsidies, the government furnishes substantial financial backing to entities engaged in green innovation. This effectively mitigates the uncertainties inherent in the R&D process of green technology innovation. It reduces the complexity of R&D for fundamental invention patent projects as well [17]. On the other hand, GEPE can deliberately steer the optimization of the allocation of regional green innovation resources. Such expenditure not only supplies innovation-supporting resources for R&D personnel and research teams but also creates a favorable policy environment and green innovation platform for the development of green capital and industries. Thus, it effectively promotes the advancement and execution of regional green innovation R&D programs [18].

It is noteworthy that, in practice, local government environmental expenditure policies typically undergo adaptive modifications by referencing neighboring jurisdictions’ governance approaches while incorporating local conditions, demonstrating pronounced strategic interactions [19,20]. The frequent interregional economic and cultural exchanges further exert profound impacts on regional green development. This phenomenon reveals close interregional spatial spillover effects in green technology innovation across regions [21,22]. Moreover, there exists certain spatial heterogeneity in environmental protection expenditures, technological innovation, and economic development among different regions. Such circumstances render the investigation of green technology innovation from an interregional perspective all the more crucial.

As key components of regional innovation activities, R&D elements also exhibit notable spatial heterogeneity across regions. Existing studies have focused on the direct effects of GEPE [23,24], ignoring the critical dynamic mechanism that propels its efficacy—the attraction and integration of cross-regional R&D elements. Specifically, talent serves as the carrier of knowledge and technology, while capital can stimulate corporate innovation. Driven by factors such as profit-seeking behavior, these elements flow across regional boundaries, forming connections among innovation entities in different regions [25]. GEPE is regarded as a strong policy signal for green development, based on signal theory, effectively guiding the cross-regional flow of scarce R&D personnel and capital, thereby inducing either the outward diffusion or inward agglomeration of innovation resources. Consequently, during the economic green transition, it is imperative to consider the impact and complement of R&D element inflows from neighboring regions, such as R&D personnel or R&D capital, on the effectiveness of local government environmental protection expenditure policies from the viewpoint of regional environmental governance. This ensures the precise support of GEPE in green innovation and the “leveraging” effect on R&D resources.

In view of this, the main questions examined in this paper are as follows:

• RQ1: Does local GEPE significantly promote regional green innovation?
• RQ2: How do R&D personnel flow and R&D capital flow moderate this relationship between GEPE and regional green innovation?

To address these issues, this article empirically examines the green innovation effect of GEPE, with particular focus on how it can alleviate talent and capital constraints in regional green innovation by adopting provincial-level panel data from China and fixed-effect models. Furthermore, the study also delves into GEPE’s impacts on provincial-level green innovation, and the spatial spillover effects it generates in neighboring regions, as well as the influence of the government’s environmental decentralization on regional green innovation. Several findings are accordingly presented in this paper. First, we find that GEPE positively affects regional green innovation in a significant way. Second, compared to the R&D capital element, the flow of R&D personnel impact on the relationship between the two more strongly. Third, GEPE’s impact on green innovation is characterized by spatial spillover effects. Moreover, an excessive degree of environmental decentralization can serve as a deterrent, substantially impeding the positive influence that GEPE exerts on regional green innovation. These findings provide theoretical support for exploring the long-term development mechanism of regional green innovation under the joint role of GEPE and R&D element flow, which is free from one region to another.

Based on the primary findings presented above, this paper offers three major scholarly contributions.

• In contrast to existing studies primarily concentrating on the micro-enterprise level [26,27], this paper adopts a macro-level perspective by examining regional innovation systems as integrated wholes to investigate the influence of GEPE on green innovation. Furthermore, this paper deeply reveals their relationship by using spatial econometric methods after fully considering GEPE’s spatial attributes. This reduces the estimation errors that may exist in previous research findings.
• This paper introduces the gravity model to measure R&D element flow and integrates GEPE, R&D element flow, and green innovation into a unified analytical framework, which deeply analyzes the catalyst effect played by R&D element flow. Meanwhile, it offers fresh ideas for further strengthening green innovation performance.
• From the perspective of environmental decentralization, this paper examines the critical moderating role of provincial governments’ environmental administrative autonomy in the relationship between GEPE and regional green innovation. Based on these findings, precise implications are offered for optimizing intergovernmental environmental governance structures and implementing differentiated regional green innovation incentive policies, thus improving the spatial allocation efficiency of environmental expenditures.

2. Literature Review and Hypothesis Development

2.1. Literature Mapping and Gap Identification

To enhance the transparency and structural rigor of the literature review, this study adopts the systematic literature review (SLR) approach proposed by Tranfield et al. [28], in conjunction with the Theory–Context–Construct–Methods (TCCM) framework developed by Paul and Rosado-Serrano [29]. Relevant publications retrieved from Web of Science (WoS) databases are systematically reviewed. The selection criteria include (1) articles published in journals indexed in the SSCI or SCI databases; (2) studies focusing on the interactions among GEPE, green innovation, and R&D elements; and (3) empirical analyses based on China or comparable emerging economies. A total of 35 core studies have been identified.

Under the TCCM framework, existing studies can be summarized as follows. Theoretically, most research is grounded in the Porter Hypothesis and the New Economic Geography theory. Contextually, the majority of studies rely on data at the city or firm level. In terms of research characteristics, prior studies mainly examine the static effects of GEPE, while the dynamic linkage between GEPE and R&D element flow has received relatively limited attention. Methodologically, spatial econometric models such as Spatial Durbin (SDM) models have been increasingly employed in recent years to measure cross-regional spillover effects. Collectively, these strands of the literature provide a solid theoretical foundation for this study.

The analysis points out a significant research gap: few studies have examined how the flow of R&D personnel and capital functions as a market-based adjustment mechanism that influences the effectiveness of GEPE in promoting green innovation. This study aims to fill this gap.

2.2. GEPE and Regional Green Innovation

GEPE is of great significance in facilitating regional green technological innovation and driving the transition towards a green economy. The topic of government environmental protection and green innovation mainly revolves around the Porter Hypothesis [30,31,32,33]. In the short term, it can effectively incentivize enterprises and research institutions to carry out technological innovation by fiscal measures such as increasing GEPE and implementing tax cuts and fee reductions [34]. This, in turn, helps to raise regional environmental standards, reduce haze pollution and the emissions of various pollutants [35,36,37], and achieve the goal of environmental governance. In the long term, GEPE influences the regional industrial structure and stimulates green innovation. As a result, it can promote economic transformation and upgrading [38,39,40] and regional green and low-carbon development as well [41]. Additionally, these effects can assist in maintaining a balance between local environmental governance and economic development [42], ultimately achieving the goal of regional sustainable development.

Compared with traditional technological innovation, green innovation activities integrate the functions of innovation and environmental protection, which is of great significance for mitigating regional resource and environmental pressures and promoting sustainable economic development [43]. GEPE, serving as both the material guarantee and direct manifestation of the government’s fulfillment of environmental protection functions, can enhance the green technology innovation output of the two main entities of regional green innovation, namely, enterprises and research institutions. Specifically, GEPE can play the following roles for regional green innovation entities. First, it provides R&D funds and reduces R&D costs [44,45]. The R&D process of green technology innovation, especially the development of fundamental invention patents, is fraught with uncertainties and requires substantial funding and talent support, and necessitates that innovation entities possess a higher capacity for risk-taking [46]. Thus, in order to promote regional green innovation, the government needs to optimize resource allocation and leverage the financial support role of regional fiscal expenditure. Moreover, GEPE can fund enterprises’ and research institutions’ green technology R&D through fiscal subsidies and direct appropriations. This stabilizes innovators’ outcome expectations and boosts their R&D confidence [47].

Second, it attracts scientific and technological talent and green capital through market-based resource allocation. In the process of regional governance, GEPE can provide corresponding innovation-supporting resources. Through various fiscal expenditure projects related to environmental management and pollution prevention, these will help address market failures in green technology innovation. They offer essential software and hardware infrastructure for green technology innovation activities, foster the development of green industries, support the cultivation of talent and research teams, and provide macro-level innovation platforms for R&D, such as the establishment of innovation parks and laboratories. In addition, GEPE reflects the government’s environmental preferences. Under conditions where there are significant differences in the economic geography and the level of green innovation across different regions [48], the differing orientations of GEPE can lead to disparities in regional public services. In regions at the forefront of environmental governance, a well-developed green and low-carbon industry environment, along with a livable environment for talent, further promotes the aggregation of green technology talent and the inflow of green capital [49].

Hypothesis 1.

Local government environmental protection expenditure can improve regional green innovation.

2.3. GEPE, R&D Element Flow and Regional Green Innovation

At the regional level, R&D elements exhibit two essential forms: stock and flow. The stock of R&D elements reflects the total amount of accumulated R&D resources within a region at a given point in time, such as the number of R&D personnel and annual R&D capital expenditures. It represents the region’s static innovation potential. In contrast, the flow of R&D elements captures the dynamic transfer and interaction of these elements across regions, serving as a core mechanism that drives knowledge spillovers and spatial reallocation [50]. As put forward by Gruber et al. [51], R&D elements encompass both R&D personnel and R&D capital. To be specific, the former refers to the flow of R&D researchers, technologists, and talent, while the latter represents the financial flow from governments, companies, and foreign investments. R&D element is realized through two key processes: inflow and outflow, and is also characterized by its scarcity and a preference for excellence. Due to regional disparities in the economic environment, innovation ecology, and other aspects, R&D resources’ distribution varies across regions and changes as R&D element flow from one region to another. Typically, R&D elements generally concentrate in regions with superior resources [52] and gravitate toward relevant fields. Furthermore, the spatial flow of R&D elements creates connections between innovative systems of each region, facilitates regional technological spillovers, and enlarges the scale of regional innovation production. This process continuously optimizes the spatial efficiency of innovation resource allocation [53,54].

With regard to R&D personnel flow, the population’s free flow is a primary pathway for achieving the equalization of basic public services and optimizing resource allocation across regions. R&D personnel are inclined to converge in regions with greater development prospects and better welfare through a form of “voting with their feet” [55]. First, higher levels of GEPE send a clear signal of the government’s long-term commitment to the green industry, effectively reducing the perceived career risk among R&D personnel and making them more willing to devote their expertise to the highly uncertain field of green technological innovation. Second, GEPE fosters high-quality employment opportunities and superior research environments by funding green innovation parks and laboratories, thereby reshaping talent behavior and attracting R&D professionals to flow into these sectors. Consequently, when R&D personnel flow is high, GEPE funds can be more effectively integrated with skilled human capital, maximizing their transformation into green innovation outcomes through mechanisms such as knowledge spillovers, employment, and learning effects.

To be specific, regions with higher GEPE attract talent inflows by fostering the expansion of green and environmental protection industries, establishing research institutes, and improving talent incentives. This effectively supplements the green technologies and innovative talent required for regional green innovation, thereby promoting its development. For one thing, R&D personnel flow facilitates technological spillovers and the diffusion of advanced technologies across regions. With the inflow of R&D personnel, the aggregate amount of innovation activities in both the destination and source regions increases, thereby enhancing the overall regional innovation level [56]. In innovation centers, the concentration of R&D personnel enables the pooling of superior innovative human resources, accelerating the diffusion of tacit knowledge over a larger area. This, in turn, fosters green innovation through the hiring and learning effects. In non-innovation center regions, the flow of R&D personnel stimulates the interaction and exchange of green innovation technologies and knowledge among different entities, intensifying competition and learning among innovation subjects. With the added impact of improved transportation accessibility, it enhances communication and connectivity between innovation centers and other regions, encourages learning from regions with higher innovation levels, promotes embedded cooperation among innovation entities in different regions, and generates a diffusion effect [57].

On the other hand, the flow of R&D personnel contributes to improving regional resource allocation efficiency. R&D resources across regions are heterogeneous and generally follow a trajectory from low-productivity regions to high-productivity ones [58]. R&D personnel flow not only helps optimize the regional innovation elements’ structure but also improves resource allocation efficiency. This flow also prevents the misallocation of R&D resources within the same region, which can negatively impact regional R&D efficiency [59], thus better promoting regional green innovation.

Hypothesis 2.

The flow of R&D personnel can promote GEPE’s impact on regional green innovation.

With regard to the flow of R&D capital, driven by market self-regulation and government intervention, it will actively flow into regions with high profit returns. Green innovation activities are typically characterized by a “dual externality” of large investment scale, long R&D cycles, and high failure risk, which makes R&D capital hesitant to enter. The role of GEPE lies in its ability to fundamentally alter the risk–return calculation of investors. On the one hand, GEPE provides the initial funding for green projects through fiscal subsidies and direct grants, significantly reducing the entry risk for capital. On the other hand, government financial support serves as a strong endorsement of project feasibility, changing market expectations among capital holders and making them more willing to invest their R&D capital in such projects. Through this leverage effect, GEPE mobilizes capital and fosters synergy between government and market forces. This collaboration ensures the establishment of a robust financial safeguard for enterprises and research institutions, effectively alleviating financial constraints in green innovation projects and, consequently, boosting regional green innovation capacities. Besides this, investment is a crucial factor in economic growth. R&D capital flow can generate innovation agglomeration effects through investment activities [60], stimulate market competition, and the cross-regional flow of green technology talents. Consequently, it forms a synergy with government fiscal expenditure, collectively enhancing the regional performance in green technological innovation.

Hypothesis 3.

R&D capital flow can promote GEPE’s influence on regional green innovation.

3. Materials and Methods

3.1. Data Source

This paper collects data from 2007 to 2018 for each province in the Chinese region, while Xizang, Hong Kong, Macao, and Taiwan are excluded due to missing data. Most data are sourced from the China statistical yearbooks, including the China City Statistical Yearbook, China Environmental Yearbook, China Environmental Statistical Yearbook, and China Science and Technology Statistical Yearbook. The other data are sourced from the China Stock Market & Accounting Research (CSMAR) database and the Wind database. Moreover, in regard to green patent information, we collect patent data from the Chinese Research Data Services (CNRDS) platform. This platform provides comprehensive patent information documented in Chinese, ensuring a rich and detailed dataset for analysis.

Additionally, it should be noted that this study includes a discussion on hysteresis effects. Considering the availability of variable data, we conduct supplementary tests using green innovation in periods t + 1 and t + 2 as the dependent variables. The observation period for the selected green innovation variables is from 2007 to 2020.

3.2. Variable Definition and Summary Description

3.2.1. Main Variable Definition

Regarding the core dependent variable, regional green innovation (Ingrva_i,t) is quantified through the aggregate count of green patent applications for each province during a specific year, with the data treated logarithmically. This approach aligns with prior studies that measure green innovation through its convergence with green technology development, utilizing green outcomes as a reliable indicator [61,62]. Based on the previous research experience [63,64], and considering that there may be possible lag time effects of patent applications and GEPE, this study also incorporates green innovation in the 1st period (Ingrva_i,t+1) and the 2nd period (Ingrva_i,t+2) as the dependent variables in supplementary tests.

GEPE (Envircost_i,t), the central explanatory variable, constitutes the most efficient and direct specialized form of expenditure by local Chinese governments in the realm of environmental governance, and it was treated logarithmically using the data provided by the China Statistical Yearbook [65]. In addition, since the national government first introduced the “Environmental Protection” functional classification under government revenue and expenditure categories in 2007, the observation period for this variable also begins in 2007.

3.2.2. Moderating Variable

This research explores the impact mechanism of R&D element flow, including R&D personnel and R&D capital mentioned above, on the relationship between GEPE and green innovation. As the main variable of this research, R&D element flow is difficult to measure directly in terms of actual interprovincial movements of personnel or capital. Therefore, the spatial association network of element flows is constructed by the gravity model, which is frequently employed to investigate the spatial interactions among macroeconomic data, as demonstrated in previous research [66,67]. This model utilizes “stock” indicators as a proxy for a region’s “mass” or its potential to attract and push factors, which are a necessary input for calculating the estimated flows within the gravity model. Building on the classic gravity model form, this study develops separate gravity models for gauging R&D personnel and capital flow, carefully considering the distinct characteristics of each type of flow.

Commonly used in labor economics to explain the flow of personnel, the term “push–pull” suggests that population flow is a consequence of the combined influence of the pulling forces in the destination region and the pushing forces in the place of origin. As demonstrated by Lesage & Fischer [68] and Chong et al. [69], the flow of personnel corresponds in direct proportion to the attractiveness of region j, from region i to region j. Based on this, this paper establishes a gravity model for calculating R&D personnel flow, selecting attractive influencing variables of each province, incorporating salary, housing price, the degree of environmental pollution, and regional education level. The gravity model formula developed in this research is

$$\begin{gathered} {RDPerFlow}_{i,j}={ln (rdper}_i)\times d_{ij}^{-2}\times \mathit{ln}\left({wage}_j-{wage}_i\right)\times \mathit{ln}\left({price}_j-{price}_i\right) \\ \times \mathit{ln}\left({pollution}_j-{pollution}_i\right)\times \mathit{ln}\left({edu}_j-{edu}_i\right) \end{gathered}$$

(1)

where province i represents the observation province, while province j refers to the other provincial-level administrative units in China. RDPerFlow_i,j denotes the level of R&D personnel flow from province i to province j. rdper_i represents the number of R&D personnel in research institutions, universities, and enterprises within province i. Wage indicates the average wage level of urban employees in either province i or province j. Price refers to the housing price level in the corresponding province, measured by the average residential sales price. Pollution reflects the environmental pollution in the respective province, measured by the industrial smoke and dust emissions. edu represents educational level in the region, measured by regional fiscal expenditure on education. d denotes the geographical distance between provinces i and j. These indicators can be obtained from the CSMAR database and the China Science and Technology Statistical Yearbook. Based on Equation (1), the total flow of R&D personnel from province i in a given year can be calculated by summing the flows, as shown in Equation (2).

$${RDPerFlow}_i={\sum }_{j=1}^n{RDPerFlow}_{i,j}$$

(2)

According to the research of Ahmed [70], the scale of capital flow is influenced by financial development. In addition, R&D capital, driven by the pursuit of profit, tends to flow toward regions with greater income [71]. Therefore, we establish a gravity model regarding R&D capital flow in this paper by selecting the average profit of industrial enterprises (profit) and the financial marketization environment (fin) as attractiveness variables. Then, R&D capital flow’s gravity model is

$${RDCapFlow}_{i,j}={ln(rdcap}_i)\times d_{ij}^{-2}\times \mathit{ln}\left({profit}_j-{profit}_i\right)\times \mathit{ln}\left({fin}_j-{fin}_i\right)$$

(3)

rdcap_i is measured by the R&D expenditure of individual provinces; profit is denoted by the large-scale industrial enterprises’ average profit; and fin is measured as the proportion of new loans, which are granted by regional financial institutions, relative to the region’s new fixed asset investment. Similarly to R&D personnel, the total flow of R&D capital is shown in Equation (4):

$${RDCapFlow}_i={\sum }_{j=1}^n{RDCapFlow}_{i,j}$$

(4)

3.3. Empirical Methodology Design

To test Hypothesis H1, we construct a model to examine how GEPE impacts regional green innovation, as follows:

$${Ingrva}_{i,t}={\beta }_0+{\beta }_1{Envircost}_{i,t}+{\beta }_{\gamma }{\Gamma }_{i,t}+{Year}_{i,t}+R{egion}_{i,t}+{\epsilon }_{i,t}$$

(5)

Furthermore, in order to test Hypotheses H2 and H3, which are about the impact mechanism of R&D element flow, we introduce the interaction term based on Equation (5), and construct the model of the moderating effect as shown in Equations (6) and (7):

$$\begin{gathered} {Ingrva}_{i,t}={\beta }_0+{\beta }_1{Envircost}_{i,t}+{\beta }_2{RDPerFlow}_{i,t}+{\beta }_3{Envircost}_{i,t}\times {{RDPerFlow}_{i,t}+\beta }_{\gamma }{ \Gamma }_{i,t}+ {Year}_{i,t} \\ +R{egion}_{i,t}+{\epsilon }_{i,t} \hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em} \end{gathered}$$

(6)

$$\begin{gathered} {Ingrva}_{i,t}={\beta }_0+{\beta }_1{Envircost}_{i,t}+{\beta }_2{RDCapFlow}_{i,t}+{\beta }_3{Envircost}_{i,t}\times {{RDCapFlow}_{i,t}+\beta }_{\gamma }{ \Gamma }_{i,t}+{Year}_{i,t} \\ +R{egion}_{i,t}+{\epsilon }_{i,t} \hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em} \end{gathered}$$

(7)

where i represents the province, and t represents the year. ${\Gamma }_{i,t}$ denotes a series of control variables, as following: industrial structure (IndStruc), measured by proportion of secondary industry output in GDP for each province in a given year; urbanization (Urbanrate), measured by proportion of urban population in total population; per capita GDP growth rate (GDPgrowth1); GDP growth rate (GDPgrowth2); industrial pollution emissions, including wastewater (Envirct1), sulfur dioxide (Envirct2), smoke and dust (Envirct3), measured by natural logarithm of industrial pollution emissions for each province in a given year. Year denotes time fixed effects. Region denotes province fixed effects.

3.4. Summary Description

Table 1 presents the summary statistics values of main variables. For the core variables, the statistical results reveal the following: regional green innovation (Ingrva_i,t), as our dependent variable, has a mean of 6.788, varying from 2.639 to 10.150, indicating significant variation in green innovation levels among the provinces included in the sample. Additionally, the average value of the green innovation variable increases annually. In particular, our main independent variable GEPE (Envircost_i,t) has a mean of 4.419 and fluctuates from 2.272 to 6.071, suggesting considerable variation in environmental protection spending across provinces and years. Concerning the moderating variable (R&D element flow), including R&D personnel flow (RDPerFlow_i,t) and R&D capital flow (RDCapFlow_i,t), the former’s minimum value is close to 0.000, the maximum value is 42.820, and the latter ranges from −0.029 to 0.051, indicating some differences in the degree of R&D element inflow and outflow between provinces. However, the median values of both R&D personnel and R&D capital flow suggest that the overall level of R&D element flow remains comparatively low.

Table 1. Summary statistics.

Variable	Obs.	Mean	Median	Std. Dev.	Min.	Max.
Ingrvai,t	360	6.788	6.796	1.595	2.639	10.150
Ingrvai,t+1	360	7.021	7.051	1.557	2.890	10.200
Ingrvai,t+2	360	7.209	7.289	1.488	3.258	10.200
Envircosti,t	360	4.419	4.529	0.743	2.272	6.071
RDPerFlowi,t	360	3.008	0.590	7.557	0.000	42.820
RDCapFlowi,t	360	−0.001	−0.000	0.010	−0.029	0.051
IndStruci,t	360	0.433	0.444	0.081	0.178	0.601
Urbanratei,t	360	0.551	0.532	0.135	0.316	0.893
GDPgrowth1i,t	360	0.121	0.106	0.064	−0.015	0.262
GDPgrowth2i,t	360	0.130	0.114	0.064	0.005	0.265
Envirct1i,t	360	10.520	10.640	1.341	6.211	12.700
Envirct2i,t	360	12.520	12.900	1.597	5.421	14.580
Envirct3i,t	360	11.820	12.200	1.977	0.000	14.890

Notes: Table 1 displays the main variables’ descriptive statistics results, encompassing observation numbers, mean, median, standard deviation, as well as minimum and maximum values.

In addition, with respect to our provincial-level control variables, industrial structure (IndStruc_i,t) ranges from 0.178 to 0.601, with the mean and median values being similar, reflecting substantial variation in the proportion of the secondary industry across provinces. Urbanization (Urbanrate_i,t) is around 0.551, ranging from 0.316 to 0.893. Furthermore, the average of per capita GDP growth rate (GDPgrowth1_i,t) is 0.121, and the average regional GDP growth rate (GDPgrowth2_i,t) is 0.130. Additionally, for the pollution-related variables (Envirct1_i,t, Envirct2_i,t, Envirct3_i,t), the means and medians are close.

Figure 1 illustrates a steady upward trend in regional green innovation (Ingrva) over the sample period, increasing from approximately 5.32 in 2007 to 8.13 in 2018. This provides a descriptive foundation for the empirical analysis and implies that China’s focus on environmental protection and innovation has yielded observable progress over time.

Figure 1. Trend of reginal green innovation.

4. Results

4.1. Baseline Regression Result

Fixed-effects regression methodology with robust standard errors is employed to check the impact of GEPE on regional green innovation, and the empirical findings are reported in Table 2. Column (1) presents the regression results in the full sample for the current year. The empirical results show that the estimated coefficient for GEPE is highly momentous and positive at the 1% level. This indicates that such expenditure exerts a pronounced promoting influence on regional green innovation, thereby confirming Hypothesis H1 and providing a positive answer to RQ1, which was raised in the Introduction. Furthermore, considering the potential lag in the impact of GEPE and the time delay in patent applications, we conduct additional tests using lagged terms of green innovation. The columns (2–3) show that the estimated coefficients of lagged green innovation variables: Ingrva_i,t+1 and Ingrva_i,t+2, remain significantly positive, further confirming the robustness of the conclusion.

Table 2. Baseline regression results.

Variables	(1) Ingrvai,t	(2) Ingrvai,t+1	(3) Ingrvai,t+2
Envircosti,t	0.185 ***	0.132 **	0.179 ***
	(3.296)	(2.166)	(2.959)
IndStruci,t	2.472 ***	2.913 ***	3.202 ***
	(4.647)	(6.077)	(7.392)
Urbanratei,t	3.240 ***	2.039 **	1.913 **
	(3.204)	(2.060)	(2.022)
GDPgrowth1i,t	−9.010 ***	−7.811 ***	−5.626 ***
	(−4.299)	(−3.886)	(−2.754)
GDPgrowth2i,t	9.513 ***	8.757 ***	5.869 ***
	(4.516)	(4.404)	(2.897)
Envirct1i,t	−0.097 **	−0.108 ***	−0.060
	(−2.037)	(−2.620)	(−1.372)
Envirct2i,t	−0.041 **	−0.020	−0.014
	(−2.010)	(−0.865)	(−0.657)
Envirct3i,t	0.022 ***	0.015 **	0.003
	(3.923)	(2.176)	(0.374)
Constant	4.419 ***	5.565 ***	5.559 ***
	(4.680)	(5.953)	(6.336)
Year FE	YES	YES	YES
Province FE	YES	YES	YES
Observations	360	360	360
R2	0.984	0.984	0.983
Adj R2	0.982	0.982	0.981

Notes: Table 2 provides the basic regression results for the exploration GEPE’s impact on regional green innovation, as well as testing green innovation’s lag effect. t-statistics are in parentheses. *** p <0.01, ** p < 0.05.

4.2. Exploration of R&D Element Flow Mechanism

The baseline regression results provide compelling evidence that GEPE significantly promotes regional green innovation. This section further investigates the effect of R&D element flow on regional green innovation in conjunction with GEPE. While GEPE directly supports green innovation through financial incentives and policy guidance, its effectiveness may vary depending on the regional capacity to absorb and utilize innovation resources. However, R&D element flow can enhance this capacity by improving the allocation efficiency of innovation inputs. Specifically, the mobility of skilled personnel facilitates knowledge exchange and spillover, while the movement of R&D capital directs financial resources toward regions with stronger innovation potential. These interactions strengthen the transmission mechanism of GEPE, enabling regions to better translate fiscal environmental efforts into substantive innovation outcomes.

Table 3 presents the exploration of R&D element flow mechanism along two dimensions: R&D personnel flow and R&D capital flow. Panel A of Table 3 reports results (1)–(3) for the mechanism of R&D personnel flow. The findings indicate that the coefficient of interactor of GEPE and R&D personnel flow is positive and significant at the 1% level. Moreover, this positive effect persists for green innovation in periods t + 1 and t + 2, indicating that R&D personnel flow amplifies the effect of GEPE on regional green innovation. In other words, the promotional impact of GEPE on green innovation becomes more evident as the flow of R&D personnel increases. Thus, Hypothesis H2 can be verified.

Table 3. R&D element flow mechanism explorations.

Panel A: R&D Personnel Flow
Variables	(1)	(2)	(3)
	Ingrvai,t	Ingrvai,t+1	Ingrvai,t+2
RDPerFlowi,t	−0.060 ***	−0.048 ***	−0.035 **
	(−4.438)	(−2.796)	(−2.276)
Envircosti,t × RDPerFlowi,t	0.016 ***	0.012 ***	0.008 **
	(5.451)	(3.116)	(2.347)
Envircosti,t	0.150 ***	0.104 *	0.157 **
	(2.743)	(1.705)	(2.583)
Control	YES	YES	YES
Year FE	YES	YES	YES
Province FE	YES	YES	YES
Observations	360	360	360
R2	0.985	0.985	0.984
Adj R2	0.983	0.983	0.981
Panel B: R&D capital flow
Variables	(1)	(2)	(3)
	Ingrvai,t	Ingrvai,t+1	Ingrvai,t+2
RDCapFlowi,t	1.504	2.370	−0.858
	(0.165)	(0.265)	(−0.167)
Envircosti,t × RDCapFlowi,t	−0.601	−0.937	−0.460
	(−0.297)	(−0.468)	(−0.376)
Envircosti,t	0.189 ***	0.138 **	0.184 ***
	(3.358)	(2.226)	(2.986)
Control	YES	YES	YES
Year FE	YES	YES	YES
Province FE	YES	YES	YES
Observations	360	360	360
R2	0.984	0.984	0.983
Adj R2	0.981	0.982	0.981

Notes: Table 3 displays the roles of R&D personnel flow and R&D capital flow mechanisms in affecting the relationship between GEPE and regional green innovation by using fixed-effects models. t-statistics are in parentheses. *** p < 0.01, ** p < 0.05, * p < 0.1.

Panel B of Table 3 reports results (1)–(3) for the moderating effect of R&D capital flow between GEPE and regional green innovation. In contrast to R&D personnel flow, the interaction term fails to reach statistical significance, demonstrating that R&D capital flow does not enhance this relationship. As a result, Hypothesis H3 is not supported. Referring to the theoretical analysis by Audretsch & Feldman [72] and Diebolt & Hippe [73], this may be due to the relatively smaller amount of knowledge and technology content embedded in R&D capital compared to personnel, making its influence on regional green innovation activities less direct. Additionally, the scale of R&D capital inflows is relatively small compared to GEPE, further limiting its impact.

Consistent with the regression results, Figure 2 further depicts the relationship between GEPE and regional green innovation across different levels of R&D element flows using fitted scatter plots. Figure 2a shows that regions with higher R&D personnel flow exhibit a steeper slope, indicating a stronger amplification effect of GEPE on regional green innovation. In contrast, the slopes in Figure 2b remain largely parallel across R&D capital–flow groups, suggesting that capital flow does not materially alter this relationship. These visual patterns reinforce the empirical findings that R&D personnel flow significantly strengthens this relationship, whereas R&D capital flow does not demonstrate a moderating role.

Figure 2. Relationship between GEPE and regional green innovation under different levels of R&D element flows (fitted scatter plots). Panel (a) groups sample regions by R&D personnel flow, while Panel (b) groups by R&D capital flow. Note: Scatter points are omitted for clarity; fitted lines represent the linear relationship between GEPE and regional green innovation across subgroups of R&D element flow.

In conclusion, the results presented in Table 3 and Figure 2 provide an answer to RQ2 from the Introduction. Specifically, the flow of R&D personnel exerts a significant positive influence on the relationship between GEPE and regional green innovation, whereas the flow of R&D capital has no significant effect on this relationship.

4.3. Robustness Test

To ensure the reliability and robustness of our fundamental results, we make use of the quantity of granted green patents (Ingrvg_i,t) as an alternative indicator of regional green innovation, instead of green patent applications. Considering that obtaining patent grants typically takes longer than applications, the fixed-effects model is re-estimated using the dependent variable from period t + 1 to t + 2. As reported in Table 4, the results remain consistent with the baseline regressions. Specifically, column (1) in Panel A shows that the coefficient of GEPE is positive and statistically significant at the 1% level. This finding provides evidence that local GEPE consistently and significantly promotes regional green innovation, whether measured by patent applications or patent grants, thereby confirming the robustness of Hypothesis H1.

Table 4. Robustness test: alternative measurement of green innovation.

Panel A: Robustness Test of Baseline Regression
Variables	(1)	(2)	(3)
	Ingrvgi,t	Ingrvgi,t+1	Ingrvgi,t+2
Envircosti,t	0.181 ***	0.077	0.113 **
	(2.862)	(1.307)	(2.355)
Control	YES	YES	YES
Year FE	YES	YES	YES
Province FE	YES	YES	YES
Observations	360	360	360
R2	0.983	0.985	0.986
Adj R2	0.980	0.982	0.983
Panel B: Robustness test of R&D personnel flow
Variables	(1)	(2)	(3)
	Ingrvgi,t	Ingrvgi,t+1	Ingrvgi,t+2
Envircosti,t × RDPerFlowi,t	0.009 ***	0.011 ***	0.009 ***
	(4.012)	(5.790)	(4.565)
Control	YES	YES	YES
Year FE	YES	YES	YES
Province FE	YES	YES	YES
Observations	360	360	360
R2	0.984	0.985	0.986
Adj R2	0.981	0.983	0.984
Panel C: Robustness test of R&D capital flow
Variables	(1)	(2)	(3)
	Ingrvgi,t	Ingrvgi,t+1	Ingrvgi,t+2
Envircosti,t × RDCapFlowi,t	−2.465	−1.975	−0.154
	(−1.070)	(−1.430)	(−0.099)
Control	YES	YES	YES
Year FE	YES	YES	YES
Province FE	YES	YES	YES
Observations	360	360	360
R2	0.983	0.985	0.986
Adj R2	0.980	0.982	0.984

Notes: Table 4 displays GEPE’s influence on regional green innovation and the role of R&D element flow between this by using alternative measurements of green innovation. t-statistics are in parentheses. *** p < 0.01, ** p < 0.05.

Moreover, Panel B confirms that the interaction term of R&D personnel flow (H2) remains significantly positive. This result indicates that the knowledge spillover and talent complementarity effects generated by R&D personnel flow serve as an important mechanism for the positive impact of GEPE on green innovation. In contrast, Panel C shows that the interaction term of R&D capital flow (H3) remains statistically insignificant. This result is consistent with the earlier regression results, suggesting that the moderating effect of R&D capital flow may be more sensitive to the measurement of green innovation and model specification.

Taken together, these findings demonstrate that both the promoting effect of GEPE and the moderating role of R&D personnel flow are robust across different measures of green innovation, which provides strong support for the central contribution of this study.

4.4. Endogenous Treatment

To mitigate potential reverse causation and omitted variable problems, we further identify the impact of GEPE on regional green innovation by using an instrumental variable approach. The average GEPE of other provinces in the same year (Envirmean_i,t) is used as an instrumental variable [74]. Theoretically, environmental protection expenditures among neighboring regions are subject to mutual imitation and path dependence, meaning that those of other provinces may influence local government expenditure decisions. However, such expenditures in other regions are unlikely to directly affect green innovation activities in local research institutions and enterprises, thereby satisfying the exogeneity requirement for valid instruments.

The two-stage least squares (2SLS) regression results based on this instrument are shown in Panels A, B, and C of Table 5, corresponding to Equations (5)–(7). The results are consistent with earlier findings and support Hypotheses H1 and H2. Furthermore, the first-stage regression results and the Kleibergen–Paap Wald rk F-statistics confirm the validity and strength of the instrumental variables.

Table 5. Endogenous treatment: instrumental variable method.

Panel A: IV Test for Baseline Regression
Variables	(1)	(2)	(3)
	Ingrvgi,t	Ingrvgi,t+1	Ingrvgi,t+2
Envircosti,t	0.185 ***	0.132 **	0.179 ***
	(3.546)	(2.331)	(3.184)
Control	YES	YES	YES
Year FE	YES	YES	YES
Province FE	YES	YES	YES
Observations	360	360	360
R2	0.984	0.984	0.983
Adj R2	0.982	0.982	0.981
Panel B: IV test for the mechanism of R&D personnel flow
Variables	(1)	(2)	(3)
	Ingrvgi,t	Ingrvgi,t+1	Ingrvgi,t+2
RDPerFlowi,t	−0.059 ***	−0.045 ***	−0.033 **
	(−4.533)	(−2.693)	(−2.118)
Envircosti,t × RDPerFlowi,t	0.015 ***	0.012 ***	0.008 **
	(5.609)	(3.076)	(2.261)
Control	0.170	0.168	0.207 *
Year FE	(1.404)	(1.362)	(1.673)
Province FE	YES	YES	YES
Observations	YES	YES	YES
R2	YES	YES	YES
Adj R2	360	360	360
Panel C: IV test for the mechanism of R&D capital flow
Variables	(1)	(2)	(3)
	Ingrvgi,t	Ingrvgi,t+1	Ingrvgi,t+2
RDCapFlowi,t	1.693	3.528	0.193
	(0.201)	(0.432)	(0.039)
Envircosti,t × RDCapFlowi,t	−0.647	−1.218	−0.715
	(−0.345)	(−0.663)	(−0.596)
Control	0.198 *	0.196 *	0.237 **
Year FE	(1.707)	(1.650)	(1.997)
Province FE	YES	YES	YES
Observations	YES	YES	YES
R2	YES	YES	YES
Adj R2	360	360	360

Notes: Table 5 displays GEPE’s influence on regional green innovation and the moderating role of R&D element flow using the instrumental variable method. t-statistics are in parentheses. *** p < 0.01, ** p < 0.05, * p < 0.1.

4.5. Further Analysis

4.5.1. Spatial Spillover Effects of GEPE on Green Innovation

Spatial factors play a crucial role in environmental and economic research [75]. Specifically, regional disparities in economic development and environmental pollution, combined with intergovernmental competition under fiscal decentralization, place pressure on local governments’ environmental protection expenditure. As a result, such expenditures are not only used for environmental protection and remediation but are also strategically directed toward attracting green capital, innovative talent, and R&D teams. This means that GEPE not only influences local environmental governance and economic development but also exhibits spatial strategic interactions, shaping the fiscal policies of neighboring regions and contributing to regional green innovation through spillover effects. Moreover, a region with more advanced environmental governance and green innovation can attract skilled professionals and R&D teams, generating exponential growth in local green innovation and potentially creating a siphon effect.

Based on this, we extend the analysis of the influence of R&D element flow by examining spatial spillover effects of GEPE on green innovation. Utilizing a spatial proximity matrix, we apply the Spatial Autoregressive Model (SAR) and Spatial Durbin Model (SDM), which are estimated using quasi-maximum likelihood estimation (QMLE) with a spatial proximity matrix. The SDM captures the effect of spatial lag for the explained variable and explanatory variables jointly [76]. This means that the model not only evaluates the effect of local environmental expenditure on local green innovation but also assesses the combined spillover impacts of environmental spending and green innovation levels in neighboring regions on local outcomes. By comparison, the SAR primarily focuses on the spatial dependence of the dependent variable. Specifically, it examines how a region’s level of green innovation is directly influenced by the innovation outputs of its neighboring areas.

The results in Table 6 demonstrate the significant and positive spatial spillover effect of GEPE (Envircost_i,t) on regional green innovation (Ingrva_i,t). Increases in local environmental spending not only directly enhance local green innovation but also significantly promote green innovation in neighboring regions. This effect remains robust when using green innovation from the following one and two years as the dependent variable.

Table 6. Spatial spillover effects of GEPE.

Panel A: Ingrvai,t
Variables	Direct Effect		Indirect Effect		Total Effect
	(1) SAR	(2) SDM	(3) SAR	(4) SDM	(5) SAR	(6) SDM
Envircosti,t	0.310 ***	0.113 **	0.235 ***	0.266 ***	0.546 ***	0.378 ***
	(6.173)	(1.973)	(4.957)	(2.711)	(6.713)	(4.347)
Control	YES	YES	YES	YES	YES	YES
Province FE	YES	YES	YES	YES	YES	YES
Observations	360	360	360	360	360	360
R2	0.542	0.486	0.542	0.486	0.542	0.486
Rho	0.457 ***	0.204 **	0.457 ***	0.204 **	0.457 ***	0.204 **
Sigma	0.049 ***	0.041 ***	0.049 ***	0.041 ***	0.049 ***	0.041 ***
Panel B: Ingrvai,t+1
Variables	Direct effect		Indirect effect		Total effect
	(1) SAR	(2) SDM	(3) SAR	(4) SDM	(5) SAR	(6) SDM
Envircosti,t	0.286 ***	0.085	0.296 ***	0.456 ***	0.582 ***	0.541 ***
	(5.566)	(1.513)	(4.981)	(4.137)	(6.050)	(5.235)
Control	YES	YES	YES	YES	YES	YES
Province FE	YES	YES	YES	YES	YES	YES
Observations	360	360	360	360	360	360
R2	0.545	0.545	0.446	0.545	0.446	0.545
Rho	0.539 ***	0.332 ***	0.539 ***	0.332 ***	0.539 ***	0.332 ***
Sigma	0.049 ***	0.040 ***	0.049 ***	0.040 ***	0.049 ***	0.040 ***
Panel C: Ingrvai,t+2
Variables	Direct effect		Indirect effect		Total effect
	(1) SAR	(2) SDM	(3) SAR	(4) SDM	(5) SAR	(6) SDM
Envircosti,t	0.250 ***	0.193 ***	0.379 ***	0.469 ***	0.629 ***	0.661 ***
	(4.882)	(3.257)	(4.511)	(2.670)	(5.012)	(3.674)
Control	YES	YES	YES	YES	YES	YES
Province FE	YES	YES	YES	YES	YES	YES
Observations	360	360	360	360	360	360
R2	0.545	0.479	0.545	0.479	0.545	0.479
Rho	0.639 ***	0.590 ***	0.639 ***	0.590 ***	0.639 ***	0.590 ***
Sigma	0.049 ***	0.044 ***	0.049 ***	0.044 ***	0.049 ***	0.044 ***

Notes: Table 6 shows the spatial spillover effects results of GEPE with regard to the impact of regional green innovation. t-statistics are in parentheses. *** p < 0.01, ** p < 0.05.

4.5.2. Moderating Role of Government Environmental Decentralization

A scientifically grounded and rational division of intergovernmental environmental governance authority is a crucial precondition and institutional foundation for addressing China’s environmental pollution challenges [77]. From the viewpoint of R&D element flow, regional disparities in environmental decentralization may create institutional “barriers” that influence the effectiveness of GEPE in promoting green innovation. Furthermore, scholars have divergent views on environmental decentralization. Some emphasize its drawbacks, arguing that centralized governance and regional coordination can incentivize governments to provide higher-quality environmental public services and prevent free-rider behavior [78], as well as reduce service provision costs. Excessive local governance power may also prompt governments to lower environmental standards for investment and tax revenue, degrading regional environmental quality. Conversely, others assert that environmental decentralization [79] allows local governments to tailor environmental governance and innovation strategies to regional conditions, thereby improving environmental quality and fostering green innovation.

On the basis of the above analysis, this study follows the methodological approaches of Meng & Chen [80] and Aziz & Bakoben [81] to examine environmental decentralization’s moderating role. The environmental decentralization is calculated as specified in Equation (8), using data from China Environmental Statistical Yearbook and the CSMAR database. In the equation, EnvirRight_i,t represents the degree of environmental decentralization in region i at time t; envpop_i,t, pop_i,t, and GDP_i,t denotes the number of environmental protection personnel, population size, and GDP of province i at time t, respectively; envpop_t, pop_t, and GDP_t represent the corresponding national-level values.

$${EnvirRight}_{i,t}=\left[\right(\left({envpop}_{i,t}\right)/\left({pop}_{i,t}\right))/(\left({envpop}_t\right)/\left({pop}_t\right)\left)\right]\times [1-({GDP}_{i,t})/({GDP}_t\left)\right]$$

(8)

The specific estimation results in Table 7 indicate that while environmental decentralization significantly promotes regional green innovation (column 1) at the significance level of 1%, the interaction term between it and GEPE shows a significant negative effect. This suggests that an increase in local authority over environmental governance weakens the positive impact of GEPE on green innovation. In other words, government environmental decentralization has a crowding effect on GEPE. This result may stem from the fact that, within a certain range, environmental decentralization can positively contribute to regional green innovation. However, in China, the relationship between environmental decentralization and green innovation is still in the phase of the “weak Porter Hypothesis”. As noted by Li & Li [82], environmental decentralization influences green innovation largely depending on the degree of fiscal autonomy available to local governments. When this fiscal autonomy is limited, the effectiveness of environmental decentralization in promoting green innovation is constrained and diminished. Crucially, this limitation forces local authorities to strategically misallocate GEPE towards short-term goals or regulatory avoidance, thereby reducing the policy precision and effectiveness of GEPE in fostering high-risk green innovation. Moreover, excessive environmental decentralization increases the potential for imitation and gaming among regional governments, which undermines the coordination between central and local authorities, as well as interprovincial environmental governance. This may affect the precision of government environmental protection spending decisions on green innovation and the effectiveness of implementation.

Table 7. Environmental decentralization moderating role test.

Variables	(1)	(2)	(3)
	Ingrvai,t	Ingrvai,t+1	Ingrvai,t+2
EnvirRighti,t	1.390 ***	1.159 **	1.601 ***
	(3.696)	(2.500)	(3.357)
Envircosti,t	0.288 ***	0.150	0.328 ***
	(3.079)	(1.421)	(2.753)
EnvirRighti,t × Envircosti,t	−0.319 ***	−0.263 **	−0.371 ***
	(−3.595)	(−2.482)	(−3.508)
Control	YES	YES	YES
Year FE	YES	YES	YES
Province FE	YES	YES	YES
Observations	270	270	270
R2	0.985	0.984	0.984
Adj R2	0.982	0.981	0.981

Notes: Table 7 shows the environmental decentralization moderating role results of GEPE with regard to the impact of green innovation. t-statistics are in parentheses. *** p < 0.01, ** p < 0.05.

Based on the results discussed above, Figure 3 shows the final analysis model of this research.

Figure 3. The final analysis model.

5. Discussion

Building on Krugman’s pioneering new economic geography (NEG) [83], which provides a foundational framework for analyzing the spatial agglomeration of economic activities, this study investigates how GEPE influences regional green innovation through the dynamic mechanism of R&D element flow. The empirical results effectively address the key research questions identified in the Introduction and offer a more integrated analytical perspective for understanding the spatial distribution and driving forces of green innovation.

First, this study empirically verifies the central role of GEPE in shaping the geographical distribution of green innovation, thereby deepening the main argument in NEG regarding how local demand externalities influence industrial activities. NEG emphasizes that the cumulative causation driven by transport costs, increasing returns to scale, and market demand constitutes the fundamental force of spatial concentration. This study finds that GEPE, as an exogenous policy intervention and a credible policy signal, strengthens this cumulative mechanism by creating local market demand for green technologies and services and by fostering spatially interactive policy effects. This policy-induced demand reduces the uncertainty of returns for green R&D, serving as a powerful centripetal force that encourages the clustering of R&D activities and related green industries. Specifically, GEPE acts as a spatial sorting mechanism, attracting R&D personnel and capital from neighboring regions, which fuels localized increasing returns and reinforces the agglomeration tendency hypothesized by NEG. The results indicate that GEPE acts not only as a direct catalyst for local green innovation but also generates spatial spillover effects that form the micro-foundation for either interregional green synergistic development or competitive divergence. This provides a new perspective for understanding the role of environmental regulation in the transformation of spatial economic structures.

Second, by distinguishing between the flow of R&D personnel and R&D capital, this study further uncovers behavioral differences and underlying mechanisms among different types of R&D elements in influencing green innovation. Although the classical NEG framework offers a foundational analytical lens for studying spatial effects of element flow, its assumptions largely center on ordinary labor and nonspecific capital. Consequently, it has limited explanatory power for the mobility of highly skilled R&D personnel and specialized R&D capital in a knowledge-based economy. The empirical results reveal that R&D personnel flow significantly strengthens the positive impact of GEPE on green innovation. This mechanism aligns closely with the technological externalities emphasized by NEG, in which knowledge spillovers occur through face-to-face communication, tacit knowledge diffusion, and human capital externalities. These effects are further amplified in the policy-driven field of green technology innovation.

By comparison, the moderating effect of R&D capital flow is not robust, suggesting that its spatial allocation logic may be more complex. Although capital is inherently profit-seeking and mobile, its allocation efficiency in the green innovation sector may be constrained by factors such as policy stability, technological maturity, and uncertainty in long-term returns. These constraints reduce its responsiveness to GEPE across regions. This distinction enriches the understanding of heterogeneous behaviors of R&D elements.

Finally, this study incorporates the perspective of environmental decentralization to examine its moderating role in the relationship between GEPE and regional green innovation. The results confirm that while environmental decentralization directly promotes green innovation, it simultaneously weakens the policy effectiveness of GEPE. This seemingly paradoxical outcome highlights the institutional complexity that must be considered when applying it in the Chinese context. When local governments are granted environmental administrative authority without corresponding fiscal coordination and regional cooperation mechanisms, excessive decentralization may lead to strategic interactions. The result highlights the critical role of effective institutional arrangements in guiding the spatial economy toward green transformation.

Despite offering a systematic assessment of how GEPE impacts regional green innovation, this study nevertheless has some room for improvement. First, green technological innovation encompasses heterogeneous modes, trajectories, and technological paradigms. The present analysis adopts an integrated perspective and therefore does not disentangle the differentiated channels through which GEPE operates, though R&D element flow, to stimulate various forms of green innovation. As a result, the policy implications generated here remain somewhat general and may offer limited guidance for tailoring institutional instruments to facilitate targeted mobility of innovation factors. Second, the measurement of R&D element flow relies on stock-based proxy indicators. While consistent with prevailing empirical practice and adequate for characterizing the spatial dynamics of element allocation, this approach is inherently constrained in capturing micro-level processes such as cross-regional mobility of R&D personnel and real-time capital transfers, which may embed important behavioral nuances.

In response to these limitations, future research may explicitly differentiate between incremental and radical forms of green technological innovation, thereby uncovering potential heterogeneity in spatial drivers and response mechanisms. Such an extension could deepen understanding of how distinct technological attributes shape the spatial organization of innovation resources and the strategic behavior of innovation entities. Furthermore, should more granular, high-frequency data on R&D element flow become available, researchers would be positioned to provide more precise empirical evidence on the micro-foundations underpinning the mechanisms proposed in this study, thereby strengthening causal inference and theoretical generalizability.

6. Conclusions

In recent years, green technological innovation, as the most vital aspect of green innovation, has gained importance in regional policy-making in pursuit of achieving competitive advantage and sustainable development. This study focuses on the role GEPE plays in fostering regional green innovation, exploring how the flow of R&D elements moderates this relationship. Employing China’s provincial panel data from 2007 to 2018, this research provides a detailed analysis of these dynamics. The results demonstrate that GEPE is crucial in promoting regional green innovation, with this effect being significantly enhanced by R&D personnel flow. By contrast, R&D capital flow does not show a comparable impact. Additionally, the study uncovers spatial spillover effects, where increased GEPE not only stimulates green innovation within the region but also benefits neighboring areas. Furthermore, we also find that excessive environmental decentralization may inhibit the effectiveness of GEPE on green innovation, underlining the importance of balanced environmental governance.

The novelty of this study lies in the following aspects: on the one hand, adopting an innovative perspective centered around R&D element flow freely, this study meticulously assesses the combined impacts of GEPE and R&D element flow on regional green innovation. It makes a discerning distinction between the knowledge spillover effects engendered by R&D personnel flow and the capital supplementation effects resulting from R&D capital flow. This differentiation is crucial for policymakers, as it provides critical information for effectively improving the spatial efficiency of government environmental protection expenditure. On the other hand, this paper innovatively incorporates geographical economics and environmental decentralization frameworks into the research paradigm. By accounting for cross-border flows of both R&D personnel and R&D capital, it expands spatial spillover effects of GEPE on green innovation in neighboring regions and the moderating role of provincial government environmental decentralization.

Given the empirical findings tested in this study, we are able to draw several valuable inferences and policy implications. To begin with, the structure of fiscal expenditures should be optimized to enhance the policy effectiveness of government environmental protection spending. The strategic orientation of GEPE should be strengthened with a focus on supporting core segments and key entities involved in green innovation activities, particularly R&D projects that are talent-intensive. Fiscal instruments should guide green innovation at its early stages by increasing incentives for research institutions and enterprises through dedicated funds, differentiated subsidies, and tax incentives. Moreover, a performance-oriented fiscal allocation mechanism should be established to ensure that public spending generates sustainable benefits by stimulating regional green innovation and alleviating constraints on talent and capital resources.

Second, the cultivation and flow mechanisms of R&D personnel should be improved to build a robust regional talent ecosystem for green innovation. The results indicate that R&D personnel flow plays a more prominent role in strengthening the relationship between GEPE and green innovation. Therefore, local governments should prioritize policies that attract and facilitate the inflow of green innovation talent. To achieve this, income tax deductions and matching research funds may be provided to high-level researchers engaged in green technology development. Scholarships and research grants tied directly to green technological advancement can be introduced to strengthen the long-term talent pipeline. Relocation subsidies and living-cost support are also recommended to attract skilled green researchers across regions.

In addition, dedicated incubation funds may be created for green start-ups, conditional upon local employment commitments and cross-regional knowledge-sharing mechanisms. Collaboration among universities, research institutes, and enterprises can be promoted through the establishment of green technology research and transformation alliances, fostering a multi-stakeholder, multi-tiered system for talent cultivation and cooperation. Such measures contribute to enhancing both the agglomeration and siphon effects of R&D personnel and, ultimately, to strengthening the regional foundation for sustained green innovation.

Third, interregional cooperation in green innovation and environmental governance should be promoted to fully leverage the spatial spillover effects of GEPE. Local governments should deepen cross-regional policy coordination, dismantle local protectionist barriers, and establish joint mechanisms for environmental governance, technology sharing, and risk monitoring. Initiatives may include jointly developing regional green innovation demonstration zones, sharing green patent pools and technology commercialization platforms, and implementing coordinated performance evaluation systems for green innovation outcomes. These efforts can shift regional competition toward cooperative competition in green development, thereby enhancing both innovation output and environmental governance efficiency across regions.