Hand in Hand or One Left Behind? The Spillovers of Green Finance on Energy Transition

Abstract

Amid the growing challenges of energy security and carbon neutrality, green finance has become an important policy instrument for promoting sustainable energy transitions. Using city-level data from China between 2009 and 2019 and a spatial Durbin model, this study not only examines local effects but also focuses on identifying the geographic and administrative boundaries of their spatial spillovers. The analysis reveals that green finance advances energy transition via fostering green innovation and optimizing industrial structures, with stronger effects in regions where public environmental awareness is higher. Pronounced regional heterogeneity is identified, as resource-abundant and economically developed cities respond more strongly to green finance. Furthermore, green finance generates positive spatial spillovers, improving the energy transition of neighboring cities within a 400 km radius, although these effects are weakened by interprovincial competition. Overall, the findings highlight the importance of regionally differentiated and coordinated green finance policies in supporting effective energy transitions.

1. Introduction

While energy consumption remains a fundamental driver of economic development, excessive reliance on fossil fuels has been a major contributor to global climate change. This tension has given rise to the “energy trilemma”, which highlights the inherent challenges faced by energy systems in simultaneously ensuring energy security, maintaining economic competitiveness, and promoting environmental sustainability [1]. Against this backdrop, the global energy system is undergoing a profound structural transformation. As the world’s largest energy consumer, China has accelerated its transition toward a more sustainable energy system by gradually reducing its dependence on carbon-intensive fuels, particularly coal, in line with its “dual-carbon” objectives of peaking carbon emissions and achieving carbon neutrality. Nevertheless, replacing fossil fuels with renewable energy on a large scale remains a complex and challenging process [2]. Progress in renewable energy technologies is often constrained by relatively low efficiency, high levels of technological and market risk, and substantial upfront investment requirements [3]. These constraints underscore the critical role of the financial sector in facilitating capital allocation and mobilizing long-term investment to support the adoption and diffusion of renewable energy technologies [4].

To promote the green economy and facilitate the energy transition, the Chinese government has established a comprehensive green finance framework, encompassing mandatory environmental information disclosure requirements, the development of diversified green financial instruments, and the implementation of regionally differentiated pilot programs. These policy initiatives have yielded substantial progress. Among the various green financial instruments, green credit has become the core pillar of China’s green finance system, while green bonds, green funds, green insurance, and carbon emissions trading have jointly enhanced market depth and vitality. By the end of 2024, the outstanding balance of green credit in China reached CNY 36.6 trillion, representing a year-on-year increase of 21.7%. In terms of sectoral allocation, infrastructure green upgrading and the clean energy sector accounted for the largest shares of green lending, at 42.84% and 27.02%, respectively [5]. In the same year, 477 green bonds were newly issued, with a total issuance volume of approximately CNY 681.43 billion. Notably, green bonds supporting the clean energy sector accounted for the largest number of issuances and the second-largest share of total issuance volume, primarily financing wind and photovoltaic power projects [6].

Through market-oriented mechanisms, green finance channels capital toward renewable energy production and research and development activities, thereby improving resource allocation efficiency, stimulating technological innovation, alleviating financing constraints and project risks, and fostering regional energy cooperation [7]. Collectively, these functions play a critical role in accelerating the energy transition and advancing sustainable development objectives. However, much of the existing literature evaluates these effects primarily from a local perspective [8,9]. This approach overlooks the potential for green finance initiatives in one region to generate spillover effects beyond the local context through economic linkages, factor mobility, and policy interactions [10].

Evaluating whether the expansion of green finance effectively supports the energy transition, as well as elucidating the underlying transmission mechanisms, is essential for achieving long-term climate goals and designing appropriate transition pathways. However, limited attention has been paid to the spatial spillover effects of green finance, namely whether and how green finance development in one city influences the energy transition performance of neighboring cities. This study extends the analysis beyond local impacts by examining whether spatial spillovers vary across geographical distances and administrative jurisdictions. Spatial analysis provides an effective framework for identifying regional heterogeneity in the distribution of benefits and costs associated with green finance. While certain regions may experience amplified gains through spillover effects, others may remain underserved or marginalized [11]. Moreover, green finance policies implemented in one locality may generate unintended externalities in neighboring regions, such as resource reallocation or competitive distortions, which necessitate careful policy coordination. Therefore, adopting an urban and regional perspective is crucial for capturing spatial disparities in the effects of green finance and for understanding how localized development trajectories shape energy transition outcomes across interconnected cities [12].

This study offers three main contributions. First, whereas most existing studies on green finance and the energy transition are conducted at the national or provincial level, this study adopts an urban-scale perspective. City-level analysis captures substantial intra-provincial heterogeneity in resource endowments and policy implementation, identifies finer spatial interaction scales, and explicitly examines inter-city competition and cooperation mechanisms that is often masked in more aggregated analyses. This localized approach offers a stronger empirical basis for formulating targeted green finance policies to support urban energy transition initiatives. Second, this study examines the spatial spillover effects and spatial spillover boundaries of green finance on energy transition in neighboring cities by constructing a segmented spatial weight matrix and employing the spatial Durbin model (SDM). Furthermore, it differentiates spillover effects between intra-provincial and inter-provincial city pairs. This spatially explicit analytical framework provides policy-relevant insights for local governments seeking to design coordinated and cross-regional green finance strategies. Third, in addition to evaluating the direct effects of green financial instruments, such as green credits and green bonds, this study investigates whether green finance pilot policies generate demonstrative or diffusion effects on surrounding regions.

The paper is structured to move from theory to empirical evidence and policy discussion. After reviewing the relevant literature and developing the hypotheses, the empirical strategy is introduced and the main results are then presented. This is followed by an analysis of the role of green finance policies in advancing energy transition, before concluding with policy implications.

2. Literature Review and Hypotheses Development

2.1. Literature Review

It is widely recognized that the energy transition toward low-carbon development serves as a critical pathway for achieving environmental preservation and high-quality economic growth [13,14]. The energy transition requires the gradual phasing out of traditional non-renewable energy sources, such as coal, oil, and natural gas, while promoting the adoption of renewable energy sources, including wind, solar, biomass, hydroelectric, and geothermal energy [15]. Numerous studies have explored the external factors shaping energy transition, such as human capital, economic growth, public environmental awareness, government actions, and technological progress [16,17,18,19,20]. Babic (2024) provided a theoretical perspective, arguing that green finance lies at the core of both the challenges and opportunities associated with the global energy transition [21].

Existing literature has investigated the direct effects of green finance on the energy transition, generally finding that green finance increases the share of renewable energy and improves energy efficiency. Empirical evidence from E7 economies suggests that green finance facilitates renewable energy adoption through improved capital allocation and technological innovation [8]. Related studies further emphasize that green finance enhances energy efficiency by promoting green technological progress and optimizing energy structures [22], with green bonds playing particularly important roles in supporting clean energy development in both the short and long run [23]. In the context of China, existing research has primarily focused on the provincial level. These studies consistently show that green finance contributes to reducing energy intensity [24] and advancing clean energy transitions [9], while the magnitude of these effects varies with regional characteristics such as urbanization [24], market development [9], and climate risk exposure [25]. Wang et al. (2024) utilized provincial data to identify key mechanisms through which green finance facilitates low-carbon energy transitions, including technological advancements in energy substitution, expansion of clean energy supply capacity, increased public environmental awareness, and the optimization of economic structures [26]. Overall, this strand of literature provides robust evidence on the local impacts and transmission mechanisms of green finance but remains largely confined to aggregated spatial units.

Beyond local effects, the spatial spillover effects of green finance on the energy transition in China remain relatively underexplored, especially at the sub-provincial level. Existing studies using SDM at the provincial level demonstrate that green finance facilitates the development of renewable energy and generates significant spillover effects on neighboring regions, in some cases exceeding local effects [27]. Wan et al. (2023) developed a low-carbon energy consumption structure index and showed that green finance supports the decarbonization of local energy systems while also generating positive spillovers to neighboring regions [28]. The study further identified green technological innovation and the strength of environmental regulation as key transmission channels driving these spatial effects [28]. Zhang et al. (2024) employed an SDM to reveal that green finance agglomeration exerts both direct and spatially mediated effects on regional carbon emission performance [29]. They further show that capital formation, energy transition, and technological advancement are key pathways through which green finance agglomeration improves carbon emission outcomes, emphasizing the importance of interprovincial coordination in fostering regional green finance clusters [29].

In addition, a smaller strand of research has examined the policy effects of green finance initiatives using quasi-natural experimental designs. Evidence from green finance pilot programs indicates that green finance promotes energy efficiency, with effects reinforced by stronger environmental regulation, higher financial efficiency, and favorable resource endowments [30], and enhances the use of renewable energy through alleviating financing constraints and fostering corporate digitalization [31]. Chen et al. (2025) studied the impact of green finance policies on the energy transition at the level of prefecture cities, showing that green finance policies exert a cross-regional influence in facilitating the energy transition [32].

In summary, some research gaps can be identified. First, existing studies on the impact of green finance on the energy transition predominantly rely on national- or provincial-level data, while analyses at the city level remain limited. Second, the spatial influence of green finance on the energy transition in China has not been thoroughly examined. Although some scholars have studied the existence of spatial spillover effects, research remains scarce on the spatial boundaries (i.e., the geographical reach of these effects) and spatial scope (i.e., the extent to which provincial boundaries affect the spillover effects). Addressing these gaps is essential for promoting cross-regional cooperation and enhancing the spatial targeting of green finance policies. Third, the effects of green finance pilot programs on energy transition in neighboring regions have received relatively little scholarly attention. Last, existing studies often depict the energy transition using a single indicator, such as the share of renewable energy or improvements in energy efficiency, overlooking the complexity of the transition process and its actual outcomes.

2.2. Hypotheses Development

Green finance constructs market-oriented solutions for energy transition [24,26]. On one hand, green finance contributes to energy efficiency improvements by bridging the funding gap in green innovation with long terms and high risks, which traditional finance fails to address [33]. On the other hand, green finance provides comprehensive financial support for green industry through differentiated financing mechanisms while restricting funding for high-pollution industries, thereby promoting a cleaner energy structure [34].

On the energy production side, capital injections drive technological breakthroughs in the energy sector, improving the efficiency of energy conversion and utilization [35]. Additionally, carbon trading markets establish a price-signal transmission mechanism that transforms the positive externalities of renewable energy projects into measurable economic benefits [36]. This mechanism attracts investors to expand renewable energy investment, accelerating the deployment of renewable energy. On the consumption side, financing constraints compel energy-intensive enterprises to restructure their production models in order to increase the economic output per unit of energy consumed, ultimately improving energy use efficiency [37]. Green finance encourages enterprises to internalize environmental costs such as carbon emissions into their strategic and operational decisions. This shift enables a transition from passive compliance with emission regulations to proactive transformation aligned with long-term sustainable development goals [38].

Moreover, in regions abundant in traditional mineral resources, the dynamics of the energy transition can be complex. On one hand, the plentiful supply of traditional energy may weaken the urgency or incentive to pursue energy transition initiatives [39]. On the other hand, the severe environmental degradation associated with resource extraction may strengthen policy emphasis on green finance, which benefits energy transition [22]. In economically developed regions, a stable financial market may enable the effectiveness of green finance instruments in promoting energy transition [24].

In conclusion, green finance facilitates the energy transition by enhancing energy efficiency, promoting the production and utilization of renewable energy, and supporting sustainable industrial transformation. However, the magnitude of its impact could exhibit regional variation. Based on these considerations, this study proposes the following hypotheses:

Hypothesis 1. 

Advancement of green finance can help promote energy transition.

We argue that green finance indirectly prompts energy transition through green innovation and industrial structure upgrading. First, technical challenges are increasingly recognized as a bottleneck in the large-scale use of renewable energy [3]. For instance, solar power generation suffers from low conversion efficiency and insufficient storage capacity, which necessitate extensive collection infrastructure and result in high utilization costs [40]. Green finance directs capital into the renewable energy sector, increasing resources devoted to the research and development of green and low-carbon technologies [41]. Moreover, green finance policies encourage enterprises to adopt green electricity [42], expanding market demand for low-carbon energy technologies. From a supply-and-demand perspective, this dynamic fosters green innovation, which in turn drives the energy transition [43].

Second, by restraining the growth of energy-intensive sectors and supporting green industries, green finance promotes industrial upgrading through more efficient capital allocation [44]. Wang and Wang (2021) proved that China’s green finance has the most significant impact on the tertiary sector, effectively promoting industrial structure upgrading [45]. Such industrial upgrading contributes to the energy transition from the demand side by shifting consumption patterns toward more sustainable and less energy-intensive sectors [46]. Therefore, we propose the following hypotheses:

Hypothesis 2. 

Green finance can help prompt energy transition through green innovation and industrial structure upgrading.

Advances in green finance within a region can stimulate energy transition in adjacent areas through spatial spillover effects. From a spatial perspective, technology diffusion effects play a vital role in this process [47]. Advanced energy technologies are disseminated to neighboring regions through technical cooperation and knowledge sharing, thereby enhancing the efficiency of renewable energy development and utilization [48]. Therefore, green finance could promote the regional spread of energy transition through technological innovation. Furthermore, green finance underpins the development of cross-regional energy infrastructure, such as power grids and natural gas pipelines, which strengthen interregional connectivity. The reliability and efficiency of such infrastructure are essential for optimizing resource allocation, as the robustness of energy transportation systems directly influences clean energy distribution [49]. The impact of green finance on energy transition is not confined to the local context. In China, provinces function as primary administrative units within which cities tend to exhibit closer connections in terms of policy transmission, industrial coordination, and resource sharing [50]. In contrast, cities located across provincial boundaries may experience weaker spillover effects due to the administrative autonomy of provincial jurisdictions.

Moreover, drawing on insights from new economic geography, the spatial diffusion of green finance is unlikely to follow a simple distance-decay logic. The “gradient effect” of technology diffusion suggests that spillovers are both distance- and hierarchy-dependent, with diffusion intensity declining as innovations spread from core to peripheral regions [51]. This attenuation effect may stem from several factors, including declining relevance of shared infrastructure, weakening policy coordination, and diminishing institutional compatibility as geographical distance increases. However, core regions with strong financial and technological agglomeration may dominate the allocation of innovative resources, generating siphoning effects that create “agglomeration shadow” in surrounding areas [52]. This process can induce one-way flows of production factors and lock-in effects, thereby reducing the accessibility of green financial resources for peripheral cities, particularly those with weaker financial infrastructure and limited absorptive capacity [53].

In summary, the spatial spillovers of green finance are unlikely to decline monotonically with distance. Instead, spillovers may exhibit a geographically bounded pattern in which moderate spatial separation allows peripheral cities to benefit from diffusion without being fully overshadowed by core-region dominance, while excessive distance weakens transmission. Therefore, we propose the following hypotheses:

Hypothesis 3a. 

Green finance not only influences energy transition locally but also through spatial spillovers.

Hypothesis 3b. 

The spatial spillover effects of green finance on the energy transition vary with geographical distance, increasing initially and then diminishing.

Based on the above hypotheses, a comprehensive theoretical framework is presented in Figure 1.

3. Methodology and Data

3.1. Model Construction

3.1.1. Benchmark Regression Model

To examine the relationship between green finance and the energy transition, we employ a fixed-effects framework. The empirical specification is given as follows:

$${{ETI}_{it}={\alpha }_0+{\alpha }_1{GF}_{it}+{\alpha }_k{Controls}_{it}+{\gamma }_t+{\mu }_i+\epsilon }_{i,t}$$

(1)

where ${ETI}_{it}$ denotes the energy transition index for province i in year t. The key explanatory variable, $GF$, represents the green finance development level. Controls are the control variables, including Lnpeople, Gn, Income, and Info. Additionally, ${\mu }_i$ denotes city fixed effects, accounting for inherent characteristics of each city that remain unchanged over time; ${\gamma }_t$ represents time fixed effects, reflecting time-specific factors impacting the outcomes; and ${\epsilon }_{i,t}$ denotes the random error term. The primary coefficient of interest is ${\alpha }_1$. A significantly positive ${\alpha }_1$ indicates that green finance positively influences energy transition, thereby supporting Hypothesis 1.

3.1.2. Mechanism Analysis Model

Green finance is theoretically expected to contribute to energy transition by enhancing green innovation and driving industrial upgrading. Therefore, the following mediation models were established:

$${M_{it}={\phi }_0+{\phi }_1{GF}_{it}+{\phi }_k{Controls}_{i,t}+{\gamma }_t+{\mu }_i+\epsilon }_{i,t}$$

(2)

$${{ETI}_{it}={\theta }_0+{\theta }_1{GF}_{it}+{\theta }_2{M}_{it}+{\theta }_k{Controls}_{i,t}+{\gamma }_t+{\mu }_i+\epsilon }_{i,t}$$

(3)

where $M$ denotes the mediating variable. If ${\phi }_1$ and ${\theta }_2$ are significant, an intermediary effect exists.

3.1.3. Moderating Models

To examine the moderating effect of public environmental awareness, an interaction term was introduced into the baseline model, resulting in the following specification:

$$\begin{array}{c}{{ETI}_{it}={\beta }_0+{\beta }_1{GF}_{it}+{\beta }_2{EA}_{it}+{\beta }_3{GF}_{it}\times {EA}_{it}+{\beta }_k{Controls}_{i,t}+{\gamma }_t+{\mu }_i+\epsilon }_{i,t}\end{array}$$

(4)

where $EA$ represents environmental awareness. A statistically significant coefficient for the interaction term $GF\times EA$ would indicate the presence of a moderating effect from public environmental awareness. The remaining variables in the model are kept the same as in the baseline specification.

3.1.4. Spatial Durbin Model

Energy transition may generate spatial spillover effects that conventional econometric methods cannot fully capture. Accordingly, we utilize the spatial Durbin model (SDM) to investigate how spatial correlation affects energy transition. The model is specified as follows:

$$\begin{array}{c}{ETI}_{it}=\rho \sum w_{ij}\times {ETI}_{jt}+{{\beta }_1{GF}_{i,t}+\beta }_k{Controls}_{it}\\ +{{\lambda }_1{\sum w_{ij}\times GF}_{i,t}+\lambda }_k\sum w_{ij}\times {Controls}_{it}+{\gamma }_t+{\mu }_i+{\epsilon }_{it}\end{array}$$

(5)

where $\rho$ denotes the spatial autocorrelation coefficient and $w_{ij}$ represents the elements of the spatial weight matrix $W$. We employ the inverse distance weight matrix to measure the spatial weight matrix, which is the reciprocal of the distance between cities, thereby assigning higher weights to cities that are geographically closer.

3.2. Variables and Data

3.2.1. Variables

The existing studies predominantly measure energy transition through the share of renewable energy consumption in total energy usage and improvements in energy efficiency [25,54,55]. However, this unidimensional approach overlooks the intrinsic complexity of energy transition. This study employs the energy transition index (ETI) proposed by Shen et al. (2023) [56]. This dataset adapts the World Economic Forum’s (2022) comprehensive ETI measure [57] to a city-level framework, capturing the energy transition data for 282 Chinese cities from 2003 to 2019. The index is divided into two components: energy system performance (ESP) and transition readiness (TR). ESP reflects how well the existing energy system is developed and its level of environmental sustainability. TR captures the factors that support continued advancement, such as economic development, investment levels, technological capacity, and human capital. More detailed indicators are shown in

Table A1. Energy transition index from Shen et al. (2023) [82].

Sub Index	Dimensions	Components	Indicators (Unit)	Lower Bound	Upper Bound	Data Sources
Energy system performance	Energy system structure	Energy mix	Share of coal in primary energy (%)	24.3%	95.2%	CEADs
		Electricity structure	Local coal consumption for power generation vs. total electricity consumption (kg standard coal equivalent/kWh)	0	2.09	CCSY, CEADs
		Energy intensity	Energy consumption per unit of GDP (standard coal equivalent/104 yuan)	0.29	9.7	CCSY, CEADs
		Energy consumption	Energy consumption per capita (standard coal equivalent per capita)	0.41	13.6	CCSY, CEADs
			Electricity consumption per capita (kWh per capita)	488	16,969	CCSY, CEADs
	Environmental sustainability	Carbon intensity	CO2 emissions per unit of GDP (t/104 yuan)	0.55	5.76	CCSY, CEADs
		Carbon emissions per capita	CO2 emissions per capita within urban territory (t/per capita)	0.99	31.3	CCSY, CEADs
		Air pollution (PM2.5)	Annual average concentration of inhalable fine particulate matter (micrograms per cubic meter)	12.3	66.5	ACAG
Transition readiness	Economic development	Economic growth	Per capita GDP (yuan)	4597	151,326	CCSY
			GDP growth rate (%)	3.5%	28.6%	CCSY
		Economic structure	Proportion of employment of mining employees in urban units at the end of the year (per 104 capita)	0	533	CCSY
			Tertiary industry as percentage to GDP (%)	21.6%	57.6%	CCSY
	Capital and investment	Capital stock	Average annual balance of net fixed assets per capita (yuan per capita)	1048.5	84,766.4	CCSY
			Proportion of urban construction land in the municipal area (%)	0.4%	34.1%	CCSY
			Per capita deposits of national banking system at the end of the year (yuan per capita)	4168.9	210,182.4	CCSY
		Investment	Total investment in fixed assets per capita (yuan)	1624.7	80,971	CCSY
			Amount of foreign capital per capita (US dollars per capita)	0.3	1007.8	CCSY
		Fiscal capacity	Public finance income per capita (yuan per capita)	167.9	14,925.3	CCSY
	Technology capability	Innovation capability	China innovation and entrepreneurship index (0–100)	4.1	97.6	PKU-ORDP
			Proportion of subscribers of internet services (%)	1%	51%	CCSY
			Number of green invention and utility model patents applied per capita in the year (number per 104 capita)	0	2.04	CNRDS
		Technology expenditure	Expenditure on science and technology per capita (yuan per capita)	0.96	651.9	CCSY
		Adaptive technology	Ratio of industrial SO2 removed (%)	0.7%	87.4%	CCSY
			Ratio of wastewater centralized treated (%)	0.0%	97.9%	CCSY
			Ratio of consumption wastes treated (%)	0.0%	100.0%	CCSY
			Ratio of industrial solid wastes treated (%)	18.0%	100.0%	CCSY
	Human capital	R&D and new economy	Proportion of people engaged in scientific research, technical services, and geological exploration industries (per 104 capita)	2.2	98.2	CCSY
			Proportion of people employed in the information transmission, computer services, and software industries (per 104 capita)	2.6	67.3	CCSY
		Educational and training capacity	Proportion of employees in the education industry (per 104 capita)	72.5	213.6	CCSY
			Number of full-time teachers in vocational secondary schools (per 104 capita)	1.1	17.2	CCSY
			Number of full-time teachers in regular institutions of higher education (per 104 capita)	0.4	49.7	CCSY
		Quality of education	Expenditure on education per capita (yuan per capita)	118.1	2831.3	CCSY
			Number of students enrolled in regular institutions of higher education (per 104 capita)	5.9	886	CCSY

. Many scholars have already utilized this dataset in related research [16,32,58].

In measuring green finance, researchers often take green credit as an indicator of overall development, while the proportion of interest payments by six energy-intensive sectors relative to total industrial interest serves as an inverse metric [59]. Some scholars construct composite indicator systems by using proxy variables for financial instruments. Lee and Lee (2022) developed a provincial-level green finance development index incorporating green credit, green securities, green insurance, and green investment, where agricultural insurance serves as a proxy for green insurance and government environmental protection public expenditure represents green investment [60]. This study focuses on two types of financial instruments, green credit and green bond, as they represent the largest segments of the green finance market and serve as the primary channels for financing environmentally sustainable projects [61]. Following Wang et al. (2023), we use the total credit issued to green listed companies within a city as a proxy variable of green credit [62]. We use the alignment of a company’s main business with the Catalogue of Industries for Green and Low-Carbon Transition (2024 edition) as the criterion for determining whether it qualifies as a green company. To alleviate the bias introduced by missing data from companies listed later, we select 563 valid corporate samples from companies listed on the A-share market after 2015. For green bonds, the total issuance volume by firms headquartered in each city is aggregated to derive the city-level scale of green bond activity. Finally, we sum green credit and green bonds and then apply a natural logarithmic transformation to construct green finance development level (GF). This transformation helps to smooth the distribution, mitigate potential heteroskedasticity, and reduce the influence of extreme values.

Given that multiple other factors can also exert an influence on energy transition, we selected the following control variables based on prior research. (1) Population number (Lnpeople) is represented by the natural logarithm of the year-end registered population. (2) Income level (Income) is assessed through the logarithm of the average earnings of employed person. (3) Government intervention (Gn) is captured by the ratio of government expenditure to GDP [63]. (4) Informatization level (Info) is determined by the number of internet access users.

Drawing on prior studies and our hypotheses, the following mediating variables were chosen: (1) green innovation (GI), represented by the logarithm of granted green invention patents; and (2) industrial structure adjustment (Ind), defined as the contribution of the tertiary industry to overall economic output.

We incorporated environmental awareness (EA) as a moderating variable. Owing to the difficulty of directly observing awareness at the city level, following Giménez-Nadal et al., EA is operationalized as the ratio of bus passenger volume to the resident population [64]. Higher public transport usage reflects stronger pro-environmental preferences and greater acceptance of low-carbon lifestyles [65].

3.2.2. Data

This study employed panel data covering 97 Chinese cities over the period 2009–2019. Following the study of Shen et al. (2023), energy transition index data was sourced from the International Society for Energy Transition Studies (ISETS: https://isets.org/data/ (accessed on 10 November 2024)) [56]. Following the study of Wang et al. (2023), Green finance development level data were derived from CCER database (http://www.ccerdata.cn/ (accessed on 15 April 2024)) [62]. Additional data were drawn from the China Statistical Yearbook, China Statistical Yearbook on Environment, China Industry Statistical Yearbook, and EPS database (https://www.epsnet.com.cn/ (accessed on 20 April 2024)). Interpolation and trend extrapolation methods were used for supplementation in instances where data were missing. Descriptive statistics for the main variables are reported in

Table 1. Descriptive Analysis.

	(1)	(2)	(3)	(4)	(5)
	N	Mean	Stdrd. Devi.	Min	Max
ETI	1067	52.791	13.105	18.874	82.724
ESP	1067	54.246	15.241	10.684	91.140
TR	1067	51.337	17.333	21.824	90.484
GF	1067	21.899	2.545	0	28.438
Lnpeople	1067	6.048	0.744	2.970	8.136
Income	1067	10.873	0.374	9.870	12.062
Gn	1067	0.146	0.095	0.010	2.702
Info	1067	4.552	1.036	1.401	8.551
GI	1067	4.933	1.820	0	9.789
Ind	1067	0.443	0.108	0.168	0.835
EA	1067	86.889	107.479	1.260	935.650

. The standard deviation of the ETI is smaller than that of the two sub-dimensions, ESP and TR, suggesting that the ETI exhibits a lower degree of dispersion and provides a more stable reflection of the overall energy transition status. In the early stages, certain cities lacked green enterprises and companies issuing green bonds, resulting in a minimum value of zero for GF.

4. Empirical Results

4.1. Baseline Results

Table 2. Baseline regression results.

	(1)	(2)
	ETI	ETI
GF	0.131 ***	0.138 ***
	(2.60)	(2.78)
LnPeople		−3.513 ***
		(−2.81)
Income		6.536 ***
		(4.91)
Gn		−1.236
		(−1.26)
Info		0.956 ***
		(3.47)
_cons	42.517 ***	−7.579
	(40.36)	(−0.47)
Fixed effect	Yes	Yes
N	1067	1067
${\mathrm{R}}^2$	0.777	0.788

Note: t statistics in parentheses. *** p < 0.01.

reports the estimated results of the baseline model. Columns (1) and (2) show that the coefficients for green finance development level remain significantly positive, regardless of whether control variables are incorporated. This aligns with our expectations that green finance promotes energy transition and provides empirical support for Hypothesis 1.

Regarding the control variables, population, income, and informatization level all have significant impacts on energy transition. Population level exerts a negative influence on energy transition, aligning with the research results of Bu et al. (2022) [66]. An increase in population size intensifies energy consumption pressure, as rising demands for residential and public service electricity create additional challenges for energy transition. In contrast, higher income positively contributes to energy transition, echoing the evidence reported by Ma et al. (2019) [67]. Increased income levels elevate consumption standards, making residents attach greater importance to air quality and ecological services, and become more willing to pay a premium for green products, thereby fostering a low-carbon consumption environment. The informatization level also plays a crucial role in energy transition. A higher informatization level promotes the widespread adoption of smart energy management systems. Through internet-connected devices such as smart meters, real-time energy usage data can be collected and analyzed to optimize energy allocation [68].

The relationship between green finance and energy transition may be endogenous, as mutual feedback effects, unobserved influences, and measurement imperfections could interfere with consistent parameter estimation [69]. To address this concern, we adopt the system generalized method of moments (sys-GMM) by Arellano and Bover (1995) [70], which mitigates endogeneity by employing lagged explanatory variables as instruments. As reported in Column (1) of

Table 3. Endogenous treatment and robustness tests.

	(1)	(2)	(3)
	Sys-GMM	Replace Independent Variable	Drop Municipalities
L.ETI	0.912 ***
	(4.28)
GF	4.202 *	0.110 **	0.133 **
	(1.83)	(2.12)	(2.52)
Control variables	Yes	Yes	Yes
_cons	41.770	−8.418	−14.742
	(0.32)	(−0.52)	(−0.90)
N	970	1067	1034
AR(1)	0.027
AR(2)	0.698
Sargan	0.650
${\mathrm{R}}^2$		0.787	0.784

Note: t statistics in parentheses. * p < 0.1, ** p < 0.05, *** p < 0.01.

, the diagnostic tests indicate that while first-order serial correlation is present in the differenced residuals, no evidence of second-order serial correlation is detected, as reflected by the significant AR(1) statistic and insignificant AR(2) statistic. Moreover, the Sargan test reports a p-value of 0.650, suggesting that the instruments satisfy the exogeneity condition and that the model is free from over-identification concerns. These findings support the validity of the system GMM framework.

To further ensure the robustness of the empirical model estimation results, we tested the following two aspects. First, we adopted green credit as a proxy variable for green finance. Since green bonds were only introduced in 2016, the lack of data for earlier years could potentially bias the regression results. Therefore, we excluded the green bond indicator and re-ran the regression. Second, the sample was changed. We excluded observations from the four municipalities directly under the Central Government (Beijing, Tianjin, Shanghai, and Chongqing) because of the large development differences between these municipalities and other cities. The robustness tests are detailed in Columns (2) and (3) of Table 3. Across all specifications, the results consistently show that green finance facilitates energy transition, providing additional support for the robustness of the baseline model.

4.2. Mechanism Analysis

To identify the transmission channels, this study focuses on two mechanisms: green innovation and industrial structure upgrading. Empirical evidence from

Table 4. Mechanism analysis model results.

	Green Innovation		Industrial Structure Upgrading
	(1)	(2)	(3)	(4)
	GI	ETI	Ind	ETI
GF	0.019 ***	0.118 **	0.001 **	0.128 **
	(2.66)	(2.40)	(2.13)	(2.58)
GI		1.018 ***
		(4.66)
Ind				9.159 ***
				(3.00)
Control variables	Yes	Yes	Yes	Yes
_cons	−2.059	−5.483	122.840 ***	−18.830
	(−0.87)	(−0.34)	(7.20)	(−1.14)
Fixed effect	Yes	Yes	Yes	Yes
N	1067	1067	1067	1067
${\mathrm{R}}^2$	0.813	0.792	0.754	0.790

Note: t statistics in parentheses. ** p < 0.05, *** p < 0.01.

suggests that the positive effect of green finance on energy transition operates partly through its capacity to foster green innovation. Specifically, green finance significantly increases green innovation, which subsequently translates into measurable improvements in energy transition performance. The Sobel–Goodman test reported in

Table 5. Sobel–Goodman mediation tests.

	Green Innovation			Industrial Structure Upgrading
	Coefficient	SE	P	Coefficient	SE	P
Indirect effect	0.345	0.058	0.000	0.320	0.056	0.000
Direct effect	0.075	0.108	0.483	0.101	0.109	0.352
Total effect	0.421	0.119	0.000	0.421	0.119	0.000

further confirms that green innovation exerts a full mediation effect on the energy transition. Given the long investment horizons, high capital intensity, and elevated risks associated with renewable energy technologies, traditional financial institutions often demonstrate risk aversion, resulting in insufficient funding for such projects [33]. Green finance effectively bridges this gap and promotes the rapid advancement of technological innovation. The enhancement facilitates greater use of renewable energy, reduces fossil fuel dependency, and aids the transition to a low-carbon energy system.

As reported in Column (3), green finance exerts a significant effect on industrial structure optimization, in line with the evidence documented by Wang and Wang (2021) [45]. Column (4) indicates that upgrading the industrial structure significantly facilitates energy transition. By promoting industrial restructuring, green finance supports the shift toward a more sustainable energy consumption structure. The Sobel–Goodman test further confirms that industrial structure upgrading exhibits a full mediation effect. These results collectively support Hypothesis 2. On the one hand, green finance aids in the industrial upgrading of energy-intensive enterprises. On the other hand, it effectively provides diversified financial support to emerging green industries. This reallocation reduces the economy’s reliance on energy-intensive, high-emission sectors and enhances the overall industrial structure. As the economy shifts toward more sustainable industries, energy consumption patterns become cleaner and more efficient, thereby promoting energy transition.

4.3. Moderating Effects of Public Environmental Awareness

Table 6. Moderating model results.

	(1)	(2)
	ETI	ETI
GF	0.147 ***	0.082
	(2.96)	(1.44)
EA	0.015 ***	−0.020
	(4.75)	(−1.25)
$\mathrm{G}\mathrm{F}\times \mathrm{E}\mathrm{A}$		0.002 ***
		(3.97)
Control variables	Yes	Yes
_cons	43.584 ***	48.293 ***
	(5.10)	(5.49)
Fixed effect	Yes	Yes
N	1067	1067
${\mathrm{R}}^2$	0.787	0.789

Note: t statistics in parentheses. *** p < 0.01.

reports the estimation results of the model with the moderating effect of public environmental awareness. Column (1) shows that higher public environmental awareness significantly accelerates the energy transition, consistent with the findings of Dabbous et al. (2025) [71], and underscoring the importance of bottom-up social forces in driving this process. Column (2) indicates that the interaction term between green finance development level and environmental awareness is significantly positive, suggesting that stronger environmental awareness amplifies the effect of green finance on the energy transition. Greater public awareness encourages environmentally responsible financial behavior, such as the purchase of green wealth management products, which expands low-cost green funding sources and increases the overall supply of green finance. At the same time, heightened public preference for green projects reduces financial institutions’ risk expectations regarding green assets, enabling them to offer more favorable financing conditions for clean energy and energy-efficiency technologies. Together, environmental awareness amplifies the effect of green finance on energy transition.

4.4. Heterogeneity Analysis

The above findings have confirmed both the direct and indirect impacts of green finance on energy transition. To further investigate whether these effects exhibit heterogeneity under varying economic development levels, we assess whether a higher green finance development level corresponds to a greater impact on energy transition. To examine this issue, a threshold regression framework is applied. Following the methodology of Hansen (1999) [72], threshold values were estimated through 300 iterations of the bootstrapping method. Evidence from

Table 7. Threshold regression model.

Threshold Test		(1)	(2)	Threshold Model Results	(3)
		Single-Threshold	Double-Threshold		ETI
Threshold value		23.859	22.261	GF ≤ 23.859	0.081
F value		35.860	19.090		(1.61)
p value		0.017	0.107	GF > 23.859	0.164 ***
Critical value	10%	24.035	19.819		(3.29)
	5%	28.609	24.643	Control variables	Yes
	1%	41.977	32.679	_cons	43.396 ***
					(41.41)
				Fixed effect	Yes
				N	1067
				${\mathrm{R}}^2$	0.785

Note: t statistics in parentheses. *** p < 0.01.

suggests that the relationship between green finance and energy transition is characterized by a single-threshold effect, whereas a second threshold is not statistically supported. Notably, green finance begins to significantly promote energy transition only once its development level rises above the threshold of 23.859. This finding implies that regions with more developed green finance systems achieve stronger advances in energy transition, which is consistent with Alharbi et al. (2023) [4].

We then examine resource heterogeneity from the perspective of resource endowments. The National Plan for Sustainable Development of Resource-based Cities (2013–2020) identifies as resource-based those cities where the local economy is largely driven by the exploitation and processing of natural resources, including forests and mineral deposits. Based on the inclusion of sample cities in the list, separate regression analyses were conducted for resource-based and non-resource-based cities. As shown in Columns (1) and (2) of

Table 8. Energy transition heterogeneity.

	(1)	(2)	(3)	(4)	(5)	(6)
	Resource-Based	Non-Resource-Based	High-ESP	Low-ESP	High-TR	Low-TR
GF	0.507 ***	0.056	0.120 *	0.171 **	0.350 ***	0.080
	(5.11)	(1.01)	(1.96)	(2.12)	(3.38)	(1.50)
Control variables	Yes	Yes	Yes	Yes	Yes	Yes
_cons	−28.893	1.491	−22.325	−9.348	−59.712 **	−37.242
	(−1.07)	(0.08)	(−0.93)	(−0.41)	(−2.27)	(−1.56)
Fixed effect	Yes	Yes	Yes	Yes	Yes	Yes
N	253	814	605	462	473	594
${\mathrm{R}}^2$	0.846	0.785	0.804	0.787	0.827	0.789

Note: t statistics in parentheses. * p < 0.1, ** p < 0.05, *** p < 0.01.

, green finance only benefits energy transition in resource-based regions. This pattern likely reflects the strong dependence of resource-based cities on extraction and processing activities, which are typically energy-intensive and highly polluting. Green finance policies in China tend to exhibit compensatory characteristics, allocating preferential financial support to regions facing greater environmental burdens and transition challenges, thereby enhancing the marginal effectiveness of green finance in promoting clean energy adoption and industrial upgrading. Furthermore, areas with significant resource endowments may possess inherent advantages in renewable energy potential and industrial foundations, making the impact of green finance on their energy transition particularly pronounced. And higher levels of green finance can help these regions leapfrog intermediate, carbon-intensive stages, further accelerating the shift away from traditional fossil-based systems [39].

To further investigate the differential effects across various regional contexts, the sample was divided based on the mean values of the two sub-dimensions of energy transition index. Columns (3) and (4) of Table 8 indicate that the responsiveness of energy transition to green finance diminishes in regions with more advanced energy system performance (High-ESP). This may be because the energy systems in such regions have limited room for further adjustment and optimization, which restricts the role of green finance. Columns (5) and (6) indicate that green finance significantly impacts energy transition only in regions with better transition readiness (High-TR). This outcome may stem from the more supportive economic and social conditions for renewable energy development in regions with higher transition readiness.

4.5. Spatial Spillover Effects

The analysis started with computing Moran’s I to evaluate the spatial dependence in energy transition. As shown in

Table 9. Moran index results.

Year	I Value	Z Value	Year	I Value	Z Value
2009	0.0644 ***	3.743	2015	0.0668 ***	3.857
2010	0.0827 ***	4.661	2016	0.0646 ***	3.743
2011	0.0709 ***	4.069	2017	0.0610 ***	3.566
2012	0.0687 ***	3.955	2018	0.0589 ***	3.459
2013	0.0734 ***	4.187	2019	0.0627 ***	3.651
2014	0.0657 ***	3.801

Note: t statistics in parentheses. *** p < 0.01.

, Moran’s I values were consistently positive and statistically significant from 2009 to 2019, supporting the use of a spatial econometric approach. To determine the appropriate model, we conducted standard spatial model selection tests. Robust LM tests for both SEM and SAR rejected the null hypothesis, indicating that the spatial Durbin model (SDM) is preferable, as it accounts for spatial dependence in both the lagged dependent variable and the error term. The Hausman test further rejected the null at the 1% level, justifying the use of a fixed-effects specification, which accounts for city-specific effects correlated with the explanatory variables. Finally, Wald and LR tests confirmed that the SDM cannot be simplified to either SAR or SEM, suggesting that spatial spillovers operate simultaneously through the dependent variable and the error structure. These results, detailed in

Table 10. Spatial model selection test results.

Test	Null Hypothesis	Statistic	Results
LM-error	SEM	33.070 ***	SDM
Robust LM-error	Robust SEM	35.363 ***	SDM
LM-lag	SAR	1.424	SAR
Robust LM-lag	Robust SAR	3.717 *	SDM
Hausman	Random effect	47.180 ***	Fixed effect
LR-SEM	SDM can be simplified to SEM	110.890 ***	SDM
LR-SAR	SDM can be simplified to SAR	101.680 ***	SDM
Wald-SEM	SDM can be simplified to SEM	112.730 ***	SDM
Wald-SAR	SDM can be simplified to SAR	102.820 ***	SDM

Note: t statistics in parentheses. * p < 0.1, *** p < 0.01.

, indicate that a two-way fixed-effects SDM is the most suitable specification.

Column (1) in

Table 11. Spatial Durbin model results.

	(1)	(2)	(3)	(4)
	Inverse Distance	Economic Distance	Administration Adjacency	Non-Administration Adjacency
GF	0.117 ***	0.077 *	0.077 *	0.024
	(2.67)	(1.71)	(1.71)	(0.50)
W $\times$ GF	1.299 ***	0.176 ***	0.330 ***	−5.735 ***
	(4.48)	(2.79)	(3.86)	(−4.01)
Control variables	Yes	Yes	Yes	Yes
Direct effect	0.149 ***	0.096 **	0.111 **	0.111 **
	(3.20)	(2.04)	(2.37)	(2.46)
Indirect effect	2.845 ***	0.261 ***	0.412 ***	−0.994 ***
	(3.55)	(3.32)	(4.41)	(−3.90)
Total effect	2.994 ***	0.357 ***	0.523 ***	−0.883 ***
	(3.67)	(3.65)	(4.67)	(−3.43)
ρ	0.510 ***	0.280 ***	0.284 ***	−5.411 ***
	(5.12)	(7.84)	(8.07)	(−9.81)
sigma2_e	5.072 ***	5.181 ***	5.127 ***	4.695 ***
	(22.95)	(22.87)	(22.64)	(21.59)
N	1067	1067	1067	1067
${\mathrm{R}}^2$	0.143	0.204	0.008	0.043

Note: t statistics in parentheses. * p < 0.1, ** p < 0.05, *** p < 0.01.

reveals that energy transition exhibits a spatial autocorrelation coefficient of 0.51, with significance confirmed at the 1% level, suggesting the presence of spatial clustering. Green finance development level has a significantly positive effect on the energy transition index, both locally and in neighboring regions, indicating the presence of spatial spillovers. To facilitate interpretation, the partial differential approach decomposes the estimated coefficients into direct effects on the local area, indirect effects on surrounding regions, and the total effect [73]. The coefficient of indirect effect is significantly positive, which indicates that enhancements in the local green finance positively influence the energy transition of neighboring cities. This is consistent with Hypothesis 3a in this study and the conclusions from Chen et al. (2025) [32].

Higher rates of economic growth are associated with elevated energy demand, highlighting the strong dependence of energy consumption on regional economic activity [74]. To better capture this relationship and reinforce the robustness of our results, an economic distance weight matrix was constructed, with matrix elements defined as the inverse of the average per capita GDP differences between city pairs from 2009 to 2019. Column (2) of Table 11 indicates that green finance significantly advances energy transition in economically proximate regions. An increase in investment in one green industry can stimulate investment demand in upstream and downstream sectors and promote the development of the entire industry chain, thereby generating industrial clustering effects and achieving coordinated regional development. Moreover, knowledge spillover effects facilitate the sharing of low-carbon energy technologies, further promoting energy transition in surrounding regions [75].

To further examine the differences in spillover effects within and across provincial boundaries, we constructed an administrative adjacency weight matrix, represented as a 0–1 matrix. In this matrix, a value of 1 indicates that two cities are located within the same province, while a value of 0 denotes otherwise. The non-administrative adjacency weight matrix uses the opposite calculation method, where a value of 1 denotes that two cities belong to different provinces, and 0 otherwise. Evidence from Columns (3) and (4) demonstrates that green finance enhances energy transition among intra-provincial cities but has a significant negative spillover effect on inter-provincial cities. This may be attributed to interprovincial resource competition, which creates a crowding-out effect that hinders the energy transition of cities outside the focal province, thereby generating a negative spatial spillover. Local protectionism in China leads provincial governments to often prioritize local economic and environmental targets, favoring domestic firms and projects over external ones, which can reduce cross-provincial capital flows [76]. Market segmentation leads to differences in regulatory standards and financial supervision across provinces, raising transaction costs for cross-regional investments and intensifying interprovincial competition [77].

To identify the boundary of the spatial spillover effects, we constructed a segmented distance weight matrix, with its elements defined as follows:

$$\begin{array}{c}{W}_{ij}=\left\{\begin{array}{l}{\displaystyle \frac{1}{d_{ij}}}, if d_{min,t}<d_{ij}<d_{max,t}, i\ne j\\ 0, if i=j or d_{ij}<d_{min,t} or d_{ij}>d_{max,t}\end{array}\right.\end{array}$$

(6)

where $d_{ij}$ represents the geographic distance between two cities, while $d_{min}$ and $d_{max}$ denote the distance thresholds. Using this matrix, we investigate how the impact of green finance on energy transition changes across distance, with SDM regressions conducted at increments of 200 km up to 1000 km. The results of the indirect effects are presented in Figure 2, where the error bars denote 95% confidence intervals, and further details available in

Table A2. Spillover boundary test results.

	(1)	(2)	(3)	(4)	(5)
	0–200 km	200–400 km	400–600 km	600–800 km	800–1000 km
GF	0.109 **	0.114 **	0.118 **	0.125 ***	0.142 ***
	(2.41)	(2. 54)	(2.55)	(2.77)	(3.08)
W*GF	0.130 *	0.201 *	−0.037	−0.164	−0.017
	(1.84)	(1.71)	(−0.27)	(−1.07)	(−0.12)
Control variables	Yes	Yes	Yes	Yes	Yes
Direct effect	0.119 **	0.128 ***	0.119 **	0.122 ***	0.145 ***
	(2.55)	(2.68)	(2.49)	(2.60)	(2.98)
Indirect effect	0.151 **	0.296 **	−0.014	−0.171	0.053
	(2.20)	(2.05)	(−0.09)	(−0.72)	(0.25)
Total effect	0.270 ***	0.424 **	0.106	−0.049	0.198
	(3.14)	(2.53)	(0.62)	(−0.19)	(0.86)
ρ	0.150 ***	0.270 ***	0.163 ***	0.377 ***	0.321 ***
	(4.34)	(5.98)	(3.14)	(6.23)	(4.88)
sigma2_e	5.416 ***	5.349 ***	5.755 ***	5.421 ***	5.722 ***
	(22.92)	(22.92)	(22.95)	(23.28)	(23.01)
N	1067	1067	1067	1067	1067
${\mathrm{R}}^2$	0.000	0.022	0.077	0.072	0.154

Note: t statistics in parentheses. * p < 0.1, ** p < 0.05, *** p < 0.01.

. The results indicate that green finance has a significant impact on energy transition of all the surrounding cities within 400 km. Green finance has proven particularly effective in boosting energy transition in cities located within a 200–400 km radius. Beyond 400 km, the positive spatial spillover effects tend to diminish, indicating that the effective boundary is approximately 400 km. The results confirm Hypothesis 3b. Central regions may exert a “siphoning” influence on neighboring peripheral areas, redirecting resources and generating agglomeration shadow zones [52]. This accounts for the observation that the spatial spillover is weaker in nearby areas and increases with greater distance. Beyond the maximum spatial range, the influence of green finance weakens as interregional connections become too distant to sustain benefits.

5. Further Discussion: Green Finance Policy

In 2017, China launched the Green Financial Reform and the Innovation Pilot Zones (GFRIPZ) in eight cities, namely Huzhou, Quzhou, Guiyang, Nanchang, Hami, Changji, Karamay, and Guangzhou. These zones support local areas in building green finance systems with different focuses and characteristics and explore the development path of the green finance system. We applied the DID model, a quasi-natural experimental method that effectively mitigates endogeneity concerns. This approach has been widely used in the empirical evaluation of policy interventions [78].

$$\begin{array}{c}{{ETI}_{it}={\sigma }_0+{\sigma }_1{Treat}_i{ \times Post}_t+{\sigma }_k{Controls}_{it}+{\gamma }_t+{\mu }_i+\epsilon }_{i,t}\end{array}$$

(7)

where ${Treat}_i{ \times Post}_t$ is a dummy variable equal to 1 if city i implements a GFRIPZ in year t, and 0 otherwise. In the estimation equation, the key coefficient of interest, ${\sigma }_1$, captures the impact of green finance pilot zones on energy transition.

To ensure the appropriateness of the DID methodology, it is necessary that treatment and control groups have similar trends before policy enactment. The parallel trend test, depicted in Figure 3 with error bars representing 95% confidence intervals, indicates no significant pre-policy differences in energy transition between the two groups. Following the introduction of the green finance pilot in 2017; however, the growth paths diverged, validating the parallel trend assumption.

As shown in

Table 12. DID model results.

	(3)	(4)
	ETI	ETI
Treat $\times$ Post	1.815 **	1.751 **
	(2.25)	(2.38)
lnpeople		−3.523 **
		(−2.45)
income		6.475 ***
		(4.10)
gn		−1.238
		(−1.37)
info		0.931 **
		(2.70)
_cons	45.167 ***	−4.013
	(3.9 × 1013)	(−0.21)
Fixed effect	Yes	Yes
N	1067	1067
${\mathrm{R}}^2$	0.777	0.787

Note: t statistics in parentheses. ** p < 0.05, *** p < 0.01.

, the effect of Treat $\times$ Post is 0.0114 and significant at 5% in the model without controls. Including control variables slightly increases this estimate to 0.0128, which continues to be significant at the 5% level.

Moreover, the estimate results of the control variables are basically consistent with those from the fixed-effects model, reinforcing the robustness of the baseline specification. This indicates that the policy has a meaningful impact on advancing energy transition and is independent of external factors, which aligns with the findings of Chen et al. (2025) [32].

To further assess the spatial impact of green finance, this study investigates whether the implementation of green finance pilot policies exerts spillover effects on non-pilot areas. The spatial Durbin DID (SDM-DID) model was further developed and extended in this study as follows:

$$\begin{gathered} {ETI}_{it}=\rho \sum w_{ij}\times {ETI}_{jt}+{{\beta }_1{Treat}_i{ \times Post}_t+\beta }_k{Controls}_{it} \\ +{{\delta }_1\sum w_{ij}\times {Treat}_i{ \times Post}_t+\delta }_k\sum w_{ij}\times {Controls}_{it}+{\gamma }_t+{\mu }_i+{\epsilon }_{it} \end{gathered}$$

(8)

where δ represents the spatial-effect coefficient associated with the policy term and the control variables, and $w$ represents the inverse-distance weight matrix.

Table 13. SDM-DID model results.

Variable	Coefficient	Variable	Coefficient
Treat $\times$ Post	1.903 **	Direct effect	2.285 **
	(2.34)		(2.49)
W $\times$ Treat $\times$ Post	10.992 *	Indirect effect	30.997 *
	(1.95)		(1.82)
Control variables	Yes	Total effect	33.281 *
			(1.90)
ρ	0.583 ***	N	1067
	(6.41)	${\mathrm{R}}^2$	0.140
sigma2_e	5.160 ***
	(22.88)

Note: t statistics in parentheses. * p < 0.1, ** p < 0.05, *** p < 0.01.

presents the estimation results of the SDM-DID model. Both the coefficients of W $\times$ Treat $\times$ Post and the indirect effects are significantly positive, indicating that the GFRIPZ policy generates positive spatial spillovers in promoting energy transition. That is, the development of green finance in pilot cities can promote energy transition in surrounding non-pilot regions. The results are consistent with those of the SDM regression on green finance development level and support Hypothesis 3a. Existing studies on green finance policies indicate that such policies can reduce the energy consumption of cities surrounding pilot regions [79] and accelerate the energy transition in adjacent areas [32]. The green finance policies of pilot cities serve as a demonstration or warning to neighboring cities. They can promote the development of the green finance system through policy imitation and learning, thereby facilitating the overall energy transition. Notably, the indirect effect far exceeds the direct effect. This result may be sensitive to the specification of the spatial weight matrix, which constitutes a potential limitation of this analysis. Further research could provide a more detailed investigation of the underlying spatial spillover patterns.

6. Conclusions

Energy transition is advanced through green finance, which stimulates green innovation and facilitates industrial structure optimization. Furthermore, higher levels of public environmental awareness reinforce this effect. The results also highlight significant regional differences, with green finance exerting a stronger impact on energy transition in resource-intensive regions and in areas where green finance is more advanced.

Moreover, the effect of green finance on energy transition demonstrates spatial spillovers, both in terms of geographical and economic distance. Inter-regional collaboration can generate greater overall benefits. Considering administrative proximity, green finance generally generates positive spillover effects on cities located within the same province. In contrast, the spillover effects of green finance on cities located in different provinces appear to be negative, potentially due to regional competition. Competition for limited financial resources and central policy support may lead to a crowding-out effect, wherein the expansion of green finance in one province reduces its growth in adjacent regions. Overall, the spillover effect of green finance varies with distance. Our findings indicate that the beneficial effect of green finance on neighboring regions strengthens with increasing distance. Its external benefits are most pronounced within an intermediate range of approximately 200–400 km, which may reflect the optimal spatial scope for infrastructure connectivity and industrial linkage networks. However, once the distance exceeds 400 km, the positive externalities of green finance decline, suggesting a threshold effect beyond which geographical separation dilutes the transmission of benefits.

Additionally, the green finance pilot policy significantly accelerates energy transition and exhibits notable spatial spillover effects. The pilot regions serve as effective demonstration zones, where the development of green finance can strongly incentivize energy transition in surrounding areas through policy diffusion.

The evidence supports several policy actions. First, policymakers should maintain focus on the energy sector and expand the service scope. The threshold regression results imply that green finance needs to reach a sufficient scale to exert a significantly positive influence on energy transition. The range of projects supported by green bonds should be broadened to include more renewable energy enterprises.

Second, based on the mechanism analysis, priority should be placed on enhancing energy technology innovation. A specialized fund for green technology R&D should be established to guide capital flows toward technological and institutional innovations in new energy system conversion, storage, and transportation. Additionally, improving the green intellectual property protection and trading market will help accelerate the commercialization of innovation outcomes.

Third, based on regional heterogeneity, the development of regionally differentiated support policies is recommended. Ensuring the long-term development of resource-based regions is essential for national economic stability. Policies should prioritize these areas by allocating more carbon reduction tools. In resource-rich areas such as Shanxi, Shaanxi, and Inner Mongolia, green finance reform pilot zones should be established, offering financial subsidies and risk compensation to support the transition of traditional energy enterprises. By leveraging the spatial spillover effects of these pilot zones, the positive influence on energy transition can be extended to neighboring cities. Rather than overexploiting resources, efforts should focus on improving coal conversion efficiency.

Fourth, cross-regional joint energy transition projects within the 400 km range should be launched to promote industrial synergy, create interconnected green energy clusters and accelerate the transformation of regional energy consumption structures. Since positive spillover effects exist within provinces, policies should encourage cooperation among cities in the same province to share best practices, resources, and technologies. However, considering the negative impact on energy transition in cities across different provinces, a unified regional framework for green finance standards and policies should be established. Overcoming administrative barriers crucial for mitigating local competition [80] and market segmentation [81]. Specifically, interprovincial coordination could be strengthened through establishing green finance cooperation platforms, which facilitate joint project approval, information sharing, and coordinated investment planning. Such platforms can help lower transaction costs associated with cross-regional capital flows.

As with all empirical studies, this study has some limitations. First, constraints in data availability hinder the direct measurement of certain green financial instruments, including green investment and green insurance. If data disclosure processes require more consistent reporting on green finance in the future, we will be able to construct a more precise analytical framework. Second, the current limited development of green finance restricts the number of cities with green enterprises. As the adoption of green financial instruments expands, future studies can increase sample sizes to enhance the robustness of the study.