The Impact of Green Finance and Financial Technology on Regional Green Energy Technological Innovation Based on the Dual Machine Learning and Spatial Econometric Models

Abstract

Regional green energy technological innovation is an important means to alleviate economic–environmental contradictions. The purpose of this study was to explore the mechanisms of green finance, financial technology, and regional green energy technological innovation. In this study, we constructed dual machine learning models, spatial econometric models, and panel threshold effect models to investigate the effects of green finance and financial technology on regional green energy technological innovation, using panel data from 266 cities nationwide from 2009 to 2021. The research findings are as follows: (1) Both green finance and financial technology significantly promote regional green energy technological innovation. (2) Based on a spatial weight matrix embedded in economic geography, both green finance and financial technology generate positive spatial spillover effects on regional green energy technological innovation. (3) The interaction between green finance and financial technology significantly contributes to regional green energy technological innovation. Financial technology can strengthen the positive local and neighboring effects of green finance on regional green energy technological innovation. (4) Based on the threshold effect of financial technology, green finance cannot significantly promote regional green energy technological innovation when financial technology is in an underdeveloped stage. With the advancement of financial technology, green finance continues to have a positive impact on regional green energy technological innovation. Based on this analysis and our conclusions, we propose practical policy recommendations that can provide a more sustainable approach to green energy technology innovation.

1. Introduction

In recent years, based on achieving “dual carbon” goals, China has considered green energy technological innovation as a crucial lever for promoting low-carbon economic development and achieving green development objectives. The reason for this lies in the ways green technology, by improving and transforming traditional methods, not only fosters regional economic growth but also compels product suppliers to enhance pollution control technologies and reinforce clean production process R&D, further achieving a win–win situation of economic growth and energy conservation along with emission reduction [1].

Given that enterprises are the primary implementers of regional green energy technological innovation, which demands substantial financial investment, it is imperative for them to leverage the financial markets. Maximizing financial resources and channeling them into clean technology fields has given rise to the concept of green finance. According to data from the People’s Bank of China, by the end of the first quarter of 2023, China’s domestic and foreign currency green loans exceeded CNY 25 trillion, and the balance of green bonds exceeded CNY 1.5 trillion, ranking the country among the top globally. This demonstrates that green finance, as a specialized financial service in response to environmental pollution, resource waste, and low utilization issues, can guide enterprises in their green technological innovation endeavors and support regional and corporate green technological innovation.

As China accelerates the pace of technological innovation, the role of financial technology (fintech) has become increasingly prominent, exerting a significant impact on regional green technological innovation [2]. Fintech refers to a new business format that utilizes advanced technology to enhance the efficiency and quality of the financial industry. It involves employing technologies like digital currency, blockchain, and machine learning to offer more efficient, convenient, and innovative financial solutions. On the one hand, fintech provides more flexible and convenient financing channels for green technological innovation [3]. On the other hand, the innovation of green technology heavily relies on the support of fintech in information services [4]. Fintech efficiently mitigates information asymmetry problems for businesses engaged in green innovation by giving innovators fast, accurate, and thorough information that helps them comprehend market needs, competitive environments, and technological developments.

Development in green finance and financial technology has provided an impetus for information exchange, capital flow, and knowledge dissemination. On the one hand, both green finance and financial technology can infuse sufficient elements into the construction of ecological civilizations and the development of eco-friendly industries, driving industrial transformation and upgrading [5]. On the other hand, the development of green finance and financial technology has provided extensive financial and technological support for research and development of energy-saving and low-carbon technologies, thereby enhancing the effectiveness of green technological innovation [6]. It can be seen that green finance and financial technology are likely to exert a specific impact on regional green technological innovation. However, current research is still deficient in terms of analyzing the underlying mechanisms. Hence, in order to explore the specific impact mechanisms, it is necessary to analyze green finance, financial technology, and regional green technological innovation within a unified research framework. In this study, we utilized panel data from 2009 to 2021, encompassing 266 prefecture-level cities and municipalities in China as the research subject. We constructed dual machine learning models, spatial econometric models, and panel threshold effect models to explore the relationships among green finance, financial technology, and regional green technological innovation.

This paper is primarily divided into the following sections: In the second part, existing research is discussed and summarized. The third section introduces the main theoretical mechanism through theoretical inference, then describes the construction and selection of the research model, and the fourth section describes the indicator system for various variables. The fifth section provides the empirical analysis of the models using the panel data. Finally, conclusions and policy recommendations are provided in the last section, along with a discussion on the marginal contributions and limitations of the study.

2. Literature Review

2.1. Review of Research on Green Finance and Regional Green Energy Technological Innovation

A review of the existing literature shows that the academic community has yet to form a consistent definition of green finance. However, it can be inferred that the essence of green finance involves integrating concepts of resource conservation and environmental protection into economic growth. It is a fresh take on finance intended to encourage synchronized growth of the natural environment, resources, and economy. Through innovative financial tools like green credit, green finance channels funds into energy-saving, environmentally friendly projects and provides financial services for these endeavors. In recent years, green finance has garnered extensive attention in academia as a pivotal driver of regional economic and environmentally sustainable development. Scholars’ research on it mainly focuses on two aspects: some scholars pay more attention to the economic benefits of the green industry, such as, for example, Hassan et al. explored the role of green finance in the low-carbon economy [7], Yin and Xu believed that green finance can promote economic growth [8], and Yang et al. studied the relationship between green finance and high-quality economic development [9]; the other part of the scholars mainly analyzed the environmental benefits of the green projects, such as the green finance project methods to obtain sustainable performance [10], the positive impact of green finance on regional ecological development in China [11], and the role of green finance in improving eco-efficiency [12].

However, more research is needed on the relationship between green finance and regional green technological innovation in the international arena. The conclusions in the existing literature regarding the impact of green finance on technological innovation can be broadly categorized into two perspectives. As a vital method to alleviate enterprise constraints, some scholars believe innovative financing can continuously support green technological innovation, which is a significant force in driving such innovations [13]. For instance, Jin and Han studied the financial performance of China’s green funds and analyzed their influence on the manufacturing industry, revealing that green funds could act as a financial force to promote green technological innovation in industries, benefiting small and medium-sized enterprises with new intelligent and green technologies [14]. He et al. demonstrated the incentivizing role of green finance in the green technological innovation activities of enterprises from the perspective of green credit [15]. Wang et al. explored the impact of issuing green bonds on green innovation and technology and suggested that issuing green bonds could alleviate corporate financing constraints and promote green technological innovation [16]. From a macro policy standpoint, several academics have examined how green financing policies affect technological innovation in the green space. Research findings suggest that enterprises guided by green finance policies significantly improve their performance in green innovation [17]. For instance, Sun and Meng analyzed the relationship between green finance policy and technological innovation from the data in the pilot zone of green finance reform [18], Zhang et al. studied the impact of green finance policy on regional green innovation [19], and Liu and Wang mainly focused on the green innovation effect in the pilot zone of green finance reform [20]. Additionally, some scholars have constructed comprehensive green finance indicator systems, incorporating various financial tools, such as green investments, insurance, and equities, to study the influence of green finance on enterprises’ green technological innovation [16].

However, some scholars have expressed contrary opinions. They argue that the development of green finance increases enterprise costs, thus reducing the liquidity of corporate funds and hindering green technological innovation. Additionally, they highlight that some enterprises merely use the “green” label and slogans while prioritizing profit maximization in their daily operations and investment decisions. These practices, they argue, neglect environmental factors and corporate social responsibility, leading to “greenwashing” and ineffective support for innovation in the green technology sector [21].

2.2. Review of Research on Financial Technology and Regional Green Energy Technological Innovation

Fintech is the sum of innovative technologies that apply emerging technologies such as artificial intelligence, blockchain, cloud computing, big data, and other emerging technologies to the financial sector to form new financial models, financial formats, and financial products. The development of technologies such as distributed computing, the Internet of Things (IoT), and biometrics provides technical support and a driving force for the transformation and change of the financial industry in the areas of delivery methods, credit assessment, payment and settlement, and risk supervision.

Existing research indicates that financial technology has a significant impact on promoting regional green innovation, primarily by reducing financing constraints, alleviating information asymmetry, and fostering industrial upgrading. Chen et al. [22] suggest that financial technology significantly contributes to easing financing constraints and directing more capital to technology innovation firms or research institutions, thereby mitigating financial sector concentration. Additionally, fintech can potentially support fundraising for large corporations by offering credit solutions or initial coin offering (ICO) options, the latter based on blockchain technology, providing more security, liquidity, and transparency compared to other financing schemes [23]. This impact is particularly noticeable in regions with relatively free sectors and a greater level of marketization [22]. Financial technology can also penetrate emerging markets with limited financial infrastructure, reducing compliance costs and enhancing the accuracy of risk assessment by strengthening identity verification, investment profiles, and consumer behavior [23].

On the other hand, the inherent characteristics of financial technology allow robust information-matching capability [24]. It can utilize digital technologies like extensive data analysis to alleviate information asymmetry within the green innovation market and indirectly enhance environmental information disclosure quality among enterprises [25]. The existing literature indicates that financial technology has expanded financial service coverage and optimized capital allocation among industries, thereby fostering upgrades in industrial structure [26]. Digital finance (DF) has given China’s industrial sector the much-needed financial boost it needs to modernize and change. According to Sheng et al., financial technology has a substantial beneficial influence on technical innovation and greatly accelerates it in manufacturing businesses [27].

The development of an advanced financial system can enhance capital allocation efficiency by channeling capital into green, low-carbon industries, thereby stimulating the vitality of green innovation for enterprises. Financial technology, known for its inclusivity, also provides financial services to individuals excluded from traditional finance. The growth of financial technology has contributed to realizing incentives for financial resource allocation mechanisms. From the perspective of Internet finance, it emphasizes the contribution of financial market growth to fostering innovation among green production enterprises [28]. The advancement of financial technology effectively addresses issues in the traditional financial system, such as inadequate resource liquidity and misallocation, improving the efficiency of unused resource mobilization among market participants [29], and enabling a timely flow of innovation resources to businesses in need of green technological innovation.

Moreover, based on a comprehensive index of green finance development, Chen [30] employed non-parametric data envelopment analysis and directional distance function (DEA-DFF) models and concluded that green finance significantly enhances the level of green productivity. Financial development and technological innovation can also elevate green total factor productivity (GTFP) [30]. Additionally, financial technology exhibits a positive correlation with green credit [24].

However, there is regional heterogeneity in how financial technology affects green technological innovation, with the developmental influence in local and neighboring areas evident only in certain regions. Financial development does not uniformly positively impact innovation activities across all regions; instead, its supportive role in innovation activities within a particular region is fully realized only after a certain level is reached [31].

Summarizing the above, we find that in recent years, there has been increasing attention to green finance and financial technology among the academic community, and the existing research tends to indicate more and more that green finance and financial technology have a certain positive impact on technological innovation, but there has not been in-depth research on it, and most studies have been stuck in the theoretical inference. Therefore, studies on the role of both in regional green energy technological innovation mechanisms are still lacking. For example, will the spatial effect of the two promote or inhibit regional green energy technology innovation? In the context of financial technology, what is the impact of green finance on regional green technology innovation? In the development process of financial technology, what changes will occur in the influence mechanism of green finance on regional green technology innovation?

3. Theoretical Mechanism

3.1. Green Finance and Regional Green Energy Technological Innovation

The dual carbon goals clearly indicate that carbon reduction will be the dominant strategy for the construction of an ecological civilization in China in the foreseeable future. High-tech and clean production technologies are crucial to achieving this strategic objective. By advancing the research, application, and dissemination of green technologies, it is possible to reduce greenhouse gas emissions, enhance resource efficiency, and foster sustainable economic development. However, such technological innovations require substantial funding. Hence, developing green finance is considered an effective way to facilitate regional green energy technological innovation, which will consequently promote carbon reduction [32,33]. Green energy technologies can be financially supported by green finance while reducing the financing costs associated with innovation activities. By introducing financial tools like green bonds and green loans, enterprises engaged in technological innovation can access funds for the research, production, and promotion of environmentally friendly, low-carbon technologies [34]. Furthermore, green finance assumes a supervisory role and forces businesses to participate in green innovation, acting as an essential tool to drive the transformation of development models toward sustainability. When providing financial support, financial institutions evaluate and disclose the environmental management practices of enterprises engaged in green technological innovation and encourage them to focus more on environmental responsibility and sustainable operations [35].

From a regional perspective, there are significant disparities in resource endowment and development levels among different regions in China. However, close economic ties and frequent flows of production factors exist among regions. The mutual influence and coordinated development among regions cannot be ignored. Therefore, further exploration is necessary to understand the spatial spillover effects of green finance on regional green energy technological innovation.

The development of green finance can stimulate local green energy technological innovation and industrial structure upgrades within a region [36]. Extending financial services and industrial networks can increase investments in neighboring regions, facilitate knowledge and information spillovers, and promote specialized divisions of labor, thus affecting the technological innovation level and industrial structure of surrounding areas [37]. From the perspective of green technological innovation, support from green finance attracts more investment and funds into the local green technology sector [38]. These funds can be utilized to develop new technologies, improve existing ones, and expand the scale of green projects. The diffusion of capital, knowledge, and technology within the local area will contribute to promoting technological innovation in other regions [39]. Concerning the industrial structure, green finance drives local enterprises toward green and sustainable development, which optimizes and upgrades the local industrial structure. This, in turn, attracts more enterprises and institutions from neighboring regions to the green technology sector, forming a virtuous cycle of industrial agglomeration [40]. Additionally, successful innovation in green energy technology by certain entities supported by green finance can have a demonstration effect, prompting interest and emulation by enterprises and institutions in other regions, thus propelling the innovation and application of green energy technology and fostering green technological development within their respective regions [41]. Therefore, green finance emerges as a critical driver of the coordinated development of regional green energy technological innovation. Technological innovation enterprises or projects supported by green finance not only drive their own and the entire region’s development but also generate a positive spillover effect on other regions. Based on the above analysis, we put forward the following hypotheses:

Hypothesis 1 (H1).

Green finance has a significantly positive impact on regional green energy technological innovation.

Hypothesis 2 (H2).

Green finance exhibits a significantly positive spatial spillover effect on regional green energy technological innovation.

3.2. Financial Technology and Regional Green Energy Technological Innovation

In 2011, the term “financial technology” was officially introduced to describe how technology-driven enterprises utilize cutting-edge technological advancements to strengthen the financial industry. Essentially, financial technology represents a shift toward technology-driven service provision, with the underlying goals revolving around optimization and transformation within the financial sector. Its dual objectives are to enhance efficiency and reduce costs, especially for enterprise users (B2B) [42]. The development of financial technology has propelled information exchange, capital flow, and knowledge dissemination. On the one hand, the “technological dividends” generated by the advancement of financial technology can be widely applied to the research and development of low-carbon and energy-saving technologies, thereby enhancing the green effects of innovation. On the other hand, financial technology can infuse sufficient elements into the development of eco-friendly industries and the construction of an ecological civilization, driving industrial upgrading and transformation.

Green innovation is characterized by high starting points, significant investments, long cycles, and high risk. The pricing of green products sometimes dissuades certain customers [43]. Research indicates that even after implementing green innovation, companies may not experience significantly improved financial performance; in some cases, there may even be a decline [44]. Organizations transitioning toward green practices require financial or other resource support to achieve breakthrough green innovation. Moreover, outside of company boundaries, bank credits involve high management and supervisory costs both during and after the loan period. Serious “market failures” have occurred in green innovation concerning investment and transformation. Given China’s current economic transformation amidst an economic downturn, many emerging knowledge-intensive enterprises face internal financial constraints, necessitating the acquisition of essential funds through external financing to speed up the development and transformation of green innovation and research outcomes.

In this context, financial technology can potentially overcome the financial constraints faced by enterprises and mitigate the negative impact of “dual externality” on green innovation. “Dual externality” refers to the impact of innovation during its initial stages continuing to affect subsequent developments [45]. It includes negative components that, to some extent, diminish enterprises’ enthusiasm for choosing green innovation. On the one hand, green innovation has a knowledge spillover effect. In this scenario, the lack of market protection for intellectual property rights might result in some companies imitating another company’s green technology, thereby weakening the incentive for the company to continue innovating [46]. On the other hand, there is the issue of trade-offs in pollutant emissions. Businesses may be hesitant to spend large amounts of money on green innovation if the cost of innovation is higher than the cost of emissions. This leads to companies evading regulations and choosing direct emission of pollutants as a way to control costs and maintain profitability [47].

Financial technology can be a driver when integrated with green innovation scenarios. Gomber et al. [48] emphasize that financial technology can create new products and services through big data, thereby reducing the cost of banks acquiring customers. On the other hand, from a risk management perspective, financial technology can operate in the credit market, reducing information asymmetry, lowering screening and monitoring costs, and restraining borrowing enterprises [49]. At the same time, by leveraging its characteristics, financial technology can assist lending institutions in establishing credit assessment models, further reducing risk assessment costs [50]. This advanced optimization process can alleviate credit constraints for relevant enterprises, improving the distribution of green credit efficiency. Additionally, it can enhance the post-lending supervision capability, thereby improving the efficiency of enterprise green investments.

Furthermore, there is a strong correlation between regional financial technology development and the driving force of financial technology behind green innovation. The number of green patents by firms increases with the degree of regional financial technology development. While optimizing modern financial systems, financial technology simultaneously offers favorable funding and financial support for enterprises’ green transformation.

In today’s sustainable development-oriented environment, finance acts as a driver of business innovation, and the field of green innovation is a constituent part of extensive innovative endeavors propelled by advancements in financial technology.

The promotion of green innovation by financial technology is primarily concentrated in regions with higher environmental regulatory standards, considerably more financial institutions, high-polluting industries, and enterprises facing stricter financing constraints. Elucidating the relationships between financial technology, green finance, and regional green technological innovation during the period of comprehensive economic and social green transformation has significant practical importance with regard to deepening financial institutions’ business operations, facilitating industrial green transformation and upgrading, and realizing China’s dual-carbon strategic goals.

This study proposes the following theoretical mechanisms based on the above analysis:

Hypothesis 3 (H3).

Financial technology has a significantly positive impact on regional green energy technological innovation.

Hypothesis 4 (H4).

Financial technology exhibits a significantly positive spatial spillover effect on regional green energy technological innovation.

3.3. Green Finance, Financial Technology, and Regional Green Energy Technological Innovation

In recent years, China’s green finance has made great strides, laying the groundwork for a green finance system primarily centered around green credit. However, this development has also encountered some issues, such as inefficiencies in green financial service delivery, the emergence of “greenwashing”, and challenges related to information asymmetry. These situations have cast doubt on the quality of green finance development [15,51]. Against the backdrop of a new technological revolution and industrial transformation, financial technology, represented by emerging technologies such as artificial intelligence, big data, cloud computing, and blockchain, is rapidly permeating the field of green finance. This integration not only empowers the development of green finance from various perspectives but also effectively addresses the various issues encountered in the process of developing green finance [52].

Financial technology can comprehensively collect enterprise information from multiple channels, effectively alleviating the information asymmetry prevalent in the field of green finance. This significantly reduces the difficulty of pre-review and the monitoring cost of green finance and enhances accuracy in identifying green entities and innovative green technology projects for financial institutions [53]. On the one hand, big data technology can capture and integrate various standardized and non-standardized data in real-time, consolidating the data into credit or green behavior information. On the other hand, blockchain technology, with its “unforgeable”, “traceable”, and “immutable” properties, can monitor the real-time flow of green funds [54]. At the same time, financial technology also uses digital technologies to provide relatively accurate forecasts of returns on various products. It can also precisely identify potential green finance demands of customers and enterprises across different scenarios, strongly propelling the innovation of green finance products and improving the efficiency of green financial services [55]. Moreover, through information processing technology, financial technology assists regulatory authorities in collecting transactional information related to green finance, enhancing the comprehensive management capability of this emerging industry. This support is crucial in preventing capital idleness and addressing issues related to greenwashing [56]. Green finance supports green projects and provides financial assistance to enterprises engaged in technological innovation, promoting the innovative development of green technology. Additionally, the extension of financial services and industrial networks plays a pivotal role in increasing investments, disseminating knowledge and information, and fostering a specialized division of labor in surrounding regions concerning green technological innovation.

Based on the above analysis, this paper posits the following hypothesis:

Hypothesis 5 (H5).

Green finance and financial technology have a significant joint positive impact on regional energy green technology innovation.

3.4. Green Finance and Regional Green Energy Innovation across Different Stages of Financial Technology Development

The promotion of regional green energy technological innovation by green finance cannot be detached from the empowerment provided by financial technology. Financial technology serves as a crucial tool for green finance to stimulate and incentivize elements crucial for technological development and innovation. It guides capital flows toward green and emerging technology industries, supports green innovation activities, and facilitates the greenization of financial fund circulation [48,57,58].

First, financial technology helps financial institutions alleviate issues such as idling capital and greenwashing caused by information asymmetry and incompleteness. In the absence of adequate financial technology (lacking relevant big data platforms and tools), investors struggle to obtain comprehensive information, making it challenging to appropriately assess the benefits and threats of regional technical advancement projects related to green energy. This limitation constrains the role of green finance in driving green energy technological innovation [59]. Additionally, financial technology based on digital technologies like big data, blockchain, and artificial intelligence incorporates economic development cycles, risk preferences, and personalized demands into the design of green financial products. It innovates financial service solutions to better meet enterprise clients’ needs at different research and development stages. However, when financial technology is underdeveloped, the capital flow tends to be restricted within traditional financial systems, making it challenging to channel funds specifically into green energy technological innovation areas. The lack of specialized green financial products and services makes it challenging for investors and fund providers to easily find opportunities for investing in green energy technological innovation [60]. Moreover, the development of financial technology dramatically simplifies and optimizes the transaction processes of green finance, reducing transaction costs while maintaining low operational risks. In contrast, underdeveloped financial technology results in complex and cumbersome transaction processes within green finance, indirectly limiting its role in driving green energy technological innovation [60].

Green finance thus can only successfully drive regional green energy technological innovation at a high degree of financial technology development. The following hypothesis is posited based on the above analysis:

Hypothesis 6 (H6).

When financial technology is underdeveloped, green finance cannot significantly drive regional green energy technological innovation. However, as the development level of financial technology increases, the positive impact of green finance on regional green energy technological innovation is continually strengthened.

4. Model Construction and Variable Selection

4.1. Model Construction

To verify the hypotheses and supplement the existing research, we constructed dual machine learning models, spatial econometric models, and panel threshold effect models to examine the impacts of fintech and green finance on regional green energy technological innovation. Additionally, we analyzed the interaction effects of spatial effects and their combined impact on regional green technology innovation. Furthermore, a panel threshold model was developed to examine the joint effect mechanism of financial technology and green finance on regional green energy technological innovation.

4.1.1. Dual Machine Learning Model Construction

To verify Hypotheses H1 and H3, we incorporated machine learning algorithms and adopted a double machine learning (DML) approach with reference to Chernozhukov et al. to correct the regularization bias present in machine learning algorithms. This method not only resolves the dimensionality issue in traditional regression analysis but also employs a partial linear regression approach that does not require the predetermined construction of control variables in the model [61]. Building on this, following the framework of Zhang and Li [62], the model was constructed as follows:

• Model 1:

  $${\mathrm{G}\mathrm{I}}_{\mathrm{i}\mathrm{t}}={\mathsf{\alpha }}_1{\mathrm{G}\mathrm{F}}_{\mathrm{i}\mathrm{t}}+\mathrm{g}\left({\mathbf{X}}_{\mathbf{i}\mathbf{t}}\right)+{\mathrm{U}}_{\mathrm{i}\mathrm{t}}, \mathrm{E}\left({\mathrm{U}}_{\mathrm{i}\mathrm{t}}|{\mathrm{G}\mathrm{F}}_{\mathrm{i}\mathrm{t}},{\mathbf{X}}_{\mathbf{i}\mathbf{t}}\right)=0$$

  (1)

• Model 2:

  $${\mathrm{G}\mathrm{I}}_{\mathrm{i}\mathrm{t}}={\mathsf{\alpha }}_2{\mathrm{F}\mathrm{T}}_{\mathrm{i}\mathrm{t}}+\mathrm{g}\left({\mathbf{X}}_{\mathbf{i}\mathbf{t}}\right)+{\mathrm{U}}_{\mathrm{i}\mathrm{t}}, \mathrm{E}\left({\mathrm{U}}_{\mathrm{i}\mathrm{t}}|{\mathrm{F}\mathrm{T}}_{\mathrm{i}\mathrm{t}},{\mathbf{X}}_{\mathbf{i}\mathbf{t}}\right)=0$$

  (2)

In the model, ${\mathsf{\alpha }}_1$, ${\mathsf{\alpha }}_2$ are the coefficients of green finance (${\mathrm{G}\mathrm{F}}_{\mathrm{i}\mathrm{t}}$) and financial technology (${\mathrm{F}\mathrm{T}}_{\mathrm{i}\mathrm{t}}$), which are used to measure the specific impact of green finance and financial technology on regional green energy technology innovation. $\mathrm{g}\left({\mathbf{X}}_{\mathbf{i}\mathbf{t}}\right)$ represents a function of multidimensional control variables. The specific functional form of $\mathrm{g}\left({\mathbf{X}}_{\mathbf{i}\mathbf{t}}\right)$ and its estimated value $\hat{\mathrm{g}}\left({\mathbf{X}}_{\mathbf{i}\mathbf{t}}\right)$ are obtained through the machine learning model. In the specific parameter estimation process, although the coefficient estimation values $\widehat{{\mathsf{\alpha }}_1}$, $\widehat{{\mathsf{\alpha }}_2}$ undergo reduced variance due to the dimensionality reduction introduced by the regularization term in the machine learning model, $\hat{\mathrm{g}}\left({\mathbf{X}}_{\mathbf{i}\mathbf{t}}\right)$ suffers from regularization bias. Hence, it is necessary to construct an auxiliary equation (taking model 1 as an example):

$${\mathrm{G}\mathrm{F}}_{\mathrm{i}\mathrm{t}}=\mathrm{m}\left({\mathbf{X}}_{\mathbf{i}\mathbf{t}}\right)+{\mathrm{V}}_{\mathrm{i}\mathrm{t}},\mathrm{E}\left({\mathrm{V}}_{\mathrm{i}\mathrm{t}}|{\mathrm{D}\mathrm{I}\mathrm{D}}_{\mathrm{i}\mathrm{t}},{\mathbf{X}}_{\mathbf{i}\mathbf{t}}\right)=0$$

(3)

In the above equation, it is necessary to employ a machine learning model to estimate $\mathrm{m}\left({\mathbf{X}}_{\mathbf{i}\mathbf{t}}\right)$ in order to obtain an estimate of this function, denoted as $\hat{\mathrm{m}}\left({\mathbf{X}}_{\mathbf{i}\mathbf{t}}\right)$. Then, an estimation of the residual term ${\mathrm{V}}_{\mathrm{i}\mathrm{t}}$ is calculated as follows:

$$\widehat{{\mathrm{V}}_{\mathrm{i}\mathrm{t}}}={\mathrm{G}\mathrm{F}}_{\mathrm{i}\mathrm{t}}-\hat{\mathrm{m}}({\mathbf{X}}_{\mathbf{i}\mathbf{t}})$$

(4)

At this point, the estimated ${\mathrm{V}}_{\mathrm{i}\mathrm{t}}$ can serve as an instrumental variable for ${\mathrm{G}\mathrm{F}}_{\mathrm{i}\mathrm{t}}$. By using machine learning, $\hat{\mathrm{g}}\left({\mathbf{X}}_{\mathbf{i}\mathbf{t}}\right)$ and $\widehat{{\mathsf{\alpha }}_1}$ can be estimated as follows:

$$\widehat{{\mathsf{\alpha }}_1}={\left(\frac{1}{\mathrm{n}}\sum _{\mathrm{i}\in \mathrm{I}\mathrm{t}\in \mathrm{T}}\widehat{{\mathrm{V}}_{\mathrm{i}\mathrm{t}}}{\mathrm{G}\mathrm{F}}_{\mathrm{i}\mathrm{t}}\right)}^{-1}\frac{1}{\mathrm{n}}\sum _{\mathrm{i}\in \mathrm{I}\mathrm{t}\in \mathrm{T}}\widehat{{\mathrm{V}}_{\mathrm{i}\mathrm{t}}}({\mathrm{G}\mathrm{I}}_{\mathrm{i}(\mathrm{t}+1)}-\hat{\mathrm{g}}({\mathbf{X}}_{\mathbf{i}\mathbf{t}}))$$

(5)

where n represents the total sample size.

4.1.2. Spatial Econometric Model Construction

Spatial econometrics takes into consideration the spatial dependency of economic variables in geographical units within a spatial panel, in contrast to traditional econometric models. To verify Hypotheses H2 and H4, we referred to the analysis of spatial econometric models by Anselin [63] and Elhorst [64] to construct a spatial autoregressive (SAR) model (SAR), spatial error model (SEM), and spatial Durbin model (SDM) to test for spatial effects. A SAR model primarily explores the dependent variable’s spatial dependence, the SEM is used to discuss the spatial effects of the error terms, and the SDM builds upon the other two by considering the spatial transmission effect of the explanatory variables on the dependent variable.

• Model 3:

  $${\mathrm{G}\mathrm{I}}_{\mathrm{i}\mathrm{t}}=\mathsf{\rho }\mathrm{W}·{\mathrm{G}\mathrm{I}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\alpha }}_1{\mathrm{G}\mathrm{F}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\alpha }}_2{\mathrm{F}\mathrm{T}}_{\mathrm{i}\mathrm{t}}+{\sum }_{\mathrm{j}}{\mathsf{\beta }}_{\mathrm{j}}{\mathrm{x}}_{\mathrm{i}\mathrm{t}}^{\mathrm{j}}+{\mathsf{\gamma }}_{\mathrm{t}}+{\mathsf{\mu }}_{\mathrm{i}\mathrm{t}}+{\mathsf{\epsilon }}_{\mathrm{i}\mathrm{t}}$$

  (6)

• Model 4:

  $${\mathrm{G}\mathrm{I}}_{\mathrm{i}\mathrm{t}}={\mathsf{\alpha }}_1{\mathrm{G}\mathrm{F}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\alpha }}_2{\mathrm{F}\mathrm{T}}_{\mathrm{i}\mathrm{t}}+{\sum }_{\mathrm{j}}{\mathsf{\beta }}_{\mathrm{j}}{\mathrm{x}}_{\mathrm{i}\mathrm{t}}^{\mathrm{j}}+{\mathsf{\gamma }}_{\mathrm{t}}+\mathsf{\lambda }\mathrm{W}·{\mathsf{\nu }}_{\mathrm{i}\mathrm{t}}+{\mathsf{\mu }}_{\mathrm{i}\mathrm{t}}+{\mathsf{\epsilon }}_{\mathrm{i}\mathrm{t}}$$

  (7)

• Model 5:

  $${\mathrm{G}\mathrm{I}}_{\mathrm{i}\mathrm{t}}=\mathsf{\rho }\mathrm{W}·{\mathrm{G}\mathrm{I}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\alpha }}_1{\mathrm{G}\mathrm{F}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\alpha }}_2{\mathrm{F}\mathrm{T}}_{\mathrm{i}\mathrm{t}}+{\sum }_{\mathrm{j}}{\mathsf{\beta }}_{\mathrm{j}}{\mathrm{x}}_{\mathrm{i}\mathrm{t}}^{\mathrm{j}}+{\mathrm{L}\mathrm{a}\mathrm{g}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\gamma }}_{\mathrm{t}}+{\mathsf{\mu }}_{\mathrm{i}\mathrm{t}}+{\mathsf{\epsilon }}_{\mathrm{i}\mathrm{t}}$$

  (8)

  $${\mathrm{L}\mathrm{a}\mathrm{g}}_{\mathrm{i}\mathrm{t}}=\mathrm{W}·({\mathsf{\theta }}_1{\mathrm{G}\mathrm{F}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\theta }}_2{\mathrm{F}\mathrm{T}}_{\mathrm{i}\mathrm{t}}+{\sum }_{\mathrm{j}}{\mathsf{\theta }}_{\mathrm{j}}{\mathrm{x}}_{\mathrm{i}\mathrm{t}}^{\mathrm{j}})$$

  (9)

In this model, ${\mathrm{F}\mathrm{T}}_{\mathrm{i}\mathrm{t}},{\mathrm{G}\mathrm{F}}_{\mathrm{i}\mathrm{t}}$, and ${\mathrm{G}\mathrm{I}}_{\mathrm{i}\mathrm{t}}$ represent the indices for fintech, green finance, and regional green energy technology innovation, respectively; ${\mathrm{x}}_{\mathrm{i}\mathrm{t}}^{\mathrm{j}}$ denotes the set of control variables; ${\mathsf{\alpha }}_1,{\mathsf{\alpha }}_2$ represent the coefficients for the explanatory variable of green finance and financial technology, respectively; $\mathsf{\rho }$, $\mathsf{\lambda }$, and $\mathsf{\theta }$ represent the spatial autoregressive coefficient, spatial spillover effect coefficient of the disturbance term (${\mathsf{\nu }}_{\mathrm{i}\mathrm{t}}$), and spatial spillover effect coefficient of explanatory and control variables, respectively; ${\mathsf{\gamma }}_{\mathrm{t}}$ signifies time fixed effects and ${\mathsf{\mu }}_{\mathrm{i}\mathrm{t}}$ indicates spatial fixed effects; and W is the spatial weight matrix utilized in the spatial econometric model. Additionally, due to potential omitted variables or observational errors, the model includes a random error term, ${\mathsf{\epsilon }}_{\mathrm{i}\mathrm{t}}$.

Following the construction of the model, to examine the interaction effect of H5, we introduced the product of the green finance index and the financial technology index (${\mathrm{G}\mathrm{F}}_{\mathrm{i}\mathrm{t}}\times {\mathrm{F}\mathrm{T}}_{\mathrm{i}\mathrm{t}}$) as an interaction term into three sets of spatial econometric models, resulting in models 6–8.

• Model 6:

  $${\mathrm{G}\mathrm{I}}_{\mathrm{i}\mathrm{t}}=\mathsf{\rho }\mathrm{W}·{\mathrm{G}\mathrm{I}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\alpha }}_1{\mathrm{G}\mathrm{F}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\alpha }}_2{\mathrm{F}\mathrm{T}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\alpha }}_3{{\mathrm{G}\mathrm{F}}_{\mathrm{i}\mathrm{t}}\times \mathrm{F}\mathrm{T}}_{\mathrm{i}\mathrm{t}}+{\sum }_{\mathrm{j}}{\mathsf{\beta }}_{\mathrm{j}}{\mathrm{x}}_{\mathrm{i}\mathrm{t}}^{\mathrm{j}}+{\mathsf{\gamma }}_{\mathrm{t}}+{\mathsf{\mu }}_{\mathrm{i}\mathrm{t}}+{\mathsf{\epsilon }}_{\mathrm{i}\mathrm{t}}$$

  (10)

• Model 7:

  $${\mathrm{G}\mathrm{I}}_{\mathrm{i}\mathrm{t}}={\mathsf{\alpha }}_1{\mathrm{G}\mathrm{F}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\alpha }}_2{\mathrm{F}\mathrm{T}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\alpha }}_3{{\mathrm{G}\mathrm{F}}_{\mathrm{i}\mathrm{t}}\times \mathrm{F}\mathrm{T}}_{\mathrm{i}\mathrm{t}}+{\sum }_{\mathrm{j}}{\mathsf{\beta }}_{\mathrm{j}}{\mathrm{x}}_{\mathrm{i}\mathrm{t}}^{\mathrm{j}}+{\mathsf{\gamma }}_{\mathrm{t}}+\mathsf{\lambda }\mathrm{W}·{\mathsf{\nu }}_{\mathrm{i}\mathrm{t}}+{\mathsf{\mu }}_{\mathrm{i}\mathrm{t}}+{\mathsf{\epsilon }}_{\mathrm{i}\mathrm{t}}$$

  (11)

• Model 8:

  $${\mathrm{G}\mathrm{I}}_{\mathrm{i}\mathrm{t}}=\mathsf{\rho }\mathrm{W}·{\mathrm{G}\mathrm{I}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\alpha }}_1{\mathrm{G}\mathrm{F}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\alpha }}_2{\mathrm{F}\mathrm{T}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\alpha }}_3{{\mathrm{G}\mathrm{F}}_{\mathrm{i}\mathrm{t}}\times \mathrm{F}\mathrm{T}}_{\mathrm{i}\mathrm{t}}+{\sum }_{\mathrm{j}}{\mathsf{\beta }}_{\mathrm{j}}{\mathrm{x}}_{\mathrm{i}\mathrm{t}}^{\mathrm{j}}+\mathsf{\theta }\mathrm{W}\left({\mathrm{G}\mathrm{F}}_{\mathrm{i}\mathrm{t}}+{\mathrm{F}\mathrm{T}}_{\mathrm{i}\mathrm{t}}+{\mathrm{G}\mathrm{F}}_{\mathrm{i}\mathrm{t}}\times {\mathrm{F}\mathrm{T}}_{\mathrm{i}\mathrm{t}}+\sum \mathsf{\beta }{\mathrm{X}}_{\mathrm{i}\mathrm{t}}\right)+{\mathsf{\gamma }}_{\mathrm{t}}+{\mathsf{\mu }}_{\mathrm{i}\mathrm{t}}+{\mathsf{\epsilon }}_{\mathrm{i}\mathrm{t}}$$

  (12)

4.1.3. Panel Threshold Model Construction

To verify the nonlinear impact of green finance on regional green technological innovation in the context of financial technology according to H6, we adopted financial technology as the threshold variable and green finance as the core explanatory variable to construct a panel threshold model based on Hansen’s threshold effect theory [65].

• Model 9:

  $${\mathrm{G}\mathrm{I}}_{\mathrm{i}\mathrm{t}}={\mathsf{\mu }}_{\mathrm{i}}+{\mathsf{\alpha }}_1{\mathrm{G}\mathrm{F}}_{\mathrm{i}\mathrm{t}}\times \mathrm{I}\mathrm{(}{\mathrm{F}\mathrm{T}}_{\mathrm{i}\mathrm{t}}\le \mathsf{\gamma }\mathrm{)}+{\mathsf{\alpha }}_2{\mathrm{G}\mathrm{F}}_{\mathrm{i}\mathrm{t}}\times \mathrm{I}\mathrm{(}{\mathrm{F}\mathrm{T}}_{\mathrm{i}\mathrm{t}}>\mathsf{\gamma }\mathrm{)}+{\sum }_{\mathrm{j}}{\mathsf{\beta }}_{\mathrm{j}}{\mathrm{x}}_{\mathrm{i}\mathrm{t}}^{\mathrm{j}}+{\mathsf{\epsilon }}_{\mathrm{i}\mathrm{t}}$$

  (13)

In the model, ${\mathsf{\mu }}_{\mathrm{i}}$ represents the intercept, $\mathrm{I}$ is the indicator function, $\mathsf{\gamma }$ is the threshold value, ${\mathsf{\alpha }}_1$ denotes the coefficient when the threshold variable is less than the threshold value $({\mathrm{F}\mathrm{T}}_{\mathrm{i}\mathrm{t}}\le \mathsf{\gamma })$, ${\mathsf{\alpha }}_2$ is the coefficient when the threshold variable is more than the threshold value $({\mathrm{F}\mathrm{T}}_{\mathrm{i}\mathrm{t}}>\mathsf{\gamma })$, and the rest of the variables are the same as in models 4–6.

In Figure 1 below, we clearly list the corresponding relationship between the theoretical mechanism and the model established in this paper.

4.2. Variable Interpretation and Data Sources

4.2.1. Interpreted Variable

Regional Green Energy Technology Innovation (GI)

This paper employs the number of green patent applications (in ten thousand units) per capita from 2009 to 2021 in 266 cities across the nation (including municipalities directly under the central government and prefecture-level cities) as the dependent variable for the regional green energy technology innovation (GI) index. Green patent data can be obtained by querying the Incopat patent database.

4.2.2. Interpret Variables

Green Finance (GF)

According to the People’s Bank of China, the preliminary formation of China’s green financial system primarily consists of seven components: green credit, green investment, green insurance, green bonds, green support, green funds, and green equity. Building upon this foundation, we constructed measurement indicators for the green finance (GF) index, as presented in

Table 1. Green financial measurement indicator system.

Level I Indicators	Characterization Indicators	Description of Indicators	Indicator Attributes
Green Credit	The Proportion of Credit for Environmental Projects	Total Provincial Environmental Project Credit/Total Provincial Credit	+
Green Investment	The Proportion of Environmental Pollution Control Investment to GDP	Environmental Pollution Control Investment/GDP	+
Green Insurance	Level of Promotion of Environmental Liability Insurance	Income from Environmental Liability Insurance/Total Premium Income	+
Green Bonds	Level of Development of Green Bonds	Total Issuance of Green Bonds/Total Issuance of All Bonds	+
Green Support	Proportion of Fiscal Expenditure on Environmental Protection	Fiscal Expenditure on Environmental Protection/General Fiscal Budget Expenditure	+
Green Fund	Proportion of Green Fund	Total Market Value of Green Funds/Total Market Value of All Funds	+
Green Equity	Depth of Development of Green Equity	Carbon Trading, Energy Rights Trading, Emission Rights Trading/Total Equity Market Trading Volume	+

. Data for these measurement indicators were sourced from authoritative institutions such as the National Bureau of Statistics, the Ministry of Science and Technology, and the People’s Bank of China, and various authoritative statistical yearbooks, including China Science and Technology Statistical Yearbook, China Energy Statistical Yearbook, China Financial Yearbook, China Agricultural Statistical Yearbook, and China Industrial Statistical Yearbook, among others.

We employed the entropy method to objectively assign weights to the indicators to avoid the subjectivity inherent in traditional expert-weighted scoring. When a single indicator affects the index system positively or negatively, we used the following formulas for standardization.

$${\mathrm{x}}_{\mathrm{i}\mathrm{j}}=\frac{{\mathrm{x}}_{\mathrm{i}\mathrm{j}}-\mathrm{m}\mathrm{i}\mathrm{n}\left\{{\mathrm{x}}_{\mathrm{j}}\right\}}{\mathrm{m}\mathrm{a}\mathrm{x}\left\{{\mathrm{x}}_{\mathrm{j}}\right\}-\mathrm{m}\mathrm{i}\mathrm{n}\left\{{\mathrm{x}}_{\mathrm{j}}\right\}}$$

(14)

$${\mathrm{x}}_{\mathrm{i}\mathrm{j}}=\frac{\mathrm{m}\mathrm{a}\mathrm{x}\left\{{\mathrm{x}}_{\mathrm{j}}\right\}-{\mathrm{x}}_{\mathrm{i}\mathrm{j}}}{\mathrm{m}\mathrm{a}\mathrm{x}\left\{{\mathrm{x}}_{\mathrm{j}}\right\}-\mathrm{m}\mathrm{i}\mathrm{n}\left\{{\mathrm{x}}_{\mathrm{j}}\right\}}$$

(15)

In the formula, ${\mathrm{x}}_{\mathrm{i}\mathrm{j}}$ represents the value after the normalization of indicator data, and $\mathrm{m}\mathrm{a}\mathrm{x}\left\{{\mathrm{x}}_{\mathrm{j}}\right\}$ and $\mathrm{m}\mathrm{i}\mathrm{n}\left\{{\mathrm{x}}_{\mathrm{j}}\right\}$ represent the maximum and minimum values of the indicator data, respectively.

$${\mathsf{\lambda }}_{\mathrm{i}\mathrm{j}}=\frac{{\mathrm{x}}_{\mathrm{i}\mathrm{j}}}{\sum _{\mathrm{i}=1}^{\mathrm{m}}{\mathrm{x}}_{\mathrm{i}\mathrm{j}}}$$

(16)

$${\mathrm{e}}_{\mathrm{j}}=-\frac{1}{\ln \mathrm{m}}\sum _{\mathrm{i}=1}^{\mathrm{m}}\left({\mathsf{\lambda }}_{\mathrm{i}\mathrm{j}}\times \ln {\mathsf{\lambda }}_{\mathrm{i}\mathrm{j}}\right),{0\le \mathrm{e}}_{\mathrm{j}}\le 1$$

(17)

$${\mathrm{w}}_{\mathrm{j}}=\frac{1-{\mathrm{e}}_{\mathrm{j}}}{\sum _{\mathrm{j}=1}^{\mathrm{m}}(1-{\mathrm{e}}_{\mathrm{j}})}$$

(18)

Financial Technology (FT)

In this paper, the financial technology (FT) index is one of the core explanatory variables in models 1–6 as well as the threshold variable in model 7. Following the approach of Li et al. [66], we measured the FT for each of the 266 Chinese cities from 2009 to 2021 using Baidu’s advanced news search terms related to financial technology. We amalgamated the research of Song et al. [67], Li et al. [68], and other scholars to establish the search terms, as presented in

Table 2. Financial technology search term library.

Variable

Financial Technology Index

Search Term

EB-level Storage, NFC Payments (Near Field Communication Payments), Cloud Computing, Internet Finance, Artificial Intelligence (AI), Billion-level Concurrency, Cyber–Physical Systems, In-memory Computing, Distributed Computing, Blockchain, Business Intelligence (BI), Image Understanding, Graph Computing, Multi-party Secure Computing, Big Data, Differential Privacy Technology, Open Banking, Heterogeneous Data, Credit Investigation/Credit Rating, Investment Decision Support Systems, Digital Currency, Data Visualization, Data Mining, Text Mining, Intelligent Customer Service, Robo-advisors, Intelligent Data Analysis, Smart Financial Contracts, Machine Learning, Stream Computing, Deep Learning, Internet of Things (IoT), Biometric Technology, Mobile Internet, Mobile Payments, Third-party Payments, Neuromorphic Computing, Green Computing, Payment and Settlement Network, Equity Crowdfunding, Natural Language Processing (NLP), Virtual Reality (VR), Integrated Architecture, Cognitive Computing, Semantic Search, Speech Recognition, Identity Verification, Quantitative Finance

. The total search volume of these terms plus one, and then the logarithmic value, served as a proxy variable for the financial technology index.

4.2.3. Control Variables

In order to obtain effective estimation results and avoid the endogeneity problem caused by omitted variables, we controlled for the influence of the other relevant variables. Among these, the environmental development (ER) index uses the ratio of industrial smoke and dust emissions to GDP as its negative indicator. The cc (FD) is represented by the ratio of a city’s year-end loan balance to its real GDP. The technological investment index (TI) describes the percentage of a city’s financial technology support relative to its GDP. The industrial structure (IS) index is calculated as the ratio of the value added from the tertiary industry to the secondary industry. The China City Statistical Yearbook covering 2009–2021 was the source of the above data. The descriptive statistical results of the variables are presented in

Table 3. Descriptive statistics of controlled variables.

Variable	Meaning	Mean	Standard Deviation	Minimum	Maximum
FD	Financial Development Index	0.940	0.592	0.112	7.45
ER	Environmental Development Index	0.002	0.021	8.7 × 10−11	1.0001
TI	Technological Investment Index	0.195	0.044	0.02	0.387
IS	Industrial Structure Index	0.956	0.509	0.139	5.297
Sample	N = 3458
Period	2009–2021

.

4.2.4. Spatial Weight Matrix

Spatial weight matrices are used to describe relationships between regions and serve as a means to determine spatial dependency in spatial econometric models. Based on how spatial weight matrices are constructed, they can be classified into three major types: (1) adjacency spatial weight matrix, in which the adjacency between two regions is represented as 1; (2) distance-based spatial weight matrix, in which the weight matrix is determined by taking the reciprocal of the length between two regions; and (3) socioeconomic structure-based spatial weight matrix, in which the reciprocal of the absolute difference in per capita GDP between two regions is used as the weight matrix.

In the mechanism analysis of this study, the spatial effects of economic variables need to be discussed based on both the economic and geographic relationships between regions. Therefore, we constructed spatial weight matrices for economic distance, adjacency, geographical distance, and economic–geographic nestedness for empirical analysis. The matrices are as follows:

$${\mathrm{W}}^{1(2,3,4)}=\left[\begin{array}{ccc}\begin{array}{cc}1& {\mathrm{w}}_{12}\\ {\mathrm{w}}_{21}& 1\end{array}& \cdots & \begin{array}{c}{\mathrm{w}}_{1\mathrm{N}}\\ {\mathrm{w}}_{2\mathrm{N}}\end{array}\\ ⋮& \ddots & ⋮\\ \begin{array}{cc}{\mathrm{w}}_{\mathrm{N}1}& {\mathrm{w}}_{\mathrm{N}2}\end{array}& \cdots & 1\end{array}\right]$$

(19)

$${\mathrm{w}}_{\mathrm{i}\mathsf{\theta }}^1=\left\{\begin{array}{c}\frac{1}{|{\mathrm{y}}_{\mathrm{i}}-{\mathrm{y}}_{\mathsf{\theta }}|} \mathrm{i}\ne \mathsf{\theta }\\ 1 \mathrm{i}=\mathsf{\theta }\end{array}\right.$$

(20)

$${\mathrm{w}}_{\mathrm{i}\mathsf{\theta }}^2=\left\{\begin{array}{c}1 \mathrm{i} \mathrm{a}\mathrm{n}\mathrm{d} \mathsf{\theta } \mathrm{a}\mathrm{r}\mathrm{e} \mathrm{a}\mathrm{d}\mathrm{j}\mathrm{a}\mathrm{c}\mathrm{e}\mathrm{n}\mathrm{t}\\ 0 \mathrm{i} \mathrm{a}\mathrm{n}\mathrm{d} \mathsf{\theta } \mathrm{a}\mathrm{r}\mathrm{e} \mathrm{n}\mathrm{o}\mathrm{t} \mathrm{a}\mathrm{d}\mathrm{j}\mathrm{a}\mathrm{c}\mathrm{e}\mathrm{n}\mathrm{t}\end{array}\right.$$

(21)

$${\mathrm{w}}_{\mathrm{i}\mathsf{\theta }}^3=\left\{\begin{array}{c}\frac{1}{{\mathrm{d}}_{\mathrm{i}\mathsf{\theta }}} \mathrm{i}\ne \mathsf{\theta }\\ 0 \mathrm{i}=\mathsf{\theta }\end{array}\right.$$

(22)

$${\mathrm{W}}^{1(\mathrm{2,3},4)}=\left\{\begin{array}{c}{\mathrm{w}}^2·\mathrm{d}\mathrm{i}\mathrm{a}\mathrm{g}\left(\frac{{\overline{\mathrm{y}}}_1}{\overline{\mathrm{y}}}\frac{{\overline{\mathrm{y}}}_2}{\overline{\mathrm{y}}}\frac{{\overline{\mathrm{y}}}_3}{\overline{\mathrm{y}}}\dots \frac{{\overline{\mathrm{y}}}_{\mathrm{N}}}{\overline{\mathrm{y}}}\right) \mathrm{i} \mathrm{a}\mathrm{n}\mathrm{d} \mathsf{\theta } \mathrm{a}\mathrm{r}\mathrm{e} \mathrm{a}\mathrm{d}\mathrm{j}\mathrm{a}\mathrm{c}\mathrm{e}\mathrm{n}\mathrm{t}\\ 0 \mathrm{i} \mathrm{a}\mathrm{n}\mathrm{d} \mathsf{\theta } \mathrm{a}\mathrm{r}\mathrm{e} \mathrm{n}\mathrm{o}\mathrm{t} \mathrm{a}\mathrm{d}\mathrm{j}\mathrm{a}\mathrm{c}\mathrm{e}\mathrm{n}\mathrm{t}\end{array}\right.$$

(23)

In the equation above, ${\mathrm{w}}_{\mathrm{i}\mathsf{\theta }}^1$ represents the economic spatial weight matrix, which can be used to represent the relationship and economic distance between regions; ${\mathrm{w}}_{\mathrm{i}\mathsf{\theta }}^2$ and ${\mathrm{w}}_{\mathrm{i}\mathsf{\theta }}^3$ denote adjacency and geographic distance spatial weight matrices, respectively, which indicate whether geographic units are adjacent and the proximity between regions. Meanwhile, the economic–geographic nested spatial weight matrix ${\mathrm{W}}^{1(\mathrm{2,3},4)}$ comprehensively illustrates the coupling of economic and geographic aspects between regions. Here, ${\mathrm{d}}_{\mathrm{i}\mathsf{\theta }}$ denotes the central geographic distance between regions i and $\mathsf{\theta }$, ${\mathrm{y}}_{\mathrm{i}}$, and ${\mathrm{y}}_{\mathsf{\theta }}$ represent the average actual GDP of a city from 2009 to 2021 (a span of 13 years), while $\overline{\mathrm{y}}$ and ${\overline{\mathrm{y}}}_{\mathrm{N}}$ represent the national average actual GDP between 2009 and 2021 and the average actual GDP of the city NN.

Because of space limitations, in this paper, we first use the economic–geographic nested spatial weight matrix to identify, evaluate, and choose spatial econometric models, as well as decompose spatial spillover effects. Subsequently, other spatial weight matrices are introduced to verify the heterogeneity of spatial spillover effects generated by economic variables based on different spatial weight matrices.

5. Empirical Results and Analysis

5.1. Baseline Regression Analysis of Double Machine Learning Models

To validate H1 and H3, with reference to Zhang and Li, we conducted an empirical analysis using a double machine learning model based on the random forest algorithm. The sample splitting ratio for machine learning was set at 1:4.

Table 4. Parameter estimation outcomes of the double machine learning model.

	Model 1	Model 2
	GI	GI
GF	0.945 ***
	(3.87)
FT		0.537 ***
		(3.81)
_cons	0.101 ***	0.0515 **
	(4.66)	(2.43)
Covariate	yes	yes
Area fixed effect	yes	yes
Time fixed effect	yes	yes
n	3458	3458
R2	-	-

t statistics in parentheses, * p < 0.10, ** p < 0.05, *** p < 0.01.

reports the parameter estimate results for models 1 and 2; the coefficients for GF in model 1 and FT in model 2 are 0.945 and 0.537, both passing the significance test at the 1% level. This outcome indicates that green finance can exert a significant positive influence on regional green energy technological innovation, and financial technology also positively contributes.

5.2. Spatial Econometric Model Analysis

5.2.1. Spatial Dependence Test

Before conducting regression analysis using spatial econometric models, it is essential to examine whether the dependent variable (regional green energy technological innovation) exhibits spatial dependence. The global Moran’s I can be used to measure the spatial dependence of economic variables. Based on the method proposed by Moran, we initially constructed the global Moran’s I using the economic–geographic nested spatial weight matrix:

$$\mathrm{M}\mathrm{o}\mathrm{r}\mathrm{a}\mathrm{n}\mathrm{’}\mathrm{s} \mathrm{I}=\frac{\sum _{\mathrm{i}=1}^{\mathrm{N}}\sum _{\mathrm{j}=1}^{\mathrm{N}}{\mathsf{\omega }}_{\mathrm{i}\mathrm{j}}\left({\mathrm{x}}_{\mathrm{i}}-\overline{\mathrm{x}}\right)\left({\mathrm{x}}_{\mathrm{j}}-\overline{\mathrm{x}}\right)}{{\mathrm{s}}^2\sum _{\mathrm{i}=1}^{\mathrm{N}}\sum _{\mathrm{j}=1}^{\mathrm{N}}{\mathsf{\omega }}_{\mathrm{i}\mathrm{j}}},$$

(24)

$$\overline{\mathrm{x}}=\frac{\sum _{\mathrm{i}=1}^{\mathrm{N}}{\mathrm{x}}_{\mathrm{i}}}{\mathrm{N}},$$

(25)

$${\mathrm{s}}^2=\frac{\sum _{\mathrm{i}=1}^{\mathrm{N}}{({\mathrm{x}}_{\mathrm{i}}-\overline{\mathrm{x}})}^2}{\mathrm{N}}$$

(26)

Next, in order to assess the significance of the results (shown in

Table 5. Global Moran’s I of the GI Index in 2009–2021.

Year	Moran’s I	p-Value	Year	Moran’s I	p-Value
2009	0.479 ***	0.000	2016	0.570 ***	0.000
2010	0.497 ***	0.000	2017	0.560 ***	0.000
2011	0.471 ***	0.000	2018	0.528 ***	0.000
2012	0.529 ***	0.000	2019	0.526 ***	0.000
2013	0.561 ***	0.000	2020	0.554 ***	0.000
2014	0.525 ***	0.000	2021	0.545 ***	0.000
2015	0.536 ***	0.000

*** p < 0.01.

and Figure 2), it is necessary to transform Moran’s I into a z-statistic. As shown in Table 5, the global Moran’s I for the regional green energy technological innovation index every year from 2009 to 2021 is consistently positive, passing the significance test at the 1% level. Consequently, based on the economic–geographic nested spatial weight matrix, we concluded that the regional green energy technological innovation index exhibits a positive spatial correlation among various cities in China. Additionally, as shown in Figure 2, most Moran scatter points are concentrated in the first and third quadrants, indicating significant high–high or low–low clustering, affirming the significant spatial dependence of GI.

5.2.2. Model Suitability Test

We constructed a spatial autoregressive (SAR) model, spatial error model (SEM), and spatial Durbin model (SDM) for spatial econometric analysis. To select a suitable model for empirical analysis, we initially performed a suitability test of these models based on the economic–geographic nested spatial weight matrix and the results are presented in

Table 6. Model selection and testing.

Test Items	Test Value	p-Value
Robust LM test (spatial lag)	62.921 ***	0.000
Robust LM test (spatial error)	374.751 ***	0.000
Hausman test	53.56 ***	0.0000
LR test for Time	2876.25 ***	0.0000
LR test for Ind	35.31 ***	0.0001
Wald test for SAR	265.68 ***	0.0000
Wald test for SEM	241.98 ***	0.0000
LR test for SAR	262.68 ***	0.0000
LR test for SEM	244.83 ***	0.0000

* p < 0.10, ** p < 0.05, *** p < 0.01.

. The test was mainly performed using LM statistics and their robust forms (Robust LM). The outcomes indicate that both the SAR and the SEM passed the significance test at the 1% level, suggesting that the SDM, which accounts for both spatial autoregression and spatial error effects, should be employed within the framework of the economic–geographic nested spatial weight matrix.

After choosing the spatial Durbin model, the Hausman test was performed. The results (Hausman = 53.56, p = 0.0000) rejected the hypothesis that random effects are employed, thus advocating for fixed effects estimation. Subsequently, an LR test demonstrated the significant superiority of the double time–space fixed effects over both regional fixed effects (LR = 35.31, p = 0.0001) and temporal fixed effects (LR = 2876.25, p = 0.0000). The results of these tests point to the selection of the spatial Durbin model with double time–space fixed effects for empirical investigation.

Post-model selection, we further verified whether the chosen spatial Durbin model would degrade into either SAR or SEM through Wald and LR tests. The tests concerning the SAR model (Wald = 265.68, p = 0.0000; LR = 262.68, p = 0.0000) and the SEM (Wald = 241.98, p = 0.0000; LR = 244.83, p = 0.0000) passed the significance test at the 1% level, indicating that the spatial Durbin model would not degenerate into either of the other models. Therefore, based on the economic–geographic nested spatial weight matrix, we finally opted for the spatial Durbin model with double time–space fixed effects for empirical analysis.

5.2.3. Analysis of Baseline Regression and Moderating Effects

After establishing the specific form of the spatial econometric model, we conducted parameter estimation for models 5 and 8 to verify the related hypotheses, H1–H5. Models 5 and 8 were decomposed into local effects and spatial spillover effects to examine whether the related mechanism assumptions would hold. Local effects were used to represent the impact of the explanatory variables on the explained variables within the same locality, while spatial spillover effects indicated the influence of explanatory variables from other regions on the explained variables within a local area (specific parameter estimation results are detailed in

Table 7. Parameter estimation results of model 5 and model 8.

	Model 5			Model 8
	GI			GI
	Local Effect	Spillover Effect	Total Effect	Local Effect	Spillover Effect	Total Effect
GF	0.697 **	9.494 **	10.19 **	0.596 *	7.190 *	7.786 *
	(2.07)	(2.00)	(2.05)	(1.85)	(1.80)	(1.86)
FT	0.362 ***	1.684 ***	2.047 ***	−0.0867	−6.264 ***	−6.350 ***
	(8.00)	(3.31)	(3.83)	(−0.56)	(−2.69)	(−2.63)
GF×FT				1.000 ***	17.42 ***	18.42 ***
				(3.05)	(3.48)	(3.55)
IS	0.294 ***	−4.091 ***	−3.798 ***	0.283 ***	−3.905 ***	−3.621 ***
	(3.18)	(−3.64)	(−3.26)	(3.10)	(−3.72)	(−3.35)
FD	0.274 ***	−1.534 *	−1.260	0.276 ***	−1.555 *	−1.279
	(4.12)	(−1.85)	(−1.48)	(4.09)	(−1.86)	(−1.49)
ER	−0.869	173.7 ***	172.8 ***	−0.127	171.8 ***	171.7 ***
	(−0.82)	(8.22)	(7.95)	(−0.12)	(8.48)	(8.26)
TI	7.819 ***	63.00 ***	70.82 ***	7.900 ***	60.03 ***	67.93 ***
	(8.66)	(5.29)	(5.72)	(8.59)	(5.66)	(6.17)
Area fixed effect	yes	yes	yes	yes	yes	yes
Time fixed effect	yes	yes	yes	yes	yes	yes
Spatial
rho	0.813 ***			0.804 ***
	(45.01)			(43.81)
Variance
sigma2_e	0.999 ***			0.998 ***
	(40.58)			(40.57)
N	3458			3458
R2	0.057			0.011

t statistics in parentheses, * p < 0.10, ** p < 0.05, *** p < 0.01.

). The spatial autoregressive coefficient in models 5 and 8 was 0.813 and 0.804, respectively, passing the 1% significance test and indicating that regional green energy technology innovation could foster synergistic inter-regional development.

The parameter estimation for model 5 revealed that the local effect coefficient for the green finance (GF) and financial technology (FT) indices was 0.697 and 0.362, respectively, passing the significance test at 5% and 1%. This indicates significant positive local effects on regional green energy technology innovation for both variables, supporting H1 and H3. Meanwhile, the spatial spillover effect coefficient for GF was 9.494 at the 5% significance level, indicating that the support of green finance in technological innovation projects or enterprises drives its own development and generates positive spillover effects in other regions, confirming H2. The spatial spillover effect coefficient for FT was 1.684 at a 1% significance level, suggesting that financial technology, represented by emerging technologies such as artificial intelligence, big data, cloud computing, and blockchain within an economic–geographic embedded spatial weight matrix, significantly contributes to the spatial transmission of regional green technology innovation, supporting H4.

Model 8, built upon model 5, introduced an interaction term between the GF and FT indices to investigate the moderating mechanism of financial technology in the path of green finance’s impact on regional green technology innovation. The local effect coefficient for the interaction term ${\mathrm{G}\mathrm{F}}_{\mathrm{i}\mathrm{t}}\times {\mathrm{F}\mathrm{T}}_{\mathrm{i}\mathrm{t}}$ was 1.000, passing the 1% significance test and indicating that financial technology can enhance the positive impact of green finance on regional green energy technology innovation, confirming H5. Furthermore, based on an economic–geographic embedded spatial weight matrix, the spatial spillover effect coefficient, with a value of 17.42, was significant at the 1% level. This result indicates that financial technology can adjust and mitigate geographic polarization effects arising from differences in economic development, financial service coverage, technology, and human resources related to green finance. This might be because financial technology plays a part in providing digital financial services, reducing transaction costs for green finance, and improving the accessibility of financial services in remote areas, thereby fostering a trickle-down effect and enabling synergistic development among regions.

5.2.4. Heterogeneity Analysis Based on Spatial Weight Matrix

The parameter estimation results of the spatial Durbin model mentioned above were derived based on an economic–geographical embedded spatial weight matrix. These results comprehensively consider the heterogeneity and correlations in both geographical and economic aspects among cities. The spatial weight matrix based on geographical distances describes the geographical relationships between regions, while the economic spatial weight matrix is derived from the economic relationships between regions. As different spatial weight matrices consider different interregional associations, we conducted parameter estimation using the economic and geographical distance spatial weight matrices separately. Further analysis of their heterogeneity is presented in

Table 8. Parameter estimation results under other spatial weight matrices.

	Economic Spatial Weight Matrix			Geographical Distance Spatial Weight Matrix
	Local Effect	Spillover Effect	Total Effect	Local Effect	Spillover Effect	Total Effect
GF	0.772 **	1.439	2.211	0.385	−6.372 *	−5.986 *
	(2.04)	(0.71)	(0.98)	(0.99)	(−1.75)	(−1.73)
FT	0.641 ***	2.627 ***	3.268 ***	0.735 ***	−2.635 ***	−1.899 ***
	(12.17)	(10.80)	(12.04)	(13.50)	(−6.54)	(−4.78)
IS	−0.0683	−1.676 ***	−1.745 ***	0.429 ***	−0.557	−0.128
	(−0.63)	(−2.96)	(−2.81)	(3.81)	(−0.84)	(−0.20)
FD	0.312 ***	−0.165	0.147	0.478 ***	−0.797	−0.318
	(4.07)	(−0.48)	(0.39)	(5.92)	(−1.32)	(−0.54)
ER	−2.957 **	37.50 ***	34.54 ***	0.205	−36.81 ***	−36.60 ***
	(−2.48)	(5.20)	(4.37)	(0.17)	(−3.81)	(−3.96)
TI	12.17 ***	39.00 ***	51.18 ***	12.85 ***	3.410	16.26 ***
	(11.67)	(6.93)	(8.22)	(12.17)	(0.58)	(2.84)
Area fixed effect	yes	yes	yes	yes	yes	yes
Time fixed effect	yes	yes	yes	yes	yes	yes
Spatial
rho	0.575 ***			2.503 ***
	(27.71)			(58.55)
Variance
sigma2_e	1.428 ***			1.537 ***
	(40.80)			(41.49)
N	3458			3458
R2	0.153			0.045

t statistics in parentheses, * p < 0.10, ** p < 0.05, *** p < 0.01.

.

According to Table 8, the local impact coefficient of the green finance (GF) index is 0.772 using the economic spatial weight matrix, which meets the 1% threshold for significance. This suggests that by considering economic relationships, green finance also promotes regional green energy technological innovation. However, the spatial lag effect coefficient of GF is insignificant; thus, determining the mechanism of spatial spillover effects of green finance from an economic relationship perspective is inconclusive. Furthermore, fintech still has a positive local and geographical spillover effect on regional green energy technological innovation, with values of 0.641 and 2.627, respectively, passing the significance test at the 1% level.

Under the geographical distance spatial weight matrix, the local effect coefficient of GF is insignificant, making the mechanism of green finance’s effect on local green energy technological innovation unclear. The spatial lag effect coefficient of GF, with a value of −6.372, is significantly negative at the 10% level. This suggests that from a geographical perspective, whether green finance generates a negative spatial spillover effect on regional green energy technological innovation is unclear. This could be due to the rapid development of green finance in some economically advanced areas, which accelerates local green energy technological innovation, causing a concentration of related talents and technologies and significant capital accumulation, resulting in a “siphoning” effect that negatively impacts the development of surrounding areas and exacerbates regional imbalances through the Matthew effect. Moreover, the local effect coefficient of the FT index is 0.735, passing the significance test at the 1% level. However, the spatial lag effect coefficient, with a value of −2.635, is significantly negative at the 1% level. This indicates that from a geographical relationship standpoint, financial technology generates a negative spatial spillover effect on regional green energy technological innovation. This might be due to financial technology companies in certain regions establishing a competitive advantage through technological innovation and data accumulation, creating a technological moat. This advantage could hinder the development of financial technology in neighboring areas, as other financial institutions and innovative companies might struggle to compete.

5.2.5. Robustness Test

This study’s sample included four municipalities directly under the central government in China: Beijing, Tianjin, Shanghai, and Chongqing. They possess a distinct status within China’s administrative framework. These cities have a developed financial infrastructure, well-established financial market systems, and more significant fiscal resources, leading to higher levels of financial market activity and stronger fiscal decision-making capability. This uniqueness might have introduced interference in the study’s findings.

5.3. Empirical Analysis of Panel Threshold Models

5.3.1. Threshold Effect Test and Threshold Determination

Based on the empirical framework of green finance–financial technology–regional green energy technological innovation, financial technology was used in this study as the threshold variable in the panel threshold model construction to further examine the nonlinear impact of financial technology development on the relationship between green finance and regional green energy technological innovation. Before estimating the parameters, it was necessary to test the threshold effects of the model and determine the corresponding threshold value. Initially, we conducted a threshold effect estimation using bootstrap resampling (bootstrap = 300) to assess the presence and reasonableness of the threshold value, as detailed in

Table 9. Robustness test.

	GI
	Local Effect	Spillover Effect	Total Effect
GF	0.669 **	6.842 *	7.511 *
	(2.07)	(1.65)	(1.72)
FT	0.205 ***	1.670 ***	1.875 ***
	(4.10)	(3.03)	(3.22)
IS	0.033	−4.205 ***	−4.172 ***
	(0.36)	(−4.08)	(−3.91)
FD	0.316 ***	−1.412 *	−1.096
	(4.94)	(−1.91)	(−1.44)
ER	−1.252	160.193 ***	158.941 ***
	(−1.24)	(8.88)	(8.54)
TI	8.324 ***	48.269 ***	56.592 ***
	(9.57)	(7.05)	(5.29)
Area fixed effect	yes	yes	yes
Time fixed effect	yes	yes	yes
Spatial
rho	0.796 ***
	(41.89)
Variance
sigma2_e	0.934 ***
	(40.61)
N	3406
R2	0.035

t statistics in parentheses, * p < 0.10, ** p < 0.05, *** p < 0.01.

. The p-values for both single and double thresholds were significant, while the p-value for the triple threshold was 0.1367, which did not pass the significance test.

Consequently, it was determined that the regression analysis should employ a double threshold effect. Based on this determination, we estimated the threshold values using the double threshold, as presented in

Table 10. Threshold effect test and determination of threshold quantity.

Threshold Variable	Threshold of the Order	F-Value	p-Value	Critical Value
				10%	5%	1%
FT	A single threshold	239.10 ***	0.0000	40.8510	48.0127	69.6030
	Double threshold	92.96 ***	0.0000	31.9721	38.0686	51.3905
	Triple threshold	47.74	0.1367	55.7794	73.2682	106.2150

t statistics in parentheses, * p < 0.10, ** p < 0.05, *** p < 0.01.

and

Table 11. Estimation results of threshold value.

Threshold Variable	Threshold of the Order	Threshold Value	95% Confidence Interval	Bootstrap	Seed Value
FT	A single threshold	−0.1322 ***	[0.1860, 0.2462]	300	201
	Double threshold	0.2148 ***	[−0.1423, −0.1282]	300	201

t statistics in parentheses, * p < 0.10, ** p < 0.05, *** p < 0.01.

. The single threshold was −0.1322, and the double threshold was 0.2148, both significant at the 1% level.

5.3.2. Likelihood Ratio Test

By plotting the likelihood ratio function, it is possible to determine the specific threshold value corresponding to a likelihood ratio statistic (LR) of 0. Figure 3 displays the results of LR statistics, where the curve represents the LR values. In Figure 3a, when LR equals 0, the threshold variable is −0.1322. In Figure 3b, when LR equals 0, the threshold variable is 0.2148. Additionally, the confidence interval for −0.1322 is [0.1860, 0.2462], and for 0.2148, it is [−0.1423, −0.1282]. Both of these intervals fall within the critical likelihood values and are lower than the 5% significance level indicated by the red line on the LR statistic curve. These confidence intervals support the estimate of a true dual threshold since they fall within the acceptable range of the threshold model’s initial premise.

5.3.3. Parameter Estimation of Panel Threshold Model

After performing threshold effect tests and determining the threshold values, we performed parameter estimation for model 9, the results of which are presented in

Table 12. Parameter estimation results of the panel threshold model.

	Model 9
	Coefficient	t-Statistic
IS (industrial structure)	0.486	1.38
FD (financial development)	0.774 ***	3.26
ER	0.245	0.88
TI	12.91 ***	4.28
GF (FT < −0.1322)	0.660	1.34
GF (−0.1322 ≤ FT < 0.2148)	2.744 ***	4.11
GF (FT ≥ 0.2148)	5.512 ***	6.83
_cons	−3.134 ***	−3.98
N	3458
R2	0.274

t statistics in parentheses, * p < 0.10, ** p < 0.05, *** p < 0.01.

. As shown in Table 12, the dual-threshold effect of the threshold variable, financial technology (FT), segments the effect of green finance on regional green energy innovation into three stages. When the value of the threshold variable, the financial technology (FT) index, is less than the single threshold value (−0.1322), the coefficient of the green finance index (GF) is 0.660, which fails to pass the significance test. This suggests that green finance fails to significantly stimulate regional green technology innovation when the level of financial technology development is low. Capital flows are constrained within traditional financial systems, making it challenging to guide funding toward green energy technological innovation. At the same time, due to incomplete information and asymmetry, investors struggle to accurately assess the risks and returns of regional green energy technological innovation projects. However, when the value of the FT index approaches the single threshold (−0.1322) and lies between the dual thresholds (−0.1322 and 0.2148), the coefficient of GF is 2.744, which passes the 1% threshold for significance.

At this point, under the influence of financial technology, green finance significantly promotes regional green energy technological innovation. Financial technology harnesses emerging technologies such as artificial intelligence, big data, cloud computing, and blockchain, among others, to drive the development of green finance from multiple perspectives. It offers more flexible and convenient financing channels and supports information services, directing capital flows toward both the green and emerging technology industries. This, in turn, bolsters the development of green energy technological innovation. When the value of the FT index exceeds the dual threshold (0.2148), the GF coefficient is 5.512, which meets the 1% threshold for significance. This indicates that the continuous advancement of financial technology will enhance the role of green finance in promoting regional green energy technological innovation.

This outcome, derived from the research framework of green finance–financial technology–regional green energy technological innovation, demonstrates that green finance does not have a consistently positive impact on regional innovation, as suggested in traditional studies. Instead, it shows a nonlinear effect on regional green innovation, evolving with the development of financial technology, thus confirming Hypothesis 6.

6. Conclusions and Policy Suggestions

6.1. Conclusions

Using an economic–geographical embedded spatial weight matrix, we employed dual machine learning models, spatial econometric models, and panel threshold effect models to analyze the relationships between green finance, financial technology, and regional green energy technological innovation, which yielded the following conclusions:

First, green finance, which serves as specialized financing to address environmental pollution, resource wastage, and low utilization, facilitates sufficient funding for green technological innovation and green enterprises to cope with increased environmental regulations. The growth of innovative green technologies in an area is encouraged by this funding. In addition, financial technology reduces financing barriers, improves funding efficiency, enhances transparency in the use of green funds, and increases efficiency in green investments, empowering enterprises in their green innovation activities.

Second, the analysis of the spatial Durbin model shows that green finance and financial science and technology have a positive spatial spillover effect on regional green technological innovation. Green finance and financial science and technology can not only promote regional development of energy technology by providing capital and information services support for enterprises engaged in technological innovation but also drive increased investment, knowledge, and information diffusion, and specialized division of labor in the surrounding areas by extending financial services and industrial networks, thus enhancing its important role in regional green energy technological innovation.

Third, there is a considerable positive spatial interaction impact between green finance and financial technology on regional green innovation, as shown by the economic–geographical embedded spatial weight matrix. This indicates that financial technology enhances the positive local and spatial lag effects of green finance on regional green energy technological innovation. Specifically, fintech can utilize emerging technologies to fully solve the information asymmetry problem in green finance and improve service efficiency, thus indirectly contributing to regional energy technological innovation.

Fourth, the promotion effect of green finance on regional green energy technological innovation is nonlinear. The available research has rarely delved into its impact pathways. As determined by empirical analysis using panel threshold models, insufficient development of financial technology, to some extent, hinders the role of green finance in promoting regional green energy technological innovation. In the absence of financial technology, capital flows may be constrained by traditional financial systems, including high financing costs, cumbersome financing processes, and insufficient understanding of green projects by financing entities, thus limiting the advancement and development of related projects. With the advancement of financial technology, the positive impact of green finance on regional green energy technological innovation will be continuously strengthened.

6.2. Suggestions

Based on the above conclusions, we suggest the following policy recommendations.

Additional assistance for green finance and financial technological development: Government organizations should focus on providing assistance for the development of cutting-edge technologies, including virtual reality, cloud services, and 5G networks, in order to strengthen the advancement of green finance and financial technology. It is crucial to encourage innovators to integrate these emerging technologies into the innovative development of green finance and financial technology. Additionally, it is essential to provide essential elements and technical support for regional green energy technological innovation, especially in terms of alleviating funding constraints and addressing information asymmetry.

Continued advancement of urban green finance: Governments should continue promoting the development of green finance in various cities. This can be achieved by implementing policy incentives that encourage urban enterprises to allocate more funds to projects related to green energy technological innovation. Additionally, offering credit advantages and interest rate incentives to companies engaged in clean environmental projects while reducing credit availability and increasing interest rates for those involved in polluting projects could encourage a shift in their production processes. At the same time, it is necessary to establish a rigorous and scientific assessment system for green finance in order to monitor and counteract any “greenwashing” behavior prevalent in the capital markets.

The progress of financial technology relies on a robust technical foundation. Further strengthening the development of financial technology infrastructure could fundamentally advance the research and development of cutting-edge financial technology and promote high-quality development. Moreover, governments should pay attention to the developmental disparities in financial technology across different regions. Establishing an environmental information-sharing platform to bridge information gaps is crucial. This will provide a basis for financial institutions’ decision making and relevant policy formulation. At the same time, it is essential to focus on evolving trends in the development of financial technology, strengthen regulatory oversight, and enhance risk prevention measures in the realm of financial technology.

It is imperative to strengthen the linkage and coordination among green finance, financial technology, and regional green energy technological innovation. This involves refining the top-level design of financial technology to establish a solid foundation for empowering green finance. Currently, China’s financial technology development is in a phase of rapid growth, yet there are gaps in the relevant guiding policies and specific directives supporting financial technology for green finance. Therefore, there should be an emphasis on formulating policies, industry norms, and technical standards for financial technology. This will guide the systematic development of financial technology and lay a solid foundation for empowering green finance.

7. Research Contributions and Limitations

7.1. Objective and Contributions

In traditional research, finance supporting technological innovation is a classic topic. However, this paper argues that there are still some gaps in the existing studies. Currently, some theoretical studies have integrated green finance and financial technology into the framework of sustainable growth theory. However, the impact of these two subfields on green technological innovation has not been considered. There is scarce research that comprehensively analyzes financial technology, green finance, and regional green energy technological innovation within a unified research framework. Therefore, there is still a gap in the study of the underlying mechanisms. Relevant studies increasingly tend to report that green finance and financial technology play crucial roles in promoting environmentally friendly economic development and have positive effects on improving the ecological environment and fostering green innovation in businesses. However, there has not been an in-depth exploration of their impact mechanisms and pathways. Studies have overlooked the fact that the influence of green finance on regional green energy technological innovation is nonlinear. The facilitating role of green finance varies as the development of financial technology progresses through different stages.

To fill the gaps in existing research and enhance our understanding of the complex mechanisms between green finance, financial technology, and regional green energy technological innovation, this research employed a comprehensive analytical framework, utilizing dual machine learning models, spatial econometric models, and panel threshold effect models as a multi-layered approach to investigate regional green energy technological innovation.

This study is the first to integrate green finance, financial technology, and regional green energy technological innovation within a unified research framework, which offers a deeper observational perspective. On the one hand, by using a spatial weighting matrix embedded in economic geography, we thoroughly considered the interconnection between regions and delved into the positive spatial spillover effects of green finance and financial technology on regional green energy technological innovation. Such a spatial analytical framework allowed us to better comprehend how these two explanatory variables are related at the spatial level. On the other hand, based on mechanism analysis and empirical research, this study proposes policy recommendations for the collaborative advancement of green finance and financial technology to enhance green energy technological development and improve regional energy efficiency. These recommendations not only emphasize the active involvement of the financial system but also address policy adjustments at the industrial level, regional coordination, and overall regional development. They provide a detailed and actionable reference for government decision making in areas such as marketization, industrial policy guidance, regional coordination, and overall regional development.

7.2. Limitations and Future Recommendations

This study has the following limitations, which could potentially be enhanced and resolved by further study:

(1)

We conducted an empirical analysis of regional green energy technological innovation solely based on panel data in China. However, green energy technological innovation exhibits heterogeneity across different industries, spatiotemporal scopes, and subsystems in different segments of technology. Future research should further explore the impact mechanisms of green finance and financial technology on green energy technological innovation in other geographical regions and industrial sectors, and more detailed technological subdivisions.

(2)

Based on the findings of relevant research, we derived mechanism hypotheses through theoretical inference, which has a certain degree of persuasiveness. Future research should apply rigorous mathematical logic, along with theories such as game theory and complex network theory, to conduct a more in-depth analysis of the specific mechanisms of the research subject. Strengthening the application of mathematical language in mechanism analysis will enhance the robustness and credibility of this study’s mechanistic explanations.

(3)

Green energy technological innovation is an evolving dynamic process influenced by multiple factors, such as market demand and policy environment. This study only captured data from 2009 to 2021, and thus, it falls short of fully capturing subsequent technological developments and innovation trends, making it challenging to extrapolate to long-term universality. Subsequent research should adopt real-time monitoring and tracking methods for technological development to ensure continuous updates over time. Additionally, further investigations should be conducted on the roles of green finance, financial technology, and other factors in green energy technological innovation.