Digital Finance, Environmental Regulation, and Green Technology Innovation: An Empirical Study of 278 Cities in China

Abstract

Digital finance provides a premises guarantee for green technology innovation, and effective environmental regulation helps to achieve green and sustainable development. This article selects Chinese urban panel data from 2011 to 2019 to explore the impact mechanism of the influence of digital finance and environmental regulation on the innovation capacity of green science and technology. It is found that extensive financing channels and the strong information-matching ability of digital finance have a significant promoting effect on local green science and technology innovation. Moreover, government environmental regulation not only facilitates the development of green technology innovation locally and in nearby regions, but also strengthens the utility of digital finance in driving green science and technology innovation. Further research found that the influence of digital finance and environmental regulation on the ability of green science and technology innovation has regional heterogeneity, and only digital finance in Central China can promote green science and technology innovation in both local and adjacent areas. Therefore, the government should continue to promote the development of digital finance, optimize environmental regulations by increasing environmental protection subsidies and creating a green innovation environment, and further stimulate willingness to innovate green technologies. At the same time, it is also important to note the coordinated development and governance with neighboring regional governments.

1. Introduction

Since the reform and opening up, China has created a number of “China miracles”, but environmental protection has been neglected in the rush to boost economic growth. After entering the new normal, China’s economy has begun to aim for high-quality development. This goal sets higher requirements for economic progress and environmental protection. However, China ranked only 120th out of 180 countries in the 2020 Environmental Performance Index, illustrating the imbalance between high-quality economic development and environmental protection. In order to balance “economic performance” and “environmental performance”, since 2012, China has gradually introduced and improved laws and regulations to protect ecological civilization, aiming to promote green transformation through strict environmental regulations.

Traditional innovation only considers technological progress and economic development, while green technology innovation also needs to take into account ecological civilization. Therefore, in order to achieve green technology innovation, the main body of innovation will face higher technical standards, capital investment, financing costs, and risks [1]. In this way, in the process of realizing green technology innovation, good financial services are needed as a prerequisite, and at the same time, there should be correct guidance from the environmental regulations by the government [2]. However, it is difficult for traditional financial services to meet the capital needs of many enterprises for green technology innovation because of its high threshold, high cost, and low efficiency [3]. Therefore, digital finance based on the continuous development of digital technology has gradually caused concern in various circles. Digital finance can solve the capital problem in the process of green technology innovation by lowering the threshold, improving resource allocation, alleviating information asymmetry, reducing transaction costs, widening inter-regional overflow channels, and using other methods [2,4,5]. So, as a new way of financial services, can digital finance drive green technology innovation to achieve sustainable development? Is digital finance affected by government environmental regulation in the process of influencing green technology innovation? Is there a spatial spillover effect of digital finance on green technology innovation?

To solve the above problems, the mechanism studied in this paper covers digital finance, environmental regulation, and green technology innovation. Based on the panel data of 278 cities from 2011 to 2019, the spatial Durbin model was constructed to explore the impact of digital finance and environmental regulation on green technology innovation. The main research contents of this paper are as follows: first, the influence of digital finance and environmental regulation on green technology innovation is discussed. Second, the moderating role of environmental regulation in the process of digital finance affecting regional green technology innovation is explored; third, the marginal effect of digital finance and environmental regulation on green technology innovation is analyzed. Fourth, 278 cities are divided into seven regions: East China, South China, North China, Central China, Southwest China, Northwest China, and Northeast China. Based on the perspective of regional heterogeneity, the influence mechanism of digital finance and environmental regulation on green technology innovation is discussed in depth.

The rest of this paper is arranged as follows: Section 2 reviews the relevant literature. Section 3 proposes the research hypothesis. Section 4 introduces the model setting and data description. Section 5 presents an empirical analysis, robust testing, and regional-heterogeneity analysis and discussion. Section 6 puts forward research conclusions and relevant policy suggestions. Section 7 puts forward research limitations and prospects.

2. Literature Review

Green technology innovation is innovation activity with high investment, high risk and long cycle. To enhance the abilities of science and technology, good financial service is the key [6,7]. Digital finance is a new financial service model that integrates traditional financial industry with big data, internet, cloud computing and other information technologies [8,9]. The G20 Advanced Principles for Digital Inclusive Finance adopted at the G20 Hangzhou Summit in 2016 advocated the development of inclusive finance relying on digital technologies and included relevant indicators of digital finance into the evaluation system of inclusive finance, which greatly promoted the development of digital finance. At present, there have been abundant discussions on the relationship between digital finance and technological innovation in academic circles. At the micro level, Lin [10] points out that fintech can reduce the financing risk caused by information asymmetry on enterprise technological innovation. Subsequently, Tang et al. [11] took Chinese listed companies as research objects and believed that digital finance could promote the output of technological innovation by broadening financing channels and reducing financing costs. At the macro level, scholars have reached a relatively consistent conclusion that digital finance has the advantages of low threshold, wide coverage, low transaction cost and high resource allocation rate, which is of far-reaching significance to realize technological innovation [12,13]. For example, Nie et al. [14] selected the SYS-GMM model and found heterogeneity in the promotion effect of digital finance on regional technological innovation. Combined with the trickle-down effect, Xu [15] found that digital finance can also drive technological innovation in neighboring areas through spatial econometric model. In addition, some scholars also discussed the relationship between digital finance and green technology innovation. For example, Yu et al. [16] pointed out that digital finance can significantly promote green technology innovation on family farms and believed that promoting the development of digital finance is of great significance to the sustainable development of agriculture. Habiba et al. [17] took 12 major countries with carbon emissions as research objects and found that green technological innovation is a key factor in reducing carbon emissions and achieving sustainable development, and digital finance can effectively promote the progress of green technological innovation. When exploring the impact of digital finance on carbon emissions, Lee [18] concluded that green technology innovation plays an intermediary role.

Environmental regulations are related environmental laws and regulations formulated by the government for the purpose of protecting the environment, aiming to guide economic subjects to make decisions to improve the environment, reduce pollutant emissions while improving the overall economic benefits, and achieve the goal of the sustainable development of technology and the environment [19]. In the 1960s, neoclassical economic theory, based on the static perspective, pointed out that under the environmental supervision of the government, enterprises need to pay a large amount of environmental protection costs, which are bound to occupy the R&D funds originally used for innovation activities of enterprises, resulting in the innovation crowding-out effect. Therefore, environmental regulation will inhibit economic development [20,21]. Porter put forward a different point of view in 1991. Porter believed that with economic development, production technology and equipment of enterprises are constantly upgrading, and the key to environmental protection has shifted from process to result. Therefore, the environment in which enterprises are located should be regarded as dynamic and the impact of environmental regulation on economic development should be studied from a dynamic perspective. Therefore, based on the dynamic perspective, the Porter hypothesis is proposed. Porter [22], as well as Porter and Vander [23] believe that strict and effective environmental regulations can guide enterprises to voluntarily strengthen their investment in green technology R&D, enhance their competitive advantages, and achieve a win–win balance between economic performance and environmental performance. Since the porter hypothesis was put forward, scholars have continued to discuss the relationship between environmental regulation and green technology innovation. However, the opinion camp is always divided into three parts: the first side mainly believes that environmental regulation can effectively promote the ability of green technology innovation based on the Porter hypothesis [24,25]. Li et al. [26] pointed out that the financing availability of large enterprises is relatively high. Therefore, in the face of strict environmental regulations, enterprises will reduce environmental costs and improve resource utilization through green technology innovation, so as to enhance their competitive advantages and achieve sustainable development. Zhang et al. [27] studied 33 countries and concluded that environmental regulation has a significant incentive effect on green patent output. The second side, supported by neoclassical economic theory, holds that environmental supervision inhibits technological innovation ability [28,29]. Lanoie et al. [30] found that the benefits generated by enterprise green technology innovation could not cover the costs generated in the process of environmental compliance. Therefore, compared with green technological innovation with high investment, high risk and long cycles, enterprises are more inclined to pay the environmental penalty. Dechezleprêtre [31] believes that environmental costs caused by environmental regulations occupy the funds originally used for innovation activities of enterprises, thus hindering the development of green technological innovation of enterprises. The third party believes that there are preconditions for the relationship between the two and emphasizes the role of environmental regulation intensity, senior executives’ environmental awareness, regional economic development level, financing and other factors [32,33].

At the same time, many scholars also pay attention to the interactive relationship between digital finance and environmental regulation. For example, Shi et al. [34] points out that the synergy between digital finance and environmental regulation can effectively improve the degree of environmental pollution and play an important role in environmental governance. Li et al. [35] showed through the study of urban panel data that the interaction between digital finance and environmental regulation is conducive to the upgrading of urban industrial structure. Wang et al. [36] point out that digital finance cannot do without the regulatory role of government intervention in the process of promoting county economic growth. In addition, Feng et al. [37] took the intensity of regional environmental regulation as the threshold variable when exploring the relationship between digital finance and green technology innovation, and found that digital finance significantly promoted of green technology innovation only in regions with stricter environmental regulation.

According to the above literature, scholars have made many achievements in the research on the relationship between digital finance and technological innovation, environmental regulation and green technological innovation. However, if we place digital finance, environmental regulation and green technology innovation in a research framework, we can find that the existing research has three characteristics. First, the literature pays more attention to the influence of digital finance on technological innovation, and less attention is paid to the influence mechanism of digital finance on green technological innovation. Second, scholars have conducted preliminary discussions on the relationship between digital finance and environmental regulation, but the discussions are few and scattered, focusing on economic development and environmental governance. Third, existing studies mostly focus on spatial spillover effects of digital finance from the perspective of spatial independence.

Compared with the existing research, this study has three main contributions. Firstly, in terms of research perspective, this paper constructs a research framework of digital finance, environmental regulation and green technological innovation, in which environmental regulation is taken as a regulating variable to provide a perspective for the discussion of the significance of digital finance. Secondly, in terms of research methods, considering the flow of financial elements, the migration behavior of enterprises and the spillover effect of technological innovation, this paper chooses the spatial Durbin model to explore the interaction between regions from the perspective of spatial correlation, further enriching the empirical research on digital finance, environmental regulation and green technological innovation. Thirdly, in terms of practical significance, this paper studies the heterogeneous impact of digital finance and environmental regulation on green technology innovation according to geographical location, providing theoretical basis for the sustainable development of each region.

3. Research Hypotheses

As a high-risk, high-investment, and long-cycle activity, green technology innovation is prone to being restricted by financing problems during its development [16,38,39]. In order to realize the improvement of green technology innovation ability, a large amount of capital is needed to support it [40]. However, the problems of traditional finance, such as information asymmetry, high threshold, and low service efficiency, all lead to its poor inclusiveness and difficulty in effectively alleviating financing difficulties [41]. Therefore, with the integration of information technology, digital finance with strong universality is gradually becoming known by all circles. Digital finance can increase the possibility of obtaining financing through a variety of ways, promote R&D investment, and strengthen green technology innovation so as to achieve high-quality economic development [42,43]. On the one hand, digital finance absorbs investors that are “large, small and scattered” in the market, that is, the long tail group [44,45], which has more financial resources and can effectively broaden supply channels. Due to technical limitations and high service costs, traditional financial markets cannot effectively absorb these investors [46]. Supported by information technology, digital finance can process massive data at low cost and low risk, lower the service threshold, and promote broader long-tail groups to join the financial market [47,48]. In addition, digital finance provides intelligent investment, supply-chain finance, consumer finance, and third-party payment, which broadens financing channels [49] and further provides the possibility of obtaining funds for green technological innovation. On the other hand, the information matching function of digital finance can alleviate information asymmetry and enhance the allocation efficiency of financial resources [50,51]. Most scholars believe that information asymmetry between the financial market and innovation subject is one of the main reasons for inefficient resource allocation. The cost of information collection reduces investors’ willingness to invest, so it is more difficult for the innovation subject to obtain external financing. Digital finance can evaluate investor credit through algorithms and big data, provide credit informatization and transparency, alleviate information asymmetry, improve the credit–resource mismatch, overcome external financing constraints, and help innovation subjects to make reasonable and effective green technology innovation decisions [52], so as to comprehensively improve regional green technology innovation.

With the continuous improvement in the development level of digital finance, due to the profit-seeking of capital and the liquidity of financial elements, digital finance can continuously radiate to neighboring areas through the “trickle-down effect”, resulting in a spatial spillover effect [53]. Especially with the support of digital technology, geographical distance is no longer one of the more difficult problems affecting innovation subjects’ access to financial services [54]. Therefore, the spatial spillover effect of digital finance strengthens the financial support and information exchange of neighboring regions, and also promotes the green technology innovation ability of neighboring regions. Based on the above, this paper proposes research Hypothesis 1:

Hypothesis 1 (H1).

Digital finance can significantly enhance urban green technology innovation. At the same time, digital finance will also help improve the green technology innovation capacity of surrounding cities.

Porter hypothesis holds that strict and effective environmental regulation can stimulate enterprises’ willingness to innovate green technology and obtain competitive advantage through improving resource utilization rate, enhancing product performance and meeting production emission standards [22,23]. When the government implements environmental regulations, enterprises need to invest a large number of research and development personnel, research and development funds, purchase environmental protection equipment, emission permits, etc., which can be collectively referred to as environmental protection costs. In order to avoid the decline in economic benefits, enterprises will add environmental protection costs back into the product price. However, companies will also lose customers as prices rise, resulting in a loss of profits. At this time, the government can force and guide enterprises to carry out green technology innovation through environmental regulation. With the upgrading of technological structure, enterprises can realize the improvement of the resource utilization rate and the reduction of production costs and administrative penalty costs, thus obtaining a greater profit margin [55]. At the same time, an enterprise’s environmental image can attract more green consumers, increase the market share, and obtain competitive advantages. In this process, the innovation income of green technology is greater than the innovation cost, resulting in the “innovation compensation” effect [56,57]. Therefore, the government can further encourage enterprises’ green innovation behavior through environmental regulation. Combined with imitative learning between governments, relocation behavior of enterprises and technology spillover effect, this paper proposes research Hypothesis 2:

Hypothesis 2 (H2).

Environmental regulation can significantly improve urban green technology innovation. At the same time, environmental regulation also helps to improve the green technology innovation capacity of surrounding cities.

The good financial supply of digital finance provides a financial guarantee for the technological innovation activities of enterprises. However, whether the innovation results can improve the competitiveness of enterprises and also give consideration to environmental protection depends on the environmental regulation of the government [58,59]. Under the constraints of environmental regulations, enterprises need to carry out green technological innovation to achieve environmental compliance. Both front-end green production innovation and back-end governance innovation require a large amount of capital [60]. At this point, if the investment in green innovation exceeds the enterprise’s expectation and the financing cost is high, the enterprise will give up green transformation and turn to the negative behavior of reducing or stopping production [61]. Digital finance provides credit support to green transformation enterprises under the guidance of government environmental regulations and facilitates the green technological innovation of enterprises with low-cost and low-threshold financial services [40,62]. Therefore, in the process of providing effective financial services, digital finance should combine the green development orientation of the government to jointly promote the ability of green technology innovation and achieve the goal of high-quality economic development. Based on the above, this study proposes research Hypothesis 3:

Hypothesis 3 (H3).

Environmental regulation positively moderates the relationship between digital finance and urban green technology innovation capability.

4. Research Design Data Description

4.1. Model Construction

This study constructs a spatial Durbin model to explore the mechanism of digital finance and environmental regulation on green technology innovation capability. The specific measurement model is as follows:

$${\mathrm{lngt}}_{\mathrm{it}}={\mathsf{\rho }\mathrm{Wlngt}}_{\mathrm{it}}+{\mathsf{\beta }}_1{\mathrm{lndf}}_{\mathrm{it}}+{\mathsf{\beta }}_2{\mathrm{lner}}_{\mathrm{it}}+{\mathsf{\theta }}_1{\mathrm{Wlndf}}_{\mathrm{it}}+{\mathsf{\gamma }\mathrm{lnX}}_{\mathrm{it}}+{\mathrm{v}}_{\mathrm{it}}$$

(1)

$$\begin{array}{c}{\mathrm{lngt}}_{\mathrm{it}}=\mathsf{\rho }\mathrm{W}{\mathrm{lngt}}_{\mathrm{it}}+{\mathsf{\beta }}_1{\mathrm{lndf}}_{\mathrm{it}}+{\mathsf{\beta }}_2{\mathrm{lner}}_{\mathrm{it}}+{\mathsf{\beta }}_3{\mathrm{lndf}}_{\mathrm{it}}·{\mathrm{lner}}_{\mathrm{it}}+\\ {\mathsf{\theta }}_1\mathrm{W}{\mathrm{lndf}}_{\mathrm{it}}+{\mathsf{\theta }}_2\mathrm{W}{\mathrm{lner}}_{\mathrm{it}}+{\mathsf{\gamma }\mathrm{lnX}}_{\mathrm{it}}+{\mathrm{v}}_{\mathrm{it}}\end{array}$$

(2)

where $\mathrm{i}$ represents a city ($\mathrm{i}=1,2,3,\cdots ,278$), $\mathrm{t}$ represents the year ($\mathrm{t}$ = 2011, 2012, 2013, …, 2019), $\mathrm{gt}$ represents green technology innovation, $\mathrm{df}$ represents digital finance, $\mathrm{er}$ denotes environmental regulation, $\mathrm{X}$ means control variables, $\mathsf{\rho }$ stands for the space autoregressive coefficient, $\mathrm{W}$ stands for the weight matrix of adjacent space, and $\mathrm{v}$ stands for the error term.

4.2. Variable Selection and Data Source

4.2.1. Explained Variable

Green Technology Innovation ($\mathrm{lngt}$): Based on Lu’s [63] opinion, this study selects the data of urban invention patents and utility model patent applications and uses the principle of entropy weight method to construct a comprehensive index to measure the level of urban green technology innovation. The specific calculation process is as follows: first, the data indicators are normalized. Second, the entropy weight method is used to calculate the weight of each index. Finally, the comprehensive index of green technology innovation in each city is calculated.

4.2.2. Core Explanatory Variable

Digital finance ($\mathrm{lndf}$): Guo et al. [64] combined the characteristics of digital finance and data availability, and constructed “Peking University Digital Inclusive Finance Index” through three first-level dimensions, 12 s-level dimensions, and 33 specific indicators by using micro-data. This index scientifically portrays the degree of development of digital inclusive finance in China. Therefore, this paper chooses its comprehensive index as the measurement index of digital inclusive finance.

Environmental regulation ($\mathrm{lner}$): Based on the ideas of Ye et al. [65], this study selected wastewater, sulfur dioxide, and smoke (powder) dust emissions for a comprehensive evaluation of environmental regulation intensity through the entropy weight method to build the index system. This indicator is a positive indicator; that is, the greater the indicator, the greater the intensity of environmental regulation.

4.2.3. Control Variables

In order to improve the scientific nature of the empirical results between digital finance, environmental regulation, and green technology innovation, a series of control variables are added. (1) Regional economic development level ($\mathrm{lngdp}$ ): measured by gross regional product; (2) urban innovation environment ($\mathrm{lnie}$): measured by the general budget of local finance; (3) degree of opening to the outside world ($\mathrm{lnod}$): measured by the gross industrial output value of foreign-invested enterprises in the region; (4) urban environmental quality ($\mathrm{lneq}$): use harmless treatment rate of household garbage to measure; (5) urban industrial structure ($\mathrm{lnis}$): the proportion of added value of the secondary industry in GDP is selected for measurement.

In consideration of data integrity and reliability, panel data of 278 Chinese cities from 2011 to 2019 were selected in this study. The data come from The Research Center for Digital Finance of Peking University and The Statistical Yearbook of Chinese Cities. In this study, all data were logarithmically processed to mitigate the impact of heteroscedasticity, extreme values, and skewness on the estimated results. Statistical results of variable description are shown in

Table 1. Descriptive statistics.

Variable	Name	Mean	Std. Dev	Min	Max
lngt	2502	4.495	1.701	0	9.906
lndf	2502	5.005	0.511	3.057	5.773
lner	2502	−3.016	0.959	−8.702	−0.312
lngdp	2502	16.605	0.898	14.243	19.760
lnie	2502	14.878	0.749	12.031	18.24
lnod	2502	13.374	2.597	0	18.834
lneq	2502	4.503	0.242	1.703	5.062
lnis	2502	3.823	0.254	2.370	4.492

.

5. Empirical Analysis

5.1. Spatial Autocorrelation Test

Before the empirical analysis, the Moran index was used to analyze the spatial autocorrelation of digital finance and the green technology innovation ability of 278 cities by using the adjacent spatial weight matrix, and the spatial econometric model was investigated. Its calculation formula was as follows:

$$I=\frac{{{\displaystyle \sum }}_{i=1}^n{{\displaystyle \sum }}_{j=1}^n{w}_{ij}\left(x_i-\overline{x}\right)\left(x_j-\overline{x}\right)}{S^2{{\displaystyle \sum }}_{i=1}^n{{\displaystyle \sum }}_{j=1}^n{w}_{ij}}$$

(3)

$$\text{Among} \text{them}: S^2=\frac{{{\displaystyle \sum }}_{i=1}^n{\left(x_i-\overline{x}\right)}^2}{n}$$

(4)

The Moran index is one of the most commonly used indicators of spatial correlation. The value of Moran index I is generally between [−1,1]. A Moran index I close to 0 indicates that the spatial distribution is random and there is no spatial autocorrelation; greater than 0 indicates positive correlation, and the larger its value, the more obvious the spatial correlation; a value less than 0 indicates negative correlation, indicating greater spatial heterogeneity. As can be seen from

Table 2. Moran ‘I index of green technology innovation and digital finance from 2011 to 2019.

Year	Digital Finance		Green Technology Innovation
	Moran Index	Z Value	Moran Index	Z Value
2011	0.060 ***	8.352	0.062 ***	16.899
2012	0.070 ***	9.743	0.061 ***	16.735
2013	0.088 ***	12.065	0.049 ***	13.550
2014	0.058 ***	8.111	0.051 ***	14.021
2015	0.068 ***	9.474	0.053 ***	14.703
2016	0.075 ***	10.343	0.056 ***	15.493
2017	0.089 ***	12.233	0.060 ***	16.419
2018	0.118 ***	16.002	0.066 ***	18.105
2019	0.126 ***	17.172	0.066 ***	18.147

Note: *** represents significant at the significance level of 1%.

, the Moran index I of digital finance and green technology innovation from 2011 to 2019 is between 0.060 and 0.126, and is significant at the 1% level, indicating a strong spatial correlation between digital finance and green technology innovation. To observe the spatial agglomeration of digital finance, this paper draws local Moran scatter plots of digital finance in 2011 and 2019, as shown in Figure 1. Figure 1 shows that digital finance has a spatial agglomeration effect and strong spatial correlation.

5.2. Model Selection

In this study, the LM test and its robustness test were used to judge the spatial distribution properties of each variable and the choice of spatial econometric model. As can be seen from the LM test results in

Table 3. LM test results.

LM Test	The Current Period		A Phase Lag
	Model (1)	Model (2)	Model (1)	Model (2)
LM Spatial error	781.159 ***	756.630 ***	674.772 ***	640.610 ***
LM Spatial lag	9.018 ***	8.410 ***	7.136 ***	6.100 **
Robust LM Spatial error	789.367 ***	765.303 ***	686.185 ***	653.170 ***
Robust LM Spatial lag	17.226 ***	17.083 ***	18.548 ***	18.660 ***

Note: ** and *** represent significant at the significance level of 5%, and 1%, respectively.

, both passed the significance test and significantly rejected the null hypothesis. The panel model with spatial effect should be selected in this paper. Secondly, the LR test of the spatial Durbin models (1) and (2) shows that the hypothesis that they degenerate into spatial error model or spatial lag model is significantly rejected, which supports the scientific selection of the spatial Durbin model. Meanwhile, Hausman test strongly rejects the null hypothesis; that is, the Durbin model with fixed effects is more suitable for this study than the Durbin model with random effects. Therefore, this paper should select the spatial Durbin model for spatial econometric analysis.

5.3. Spatial Effect Analysis

5.3.1. Spatial Model Results

This study examined the relationship between digital finance, environmental regulation, and green technology innovation using the spatial Durbin model with time–city dual fixations. Model (1) mainly examines the impact of two explanatory variables on green technology innovation, while Model (2) includes the interaction term of digital finance and environmental regulation, and comprehensively considers the interaction relationship among the three. As can be seen from the results of Model (1) in

Table 4. Regression results of Durbin model in fixed-effect space.

The Variable Name	The Current Period		A Phase Lag
	Model (1)	Model (2)	Model (3)	Model (4)
lndf	2.721 *** (22.56)	3.096 *** (19.30)	2.660 *** (21.21)	3.056 *** (18.15)
lner	0.092 *** (5.02)	−0.423 *** (−2.89)	0.085 *** (4.18)	−0.457 *** (−2.95)
lndf*lner		0.102 *** (3.55)		0.110 *** (3.53)
lngdp	1.156 *** (22.96)	1.147 *** (22.65)	1.146 *** (20.98)	1.133 *** (20.88)
lnie	0.064 (1.26)	0.068 (1.33)	0.090 (1.65)	0.095 ** (1.74)
lnod	0.024 *** (2.94)	0.026 *** (3.03)	0.022 *** (2.43)	0.023 *** (2.56)
lneq	0.133 ** (2.24)	0.142 ** (2.36)	0.139 *** (2.25)	0.147** (2.40)
lnis	−0.358 *** (−3.86)	−0.349 *** (−4.62)	−0.329 *** (−3.93)	−0.324 *** (−3.86)
w*lndf	−0.198 (−0.44)	−0.260 (−0.34)	−0.229 (−0.48)	−0.669 (−0.79)
w*lner	0.252 * (1.83)	0.491 (0.49)	0.104 (0.69)	0.961 (0.89)
w*lngt	0.339 *** (3.60)	0.316 *** (2.27)	0.334 *** (3.53)	0.297 ** (2.08)
R2	0.476	0.482	0.485	0.499
observations	2502	2502	2224	2224
time	fixed	fixed	fixed	fixed
city	fixed	fixed	fixed	fixed

Note: *, ** and *** represent significant at the significance levels of 10%, 5%, and 1%, respectively, and t-statistics in parentheses.

, the regression coefficient of digital finance on local green technology innovation is 2.721, which passes the significance test. However, the influence of digital finance on green technology innovation in neighboring areas is not significant. Part of hypothesis 1 is verified. This indicates that digital finance can only promote local green technology innovation. The wide financing channels and strong information-matching ability of digital inclusive finance stimulate the willingness of innovation subjects to green innovation, so it has a significant positive impact on the local green technology innovation ability. However, the influence of digital finance on green technology innovation in neighboring areas is not significant. This indicates that green technology innovation is only affected by the development of digital finance in this region, and is not affected by the development of digital finance in other regions, which is consistent with the research conclusion of Zhang et al. [66]. The possible reason that lies in the difference between the development level of inter-regional digital finance and the degree of government interaction leads to the regional heterogeneity of the spatial spillover effect of digital financial development. Combined with the results of regional heterogeneity analysis in Section 5.5, it can be seen that the significant inhibition effect in southwest China and northwest China may offset the significant promotion effect in central China, resulting in the insignificant total sample estimation coefficient. The regression coefficient of environmental regulation on local green technology innovation was 0.092, which passed the significance test. Environmental regulation promotes local green technology innovation Moreover, environmental regulation also has a positive impact on green technology innovation in neighboring areas. On the one hand, with the increase in government environmental regulation intensity, enterprises will achieve the effect of reducing environmental protection costs and improving resource utilization rate through green technological innovation, aiming to achieve the common progress of economic benefits and environmental benefits through the “innovation compensation” effect. On the other hand, in order to avoid excessive expenditure in environmental costs, some small and medium-sized high-tech enterprises move to the neighboring areas with relatively low environmental regulation intensity. Therefore, the flow of capital, information, technology and personnel promotes green technology innovation in neighboring areas. Therefore, hypothesis 2 is supported. The regression coefficient of the interaction term between digital finance and environmental regulation in Model (2) is significantly positive, indicating that environmental regulation plays a positive moderating role in the process of digital finance affecting local green technology innovation. Hypothesis 3 is supported. That is, the government’s environmental regulation can play a positive role in the process of digital finance promoting green technology innovation. The empirical study shows that the level of economic development, the degree of urban openness, and the quality of urban environment all have a significant promoting effect on green technology innovation, while the industrial structure has a significant positive effect on green technology innovation. This indicates that the higher the proportion of secondary industry is, the more unfavorable it is to the progress of urban green technology innovation level. The coefficient of the spatial Durbin model passed the significance test at the 1% level, indicating that the level of local green technology innovation contributes to the improvement of the level of green technology innovation in neighboring areas; that is, there is a spatial spillover effect of green technology innovation.

Considering that digital finance and environmental regulation may have a lag effect on green technology innovation, this study adopts digital finance and environmental regulation with a lag of one stage to conduct re-regression on Model (1) and Model (2). The test results are shown in Model (3) and Model (4) in Table 4. Based on Table 4, it can be seen that, compared with Model (1) and Model (2), the test result of one lag period is basically consistent with that of the current period. Therefore, the following robustness test adopts lagged one-phase variables to further test the model.

5.3.2. Spatial Effect Decomposition

To further illustrate the marginal effects of digital finance and environmental regulation on green technology innovation, this study performs a spatial effect decomposition and divides the changes into direct, indirect and total effects. The direct effects include the direct impact of explanatory variables on local green technology innovation and the feedback effect of neighboring explained variables on local green technology innovation. Indirect effects reflect the influence of local explanatory variables on green technology innovation in neighboring areas

Table 5. Spatial effect decomposition.

Variable	Direct	Indirect	Total
lndf	2.725 *** (22.48)	1.072 (1.18)	3.798 *** (4.09)
lner	0.091 *** (5.85)	0.420 ** (2.10)	0.511 ** (2.55)

Note: ** and *** represent significant at the significance levels of 5%, and 1%, respectively, and t-statistics in parentheses.

shows the decomposition results of spatial effects of digital financial and environmental regulations. According to the direct-effect test results, digital finance has a significant positive influence on local green technology innovation; that is, every 1% increase in the development level of digital finance can improve the local green technology innovation level by 2.725%. Environmental regulation plays an important role in promoting local green technology innovation; that is, when the intensity of environmental regulation increases by 1%, the level of local green innovation will increase by 0.091%. Compared with the parameter estimation of the fixed effect of the spatial Durbin model in Table 4, it can be seen that there are some differences between the parameter estimation results of digital finance and environmental regulation. For example, the direct effect of digital finance on local green technology innovation is 2.725, while the regression coefficient estimated by the spatial Durbin model is 2.721. The difference between the two is caused by the feedback effect of digital finance on green technology innovation in nearby areas.

The estimation results of indirect effects show that the environmental regulation has a significant positive spillover effect, while the spillover effects of digital finance do not pass the significance test. Each 1% increase in the intensity of environmental regulation has a 0.612% promotion effect on the green technology innovation ability of neighboring areas.

5.4. Robustness Test

To maintain the reliability of the regression results, data around 3% of the sample maximum and minimum values were excluded for robustness testing, and the results of each indicator after excluding outliers are analyzed in detail in the columns of

Table 6. Robustness test.

Variable	Eliminating Outliers		Variable	Eliminating Outliers
	Current Period(1)	Lag Period (2)		Current Period(1)	Lag Period (2)
lndf	3.320 *** (19.36)	3.056 *** (18.15)	lneq	0.171 ** (2.20)	0.147 *** (2.40)
lner	−0.264 * (−1.94)	−0.457 *** (−2.95)	lnis	−0.291 *** (−3.55)	−0.324 *** (−3.86)
lndf*lner	0.068 ** (2.55)	0.109 *** (3.53)	w*lndf	−0.525 (−0.50)	−0.668 (−0.79)
lngdp	0.963 *** (18.40)	1.132 *** (20.88)	w*lner	1.187 (0.76)	0.960 (0.89)
lnie	0.203 *** (3.73)	0.095 * (1.74)	w*lngt	0.206 (1.02)	0.296 ** (2.08)
lnod	0.075 *** (6.75)	0.023 *** (2.56)

Note: *, ** and *** represent significant at the significance level of 10%, 5%, and 1%, respectively, and t-statistics in parentheses.

. From the results in Table 6, it is known that the estimated coefficient values of the variables remain significant, the coefficient fluctuation range is not large, and the sign of the positive and negative have not changed. It is not difficult to see that the results are basically consistent with the previous spatial regression results, which further confirms the robustness of the empirical results in this study.

5.5. Heterogeneity Analysis

To further analyze the regional differences in digital finance and environmental regulation on green innovation, 278 cities were divided into seven parts, namely East China, South China, North China, Central China, Southwest China, Northwest China, and Northeast China, and each region was tested. Specific test results are shown in

Table 7. Heterogeneity analysis of the impact of regional digital finance and environmental regulation on green technology innovation.

Variable		Northeast China	East China	North China	South China	Central China	Northwest China	Southwest China
lndf	Direct	0.414 ** (2.24)	0.779 *** (5.31)	−0.269 (−0.91)	1.361 *** (4.36)	0.611 ** (2.26)	0.769 ** (2.31)	0.387 * (1.82)
	Indirect	−0.174 (−0.83)	−0.813 (−1.43)	−0.234 (−0.63)	−0.472 (−0.99)	0.064 ** (2.26)	−0.772 ** (−1.79)	−1.372 ** (−2.33)
	Total	0.240 ** (1.79)	−0.034 (−0.06)	−0.504 * (−1.77)	0.889 ** (2.22)	0.676 *** (4.11)	−0.002 (−0.01)	−0.985 * (−1.68)
lner	Direct	−0.040 (−0.82)	0.114 *** (3.19)	−0.022 (−0.55)	−0.022 (−0.41)	−0.026 (−0.67)	−0.072 (−1.21)	−0.047 (−1.15)
	Indirect	0.103 * (1.67)	0.376 ** (2.23)	−0.246 ** (−2.24)	0.172 (0.63)	0.206 *** (4.01)	0.010 (0.09)	0.036 (0.23)
	Total	0.063 (1.79)	0.489 *** (2.98)	−0.268 ** (−2.44)	0.149 (0.55)	0.179 *** (5.55)	−0.062 (−0.54)	−0.011 (−0.08)

Note: *, ** and *** represent significant at the significance level of 10%, 5%, and 1%, respectively, and t-statistics in parentheses.

. By comparing the total effect of digital finance, we found that smart finance in Northeast, South and Central China has a significant contribution to green technology innovation. Moreover, from the elastic coefficient, digital finance has the best promotion effect on South China. On the contrary, digital finance inhibits green technology innovation in North China and Southwest China. The elasticity coefficient of Southwest China is –0.985, indicating that the level of technology innovation in Southwest China will decrease by 0.985% when digital finance increases by 1%. To be specific, digital finance can promote local green technology innovation except in North China, and digital finance has a spillover effect only in Central China, Northwest China, and Southwest China. In conclusion, there is regional heterogeneity in the impact of digital finance on China’s green technology innovation in China, so it is difficult to comprehensively promote green innovation

According to the results in Table 7, we find that the total effect of environmental regulation only passes the significance test in East China, North China, and Central China. Among them, the total effect of environmental regulation in East China and Central China is positive, indicating that environmental regulation has an inhibitory effect on green technology innovation. Moreover, the promotion effect in East China is greater than that in Central China. Xu et al. [67] believe that the role of environmental regulation is closely related to the degree of economic development. East China has a high-quality economy, so its environmental regulations are more scientific and perfect. Therefore, under strict and effective supervision, green technology innovation can be further promoted. Central China is in the middle of the economic development level, because it is largely East China that undertakes energy-intensive industries. Therefore, its regulation is limited. When environmental regulation is strengthened in central China, its promotion effect on green technology innovation is less than that in eastern China. On the contrary, environmental regulation inhibits the progress of green technology innovation in North China. There are negative spillover effects that hinder the progress of green technology innovation in neighboring areas. The environmental regulations in East China and Central China not only hinder the local innovation and progress but also inhibit the neighboring regions. Xin [68] believes that North China is an important political center of China, where a large number of technological enterprises gather together. Therefore, when the intensity of environmental regulation increases in North China, some enterprises with strong pollution move to neighboring areas, which ultimately inhibits the level of green technology innovation in local and surrounding areas.

5.6. Discussion of Empirical Results

This paper examines the influence mechanism among digital finance, environmental regulation and technological innovation by constructing a spatial Durbin model. It can be seen from the robustness test results in Table 6 that the model has passed the test, indicating that the conclusion is true and reliable. Meanwhile, according to the empirical results, although digital finance has a significant role in promoting local green technology innovation, the spatial spillover effect on neighboring areas fails to pass the test. This conclusion is consistent with Xie’s [69] research on digital finance and regional technological innovation based on provincial panel data. Although digital finance has a significant spatial agglomeration effect, it fails to drive green technology innovation in neighboring areas. Secondly, the empirical results show that environmental regulation can not only promote local green technology innovation progress, but also stimulate the improvement of green technology innovation levels in neighboring areas through a positive spatial spillover effect. This is consistent with the conclusions of Zheng et al. [70] analyzing the impact of environmental regulation on industrial green innovation and Zhang et al. [71] examining environmental regulation and environmental governance. Thirdly, the study also shows that environmental regulation can strengthen the promotion effect of digital finance on green technology innovation. That is, in the process of digital finance affecting green technology innovation, environmental regulation plays a positive moderating role. Shi et al. [37], Li et al. [35], Wang et al. [36] reached similar conclusions when exploring the impact of digital finance and environmental regulation on environmental pollution, industrial structure upgrading and economic growth. Finally, the relationship among digital finance, environmental regulation and green technology innovation is found to have regional heterogeneity. Green technology innovation simultaneously considers technological progress, economic performance and environmental performance. Thus, spurred by digital finance and environmental regulation, companies can make more profits from cleaner methods of production. These achieve sustainable economic development through green technology innovation. In summary, the research in this paper further enriches the research related to digital finance, environmental regulation and green technology innovation, and at the same time provides a theoretical basis for the government to adopt relevant mechanisms and thus achieve regional green transformation and upgrading.

6. Conclusions and Suggestions

This paper selects panel data of 278 cities in China from 2011–2019 and builds a spatial Durbin model based on a spatial correlation perspective to empirically investigate the relationship between digital finance, environmental regulation and green technology innovation and conducted robustness tests. Then, considering the regional heterogeneity, 278 cities were divided into seven parts according to geographical location, and the relationship among the three areas was discussed, respectively. The results are as follows.

• Digital finance has an important role to play in promoting local green technology innovation. It is obvious that the low superlative threshold, low cost, high efficiency and informatization of digital finance encourage local enterprises’ green technology innovation through channels such as improving financing availability, reducing financing cost and transaction time, and improving resource allocation rate.
• Government environmental regulation facilitates the development of green technology innovation in local and adjacent areas. For one thing, it shows that the Porter hypothesis is valid in China. For another, environmental governance also reflects the relationship between learning and competition among local governments in China. When local governments force companies to innovate in green technologies by enforcing strict environmental regulations, neighboring governments also strengthen environmental regulations to achieve high-quality development.
• Environmental regulation plays a positive moderating role in the process of digital finance affecting green technological innovation. That is, environmental regulation plays a positive moderating role in the process of digital finance affecting green technology innovation. It shows that in the process of digital finance promoting green technology innovation, government environmental regulation plays an important guiding role.
• There is regional heterogeneity in the relationship between digital finance, environmental regulation, and green technology innovation. Among them, the environmental regulation in North China inhibits the local green technology innovation the most; Digital finance in Central China can not only promote green technology innovation in the region but also green technology innovation in neighboring regions through a spillover effect.
• The development of the secondary industry hinders the progress of green industry and further inhibits the level of urban green technology innovation.

In summary, we put forward the following policy recommendations: First, the government should continue to promote the development of digital finance and accelerate the innovative integration of finance and technology, on the basis of improving digital finance development infrastructure, promoting the construction of credit evaluation system, and guiding more practitioners to join. Additionally, it is essential to standardize the financial market service system and strengthen information protection. Second, the government should fully consider regional heterogeneity when formulating environmental regulations, combining regional characteristics to guide enterprises through green technology innovation to through environmental subsidies and policy publicity, so as to realize the coordination of environmental protection and economic progress. Therefore, local governments should break the restrictions of administrative regions and strengthen the communication and cooperation between regions when formulating and implementing environmental regulations. We should give full play to the role of environmental regulations in improving green technological innovation, and work together to achieve green upgrading and transformation. Third, the government should vigorously promote the transformation of the secondary industry. To achieve high-quality economic development, the government needs create a good industrial innovation environment and stimulate the willingness of the secondary industry to innovate.

7. Research Limitations and Future Research

There are some limitations in this study. First, due to limited data availability, this paper and a large number of existing studies in the construction of green technology innovation comprehensive evaluation index only consider the relevant data of green patents, not the R & D personnel in the process of innovation, R & D funds and R & D results of green product sales and other related data into the evaluation system. In the future, data will continue to be mined to further improve the comprehensive index of green technological innovation. Second, the study in this paper focuses more on the impact of digital finance on green technology innovation, and therefore does not provide a detailed delineation of environmental regulation. In the later research, environmental regulation should be divided into command-and-control type, market incentive type and voluntary type according to the different regulatory tools, so as to further explore the heterogeneous impact of environmental regulation.