Introduction

Urbanization and globalization have been advanced in recent years; energy consumption has grown, leading to an increase in various ecological pollutants (Irfan et al. 2022; Luo et al. 2022), carbon dioxide (CO2) (Bai et al. 2022). China Petroleum claims that annual worldwide CO2 emissions have increased of CO2 in 2018. To better comprehend the situation, generate worldwide displaying allocation of CO2 emissions in 2018. As a method of reducing the negative impacts of the greenhouse effect, administrations throughout the globe have been pushing hard for a transition (Zeng et al. 2022; Zheng et al. 2022). Further, the signing of confirms the commitment of all governments to lowering carbon emissions (Liddell et al. 2012). To be more specific, a low-carbon economy cannot be achieved without first making the transition to renewable energy. CO2 emissions must be lowered to slow down the acceleration of global warming. State strategies that act to reduce emissions of greenhouse gasses (GHG) and to prepare for the increasingly foreseeable consequences are now top priorities in many countries. For this reason, ecological protection and the SDGs have moved to the top of the agendas of national governments throughout the globe, since the protection and restoration of the world’s ecosystems are crucial to the financial and social well-being of countries.

Even though many specialists have paid attention to the topic, very little research has offered credible and consistent evaluation methods for energy resilience. Additionally, little research has looked at the importance of energy resilience in the field of ecology (Irfan et al. 2019). Three items have piqued our interest in light of the above: the first order of business is to figure out how to assess the state of the nation’s energy infrastructure more precisely. Can the greenhouse impact be affected by a resilient energy system? If so, what is the precise mechanism of effect, and are there any implications for asymmetry or spatial heterogeneity among energy resilience and CO2 emissions (Xu et al. 2022b)? The correctness depends heavily on a discussion of the following issues. As such, we use a cross-sectional dataset consisting of information from 108 nations in 2018 to empirically assess CO2 emissions. Han et al. (2022a) did this by first analyzing by breaking down the whole effect of energy resilience on global CO2 emissions. Jin et al. (2022) also conduct asymmetrical and regional heterogeneous assessments of the relationship between energy resilience and CO2 emissions to account for potential variability.

The need for technological advancement and financial outlay may be affected by the ease with which people can acquire and share knowledge in a prosperous and all-encompassing economic sector. The World Bank considers the availability of appropriate and affordable financial services and products to be an enabler of the seventh goal of the seventeen SDGs (Zhao et al. 2020a); the stated aim of which is to ensure that businesses and individuals have access to credit and insurance, conduct economic transfers, make payments, and build savings. A further definition of financial involvement is the provision of high-quality services to underserved communities at low cost (Zhao et al. 2020b). Generally speaking, the availability and use of financial services are the most telling indicators of a country’s progress toward economic inclusion. Financial inclusion initiatives may have grown in certain countries (Han et al. 2022c). By expanding access to banking services, proponents of economic inclusion help small businesses boost productivity and expand more quickly. Furthermore, data from shows that cost is a major barrier to people making the switch to cleaner energy at home.

ICT has evolved as a basic pillar of contemporary life, enabling the transition from the industrial era to the information era (Dai et al. 2022). In addition, ICT is considered a driver of change toward a more equitable, inclusive, sustainable, and competitive financial and social order. In the economics literature, it is generally agreed that the adoption of ICT has had a positive and statistically significant influence on effectiveness growth. Similarly, expenditures in information and communication technologies (ICT) might lead to cleaner, more sustainable commercial processes, which in turn could lead to greater carbon efficiency (Jiang et al. 2022). To reduce energy intensity and greenhouse gas emissions, ICT plays a critical role in promoting the dissemination of innovative ideas and technology (Liu et al. 2022b). Many contemporary scholarly conversations have focused on the potential of information and communication technology to develop communities that are stronger, more inclusive, and more environmentally sustainable (Liu et al. 2021b). Therefore, ICT may aid in achieving the SDGs by enhancing the financial, social, and environmental effectiveness with which these objectives, such as speeding financial growth, reducing poverty, and limiting climate change, are pursued (Wang et al. 2022). Many individuals think that ICT can free financial growth from its dependence on carbon dioxide (CO2) emissions.

There is empirical evidence that shows how the proliferation of ICTs has contributed to both financial growth and lower carbon dioxide emissions. The first group has studied the link between ICT and financial development within the framework of neoclassical and endogenous growth models. Economic expansion is linked to the spread of information and communication technologies, according to available data (Xu et al. 2022a). The second team has investigated the EKC strategy’s potential to reduce CO2 emissions by studying the interplay between the growth of ICT and economic development. The impact of information and communication technology on carbon dioxide emissions is not yet well understood. Some studies suggest that ICT aids in lowering carbon dioxide emissions and encouraging countries to green their economies, whereas other studies find that greater ICT use and penetration are harmful to the environment (Liu et al. 2022a).

Third, this research shows that, from the standpoint of green innovation, the influence of digital finance on green innovation is significantly less affected by characteristics like the level of regional economic growth and the form of company property rights (state-owned or private). Using data collected from 30 Chinese provinces between 2000 and 2017, this study experimentally investigates the effect of digital finance on the quantity and quality of green technical innovation. The literature reviews to this work is described in the “Literature review” section. The “Empirical model and study variable” section presents the empirical methods, study variables, and descriptive statistics results. The “Empirical results and discussion” section presents the empirical results and discussion. The “Conclusion and policy implications” section presents conclusion and policy implication.

Literature review

Research on energy resilience

Academic focus on energy resilience has grown over the last several years as the study of resilience has spread across the energy industry. The best approach to physically measure resilience is the subject of heated scholarly dispute. For this reason, many researchers are stumped by the challenge of determining how to comprehensively.

Mohsin et al. (2022b) proposed a theoretical method for analyzing and carrying out effective evaluations of urban energy resilience. Further, they classified the different energy resilience principles in terms of the four primary sustainability metrics: availability, cost, quality, and tolerance. Studies like Sun et al.’s (2019a) study just explains the disciplines to analyzed energy resilience; instead of providing the criteria for doing so, they only supply the assessment itself. Given these factors, we provided a novel, multi-stage method for energy resilience analysis based on the following three procedures: factual assessment (Ikram et al., 2019), value calculation, and comparison analysis. Whoever suggested that takes into account regional differences to provide energy security in metropolitan areas.

There is sufficient evidence that green innovation is on the increase in both developed and developing nations (Abbas et al. 2022). Based on their research, they determined that when it comes to this kind of technology, the advanced economies of Western Europe and North America are light years ahead of the rest of the world. Since developments in green and low-carbon technology are intricately related to policy regulations, it has also become a new pillar of sustainable growth. Research on BRICS economies’ carbon emissions and green technologies from 1994 to 2016 is presented (Iram et al., 2020). Green technology is effective in lowering carbon emissions, particularly in the highest emission quantiles. The link between transmitting low-carbon technology and reducing emissions is highlighted by Sun et al. (2019b). Global warming may be considerably mitigated by the dissemination of climate-related technology. Using the EKC curve as a lens analyzed the contribution of ICT to meet sustainable growth goals in the context of solving environmental issues. An analysis of moment-quantile regression data demonstrates that ICT has a significant role in preventing environmental degradation. Mohsin et al. (2022a) urged to invest more in incorporating environmentally friendly technologies into the industrial sector at large. Expenditure in research and development as well as training of human resources is essential if ICT innovations are to be focused on sustainable business solutions. Furthermore, global innovations have been applauded for their capacity to lower emissions, and various studies have explored the dynamic and causal connection between global technology and ecological concerns (Zhang et al. 2021).

Agyekum et al. (2021) looked at the correlation between financial growth and two pollutants from 1990 to 2015 and determined their causal relationship. Their research lends support to the idea that emissions of two pollutants are cointegrated with GDP over time. It was found that the relationship between SO2 emissions and GDP had the shape of an inverted U. As for CO2 emissions, it seems that things are improving as the two continue to trend upward in tandem with GDP. To further understand the relationship among CO2 emissions, GDP, energy consumption, and commerce used the cointegration approach (Mohsin et al., 2020). Additional support for the EKC model is provided by the finding that pollution emissions increase in tandem with GDP initially, before leveling off after the economy reaches a certain degree of stability. The cointegration approach is then used to examine the association between economic development, carbon dioxide emissions, and several other variables in Tunisia between 1969 and 2012. So, we can rule out the EKC hypothesis. Finally, a more recent study (Miao et al. 2021) examines the correlation between CO2 emissions, information and communication technology, and total factor productivity in Tunisia between 1995 and 2018. Integrating their data demonstrated that the EKC hypothesis for CO2 emissions was not supported by their results. To wrap up, Iqbal et al. (2019) look at the impact that technological innovation had on CO2 emissions in Vietnam between 1970 and 2016 by using the number of patents granted as a proxy. When using the ARDL technique with a break, the EKC hypothesis does not hold up in developing economies. Granger causality studies reveal that although technology has little direct effect on pollution, it has a significant indirect effect by dampening the beneficial consumption on CO2 emissions.

The literature on financial inclusion

Financial inclusion requires supplying underserved groups with necessary and affordable services and commodities, such as credit, insurance, economic transactions, payments, and savings (Shaari et al., 2012). There has been a surge in recent years of interest in the idea of financial inclusion among researchers, but no universally accepted technique of computation has been established to evaluate this trend (Iqbal et al. 2019). Generally speaking, economic service access and use are the two primary indicators of financial inclusion. The majority of studies examine the relationship between economic inclusion and other factors including GDP growth, new product growth, job creation, income inequality, and environmental balance (Ullah et al., 2020). However, little research that does look at this topic finds that increasing people’s access to financial services considerably reduces energy poverty in both Ghana and Turkey. However, most prior studies have neglected to include financial inclusion as a factor in energy use or energy transition.

The nexus of financial inclusion and renewable energy

Poor credit ratings make it difficult for small and micro-sized enterprises (SMEs) to engage in R&D (Canadell et al. 2007). Without adequate funding, small- and medium-sized enterprises (SMEs) in the renewable energy sector will not be able to take advantage of the opportunities presented by clean energy technology innovation to increase the rate of renewable energy manufacturing and decrease the cost of renewable energy manufacturing. Moreover, from the perspective of the consumer, increased economic participation will make it possible for individuals and small businesses to purchase sustainable energy equipment like solar water heaters (Dilanchiev and Taktakishvili 2022). Increased availability at more affordable (Udemba et al. 2021) prices will lead to greater adoption of renewable energy sources.

Furthermore, the demand for renewable energy is affected by a wide range of factors. Renewable energy development in China is favorably connected with GDP growth (Zhang and Dilanchiev 2022). New studies show that exports per capita have a positive influence on renewable energy use in the central and eastern regions, but a harmful effect in the western regions.

Empirical model and study variable

Econometric analysis

This section measuring the impact of green digital finance on green innovation and renewable energy described can be expressed follow as.

Empirical model is particular as follows:

$${y}_{i,t}= {\alpha }_{0}+{\beta }_{1}{\mathrm{GDF}}_{it-1}+ {\sum }_{j=1}^{{p}_{i}}{X}_{it}\gamma + {u}_{i}+{\varepsilon }_{it}$$
(1)

where \(Y\) presents the study explained and advance variables and the quality and quantity of green innovation and digital innovation (Yuman et al. 2018) The China cities are represented by the subscript \(i\). The study year is indicated by the \(t\). In the selected variables of the individual specific intercepts and time trends, several pooled panel unit root tests with different specifications were produced. The model Eq. (1) null hypothesis is equal to 0. The null hypothesis is dismissed if it is significant at the 5% level. It suggests that the variables discussed have an impact on digital finance. \(X\) stands for a few control variables. Unobservable factors at the individual level and temporal trend are added to Eq. (2). An effort to manage them (one) is the word that represents random error and the time-fixed effect and denotes the cities fixed effect of China cities \(i\). Although the intercept and trend may vary throughout individual sequences, their test imposes uniformity on the digital finance coefficient, which indicates the presence or absence of a unit root issue. The ADF regression (Bai et al. 2020) to investigate the unit root hypothesis is shown in the steps below. We apply the ADF regression separately for all China’s cities.

The reduced-form VAR model is:

$${y}_{i,t}= {\alpha }_{i}+\mathrm{GDF}{y}_{it-1}+ {\sum }_{j=1}^{{p}_{i}}{\alpha }_{i,j}{y}_{i,t-j}+{\varepsilon }_{i,j}$$
(2)

In addition, specific countries are allowed to use the lag order \({p}_{i}\). Having allowed the full lag order, then use the t-statistics for \(ij\) b to decide whether a smaller lag order is desired on how the appropriate lag period is (Kathuria and Sabat 2020) determined. Save the residuals from running two different regressions (Eq. (2)) which are regressed. Equations (1) and (2) reflect the direct effects of green digital finance (GDF). The following mediating impact model is created by including the mediating variable to further examine the potential indirect influence mechanisms of digital finance on green innovations.

$${\widetilde{{y}_{i,t}= {\alpha }_{0}+{\beta }_{1}{\mathrm{GTI}}_{it-1}+ {\sum }_{j=1}^{{p}_{i}}{X}_{it}\gamma + {u}_{i}+{\varepsilon }_{it} }\eta }_{it}=\frac{{\widetilde{\eta }}_{it}}{{\widehat{\sigma }}_{ei}} ,{\widetilde{\eta }}_{it-1}=\frac{{\widetilde{\eta }}_{it-1}}{{\widehat{\sigma }}_{ei}}$$
(3)

Model 2 null hypothesis is equal to 0. Model 3 null hypothesis is equal to 0. In the event that both are significant at the 10% level, there is a potential mediating effect. If it is statically important but not significant, it has a complete significant influence.

Multilevel model

Introducing the panel unit root testing (Muggeridge et al. 2014; Economidou et al. 2020; Nosheen et al. 2021; Silva et al. 2021), this testing extends the augmented Dickey-Fuller (ADF) test to each series, permitting each sequence to have short-run dynamics (Agri et al., 2018). However, the aggregate t-test statistics take the statistical mean of all distinct China cities.

$$\Delta {x}_{i,t}=\overline{\omega }j+\overline{\omega }i {x}_{i,t-1}+{\sum }_{j=1}^{{p}_{i}}{\varnothing }_{i,j}\Delta {x}_{i,t-j}+{\nu }_{i,t}$$
(4)

For China cities, the augmented Dickey-Fuller (ADF) test (Anser et al. 2020a; Ram et al. 2020; Zeng et al. 2020) has distinct modification lags that are present in many observations; \(E\left({t}_{T}\right)\) and \(\mathrm{var}\left({t}_{T}\right)\) are then followed by the tabulated averages for the respective group, such as \(E\left({t}_{T},{P}_{i}\right)\) and \(\mathrm{var}\left({t}_{T},{P}_{i}\right)\), respectively. The IPS testing permits variation within the value under its alternate hypothesis \(\overline{\omega }i\).

$$\lambda =-2{\sum }_{i=1}^{n}{\mathrm{log}}_{e}\left({\mathrm{FDI}\_\mathrm{GDF}\_\mathrm{GTI}p}_{i}\right)\sim {x}_{{2}_{n}}^{2}\left(d.f.\right)$$
(5)

where \({p}_{i}\) is the likelihood value for unit \(i\) from the ADF unit root test. The MW unit root test outperforms the IPS, the unit root test, since it is adaptive to lag length selection in individual ADF regression analysis.

Variable selection

Dependent variable

The survey points are used to quantify the dependent variable responses in models one and three (in terms of income) and the corresponding changes in the independent variable as represented in models two and four as the predictor selection the high (GDP) level in China cities; green technological developments are discernible. The factors that have been explained are the amount of green technological innovation (GTI) one and two. The quantity of green innovation is now measured by the total number of green patent applications, following Lindsey et al. (2020) and Zhang and Zhou (2020), according to the classification system used by the China National Intellectual Property Administration.

Independent variables

This paper used (Gaglione et al. 2020) crisis strategy conceptualization (Yarovaya et al. 2021) expanded by focusing solely on continuing enterprises. This paper’s main independent variable is green digital finance (GDF). In Peking University (Poortinga et al. 2019), green digital finance index (GDFI) of China is used to measure digital finance as a result. Firstly, green digital finance (GDF1) stands for “green digital finance breadth,” which primarily refers to the quantity of electronic accounts. Second, green digital finance (GDF2), which shows the breadth of usage, which reflects how digital finance is used for credit, investments, insurance, and payments. Thirdly, DIF3 illustrates the degree to which digital finance has been digitized while exhibiting conveniences.

Final control variable and mediating variable

There are five control variables present. We calculate the proportion of R&D employees to all other employees in each sector. Second, we employ the capital-labor ratio (KL). Thirdly, we incorporate a foreign investment (FDI) variable. Second, all technological innovation requires top-notch human resources. According to Acheampong et al. (2020), the quantity of high school students is used to calculate human capital (HC). Third, the local green economic growth may be strongly impacted by a city’s investment structure using the fixed asset investment to GDP ratio. Fourth, the relationship among FDI and green growth is explained by the “emission sanctuary” and “environmental damage bubble” ideas. This paper uses the per capita loan balance of financial institutions which serves as a proxy for financing limitations (FL). The smaller the financial constraint for the China cities, the greater the loan balance per capita.

Data sources and descriptive results

This paper’s main focus on the green digital finance, using panel data from 30 green Chinese cities over the period before January 2000 to March 2017 (Table 1) made up the research sample. The source of the main sample data on green digital finance index (DFI) is (Thomä and Hilke 2018). The data collected mostly from the World Bank, Chinese research data services, and China city statistical yearbook database provided the raw data for the control variables. Table 1 presents the summary statistics.

Table 1 Descriptive analysis results
Full size table

Descriptive analysis

Table 1 column for standard deviation shows the green digital finance (1), the highest value at 0.1298, followed by natural green digital finance (1) at 0.1279. The green technology innovation (1) markets give the lowest standard deviation value at 0.1190.

Empirical results and discussion

Estimated of spatial effects

First, we test for cross-sectional dependency in this analysis. The effect of spatial test developed by Huang et al. (2021) was used. This test compares the null hypothesis (H0) of no significant CSD exit with the alternative (H) of substantiality. Table 2 displays the results supporting either the null or alternative hypothesis. For climate technology, economic industry effectiveness, access to economic organizations, green bonds, and carbon emission per person, the null hypothesis of no effect of spatial test is rejected at the 1 percentage importance level.

Table 2 Multilevel spatial effects results
Full size table

Analysis of the data’s unit root properties or stationarity qualities using (van der Ploeg and Rezai 2020) unit root tests follows an inspection of the effect of spatial test. Whenever effect of spatial test, variability in the slope, or architectural fractures are present, they have offered helpful conceptual and actual ideas for the use of proposed tests. The results of the Pesaran unit root test are shown in Table 2; they corroborate the level-stationarity of all variables, but they do not account for architectural discontinuities in the information. The findings of the cointegration (Kalli and Griffin 2015) test developed by Anser et al. (2020b) indicate, however, that the null hypothesis of data stationarity is not supported. Incorporating architectural changes, as examined by the Pm and P statistics in the bottom half of Table 2, confirms the unit root in all model variables.

Third, using a variant (Garrett-Peltier 2017) introduced by Pesaran and Yamagata, we analyzed the slope heterogeneity in Table 3. An argument that the results under subsequent analysis would be viewed as improper and prejudiced if slope heterogeneity is not addressed provides evidence for the relevance of slope heterogeneity. Another convincing case for slope heterogeneity estimate is made by Balakrishnan et al. (2020). Specifically, both the null and alternative hypotheses were proposed, with the former indicating that the slope coefficients are all the same and the latter supposing that there is some degree of heterogeneity in the slopes. Table 3 shows the results that may be used to argue for or against H0 and H1. Slope heterogeneity is supported by the data, as seen by the meaningful t at the 1% level, indicating that hypothesis 1 is correct.

Table 3 Regression and fixed effects results and GTI and GFIN
Full size table

Assessment of the major findings

Chinese regions throughout the world between 1994 and 2013 analyzed the impact of green technology on carbon emissions, classifying the countries into high- and low-income categories (Pang et al. 2015). An appealing empirical result was presented with the assertion that in low-income nations, green technology innovation does not substantially help to lower environmental deterioration. Han et al. (2022b) do similar research on the Turkish economy, this time looking at how green technology affects carbon output. They found actual evidence supporting the hypothesis that green technology and associated developments do, in fact, lower carbon dioxide emissions. Mi et al. (2019a) have taken into account the development of renewable energy technology and validated its important impact in lowering carbon emissions throughout China’s 26 provinces. All of this explanation should make it clear that the process underlying climate technology and associated technology reduces reliance on conventional innovations, which rely more heavily on conventional energy sources that cause greater ecological damage and results of direct and indirect effects in Table 4.

Table 4 Results of direct and indirect effects
Full size table

Before conducting benchmark regression, it is crucial to conduct tests for multicollinearity and heteroscedasticity. While the p value indicates strong heteroscedasticity, the variance inflation factors (VIFs) suggest that the chosen variables are not multicollinear. For estimation results, and we use the comprehensive feasible generalized least squares (FGLS) method as the benchmark (Mamat et al. 2019) method because it simultaneously accounts for intergroup heteroscedasticity, intragroup autocorrelation, and coincident correlation. Also, substitution analyzed the effects of these three factors on CO2 emissions. Table 4’s last three columns include the outcomes of our estimations.

Empirical estimates in the green finance demand equation

According to the data in the first column of Table 5, increasing energy resilience by 1% may boost CO2 emissions by 0.2%. This contradicts our predictions since we had assumed that enhanced energy resilience would not result in a commensurate increase in CO2 emissions and green finance (Mi et al. 2019b). One possible explanation is that increasing the energy system’s resilience helps assure a steady supply of energy and satisfies market demand, even in the face of shocks, allowing businesses to keep producing as usual despite the disruption. For instance, the worldwide spread of the COVID-19 virus had a significant effect on the world’s energy infrastructure, negatively affecting the extraction of fossil fuels, the importation of foreign energy, and the availability of electricity in many countries. Simultaneously, most nations’ output and operations have been hampered and green technology innovation (GTI) by the city closure policy. Isolated households’ energy use is far lower than that of (Sueyoshi and Goto 2013) commercial companies. Negative financial development rates are the result in most nations, and this substantially lowers CO2 emissions and green technology innovation (GTI). In conclusion, the regular functioning of the market may be ensured by social stability and high energy resilience, which might raise energy use and exacerbate the greenhouse impact.

Table 5 Regression results of the aggregate green innovation
Full size table

Green technology innovation in promoting

Green technology innovation (GTI) nexus has a very positive coefficient of the energy access index, suggesting that easy access to energy may lead to an increase in CO2 emissions, and green technology innovation is shown in Table 6. The ease with which people and businesses can obtain energy has the potential to greatly increase energy usage, particularly in the case of fossil fuels, stable fuel supply, and manufacturing base for (Liu et al. 2021a) the growth of businesses, particularly in the production sector. In particular, highlighting the beneficial impact of energy availability has been validated by a large number of experts (Padilla-Rivera et al. 2018). Second, increased renewable energy resilience may lessen the greenhouse impact, but this is not economically important. The ecological environment may be efficiently protected by the widespread use of renewable energy, which is typically clean and recyclable. However, at the moment, renewable energy is still in the stage of R&D and fast promotion in many nations, and its role in lowering carbon emissions has not been explicitly recognized.

Table 6 Regression results of low value-added and pollution-intensive (LP) sectors
Full size table

Third, from the very last column of Table 6, we learn that a 1% improvement in energy efficiency’s resilience promotes CO2 emissions by 0.177%. This means that better resilience in energy effectiveness exacerbates the greenhouse impact. Energy effectiveness improvements, on the one hand, may boost national economies by GTI, DFI, and FIN maximizing the value of every kilowatt hour of energy used (Sueyoshi et al. 2017). High energy effectiveness, on the other hand, may hasten the transition of the economy from the capital- to labor-intensive sectors, boost employment for locals, and encourage the launch and operation of businesses, all of which contribute to a reduction in the global greenhouse effect.

Both financial expansion substantial connections concern the control factors; in other words, financial development decreases CO2 emissions at the outset but then has a positive effect on CO2 emissions after it reaches a certain threshold. This runs counter to the EKC hypothesis’s final verdict. The growth of CO2 emissions may be accelerated by several factors, including changes in commercial architecture, financial aggregate, labor (Bai et al. 2018) force, and urbanization. Large amounts of energy are needed for things like the expansion of the service sector, the growth of the economy as a whole, the availability of a large pool of potential workers, and the progress of urbanization.

Given the presence of heteroscedasticity, using the ordinary least squares (OLS) technique and a strategy is more reliable and effective than others for handling heteroscedasticity, making it a better choice for use in general. Standard errors and regression coefficients are consistently estimated (Ramli et al. 2016). Table 7 displays the actual findings, which use both total factors; importance is consistent with, as shown above. This helps to highlight the reliability of the empirical results.

Table 7 Calculation of the goodness-of-fit index
Full size table

The control variables are also worth discussing in this context. Concerning the socio-economic development of the focal province, the impact of the urbanization level (Chen and Zhang 2021) was found to be significantly positive, but the impact of this urbanization was negative for adjacent provinces. These results indicate that socio-economic development in focal areas is fueled by the urbanization process, reflecting the central role played by urbanization in economic and social development more generally, as seen in Fig. 1. In the meantime, renewable energy (RE) accounts for 49.6% (0.586) of the change in green innovation.

Fig. 1
figure 1

Renewable energy fluctuation in green technology innovation

Full size image

The “digital divide” is the gap in access and usage of information and communication technologies (ICT) that exists between people of different socioeconomic backgrounds, living in different regions. There is a widespread digital gap across areas, and its very existence has stymied progress in many places (Alemzero et al. 2020). In this scenario, a province with a high rate of technology development growth might operate as a magnet for skilled workers, investors, and cutting-edge equipment, therefore, hastening the circulation of all three within the area. Since various regions adopt and upgrade their technologies at different rates, the existing economical division will only expand. This variation mirrors the trend that emerged from this study’s findings: advancements in information and communication technology in one location tend to stifle financial growth in neighboring areas. In addition to the digital divide, the delay in the widespread adoption of ICT is a contributing factor in the emergence of paradoxical outcomes. It is hard to gage the long-term benefits of ICT on surrounding communities. Direct good benefits from nearby places do not immediately manifest. Positive results may be expected from ICT adoption; nevertheless, not all areas will experience the full advantages of ICT immediately (Mayer et al. 2020). Nonetheless, it is assumed that the negative effects of technology development in the focus area may be mitigated and converted into an active influence, given the prevalence of the good benefits of technology innovation. Long-term implications of green technology innovation (GTI) are also taken into account, which leads to a wider variety of technological innovation applications and a more even distribution of technology development across sectors.

Green finance and renewable energy estimations

With the acceleration of peri-urban urbanization, a fast-growing area may continually expand, increasing the urbanization of land and leading to population increase of total population due not only to the natural growth of the area but also to population migration by those in pursuit of better job opportunities and high-quality public services. The result is an overall improvement of the area’s socio-economic level, consequently, a dampening of the development and green finance and renewable energy in neighboring regions, at least to some extent in Table 8. This interpretation can also be used to explain the effect on socio-economic development but in the opposite way. This is because a higher population density will negatively impact socio-economic development in focal areas but may enhance levels of development and green finance and renewable energy in neighboring areas.

Table 8 Robustness test results with the alternative financial constraint index
Full size table

Table 8 summarizes findings on the robustness test with the alternative financial constraint index. Changes in the explanatory variable in study regions have consequences for the explained variable, which are reflected in the direct impacts. Indirect effects reveal how changes in the explanatory factor influence social and financial growth in neighboring (Mafauzy 2000) regions. When we talk about the “total impacts,” we are talking about the aggregate of the direct and indirect effects, which has an averaged influence on the overall macroeconomic growth. As a result, the earliest stages of the outbreak had a beneficial effect on ecological quality by decreasing CO2 emissions and pollutants. Carbon emissions increase and decrease with financial activity in the short run, as shown by the research. The second theory is the Pandemic Pollution Haven theory, which states that an increase in GTI inflows in the decrease of coronavirus infections will increase carbon emissions owing to the increase in commercialization activities in nations. Our findings are entirely consistent with this theory. Our findings also provide new light on the link between DIF emissions and the incidence of COVID-19, showing that the association is statistically negligible over the medium term. We determine that the duration of this connection is either brief (caused by economic activity) or indefinite (due to containment measures).

Influence mechanism analysis

The connection between carbon dioxide emissions (CO2), carbon allowances, and productivity (P) explains these patterns over the medium term. Research shows that coal consumption increases as allowance prices. Low allowance prices encourage the purchase of permits and, by extension, increased CO2 input. In the wake of the COVID-19 pandemic, a sudden spike in oil demand developed as a result of the worldwide decrease in supply. When oil demand unexpectedly falls, as it does in response to negative demand shocks, the price of oil falls as well. As a consequence, the price of carbon permits dropped in the near and medium term, as did the demand for such permits. The FIN findings of analysis of influencing mechanism in Table 9. More assets spent on clean energy means less pollution, as seen by the data, which points to a substantial negative comovement among GDF and GTI over the long term. Awareness of the relative concerns of global warming among administrations, shareholders, and citizens has been raised as a result of these initiatives. The analysis confirms that by using renewable energy, we can cut our CO2 emissions by as much as 70% points by the year 2040.

Table 9 Analysis of influencing mechanism
Full size table

Robustness test analysis

This section evaluates how robust our main findings are. While environmental pollution and GTIs are also influenced by other factors, such as economic growth and green digital finance expenditure levels, spatial weight matrix W1 represents the influence of geographic closeness. To fully and accurately depict geographical spillover effects, spatial weight matrices from various viewpoints are required. Economic links are often thought to be stronger the closer the local and surrounding regions’ economic growth rates are. As a result, it is possible to develop the spatial weight matrix of economic distance (W2) and green digital finance matrix (W3) in accordance with regional differences. The robustness tests of the aforementioned empirical findings are also shown together with the predicted regression coefficients of the dynamic panel model under W2 and W3. Table 10 displays the robustness tests for spatial decomposition effects and demonstrates that the significance and sign of the regression parameters do not vary considerably.

Table 10 Total effects of green technology innovation’s variables
Full size table

The significance test is not passed by the GTI coefficients in columns (2) and (4), which are 0.167 and 0.81 respectively. In comparison, columns (1) and (3) had DFII coefficients of 0.457 and 0.408 with significant levels of at least 5%. All of these point to the fact that regions with strict financial and environmental regulations will experience a greater impact from digital banking on regional GFIN. In other words, the financial and environmental oversight provided by the Chinese government can encourage digital finance to contribute positively to energy conservation and emission reduction.

Conclusion and policy implications

Main findings and conclusion

This article examines the innovation in the green sector frequently facing a financial conundrum. Production of renewable energy is the eighth sustainable development, based on data from (2000 to 2017) the 30 Chinese provinces; we examine the effects of green digital finance on green innovation on protection of environment using influence mechanism analysis. Digital finance, which has become a major driver of green innovations in China, may first increase the number and quality of green technical innovation.

Secondly, the empirical study demonstrates that the promotion effect of digital finance on the efficiency of renewable energy markets is greater than the inhibitory effect, making the total effect less obviously favorable. In other results, the elasticity of lnGFDI is significant at the 5% level and is 0.1545 and 0.1880 in the present and 1-year delayed periods, respectively. Further, the average total effect of FDI on the effectiveness of green innovation is 0.008, with an average encouraging effect of 0.0051 and an average inhibiting effect of 0.0039. Further, the average total effect of FDI on the effectiveness of green innovation is 0.008, with an average encouraging effect of 0.0051 and an average inhibiting effect of 0.0039. Our empirical findings illuminate the need for coordinated urban digital transformation and green economic development in China.

Thirdly, empirically demonstrate that green technology innovation is a powerful catalyst for promoting the cleaner upgrading directly, both for the overall industry and for various groups of industrial sectors, with the exception of the LC sectors, where the green innovation-driven effect is not significant. It shows that Chinese businesses have amassed a wealth of experience in developing and implementing such cutting-edge technologies, and they have made outstanding advancements in green technology. Additionally, the capital-labor ratio and comparative study of the coefficients of green innovation show that China’s industry is moving away from a model that is driven by unsustainable factors and toward one that is driven by green innovation.

Policy implications

This study offers empirical support for the use of digital finance to foster green technological innovation, opening up new perspectives on green innovation promotion in China.

  1. 1.

    China’s current scenario nevertheless has several practical issues, such as poor innovation quality and uneven regional development.

  2. 2.

    The issues raised above can be effectively resolved with the help of digital finance. We offer some policy recommendations in light of the aforementioned findings. Digital infrastructure is required to support digital money.

  3. 3.

    In China, further promotion of digital infrastructure construction is required. A number of new digital technologies, including 5G and the Internet of Things, need to be widely adopted, and investment in research and development must increase.

  4. 4.

    Not to mention, one of the main objectives should be to promote the concentration of cutting-edge resources in the high value-added and greener industries, followed by the increase of their production scales. Given that the HC sectors look to be a leader in the green-innovation-driven paradigm, the Chinese government should seek to construct and strengthen a long-term development pattern of inter-sectoral collaboration between green innovations. The growth of digital finance not only broadens sources of capital for businesses to innovate but also offers residents easy access to loans. Rich liquidity encourages locals to purchase specific high-emission commodities, including cars and air conditioners, which causes large carbon emissions.

Future research

Future studies may examine the degree to which the consumption effect cancels out the innovation impact, which has received a significant lot of attention in our work, and how this affects the green innovation.