Introduction

As industrialization and urbanization continue to advance, human activities and rising energy demand have led to increased carbon emissions, exacerbating a series of climate chain reactions characterized by climate warming (Aslam et al. 2021; Tan et al. 2021; Razzaq et al. 2023). China has pledged to achieve its carbon peak by around 2030 and achieve carbon neutrality by 2060 to actively combat global warming, fully commit to energy saving and resource efficiency, and uphold the Paris Agreement (Irfan et al. 2022; Saqib et al. 2023; Liu et al. 2022a). This target indicates that as China’s environmental carrying capacity is decreasing, the traditional development model of not taking into account environmental costs is now unsustainable, and full attention needs to be paid to total factor carbon productivity, which considers carbon emission as undesired output (Wang and Yi 2022; Hao et al. 2021; Cheng et al. 2022). Given the idea that green development was first explicitly put forward in the 13th Five-Year Plan, which aims at low-carbon development by building a green financial system, developing green trusts, securities, funds, etc. the 19th National Congress report proposed “building a market-oriented green technology innovation system, developing green finance, and growing clean production, clean energy, and energy conservation and environmental protection industries.” The Chinese government is the first market economy in the world to establish a systematic theoretical research framework for green finance policies and measures, but the current in-depth research on green finance policies and measures focuses on qualitative analysis methods, with relatively little empirical evidence, which lags behind the perfect policy system of green finance GF (Lv et al. 2021a, b; Cao et al. 2022). At the stage of the 14th Five-Year Plan, under the ambitious goal of “double carbon,” improving carbon emission efficiency has become a key factor for high-quality economic development, and investigating the value, efficacy, and repeatability of green finance in the conservation of energy and emission control is extremely important from a practical standpoint (Chai et al. 2022).

As an essential measure to support the green development system through financial supply-side reform (Lee and Lee 2022), could GF serve with both objectives of “carbon peak” and “carbon neutrality”? In this research, we measured the GF and carbon emission efficiency (CEE) data and explored the mechanism, heterogeneity, and influence of GF on CEE by using the dynamic panel model and spatial econometric model. The paper aims to offer a theoretical basis and empirical support for GF to benefit the construction of a green and low-carbon economy under the dual carbon target.

The marginal contributions of this paper may be: firstly, it expands the research framework of GF on CEE and explores the mechanism and effect mechanism of GF on CEE; secondly, it distinguishes the heterogeneity of the influence of GF on CEE in cities with diverse geographical locations, economic development levels, and green finance levels, and provides a reference for cities to implement differentiated green finance policies to help green low-carbon development.

Literature review

The theoretic research on GF has progressed relatively slowly, mostly from the perspectives of the implication of GF, the creation and evaluation of the system, and the theoretical exploration of future development (Sachs et al. 2019; Berrou et al. 2019; Migliorelli and Dessertine 2019). Until the last 5 years, as green development has received great attention again, the study on GF has become active, and the relationship between GF and green development has become a core issue in this field (Zhang et al. 2019). “Financing of investments that produce environmental advantages” is the definition of GF. Since its inception, international organizations and national governments have been having extensive discussions about GF, which is a clear indication of its crucial role in policy (Lindenberg 2014). Academic researchers’ interest in the topic has also grown. Zhang et al. (2021a, b) claim that green credit rules reward “high and low” enterprise short-term financing activity while penalizing long-term financing and investment behavior and also help reduce sulfur dioxide and wastewater discharge. More recent empirical studies tend to argue that GF can play a positive role in curbing heavy polluters and improving the ecological environment (Wang and Zhi 2016; Wang et al. 2021a, b; Huang and Chen 2022). Some studies have also explored the connection between economic growth and carbon emission. The results confirm that GF is the best financial strategy to reduce the output of carbon emissions. According to Ren et al. (2020), China’s industry of GF grew quickly, and advancements in the GF index and rising non-fossil energy usage helped to lower carbon intensity.

To reach China’s objective of becoming carbon neutral by 2060, the need for the improvement of CEE is becoming more and more recognized by academics both domestically and internationally. In the previous related studies, the research direction and perspective emphasize primarily three aspects, the definition of CEE, the measurement of CEE, and the influencing factors. Generally, CEE is divided into single-factor CEE and full-factor CEE. The three basic categories used to classify single-factor CEE. One is carbon productivity, which Kaya and Yokobori initially planned to express carbon productivity considering the GDP to CE ratio (Kaya and Yokobori 1997), i.e., the value of GDP generated per unit of CE. Second is the carbon index, which was put up by Mielnik and Goldemberg (1999) to evaluate CEE in terms of the proportion of overall energy use to overall CE, i.e., the CE for each unit of energy consumed. The third component is carbon intensity, or CE per unit of GDP growth (Ang 1999). Because single-factor efficiency can only represent how energy and economic efficiency are related, ignoring the consideration of the effect of other factors like capital, labor, and technology on the significance of economic growth (Lv et al. 2021a, b), total factor productivity takes into account the situation of multiple inputs (Zhao et al. 2022a, b; Wang et al. 2021a, b), and the calculation results are more scientific and reasonable than single-factor efficiency, scholars began to use the stochastic frontier analysis (SFA model) of parametric method and the non-parametric method of data envelopment analysis (DEA model) to mainly measure the total factor carbon efficiency to conduct research. In previous studies, it is generally agreed that openness to the outside world, government intervention, enterprise ownership structure, technological progress, and enterprise size have favorable effects on CEE; economic size, energy consumption structure and industrial structure, urbanization rate, and endowment structure have greater negative effects on CEE (Sun and Huang 2020; Yu and Zhang 2021).

To sum up, the literature already in existence has provided some groundwork for the study of GF and CEE, but there are still some gaps: firstly, scholars focus more on theoretical studies related to green finance, and there are relatively few empirical studies; secondly, there are studies related to the association between GF and green development, but the influence of GF on the overall performance of carbon emission is neglected. Therefore, using data from China’s provincial panel between 2007 and 2019, this paper measured the green finance index by the entropy method and the CEE with carbon emission as the non-desired output by the Super-SBM model.

Theoretical analysis and research hypothesis

The direct influence of GF on CEE

China’s economy has transitioned from fast growth to high-quality development; green finance has received increasing attention to find an effective power source during the economic transformation. Zhou et al. (2020) discovered that green financing can result in a situation where both economic growth and the environment benefit. They did this by using global principal component analysis to build a GF development index. According to Soundarrajan and Vivek (2016), GF is an essential component of the low-carbon green economy. It is a market-based loan or investment project that evaluates risks in light of environmental impact or uses environmental incentives while making business choices. It thus connects finance, environmental improvement, and economic growth. Iqbal et al. (2021) argue that to control global warming, the transition to green energy requires a lot of GF. Green finance could aid in lowering the environmental pollution. Among the intellectual studies on the influence of GF on the output of carbon emission, some researchers have pointed out that financial development could lower the output of carbon emission through paths such as encouraging business technology innovation and raising awareness of nature conservation (Zhang et al. 2021a, b). Using data from different countries, Boutabba (2014) found that financial deepening suppressed the output of carbon emissions. Xiu et al. (2015) argued that financial regulation initiatives help to encourage saving and lowering pollutants under industrial development constraints. Increases in the use of renewable power, real exchange rate, and financial deepening all result in a drop in carbon pollution, according to Dogan and Seker (2016), while increases in the use of non-renewable energy increase the amount of pollution. Ren et al. (2020) investigated the improvement of the growing index of GF and the utilization of non-fossil power. Shen et al. (2021) investigated that green inputs showed a negative relationship with carbon dioxide emissions. Chen and Chen (2021) investigated that after the introduction of the GF policy, the interest-bearing corporate debt investment and corporate gearing of heavy polluters were significantly reduced, the corresponding financing cost became larger, and the new incoming investment was significantly reduced. When the investment is restrained, highly polluted companies will have less capacity to produce, and energy consumption and pollutant emission will be reduced at the same time.

According to the constraint theory, green investment as a special fund for pollution control has special characteristics and cannot be invested in production and operation (Ren et al. 2022a, b). Under the condition that the existing capital and labor factors remain unchanged, early on in the creation of GF, green investment is low, and enterprises need more energy consumption for regional economic growth to maximize short-term profits, which results in an energy rebound effect and leads to the rise in the output of carbon emission. In line with the scale effect theory, with the improvement of GF, the scale of green investment expands, the production process of green enterprises becomes more and more perfect under the effective control of R&D cost, and the energy utilization efficiency is significantly improved, which can effectively lower the output of carbon emission in the production process. The findings of the current study suggest that the advancement of GF will be beneficial to enhance the performance of carbon pollution on the whole. Thus, the following hypotheses are proposed:

  • H1: GF improves the CEE.

Mechanisms of GF’s impact on CEE

With the release of green bonds and trusts, GF has created a new avenue for financial investment in green sectors, such as those that conserve energy and protect the environment (Falcone 2020). The gray correlation approach is used by Wang and Wang (2021) to empirically investigate the association between GF and upgrading of the industrialization in China. The goal of green finance is to assist environmentally friendly initiatives financially, advance technical advancement, and foster environmentally friendly and sustainable economic growth. For China’s “three high” excess industries, the investment environment risk will further increase the investment environment cost of enterprises, strengthen the government investment environment constraints, and curb the financing of high energy-consuming enterprises; green finance will use the effect of government investment environment penalties and corporate financing environment, inhibiting influence to encourage the green transformation of the economy so that the unit energy consumption intensity significantly decreased, the productivity of the unit resource consumption corresponding to increase (Qi and Qi 2020; Du et al. 2022). Green finance promotes the green and ecological life and consumption of residents, influences the activities of enterprises by using the requirements of green consumption at the end and market-based economic incentives, forces the optimization and upgrading of front-end industries, increases the efficient supply of green and ecological goods, and improves the efficiency of energy consumption (Sun and Chen 2022). Green financial capital is tilted toward growth and innovative industries, and market players are guided by social expectations to spontaneously encourage improving and modernizing industrial structures, which in turn promotes green development. Gu et al. (2021) established the VAR model, super efficiency DEA, and the Tobit regression model. Empirical examination shows that GF is quite effective at encouraging industrial modernization and transformation.

Green finance itself belongs to the service industry; firstly, its development improves its gross product and thus increases the proportion of the three industries; secondly, it promotes the migration of production factors to more efficient sectors and encourages the modernization of industrial infrastructure, and the three industries develop to contribute to a high degree of economic efficiency, low consumption, and less pollution; it makes the allocation of financial resources more effective and encourages the rationalization of industrial structure and thus improves the efficiency of various resource factors such as energy. It makes the allocation of financial resources more effective, advocates the rationalization of industrial structure and thus enhances the application efficiency of various resource factors such as energy, promotes economic and social development, and has a favorable impact on carbon emission performance (Wang et al. 2019). Accordingly, we suggest the following scenario:

  • H2: GF improves the CEE by promoting industrial structure upgrading.

Study design

Model construction

Using the aforementioned theoretical study and research premise as a foundation, we may investigate how GF affects CEE. This paper constructs a dynamic panel model as follows:

$$CE{E}_{it}={\beta }_{0}+{\beta }_{1}CE{E}_{it-1}+{\beta }_{2}{\mathrm{green}}_{it}+{\beta }_{n}{X}_{it}+{\varepsilon }_{it}$$
(1)

In Eq. 1, \(i\) denotes city, \(t\) indicates year, CEE denotes carbon emission efficiency, green denotes green finance, X denotes control variable, and \(\varepsilon\) denotes random disturbance term.

In addition, utilizing the previous mechanism of action study, in addition to the above-mentioned direct effects, to prove that GF increases CEE by optimizing industrial structure, this paper draws on Wu et al. (2021) to test the mediating effect and constructs the following equation in three steps:

$${{\mathrm{indup}}_{it}=\varphi }_{0}+{\varphi }_{1}{\mathrm{green}}_{it}+{\varphi }_{n}{X}_{it}+{\varepsilon }_{it}$$
(2)
$${CEE}_{it}={\vartheta }_{0}+{\vartheta }_{1}{\mathrm{green}}_{it}+{\vartheta }_{2}{{\mathrm{indup}}_{it}+\vartheta }_{n}{X}_{it}+{\varepsilon }_{it}$$
(3)

Equations 1, 2, and 3 constitute the mediating model. \(indup\) represents the mediating variable, and if \({\varphi }_{1}\) and ϑ2 are significant simultaneously, then there is a mediating effect φ1 × ϑ2 on the impact of GF on CEE.

GF and CEE in one region may also affect neighboring or more interlinked locations by spatial spillover effects. By using the research conducted by Le Sage and Pace (2009), with this essay, the spatial Durbin model is chosen and set up as follows:

$$CEE_{it}=\alpha_0+\rho W\cdot CEE_{it}+\alpha_1\;green_{it}+\Sigma\;aX_{it}+\theta W\left(green_{it}+\Sigma\;\alpha X_{it}\right)+\gamma_t+u_{it}+\varepsilon_{it}$$
(4)

where ρ depicts the lagged regression coefficient of explained variables, θ depicts the lagged regression coefficient of explained variables;\(\gamma\) represents the time-fixed effect;\(u\) depicts the individual fixed effect; and W is the economic spatial weight matrix, which preferably able to capture the level of economic asymmetry among areas.

Variable measures and descriptions

Explanatory variables

Carbon emission efficiency (CEE).

Utilizing the findings of Ge et al. (2022) and Meng et al. (2022), the essay constructs a measurement system of CEE based on the Super-SBM model, using DEA Solver Pro 5.0 software, as shown in Table 1. Among them, following Ren and Wu (2022), the capital stock is used to compute the capital, using 2006 as the starting point and the “perpetual inventory method.” Following Wu et al. (2020), for labor force indicators, the total number of employed persons in the three industries in the current year is selected for measurement. The expected output is the gross domestic product (GDP) of each province and region, and the real GDP of each region is deflated by taking 2006 as the base period to account for price changes. The non-desired output factor is the output of carbon emission of each province, and the output of carbon emission from fossil fuel consumption in 30 provinces in mainland China is measured by the approach outlined in the IPCC Guidelines for National Greenhouse Gas Inventories from 2006 (Hao et al. 2023).

Table 1 The measurement system of CEE
Full size table

Explanatory variables

Green finance (green)

In this study, as shown in Table 2, according to the data obtained, five categories of financial services using the entropy value method can calculate the index of GF.

Table 2 Green financial evaluation index
Full size table

Mediating variables

Industrial structure upgrading (indup)

China’s economy has developed to this stage, and industrial development shows a significant increase in the development rate of the tertiary industry. Referring to Liu et al. (2021), industrial structure upgrading is quantified using the ratio of the tertiary industry’s value added to the secondary industry’s value added.

Control variables

Considering that CEE is impacted by a variety of things, following Xiao and Liu (2022), Liu et al. (2022a, b), and Ren et al. (2022a, b), some control variables are introduced in the paper to reduce the bias caused by omitted variables. Among them, the logarithm of per capita GDP is applicable to determine the amount of economic growth (lnpgdp); the degree of foreign openness (lnfdi) is determined by taking the logarithm of foreign direct investment; and government intervention (gov) is the government’s share of GDP as measured by spending.

Data sources and descriptive statistics

The primary sources of the original data are the China Statistical Yearbook, China Energy Statistical Yearbook, and historical statistical yearbooks of the provinces, autonomous areas, and municipalities directly under the central government. Among them, the data of green financial index measurement are from the annual statistical reports of the banking industry in the past years, Wind and CSMAR databases. Table 3 displays the descriptive statistics for the primary variables.

Table 3 Descriptive statistics of main variables
Full size table

Empirical results and analysis

Baseline regression results

This essay first uses a dynamic panel model to assess the influence of GF on CEE. Columns (1) and (2) of Table 4 show, correspondingly, the statistical outcomes of the OLS model without and with control variables utilizing dynamic lagged variables. Columns (3) and (4) show the empirical results of the system GMM model. The CEE regression coefficients for the lag period may be shown to be both statistically notable and favorable at the 1% level, indicating that the CEE during the present time is significantly influenced by the value of the previous period, which has a strong cumulative circularity, which also indicates that the estimation of dynamic panel model is reasonable. All of the GF’s coefficients on CEE are significant and positive, and thus hypothesis 1 is valid. Unlike the first-order autocorrelation statistics, the p-value of the second-order autocorrelation statistics is greater than 0.1. Therefore, the original hypothesis of “no autocorrelation of the disturbance term” is accepted; and the Hansen statistic is not significant, which indicates that the selected instrumental variables are valid and there is no over-identification. Our findings are similar to those of Zhao et al. (2022a, b), Lu et al. (2022), and Sharif et al. (2022). GF, through green loans and securities, broadens the financing options available to businesses engaged in energy conservation and environmental safeguarding to support their development (Meo and Abd Karim 2022; Sun et al. 2022). For the “three high” industries, their environmental protection business risks will increase the cost of corporate financing, increase the investment and financing constraints, and curb the input of high energy-consuming SMEs (Soundarrajan and Vivek 2016; Xiong and Sun 2022). Green finance accelerates the green transformation through the financing penalty effect and investment disincentive effect, which is also conducive to the enhancement of CEE.

Table 4 Baseline regression results
Full size table

Analysis of the intermediary effect

Theoretical studies suggest that green finance enhances carbon performance through industrial structure effects. To determine if this mechanism is reliable, we employ the mediating effect model. In Table 5, we can see that GF positively influences industrial structure improvement and that both green financing and industrial structure improvement positively impact CEE, which means that GF can improve CEE by promoting industrial upgrading; thus, hypothesis 2 is valid. Green finance itself belongs to the service sector and its development has increased its gross product and thus the share of the three sectors. Green finance facilitates the migration of factors of production to more efficient sectors and encourages the modernization of industrial infrastructure (Gu et al. 2021; Sun et al. 2022). It makes the allocation of financial resources more efficient, encourages industrial structure upgrading, and has a favorable impact on carbon emission efficiency (Wang et al. 2019).

Table 5 Mediation impact of GF on CEE
Full size table

Analysis of spatial spillover effects

In this essay, utilizing the research done by Xu et al. (2022), we use the economic spatial weight matrix and discuss the spatial clustering status of the model explanatory and explanatory variables through the global Moran index. The outcomes are displayed in Table 6. In the table, it is apparent that the observed values of GF and CEE are positive in the observation period of 2007–2019 (except for 2018), which indicates that the observed values have a considerable positive spatial connection between the two.

Table 6 Moran’s I for GF and CEE
Full size table

Within this essay, we shall choose a spatial econometric model that is appropriate for this paper and evaluate its reliability. The study uses the approach described by Le Sage and Pace (2009) to test whether there is spatial autocorrelation, the spatial autoregressive model passes the LM statistic and its robust form. Secondly, the Wald test is used to compare the applicability of the SDM. The results of the aforementioned experiments are presented in Table 7, and it is clear that the SDM is the best model.

Table 7 Identification and testing of spatial econometric models
Full size table

The spatial Durbin model’s regression outcomes are displayed in Table 8, from which we can see that the main utility coefficients of GF are all beneficial and notable at the 1% level, indicating that GF has been shown to considerably improve regional carbon efficiency, which again verifies hypothesis 1. Additionally, the coefficient of W × green is substantial and positive, demonstrating that GF has a favorable geographical spillover impact, which not only improves local CEE but also significantly contributes to the economic. The coefficient of W × green is also significant.

Table 8 Spatial Durbin model regression results
Full size table

Heterogeneity analysis

Although it is clear that GF and CEE are strongly correlated on a global scale and may differ significantly depending on the region’s geography, level of economic development, and intensity of GF policies, the regression results of GF heterogeneity on CEE are displayed in Table 9. Firstly, to examine how GF affects local CEE, the data are split into eastern and central-western samples. It is evident that GF considerably improves CEE at the 5% level in eastern provinces, but not in the central-west. The possible reason is that the capital market in the eastern region has been more mature through its long-term development, which is more likely to attract scarce financial resources to the east and has profoundly influenced the economic growth and capital pattern of the country; the Midwest areas are also developing their economies based on energy consumption, but the development level of GF lags behind that of the east and central regions, so a mature financial market has not yet been formed, and it cannot have a notable influence on the progress of regional CEE. The development of GF in the Midwest areas is also founded on energy consumption.

Table 9 Heterogeneity of the effect of GF on CEE
Full size table

To further clarify the influence of GF on regional CEE at different development stages, the total sample was divided into two smaller samples of high and low economic development levels following the median GDP per capita and then divided into two smaller samples of high and low GF levels following the median GF index, and regression analysis was conducted separately. The regression analysis shows that the effect of GF on CEE is positive in areas with a high economic level and high GF level, but not notable in areas with a low economic level and low GF level. One explanation could be that the economy is still in its infancy, and the high speed of development must consume huge energy resources and form huge carbon emission, so the environmental performance decreases; when the development reaches a certain amount of living level, citizens and government departments will increasingly focus on environmental health and seek green development, and government departments will focus on pollution control and promote carbon emission reduction. It is also in line with the “Environmental Kuznets Curve” (EKC) effect, which has been widely studied (An et al. 2021). The application of GF will push for improved industrial structure, improve the efficiency of factor allocation, and control carbon emission while economic development (Xiong and Sun 2022).

Robustness tests

Robustness tests for samples that may cause interference are excluded

In this paper, 30 provinces and cities in mainland China are selected as the sample, among which Guangdong, Zhejiang, Jiangsu, and Shandong provinces are among the more developed provinces with the top four GDPs in the 30 provinces. Because of their rapid economic development, extensive capital markets, large financial market volume, and active financial activities, which may obstruct the study for this work, the specimens of these four provinces are firstly excluded from this paper, and the results of the re-run regression are shown in Table 10, and the main variables’ signals and importance match those in the earlier paper, which proves the robustness of the results.

Table 10 Robustness test I: eliminating samples that might interact
Full size table

Excluding the first-year data robustness test

The time frame of this paper is 2007–2019; the 2008 global economic crisis prompted the Chinese government to modify the extent and scope of financial reform, focusing on prudential supervision of the financial system on the macro level, tightening the credit risk audit of banks, tightening the risk management and internal control of commercial banks, and changing the capital and financial markets compared to 2008. To avoid the possible disturbance of data before 2008, the paper excludes the sample data from 2007 and adjusts the data before estimating the parameters again. Table 11 again proves the robustness of the empirical results, and the sign and significance of the main variables remain the same as above.

Table 11 Robustness test II: excluding the first-year data
Full size table

Main conclusions and policy recommendations

Using Chinese provincial panel data from 2007 to 2019, this paper measured the GF index using the entropy method and carbon emission efficiency with carbon emission as non-expected output using the Super-SBM method, explored the influence of GF on CEE using the dynamic panel model and spatial Durbin model with full consideration of the mechanism of the variables’ effects, and reached the following research conclusions: first, GF significantly improves CEE; second, the modernization of the industrial structure is a key mediator in the transmission of GF to CEE; third, GF significantly reduces CEE in eastern regions, high economic development level, and large cities with strong GF policies, while it does not significantly affect CEE in the Midwest, low economic development level, and low level of green finance policy support. Fourth, GF has a notable positive spatial spillover impact on the CEE of provinces with close economic ties. The findings of this paper expand the mechanism of the impact of GF on CEE, and the following are the policy ramifications.

First, the orderly guidance of GF helps low-carbon economic growth. In the process of promoting the construction of GF, the flow of financial resources and capital into new enterprises must be increased comparatively. In the eastern region, we should promote the optimization of industrial structure, attract green capital to support technological innovation in the energy field, strengthen the construction of energy networks, and improve energy utilization efficiency by various means such as reducing losses caused by energy transmission over long distances; in the Midwest regions, we could further accept the industrial migration from the eastern development regions, empower China’s traditional resource factors with data factors, activate economic potential, optimize the industrial structure, and improve the utilization efficiency of various resource factors. In the Midwest regions, we should further accept the industrial migration from the eastern development regions, empower China’s traditional resource elements with data elements, activate economic potential, optimize the industrial structure, improve the efficiency of various resource elements, and achieve green development.

Second, we should reduce government intervention in the financial market and incorporate green GDP into the government performance appraisal to prevent the detrimental effects of government “bottom-up competition” on regional CEE, break the path dependence created by traditional high energy-consumption industries and “zombie firms,” strengthen the performance management of both energy efficiency and ecological sustainability, increase and rationalize the use of local government spending on preventing pollution and pollution prevention, and stimulate local governments to promote green and ecological economic development. Similarly, they are required to set “green technology thresholds for environmental protection” and strict energy-saving access when introducing foreign investment in construction projects to tighten the construction of high-pollution, high-energy-consumption projects from the root.

Third, we should strengthen exchanges and cooperation. Green finance can become an essential tool to enhance the CEE and can better support sustainable economic development. China has a relatively mature green financial policy system, which should strengthen inter-regional exchanges and cooperation. Regions with rich practical experience in the field of GF should play a leading role in actively promoting GF policies so that they can be gradually promoted and implemented nationwide to better cope with global environmental pressure.