Financial Development, Heterogeneous Technological Progress, and Carbon Emissions: An Empirical Analysis Based on Provincial Panel Data in China

Abstract

Global warming, caused by an increase in carbon emissions, has attracted considerable attention worldwide. In addition, financial development affects technological progress and carbon emissions. Despite numerous works that explore the impact of financial development on technological progress and carbon emissions, few have integrated the three into a unified framework of research. To fill this gap, this study constructed a mediation effect model for empirical analysis based onthe provincial panel data of 30 provinces and cities in China from 2009 to 2021.Taking into account the regional differences across China, this study explored the impact of financial development on carbon emissions and the intermediary role that heterogeneous technological progress plays within. The results showed that at the national level, and in eastern and central China, the relationship between financial development and carbon emissions conformed to an inverted U, the environmental Kuznets curve, whereas in western China, carbon emissions were found to linearly increase with financial development. Among the variables of technological progress that served as mediators, generalized technology progress, environmental technology progress, energy technology progress, capital embodied technology progress, and FDI technology spillover were the transmission paths for the impact of financial development on carbon emissions at the national level. However, the effect of these variables of technological progress on the impact of financial development on carbon emissions varied among the different regions. This paper aims to provide some inspiration to reduce carbon emissions through financial development and prevent “one-size-fits-all” policies for technological advances or overall planning without considering regional differences.

1. Introduction

The fifth anniversary of the Paris Agreement was on 22 April 2021, which marked the beginning of the climate summit attended by world leaders. At the summit, climate change, renewable energy, and green economic development became the focus of discussions and received attention from all countries [1]. In the speech, “Jointly Building a Community of Life between Man and Nature“ that Chinese President Xi Jinping delivered at the summit, he elaborated on the importance of China, atypical developing country, to shoulder the mission of promoting human-nature harmony and liberating productive forces. Because of its reform and opening-up policy, China has achieved rapid industrialization, urbanization, and economic growth in the early stage of development, but at a cost ofhigh energy consumption and environmental pollution. The Environmental Performance Index (EPI) assessment results released by global authorities in 2020 showed that China’s EPI ranking remained low across the globe. Meanwhile, as extreme weather conditions increase, it is important to improve energy utilization, accelerate technological progress, reduce carbon emissions, and reach the carbon peak as soon as possible [2,3,4]. In China’s Standards Plan, economic development, sci-tech progress, and ecological conservation are given equal importance. Developing a low-carbon economy is the only way toward energy conservation and emission reduction, and economic development relies partly on financial development. Finance plays a crucial part in optimizing the allocation of funds, guiding industrial restructuring, and improving the marginal productivity of capital in modern economic systems [5,6]. Many scholars have put the regional financial development and low-carbon economy within the same frame work of research and explored the relationship between them; however, no unified conclusions were reached [7,8]. In view of this, this study selected the provincial-level panel data of 30 provinces and cities in China from 2009 to 2021 as samples. The samples were divided into sub-samples by region to further explore the relationship between financial development and carbon emissions, as well as the intermediary role of heterogeneous technological progress.

The remaining parts of this paper are arranged as follows: Section 2 is the literature review;the methodology and data are described in Section 3; Section 4 presents the empirical results, and Section 5 provides the conclusions and policy recommendations.

2. Literature Review

Previous works have confirmed the impact of financial development on carbon emissions. Grossman et al. and Neaguet al. explored the changes in environmental variables in the early stage of economic growth and proposed the inverted U-shaped environmental Kuznets curve (EKC) that embodies the relationship between economic development and environmental pollution [9,10]. Financial development and economic development are in a mutually-reinforcing relationship and financial development promotes technological progress and improves the total output of domestic goods and services [11,12]. Levine et al., believing the inevitability of “market failure”, argued that finance increases the efficiency of resource allocation and plays an intermediary role in the market economy [13]. Industrial enterprises also benefit from financial development: a more favorable financing environment increases the production scale, which results in the increased consumption of raw materials and energies such as coal, carbon, and crude oil [14,15]. Some hold the view that financial development is accompanied by increased energy consumption: the consumer credit market often sees an upsurge as consumer demand for cars and houses increases, which then leads to an increase in carbon emissions [16]. Some others maintain that financial development alleviates the financing challenges of industrial enterprises by providing a more favorable environment for loan lending and R&D investment [17]. Thus, it is feasible to increase the technical strength of industrial enterprises to improve their resource utilization rate and reduce their unit energy consumption, thus reaching the goal of reducing carbon emissions [18,19,20]. Xu and Song concluded through empirical research that the relationship between economic growth and carbon emissions is nonlinear, and there is a time inflection point in their correlation [21]. For some regions in China, the correlation between economic growth and carbon emissions complies with the inverted U-shaped EKC, but this is hardly the case in other regions. To explore whether the relationship between China’s financial development and carbon dioxide emissions follows the EKC, Hu et al. introduced the square term of the financial development proxy variable into the empirical model they designed and concluded that the relationship curve between financial development and carbon emissions first rose before leveling off, which is consistent with the EKC [22].

Developing a low-carbon economy and reducing carbon emissions relies on technological progress. Previous works interpreted the contribution of technological progress to reduce carbon emissions from different perspectives. Sun et al. (2010) defined technological progress as the major contributor to the improvement of total factor productivity under carbon intensity constraints [23]. Li et al. (2018) believe that improved carbon efficiency is mainly boosted by technological advances [24]. In a study by Zhao et al. (2020), technological progress is defined as the efficient use of energy to achieve carbon emission reduction [25]. Some technological changes may exacerbate pollution, while others aid environmental protection, so the impact of technological advances on carbon emissions is heterogeneous [26]. Technological progress is also divided into many types and cannot be represented by just a single indicator. After dividing technological progress into energy use technology, CO₂ emission technology, and broad technological innovation, it has been found that progress inCO₂ emission technology has the most significant inhibitory effect on carbon emission reduction [27,28,29].

The effects of technological progress on carbon emissions also differ between industries. Zhang et al. (2017) selected 37 industries as samples for analysis and found that technological progress plays a stronger role in reducing carbon emissions among industries with high energy efficiency [30]. Some researchers have employed mathematical models to analyze the relationship between technological progress and carbon emission reduction. For example, Zhu et al. (2010) applied the STIRPAT model to investigate the impacts of population, wealth, and technology on carbon emissions [31] and found, through empirical research, that the impact of technological progress on carbon emission was not significant among the samples within the selected period of time and concluded that the potential of technological progress on carbon emission reduction is unlimited in the future. Wei and Yang(2010), by applying endogenous growth theory to an empirical study of the impact of technological progress on carbon dioxide emissions, found that technological progress was conducive to carbon emission reduction, though with regional differences [32]. As it often takes time to observe the effect of technological progress, it is necessary to encourage technological innovation to assist in carbon emission reduction [33]. In their work that explored the relationship between environmental development and technological change, Jaffe et al. (2002) found that the relationship between technological progress and carbon emission reduction was complex, as technological progress may either increase or reduce carbon emissions [34]. Since the economic reform and opening-up, China has witnessed industrial restructuring and economic booms, each with contributing factors to carbon emissions that vary at different stages of economic growth. Lu et al. (2013) divided economic growth into five stages and investigated the contributing factors to carbon emissions over these stages, indirectly explaining the varied effect of technological progress on carbon emission reduction in different stages of economic development [35].

Recently, there has been a trend to incorporate financial development, technological progress, and carbon emissions into an integrated framework ofresearch. Research on the development of a regional low-carbon economy based on these three elements has attracted increasing attention across the globe [36,37,38]. Ma et al. (2018) [39] constructed a spatial panel data model, decomposed financial development into indicators of financial development scale and those of financial development structure, and selected multiple indicators as the representative variables of technological progress. They found that the financial development structure indicators and the number of patent grants which represent technological progress and technology spillover from foreign investors have a positive impact on low-carbon economic development. Chen and Hu(2020) believe that financial development or technological progress alone cannot effectively promote carbon emission reduction; rather, a synergy of the two is necessary to reach the goal of carbon emission reduction [40]. Some other scholars argue that the technological progress that contributes to carbon emission reduction is not universal, but biased, such as green technological progress and technological innovation [41,42,43,44,45]. Yan et al. (2016) built an endogenous growth model and found that financial development contributes to technological progress and innovation which then promotes the transformation into a low-carbon and energy-saving economy [46]. However, few studies have incorporated technological progress into the path of the financial development effect on carbon emissions and the impact of the heterogeneity of technological progress on carbon emission reduction was rarely explored.

The possible contributions of this study are as follows: (1) the financial development theory and the environmental Kuznets curve are combined to construct a mediating effect model that uses technological progress as a transmission path for the impact of financial development on carbon emissions; (2) indicators of technical progress are decomposed into different variables, including the energy utilization rate, capital utilization rate, and foreign investment conversion rate, to avoid biases caused by analysis based on single indicators; and (3) considering the regional differences across the vast territory of China, this study explores what types of technological progress can better promote carbon emission reduction through financial development in the eastern, central, and western regions in China, which facilitates the formulation of region-specific green financial policies.

3. Methodology and Data

3.1. Introduction to theIntermediary Variable Method

This study draws on the works by MacKinnon et al. (2002) and Wen et al. (2004) on the empirical test method of mediation effect, as shown in Figure 1 [47,48]. If an independent variable $X$ can affect a dependent variable $Y$, but in an indirect way by affecting the variable $M$ first, then $M$ is called an intermediary variable. According to the mediation effect theory, the method of sequentially testing the regression coefficients is as follows: if the coefficient $c$ in Equation (1) in Figure 1 is significant, $X$ is considered to have a significant influence on $Y$; if $a$ in Equation (2) is significant, $X$ significantly affects $M$; if the coefficient $c\text{'}$ in Equation (3) is completely insignificant and the coefficient $b$ is significant, the influence of $X$ on $Y$ is considered to be completely realized by $M$. In this case, the intermediary variable $M$ is called a complete intermediary variable, that is, a complete mediation effect exists. If both $c\text{'}$ and $b$ in Equation (3) are significant, it indicates that there is an incomplete mediating effect and $M$ is a partial mediating variable between $X$ and $Y$.

$$Y=\mathrm{c}X+{\epsilon }_1$$

(1)

$$M=aX+{\epsilon }_2$$

(2)

$$Y=\mathrm{c}\text{'}X+bM+{\epsilon }_3$$

(3)

3.2. Empirical Model

According to the test theory of intermediary variables, (Figure 1), first, an econometric model isconstructed to test the relationshipbetween financial development and carbon emissions.

$$C{I}_{it}={\alpha }_0+{\alpha }_{1}F{D}_{i,t-1}+{\alpha }_2{F{D}^2}_{i,t-1}+{\displaystyle \sum {\alpha }_j{Z}_{it}}+v_i+v_t+{\epsilon }_{it}$$

(4)

Then, the mediation effect model of financial development and technological progress, as well as financial development, technological progress, and carbon emissions, is constructed as follows:

$$T_{it}={\beta }_0+{\beta }_{1}F{D}_{i,t-1}+{\displaystyle \sum {\beta }_j{Z}_{it}}+v_i+v_t+{\epsilon }_{it}$$

(5)

$$C{I}_{it}={\gamma }_0+{\gamma }_{1}F{D}_{i,t-1}+{\gamma }_2{FD}_{i,t-1}^2+{\gamma }_3{T}_{i,t-1}+{\displaystyle \sum {\gamma }_j{Z}_{it}}+v_i+v_t+{\epsilon }_{it}$$

(6)

where $i$ represents the province, $t$ represents the year, and the dependent variable $CI$ represents the carbon emission intensity, which is the explained variable; represents the carbon emission intensity, which is the explained variable; $FD$ represents the independent variable financial development, which is the core explanatory variable; $T$ represents technological progress; $Z$ is the control variable, including regional openness ($Open$), human capital ($Hc$), infrastructure level ($Infr$), industrial structure ($Stru$), and urbanization level ($Urb$); $Vi$ represents the individual effect, $Vt$ represents the time effect,$\epsilon$ represents the random error term, and $a_0$ is the constant term of Model (3). The square term of financial development ($F{D}^2$) is added to the model to verify whether there is an inverted U-shaped relationship between financial development and carbon emissions. If the coefficient $a_2$ of the square term ($F{D}^2$) is negative, it is considered that there is an inverted U-shaped relationship between financial development and carbon emissions, which is consistent with the relationship between economic development and pollutant emissionsemphasized by the environmental Kuznets curve (EKC). To enhance the robustness of the model results and prevent reverse causality, the core explanatory variable, financial development ($FD$), is selected to lag for one period. ${\beta }_0$ and ${\gamma }_0$ are the constant terms of Models (5) and (6), respectively. If $a_1$, ${\beta }_1$, ${\gamma }_1$, and ${\gamma }_3$ are all significant in the regression results of Models (4), (5), and (6), it indicates that financial development indirectly affects carbon emissions by influencing technological progress. In thiscase, technological progress, as a mediator variable, has a partial mediation effect. If $a_1$, ${\beta }_1$, and ${\gamma }_3$ are significant but ${\gamma }_1$ is no longer significant, it means that the influencing path of financial development to carbon emissions is completely realized through technological progress, and technological progress plays a completely mediating role.

3.3. Data Sources

This study usedthe provincial panel data of 30 provinces and cities in China from 2009 to 2021. Owing tothe lack of statistical data on Hong Kong, Macao, and Taiwan, these regions were excluded from the sample. The various energy consumption data involved in the carbon emission calculation were all from the “China Energy Statistical Yearbook” and thestatistics released by Intergovernmental Panel on Climate Change (IPCC). The data onthedeposits and loans of financial institutions were obtained from the official website of the People’s Bank of China. The statistical data on domestic patent certificationcame from the “China Science and Technology Statistical Yearbook”.Data onthefixed asset investment came from the “Statistical Yearbook of China’s Fixed Asset Investment”. Data ontheforeign direct investment (FDI) were obtained from the Wind database and the “China Foreign Economic Statistics Yearbook”. The regional GDP figures and other data are from the “China Statistical Yearbook”.

3.4. Variable Selection

(1)

Explained variables

In this study, the ratio of total carbon emissions to GDP was defined as the index of carbon emission intensity, whichquantifies the degree of carbon emissions. Alarger carbon emission intensity indicatesmore CO₂ per unit output, whichindicatesastronger needto develop a low-carbon economy and take a route of intensive economic growth. The estimation of carbon emissions refers to the formulation method of IPCC, and the specific calculation method is as follows:

$$C{O}_2={\displaystyle \sum _{i=1}^{14}{E}_i}\times NC{V}_i\times CE{F}_i$$

(7)

$$CE{F}_i=C{C}_i\times CO{F}_i\times (44/12)$$

(8)

where $C{O}_2$ represents the amount of carbon dioxide emissions to be estimated, $i$ represents the type of energy; ${\displaystyle \sum _{i=1}^{14}{E}_i}$ represents the burning consumption of the i-th energy fuel; $NC{V}_i$ represents the average low calorific value of the i-th energy; $CE{F}_i$ represents the carbon dioxide emission factor of the i-th energy; $C{C}_i$ is the carbon content of the i-th energy; and $CO{F}_i$ represents the carbon oxidation factor of the i-th energy; finally, $(44/12)$ is the molecular weight ratio of carbon dioxide to carbon. The indicators involved in the carbon emission calculation method are listed in

Table 1. The average low calorific value and carbon dioxide emission factor of different energy sources.

Energy Name	Coal	Coke	Coke Oven Gas	Blast Furnace Gas	Converter Gas	Other Gas	Crude Oil
$NCV$(kj/kg)	20,908	28,435	17,981	3855	8585	18,273.6	41,816
$CEF$(Kg/TJ)	95,977	105,996	44,367	259,600	181,867	44,367	73,333
Energy name	Gasoline	Kerosene	Diesel oil	Fuel oil	Liquefied petroleum gas (LPG)	Natural gas	Liquefied natural gas
$NCV$(kj/kg)	43,070	43,070	42,652	41,816	50,179	38,931	44,200
$CEF$(Kg/TJ)	70,033	71,500	74,067	77,367	63,067	56,100	64,167

Source: China Energy Statistical Yearbook and IPCC.

.

(2)

Explanatory variables

The core explanatory variable of this studywas financial development. Financial development not only refers to changes in the financial transaction flow at successive stages but also includes relative changes in the financial structure across time. Many quantitative indicators have been developed to measure financial development. In this study, the financial correlation ratio (FIR) proposed by Chen et al. (2020) was used as theindicator of financial development [37]. The deposit and loan balance of financial institutions represents the total amount of financial transaction activities, andthe regionalGDPrepresents the total amount of economic activity in the region.Since the financial correlation ratio (FIR) is the ratio of the total amount of financial activities to the total amount of economic activities within a certain period, this study set the proxy variable of financial development as the ratio of the balance of deposits and loans of financial institutions to GDP, which is directly proportional to the development of monetary and financial markets.

(3)

Indicators of heterogeneous technological progress

To analyze the difference between different types of technological progress as transmission paths for the impact of financial development oncarbon emission transmission paths, this study selected five variables to represent technological progress ($T$) and constructed indicators for heterogeneous technological progress, including generalized technological progress ($Tech$), environmental technological progress ($ET$), energy technology progress ($IE$), capital embodied technology progress ($KE$), and FDI technology spillover ($TFDI$). To ensure the stability and convergence of the data, the logarithmic form of this variable was adopted. Among the indicators, $Tech$ is represented by the number of grantedpatents in China, and its value is proportional to the technological strength; that is, a larger value of $Tech$, indicates more technological innovation achievements, higher production efficiency, and more technical strength. $ET$ is expressed as the ratio of carbon emissions to energy consumption. A larger value of $ET$, indicates a higher carbon emission per unitofenergy consumption and lower technological strength. $IE$ is expressed in terms of energy consumption per unit of GDP, and a larger value of $IE$ indicates more energy consumed to create a unit of GDP, which means a lower level of technology. $KE$ is expressed as the ratio of GDP tothe physical capital stock. For the calculation of the physical capital stock, the perpetual inventory methodproposed byShan (2008) and Zhang (2015) was used in this paper [49,50]. A larger value of $KE$ means a higher GDP per unit of physical capital stock, a higher utilization rate of material capital, and a higher level of technological progress. $TFDI$ is expressed as the ratio of actual FDIof a province converted into RMB at the standard exchange rate to the GDP of the province in question. $TFDI$ reflects the development of low-carbon technology, while the introduction ofnew technologies from other countries promotes the updating of corporate management concepts. A higher value of $TFDI$ means a higher level of technological progress. Thus, the variable of $T$ is represented by the following formula:

$$T=T(Tech,ET,IE,KE,TFDI)$$

(9)

(4)

Control variables

To ensure the reliability and robustness ofthe research result, this study selects the degree of regional openness $\left(Open\right)$, human capital $\left(Hc\right)$, infrastructure level ($Infr$), industrial structure $\left(Stru\right)$, and urbanization level $\left(Urb\right)$ as control variables. The degree of regional openness $\left(Open\right)$ is expressed as the ratio of the total import and export volumes of each province to their respective GDP. Human capital $\left(Hc\right)$ is expressed as the average years of education of employees in each province.The infrastructure level $\left(Infr\right)$ is measured by the total mileage of roads and railways per square kilometer in each province. The industrial structure $\left(Stru\right)$ is expressed as the ratio of the added value of the secondary industry in each province to its annual GDP [22].The urbanization level $\left(Urb\right)$ is expressed as the ratio of the urban population of each province to its total population.

3.5. Descriptive Statistics and Correlation Analysis

Table 2. The mean value, standard deviation, and correlation coefficient matrix of each variable.

Variable	CI	FD	Tech	ET	IE	KE	TFDI
CI	1.000
FD	−0.273 ***	1.000
Tech	−0.655 ***	0.237 ***	1.000
ET	0.592 ***	−0.332 ***	−0.247 ***	1.000
IE	0.848 ***	−0.233 ***	−0.741 ***	0.200 ***	1.000
KE	−0.310 ***	−0.044	0.345 ***	−0.171 ***	−0.276 ***	1.000
TFDI	−0.304 ***	0.120 **	0.273 ***	0.018	−0.411 ***	0.324 ***	1.000
Open	−0.361 ***	0.520 ***	0.440 ***	−0.215 ***	−0.399 ***	0.441 ***	0.505 ***
Hc	−0.361 ***	0.620 ***	0.502 ***	−0.092 *	−0.528 ***	0.013	0.429 ***
Infr	−0.105 **	−0.450 ***	0.339 ***	−0.018	−0.134 ***	0.113 ***	−0.315 ***
Stru	0.273 ***	−0.670 ***	−0.065	0.206 ***	0.262 ***	0.045	−0.025
Urb	−0.414 ***	0.631 ***	0.545 ***	−0.171 ***	−0.524 ***	0.075	0.561 ***
MEAN	2.454	2.928	9.397	2.357	1.115	0.543	0.022
SD	1.668	1.092	1.590	0.580	0.633	0.118	0.017
Variable	Open	Hc	Infr	Stru	Urb
Open	1.000
Hc	0.564 ***	1.000
Infr	−0.382 ***	−0.325 ***	1.000
Stru	−0.252 ***	−0.415 ***	0.259 ***	1.000
Urb	0.737 ***	0.875 ***	−0.445 ***	−0.326 ***	1.000
MEAN	0.367	8.841	14.131	0.458	54.088
SD	0.311	0.986	7.494	0.083	13.596

The standard error is in parenthesis. *** p < 0.01, ** p < 0.05, * p < 0.1.

showsthe mean, standard deviation, and Pearson correlation analysis results of the main variables. In the correlation coefficient matrix of Table 2, the core explanatory variable $FD$ has significant and negative effects on $CI$, with a correlation coefficient at −0.273, which means that for every unit increase in the level of financial development, the carbon emission intensity will decrease by 0.273 units. Among heterogeneous technological progress indicators, $Tech$, $KE$, and $TFDI$ show a significantly negative correlation to $CI$, indicating that the improvement of the generalized technological progress, capital embodied technological progress, and the FDI technology spillover level will reduce the carbon emission intensity. Additionally, $ET$ and $IE$ show a significantly positive correlation to $CI$, with a correlation coefficient of 0.592 and 0.848, respectively, which meansthatwhen the value of the environmental technology progress variable and energy technology progress variable increases by one unit, the carbon emission intensity increases by 0.592 and 0.848 units, respectively. Since environmental technology progress and energy technology progress are reverse variables, progress in environmental technology and energy technology will lead to a decline in carbon emission intensity in the correlation analysis. Among the control variables, $Stru$ and $CI$ show a positive correlation, indicating that the increase in the ratio of the added value of the secondary industry to GDP will lead to an increase in carbon emission intensity; the other control variables are all negatively correlated to $CI$ in varying degrees.

China has a vast territory under its control, and it is necessary to explore the regional differences in the correlation between carbon emission intensity and heterogeneous technology levels across the nation. The sample data were divided into eastern, central, and western regions according to their geographical locations. Figure 2 shows the mean values of carbon emission intensity in these different regions and nationwide. As the figure shows, the western part of China marks the largest value of carbon emission intensity, followed by the central region and then the eastern region, that is, the carbon emissions required to create a unit of GDP are the highest in the western regionand lowest in the eastern provinces and cities of China. This can be partly attributed to the fact that eastern provinces and cities in China have witnessed a booming tertiary industry and a higher added value of hi-tech industries than their central and western counterparts, whereas the secondary industry accounts for the principal contributor to the total economic output value of the central and western regions. In particular, most areas in western China have adopted an extensive mode of economic growth and have high energy consumption. Figure 3 shows the mean values of heterogeneous technological progress variables for each region in China and nationwide. As the figure shows, the values of these variables differ across regions: the variable generalized technological progress $\left(Tech\right)$ marks the largest regional differences, with its value much higher in the eastern region than in the middle and western regions, indicating more granted patents in eastern provinces and cities than in their central and western counterparts. Part of the reason isthecontinuously improved infrastructure and economic development in eastern China in recent years, as well as the region’s increasing appeal to high-level talents and capacity forindependent innovation. Among the variables ofheterogeneous technological progress, the value of the proxy variable, energy technology progress $\left(IE\right)$, shows a trend of increasing from east to west, which means that the energy consumption per unit of GDP is the lowest in the east and highest in the west. This may be because of the high production efficiency in eastern Chinaand the region’s gradual shift from extensive production to intensive production. In the descriptive analysis of the environmental technology progress $\left(ET\right)$ across regions, it is found that this variable is higher in central China than in other regions. This means that the provinces and cities inthe central regionhavethe highest carbon emissions per unit of energy consumption across China, which can be attributed to the coal-based energy structure and the dependence on resource-intensive industries in the central region. All the other variables of technological progress are higher in the east and lower in the west, which is consistent with the differentiated levels of economic development across China—the eastern part is most developed, followed by the central regions and then the western provinces.

4. Empirical Results and Discussion

4.1. Principal Effect Test

Before regression analysis, the variance expansion factor $VIF=4.10$ (less than 10) of each variable was measured, indicating that multicollinearity between variables did not exist. Then, the result of the Hausman test significantly rejected the null hypothesis of random effects, so the fixed effects model was used in this study.

Table 3. The principal effect regression results from the impact of financial development on carbon emissions.

Variable	Nationwide (1)	Eastern Region (2)	Middle Region (3)	Western Region (4)
L.FD	0.321**	−0.849 **	2.156 **	0.217 *
	(2.53)	(−2.02)	(2.31)	(1.66)
L.FD2	−0.048 ***	−0.041 *	−0.450 **	0.005
	(−3.59)	(−1.81)	(−2.57)	(0.05)
Constant	6.405 **	14.73 ***	7.651 ***	12.186 ***
	(2.43)	(7.71)	(3.99)	(9.57)
Control variables	Yes	Yes	Yes	Yes
Individual fixed effect	Yes	Yes	Yes	Yes
Year fixed effect	Yes	Yes	Yes	Yes
R-squared	0.731	0.596	0.601	0.750
Number of obs	360	132	96	132

The standard error is in parenthesis. *** p < 0.01, ** p < 0.05, * p < 0.1.

displays the impact of financial development on carbon emissions and the regression results of the principal effect model (4). To further analyze the impact of financial development on carbon emissions among different regions, the sample datawere divided by regions. Columns (1), (2), (3), and (4) show the principal effect regression results of the impact of financial development on carbon emissions nationwide, in the east, the middle, and the west, respectively. Nationwide, the first-order coefficient ${\alpha }_1=0.321(p<0.05)$ of the lagging first-order core variable $L.FD$ is significantly positive, and the coefficient ${\alpha }_2=-0.048(p<0.01)$ of the quadratic term $L.F{D}^2$ is significantly negative. This conforms to the inverted U-shaped characteristic of the environmental Kuznets curve (EKC), indicating that financial development promotes carbon emissions in the short term but reduces carbon emissions in the long term. This is a result of the interplay between the expansion effect of financial development on carbon emissions and the impact oftechnological progress; across the nation and in the short run, the expansion effect of financial development on carbon emissions is greater than theimpact of technological progress, that is, the energy consumption increased by economic growth and increased production boosted by financial development is greater than carbon emissions per unit of GDP reducedbytheimproved financial development and technology progress. The regional regression results of the principal effect model were also obtained. In eastern China, the coefficient ${\alpha }_{2east}=-0.041(p<0.1)$ of the quadratic term $L.F{D}^2$ of the lagging first-order core variable is negative, indicating that the relationship between financial development and carbon emissions in eastern provinces also conforms to the inverted U-shapedEKC; the coefficient of the first term $L.FD$ is ${\alpha }_{1east}=-0.849(p<0.05)$, which indicates that the financial development in eastern provinces plays a role in reducing carbon emissions, and the impact of technology advances is greater than the expansion effect of financial development. For the central provinces, the coefficient of the quadratic term $L.F{D}^2$ of the lagging first-order core variable is ${\alpha }_{2middle}=-0.450(p<0.05)$ and the coefficient of the primary term $L.FD$ is ${\alpha }_{1middle}=2.156(p<0.05)$, which indicates that the relationship between financial development and carbon emissions in central China conforms to the inverted U-shaped EKC.However, at this stage (i.e., in the short term), financial development has not yet achieved the effect ofcarbon emission reduction. The current financial development has increased the carbon emission intensity but it will reduce the carbon emission intensity in the long run. For western provinces, the regression coefficient of the quadratic term $L.F{D}^2$ of the lagging first-order core variable is not significant, and the coefficient of the first order $L.FD$, which is ${\alpha }_{1west}=0.127(p<0.1)$, is significantly positive, indicating that financial development and carbon emissions are linearly positively correlated there, and hence, financial development leads to increased carbon emission intensity. In the principal effect, the regression results of Model (4) reveal thatfinancial development and carbon emissions show a significant correlation in all three regions and nationwide, indicating that financial development in each region significantly affects carbon emissions.

4.2. Mediating Effect Test

After the principal effect test on the relationship between financial development and carbon emissions, the mediation effect test path theory was employed to test the mediation effect of variables.

Table 4. The regression results of the mediating effect of heterogeneous technological progress at the national level.

Variable	Tech	CI1	ET	CI2	IE	CI3	KE	CI4	TFDI	CI5
	(5)	(6)	(7)	(8)	(9)	(10)	(11)	(12)	(13)	(14)
L.FD	0.258 ***	0.292 **	−0.320 ***	0.070	−0.107 ***	0.319 *	0.013 *	0.267 **	0.011 *	0.280 *
	(5.11)	(2.32)	(−4.34)	(0.21)	(−3.13)	(1.68)	(1.67)	(2.49)	(1.68)	(1.66)
L.FD2		−0.048 *		−0.001		−0.045		−0.042		−0.045
		(−1.67)		(−0.01)		(−1.06)		(−0.79)		(−1.03)
Tech		0.271 *
		(1.91)
ET				0.742 ***
				(6.96)
IE						0.812 **
						(2.29)
KE								−0.189 *
								(−1.65)
TFDI										−3.696 *
										(−1.66)
Constant	−0.325	6.136 **	3.950 ***	6.008 **	5.830 ***	4.226 *	0.636 ***	7.202 ***	0.025	7.123 ***
	(−0.73)	(2.57)	(6.04)	(2.57)	(19.13)	(1.66)	(7.67)	(2.87)	(1.64)	(2.89)
Control variables	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Individual fixed effect	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Year fixed effect	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
R-squared	0.897	0.733	0.134	0.774	0.733	0.756	0.555	0.728	0.117	0.731
Number of obs	360	360	360	360	360	360	360	360	360	360

The standard errors are in parenthesis. *** p < 0.01, ** p < 0.05, * p < 0.1.

shows the regression results of the mediation effect of Models (5) and (6) of heterogeneous technological progress at the national level. As shown in Column (5) in the table, the regression coefficient of $L.FD$ is ${\beta }_1=0.258$, and it is significant at the 1% level, indicating that financial development promotes generalizedtechnological progress.Then, the intermediary effect of generalized technological progressis seenin the correlation between financial development and carbon emission intensity. In Column (6), when the variable $Tech$ is controlled, the regression coefficient of $L.FD$ on $CI$ is significant at the 5% level, and ${\gamma }_1=0.292\prec 0.321$ is lower than that of $L.FD$ on $CI$ in the principal effect Model (4). This shows that there is an indirect transmission path where financial development affects carbon emissions through generalized technological progress. In this case, the intermediary effect coefficient is about 0.070 (0.271 ∗ 0.258) and the explanatory strength is approximately 21.78% (0.070/0.321), indicating that generalized technological progress plays a partial intermediaryrole in the impact of financial development on carbon emissions. This result is consistent with China’s national conditions, mainly because financial development has facilitated financing initiatives of enterprises, thus allowing enterprises to invest moreinscientific research and increase the number of granted patents. However, as the nation is still shifting from an extensive mode of economic development to an intensive mode, and the awareness of environmental protection is still weak among the masses, financial development does not support low-carbon economic development through the path of generalized technological progress. In Column (7), the proxy variable $L.FD$ of financial development has a significantly negative correlation to the environmental technology progress variable $ET$ at the 1% level.This is because environmental technology progress is an inverse variablerepresenting carbon emissions per unit of energy consumption, showing that financial development has led to improved environmental technology progressat the national level, which helps to reduce carbon emissions per unit of energy consumption. In Column (8), with the proxy variable of environmental technological progress ($ET$) controlled, the regression coefficient of the impact of financial development on carbon emissions is 0.070 and not significant, indicating that $ET$ is a complete intermediary variable and has a complete intermediary effect; that is, the impact of financial development on carbon emissions is completely realized through environmental technological progress. In Column (9), the regression coefficient of $L.FD$ to $IE$ is significantly negative, which indicates that financial development reduces energy consumption per unit of GDP. In Column (10), with the proxy variable $IE$ controlled, the regression coefficient of financial development on carbon emissions is 0.319, which is significant at the 10% level, indicating that energy technology progress is an intermediary variable for the impact of financial development on carbon emissions, and its mediating effect is about −0.087 (−0.107 ∗ 0.812).The sign is opposite to the principal effect regression coefficient (0.321) which means that the progress in energy technology plays a role in diluting the total effect of financial development on carbon emissions. Columns (11) and (13) are the regression coefficients of the $L.FD$ for $KE$ and $TFDI$, respectively. The values are 0.013 and 0.011 and both show 1% significance, indicating that capital-embodied technological progress and FDI technological spillover are mediating variables in the relationship between financial development and carbon emissions. The mediating effects are −0.002 (−0.189 ∗ 0.013) and −0.041 (−3.696 ∗ 0.011), respectively, which is opposite to the principal effect regression coefficient (0.321). This means that the increasedcapital utilization rate and adoption of advanced production technologies brought by foreign investment can alleviate the increased carbon emissions caused by financial development at the national level.

4.3. Subsample Mediation Effect Test

Given the differences in the environment, energy resources, climate, and economic development levels across regionsinChina, the regression results of the mediation effect of heterogeneous technological progress in different regions were also analyzed here.

Table 5. The regression results of the mediating effect of heterogeneous technological progress in the eastern region.

Variable	Tech	CI1	ET	CI2	IE	CI3	KE	CI4	TFDI	CI5
	(15)	(16)	(17)	(18)	(19)	(20)	(21)	(22)	(23)	(24)
L.FD	0.233 ***	−0.739 *	−0.642 ***	−0.252 **	−0.040 *	−0.797 **	0.006 *	−0.786 *	0.013 *	−0.802 **
	(3.48)	(−1.75)	(−4.86)	(−2.19)	(−1.68)	(−2.05)	(1.66)	(−1.81)	(1.71)	(−2.14)
L.FD2		0.037		−0.012		−0.045		0.033		0.049
		(0.88)		(−0.59)		(−1.15)		(0.74)		(1.14)
Tech		−0.434 **
		(−2.26)
ET				0.861 ***
				(18.68)
IE						2.112 ***
						(4.46)
KE								0.633 *
								(1.76)
TFDI										−4.509 *
										(−1.66)
Constant	−0.467	13.478 ***	10.043 ***	8.075 ***	3.666 ***	8.476 ***	0.688 ***	14.019 ***	0.062	15.104 ***
	(−0.49)	(7.19)	(5.30)	(4.37)	(10.74)	(3.65)	(4.19)	(6.76)	(1.38)	(8.16)
Control variables	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Individual fixed effect	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Year fixed effect	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
R-squared	0.891	0.609	0.334	0.707	0.836	0.640	0.575	0.593	0.368	0.731
Number of obs	132	132	132	132	132	132	132	132	132	132

The standard errors are in parenthesis. *** p < 0.01, ** p < 0.05, * p < 0.1.

shows the regression results for subsamplesin the eastern region. The regression results of financial development to heterogeneous technological progress are all significant, indicating that in eastern China, generalized technological progress, environmental technological progress, energy technological progress, capital-embodied technological progress, and FDI technology spillover all play a mediating role in the impact of financial development on carbon emissions. It is noteworthy that the mediating effect of ET is about −0.553 (−0.642 ∗ 0.861), and its explanatory power is approximately 65.11% (−0.553/−0.849). This shows that the path of technological progress through which financial development reduces carbon emissions in eastern China relies mainly ontheenvironmental technology progress level. The major reason is that the eastern part has always been the pilot zone of green economic development, which is basically consistent with what is specified in the “China Green Economic Development Report”.

Table 6. The regression results of the mediating effect of heterogeneous technological progress in central China.

Variable	Tech	CI1	ET	CI2	IE	CI3	KE	CI4	TFDI	CI5
	(25)	(26)	(27)	(28)	(29)	(30)	(31)	(32)	(33)	(34)
L.FD	0.415 *	0.317 ***	0.314 **	0.601	−0.118 *	1.403 **	0.024 **	2.066 ***	0.002 *	1.639 *
	(1.96)	(3.20)	(2.09)	(0.82)	(−1.69)	(2.25)	(2.13)	(2.72)	(1.71)	(1.70)
L.FD2		−0.615 ***		−0.237 *		−0.235 **		−0.497 ***		−0.330 *
		(−3.38)		(−1.76)		(−1.99)		(−3.04)		(−1.79)
Tech		−4.90 **
		(−2.50)
ET				1.619 ***
				(7.91)
IE						2.551 ***
						(10.24)
KE								3.239 ***
								(3.68)
TFDI										34.875 *
										(1.82)
Constant	−2.407 **	9.349 ***	0.424	9.493 ***	5.397 ***	−1.237	0.606 ***	6.175 ***	−0.010	10.354 ***
	(−2.51)	(5.01)	(0.62)	(7.21)	(11.04)	(−0.68)	(4.14)	(3.63)	(−1.01)	(6.19)
Control variables	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Individual fixed effect	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Year fixed effect	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
R-squared	0.893	0.575	0.223	0.744	0.799	0.774	0.749	0.669	0.444	0.586
Number of obs	96	96	96	96	96	96	96	96	96	96

The standard errors are in parenthesis. *** p < 0.01, ** p < 0.05, * p < 0.1.

shows the regression results for the mediating effect of heterogeneous technological progress in provinces in central China. In the middle region, the regression coefficient of the variable $L.FD$ to $ET$ is 0.314, and it is significant at the 5% level, which means that one unit increase in $L.FD$ corresponds to an increase in $ET$ by 0.314 units. However, with the variable $ET$ controlled, the correlation between financial development andcarbon emissionsis no longer significant, indicating that financial development in provinces in central China has led to an increase in carbon emissions per unit of energy consumption. The transmission path ofthe impact of financial development on carbon emissions is mainly through the advances in environmental technology, and, thence, the progressin environmental technology has a complete mediation effect. This is mainly because the central region is currently dominated by the secondary, especially heavy industry, which relies highly on fossil fuels andother high-carbon energy sources. The regression results of financial development on the other heterogeneous progress variables are all significant, indicating that all the selectedvariables of technological progress are the transmission paths for the impact of financial development on carbon emissions. Among them, the mediating effectsof the variables Tech and IE (−2.034 and −0.301) have opposite signs of the principal effect (2.156). This shows that financial development plays a dilutive role in the carbon transmission effect. Financial development can increase the number ofgranted patents and reduce energy consumption per unit of GDP in the central region. In other words, financial development can support the low-carbon economic development of the region by boosting generalized technological progress and energy technological progress.The mediating effect of capital-embodied technological progress and FDI technology spillover in central China is estimated to account for only 3.61% (0.024 ∗ 3.239/2.156) and 3.24% (0.002 ∗ 34.875/2.156) of the total effect, respectively, which is consistent with the fact that provinces in central China have a small physical capital stock, a lowutilization rateof capital, and insufficient foreign investment. This is partly due to the national development strategies released by the Chinese government in the 1990s that give priority to coastal areas in the east and provide assistance to westernprovinces. Despite the “Rise of Central China” plan that the central government proposed in these years, it will take time before the plan makes some actual achievements.

Table 7. The regression results of the mediating effect of heterogeneous technological progress in western China.

Variable	Tech	CI1	ET	CI2	IE	CI3	KE	CI4	TFDI	CI5
	(35)	(36)	(37)	(38)	(39)	(40)	(41)	(42)	(43)	(44)
L.FD	0.198 **	0.089 *	0.031 *	0.191 *	−0.118	−0.089	0.046 ***	0.047 *	0.001	0.123
	(2.58)	(1.73)	(1.71)	(1.65)	(−1.09)	(−0.14)	(2.98)	(1.89)	(0.05)	(0.18)
L.FD2		−0.005		0.089		−0.022		0.062		−0.016
		(−0.05)		(1.07)		(0.24)		(0.68)		(−0.16)
Tech		−0.211 *
		(−1.66)
ET				0.852 ***
				(6.52)
IE						0.525 ***
						(2.68)
KE								3.712 ***
								(3.96)
TFDI										10.264
										(0.99)
Constant	1.328 **	12.271 ***	2.214 ***	11.355 ***	5.397 ***	8.148 ***	0.803 ***	14.318 ***	−0.008	12.268 ***
	(2.41)	(10.35)	(3.06)	(9.32)	(11.04)	(4.92)	(7.24)	(11.46)	(−0.75)	(10.49)
Control variables	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Individual fixed effect	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Year fixed effect	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
R-squared	0.935	0.751	0.102	0.759	0.799	0.771	0.649	0.777	0.308	0.752
Number of obs	132	132	132	132	132	132	132	132	132	132

The standard errorsare in parenthesis. *** p < 0.01, ** p < 0.05, * p < 0.1.

shows the regression results for the mediating effect of heterogeneous technological progress in western China. Among the indicators ofheterogeneous technological progress, the regression results of $L.FD$ on variables $IE$ and $TFDI$ are not significant. This is partly because the financial market in the western regionof China is underdeveloped, and large-scale projects there are mainly fundedby the state; in addition, the region does notappeal to foreign investors due to its unfavorable geographical conditions. As shown in Table 7, the regression results of $L.FD$ on the variables $Tech$, $ET$ and $KE$ are significant. Among them, the mediating effect of $Tech$ is −0.042 (−0.211 ∗ 0.198), whichhas an opposite sign to the principal effect (0.217), indicating that generalized technological progress dilutes the impact of financial development on carbon emissions as a mediating variable. As the central government attempts to boost the development of its western region, projects with advantageous industries have been initiated, such as the West-East Gas Pipeline, Qinghai-Tibet Railway, construction of wind power plants, andthedevelopment of alternative new energies, which all rely on patented technologies. Financial investment institutions, guided by national policies, invest more in the development of characteristic industries there, thus contributing to the development of a low-carbon economy. In addition, environmental technological progress and capital-embodied technological progress are intermediary transmission paths for the impact of financial development on carbon emissions in the western region.

Figure 4 is a summary of the research results.

5. Conclusions and Recommendations

This paper fully considered the heterogeneity of technological progress, selected indicators of different technological progress into the mediating effect model, and usedthe data of 30 provinces from 2009 to 2021 to build a provincial panel data model for empirical analysis. The main research conclusions are as follows: (1) forprovinces in the eastern and central regions, the relationship between financial development and carbon emissions conforms to the inverted U-shape of the environmental Kuznets curve, and, ultimately, financial development will help reduce carbon emissions in the long run. For the western regions, financial development increases linearly with carbon emissions; (2) different types of technological progress have varied effects on the impact of financial development on carbon emissions. At the national level, generalizedtechnology progress, environmental technology progress, energy technology progress, capital-embodied technology progress, and FDI technology spillover are all the transmission paths for the impact of financial development on carbon emissionswhere environmental technology progress is the complete transmission path of financial development to carbon emissions; and (3) in the eastern regions, generalizedtechnological progress, environmental technological progress, energy technological progress, capital-embodied technological progress, and FDI technological spillover are all part of the transmission path of the financial development on carbon emissions, and environmental technological progress has the strongest explanatory power.In the centralregion, environmental technology progress is acomplete transmission path of the impact offinancial development on carbon emissions, and financial development can play a role in reducing thecarbon emission intensity through the transmission paths of generalized technological progress and energy technology progress. In the western region, the impact of financial development on carbon emissions mainly relies on the transmission paths of generalized technological progress, environmental technological progress, and capital-embodied technological progress, where generalized technological progress can reduce the carbon emission intensity caused by financial development.

Given the above conclusions, the following recommendations are provided to develop a low-carbon economy: (1) the local governments need to guide financial services to the real economy and strengthen the reform and innovation of green finance. For example, they can release more favorable policies for the development of green financial services, such as green funds, green trusts, and green leases; meanwhile, more financing channels can be developed to support low-carbon and green economic and technological initiatives to reduce the risk and cost of financing; (2) local administrations need to promote advances of technologies but keep in mindthe heterogeneity of technological progress to avoid “one-size-fits-all” policies. More attention should be paid to advancement in environmental technology as it was found through empirical analysis to have contributed considerably tothe development of a low-carbon economy. Policy-guided investment into sci-tech R&D was found to have reduced carbon emissions per unit of energy consumption at both the national and regional levels, where advances in environmental technology are an important transmission path for the impact of financial development on carbon emissions. In terms of the mediation effect of technological progress in the middle and western regions, the types of technological progress paths affected by financial development are less than those in the eastern region. Therefore, the middle and western regions should improve the utilization rate of resources and pay more attention to the use of clean energy. The eastern region should actively take advantage of its geographical location, appropriately introduce advanced management concepts and carbon emission technologies from FDI, increase the utilization rate of human and material capital, and extend the adoption of advanced technologies to the middle and western regions; and (3) initiatives for industrial restructuring, development of tertiary industries and low-energy-consuming industries, and market-based controlof greenhouse gas emissions should be put in place across the whole nation. For example, the central government and local administrations can establish monetary policy tools to support carbon emission reduction, encourage the online trading of carbon emissions with the power generation industry as the pilot sector, and extend the practice across different industries as well as channel more social funds to carbon emissions reductionendeavors.