Can green credit inhibit regional carbon emissions? evidence from China

Abstract

During the inhibition path of carbon emissions, the role of green finance is of increasing concerns, among which, green credit is regarded as crucial instrument. However, there exist insufficient empirical explorations on effects of green credit. Therefore, this study introduced intergovernmental panel on climate change method to calculate regional carbon emissions based on eight fossil fuels from 2008 to 2019 in China. Subsequently, spatial measurement, threshold regression, and intermediary model were further applied to examine how green credit may affect carbon emissions regarding the restraining effect, threshold effect, transmission mechanism, and spatial heterogeneity. The results show that: (1) green credit can effectively inhibited China's provincial carbon emissions although there existed regional inconsistency. (2) Appropriate levels of environmental regulation and marketization can enhance the repress effect of green credit. (3) Green credit can suppress provincial carbon emissions through optimizing the energy consumption structure, while there exists no intermediary effect of industrial structure upgrading during impact path. (4) Considering endogeneity, green credit can still significantly inhibit regional carbon emissions. These findings further enrich the current literature and provide references for policy design of constructing efficient carbon neutralization path.

Keywords

Green credit carbon emission spatial measurement intermediary effect threshold effect transmission mechanism

Introduction

Carbon emissions are negative externalities associated with economic activities. Long-term uncontrolled carbon emissions have caused extreme climate including greenhouse effect, melting glaciers, droughts, and floods, which have become the global challenge faced by mankind.^1,2 Strengthening the global climate governance cooperation and promoting low-carbon economy have become the primary strategies of many countries.^3,4 As the world's major carbon emitter, the carbon emission in China reached 10.15 billion tons in 2021, accounting for 30.2% of the global total emission.^5,6 Aiming at low-carbon economy, China has actively responded to global carbon emission reduction actions, and made solemn commitment in the Paris Agreement on Climate Change to achieve carbon peaking by 2030 and carbon neutrality by 2060 (namely “30·60” goal).^7,8 However, compared with developed countries, developing countries face severe challenges especially considering uncertainties resulted from post-epidemic era.^9,10 Therefore, it is importance for developing countries to formulate an efficient carbon neutralization policy system with reference to local conditions. In particular, coupling industry-finance-market-policy, China should develop scientific carbon control policy system and achieve the “30 · 60” goal.

Regarding the “30 · 60” goal, a series of researches have been conducted on path explorations, advocating the importance of economic structural adjustment, energy replacement, technological innovation, and green innovation on carbon emission reduction.^5,11–15 However, the scientific path plan urgently needs sufficient funds to be escorted. In order to balance the relationship between carbon emissions and finance, the academic community began to pay attention to the relationship between green finance and carbon emissions, since green finance is regarded as the important instrument of carbon control.^16,17 As the crucial instrument of green finance, green credit utilizes credit leverage to restrict the development of pollution- and energy-intensive industries, and thus reducing environmental risks.¹⁸ Compared with traditional credit policy, green credit integrates the strategies of environmental protection and sustainable development with credit service system, which stimulates social funds into green industries and reduces the cost of industrial energy transformation, thus minimizing the total cost of carbon neutralization society.^19–21 However, the existing literature regarding green credit and carbon emission reduction remain at the theoretical level, most of which, focuses on the relationship between traditional credit policy and environment governance, while the conclusions are inconsistent due to differences in research perspectives, research subjects, and green characteristics.²²

On the other hand, faced by “30 · 60” strategic goal, there exists various factors affecting carbon emission reduction, and thus it is necessary to further deepen the relationship between green credit and carbon emissions. In particular, spatial heterogeneity resulted from different regional element systems should be clarified; and potential transmission mechanism should be captured as well as considering both market and government impact on the inhibition path. The abovementioned research gaps can reduce the integrity of the research pedigree. Therefore, the main contributions of current research include: ① systematically portray how green credit reduces carbon emissions; ② clarify the threshold effect and intermediary transmission paths of repress impact; ③ reveal the associated spatial heterogeneity; ④ provide empirical evidence for countries with economies in transition. The rest of this research is organized as follows: theory and hypotheses are outlined in Section 2, while methods and data are elaborated in Section 3. Section 4 indicates the results and discussion. Section 5 draws the conclusions.

Theories and hypotheses

Green credit and carbon emission

Green credit presents distinct properties of top-level policy, which restricts the extension of pollution-intensive industries and prevents environmental risks through credit leverage. Gianfrate and Peri²³ argued that green credit is the crucial policy instrument to enhance fund mobility and economic transformation as well as carbon emission reduction. Glomsrød and Wei²⁴ claimed that green credit could achieve carbon emission reduction of 470 million tons by 2030 if it could be operated on a reasonable track. In addition, green credit restricts the provision of loan and implements punitive high interest rate for pollution- and energy-intensive industries as well as surplus industries, and thus increasing their financial costs,²⁵ which subsequently stimulate enterprises to promote green production.¹¹ In detail, through fund formation mechanism and fund oriented mechanism, green credit provides increasing funds to emerging industries of energy conservation and environmental protection, while restricting loans to energy- and pollution-intensive industries. In the short term, large-scale capital can be raised to promote green industries and thus accelerating the industrial transformation and upgrading as well as achieving carbon emission reduction ^18,^26,²⁷. Moreover, through credit catalytic mechanism, not only short-term profitable projects but also long-term sustainable projects can receive investment, generating catalytic effect on industrial transformation and upgrading, which subsequently reduce carbon emissions. Therefore, the first hypothesis can be proposed:

H1:

Green credit can significantly curb provincial carbon emission.

Threshold effects of degree of marketization and environmental regulation

There exist preconditions for the inhibition effect of green credit on regional carbon emissions. Regarding economic system in China, government regulations and marketization are important factors affecting the repress impact of green credit.^10,18 As a financial policy instrument, with the deepening of financial reform, the marketization characteristics of green credit have become increasingly prominent. In order to vigorously promote carbon emission reduction, the government should reduce the access threshold and administrative control to activate market flexibility through policy, investment promotion and other aspects.⁹^,26 Yi Gang, from the central bank of China, indicated that the role of the market should be fully extracted to improve the financial system and consolidate the impact path⁹ (Lyu et al., 2022). Ruiz et al.²⁸ claimed that in the immature stage of market-oriented development, repress impact of green credit depends on policy regulation as the remedy. Environmental regulation can be an important impact factor of green credit on carbon emissions since it can promote energy conservation, emission reduction, and low-carbon development of microeconomic entities, as well as increase the cost of pollution and institutional compliance. Romano et al.²⁹ further argued that active green environmental regulation policy could promote investment in green production and achieve the purpose of carbon emission reduction. In addition, Xia and Li³⁰ explored the micro level of enterprises and believed that in the immature stage of the market, environmental regulation is one of the effective means to optimize the efficiency of green credit allocation and reduce carbon emissions. Therefore, as an external constraint, environmental regulation can affect the costs and profits of economic activities, thus affecting the green credit decisions of microeconomic entities. However, different intensities of environmental regulation may result in asymmetrical costs and benefits of economic activities, which can dynamically affect the carbon reduction effect of green credit. Hence, environmental regulation (ER) and degree of marketization (DM) may directly affect the effectiveness of green credit in the path of carbon emission reduction and the following two hypotheses are proposed:

H2:

DM generates threshold effect during the inhibition process of green credit on provincial carbon emission.

H3:

ER has threshold effect during inhibition process of green credit on provincial carbon emission.

Transmission mechanism of green credit

Green credit can significantly promote the optimization of regional energy consumption structure (ECS), which is the key factor of carbon emission reduction.¹⁹ ECS is closely related to the quality of the environment, since large proportion of coal, oil, and other fossil fuels can cause excessive carbon emissions. Therefore, ECS optimization can effectively promote carbon emission reduction.^25,31 Green credit can generate “crowding-out effect” on energy- and pollution-intensive industries. In detail, environmental pollution can be internalized as the financing cost of polluting enterprises, thus accelerating the upgrading of carbon emission equipment and technology, enhancing the substitution of clean energy for fossil energy consumption, promoting ECS, and subsequently achieving carbon emission reduction. In addition, green credit plays capital-oriented role by providing low-cost credit through structural monetary policy, diverting capital flow into green industries, and promoting green industry expansion. Subsequently, the production and consumption of clean energy can be enhanced and ECS can be optimized, thus reducing carbon emissions.²⁶ Hence, H4 is proposed as follows:

H4:

ECS plays an intermediary role during the repress impact of green credit.

Industrial structure upgrading (ISU) remains one of the important paths to reduce energy consumption as well as carbon emissions. Green credit can divert capital to different enterprises through differential interest rates, which is essentially the capital flow process among various industries, thus resulting in differences in the development of industrial sectors and industrial structures.³² Particularly, green credit can promote both replacement of traditional energy with clean energy and environmental pollution projects by environmental protection projects.³³ From macroeconomic perspective, green credit can be conducive to the elimination of energy- and pollution- intensive industries, and promote the capital transfer from polluting industries to low-energy and clean industries, thus contributing to regional energy conservation and carbon emission reduction.^{2, 34} Furthermore, green credit contributes to forcing pollution-intensive enterprises to carry out technological innovation and product/service upgrading, forming green incentive effect on the industry. Capital flow from credit institutions to green enterprises generates a signal to the society to develop green economy. As a policy instrument, green credit creates strong guiding effect on industrial investment, diverting capital flow to energy-saving and environment-friendly industries, such as high-tech industry, knowledge- or service-based economy, and so on, which facilitates ISU as well as carbon emission reduction.²¹ Therefore, the following hypothesis is proposed:

H5:

ISU plays an intermediary role during the repress impact of green credit.

Theoretical mechanism model

According to abovementioned analysis, a theoretical mechanism model is presented in Figure 1 to portray how green credit repress carbon emissions. In detail, the inhibition path, transmission path, and possible threshold effect of green credit on carbon emissions are integrated to visually present the research context. The logical inhibition path of green credit on carbon emissions is stereoscopically displayed, which echoes the research hypothesis.

Figure 1.

Impact mechanism of green credit on carbon emissions.

Methods and data

Model construction

(1) Spatial metrology model setting:

SDM model is prior to SAR and SEM model since it integrates the advantages of both models. Therefore, SDM was introduced to explore the transmission mechanism and spatial correlation between green credit and provincial carbon emissions in China. The model construction is shown as follows:

C O_{2 i t} = β_{0} + ρ W_{i t} C O_{2 i t} + β_{1} G C_{i t} + β_{2} X_{i t} + θ_{1} W_{i t} G C_{i t} + θ_{2} W_{i t} X_{i t} + μ_{i} + ν_{i} + ε_{i t}

(1)

In detail,

C O_{2 i t}, G C_{i t}

refer to the carbon emission and green credit of region i in year t, respectively.

X_{i t}

is the set of control variables;

W_{i j}

refers to the spatial weight matrix;

μ_{i}

、

ν_{i}

refer to the fixed effect of provinces and time, respectively; and

ε_{it}

represents the random disturbance term. Regarding formula 1, when

θ_{i}

equals to 0, SDM model will retrograde to SAR model; when

θ_{i} = - β_{i}

, SDM model will retrograde to SEM model.

(2) Exploratory spatial analysis

Examining spatial correlation of economic variables remains the key step in establishing spatial econometric models. Moran I index can reflect the similarity degree of spatial adjacency or spatial adjacency regional unit attribute values, which is usually applied to test the spatial correlation of variables. In detail, the Global Moran's I ∈[1-1], I ∈ [-1-0) represents negative relationship while I ∈ (0-1] indicated positive relationship.

I = \frac{n \sum_{i = 1}^{n} \sum_{j = 1}^{n} w_{i j} (x_{i} - \bar{x}) (x_{j} - \bar{x})}{\sum_{i = 1}^{n} \sum_{j = 1}^{n} w_{i j} \sum_{i = 1}^{n} {(x_{i} - \bar{x})}^{2}} \frac{n \sum_{i = 1}^{n} \sum_{j = 1}^{n} w_{i j} (x_{i} - \bar{x}) (x_{j} - \bar{x})}{s^{2} \sum_{i = 1}^{n} \sum_{j = 1}^{n} w_{i j}}

(2)

In detail, n refers to the number of provinces;

w_{i j}

is the weight matrix;

x_{i}, x_{j}

represent the attributes of provinces;

\bar{x} = \frac{1}{n} \sum_{i = 1}^{n} x_{i}

refers to the average value of attributes; and

S^{2} = \frac{1}{n} \sum_{i = 1}^{n} {(x_{i} - \bar{x})}^{2}

is the variance of attributes.

(3) Spatial matrix

The establishment of weight matrix is the premise of introducing spatial econometric model for empirical analysis since reflection of location relevance stays as the core of the empirical analysis of spatial econometric model. Common spatial weight matrices include proximity, geographical, and economic weight matrices. Since carbon emissions are not only generated from local but also from adjacent regions, this study explored the relationships among variables taking the geographical weight distance into account. Hence, the geospatial matrix is constructed as follows. In detail, $d_{i j}$ refers to the spatial distance between province i and province j, indicating that the spatial dependence between provinces decreases gradually with the increase of distance.

Geographical distance : W_{i j} = 1 / d_{i j}^{2} \cdot i \neq j

(3)

(4) Threshold model

Impact of green credit may be affected by external factors. Therefore, ER and DM were treated as threshold variables to conduct the threshold model. Subsequently, possible threshold effects under the situations with different threshold variables can be portrayed.

C O_{2} = a_{0} + a_{1} G C {E R \leq ϕ} + a_{2} G C {E R \geq ϕ} + τ X_{it} + ε_{it}

(4)

C O_{2} = a_{0} + a_{1} G C {M D \leq ϕ} + a_{2} G C {M D \geq ϕ} + τ X_{it} + ε_{it}

(5)

In detail,

C O_{2}, G C

refer to the carbon emissions and green credit of region i in year t, respectively; ER and MD represent their threshold values, respectively;

X_{i t}

refers to control variable set and

ε_{i t}

represents random disturbance term.

Data and index selection

Based on provincial panel data, this research examined the relationship between green credit and China's provincial carbon emissions. In order to avoid large differences in regression coefficients, some variables were logarithmically processed to eliminate the impact of heteroscedasticy, and interpolation processing was carried out for individual missing data. Relevant variable data were collected from China Statistical Yearbooks. The statistical description of variables is shown in Table 1. In particular, the deviation of each variable (except for the natural logarithmic variable) is relatively small, which implies that the fluctuations of variables such as carbon emissions, green credit, ISU, Foreign direct investment (FDI), environmental regulation, and marketization degree of China's province remains relatively stable; while the maximum value, minimum value, mean value and deviation value of each variable are significantly different, implying the significant differences among carbon emissions, green credit and related variables, despite overall growth trend remains significant.

Table 1.

Descriptive statistics.

Variable	Obs	Mean	Std. Dev.	Min	Max
CO2	360	9.086	0.744	7.036	10.655
GC	360	0.469	0.149	0.094	0.808
FDI	360	12.719	1.649	7.31	15.09
ISU	360	2.348	0.131	2.102	2.832
ECS	360	0.414	0.152	0.012	0.724
ER	360	1.451	0.704	0.373	4.844
DM	360	6.445	1.927	2.330	11.400
URB	360	55.521	13.186	29.110	89.600
ID	360	45.232	8.681	16.200	61.500

(1) Explained variables

Total carbon emission (CO₂): Aiming at scientifically characterize provincial carbon emission, the consumptions of eight selected fossil fuels^a were converted into 10000 tons of standard coal. Subsequently, the intergovernmental panel on climate change method is introduced for calculating and the specific formula is shown in Formula 6, where $i, j, t$ refer to province, energy category and year respectively; $δ$ represents the coefficient of the corresponding energy converted into 10000 tons of standard coal; $Q_{i j t}$ indicates the energy consumption by region; $C O_{2}$ refers to total carbon emission (unit of 10000 tons); $β_{j}$ refers to carbon emission coefficient corresponding to eight energy sources.

C O_{2 i t} = \sum_{i = 1}^{8} δ_{j} * Q_{i j t} * β_{j}

(6)

(2) Core explanatory variable

Green credit (GC): Currently, indirect measurement such as ratio of the total interest expenditure of credit obtained by all heavily polluted industries in the region to the interest of industrial sectors in the region is usually applied to reflect GC. Since GC may cause less credit resources flowing into the heavily polluting industries, this research adapted Xie and Liu's work²¹ and selected the proportion of interest expenditure of non six energy-intensive industries in the total industrial interest expenditure of each province to characterize the level of green credit in China's provinces.

(3) Threshold variables

Environmental regulation (ER): Appropriate ER is not only conducive to protecting the environment, but also stimulating technological innovation, and thus reducing carbon emissions. This research adopted the relationship between provincial GDP and provincial total energy consumption to measure ER.³⁵

Degree of marketization (DM) Higher level of marketization can enhance green development and curb regional carbon emissions. With reference to Fan et al.'s³⁶ work, the marketization index was adopted.

(4) Control variables

Foreign direct investment (FDI): FDI is an important factor affecting carbon emissions, although its impact mechanism remains complex and uncertain. This research adopted the ratio of actual regional foreign capital to regional GDP to measure FDI.³⁷

Industrial level (ID): Industrialization is considered as main source of regional carbon emissions.

The research introduced the ratio of secondary industry in regional GDP to measure ID.³⁸

Industrial structure upgrading (ISU): ISU remains the important factor affecting industrial carbon emissions. Since there exists no consistent measurement of ISU, this study adapted Wang et al.'s³⁹ measurement and the specific method is shown as follows. In detail, $W_{1}, W_{2}, W_{3}$ represent the proportions of the primary sector, secondary sector, and tertiary sector, respectively, with index weight of 1,2,3. Greater ISU index reflects higher upgrading level.

I S U = W_{1} + 2 W_{2} + 3 W_{3}

(7)

Urbanization level (URB): Urbanization may consume large amount of steel and cement, which is the main carbon emitter. This research referred to Liu and Xie's³⁸ work and adopted the ratio of urban population to total population at the end of the year to reflect the URB.

Energy consumption structure (ECS): Energy consumption remains the main source of carbon emissions. The ECS optimization can effectively reduce emissions. Hence, this research applied the ratio of coal consumption to total energy consumption to characterize ECS.

Results and discussion

Statistical analysis

(1) Unit root test

To ensure scientific research, LLC and Fisher unit root tests were applied before introducing the spatial econometric model to verify the stability of variables. Table 2 presents the results. In particular, according to the results of LLC and Fisher tests, all variables except ER pass the significance test. Therefore, there exists no unit root for the selected variables except for ER. Thus, selected data can be concluded to be relatively stable and the regression results are overall reliable. However, the existence of unit root of ER may not reject the existence of cointegration, and further cointegration testing is required.

Table 2.

Unit root test.

Method	LLC		Fisher		Method	LLC		Fisher
Variable	Level	First order	Level	First order	Variable	Level	First order	Level	First order
CO2	−6.728 ***	−8.957 ***	81.967 **	93.512 ***	ISU	−9.358 ***	−8.198 ***	88.665 ***	97.086 ***
GC	−6.683 ***	−8.511 ***	76.730 **	80.307 ***	URB	−11.452 ***	−10.032 ***	392.600 ***	412.267***
ID	−9.221 ***	−11.192 ***	58.403 *	96.496 ***	FDI	−5.787 ***	−6.124 ***	73.611 *	96.496 ***
Threshold variable					Intermediary variable
DM	−13.761 ***	−13.725 ***	199.770 ***	174.210 ***	ECS	−12.042 ***	−11.709 ***	190.481 ***	168.500 ***
ER	−3.170 ***	−3.674 ***	31.674	73.409 **	ISU	−9.358 ***	−8.198 ***	88.665 ***	97.086 ***

***p < 0.01, **p < 0.05, *p < 0.1.

(2) Cointegration test

Targeting at long-term stability among variables, Pedroni test and Westerlund test were introduced to conduct homogeneous and heterogeneous panel tests respectively, of which, the null hypotheses refer to “there exists no cointegration relationship.” The test results presented in Table 3 indicate that the P-values of Pedroni test and Westerlund test are less than 0.01, rejecting null hypotheses. Therefore, there exists long-term cointegration relationship among variables selected and the econometric model can be applied for regressive analysis.

Table 3.

Cointegration test.

Pedroni test	T	Westerlund test	T	P
Modified Phillips–Perron t	8.296	Variance ratio	7.049	0.000
Phillips–Perron t	−5.112
Augmented Dickey–Fuller t	−3.967

(3) Spatial correlation test

The carbon emissions of neighboring regions can indirectly affect the carbon concentration of neighboring provinces due to the impact of airflow, water flow, and industrial transfer. Therefore, with application of geographic weight matrix, Moran's I of green credit and carbon emissions were calculated. As can be seen from Table 4, the Moran’ I indexes of green credit and carbon emissions from 2008 to 2019 pass the significance test and fall between [0.406−0.701], [0.287−0.329], respectively, indicating the existence of positive spatial autocorrelation. Therefore, it is rational to apply spatial econometric model.

Table 4.

Moran's I index of GC and CO₂ emission from 2008 to 2019.

Year	GC	$C O_{2}$	Year	GC	$C O_{2}$	Year	GC	$C O_{2}$
2008	0.406*** （2.716）	0.329*** （2.251）	2012	0.679*** （4.415）	0.292** （2.026）	2016	0.631*** （4.089）	0.304** （2.081）
2009	0.522*** （3.423）	0.326*** （2.236）	2013	0.655*** （4.279）	0.284** （1.973）	2017	0.506*** （3.295）	0.299** （2.053）
2010	0.556*** （3.684）	0.313*** （2.169）	2014	0.604*** （3.939）	0.287** （1.988）	2018	0.516*** （3.376）	0.308** （2.114）
2011	0.701*** （4.569）	0.306** （2.121）	2015	0.574*** （3.765）	0.309** （2.119）	2019	0.439*** （2.898）	0.298 ** （2.054）

***p < 0.01, **p < 0.05, *p < 0.1.

In order to further analyze the spatial heterogeneity of GC and CO₂ emissions, the Moran's I scatter diagrams of green credit and carbon emissions were portrayed through STATA. As shown in Figures 2 and 3, the positive spatial autocorrelation of GC and CO₂ emissions have been further verified since the selected variables are clustered in the first and third quadrants.

Figure 2.

Moran's I scatter of carbon emission.

Figure 3.

Moran's I scatter of green credit.

Analysis of direct effects

(1) Selection of spatial model

During the process of exploring the relationship between GC and CO₂ emissions, best-fitted spatial model should be determined at first. With reference to Anselin's⁴⁰ research, Hausman test result shows that the selected model passes the 1% significance test, indicating the fixed effect model to be more suitable. In addition, as shown in Table 5, the statistics of LM and R-LM tests pass the significance test, where the statistics of SAR model is greater than that of SEM model, indicating SAR to be more appropriate. Moreover, the statistics of Wald and LR tests pass the significance tests, showing that SDM model cannot degenerate into SAR or SEM models. Therefore, the subsequent analysis mainly refers to SDM model.

Table 5.

Model testing.

Method	Statistics	Probability	Method	Statistics
LM-Spatial_Lag	35.5402	0.000	Wald-Spatial_Lag	15.49
Robust-LM-Spatial_Lag	26.1649	0.000	LR-Spatial_Lag	23.66
LM-Spatial_Erro	20.8632	0.000	Wald-Spatial_Erro	29.03
Robust-LM-Spatial_Erro	11.4879	0.001	LR-Spatial_Erro	40.47

(2) Relationship between green credit and carbon emission

The regression results are presented in Table 6, the impact path of green credit on carbon emission refers to −0.302***, passing the 1% significance test. The possible reasons include that green credit can reduce the proportion of non-green industry credit through capital flow selection; promote ISU; and eventually achieve industrial carbon emissions reduction. Therefore, H1 is accepted. Regarding control variable, FDI significantly suppresses regional carbon emissions (−0.029**), which verifies the pollution halo theory and denies the pollution paradise theory. On the other hand, industrial structure upgrading has restrained regional carbon emissions (−0.218) to a certain extent while the impact is not significant, indicating that regional industrial structure upgrading is insufficient. In addition, the level of urbanization and industrialization remain the major carbon sources, which is also the root of government commitment to the construction of new-type urbanization and industrial transformation.

Table 6.

Green credit and carbon emission estimation results.

$C O_{2}$	$β$	Z	$β$	Z	$β$	Z
$C O_{2}$	SDM		SAR		SEM
GC	−0.302***	−3.110	−0.241***	−2.510	−0.238**	−2.440
ISU	−0.218	−0.870	−0.246	−0.990	−0.296	−1.180
FDI	−0.029**	−2.220	−0.033***	−2.470	−0.031**	−2.290
URB	0.008**	2.200	0.012***	3.820	0.012***	3.790
ID	0.001	0.340	0.001	0.580	0.001	0.450
W*GC	0.088	0.640
W*ISU	1.350***	3.510
W*FDI	−0.049***	−2.450
W*URB	0.004	0.730
W*ID	0.012***	3.270
$ρ$	0.146* *	2.260	0.192***	3.130
Fixed effect	Time fixed control	Space fixed control	Time fixed control	Space fixed control	Time fixed control	Space fixed control

***p < 0.01, **p < 0.05, *p < 0.1.

Endogenous test

Considering endogenous problems such as two-way causality or omitted variables, which may hinder the reliability of research conclusions, SYS-GMM model was applied. In particular, lag of one period of CO₂ emissions (L.CO₂) was considered as instrument variable (IV), while lag of one period of green credit was considered as explanatory variable to minimize the research deviation caused by the existence of endogenous problems. In detail, as can be seen from Table 7, the coefficient of L.CO₂ remains positive, verifying the robustness of current research and reflecting the accumulation of CO₂ emissions in regions. Hence, application of SYS-GMM model is rational. On the other hand, the result of AR (2) suggests that the random error term does not have second-order sequence correlation, while Sargan test suggests the validity of IV. Furthermore, coefficient of GC is −0.369, passing 1% significance test. Thus, GC can still effectively restrain CO₂ emissions, while its impact path has been enhanced after concerning the relevant endogenous problems, which further verifies the robustness of current research.

Table 7.

Endogenous tests.

$C O_{2}$	$β$	StErr.	t-value	p-value	Confidence Interval		Sig
L.CO₂	0.901	0.028	32.220	0.000	0.847	0.956	***
GC	−0.369	0.051	7.264	0.000	0.269	0.468	***
L.GC	−0.107	0.037	−2.901	0.004	−0.179	−0.035	***
ISU	0.074	0.162	0.457	0.649	−0.244	0.392
FDI	−0.031	0.006	−5.673	0.000	−0.042	−0.021	***
URB	−0.001	0.000	−0.869	0.385	−0.004	0.001
ID	0.004	0.002	2.422	0.016	0.001	0.007	**
Constant	0.935	0.393	2.380	0.017	0.165	1.705	**
Sargan	chi2	P > chi2	AR(1)	Z	P > chi2	AR(2)	Z	P > chi2
Sargan	26.083	0.9998	AR(1)	−3.791	0.000	AR(2)	00.775	0.4381

***p < 0.01, **p < 0.05, *p < 0.1.

Threshold effect analysis

There exist external factors affecting the repress impact of GC, among which government regulation and DM are considered as core factors.¹⁰ Therefore, possible threshold effects are explored considering environmental regulation and marketization degree as threshold variables in order to further clarify the internal mechanism of inhibiting process.

(1) Degree of marketization

Government cannot solely ensure the healthy development of green credit system while the market also plays important role. The DM is a necessary prerequisite, which can facilitate financial system innovation; innovate financial derivatives; optimize the allocation of credit resources; promote green industry and low-carbon development. However, it is still doubtful to what extent marketization will affect impact path of GC on CO₂ emissions. Therefore, this study explored the threshold effect of marketization degree with reference to Hansen's work,⁴¹ of which, the results are presented in Table 8.

Table 8.

Threshold effect test.

Threshold value: Degree of Marketization
Threshold	Threshold value	Confidence interval	Fstat	Prob	Crit10	Crit5	Crit1
Single	4.310	4.150–4.330	41.430	0.036	32.224	38.421	56.715
Double	6.570	6.410–6.590	24.410	0.118	25.856	34.706	45.871
Threshold value: Environmental regulation
Threshold	Threshold value	Confidence interval	Fstat	Prob	Crit10	Crit5	Crit1
Single	0.472	0.447–0.513	38.800	0.099	37.878	48.150	66.027
Double	0.696	0.657–0.994	22.090	0.258	36.880	53.889	76.581

In detail, single threshold effect of MD pass the 5% significance test with the corresponding P-value to be 0.036; while the double threshold effect is insignificant with corresponding P-value to be 0.018. Hence, the single threshold effect of MD exists with its value being 4.310 that falls into 95% confidence interval. Regarding Figure 4, the accuracy of threshold value has been further verified. Thus, it can be concluded that the threshold effect estimation of marketization degree is rational.

Figure 4.

Relationship between threshold value and LR.

On the other hand, Table 9 M1 indicates the regression results of the impact of GC on CO₂ emissions by taking DM as threshold variable. When DM < 4.310, low level of marketization restricts the repress impact of green credit; when DM > 4.310, for every 1% increase in green credit, carbon emission reduction will be suppressed by 0.280%. Therefore, H2 is accepted.

Table 9.

Threshold effect regression results.

M1:Threshold vale of DM					M2:Threshold vale of ER
$C O_{2}$		$β$	Z	P > t	$C O_{2}$		$β$	SE	P > t
ISU		0.547	2.080	0.038	ISU		0.234	0.920	0.358
FDI		0.002	0.110	0.914	FDI		−0.006	−0.420	0.674
URB		0.023	9.490	0.000	URB		0.021	8.240	0.000
ID		0.007	3.020	0.003	ID		0.007	2.750	0.006
GC	DM < 4.310	0.191	1.970	0.050	GC	ER < 0.472	−0.312	−2.330	0.020
GC	DM > 4.310	−0.280	−1.910	0.057	GC	ER > 0.472	0.085	0.891	0.372

***p < 0.01, **p < 0.05, *p < 0.1.

(2) Environmental regulation.

Green credit can grow healthily under the sufficient marketization, while the environmental regulation is necessary before market is fully mature. The government promotes green credit system and enforces environmental liability insurance through guarantee discount and other policy systems to provide driving force for green industry and restrict non-green industry. However, due to differences in regional economic structures, ER remains inconsistent, which may restrict its implementation effect. Thus, ER was considered as threshold variable to examine the possible threshold effect.

As shown in Table 8, the single threshold of ER is significant at the 10% level, with P-value being 0.099; while double threshold effect is insignificant with corresponding P-value being 0.258. Therefore, there exists single threshold of ER, of which, the threshold value is 0.472, falling within the 95% confidence interval. Thus, threshold value division is considered reasonable. Furthermore, referring to Figure 4, the rationality of threshold value has been verified.

In addition, Table 9 M1 presents the regression results of the impact of GC on CO₂ emissions taking ER as threshold variable. In detail, when ER > 0.472, the impact of green credit on carbon emission changes from inhibition to stimulation. Therefore, ER has a double-sided sword effect, the scale of which, should raised attention by local governments. Appropriate ER scale can promote green credit and cover the shortages brought by insufficient marketization, thus achieving the purpose of restraining regional carbon emissions. However, environmental regulation that exceeds the affordability of the market may result in resistance to the development of GC and CO₂ reduction. Therefore, H3 is accepted.

Spatial heterogeneity analysis

China has a vast territory, which exists significant differences in regional factor resources as well as green credit systems. The repressing effect of GC in different regions may be inconsistent, which increases the difficulty of carbon emission control. Aiming at clarifying the differences in repress impact of GC, this research divided regions into coastal and inland economic zones to verify spatial heterogeneity, which can facilitate the formulation of national policies and achieve synchronization of regional carbon emission control.

As shown in Table 10, the impact of GC on carbon emission in coastal areas (−0.304***) is more significant than that in inland areas (−0.022). Possible reasons include that the country has carried out industrial transformation and structural upgrading in coastal areas to develop into pacesetters. Compared with inland area, coastal area enjoys more complete green credit system, which can alleviate financing constraints and curb regional carbon emissions by implementing policy dividends to green industries. Regarding inland area, which is the carrier of pollution- and energy-intensive industries, the transformation of economic structure is restricted as well as the development of green credit. Therefore, carbon emission reduction in inland area remains severe.

Table 10.

Spatial heterogeneity analyses.

$C O_{2}$	Coastal area			$C O_{2}$	Inland area
$C O_{2}$	$β$	SE.	Z	$C O_{2}$	$β$	SE	Z
GC	−0.304***	0.075	−4.070	GC	−0.022	0.148	−0.150
ISU	1.059***	0.332	3.190	ISU	−0.515*	0.309	−1.670
FDI	−0.042***	0.012	−3.470	FDI	0.015	0.020	0.740
URB	0.013***	0.002	6.020	URB	−0.027***	0.009	−3.030
ID	0.016***	0.003	5.310	ID	0.002	0.004	0.430
W*GC	−0.190**	0.082	−2.320	W*GC	−0.208	0.275	−0.760
W*ISU	1.478***	0.392	3.770	W*ISU	1.592***	0.406	3.920
W*FDI	−0.023	0.025	−0.940	W*FDI	−0.096***	0.028	−3.450
W*URB	0.004	0.005	0.920	W*URB	0.030***	0.011	2.660
W*ID	0.027***	0.004	7.170	W*ID	0.005	0.004	1.160
$ρ$	−0.337***	0.075	−4.520	$ρ$	0.221**	0.079	2.820
Fixed effect	Time fixed control	Space fixed control		Fixed effect	Time fixed control		Space fixed control

***p < 0.01, **p < 0.05, *p < 0.1.

Transmission mechanism analysis

The realization of carbon emission reduction stimulated by green credit requires the participation of intermediary variables. To scientifically verify the inhibition path of GC, this study regarded ECS and ISU as intermediary variables to conduct further regression analysis. The results are presented in Table 11.

Table 11.

Analysis on the transmission mechanism of green credit on carbon emission.

Intermediary transmission effect of ECS				Intermediary transmission effect of ISU
Variable	M1: $C O_{2}$	M2: ECS	M3 $C O_{2}$	Variable	M4: $C O_{2}$	M5:ISU	M6: $C O_{2}$
GC	−0.302*** (−3.110)	−0.109*** (−3.090)	−0.158** (−1.730)	GC	−0.259*** (−2.630)	−0.015 (−0.730)	−0.302*** (−3.110)
ECS			1.022*** (7.570)	ISU			−0.218 (−0.870)
ISU	−0.218 (−0.870)	−0.398*** (−4.390)	0.007 (0.030)	FDI	−0.029** (−2.160)	0.002 (0.760)	−0.029** (−2.220)
FDI	−0.029** (−2.220)	0.003 (0.670)	−0.036*** (−2.950)	URB	0.008** (2.400)	0.001 (1.260)	0.008** (2.200)
URB	0.008** (2.200)	−0.005*** (−3.980)	0.015*** (4.500)	ID	0.003 (1.190)	−0.006*** (−12.800)	0.001 (0.340)
ID	0.001 (0.340)	−0.003*** (−3.660)	0.002 (0.800)	W*GC	0.005 (0.040)	0.082*** (2.860)	0.088 (0.640)
W*GC	0.088 (0.640)	−0.056 (−1.130)	0.231* (1.800)	W*ISU			1.350*** (3.510)
W*ECS			0.925*** (3.480)	W*FDI	−0.036* (−1.790)	−0.002 (−0.470)	−0.049*** (−2.450)
W*ISU	1.350*** (3.510)	0.435*** (3.110)	1.188*** (3.250)	W*URB	0.005 (1.110)	0.003*** (3.100)	0.004 (0.730)
W*FDI	−0.049*** (−2.450)	−0.003 (−0.450)	−0.040** (−2.130)	W*ID	0.004 (1.370)	0.001* (1.630)	0.012*** (3.270)
W*URB	0.004 (0.730)	−0.000 (−0.060)	0.008* (1.660)
W*ID	0.012*** (3.270)	0.005*** (4.030)	0.009*** (2.540)
$ρ$	0.146** (2.260)	0.009 (0.150)	0.052 (0.790)	$ρ$	0.142* * (2.210)	−0.070 (−1.200)	0.146** (2.260)

***p < 0.01, **p < 0.05, *p < 0.1.

(1) Intermediary transmission effect of energy consumption structure

As shown in Table 11 M2, the coefficient of GC on ECS remains significant, accounting to −0.109, indicating that GC can promote ECS through accelerating the utilization of clean energy and thus reducing CO₂ emissions. The coefficient of GC on CO₂ refers to −0.302, passing 1% significance test. On the other hand, M3 presents the results after considering ECS, where the impact path of GC on CO₂ decreases from −0.302 to −0.158. Therefore, ECS generates partial intermediary effect during the process of curbing CO₂ through green credit. Moreover, Table 12 shows the results of Bootstrap test. The indirect effect of GC on CO₂ refers to −0.451, passing the 1% significance test, with the confidence interval not covering zero, which further verifies the partial intermediary effect. Thus, H4 is accepted. The possible reasons may include: GC can facilitate the capital flow into green industry and form industrial clusters; GC may compress credit capital of traditional industries and stimulate their transform to green industries; sustainable green credit policy can promote competition levels and encourage fastening carbon emission reduction.

Table 12.

Robustness test: bootstrap test.

Energy consumption structure
Sobel Test	Indirect effect	$β$	T	P	Confidence interval	(−0.807: −0.241)
Sobel Test	Indirect effect	−0.451***	(3.683)	BC	Confidence interval	(−0.960: −0.490)
Industrial structure upgrading
Sobel Test	Indirect effect	$β$	T	P	Confidence interval	(−0.048–0.272)
Sobel Test	Indirect effect	0.103	(4.782）	BC	Confidence interval	(−0.082–0.088)

***p < 0.01, **p < 0.05, *p < 0.1.

(2) Intermediary transmission effect of industrial structure upgrading

As can be seen from Table 11 M4, the impact path of GC remains significant, referring to −0.259. Considering ISU, M6 shows that the impact path of GC increases from −0.259 to 0.302, which has been strengthened. Moreover, M5 indicates that CG cannot promote ISU. Therefore, there exists no intermediary transmission effect of ISU in the repressing process of GC. Table 12 shows the results of Bootstrap test. The indirect effect of GC on CO₂ refers to 0.103, remaining insignificant, with the confidence interval covering zero, which further verifies the inexistence of intermediary effect. Hence, H5 is rejected. Theoretically, GC provides financial support for ISU and promotes the development of green industry, which subsequently results to effective carbon emission reduction. However, in practice, due to the profit driven by traditional industries, the development of regional GC and ISU has not reached the optimal equilibrium. Therefore, increasing efforts should be paid to effective promotion of industrial structure upgrading.

Conclusions and policy implications

Conclusions

Pursing the “30·60” goal, this research explored how green credit inhibits provincial carbon emissions based on provincial panel data collected from China. In particular, the inhibition effect, threshold effect, and intermediary transmission effect of green credit on carbon emissions were verified. Furthermore, considering regional heterogeneity, differences in impact paths of green credit were clarified. The research findings include: ① Green credit can effectively inhibit China's provincial carbon emissions although there exists spatial heterogeneity, which is consistent with theoretical analysis carried out by Glomsrød and Wei²⁴ and empirical study conducted by Hu³¹. ② During inhibition process of green credit, marketization degree and environmental regulation generate the threshold effects, which is consistent with existing literature.^28–30 ③ Green credit can reduce the proportion of coal consumption and optimize the energy consumption structure to curb regional carbon emissions. However, the intermediary transmission effect of current ISU tends to be insignificant, failing to achieve the purpose of carbon emission reduction. Although the results indicate the existence of intermediary transmission path of industrial structure upgrading, which is verified by current literature, the insignificance of intermediary transmission effect is inconsistent with existing literature.^21,32 Possible reasons may include selection of indicators for industrial structure upgrading or control variables of panel data. Therefore, it is necessary to divide the industrial structure into indicators such as industrial structure upgrading, industrial structure rationalization, and industrial structure adjustment for comparative analysis to clarify the role of industrial structure. ④ Considering endogeneity, green credit can still significantly curb regional carbon emissions, which further verifies the robustness of the research.

Policy implications

Promoting green credit system: Regional government should clarify policy requirements and access standards of green credit policy. Stimulating and restricting mechanisms need to be promoted to optimize the green credit policy and encourage capital flowing into green industry. Innovative green credit products and services can be developed as well as the green credit standard system. Restrictions on traditional industries can be strengthened to create incentives on green investments and green transformation.

Defining scope of threshold effect: Both environmental regulation and marketization should be thoroughly considered to strengthen the preconditions of inhibition process and their intensities should be scientifically defined to consolidate the carbon emission reduction inhibition path and enhance effectiveness of regional carbon emission reduction governance.

Portraying transmission mechanism of CO₂ emission reduction: Green credit reform needs to be deepened to innovate credit products and optimize the allocation efficiency of credit resources. In detail, financial guarantee can be issued for energy structure transformation and industrial structure upgrading, in order to optimize the in-depth transformation of industrial structure and deepen the carbon emission suppression path of green credit.

Clarifying spatial heterogeneity: Regional differences regarding carbon emission reduction and their inhibition paths should be examined. According to the regional characteristics, the carbon emission target should be set differently under the background of innovation driven. Therefore, rational green credit policy system can be formulated consistent with regional development to ensure the reliability and validity of the policy. Regional carbon emission supervision mechanism can be established to accurately grasp the heterogeneity of carbon emission inhibition and adjust the strength of green credit policies, thus optimizing the efficiency of carbon emission reduction policy.

Footnotes

Availability of data and materials

The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Consent for publication

Not applicable.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Ethics approval and consent to participate

Not applicable.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the National Social Science Fund of GuangXi Province, National Natural Science Foundation of China, (grant numbers 20FJL003, 72064009, 72262010, 72264008).

ORCID iD

Yi Xu

Notes

Lingjun Guo is an associate professor in Accounting. His area of research is ecological environment governance and green finance.

Wenyu Tan is an associate professor in Accounting. Her area of research is ecological environment governance and green finance.

Yi Xu is a Lecturer in Finance. Her area of research is green finance, green development and sustainability.

Qinchen Tang is a research associate in Finance. Her area of research is green finance.

Guangfu Liu is a professor in Management Science and Engineering. His area of research is renewable resources and circular economy.

References

Guo

. Carbon neutrality by 2060 leads to fundamental changes in China's economic system. Journal of Beijing University of Technology(Social Sciences Edition) 2021; 05: 1–13.

Anser

Usman

Does air pollution affect clean production of sustainable environmental agenda through low carbon energy financing? Evidence from ASEAN countries. Energy & Environment 2022; 33(3): 472–486.

Fei

Yang

, et al. Analysis of pollution and carbon reduction paths in industrial parks under the carbon peak and carbon neutrality goals. Environmental Protection 2021; 49: 61–63.

Yang

Shang

Deng

. Estimation, decomposition and reduction potential calculation of carbon emissions from urban construction land: evidence from 30 provinces in China during 2000–2018. Environment Development and Sustainability 2022; 24: 7958–7975.

Liu

Tao

, et al. The influence of ecological civilization construction to carbon emission intensity in China's provinces. China Population Resources and Environment 2019; 29: 50–56.

Gui

Xue

;, et al. Research on carbon emissions reduction strategy considering government subsidy and free riding behavior. Environ Eng Sci 2021; 39: 329–341.

Liu

. Preface to the column “research on China's carbon peak”. Resource Science 2021; 43: 637–638.

Wang

, et.al. Does the low carbon city pilot policy achieve the synergistic effect of pollution and carbon reduction? Energy & Environment 2022.

Cai et al. China’s carbon emission pathway under the carbon neutrality target. China Population,Resources and Environment 2021; 31: 7–14.

10.

Wang

. The impact of environmental pollution and green finance on the high-quality development of energy based on spatial dubin model. Resources Policy 2021;74:102451.

11.

Jiang

Wang

, et al. Research on the carbon emission reduction effect of China's green finance development:taking green credit and green venture capital as examples. Financial Forum 2020; 25: 39–48.

12.

Sun

. The impact and transmission mechanism of energy resource endowment on carbon emissions: an empirical analysis based on a spatial econometric model. Business Research 2020; 518(06): 65–74.

13.

Lin

Bian

Wang

. The spatiotemporal evolution characteristics and influencing factors of industrial carbon emission efficiency in Beijing-tianjin-hebei region. Econ Geogr 2021; 41: 187–195.

14.

Song

, et al. Does a carbon emissions trading scheme spur urban green innovation? Evidence from a quasi-natural experiment in China. Energy & Environment 2022; 33: 640–662.

15.

Milindi

Inglesi-Lotz

. Impact of technological progress on carbon emissions in different country income groups. Energy & Environment. Epub ahead of print 2022. DOI:10.1177/0958305X221087507.

16.

D’Orazio

Popoyan

Fostering green investments and tackling climate-related fnancial risks: which role for macroprudential policies?

Ecological Economy 2019; 160 (JUN): 25–37.

17.

Wang

. Low-carbon transformation under the vision of carbon neutrality. Environmental Conformity Assessment 2021; 13: 13–15.

18.

Zheng

. How does green credit affect carbon emissions in China? A theoretical analysis framework and empirical study. Environmental Science and Pollution Research 2022; 29: 59712–59726.

19.

Liu

Ren

. Research on the impact of green credit on low-carbon energy consumption structure. Wuhan Finance 2019; 239(11): 66–70.

20.

Zhu

Zhang

Sun

, et al. Environmental governance constraints and China's economic growth: an empirical analysis of carbon emission control. China Soft Science 2018; 330(06): 163–171.

21.

Xie

Liu

. How does green credit affect the growth of China's green economy? China Population, Resources and Environment 2019; 29: 83–90.

22.

Yan

Wang

. The micro efects of green credit policy — from the perspective of technological innovation and resource reallocation. China Ind Econ 2021; 1: 174–192.

23.

Gianfrate

Peri

. The green advantage: exploring the convenience of issuing green bonds. J Cleaner Prod 2019; 219: 127–135.

24.

Glomsrød

Wei

. Business as UNusual: the implications of fossil divestment and green bonds for financial flows, economic growth and energy market. Energy Sustain Dev 2018; 44: 1–10.

25.

Chen

. Has China's green credit policy been implemented–an analysis of the loan scale and costs based on “two highs and one surplus” enterprises. Contemporary Finance 2019; 06:118–129.

26.

Lyu

. How does green credit promote carbon reduction? A mediated model. Front Environ Sci 2022; 2022(10): 1–10.

27.

Qin

Cao

. Carbon emission reduction effects of green credit policies: Empirical evidence from China. Front. Environ. Sci. 2022; 10: 1–12.

28.

Ruiz

Arboleda

Botero

. A proposal for green financing as a mechanism to increase private participation in sustainable water infrastructure systems: the Colombian case. Procedia Eng 2016; 145: 180–187.

29.

Romano

Scandurra

Carfora

, et al. Renewable investments: the impact of green policies in developing and developed countries. Renewable Sustainable Energy Rev 2017; 68: 738–747.

30.

Xia

. Environmental investment, environmental policy and allocative efficiency of green finance. Research on Technical Economy and Management 2019; 276(07): 68–72.

31.

Mudakkar

Zaman

Khan

, et al. Energy for economic growth, industrialization, environment and natural resources: living with just enough. Renewable Sustainable Energy Rev 2013; 25: 580–595.

32.

Jiang

Zhong

. Impact of green credit on industrial structure in China: theoretical mechanism and empirical analysis. Environ. Sci. Pollut. Res 2020; 27: 10506–10519.

33.

Wang

Xiang

QF.

Adjustment of industrial structure and the potential assessment of energy saving and carbon reduction. China Industrial Economy 2014; 310(01): 44–56.

34.

Song

Xie

Shen

. Impact of green credit on high-efficiency utilization of energy in China considering environmental constraints. Energy Policy 2021; 153: 112267.

35.

Kang

. Bilateral effects of environmental regulation on green innovation. China Population, Resources and Environment 2020; 30(10): 93–104.

36.

Fan

Wang

Zhang

.and Zhu HP. Report on the relative progress of marketization in various regions of China. Economic Research 2003; 03: 9–18 + 89.

37.

Guo

Han

. Foreign direct investment, environmental control and environmental pollution. International Trade Issues 2008; 308(08):111–118.

38.

Liu

Xie

. Analysis of the double threshold effect of industrialization and urbanization in the proceed of environmental economic aggregation. Manage Rev 2017; 29: 21–33.

39.

Wang

Liu

Peng

. Research on the effect of industrial structure upgrading of population aging. China Industrial Economy 2015; 332(11): 47–61.

40.

Anselin

. Spatial Econometrics :Methods and Models. Dordrecht:Kluwer Academic Publishers, 1988, pp256-357.

41.

Hansen

. Threshold effects in non-dynamic panels: estimation, testing, and inference. J Econom 1999, 93(2):345-368.