Do Tradable Green Certificates Promote Regional Carbon Emissions Reduction for Sustainable Development? Evidence from China

Abstract

The tradable green certificate (TGC) scheme is an important approach for mitigating carbon emissions within the context of a renewable energy development strategy and regional sustainable development. However, studies investigating the role of TGCs in encouraging carbon emissions reduction in China are limited and inconclusive due to ignoring the interference of other renewable energy policies and little distinguishing the impact of different green certificates. Using Chinese provincial data from 2013 to 2023, this study employs a difference-in-differences strategy to estimate the effect of the TGC policy on regional carbon emissions. The results reveal that the TGC policy significantly reduces provincial carbon emissions, and this reduction is predominantly contributed by certificate-electricity integration green certificates rather than certificate-electricity separation certificates. A 1% increase in the provincial trade volume of certificate-electricity integration green certificates can reduce total provincial carbon emissions by 0.8–1.3%. These findings hold across a series of rigorous robustness tests. This study also explains the different effects between certificate-electricity integration and certificate-electricity separation green certificates by the concept of additionality. To effectively reduce carbon emissions in the future, the TGC system must meet the requirement of additionality. These insights can provide reference for the improvement of TGC policy to better achieve the carbon reduction objective and sustainable development.

1. Introduction

Tradable green certificate (TGC) systems have been implemented in several countries as a policy measure to promote the development of renewable energy, reduce reliance on traditional fossil energies, and achieve sustainable development [1]. The system issues renewable energy producers TGCs for each unit of electricity generated from eligible renewable energy sources such as solar, wind, biomass, or hydro power. TGCs represent the environmental attributes of renewable electricity and can be bought, sold, or traded on the market separately from the physical electricity to incentivize the generation of renewable energy by earning financial benefits [2]. Although scholars agree that TGC systems can reduce financial pressure on the government to provide subsidies [3], promote market-based investment [4], and increase the installed capacity of renewable energy systems [5], providing a good incentive for the development of renewable energy, previous research fails to provide a consistent answer regarding whether TGC systems promote actual carbon emissions reduction. This study investigates the impact of China’s TGC system on regional carbon emissions to examine the effectiveness of TGCs, and proposes targeted suggestions for future TGC policy to efficiently reduce regional carbon emissions and achieve regional sustainable development [6].

Under the TGC system, renewable energy generation companies obtain environmental value benefits by selling green certificates [7], which helps companies increase their investment in renewable energy and further reduce carbon emissions [8,9]. Electricity users purchase and hold green certificates to demonstrate that they have consumed renewable energy [10], which encourages companies to use more environmentally friendly energy sources, promoting the development of the renewable energy market, reducing carbon emissions, and promoting the development of a low-carbon economy [11,12]. However, previous research does not reach a common consensus on the carbon emissions reduction effects of the TGC system. Huang et al. [13] propose that the TGC system can increase renewable energy installed capacity and replace fossil fuel energy, reducing the carbon emissions. Feng et al. [14] argue that under the TGC system, the growth rate of electricity industry carbon emissions may diminish. Pan and Dong [15] use a dynamic computable general equilibrium (CGE) model to explore the coupling effect of TGC trading on a low-carbon economy, finding that the TGC system reduces carbon emissions and mitigates negative environmental externalities. In contrast, some scholars argue that the TGC system has not been effective in promoting carbon emissions reduction [16]. Using information disclosed by 115 companies, Bjørn et al. [17] find that companies purchasing green certificates claiming that they use renewable electricity with zero carbon emissions exaggerate actual emissions reduction and do not adopt additional renewable energy generation, and therefore cannot effectively contribute to regional carbon emissions reduction. Brander et al. [18] plot a green certificate supply–demand curve, suggesting that the green certificate market has difficulty in generating additional renewable energy investment and increasing renewable energy generation, failing to promote regional carbon emissions reduction. Liu et al. [19] examine the TGC market, finding that TGC does not significantly promote industrial enterprises’ green innovation and has no impact on carbon emissions reduction. Yan et al. [20] also argue that TGC trading in China has an uncertain effect on carbon emissions reduction and requires systematic reassessment, and the availability of additionality is key to advancing the ability of TGCs to reduce carbon emissions and improve the environment.

Why does the existing literature come to inconsistent conclusions? There may be two important reasons. Notably, when research examines the carbon reduction effects of TGC systems, the potential interference and impact of other renewable energy-related policies is often ignored, which may be an important reason for previous inconsistent conclusions. The impact of RPS policy must be considered when investigating China’s TGC system [21]. The RPS policy specifies the proportion of renewable energy power that must be reached in electricity consumption [22], determines the renewable energy consumption obligations of fossil energy producers, renewable energy producers, grid enterprises, and consumers, and TGC purchase by obligated entities becomes a necessary means to achieve consumption goals [23]. This means that the RPS policy can promote the effective consumption of renewable energy power, reducing carbon emissions [24]. Another reason for the inconsistent conclusions regarding the effect of the TGC system in China may be that scholars do not distinguish between the different categories of green certificates. As noted, China’s TGC system has successively implemented two TGC models (i.e., certificate-electricity integration and certificate-electricity separation). Significant differences exist in the additionality of carbon reduction between these two models [25]. If the two models are not distinguished, their different roles will be mixed, leading to biased conclusions. Bjørn et al. [17] argue that only the certificate-electricity integration model exerts an additional effect on actual carbon emissions reduction, but this perspective requires further empirical evidence for support.

To sum up, existing studies examine the impact of green certificate trading on carbon emissions reduction but do not reach consistent conclusions, leaving research gaps that merit further investigation. First, previous studies do not make a clear distinction between the effects of TGCs on carbon emissions reduction and similar policies such as the renewable portfolio standard (RPS) in China. Second, limited research engages comparative studies on the effects of TGCs on carbon emissions under certificate-electricity integration and certificate-electricity separation models. Third, although some literature examines the effect of TGC trading on carbon emissions reduction, it does not further explain the underlying mechanisms.

In order to achieve the goal of carbon emissions reduction and sustainable development, this study empirically estimates the effects of the TGC system on carbon emissions and explores the possible mechanisms based on the actual circumstances in China, which is currently the largest carbon emitter in the world. First, based on the different trading characteristics of China’s TGC market in each province, this study employs a difference-in-differences (DID) strategy to accurately estimate the impact of TGCs on carbon emissions at the provincial level and also applies the DID strategy to strictly control for the RPS system, to exclude the interfering effects. Second, this study respectively analyzes the different effects of the two TGC models (certificate-electricity integration and certificate-electricity separation) on carbon emissions. Finally, based on the concept of carbon emissions reduction and sustainable development, this study examines and explains the effects of TGCs on carbon emissions reduction. The results demonstrate that although the TGC system in China has effectively promoted overall carbon emissions reduction, these effects are primarily attributable to certificate-electricity integration TGCs rather than certificate-electricity separation TGCs, which implies that only certificate-electricity integration TGCs have actual carbon emissions reduction additionality, and future policies should encourage the issuance of integrated green certificates.

Compared with the existing literature, this study makes three marginal contributions. First, it assesses the effects of the TGC system on China’s carbon emissions reduction comprehensively and accurately, and enriches the related research in the field. This study takes the implementation of the green certificate policy as a quasi-natural experiment and constructs a DID model for estimation, and also constructs an interaction term to exclude the possible influence of the RPS system, providing more scientific evidence for assessing the effect of the TGC policy. Second, this study specifically analyzes the mechanism of carbon emissions reduction by dividing certificate-electricity integration and certificate-electricity separation TGCs and explains the limitations of the latter from the perspective of additionality. Third, this study concludes that certificate-electricity separation TGCs do not significantly promote carbon emissions reduction, which provides valuable insights and practical reference for improving related TGC policy, aiding regional governments’ carbon emissions reduction and achieving sustainable development.

The remainder of the paper is organized as follows. Section 2 introduces the TGC system in China; Section 3 details the study’s materials and methods; Section 4 outlines the empirical results; Section 5 is the discussion, and Section 6 is the conclusion and policy implications.

2. The TGC System in China

To promote the large-scale development of renewable energy, the Chinese government has been subsidizing grid connection of wind and solar power generation since 2006. However, with advances in new energy generation technology, the costs of solar and wind power have achieved grid parity [26], and continuous massive subsidies have put enormous pressure on the government’s finances. In 2017, the Chinese government implemented the TGC system to alleviate the financial pressure caused by renewable energy subsidies, hoping to employ market approaches to obtain additional benefits for the environmental rights of green power in place of government subsidies.

TGCs represent the environmental value of renewable energy power, where sellers (renewable energy companies) can obtain environmental value benefits through TGC trading and buyers (electricity companies or individuals) obtain environmental rights and can demonstrate that they are using green, low-carbon renewable energy. The original intent of the TGC system was to promote renewable energy development through issuing certificates to renewable energy companies in the form of financial subsidies. This increased funding support from the TGC market can encourage power generation companies to invest more in energy storage equipment technology innovation to produce more renewable energy power, ultimately reducing carbon emissions and promoting sustainable development [27,28].

In the initial stage, the Chinese TGC system adopted the certificate-electricity separation model, meaning that renewable energy (i.e., green electricity; GE) and TGCs were traded separately. In this case, companies can sell TGCs to other companies that need to demonstrate the use of renewable energy on the TGC trading market, while retaining the corresponding electricity for internal use or selling it in the electricity market, and the purchaser of this portion of electricity cannot claim that they have used renewable energy. This trading model decouples TGC prices and electricity prices, leading to fluctuations in TGC market prices and low trading volume. China established the RPS in 2019 to further promote the development of renewable energy, setting market shares for renewable energy generation for each province. RPS implementation enabled renewable energy power demanders to meet quotas by purchasing TGCs, which increased the TGC trading volume [24]. Under the RPS, the TGC system is still based on the certificate-electricity separation model, and the trading activity of the TGC market did not change significantly, indicating that the TGC policy implementation effect was still not substantial enough [25].

In September 2021, directed toward sustainable development, China began to pilot and explore the certificate-electricity integration model, which binds TGC and GE trading together. In this case, while settling the results of green power trading, power users also obtain the corresponding TGCs for the settlement of GE, ensuring that the GE traded is directly used for the purchaser’s electricity demand. The certificate-electricity integration model improves market transparency, reduces costs, promotes market liquidity, and increases the TGC trading volume and advances the development and use of renewable energy. The framework of China’s TGC system is illustrated in Figure 1.

At the same time, due to differing distribution of renewable energy resources and electricity demand in each province, the number of green certificates traded varies considerably from province to province in the Chinese market. Figure 2 shows the cumulative volume of TGC trading by province in China up to December 2023 [The data are obtained from the China Green Power Certificate Trading Platform: https://www.greenenergy.org.cn/, (accessed on 24 July 2024)]. Eight provinces, including Ningxia, Jilin, and Hebei, among others, have traded more than one million certificates, among which Ningxia has the highest trading volume, with more than 3 million certificates, followed closely by Jilin. Notably, 20 regions have traded less than 500,000 certificates, and 6 regions, including Hainan province, traded less than 10,000 certificates. Statistics on the volume of TGCs traded in each province show significantly different impacts of the TGC system, indicating that carbon emissions reduction and sustainable development effects can be determined by comparing the impacts on high- and low-volume provinces.

3. Materials and Methods

3.1. Theoretical Framework and Research Design

Based on the theory of externalities, there are “carbon externalities” because the impacts of the economic activities of enterprises on carbon emissions are not reflected in production costs. In order to internalize these carbon externalities, the TGC system, as an important policy innovation mechanism, plays an important role in promoting carbon emission reduction and sustainable development. Under the green certificate system, the government requires electricity companies to include a certain percentage of renewable energy power in their power supply mix by setting renewable energy quotas, thus creating market demand for green certificates. It is conducive to raising the price of green certificates, promoting the development and construction of more renewable energy projects, and thus promoting carbon emission reduction. In addition, enterprises can realize their own carbon emission reduction targets through the purchase of green certificates, which is conducive to reducing the cost of carbon emission reduction and enhancing the efficiency of carbon emission reduction and sustainable development.

This study explores the effects of China’s TGC system on regional carbon emissions. Since the official implementation of the TGC system in 2017, China’s TGC market has presented the problem of large differences in trading volume between different regions. According to Figure 2, only eight provinces had a total TGC volume of more than one million certificates, including Jilin, Heilongjiang, Ningxia, Hebei, Liaoning, Anhui, Shandong, and Jiangsu, and other provinces are less than one million. The TGC system has not produced substantial effects in the provinces with less trading volume, which constitutes the condition for the quasi-natural experiments in this study using a DID strategy based on the varied TGC trading volume in different provinces. It is more accurate for the DID strategy to estimate the impact of the TGC policy on regional carbon emissions by controlling for potentially differing variables in the experimental and control groups and excluding possible influencing factors.

The study’s research framework is presented in Figure 3. First, a DID strategy is constructed to capture the impact of the TGC system on carbon emissions under circumstances with and without controlling for the RPS policy, respectively. Second, a series of robustness tests are conducted for the DID estimated results, including common trend, placebo, and heterogeneity tests, and excluding the effects of contemporaneous policies to ensure the accuracy of the baseline regression results. Finally, the possible mechanisms of TGCs reducing carbon emissions are further examined from the perspective of additionality.

3.2. Models

Although previous studies adopt different methods to assess the impact of TGC systems on carbon emissions such as qualitative analysis [14,16,24], ordinary least squares regression with fixed effects [17,19,23], the CGE model [15,22], and supply–demand curve analysis [20,25], studies fail to control for the influences of other policy factors. The DID method can control for individual and time-fixed effects through within- and between-group differences, separating policy effects from other factors that may affect the empirical results, to address the endogeneity problem [29].

This study takes the eight provinces with TGC trading above one million as the experimental group and those with less than one million as the control group. The effect of the TGC system on carbon emissions can be accurately captured by comparing the differences between the experimental and control groups before and after TGC system implementation. The DID model is set as follows:

$$C{E}_{\mathrm{it}}={\alpha }_0+{\alpha }_{1}G{P}_{\mathrm{i}}\ast P{\mathrm{ost}}_{\mathrm{t}}+{\alpha }_2{X}_{\mathrm{it}}+{\gamma }_{\mathrm{i}}+{\lambda }_{\mathrm{t}}+{\epsilon }_{\mathrm{it}}$$

(1)

where ${CE}_{it}$ is the total carbon emissions accounting of province I in year t; ${GP}_i\ast {Post}_t$ is the DID term in which ${GP}_i$ represents whether the TGC system has a notable impact on province i. When the volume of TGC trading in province i is more than one million, ${GP}_i$ equals 1, otherwise it equals 0. ${Post}_t$ is a time dummy variable that measures the policy shock of the TGC system. When the year is after 2017, it is assigned a value of 1, and 0 otherwise. The coefficient ${\alpha }_1$ is the estimated coefficient that measures the effect we are concerned with. $X_{it}$ is a set of control variables related to the regional TGC system and carbon emissions. ${\gamma }_i$ and ${\lambda }_t$ are province- and time-fixed effects, respectively, and ${\epsilon }_{it}$ is the error term.

Since the RPS policy and the TGC system overlap in time and are strongly correlated, the estimation of Equation (1) is likely to be influenced by the RPS policy, which may lead to biased estimation. This study constructs a new interaction term with the renewable energy portfolio indicators of each province and the year dummy variables, introducing it into Equation (1) as follows:

$$C{E}_{\mathrm{it}}={\beta }_0+{\beta }_{1}G{P}_{\mathrm{i}}\ast P{\mathrm{ost}}_{\mathrm{t}}+{\beta }_2{Q}_{\mathrm{i}}\ast T_{\mathrm{t}}+{\beta }_3{X}_{\mathrm{it}}+{\gamma }_{\mathrm{i}}+{\lambda }_{\mathrm{t}}+{\epsilon }_{\mathrm{it}}$$

(2)

where $Q_i\ast T_t$ is the interaction term to control for the RPS policy, in which $Q_i$ is the indicator quota of RPS in each province, and $T_t$ represents the year dummy variable of the implementation of RPS policy, which is 1 when the year is before 2019 and 0 otherwise. Controlling for the RPS policy will obtain more accurate results.

3.3. Variables and Data

3.3.1. TGC Volume

The TGC transaction volume in each province is obtained through the China Green Power Certificate and Trading Platform. Total volume includes wind power and photovoltaic TGC transactions. This study calculates the total number of TGCs traded in each province and divides the treatment and control groups according to whether a province exceeded one million transactions to apply the DID strategy. According to the standard of whether GE and TGCs are integrated, the number of green electricity certificates (GECs) and nongreen electricity certificates (NGECs) in each year from 2013 to 2023 is classified, which can be used to identify the impact of different types of TGC trading on carbon emissions.

3.3.2. Carbon Emissions

This study applies two methods to measure each provincial region’s annual carbon emissions. The first method applies the Intergovernmental Panel on Climate Change (IPCC) energy factor conversion method, which is as follows:

$$C{E}_{\mathrm{it}}={\displaystyle \sum E_{\mathrm{itj}}\ast {\eta }_{\mathrm{j}}}$$

(3)

where ${CE}_{it}$ is the total carbon emissions of province i in year t, E_itj is the consumption of energy j in province i in year t, and ${\eta }_j$ is the carbon emissions coefficient of energy j. As the consumption of various energy sources are physical statistics in the original statistics, they must be converted into standard statistics when measuring carbon emissions. According to the China Energy Statistical Yearbook, energy consumption types are divided into nine categories, including raw coal, coke, crude oil, gasoline, kerosene, diesel oil, fuel oil, natural gas, and electricity, and the associated carbon emissions coefficients are obtained from the China Energy Statistical Yearbook.

The second method applies the publicly available carbon emissions database produced by [30], which is from the global atmospheric carbon dioxide concentration simulation grid dataset. Based on the remotely sensed product of troposphere carbon concentration from 2002 to 2023 (AIRx3C2M 005), the dataset is developed using an improved sinusoidal model for each grid cell. This study calculates each Chinese province’s annual carbon emissions using ArcGIS 10.8 software and this dataset.

3.3.3. RPS Indicator Quota

Renewable energy portfolio indicators are obtained from the National Energy Administration. According to the RPS policy, the National Energy Administration regulates renewable energy power portfolio indicators for each provincial administrative region at an annual amount for each province based on the renewable energy resources, national energy planning, cross-provincial and cross-regional transmission channel construction, operation conditions, and other factors.

3.3.4. Other Control Variables

This study also controls for the following macro variables related to provincial energy consumption. (1) Economic development. Faster economic development means higher production and consumption, which causes more energy consumption and carbon emissions. This study measures economic development using each province’s gross domestic product (GDP). (2) Population. Zhang et al. [31] assert that an inverted U-shaped curve relationship exists between regional population size and carbon emissions. (3) Industrial structure. This study uses the share of secondary industry output to measure industrial structure because the secondary industry includes manufacturing, construction, and other production and processing-based industries, which generally produce a large amount of greenhouse gas emissions [32]. (4) Urbanization rate. It is generally believed that urbanization is significantly and positively related to carbon emissions [33]. The process of urbanization involves the use of large amounts of energy in buildings, transportation, and industry, which also causes environmental problems such as changes in land use, traffic congestion, and waste pollution, and impacts the volume of regional carbon emissions. The descriptive statistics of variables are listed in

Table 1. Descriptive statistics.

Variables	Unit	Obs	Mean	Std. Dev	Min	Max
TGC volume(ln)	One certificate	330	13.534	1.564	0.693	15.385
Green TGC volume(ln)(GE)	One certificate	330	13.245	1.345	0	14.943
Nongreen TGC volume (ln)(NGE)	One certificate	330	12.142	1.264	0	14.298
Carbon emission(ln)(IPCC accounting)	1000 kg	330	10.797	0.364	9.005	11.302
Carbon emissions(ln) (remotely sensed dataset)	1000 kg	330	10.864	0.224	9.424	11.642
RPS	%	330	0.125	0.046	0.014	0.254
GDP(ln)	million yuan	330	10.627	0.865	7.543	11.564
Population size(ln)	million people	330	8.642	0.764	6.536	9.775
Share of secondary industry	%	330	0.475	0.077	0.430	0.676
Urbanization rate	%	330	0.556	0.068	0.332	0.869

.

4. Results

4.1. Baseline Results

The benchmark regressions are conducted using Equations (1) and (2), and the results are presented in

Table 2. Baseline results.

Variables	Explained Variable: Carbon Emission(ln)(IPCC Accounting)				Explained Variable: Carbon Emission(ln)(Remotely Sensed Dataset)
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
GP × Post	1.413 ***	1.137 ***	0.656 ***	0.558 ***	1.543 ***	1.325 ***	0.463 ***	0.356 ***
	(0.125)	(0.114)	(0.147)	(0.153)	(0.145)	(0.155)	(0.133)	(0.136)
Q × T			0.948 ***	0.845 ***			0.955 ***	0.844 ***
			(0.243)	(0.241)			(0.243)	(0.264)
GDP		0.143 **		0.075 ***		0.164 **		0.044 **
		(0.071)		(0.022)		(0.079)		(0.021)
Population size		0.024		0.014		0.027		0.016
		(0.052)		(0.011)		(0.058)		(0.011)
Share of secondary industry		0.243 ***		0.186 **		0.298 **		0.127 **
		(0.086)		(0.087)		(0.145)		(0.067)
Urbanization rate		0.045		0.022		0.044		0.032
		0.033		0.021		(0.041)		(0.030)
Area-fixed effect	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Time-fixed effect	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
N	330	330	330	330	330	330	330	330
R2	0.724	0.778	0.815	0.817	0.754	0.776	0.816	0.848

Note: ** p < 0.05, *** p < 0.01. Control variables include GDP, population size, share of secondary industry output, urbanization rate, etc.

. Columns (1)–(4) in Table 2 are the regression results using the IPCC carbon emissions accounting data, while columns (5)–(8) are the estimation results based on carbon emissions calculated from the remotely sensed dataset.

In the IPCC accounting data regression results, the estimated coefficients of the DID interaction term for the TGC system are all positive and significant at the 1% level, indicating that the TGC system significantly contributes to carbon emissions reduction in each province. Further comparison of the estimation results indicates that after controlling for the RPS policy, although the estimated coefficients of the DID term of the TGC system are still significantly positive at the 1% level, the absolute values decrease significantly, while the estimated coefficients of the RPS interaction terms (Q × T) are also significantly positive. This indicates that both the TGC system and the RPS policy can effectively promote regional carbon emissions reduction. Regardless of whether the IPCC data or the remotely sensed dataset are employed, the coefficients of the concerned interaction term (GP × Post) are significantly positive, suggesting that the TGC system significantly reduces regional carbon emissions, and this conclusion still holds after controlling for the RPS policy.

4.2. Common Trend Test

The basic prerequisite assumption for the DID approach to test the validity of a treatment effect is the presence of a pretreatment common trend, confirming that the explained variables in treatment and control groups have similar trends before the exogenous policy shock. The time effect in the control group can be used to eliminate unobservable time effects in the treatment group only if the common trend assumption holds. This study uses the event study method used by [34], where year dummy variables are regressed as explanatory variables for 4 years before and after policy implementation, using the year of policy implementation as the base group. In Figure 4 (using IPCC accounting data) and Figure 5 (using remotely sensed dataset), the coefficients reflect the differences between the treatment and control groups in a given year, with the dashed line representing the 95% confidence interval, revealing that before TGC implementation in 2017, the estimated coefficients fluctuate around 0 (the 95% confidence interval includes the value of 0). This indicates that the pre-TGC policy carbon emissions of treatment and control groups do not significantly differ and the basic regression satisfies the ex-ante common trend hypothesis.

4.3. Placebo Test

To further exclude the effect of changes in carbon emissions being influenced by other policies or stochastic factors, this study conducts a placebo test using the method of randomly generated experiments proposed by [35,36]. The study’s sample includes 30 regions, 8 of which are regions with over one million TGCs traded. Accordingly, 8 of the 30 regions are randomly selected and set as a pseudo treatment group. To avoid the interference of other small probability events on the estimation results, this study repeats the above process 500 times for regression analysis. Figure 6 and Figure 7 present the distribution of the estimated coefficients for the 500 randomly generated treatment groups. The results of the IPCC accounting data reveal that the distribution of the regression coefficients obtained based on the random sample estimation is around 0, conforming with a normal distribution. Combined with the baseline estimated coefficient of 0.558 in column (4) of Table 2, the actual estimated coefficients of this real sample differ significantly in size from the mean value of 0 of the estimated coefficients in the placebo test. This implies that since it is a small probability event that other factors influenced the estimated results of the benchmark regression, it can be assumed that the contribution effect of the TGC system on carbon emissions reduction is not disturbed by omitted variables. The results of the remotely sensed dataset reveal that the kernel density estimates of the coefficients of the stochastic process are distributed around 0, conforming with a normal distribution. The estimated coefficient of the benchmark in column (8) of Table 2 is 0.356, which is larger than the vast majority of simulated values and can be considered extreme. This indicates that the true estimation results are unlikely to be driven by chance alone, and the results of the benchmark regression are robust and valid.

4.4. Heterogeneity Test

The previous analysis demonstrates that the TGC system can significantly reduce the regional carbon emissions level. However, do the carbon emissions reduction effects of TGCs vary across different types of green certificates? Examining this question can help to understand the mechanism and boundary conditions of the TGC system more comprehensively. Based on the criteria of whether renewable GE and GECs are certificate-electricity-integrated, this study divides GECs into GE and NGE certificates. The interaction term of DID and NGE green certificates is added to Equations (1) and (2) for further estimations, and the results are presented in

Table 3. Impact test of NGE TGCs on regional carbon emissions.

Variables	Explained Variable: Carbon Emission(ln)(IPCC Accounting)		Explained Variable: Carbon Emission(ln)(Remotely Sensed Dataset)
	(1)	(2)	(5)	(6)
GP × Post × NGE	0.345	0.313	0.242	0.246
	(0.753)	(0.754)	(0.646)	(0.662)
GE × Post	0.716	0.687	0.642	0.610
	(0.853)	(0.854)	(0.786)	(0.757)
NGE	0.137	0.165	0.157	0.186
	(0.343)	(0.363)	(0.355)	(0.324)
Q × T		0.846 ***		0.974 ***
Control variables		(0.263)		(0.175)
Province-fixed effect	Yes	Yes	Yes	Yes
Time-fixed effect	Yes	Yes	Yes	Yes
N	330	330	330	330
R2	0.847	0.848	0.828	0.831

Note: *** p < 0.01. Control variables include GDP, population size, share of secondary industry output, urbanization rate, etc.

. The NGE TGCs do not effectively reduce regional carbon emissions, which may be due to a lack of additionality of NGE TGCs that will be discussed in further detail in Section 6. However, the estimated coefficient of Q × T is significantly positive, indicating that regional carbon emissions decrease under the RPS. This may be because the RPS makes regional CE consumption a mandatory constraint, which can significantly improve the level of renewable energy consumption and decrease regional carbon emissions. Therefore, NGE TGCs do not effectively promote regional carbon emissions reduction, and the RPS causes regional carbon emissions reduction under this trading model.

This study next introduces the interaction term of the DID term and GE TGCs into Equations (1) and (2) for regression to identify the effect of GE TGC trading on regional carbon emissions, and the results are presented in

Table 4. Impact test of GE TGCs on regional carbon emissions.

Variables	Explained Variable: Carbon Emission(ln)(IPCC Accounting)		Explained Variable: Carbon Emission(ln)(Remotely Sensed Dataset)
	(1)	(2)	(5)	(6)
GP × Post × NGE	1.632 ***	1.367 ***	0.848 ***	0.826 ***
	(0.237)	(0.253)	(0.144)	(0.178)
GE × Post	0.742	0.753	0.624	0.626
	(0.853)	(0.823)	(0.745)	(0.745)
GE	0.375	0.357	0.476	0.475
	(0.453)	(0.475)	(0.448)	(0.435)
Q × T		0.863 ***		1.073 ***
Control variables		(0.222)		(0.138)
Province-fixed effect	Yes	Yes	Yes	Yes
Time-fixed effect	Yes	Yes	Yes	Yes
N	330	330	330	330
R2	0.826	0.839	0.826	0.834

Note: *** p < 0.01. Control variables include GDP, population size, share of secondary industry output, urbanization rate, etc.

. The estimated results under IPCC accounting and the remotely sensed dataset show that GE TGCs can significantly contribute to regional carbon emissions reduction. Specifically, a 1% increase in GE TGC trades can reduce regional carbon emissions by 0.8–1.3% on average, and this conclusion still holds after controlling for the RPS policy. Compared with the NGE TGCs, GE TGCs improve market transparency, reduce costs, and promote market liquidity, promoting an increase in the volume of TGC trade. This can generate more subsidy funds for enterprises to further increase the development and use of renewable energy, and further decrease carbon emissions.

4.5. Additional Robustness Tests

4.5.1. Excluding Municipalities

Municipalities in China are generally better equipped with economic development policies and financial resources, and are more likely to be pilot areas for national policies and may have better policy implementation effects than other cities, leading to biased estimates. To exclude this potential interference, this study removes the four major municipalities of Beijing, Shanghai, Tianjin, and Chongqing from the sample for a robustness test. The results of the GE TGC models are presented in columns (1) and (4) of

Table 5. Exclude municipalities, PSM-DID and reconstruct the control group (GE TGC trading).

Variables	Explained Variable: Carbon Emission(ln)(IPCC Accounting)			Explained Variable: Carbon Emission(ln)(Remotely Sensed Dataset)
	(1)	(2)	(3)	(4)	(5)	(6)
GP × Post × GE	0.864 ***	1.545 ***	1.743 ***	0.733 ***	0.963 ***	1.128 ***
	(0.153)	(0.423)	(0.054)	(0.212)	(0.254)	(0.135)
GP × Post	0.845	0.937 ***	1.045 **	0.836	0.626 ***	0.827 ***
	(0.834)	(0.234)	(0.492)	(0.764)	(0.228)	(0.225)
GE	0.257	0.344	0.547	0.243	0.863	0.767
	(0.463)	(0.535)	(0.374)	(0.638)	(0.745)	(0.692)
Q × T	1.126 ***	0.926 ***	0.848 **	1.154 ***	1.757 ***	1.757 ***
	(0.248)	(0.285)	(0.434)	(0.232)	(0.462)	(0.465)
Control variables	Yes	Yes	Yes	Yes	Yes	Yes
Province-fixed effect	Yes	Yes	Yes	Yes	Yes	Yes
Time-fixed effect	Yes	Yes	Yes	Yes	Yes	Yes
N	257	164	227	257	164	227
R2	0.886	0.917	0.910	0.886	0.927	0.914

Note: ** p < 0.05, *** p < 0.01. Control variables include GDP, population size, share of secondary industry output, urbanization rate, etc.

, revealing significantly positive results, which indicates that the promotional effect of GE TGCs on renewable energy investment remains significant after excluding the influence of regional economic development. The results of the NGE TGC models are shown in columns (1) and (4) of

Table 6. Exclude municipalities, PSM-DID and reconstruct the control group (NGE TGC trading).

Variables	Explained Variable: Carbon Emission(ln)(IPCC Accounting)			Explained Variable: Carbon Emission(ln)(Remotely Sensed Dataset)
	(1)	(2)	(3)	(4)	(5)	(6)
GP × Post × GE	0.456	0.567	0.739	0.314	0.344	0.327
	(0.727)	(0.464)	(0.454)	(0.463)	(0.627)	(0.634)
GP × Post	1.143	0.983	1.033	0.646	0.538	0.535
	(0.686)	(0.777)	(0.847)	(0.668)	(0.568)	(0.579)
NGE	0.677	0.438	0.744	0.435	0.354	0.244
	(0.864)	(0.354)	(0.685)	(0.644)	(0.533)	(0.466)
Q × T	0.852 ***	0.949 ***	0.863 **	1.868 ***	1.576 **	1.257 ***
	(0.234)	(0.325)	(0.413)	(0.475)	(0.553)	(0.353)
Control variables	Yes	Yes	Yes	Yes	Yes	Yes
Province-fixed effect	Yes	Yes	Yes	Yes	Yes	Yes
Time-fixed effect	Yes	Yes	Yes	Yes	Yes	Yes
N	257	164	227	257	164	227
R2	0.884	0.912	0.893	0.871	0.921	0.903

Note: ** p < 0.05, *** p < 0.01. Control variables include GDP, population size, share of secondary industry output, urbanization rate, etc.

, and the coefficients are insignificant, implying that the effect of these certificates on regional carbon emissions is still insignificant after excluding the effect of regional economic development, which is consistent with the previous conclusions.

4.5.2. Propensity Score Matching-DID Estimation

Considering the possible interference of unobservable characteristics among different regions, this paper applies the PSM-DID method to further mitigate the endogeneity of the models. The results for GE TGCs are presented in columns (2) and (5) in Table 5. The coefficients of DID are significantly positive in both models, supporting the previous finding that GE TGCs promote regional carbon emissions reduction. The results for NGE TGCs are shown in columns (2) and (5) in Table 6. The coefficients of DID are still insignificant, confirming that NGE TGCs have no significant impact on carbon emissions after applying PSM-DID estimation.

4.5.3. Reconstructing the Control Group

Considering that some bias may occur in dividing the experimental and control group based on the total number of TGCs traded exceeding one million, this study further excludes provinces and regions (including Hubei, Guangxi, Qinghai, Gansu, Shanxi, Guangdong, and Shaanxi) with the total number of transactions between 100,000 certificates and one million TGCs from the original control group, and retains only provinces with the total number below 100,000 TGCs as the control group. These provinces have fewer TGCs (less than 100,000) than the benchmark experimental group (more than one million) and can be considered to be negligibly affected by the TGC system compared to the experimental provinces. The results for the GE TGCs are presented in columns (3) and (6) in Table 5, and are also significantly positive in both models, implying that the baseline conclusion that GE TGCs promote carbon emissions reduction is robust. The results for NGE TGCs are shown in columns (3) and (6) in Table 6, and the regression coefficients remain insignificant, indicating that the NGE TGCs have no significant effect on regional carbon emissions reduction.

4.6. Excluding the Effects of Contemporaneous Policies

TGC system implementation occurred in a critical period in which China’s pollution control policies are continuously strengthened. Other environmental policies launched during the same period may also have an impact on the effects of TGC and regional carbon emissions. This study considers the carbon trading pilot policy, low-carbon pilot policy, the central environmental protection inspector, and key regional air pollution control policies, introducing dummy variables for each of these policies into the benchmark regression equations.

The Chinese government launched a carbon emissions trading pilot policy in 2011, officially approving Beijing, Tianjin, Shanghai, Chongqing, Hubei, Guangdong, and Shenzhen as pilot regions. The carbon emissions trading pilot policy can stimulate energy-intensive industries to save energy use and adopt measures to reduce carbon emissions, further affecting regional carbon emissions [37]. This study introduces a dummy variable for carbon trading pilot regions to the baseline regression equation, presenting the estimated results for GE TGC trading in columns (1) and (5) in

Table 7. Exclude the impact of related policies on GE TGC trading in the same period.

Variables	Explained Variable: Carbon Emission(ln)(IPCC Accounting)				Explained Variable: Carbon Emission(ln)(Remotely Sensed Dataset)
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
GP × Post × GE	1.127 ***	1.244 ***	0.836 ***	1.236 ***	0.838 ***	0.846 ***	0.818 ***	0.846 ***
	(0.313)	(0.243)	(0.131)	(0.134)	(0.233)	(0.275)	(0.237)	(0.258)
Control variables	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Province-fixed effect	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Time-fixed effect	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
N	330	330	330	330	330	330	330	330
R2	0.854	0.854	0.862	0.864	0.832	0.833	0.828	0.832

Note: *** p < 0.01. Control variables include GDP, population size, share of secondary industry output, urbanization rate, etc.

. The regression coefficients of carbon emissions reduction remain significantly positive. The estimated results for NGE TGC trading are shown in columns (1) and (5) in

Table 8. Exclude the impact of related policies on NGE TGC trading during the same period.

Variables	Explained Variable: Carbon Emission(ln)(IPCC Accounting)				Explained Variable: Carbon Emission(ln)(Remotely Sensed Dataset)
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
GP × Post × NGE	0.316	0.352	0.321	0.363	0.257	0.253	0.267	0.246
	(0.438)	(0.465)	(0.427)	(0.452)	(0.248)	(0.245)	(0.244)	(0.247)
Control variables	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Province-fixed effect	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Time-fixed effect	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
N	330	330	330	330	330	330	330	330
R2	0.836	0.832	0.832	0.834	0.825	0.825	0.822	0.822

, and the coefficients of carbon emissions reduction in both models remain insignificant, verifying the robustness of the baseline conclusions of this study.

Second, this study examines the low-carbon pilot policy. In 2010, the Chinese government issued a policy on low-carbon cities and selected eight cities in five provinces as national low-carbon pilot cities. This scope was further expanded to 80 cities in 6 provinces in 2017. Ma et al. [38] propose that the low-carbon pilot cities policy formulates renewable energy development plans and renewable energy policies targeting technology research and development to inspire renewable energy innovation in various industries and accelerate the development of the renewable energy industry in a continuous industrial energy transition in low-carbon cities, effectively promoting regional carbon emissions reduction. This study introduces a dummy variable for provinces with a low-carbon pilot city to the benchmark estimation models, presenting results of GE TGCs in columns (2) and (6) in Table 7. The GE TGCs significantly contribute to regional carbon emissions reduction. The results in columns (2) and (6) in Table 8 show that the NGE TGC system does not have a significantly positive effect on regional carbon emissions reduction, once again indicating that the baseline results remain robust after controlling for low-carbon pilot policies.

Third, the central environmental protection inspector. In 2016, China issued the Central Environmental Protection Inspectorate (CEPI) policy, known as the environmental protection inspector. The CEPI was established under the leadership of the Ministry of Environmental Protection, with the participation of leaders from the Central Commission for Discipline Inspection and the Organization Department of the Central Committee, as an environmental protection inspector on behalf of the Communist Party of China and the State Council for provinces’ Party committees and governments. The inspectors requested the establishment of a sound energy policy system for green development, accelerated energy structure adjustment, high-quality development of renewable energy, and effective reduction of carbon emissions. Similarly, this study’s controls for this policy can be controlled by introducing a dummy variable into the benchmark models. The results of GE TGCs are shown in columns (3) and (7) in Table 7, and the estimated coefficients of carbon emissions are significantly positive at the 1% level, while the regression coefficients of NGE TGCs affecting carbon emissions remain insignificant, as shown in columns (3) and (7) in Table 8. The central environmental protection inspectors policy has no significant effect on the influence of GE and NGE TGCs on carbon emissions.

The Ministry of Ecology and Environment of the Chinese government issued the Atmospheric Pollution Transmission Channel Cities of the Beijing–Tianjin–Hebei Region policy which consists of 26 prefecture-level cities under the jurisdiction of 4 surrounding provinces (Hebei, Shanxi, Shandong, and Henan), with Beijing and Tianjin as the center. The policy aims to promote and use clean production such as expanding the use of renewable energy to control air pollution in the future as the primary approach to reducing carbon emissions [39]. Introducing dummy variables, this study controls for the impacts of this policy and conduct regressions. The results of GE TGCs are shown in columns (4) and (8) in Table 7, and the estimated coefficients of carbon emissions are significantly positive, indicating that the key regional air pollution prevention and control policy does not significantly interfere with the impact of GE TGCs on regional carbon emissions reduction. While columns (4) and (8) in Table 8 show that the estimated coefficients of the impact of NGE TGCs on regional carbon emissions remain insignificant, confirming that the benchmark conclusions are robust after controlling for the effects of key regional air pollution control policies.

5. Discussion

Assessing the actual effect of TGC policy on regional carbon emissions reduction is of great significance for achieving sustainable development. The results of DID estimations presented above demonstrate that although the TGC system promotes regional carbon emissions reduction overall, its carbon reduction effects are primarily attributable to GE TGCs rather than NGE TGCs. Previous studies analyze TGCs as an inseparable single object, ignoring different models [14,20,40], obtaining the inconsistent conclusions on the effects of the TGC system on regional carbon emissions reduction and sustainable development [13,15,17]. In contrast to this research, this study further investigates the structural characteristics of the TGCs, i.e., the effects of different TGC categories on carbon emissions reduction, determining that the effects may vary across these categories. Only the certificate-electricity integration TGC model has real carbon emissions reduction effects, whereas the certificate-electricity separation TGC model does not. From this perspective, this is a novel finding that reaches beyond the existing literature. This section further analyzes the possible reasons that only GE TGCs promote regional carbon emissions reduction from perspectives of the additionality of carbon emissions reduction and the supply–demand relationship of TGCs.

5.1. The Additionality of GE and NGE TGCs

According to Bjørn et al. [17], the TGC system produces real carbon emissions reduction effects only when the TGC possesses “additionality”, which stems from externality theory in economics. The IPCC defines additionality as the “reduction in CO₂ emissions relative to a certain baseline scenario”. A GE TGC that possesses additionality means that the emissions reduction effect brought by GE TGCs is additional and did not exist prior to TGC system implementation. Under certificate-electricity integration trading, since TGCs are bundled with GE power sales, the TGC purchaser is the consumer of GE, representing a direct reduction in carbon emissions, and if the enterprise does not purchase TGCs, the corresponding emissions reduction effect will not occur [41].

In contrast, when NGE TGC trading does not have additionality, it means that the associated emissions reduction effects are not additional because these emissions reduction effects already existed and were included in other emissions reduction targets. As the TGCs and GE are not bundled together for trading, renewable energy generation companies may have already included the corresponding emissions reduction amounts in other emissions reduction targets before trading NGE TGCs. In this case, the purchaser cannot achieve additional emissions reduction by trading NGE TGCs; therefore, such trading cannot be considered as having additionality.

5.2. TGC Supply–Demand

According to Brander et al. [18], when TGC supply in the market exceeds demand, the excess TGCs in the market will not have a practical reduction effect on current carbon emissions. Since the carbon emissions reduction of the TGCs are included in the regional power grid when a green certificate is produced, enterprises that purchase those excess TGCs will not actually reduce new carbon emissions, but will only re-count the previous carbon reduction for themselves again [42].

In the current Chinese TGC market, TGC supply far exceeds demand. On the demand side, the demand for GE TGCs is higher compared to NGE TGCs because they can improve market transparency, reduce costs, and promote market liquidity, and demand is higher. However, on the supply side, the supply of NGE TGCs is higher, primarily because a large number of TGCs were issued at the early stage of policy implementation, and a massive number of NGE TGCs were issued. Since the initial phase was based on the certificate-electricity separation TGC trading model, there are a huge number of NGE TGCs [25]. Consequently, we conclude that the supply of NGE TGCs far exceeds demand, and this state of oversupply will result in a lack of additional motivation for enterprises to develop renewable energy, failing to effectively promote carbon emissions reduction and sustainable development.

Only with the certificate-electricity integration TGC model can TGC trades go with the consumption of GE to have a substitution effect on the original coal power and reduce carbon emissions. Therefore, the fact that TGC supply in the Chinese market far exceeds demand is also a significant reason why the TGC policy does not have an emissions reduction effect under the certificate-electricity separation model.

6. Conclusions and Implications

It is crucial to assess the actual carbon emissions reduction effect of China’s existing TGC system to improve the future TGC system and effectively promote carbon emissions reduction and sustainable development. Although the existing literature presents some empirical studies, such research fails to reach consistent conclusions due to lack of accurate assessment strategies or structural decomposition of TGCs. This study constructs a DID strategy to precisely estimate the carbon emissions reduction effects of the TGC system in China. By controlling for the impact of RPS policy and using different carbon emissions accounting methods, the results demonstrate that the TGC system has significantly reduced carbon emissions in provincial regions in China. This study further divides TGCs into GE and NGE certificates for analysis, determining that the carbon emissions reduction effect of the TGC system is attributable to GE TGCs rather than NGE TGCs.

The study also references additionality properties to explain the differentiated results. The carbon emissions reduction of NGE TGCs has been accounted for in real-time electricity trading. If the physical (electricity) and environmental (carbon emissions reduction) attributes of GE are separately introduced into the electricity market and the TGC market in different periods (i.e., NGE TGCs), the amount of carbon emissions reduction will be double-counted, so that the NGE TGCs are not additional. GE TGCs combine physical and environmental attributes and will not calculate carbon emissions reduction twice. Moreover, China’s current oversupply situation in the TGC market includes more NGE TGCs, further weakening the additionality of TGC market.

The study scientifically evaluates the carbon reduction effect of TGC policy and enriches the research on the TGC system and carbon reduction. In addition, this study provides important insights for improving the TGC system to reduce carbon emissions and reach sustainable development goals in the future. Notably, some international organizations such as the European Union and the Climate Group RE100 (100% renewable electricity) have not fully recognized the validity of China’s TGCs due to unreliable additionality from the potential double-counted risk of carbon emissions reduction. In the future, China’s TGC system should focus on three primary considerations. First, the government should cease issuing NGE TGCs and purchase existing NGE TGCs via government procurement. Second, the separation of TGC and GE markets may increase the complexity and time cost of intermediate transactions, and may also lead to additional transaction costs. To address these challenges, the government should integrate the current TGC and GE markets and establish a unified GE TGC trading market to ensure the additionality of TGC trading. Finally, the government should actively align with international TGC standards to promote the recognition and trading of Chinese TGCs in the international market.

Of course, this study has some deficiencies, such as the lack of in-depth data support for the discussion on TGC additionality, and follow-up studies can further conduct empirical research on the additionality of the TGC system. In addition, follow-up studies can conduct further research on recommendations for differentiated management strategies tailored to different provinces.