Do Green Transfer Payments Contribute to Carbon Emission Reduction?

Abstract

Reducing carbon emissions is the top priority for mankind for dealing with climate issues. The Chinese government selected 30 demonstration cities in three batches and provided these cities green special transfer payment funds for three years to build green projects and achieve energy saving and emission reduction (ESER). This policy provides a great opportunity to study whether green transfer payments play an important role in carbon reduction, which has received little attention before. Additionally, the central government set a series of fiscal performance assessment indicators, including the ESER effect, the completion of green projects, and long-term mechanism construction in order to evaluate the effectiveness of the use of funds. This article creatively conducts theoretical analysis from the perspective of performance assessment, takes ESER special transfer payment policy as a quasi-natural experiment, and uses the panel data of 284 cities in China from 2007 to 2017 and 2019 to verify the impact of the green transfer payment on carbon emissions in and after demonstration periods and its mechanisms with the staggered DID method and the new DID Multiplegt (DIDM) model. This article found that green transfer payments could reduce carbon emissions in demonstration cities, and this effect still existed even after policy withdrawal. Mechanism analysis further corroborates that the carbon-reduction effect of green transfer payments could be achieved by increasing the urban green area and improving energy efficiency. Heterogeneity analysis reveals that green transfers have a greater carbon reduction effect on demonstration cities, which receives more green transfer payment funds and has a higher level of financial development. Recommendations related to setting proper performance assessment ESER indicators include improving the management of ESER transfer funds and maximizing the cost–benefit ratio of fiscal funds, which are proposed according to the research conclusions.

1. Introduction

The trend in global warming is increasingly obvious. Climate warming will lead to an increase in extreme weather events and the frequent occurrence of climate disasters [1]. Carbon dioxide and other greenhouse gases produced by human activities are the main cause of global warming [2]. Therefore, mitigating carbon emissions brook no delay.

In order to alleviate global warming and reduce carbon emissions, scholars analyzed the factors affecting carbon emissions. From the perspective of economic and industrial development, Liu and Xiao (2018) verified the EKC hypothesis in China and found the relationship between unit GDP and carbon emissions to show an inverse U-shaped curve, obviously [3]. Yang et al. (2022) revealed that the optimization of the industrial structure significantly inhibited the growth of carbon emissions [4]. From the view of urban planning, Zhang et al. (2016) developed a computable urban economic model, reflecting that land use and transport affect carbon emissions [5]. In addition, Fernando and Hor (2017) thought that energy audit and energy efficiency were two critical factors for reducing carbon emissions [6]. Chen X and Chen Z (2021) found that the development of green finance contributed to carbon emission reduction [7].

Governments have a duty to introduce various carbon emission reduction policies to fight climate change and mitigate carbon emissions [8,9]. In addition to market-oriented environmental regulation, such as the carbon emission trading scheme [10,11,12,13] and tax policies containing carbon tax [14,15,16], environmental tax [17,18,19], and energy tax [20]. Fiscal policies also have an impact on carbon reduction [21]. In 2011, China proposed a special transfer payment policy for energy saving and emission reduction (henceforth ESER). These special transfer payments refer to the special subsidy fund set up by the superior government (generally refers to the central, provincial, and municipal governments in China) to achieve specific macro policy objectives and compensate for some affairs entrusted to the subordinate government. The fund receivers use the fund according to the specified purpose. According to the ESER policy, the central government was issued to transfer payment funds to local governments of demonstration cities in three batches to carry out ESER urban construction. Demonstration periods of each batch lasted three years. After 2017, the demonstration periods of the demonstration cities in the three batches all ended, and the demonstration cities no longer received ESER transfers. Furthermore, in order to test the usage effect of transfer payment funds, a series of performance assessment criteria were set, and the demonstration cities were scored according to the criteria, with corresponding rewards and punishments on the basis of the scores.

With respect to ESER policy, there has been research focusing on its environmental and economic consequences. Lin and Zhu found that the ESER policy could effectively reduce pollutant emissions while ensuring stable economic growth, but the event study revealed that the reduction effect only existed in the demonstration period, which meant that this policy was not sustainable [22]. Zhu et al. explained that this policy reduced industrial SO2 (sulfur dioxide) emissions by 23.8% on average and industrial wastewater emissions by 17.5% on average [23]. Xue and Chen proposed that the policy reduced carbon emissions by 5.4% [24], while Xu et al. thought this figure could be 15.26%, and the policy effect was sustainable [25]. However, both of them thought that the ESER policy achieved a carbon-reduction effect by promoting energy-saving effects, optimizing structure effects, and strengthening green technique effects [24,25]. Tian et al. revealed that this policy could promote the transformation and upgrading of enterprises [26].

By sorting out the literature, we found that research on the factors affecting carbon emissions and the impact of ESER policy is abundant. However, the current research results of the ESER policy’s effect are still controversial. First, the calculation results of the percentage of carbon emission reduction caused by this policy varied greatly. Second, after 3 years of demonstration periods, whether the policy effect would last remained to be seen. Third, the existing literature with one accord supported the fact that promoting the energy-saving effect, optimizing the structure effect, and strengthening the green technique effect were mechanisms of the policy’s impact on carbon reduction. However, these were not the only mechanisms.

Based on the panel data of 284 cities in China from 2007 to 2017 (The latest data are up to 2017 because the ESER policy was completely withdrawn after 2017), this article takes the ESER special transfer payment policy as a quasi-natural experiment, puts forward assumptions through theoretical analysis from the perspective of fiscal performance assessment, and uses a staggered DID model and multiplegt (DIDM) model to verify the impact of green transfers on carbon emissions and its mechanisms. Heterogeneous analysis was carried out in terms of cities’ receiving an amount of green transfer payment funds and financial development levels. Additionally, whether the effect of this policy was sustainable was tested by extending the time window to the year 2019.

The main contributions of this paper are summarized as follows: (1) a new empirical method is used, which can effectively avoid estimation errors and improve the accuracy of the estimation results. On the one hand, the existing research tends to use the two-way fixed effect (henceforth TWFE) DID model to calculate the specific value of carbon reduction caused by the policy. However, many scholars found TWFE to cause “negative weight” when there were multiple policy time coexistences, which resulted in wrong estimated coefficients. On the other hand, in addition to staggered implementation characteristics, the transfer payment policy studied in this paper also had exit situations. In policy withdrawal situations, the carryover effect should be tested as well. Therefore, the robust estimator DIDM proposed by de Chaisemartin and D’Haultfoeuille (2020) [27] was adopted to explore and solve the error caused by multiple policy time coexistences and policy withdrawal. (2) This article empirically tests the carbon reduction effect of the ESER transfer policy during and after the demonstration period, which provides a reference for the sustainable and long-term effects of the policy. (3) In addition to verifying that ESER transfers reduce carbon emissions by improving energy efficiency, this paper found a new influential mechanism that green transfers use to reduce carbon emissions by increasing urban green areas. (4) In terms of demonstration cities with different receiving amounts of green transfer payment funds and different financial development levels, ESER transfers have a heterogeneous effect on their carbon reduction. Demonstration cities receiving more green transfer payment funds with higher levels of financial development have greater effects on carbon reduction.

The remainder of this article is listed as follows. Section 2 introduces China’s ESER special transfer payment policy, carries out theoretical analyses, and proposes research hypotheses. In Section 3, the model and data are introduced. The results of empirical tests and robustness tests are performed in Section 4, where influential mechanism analysis and heterogeneity analysis are also conducted. The last section elaborates on the conclusions and offers suggestions.

2. Background, Theoretical Analysis and Research Hypotheses

2.1. Background of Green Transfers for ESER

Saving resources and protecting the environment are China’s basic national policies. To achieve the targets of carbon emission reduction, since 2011, The Ministry of Finance and the National Development and Reform Commission have selected 30 cities in three batches as demonstration cities to carry out ESER special transfer payment policies. Through sorting out the list of three batches of demonstration cities, this article found that the demonstration cities covered 27 provinces (municipalities and autonomous regions) around the country and were located in the east, central, and western regions. At the same time, these cities had obvious differences in urban scale and the characteristics of carbon emission. There are 10 cities belonging to municipalities directly under the Central Government and urban agglomeration (Class I) or cities specifically listed in the plan and provincial capital cities (Class II). The other 20 ones ordinary cities (Class III). Carbon emissions of 10 cities in Class I and Class II accounted for 55% of the total carbon emissions of all the demonstration cities, while another 20 cities occupied the remaining 45%. The specific list of cities is shown in

Table 1. List of demonstration cities.

Batch No.	Demonstration Cities	Establish Year	Periods of Receiving Transfers
1	Beijing, Shenzhen, Chongqing, Hangzhou, Changsha, Guiyang, Jilin, Xinyu	2011	2012–2014
2	Shijiazhuang, Tangshan, Tieling, Qiqihar, Tongling, Nanping, Jingmen, Shaoguan, Dongguan, Tongchuan	2013	2014–2016
3	Tianjin, Linfen, Baotou, Xuzhou, Liaocheng, Hebi, Meizhou, Nanning, Deyang, Lanzhou, Haidong, Urumqi	2014	2015–2017

. Demonstration periods over the last three years: during the last three years, the central government has given special transfer payment funds (ESER transfers) to the typical green demonstration projects declared and filed by the demonstration cities. Among them, the amount of ESER transfers were determined according to the nature of the demonstration cities. Cities in Class I receive 600 million yuan each year. Class II cities receive 500 million yuan each year, and other cities receive 400 million yuan each year. The ESER transfers are used by the demonstration cities themselves. The central government also carries out comprehensive performance assessments on demonstration cities to ensure the effectiveness of ESER. Comprehensive demonstration performance assessments are divided into the annual performance assessment and overall performance assessment. Among them, the annual assessment results are linked to next year’s ESER transfers. For cities with excellent assessment results, the central government gives an additional 20% of ESER transfers. On the contrary, 20% of ESER transfers are deducted for the cities whose assessment results are unqualified. The overall assessment results are linked to the title of the demonstration city and ESER transfers. For cities that fail to meet the overall ESER targets or have other serious circumstances, their demonstration qualification will be canceled, and all the ESER transfers deducted. Performance assessment includes three aspects: the ESER effect, the completion of green projects, and long-term mechanism construction. Therefore, demonstration cities carry out ESER work mainly based on these three assessment factors.

2.2. Theoretical Analyses and Research Hypotheses

The first assessment factor, the ESER effect, includes indicators such as carbon emissions, energy consumption per unit of GDP, and the main pollutant discharge amount. Carbon emissions, needless to say, and energy consumption per unit of GDP directly link with carbon emissions. The decrease in this indicator can effectively reduce carbon emissions. If carbon emissions or the actual amount of energy consumption per unit of GDP are lower than the target ones, they will affect the assessment results of demonstration cities. Under the supervision of the assessment, demonstration cities positively achieve the target aims, which also leads to a reduction in carbon emissions. So, this article proposes the hypothesis:

H1: 

Green transfers play a role in reducing carbon emissions.

The second assessment factor, the completion of green projects, means that the actual investment completion of typical green projects is declared by demonstration cities themselves. Green projects that are declared mainly contain six aspects: low-carbon-industry development, traffic cleaning, green urban construction, service intensification, the quantification of main pollutants, and scaling up renewable energy utilization. These green projects can realize carbon emission reduction by increasing urban green areas and improving energy efficiency.

Especially on the one hand, green transfers can reduce carbon emissions by increasing the urban green area. Green urban construction includes green buildings, urban green land, and park and wetland construction, which can add vegetation coverage, increase carbon sinks and contribute to carbon emission reduction [28,29,30]. For example, Deyang City carried out the landform restoration and vegetation transplantation for ecological function degradation sites, built three new wetland parks, 1536 km of greenway, and added 760 hectares of green area during the demonstration periods. Additionally, Liaocheng City carried out the Zhougonghe Wetland Project, which included the construction of a subsurface flow wetland, surface flow wetland, and river corridor wetland. At the same time, Tongchuan City implemented the greening of urban construction. All of these projects could increase urban green areas, parks, and wetlands, which can achieve direct carbon fixation by isolating and storing carbon dioxide in the atmosphere through plant photosynthesis, providing a direct way for cities to increase sinks [31]. By enhancing the carbon sink function of urban green areas, the purpose of accomplishing the carbon and oxygen balance in the city or alleviating the carbon sink pressure of forest land outside the city is conducive to the sustainable renewal of urban green area inventory, reducing the impact of the increase in carbon dioxide concentration on human production and life, and playing an important role in achieving the goal of carbon emission reductions. Therefore, this article proposes the following hypothesis:

H2: 

Green transfers reduce carbon emissions by increasing urban green area.

On the other hand, green transfers reduce carbon emissions by improving energy efficiency. Energy consumption is one of the main sources of carbon emissions [32]. Additionally, improving energy efficiency is an essential way to reduce carbon emissions. For example, low-carbon-industry development means that enterprises are supported to apply for advanced energy-saving and environmental protection equipment and eliminate under-developed production capacity. The entry threshold is raised for industries with high energy consumption and high emissions and the energy consumption quota for major energy-consuming products. Because carrying out investment projects based on the internalization of environmental costs has positive externalities, which is helpful for the whole environment, but has little benefit for the enterprises themselves, even having a negative effect on their profitability [33]. ESER transfers can not only subsidize enterprises to replace energy-saving equipment, but can also attract social capital to invest in the energy saving and environmental protection industry, which provides sufficient financing channels for enterprises to increase cash flow. Then, enterprises can introduce advanced and environmentally friendly production equipment and technology, produce green products and develop a circular economy with relatively small risks. This contributes to carbon emission reduction. Additionally, traffic cleaning means promoting energy-saving and new energy vehicles and developing public transport systems. ESER transfers provide automobile enterprises and consumers fiscal subsidies, which encourage people to purchase energy-saving and new energy vehicles. It can also strengthen the construction of public transport. Energy-saving and new energy vehicles can directly promote energy efficiency and mitigate carbon emissions [34]. Additionally, a perfect public transport system improves the willingness of people to travel by public transport. As it is convenient to travel by public transport, the opportunity to drive private cars will be reduced, which is conducive to energy saving and emission reduction. Scaling up renewable energy utilization can also improve energy efficiency. Because in the process of the delivery of basic energy services, extraction, conversion, transportation, transmission, and terminal use generate potential losses. The opportunity to improve the energy efficiency of the whole system exists at every link. First of all, renewable energy, such as wind energy, solar energy, and hydropower, which do not need fuel, will naturally improve efficiency because there is no need for heat conversion. Secondly, the deployment of distributed renewable energy will increase the proportion of renewable energy, and the other primary energy required to provide the same level of energy services will be reduced accordingly, thus minimizing the environmental and economic costs of the entire system and improving energy efficiency [35]. All in all, these green projects supported by ESER transfers can improve energy efficiency, and eliminate under-developed production capacity, which reduces carbon emissions [36,37]. Therefore, this article proposes the following hypothesis:

H3: 

Green transfers reduce carbon emissions by improving energy efficiency.

The third assessment factor is a long-term mechanism construction which contains building local energy saving, an environmental protection market, and carbon emission trading mechanisms in demonstration cities. As shown by a large number of studies, carbon emission trading schemes can improve carbon productivity [38,39,40] and reduce carbon emissions [10,11,12,13]. Moreover, once the carbon emission trading market is established with the support of transfer payment funds, it can be used for a long time and plays a sustainable role in carbon emission reduction, which means that the carbon reduction effect of ESER transfers is long-term effectively. Therefore, this article proposes the following hypothesis:

H4: 

Carbon reduction effect of green transfers is sustainable even after the demonstration periods.

While receiving ESER transfers from the central government, provincial and municipal governments also provide counterpart funds for the construction of green projects in demonstration cities. Take the first batch of demonstration cities as an example. In 2012 alone, the central government allocated 4 billion yuan of ESER transfers to eight of the first batch of demonstration cities, and the provincial government, where the eight demonstration cities were located, also allocated more than 20 billion yuan of counterpart funds. Demonstration cities integrated funds from different sources and made comprehensive utilization of them. For example, as a subordinate county-level city of Deyang City, Mianzhu City coordinated ESER funds at the central, provincial, and municipal levels and cooperated with Deyang to promote the implementation of 24 industrial low-carbon projects, which avoided the “fragmentation” of the use of funds and gave full play to the combined forces of various ESER transfers. Generally speaking, the more ESER transfer payment funds were received from the central government, the more counterpart funds will be provided by the provincial government. For example, some of the demonstration cities that belong to Class I will not only receive more ESER transfer payment funds from the central government each year but also have more corresponding provincial and municipal counterpart funds. Additionally, more green and low-carbon projects need to be built in such cities. What’s more, governments of demonstration cities are also encouraged to make ESER transfers play the leverage role, take advantage of local financial markets, and use ESER transfers to leverage social capital to participate in green and low-carbon construction. For example, Deyang City innovated the mode of capital investment, attracted the participation of social capital, and deeply explored the application of the public-private partnership (PPP) model in the typical ESER demonstration projects. From 2015 to 2016, Deyang City made the central ESER transfers of 39.05 million yuan and played the leverage role, leveraging the social investment of 2.879 billion yuan in typical demonstration projects, achieving a driving effect of 1:73.73.

Therefore, on the one hand, the more ESER transfer payment funds at the central, provincial, and municipal levels that are received, the greener projects need to be built, and the stronger the carbon reduction effect is. On the other hand, the more social funds are guided by green transfer payment funds to invest in green, low-carbon projects, the better the effect of ESER demonstration cities’ construction is. The number of social funds that cities can attract to invest in green projects is related to the financial development level of the cities. Green transfer payments can drive more social funds in demonstration cities with better financial development levels. The more funds are invested in green project construction, the better the carbon reduction effect is. So this article proposes hypotheses H5 and H6.

H5: 

Demonstration cities receiving more green transfer payment funds have greater effect of carbon reduction.

H6: 

Demonstration cities with higher level of financial development have greater effect of carbon reduction.

3. Study Design

3.1. Model Setting

There were 29 cities (Since Haidong City only became a prefecture-level city in 2013, considering the lack of data before 2013, Haidong is not included in the sample) approved as demonstration cities in three batches from the 284 sample cities, which provides a good “quasi natural experiment”. As the cities were approved by stages, the years that demonstration cities received transfers were also different. Therefore, this article refers to existing research [22] using the staggered difference-in-differences method to test the average carbon reduction effect of the green transfer payments by comparing carbon emissions between demonstration cities and the other cities before and after receiving ESER transfers. Additionally, the specific formula (Equation (1)) is listed below.

$${\mathrm{CE}}_{\mathrm{i},\mathrm{t}}{=\mathsf{\beta }}_{0 }{+\mathsf{\beta }}_1{\mathrm{Policy}}_{\mathrm{i},\mathrm{t} }{+\mathsf{\gamma }}_{\mathrm{i}}{\mathrm{X}}_{\mathrm{i},\mathrm{t}}{+\mathsf{\lambda }}_{\mathrm{t} }{+\mathsf{\mu }}_{\mathrm{i} }{+\mathsf{\epsilon }}_{\mathrm{i},\mathrm{t}}$$

(1)

Here, t and i separately represent the year and city. The CE is the dependent variable, which denotes the carbon emissions. The policy is the dummy variable, policy = 1 means that city i belongs to demonstration cities and has received the transfers in that year. Otherwise, policy = 0. X includes the control variables that probably affect the carbon emissions of cities. $\mathsf{\lambda }$ is the time-fixed effect and $\mathsf{\mu }$ is the city-fixed effect. $\mathsf{\epsilon }$ is the random disturbance term. This model estimates the specific impact of ESER transfers on the carbon emissions of cities by observing the coefficient of ${\mathsf{\beta }}_1$.

3.2. Variables and Data

Explained variables are the logarithm of carbon emissions (LnCO₂) and the logarithm of per capita carbon emissions (LnPCO₂). The data of carbon emissions adopted in calculation directly sources from the existing research results [41]. Compared with previous studies, Chen et al. used a particle swarm optimization—back propagation (PSO-BP) algorithm to calculate the carbon emissions and unified the scale of DMSP/OLS and NPP/VIIRS satellite images in the calculation process [41], so the calculation results are more accurate. In addition, this article uses the data of carbon emissions from Carbon Emission Accounts & Datasets (CEADs), which are also widely used by Chinese scholars, for the robustness test.

The explanatory variable is the dummy variable Policy. The list of demonstration cities and demonstration periods of these cities are manually collected by the author on the government websites such as the Ministry of Finance and listed in Table 1.

Control variables are selected referring to existing literature [34,42,43,44], include economic development level (Pgdp), which is measured by the logarithm of per capita GDP; industrialization (Ind), which is measured by proportion of secondary industry in GDP; fiscal expenditure on technology (Tec), which is measured by proportion of fiscal expenditure on technology in fiscal expenditure; population density (Pop), which is measured by the logarithm of population density; level of openness (Open), which is measured by the logarithm of the amount of foreign direct investment; environmental regulation (Er), which is measured by composite index of emission of industrial wastewater, SO₂ and smoke calculated by entropy method All of the data of the cities comes from “China City Statistical Yearbook”, CSMAR and EPS databases (China Stock Market Accounting Research (CSMAR) and Economy Prediction System (EPS) databases are research-based accurate databases in the economic and financial fields developed by China in accordance with China’s actual national conditions. They have been widely used by Chinese researchers). The descriptive statistics of the main variables are listed in

Table 2. The descriptive statistics of main variables.

Variable	Demonstration Cities					Other Cities
	Obs	Mean	Std. Dev.	Min	Max	Obs	Mean	Std. Dev.	Min	Max
CO2	319	4357.981	3513.916	401.619	15,375.227	2805	2624.184	2216.014	209.075	23,071.17
LnCO2	319	8.04	0.877	5.996	9.641	2805	7.598	0.745	5.343	10.046
PCO2	319	879.9267	534.5904	234.878	2647.899	2789	739.636	699.901	59.213	7631.968
LnPCO2	319	6.6225	0.552	5.459	7.882	2789	6.347	0.6923	4.08	8.94
Policy	319	0.273	0.446	0	1	2805	0	0	0	0
Ind	318	48.07	11.776	19.01	75.18	2795	49.791	12.116	9.74	90.97
Pgdp	318	10.701	0.64	9.208	13.056	2795	10.363	0.685	4.595	12.456
Tec	306	2.384	5.881	0.109	100	2659	1.897	6.899	0.067	100
Open	319	10.378	2.66	0	14.941	2805	9.089	3.147	0	14.431
Pop	318	8.005	0.742	5.67	9.383	2795	7.992	0.747	5.513	9.908
Er	319	0.685	0.540	0.0005	2.585	2805	0.705	0.544	0	2.59

.

This paper uses the data from 284 prefecture-level cities from 2007 to 2017, applying the staggered DID model to study the impacts of green transfers on carbon reduction. However, considering that demonstration cities were approved in three batches, there are multiple policy time coexistences. Additionally, after receiving transfers from the central government for 3 years, demonstration cities stopped being treated. Therefore, the sample cannot be simply divided into two groups: the treatment group and the control group. In the case of multiple policy time coexistences and policy withdrawal, the staggered DID estimator using the two-way fixed effect model may have serious errors. On the one hand, the two-way fixed effect model causes “negative weight” when there are multiple policy time coexistences [27,45,46]. On the other hand, demonstration cities stopped being treated after three years, which makes the explanatory variable policy of those cities move from zero to one. Therefore, each batch of the treatment group will become the control group after three years. If there were immediately no impacts in green transfers on carbon emissions of demonstration cities once the ESER policy was withdrawn, the estimator of the two-way fixed effect model would be accurate. However, if the carry-over effect existed, using the two-way fixed effect model would cause bias.

From the perspective of “negative weight”, many scholars have proposed new methods to identify and solve this problem [27,45,47,48]. Additionally, the method DIDM is the most flexible and can be used when the policy is canceled [27]. From the perspective of the carryover effect, the most common solution at present is to delete the sample after the policy withdrawal. Considering the problems of both “negative weight” and carry-over effect, this article, referring to existing research [26], excludes the samples of the first batch of demonstration cities in 2015–2017 and the second batch of demonstration cities in 2017 from the basic regression and follow-up analysis. In the part of Robustness Check, this paper adds the samples of the first batch of demonstration cities in 2015–2017 and the second batch of demonstration cities in 2017 that are back and back and using the DIDM model to check the robustness of staggered DID model used for the case of policy implementation and withdrawal. Additionally, the time window is extended to the year 2019 to test whether the policy has a long-term effect even after withdrawal.

4. Empirical Results Analysis

4.1. Basic Regression Results Analysis

Table 3. The empirical results of basic regression.

Variable	(1) Lnco2	(2) Lnco2	(3) Lnpco2	(4) Lnpco2
Policy	−0.0473 ***	−0.0434 **	−0.0633 ***	−0.0524 **
	(0.0087)	(0.0000)	(0.0055)	(0.0000)
_cons	7.6380 ***	6.7918 ***	6.3749 ***	5.4449 ***
	(0.0000)	(0.0000)	(0.0000)	(0.0000)
Control variable	No	Yes	No	Yes
City fixed effect	Yes	Yes	Yes	Yes
Year fixed effect	Yes	Yes	Yes	Yes
N	3090	2932	3074	2916
adj. R2	0.9898	0.9904	0.9844	0.9863

p-values in parentheses ** p < 0.05, *** p < 0.01.

shows the results of basic regression. Whether the control variables are added or not, green transfers have a reducing effect on the total carbon emissions and per capita carbon emissions of the demonstration cities. The estimated coefficients of the Policy are significant, at least at the level of 5%. Additionally, the $R^2$ value is 99.04% and 98.63%, indicating that our regression model fits actualities accurately. Green transfers reduce about 4.34% of the total carbon emissions and about 5.24% of per capita carbon emissions in demonstration cities. H1 is proved.

4.2. Robustness Check

4.2.1. Parallel Trend Analysis

Parallel trend assumption is an important prerequisite for using the DID model [49]. It requires the trends of the treatment group and the control group to not be significantly different from each other before being affected by the policy. Therefore, this article built the following model to test.

$${\mathrm{CE}}_{\mathrm{i},\mathrm{t}}{=\mathsf{\beta }}_0+{\displaystyle \sum }_{\mathrm{j}=-5}^2{\mathsf{\beta }}_{\mathrm{j}}{\mathrm{Policy}}_{\mathrm{i},\mathrm{t}}^{\mathrm{j}}{+\mathsf{\gamma }}_{\mathrm{i}}{\mathrm{X}}_{\mathrm{i},\mathrm{t} }{+\mathsf{\lambda }}_{\mathrm{t}}{+\mathsf{\mu }}_{\mathrm{i}}{+\mathsf{\epsilon }}_{\mathrm{i},\mathrm{t}}$$

(2)

In Equation (2), the variable ${\mathrm{Policy}}_{\mathrm{i},\mathrm{t}}^{\mathrm{j}}$ is a dummy variable that is generated referring to the relative year of policy implementation. If a city becomes a demonstration city in period t + j, ${\mathrm{the} \mathrm{Policy}}_{\mathrm{i},\mathrm{t}}^{\mathrm{j}}$ will be one; otherwise, it will be zero. The estimator focus of this paper is ${\mathsf{\beta }}_{\mathrm{j}}$. It reflects the impacts on carbon emissions before and after the implementation of the green transfer payment policy. If it is not significantly different from 0 during the period of j < 0 and means that the sample in this paper meets the parallel trend hypothesis. On the contrary, it means that the parallel trend assumption is not satisfied. In addition, when j ≥ 0, it can also depict the dynamic effects of the construction of demonstration cities.

From Figure 1 and Figure 2, whatever the explained variable is in the total carbon emissions or per capita carbon emissions, the regression coefficients of periods before the implementation of this policy fluctuate around zero, and there are no significant changes in this trend, at least at the 5% statistical level. The results reveal that the explained variables of the treatment group and the control group satisfy the parallel trend assumption before the policy’s implementation. Carbon emissions did not change significantly in the year when the green transfers were first received, indicating that the construction in demonstration cities had not yet begun. However, one year later, carbon emissions reduced sharply. Additionally, for the total carbon emissions, the coefficients were significantly different from zero after the implementation of the policy, indicating that the impacts of this policy were continuous. For per capita carbon emissions, the confidence interval in the second year after the implementation of the policy contained zero, which means that the impacts of this policy may disappear. Therefore, whether the green transfer payment policy has long-term effects remains to be seen. In conclusion, the parallel trend assumption is satisfied. The results in Table 3 are robust.

4.2.2. Placebo Test

In order to further verify that the reduction in carbon emissions in demonstration cities was affected by the green transfer payment for ESER rather than other unobservable factors. This article advances the policy implementation year by 1, 2, and 3 years, respectively, according to the distribution of variable Policy in the basic regression and the use of Equation (1) to perform regression estimation again. Listed in

Table 4. Results of placebo test.

Variable	(1)	(2)	(3)	(4)	(5)	(6)
	Lnco2	Lnpco2	Lnco2	Lnpco2	Lnco2	Lnpco2
policy	−0.0149	−0.0147	0.0067	0.0125	0.0173	0.0286
	(0.1157)	(0.1325)	(0.5708)	(0.3602)	(0.2796)	(0.1283)
_cons	6.7538 ***	5.3248 ***	6.7568 ***	5.3275 ***	6.7585 ***	5.3302 ***
	(0.0000)	(0.0000)	(0.0000)	(0.0000)	(0.0000)	(0.0000)
Fake Policy	Pre1	Pre1	Pre2	Pre2	Pre3	Pre3
Control var	Yes	Yes	Yes	Yes	Yes	Yes
City fixed effect	Yes	Yes	Yes	Yes	Yes	Yes
Year fixed effect	Yes	Yes	Yes	Yes	Yes	Yes
N	2965	2949	2965	2949	2965	2949
adj. R2	0.9903	0.9850	0.9903	0.9850	0.9903	0.9851

p-values in parentheses *** p < 0.01.

are the results. The first two columns are estimated coefficients whose policy implementation year is put forward one year, columns (3) and (4) are estimated coefficients whose policy implementation year is put forward two years, and so on.

It can be seen that all of the estimated coefficients are not significant, at least at the level of 10%. This shows the conclusion that the green transfer payment for ESER can reduce carbon emissions in demonstration cities and is not accidental. It has passed the placebo test, further proving the robustness of the estimated results in this paper.

4.2.3. Replace Explained Variable

In order to avoid the lack of robustness of the empirical results caused by a single method of computation for the explained variable, this paper refers to existing research [50] and uses the data of carbon emissions from Carbon Emission Accounts & Datasets (CEADs) as a substitute for the explained variable to conduct empirical analysis again. The empirical results prove the same conclusions and also pass the parallel trend test, which proves that the research conclusions of this paper are robust. The results are shown in

Table 5. Empirical results of the robust check replacing explained variable.

Variable	(1) Lnco2	(2) Lnco2	(3) Lnpco2	(4) Lnpco2
Policy	−0.0466 ***	−0.0434 **	−0.0628 ***	−0.0524 **
	(0.0087)	(0.0189)	(0.0055)	(0.0193)
_cons	12.2438 ***	11.3969 ***	6.3759 ***	5.4449 ***
	(0.0000)	(0.0000)	(0.0000)	(0.0000)
Control variable	No	Yes	No	Yes
City fixed effect	Yes	Yes	Yes	Yes
Year fixed effect	Yes	Yes	Yes	Yes
N	3080	2932	3064	2916
adj. R2	0.9899	0.9904	0.9844	0.9863

p-values in parentheses ** p < 0.05, *** p < 0.01.

.

4.2.4. Empirical Methods Substitution

Unlike traditional DID, which requires that the sample of the treatment group is treated at the same time, and staggered DID, which assumes that the treatment effect does not change, the beginning and ending year of the policy in each batch of demonstration cities are different, which brings great difficulties to the identification, estimation, and inference of the treatment effect. As mentioned above, the latest theoretical literature points out that in the case of staggered treatments, the heterogeneity of the treatment effect among groups and time dimensions may cause negative weight and bias using a two-way fixed effect model. At the same time, this research also puts forward some robust estimators to deal with “negative weight”. Among them, the multiplication DID model of multi-period and multiple units (DIDM) is the most flexible [27]. It can identify the “negative weight” and test the carryover effect after the policy is canceled. Therefore, this article refers to the existing literature, reserves a sample of demonstration cities after policy withdrawal, and uses the DIDM model to test the carbon reduction effect of the green transfer payment and again to check the robustness of the conclusion. This article also extends the time window to the year 2019 and identifies whether the policy has long-term effects after it was eliminated [26].

The DIDM model can tell whether the weight of the average treatment effect on the treated (ATT) was positive or negative. Additionally, all the 87 weights of ATT in basic regression are positive none of them is negative, which indicates that the basic regression used by this article has no “negative weight” and is robust. Furthermore,

Table 6. Results of DIDM model.

Period	LnCO2
	2007–2017			2007–2019
	Estimate	Lower Bound of 95% Confidence Interval	Upper Bound of 95% Confidence Interval	Estimate	Lower Bound of 95% Confidence Interval	Upper Bound of 95% Confidence Interval
Effect_0	−0.0060	−0.0178	0.0057	−0.0073	−0.0189	0.0045
Effect_1	−0.0275	−0.0551	0.0001	−0.0280	−0.0560	−0.0000
Effect_2	−0.0306	−0.0580	−0.0031	−0.0298	−0.0564	−0.0032
Effect_3	−0.0661	−0.0993	−0.0329	−0.0448	−0.0741	−0.0156
Effect_4	−0.0702	−0.1265	−0.0139	−0.0522	−0.0921	−0.0124
ATT	−0.0414	−0.0699	−0.0129	−0.0540	−0.0903	−0.0178
Period	LnPCO2
	2007–2017			2007–2019
	Estimate	Lower Bound of 95% Confidence Interval	Upper Bound of 95% Confidence Interval	Estimate	Lower Bound of 95% Confidence Interval	Upper Bound of 95% Confidence Interval
Effect_0	−0.0066	−0.0221	0.0089	−0.0081	−0.0207	0.0044
Effect_1	−0.0345	−0.0711	0.0021	−0.0354	−0.0671	−0.0036
Effect_2	−0.0345	−0.0707	0.0017	−0.0342	−0.0673	−0.0012
Effect_3	−0.0858	−0.1443	−0.0272	−0.0525	−0.0947	−0.0102
Effect_4	−0.1367	−0.2534	−0.0201	−0.0652	−0.1221	−0.0084
ATT	−0.0556	−0.1028	−0.0084	−0.0652	−0.1194	−0.0109

shows the average treatment effect on the treated, which was calculated by relative time to the period where treatment first changed. The year before receiving the green transfers is set as the base year for event analysis. It can be seen that the trend in carbon reduction for the demonstration cities in the DIDM model is similar to those of basic regression. Additionally, both the total and per capita carbon emissions in demonstration cities decreased significantly after stopping and receiving the green transfers, as the confidence intervals did not contain zero, which indicates that the green transfers had an obvious carry-over effect on carbon reduction. Therefore, it is proper to delete the samples of demonstration cities no longer receiving ESER transfers. At the same time, green transfers reduce total carbon emissions by about 4.14% and mitigate per capita carbon emissions by about 5.56%, which is similar to the results of basic regression. It means that the results of basic regression are robust. Additionally, from the estimated results in the third period and the fourth period after treatment first changed, this article draws a conclusion that the carbon reduction effects of the demonstration cities’ construction are continuous and effective in the long term. Additionally, ESER transfers can reduce total carbon emissions by about 5.4% and decrease per capita carbon emissions by about 6.52% as of 2019. H4 was proved.

4.3. Mechanism Interpretation

As mentioned in Section 2, this article proposes that green transfers reduce carbon emissions by increasing the urban green area and improving energy efficiency. In this part, an empirical method was used to check these two mechanisms. Additionally, the specific model (Equation (3)) is listed below.

$${\mathrm{M}}_{\mathrm{i},\mathrm{t}}{=\mathsf{\beta }}_0{+\mathsf{\beta }}_1{\mathrm{Policy}}_{\mathrm{i},\mathrm{t}}{+\mathsf{\gamma }}_{\mathrm{i}}{\mathrm{X}}_{\mathrm{i},\mathrm{t}}{+\mathsf{\lambda }}_{\mathrm{t}}{+\mathsf{\mu }}_{\mathrm{i}}{+\mathsf{\epsilon }}_{\mathrm{i},\mathrm{t}}$$

(3)

where M is the mechanism variable, and the definitions of other variables are the same as those in Equation (1).

4.3.1. Green Transfers and Urban Green Area

In order to check whether green transfers reduce carbon emissions by increasing the urban green area, this article uses two indicators: the area of parks and green land (Park) and the green covered area in the completed area (Greenbulit), to measure the value of an urban green area. The empirical results are shown in

Table 7. Results of mechanical test.

Variable	(1)	(2)	(3)	(4)
	Park	Greenbuilt	Energy_GDP	Power_GDP
Policy	799.9490 **	2235.9686 **	−0.0511 **	−0.0315 **
	(0.0386)	(0.0127)	(0.0470)	(0.0184)
_cons	3642.7677 **	16,612.8114 *	954.0721 **	0.5140 **
	(0.0500)	(0.0506)	(0.0121)	(0.0118)
Control var	Yes	Yes	Yes	Yes
City fixed effect	Yes	Yes	Yes	Yes
Year fixed effect	Yes	Yes	Yes	Yes
N	2892	2868	2910	2850
adj. R2	0.8976	0.9474	0.4150	0.5224

p-values in parentheses * p < 0.10, ** p < 0.05,

.

The results reveal that green transfers can make demonstration cities increase their areas of parks and green land by about 800 hectares and the increasedgreen covered area in the completed area by about 2236 hectares. Therefore, the results reveal that green transfers play a role in increasing urban green area, which increases urban carbon sinks and reduces carbon emissions in demonstration cities. H2 was proved.

4.3.2. Green Transfers and Energy Efficiency

As mentioned above, energy consumption is a main source of carbon emissions. Additionally, improving energy efficiency can reduce carbon emissions. Energy consumption per unit of GDP (Energy_GDP) is widely used by existing research to verify energy efficiency [24,25]. However, the official statistics of energy consumption per unit of GDP at the urban level are unavailable. It is often confirmed by calculation. This article refers to the method designed by Chen [51]. To avoid the serious bias caused by calculation processes, this article regards the power consumption per unit of GDP (Power_GDP) disclosed in the China City Statistical Yearbook as an examination. The lower the value of the indicator is, the higher the electric energy utilization efficiency becomes, which can reflect changes in energy efficiency to some extent. Additionally, the empirical results are shown in Table 7.

Green transfers can reduce energy consumption per unit of GDP and power consumption per unit of GDP in demonstration cities, and the results are significant at the level of 5%. Therefore, green transfers reduce carbon emissions by improving energy efficiency. H3 was proved.

4.4. Heterogeneity Analysis

In Section 2, this article proposes that demonstration cities receiving more green transfer payment funds have greater effects on carbon reduction. Demonstration cities in Class I receive 600 million yuan each year, those in Class II receive 500 million yuan each year, and other cities receive 400 million yuan each year. Additionally, sources of ESER transfers include the central government transfer payment and the counterpart funds from provincial and municipal public finance. In general, the more ESER transfer payment funds that are received from the central government, the more counterpart funds are provided by the provincial and municipal governments. Therefore, compared with other cities, demonstration cities, which belong to Class I and Class II, have more ESER transfer payment funds at their disposal. This article divides all the demonstration cities into two groups and regresses again. The cities that receive central ESER transfers of 600 million yuan and 500 million yuan each year are divided into cities receiving more ESER transfers, and the other ones fall into a group of cities receiving fewer ESER transfers. The regression results are shown in

Table 8. Results of heterogeneity analysis.

Variable	Cities Receiving More ESER Transfers		Cities Receiving Less ESER Transfers		Cities with Higher Financial Development Level		Cities with Lower Financial Development Level
	Lnco2	Lnpco2	Lnco2	Lnpco2	Lnco2	Lnpco2	Lnco2	Lnpco2
Policy	−0.0814 **	−0.1063 **	−0.0238	−0.0243	−0.0584 **	−0.0830 **	−0.0296	−0.0231
	(0.0280)	(0.0262)	(0.1775)	(0.2013)	(0.0496)	(0.0238)	(0.1348)	(0.2727)
_cons	6.7639 ***	5.4110 ***	6.7229 ***	5.3734 ***	6.8074 ***	5.4565 ***	6.6858 ***	5.3413 ***
	(0.0000)	(0.0000)	(0.0000)	(0.0000)	(0.0000)	(0.0000)	(0.0000)	(0.0000)
Control var	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
City fixed effect	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Year fixed effect	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
N	2748	2732	2843	2827	2790	2774	2801	2785
adj. R2	0.9901	0.9861	0.9900	0.9868	0.9903	0.9864	0.9898	0.9864

p-values in parentheses ** p < 0.05, *** p < 0.01.

. The estimated coefficient of the variable policy shows that the carbon emissions of the demonstration cities receiving more ESER transfers were significantly decreased, while carbon emissions of the other group of demonstration cities were also reduced, but not significantly so. H5 was proved; that is, demonstration cities receiving more green transfer payment funds have greater effects on carbon reduction. This result indicates that the carbon reduction effect of ESER is mainly caused by demonstration cities in Class I and Class II, whose carbon emissions makeup 55% of the total in all the demonstration cities, while demonstration cities in Class III contribute less to carbon reduction. It may also be because municipalities and provincial capital cities tend to have greater pressure on fiscal performance assessments and manage funds more standardly than ordinary cities.

As mentioned above, the green and low-carbon construction of demonstration cities not only depends on the subsidies of fiscal transfer payment funds but also leverages social capital investment through the financial market. Therefore, the financial development level of the demonstration city itself also plays an important role in carbon emissions reduction. If the financial development level of the demonstration city is higher, the transfer payment funds can drive more social funds to invest in green and low-carbon construction. Therefore, this paper referring to the existing literature uses the total amount of deposits and loans per capita in cities to measure the financial development level of cities [52] and divides the demonstration cities into two groups according to this indicator. One of the groups contains demonstration cities with higher financial development levels, and the other one includes demonstration cities with lower financial development levels. Regression results are shown in Table 8. ESER transfers significantly reduce the carbon emissions of demonstration cities with higher financial development levels but also have little impact on demonstration cities with a lower financial development level. H6 was proved.

5. Conclusions and Implications

5.1. Research Conclusions

In order to test whether green transfers reduce carbon emissions or not, this article takes ESER’s special transfer payment policy as a quasi-natural experiment and uses the panel data of 284 cities in China from 2007 to 2017 to verify the impact of green transfers on carbon emissions. Additionally, its mechanisms are systematically investigated. The main conclusions are as follows: (1) Green transfers play a role in mitigating carbon emissions, and the ESER special transfer payment policy can reduce the total carbon emissions by about 4.34% and decrease per capita carbon emissions by about 5.24% in demonstration cities. After a series of robustness tests, this conclusion is still reliable. (2) Based on a new DID Multiplegt (DIDM) model, the event study shows that carbon emissions in demonstration cities follow a decreasing trend over time, which is also significant after the transfers are stopped, indicating that ESER special transfer payment policy has a sustainable effect on carbon emissions reduction. Additionally, ESER transfers can reduce total carbon emissions by about 5.4% and decrease per capita carbon emissions by about 6.52% as of 2019. (3) Mechanism analysis further corroborates that the carbon-reduction effect of green transfers could be achieved through increasing urban green areas and improving energy efficiency. (4) Heterogeneity analysis reveals that green transfers have greater carbon reduction effects on demonstration cities that receive more green transfer payment funds or have a higher level of financial development. It indicates that cities that have more fiscal funds or are able to make full use of local financial markets to leverage a larger amount of social capital to invest in ESER projects and mitigate carbon emissions much better.

At the same time, green projects were declared and filed by the demonstration cities themselves. Which green projects are built in each city, specific construction details of these projects, and the number of transfer payments invested in each of the projects are not available, which hinders this paper to make further efforts in finding out the specific use effect of ESER transfer payments. Additionally, this is the main limitation of this article.

5.2. Implications

Based on the existing research conclusions, this paper puts forward the following policy suggestions:

Above all, set proper performance assessment indicators were related to ESER to make sure that the fiscal funds played an important role in achieving carbon emission reduction. Fiscal funds should be effective in resource allocation, improving the incentive and coordination mechanism between the central and local governments, standardizing the transfer payment system for environmental protection, strengthening the integration of funds and policies, and opening the blocking points of policies and funds in the resource allocation. In the past, the assessment and supervision of local governments were based on economic development. This time, the performance evaluation of the green special transfer payment took the ESER effect as the indicator and achieved good environmental effects. Therefore, the central government should standardize the performance assessment and accountability system and establish a scientific, reasonable, and easy-to-quantify assessment and evaluation system. Proper performance assessments can improve the use efficiency of transfer payment funds.

Furthermore, local governments should use ESER transfer payments seriously and take advantage of the funds according to their own circumstances. Green transfers should be invested in projects that do work in mitigating carbon emissions. Additionally, the local government should disclose the project construction and funds’ usage in detail, improve the transparency of fiscal revenue and expenditure, and actively accept public supervision. Cities can also choose to introduce fiscal policies encouraging urban green area construction and energy efficiency improvement. The carbon-reduction effect of green transfers is achieved through increasing urban green areas and improving energy efficiency. On the one hand, local governments should strengthen municipal construction and increase carbon sinks. Local governments can take steps to strengthen the construction of urban protective green spaces, implement the ecological and landscape restoration project of urban damaged abandoned land and promote projects such as road greening, residential area greening landscape improvement, and municipal infrastructure construction improvement. On the other hand, energy efficiency should be improved continuously. Enterprises are also the main body of energy saving. To improve energy efficiency, enterprises need to carry out green technology innovation, which is a risky activity or re-place environmentally friendly production equipment. Local public finance should introduce policies to build industry-university research cooperation mechanisms and reduce the risks of enterprises’ innovation, or increase their cash flow to replace environmentally friendly production equipment with high energy efficiency.

Last but not least, transfer payment funds should exert leverage as much as possible to drive social capital and build market trading mechanisms to realize carbon emission reduction. Only relying on fiscal subsidies or transfers to achieve carbon reduction will cause huge financial pressure on the government. Enterprises are the main body of carbon reduction. Local government should innovate the system and mechanism, make the limited financial funds fully cooperate with social capital, establish appropriate market trading means, such as carbon emission trading scheme, and let enterprises realize long-term carbon emission reduction through spontaneous ESER according to market changes, and maximize the cost–benefit ratio of fiscal funds.

Finally, ESER special transfer payment policies could be promoted to non-demonstration cities in view of the good results achieved in the demonstration cities. Additionally, non-demonstration cities need to learn from the experience of demonstration cities and declare green projects that are combined with their own characteristics to reduce carbon emissions. Additionally, the formulation of ESER transfer payment policies should pay more attention to fairness. At present, most of the policies and resources are inclined to cities with a higher level of economic development or financial development, municipalities, and provincial capital cities. How to further promote the fairness of environmental policy is the direction of future research.

Other countries with similar conditions to China could refer to this experience and carry out appropriate fiscal policies to achieve carbon reduction.