How economic growth pressure impact carbon emissions: Evidence for China

Abstract

The article explores the impact of economic growth pressure on carbon emissions based on panel data from China’s 277 cities. Moreover, the article analyzes the underlying influence mechanisms as well as regional heterogeneity. The results demonstrate that economic growth pressure significantly increases carbon emissions. Technological innovation and foreign trade constitute the channels through which economic growth pressure affects carbon emissions, but the mediating mechanism of industrial structure upgrading does not exist. Concretely, economic growth pressure increases carbon emissions by reducing technological innovation and foreign trade. In Western China, economic growth pressure has the highest impact on carbon emissions. In central and western China, economic growth pressure has a significant positive effect on carbon emissions. On the contrary, the effect of economic growth pressure on carbon emissions is significantly negative in Eastern China. In Northeast China, the positive effect of economic growth pressure on carbon emissions is statistically insignificant.

Keywords:

• economic growth pressure
• CO₂ emissions
• CO₂ emissions per capita
• political promotion tournament
• mediating effects
• regional heterogeneity

JEL CODES:

• A10
• D01
• Q54

1. Introduction

China contributes the largest share of global CO₂ emissions as the world’s largest developing country. On September 22, 2020, China announced its carbon peaking by 2030 and carbon neutrality by 2060, namely China’s ambitious goal of "double carbon". As early as the Copenhagen Climate Conference in 2009, the Chinese government proposed a plan to reduce both CO₂ emissions and carbon intensity, and then CO₂ emissions reduction became an important performance indicator for environmental governance and even local assessment. Under the governance system of "centralized political power and decentralized economic power," local officials in China have a decisive influence over economic reforms and resource allocation and control (Bo, 2020). Therefore, they can directly or indirectly influence regional CO₂ emissions. Although the political promotion tournament theory offers a variety of economic explanations for the behavior of China’s local officials (Zhou, 2007), economic growth, which is relatively easy to measure, remains the most important indicator of local development performance. The GDP-first assessment system undoubtedly creates important incentives for local officials, and the pressure for local officials’ promotion comes mainly from the economic growth in their jurisdictions (Lu et al., 2022).

Numerous existing studies have confirmed that economic growth often contradicts CO₂ emissions control (e.g. Kais & Sami, 2016; Shahbaz et al., 2013; Zhang & Zhang, 2021). However, with the increasing importance of climate governance, the Chinese government has paid more and more attention to environmental protection, and accordingly, CO₂ emissions reduction has also become an important political performance assessment indicator. Thus, under the dual pressure of economic growth and CO₂ emissions reduction, how local governments make decisions constitutes a critical issue. The trade-off between GDP primacy and CO₂ emissions reduction has become a contradictory choice for local officials in the face of promotion pressure. Nevertheless, the traditional notion of GDP primacy is still deeply entrenched in China; therefore, there are frequent cases of local governments being held accountable for environmental problems caused by economic growth. Consequently, it is necessary to understand the choices of local governments under the economic development pressure in the context of increasingly stringent environmental regulations and the rise of the "double carbon" target as a national strategy. Although many studies have examined the relationship between China’s economic growth and CO₂ emissions, to our best knowledge, few studies have focused on the effect of economic growth pressure (EGP) on CO₂ emissions.

Economic growth targets reflect officials’ decision-making behavior more than economic development under the political promotion tournament theory. The jurisdictional growth targets, which are directly set by and accountable to officials, are more direct evidence of officials’ influence on economic growth. Economic growth targets serve as both an assessment criterion for superiors to subordinates and a performance commitment from subordinates to superiors. Therefore, when localities fail to achieve economic growth targets, the pressure on officials increases significantly. In other words, officials’ pursuit of political promotion is internalized in the pressure for economic growth in their jurisdictions. The EGP is characterized by setting economic growth targets, which is clearly different from economic development. Given this background, the article’s primary purpose is to investigate how EGP affects regional CO2 emissions. As mentioned earlier, local governments have absolute power in economic reforms and resource allocation, so we further explore how economic growth pressure affects CO2 emissions through reforming the economic system and allocating resources. Finally, due to the significant developmental differences within China, the impact of EGP on CO2 emissions is theoretically regionally heterogeneous. The article also attempts to enrich the knowledge of the relationship between EGP and CO2 emissions and contribute to relevant decision-making by investigating such heterogeneous effects. In sum, the article addresses three theoretical questions: the effect of EGP on CO₂ emissions, the channels through which EGP affects CO₂ emissions, and the possible regional heterogeneity of EGP’s influence on CO₂ emissions.

By addressing the above questions, the article has the following significant contributions. Firstly, the article is the first to investigate the effect of EGP on CO₂ emissions using econometric methods, which helps to understand the endogenous decision-making of local officials. Because economic growth pressure is an important representation of the political promotion tournament theory, this study also significantly extends the scope of the political promotion tournament application. Secondly, the article discusses the influence mechanism under which how local EGP affects CO₂ emissions using mediating effects models, thus deepening and enriching the theoretical understanding of the relationship between economic growth and CO₂ emissions. Thirdly, this study also explores the regional heterogeneity of the effects of EGP on CO₂ emissions through the grouped regression method, thus enriching the knowledge of such effects in different socioeconomic contexts and contributing to decision-making adapted to local conditions. Finally, from the practical perspective, the article’s findings will improve a more rational promotion mechanism that accommodates the double dividend of economic development and CO₂ emissions control. From the methodological perspective, this paper provides more fine-grained evidence on the relationship between EGP and CO₂ emissions using city-level panel data. Moreover, we address the endogeneity of the model through instrumental variables tests and confirm the reliability of the paper’s findings through various other robustness tests.

2. Literature review

2.1. Influence factors of CO₂ emissions

The factors causing CO₂ emissions include two aspects. One is the market environment factors, among which economic growth has received the most attention. The other is the policy environment factors, namely some governmental decisions.

Extensive prior studies have explored the influence of the economy on CO₂ emissions. Economic development requires significant amounts of energy, especially fossil energy and resources, in production, consumption and transport, resulting in CO₂ emissions (Mujtaba et al., 2022; Jia et al., 2022; Huo et al., 2022). In these studies, GDP per capita is often used to measure regional economic growth. Zhang and Zhang (2021), Shahbaz et al. (2013), and Kasman and Duman (2015) argued that economic growth Granger causes CO₂ emissions. Kais and Sami (2016) and Razzaq et al. (2021) supported the significant positive effects of economic development on CO₂ emissions. Economic growth measured as per capita income is also available, and the conclusions confirm its positive impact on CO₂ emissions (Ozturk & Acaravci, 2013). However, the positive effects of economic development on CO₂ emissions do not always exist. The lack of long-term causality or the inverted U-shaped link between the two indicates the dynamic nature of the relationship (Soytas & Sari, 2009; Arouri et al., 2012). Besides, Acheampong (2018) argued that economic development negatively affects CO₂ emissions, but this is region-specific and exists only globally and in Latin America.

Also, renewable energy use significantly reduces CO₂ emissions (Mujtaba et al., 2022; Zhang et al., 2022; Feng & Nie, 2022; Sun et al., 2022; Balsalobre-Lorente et al., 2022; Kirikkaleli et al., 2022). Other market environment elements include green technology (Milindi & Inglesi-Lotz, 2022; Wang et al., 2022; Sun et al., 2022; Koseoglu et al., 2022; Khattak et al., 2022), industrial structure adjustment (Zhao et al., 2022; Huo et al., 2022; Li et al., 2021), trade openness (Li & Haneklaus, 2022; Udeagha & Ngepah, 2022; Wang & Zhang, 2021), foreign direct investment (Djellouli et al., 2022; Balsalobre-Lorente et al., 2022), development assistance (Wang et al., 2022), finance (Ozturk & Acaravci, 2013; Wang et al., 2022; Lv et al.,2022; Kirikkaleli et al., 2022), geopolitical risk (Adebayo et al., 2022), and population (Razzaq et al., 2021; Zhang & Cheng, 2009).

Undoubtedly, government decisions significantly impact CO₂ emissions, particularly in terms of the development and promotion of climate and energy policies. Carbon taxes and the CO₂ emissions trading scheme (ETS) have received the most attention and are considered the most effective climate policy tools for mitigating CO₂ emissions (Haites, 2018). In China, ETS is by far the only mandatory and considered a key tool for achieving its emissions reduction targets (Fang et al., 2021; Cao et al., 2019). Liu et al. (2022), Zhang and Zhang (2019), and Zhang et al. (2022) found that ETS can effectively reduce CO₂ emissions and carbon intensity in China. Besides, Deng et al. (2018) argued that the ETS has put some pressure on Chinese companies, which has led them to set purposeful reduction targets. Zhang and Cheng (2021) suggested that ETS has a negative impact on CO₂ emissions from the service sectors in China, but such an effect varies across regions and industries. The ETSs in Latin America and the European Union have also effectively reduced CO₂ emissions (Oliveira et al., 2020; Pietzcker et al., 2021).

Carbon taxes are another effective tool for reducing emissions and are considered more efficient than ETS due to their easy implementation and broader coverage (Jia & Lin, 2020). Ahmadi et al. (2022) concluded that a carbon tax could stimulate economic growth while achieving emission reductions. Although carbon taxes have yet to be implemented in China to date, many studies have examined their possible impact in China due to their excellent performance in reducing emissions. For example, Ding et al. (2019) and Zhang and Zhang (2018) found that a higher carbon tax rate helps reduce China’s CO₂ emissions. Of course, carbon taxes are not a single policy instrument per se but are accompanied by their own diversification such as tax offsets (Wang-Helmreich & Kreibich, 2019), differentiated tax rates (Zhang & Zhang, 2018), as well as a mix of other policy instruments such as ETS (Zhang et al., 2022) and energy subsidies (Yi & Li, 2018; Jia & Lin, 2021). Energy reforms, particularly energy subsidies, can also contribute to a low-carbon transition. For instance, Yi and Li (2018) evidenced that subsidizing low-energy products helps save energy and reduce emissions. However, support for fossil energy has a negative impact on CO₂ emissions reduction (Lin & Ouyang, 2014; Lin & Xu, 2019; Husaini et al., 2021). In summary, renewable energy subsidies are widely used worldwide to stimulate more renewable energy use and reduce reliance on fossil energy, ultimately promoting CO₂ emissions reduction (Kalkuhl et al., 2013; Zhang et al., 2017).

2.2. EGP and CO₂ emissions

Although the literature has identified the impact of government actions on CO₂ emissions, it has yet to explore the endogenous dynamics of this behavior. These actions are more often the result of administrative directives from above. For example, China’s various new energy developments and ETS are top-down executive orders, subject to the central government decisions, and ultimately implemented in an orderly and selective manner at the local level. Theoretically, local governments have no spontaneous endogenous motivation for CO₂ emissions mitigation, especially under the promotion assessment mechanism centered on economic growth; on the contrary, they have a vigorous pursuit of economic development, that is, GDP growth (Lu et al., 2022). In the theoretical framework of the political promotion tournament in China, EDT is closely linked to the promotion assessment of officials, thus local governments at all levels in China guide and manage the economic growth of their jurisdictions by setting EDT (Su et al., 2012). Concretely, the local governments refer to higher-level government’s EDT and at the same time set their EDTs based on the incentives they face. The EDT is the primary basis for macroeconomic management and economic policy orientation for the coming period. As a result, local heads of government have a solid incentive to set high EDT in order to meet their potential political advancement. This pattern of EDT setting leads to significant EGP on local governments.

EGP systematically influences the policy orientation and intensity of local governments. When the EDP rises, local officials will be more active in designing and introducing policies to promote economic growth, together with a variety of resources and instruments (Zhou, 2007; Wu et al., 2021). In this way, EGP will have diverse consequences. For example, Wang et al. (2021) indicated that EGP contributes to innovation capacity but shifts to curb it as EGP increases. The effect is more pronounced in regions with a higher degree of marketization. Wang et al. (2021) evidenced that when local leaders are conservative and short-sighted in the promotion tournament, local enterprises prefer to avoid venture long-run innovative investments. Li et al. (2021) discovered that the greater the EDT is, the more local governments rely on and allocate better resources to firms that can contribute more quickly to local economic growth. Zhu and Lin (2022) and Lu, Ya and Wang et al. (2022) recently found the significant effects of EGP on energy performance and land approval, respectively.

In conclusion, the factors influencing CO₂ emissions and the diversified effects of EGP have received more attention; however, the role of EGP in affecting CO₂ emissions has rarely been investigated. Hence, there is an urgent need for us to explore the impact of EGP on CO₂ emissions and thus better reveal the endogenous motivation of officials to reduce carbon emissions in the context of the political promotion competition and the achievement of the "double carbon" target. Moreover, in contrast to the existing effects of EGP, we further focus on the transmission mechanisms by which EGP affects CO₂ emissions, thus enriching the meaning of the "political promotion tournament" under the background of low-carbon development. Finally, of course, the heterogeneity found in the influence on CO₂ emissions and the impact of EGP has inspired this paper to explore the heterogeneity of the impact of EGP on CO₂ emissions.

3. Methodologies

We build an econometric model to investigate the effect of EGP on CO₂ emissions. To more robustly examine such effects, we introduce various control variables and control two-way fixed effects, as formulated in the model (1). Introducing control variables is to prevent as much as possible the endogeneity problems caused by omitted variables since omitted variables may affect both the explanatory and explained variables and lead to estimation bias. The usual control variables are observable and vary over time and area; thus, we control unobservable and non-time-varying factors and thus control two-way fixed effects. (1) $$\mathit{carbo}{n}_{i,t}=\alpha +{\beta }_1{p}_{i,t}+\phi \mathit{co}{\mathit{n}}_{i,t}+{\gamma }_i+{\lambda }_t+{\epsilon }_{i,t},$$(1) where carbon denotes CO₂ emissions, p denotes the EGP, con denotes other potential factors affecting CO₂ emissions, namely the control variables in the model (1), the subscripts i and t denote region and time. γ_i and λ_t represent region fixed effects and time fixed effects, ε_i,t represents the stochastic error term. The coefficient β₁ captures the effect of EGP on CO₂ emissions.

Typically, carbon emissions are measured as CO₂ emissions per capita. In order to perform a robustness check on the findings of the model (1), we additionally use total CO₂ emissions to represent regional CO₂ emissions as a new independent variable, which is also accepted in existing studies (e.g. Liu et al., 2022; Zhang et al., 2022). EGP is expressed as the ratio of the EDT to last year’s economic growth rate. Usually, local governments set growth targets based on past development practices and current year development plans, which put growth pressure on local governments. Therefore, the EDT is a visual indicator of the growth pressure on local governments (Li et al., 2021). Obviously, the higher ratio means a greater EGP. In order to ensure a common trend in the growth pressure data, we further process the relevant data according to equation (2)(2) $$p_{i,t}=(\mathit{goa}{l}_{i,t}+1)/(\mathit{growt}{h}_{i,t-1}+1)-1.$$(2) , taking into account that some cities had negative economic growth rates in individual years. (2) $$p_{i,t}=(\mathit{goa}{l}_{i,t}+1)/(\mathit{growt}{h}_{i,t-1}+1)-1.$$(2)

Here, goal represents the EDT, and growth represents last year’s economic growth rate. We respectively add 1 to the current year’s economic growth target and the previous year’s actual growth rate and then use the new data to get the ratio of the two. This ratio is then minus 1 to obtain final data on EGP. Also, we choose a new proxy for EGP to perform robustness checks. The new proxy is measured as the difference between the EDT and last year’s economic growth rate. Similarly, the larger this difference is, the greater the EGP is.

The control variables in model (1) include population (Razzaq et al., 2021; Zhang & Cheng, 2009), economic development (Huo et al., 2022; Acheampong, 2018; Razzaq et al., 2021), green technology innovation (Milindi & Inglesi-Lotz, 2022; Koseoglu et al., 2022; Khattak et al., 2022), and industrial structure (Zhao et al., 2022; Huo et al., 2022; Li et al., 2021) respectively indicated by the share of the secondary industry and the share of the tertiary industry. The significant effects of these above factors on CO₂ emissions have been confirmed in prior studies, as explained previously. The population is measured by the number of resident people; economic development is measured as GDP per capita; green technology innovation is measured as the number of green patents granted. The share of the secondary industry is the proportion of secondary sectors’ value added to GDP; the share of the tertiary industry is the ratio of the value added of the tertiary sectors to GDP.

We use Chinese city-level panel data for our empirical analysis. Compared to provincial panel data, city-level panel data have more observations and less variation, which effectively reduces the estimation bias of the model. Data on CO₂ emissions are derived from Chen et al. (2020) who calculated China’s county-level CO₂ emissions. We aggregate county-level data into city-level data. Data on EDT and annual growth rates are taken from the governments’ work reports at the beginning of the year. Data on GDP per capita, the share of the secondary industry, and the share of the tertiary industry are collected from China City Statistical Yearbook. Data on population are obtained from China’s provincial statistical yearbooks. We obtain the number of green patents granted from the patent database of the China National Intellectual Property Administration. We identify green patents according to the World Intellectual Property Organization in 2010 and summarize the number of green patents granted by the city.

As CO₂ emissions data are only available up to 2017, our panel dataset is also up to 2017. In addition, among the 297 prefecture-level and above cities in China, there are some cities, especially those located in central and western China and border areas, for which statistics are significantly missing. Therefore, considering the limited data availability, we collect panel data from 2005 to 2017 for 277 cities including four provincial-level municipalities, namely Beijing, Tianjin, Shanghai, and Chongqing. In addition, we eliminate the price effect based on constant 2005 prices. Table 1 reports the descriptive statistics for each variable. Among these, industrial structure upgrading, numbers of the patent granted, and trading openness are the mediating variables to be examined, as explained further below. Due to the significant differences in the unit magnitudes of the different variables, we take the natural logarithm of population, numbers of the green patents granted, GDP per capita, numbers of patents granted, and trade openness. Notably, the minimum number of green patent granted is 0, so for logarithmic purposes, all green patents granted values are added by 1.

Table 1. Descriptive statistics (observations: 3601).

Variable	Mean	Median	Max.	Min.	Std. Dev.
Carbon emissions per capita (ton)	6.865	5.725	55.218	0.471	5.629
Total carbon emissions (million tons)	26.679	19.984	230.712	1.723	23.706
Economic growth pressure	−0.004	−0.005	0.231	−0.127	0.024
Alternative economic growth pressure	−0.005	−0.005	0.207	−0.167	0.027
Population (10,000 persons)	441	362	3075	18	326
Numbers of green patent granted	200	32	13360	0	674
GDP per capita (¥)	30202	23546	198005	2730	22296
The share of the secondary industry	0.4859	0.4880	0.9097	0.0900	0.1101
The share of the tertiary industry	0.3753	0.3649	0.8060	0.0342	0.0912
Industrial structure upgrading	0.857	0.768	4.360	0.075	0.454
Numbers of patent granted	3020	511	106767	0	8228
Trade openness (¥)	9435	1793	276794	1	23944

Source: Authors own calculations.

Table 1 shows that CO₂ emissions per capita and EGP differ significantly across cities and years. The former varies from 0.471 to 55.218; the latter varies from −0.127 to 0.231. Other variables’ data also vary significantly, reflecting the huge development differences within China. This makes it necessary to explore the relationship between EGP and CO₂ emissions and the underlying influence mechanisms and also suggests the regional heterogeneity of this relationship.

4. Results

4.1. Empirical results

Table 2 reports the empirical findings according to model (1) using the classic ordinary least squares (OLS) estimation method. Column 1 shows the regression results. The results indicate that EGP positively affects CO₂ emissions per capita at the 1% significance level. Therefore, as EGP increases, CO₂ emissions per capita also increase. In other words, in the face of EGP, local authorities may sacrifice CO₂ emissions control. Columns (2) to (5) introduce in turn control variables: population, GDP per capita, the numbers of the green patents granted, the secondary industry, and the tertiary industry. Compared to column (1), the coefficient of EGP does not change significantly but remains significantly positive. This result highlights the contradiction between EGP and CO₂ emissions reduction at the national level. This suggests that, although China is performing increasingly diverse political assessments, local governments are still willing to promote economic development at the cost of the ecology to secure a political promotion. Therefore, local officials still widely accept the idea that increased GDP leads to promotion.

Table 2. The effects of economic growth pressure on carbon emissions.

Variables	Dependent variable: Carbon emissions per capita
	(1)	(2)	(3)	(4)	(5)	(6)
Pressure	3.764***	3.798***	5.162***	5.139***	4.798***	4.818***
	(1.146)	(1.146)	(1.192)	(1.192)	(1.192)	(1.192)
lnpopulation		−0.935**	−0.343	−0.288	−0.518	−0.591
		(0.433)	(0.456)	(0.461)	(0.463)	(0.466)
lnGDP_per_capita			0.808***	0.825***	1.320***	1.305***
			(0.200)	(0.201)	(0.230)0	(0.230)
Lngreen_patent				−0.047	−0.048	−0.044
				(0.056)	(0.056)	(0.056)
Secondary					−2.800***	−3.649***
					(0.642)	(0.888)
Tertiary						−1.537
						(1.111)
Constant	6.881***	12.369***	0.759	0.427	−1.845	−0.293
	(0.023)	(2.543)	(3.834)	(3.854)	(3.880)	(4.038)
City fixed effects	Yes	Yes	Yes	Yes	Yes	Yes
Time fixed effects	Yes	Yes	Yes	Yes	Yes	Yes
Obs.	3601	3601	3601	3601	3601	3601
R-squared	0.942	0.942	0.943	0.943	0.943	0.943

Robust Std. Errors are in brackets. *, **, and *** respectively denote the significance level of 10%, 5%, and 1%.

Source: Authors own calculations.

4.2. Robustness tests

To verify the sensitivity of the above findings to the variables, samples, and estimation methods, as well as account for possible endogeneity problems in the model (1), we perform the following robustness tests.

4.2.1. Alternative explanatory variables

As indicated above, the new indicator of EGP is measured as the difference between the current year’s EDT and last year’s economic growth rate. The new results are shown in Table 3. We found that the coefficients of the new EGP differ somewhat from the results in Table 2 but remain significantly positive. This suggests that changing the measure of the explanatory variable does not change the nexus of EGP and CO₂ emissions; therefore, the results in Table 2 are robust.

Table 3. Robustness check: Alternative economic growth pressure.

Variables	Dependent variable: Carbon emissions per capita
	(1)	(2)	(3)	(4)	(5)	(6)
Alternative pressure	3.831***	3.889***	5.040***	5.026***	4.692***	4.714***
	(1.055)	(1.055)	(1.090)	(1.090)	(1.090)	(1.090)
lnpopulation		−0.955**	−0.372	−0.315	−0.541	−0.615
		(0.433)	(0.456)	(0.460)	(0.462)	(0.465)
lnGDP_per_capita			0.802***	0.821***	1.310***	1.296***
			(0.198)	(0.199)	(0.229)	(0.229)
Lngreen_patent				−0.049	−0.049	−0.046
				(0.056)	(0.056)	(0.056)
Secondary					−2.774***	−3.630***
					(0.642)	(0.888)
Tertiary						−1.551
						(1.111)
Constant	6.886***	12.491***	0.986	0.638	−1.623	−0.056
	(0.024)	(2.543)	(3.814)	(3.835)	(3.861)	(4.020)
City fixed effects	Yes	Yes	Yes	Yes	Yes	Yes
Time fixed effects	Yes	Yes	Yes	Yes	Yes	Yes
Obs.	3601	3601	3601	3601	3601	3601
R-squared	0.942	0.943	0.943	0.943	0.943	0.944

Robust Std. Errors are in brackets. *, **, and *** respectively denote the significance level of 10%, 5%, and 1%.

Source: Authors own calculations.

4.2.2. Alternative explained variable and samples

We continue to use a different dependent variable to measure the influence of EGP on CO₂ emissions. We use total CO₂ emissions instead of CO₂ emissions per capita. In addition, we re-run the regressions with different samples to examine the sensitivity of the findings to the sample. First, we remove municipalities and only consider prefecture cities in the sample. Second, we shorten the overall sample time by keeping only the sample between 2005 and 2016. Column 1 in Table 4 shows that EGP still significantly positively affects CO₂ emissions. Therefore, changing the explained variable does not change the influence of EGP on CO₂ emissions. Columns 2 and 3 report that after removing municipalities and reducing the sample time, the impact of EGP on CO₂ emissions is still significantly positive.

Table 4. Robustness check: New explained variable and samples.

Variables	Dependent variable: Carbon emissions	Dependent variable: Carbon emissions per capita
	(1)	(2)	(3)
Pressure	4.439***	3.687***	4.834***
	(1.393)	(1.173)	(1.268)
lnpopulation	20.666***	−0.664	−0.256
	(1.326)	(0.462)	(0.499)
lnGDP_per_capita	4.317***	0.317***	1.837***
	(0.657)	(0.080)	(0.256)
Lngreen_patent	0.603***	−0.022	−0.097*
	(0.159)	(0.047)	(0.057)
Secondary	−7.874***	−2.398***	−6.838***
	(2.529)	(0.848)	(1.279)
Tertiary	7.573**	−1.466	−4.362***
	(3.165)	(1.117)	(1.376)
Constant	−139.219***	9.355***	−4.840
	(11.497)	(3.080)	(4.302)
City fixed effects	Yes	Yes	Yes
Time fixed effects	Yes	Yes	Yes
Obs.	3601	3549	3324
R-squared	0.974	0.943	0.944

Robust Std. Errors are in brackets. *, **, and *** respectively denote the significance level of 10%, 5%, and 1%.

Source: Authors own calculations.

4.2.3. Alternative estimation methods

If the relationship between EGP and CO₂ emissions changes significantly using different estimation methods, the results of Table 2 still need to be discussed more. So we examine the sensitivity of our results to the estimation method. To do so, we estimate model (1) using RLS and SGMM, respectively. RLS is used to overcome the possible influence of outliers on the regression results, while SGMM is used to avoid the bias caused by traditional OLS estimation and address the endogeneity of the model to some extent. The new estimation results are shown in columns (1) and (2) in Table 5, respectively. Table 5 also confirms the significant positive impact of EGP on CO₂ emissions.

Table 5. Robustness check: New estimation methods.

Variables	Dependent variable: Carbon emissions per capita
	(1)	(2)
Pressure	20.147***	2.539***
	(1.524)	(0.133)
lnpopulation	−0.800***	5.487***
	(0.084)	(0.217)
lnGDP_per_capita	2.190***	1.040***
	(0.113)	(0.026)
Lngreen_patent	−0.069	−0.032***
	(0.049)	(0.011)
Secondary	6.033***	−4.862***
	(0.602)	(0.223)
Tertiary	3.380***	−9.941***
	(0.722)	(0.233)
Constant	−15.370***
	(1.260)
Obs.	3601	3046
R-squared	0.302
AR(1) p-value		0.000
AR(2) p-value		0.666
Hansen test		1.000

Robust Std. Errors are in brackets. *, **, and *** respectively denote the significance level of 10%, 5%, and 1%.

Source: Authors own calculations.

4.2.4. Endogeneity test

Finally, we address the possible endogeneity problem of the model (1) using instrumental variable tests. Generally speaking, the GMM only addresses the endogeneity of the lagged explained variable, while explanatory variables’ endogeneity still needs to be considered separately. The endogeneity problem of the independent variable has two aspects. One is the possible omitted variables in the model (1). Although we have controlled the effect of other factors, we cannot exhaust all the influencing factors in model (1); thus, there is still the problem of possible omitted variables. In addition, model (1) may have a reverse causality, i.e., CO₂ emissions may conversely affect EGP, leading to a correlation between the two.

We perform the instrumental variable test to overcome the possible endogeneity problems. We select instrumental variables using the provincial EGP and the mean EGP of the provinces’ remaining cities. Since Beijing, Shanghai, Tianjin, and Chongqing are provincial regions, they are removed from the endogeneity test sample. The reason for this choice of instrumental variables lies in that local EDT is set to a certain extent by the guidance of higher-level governments. In this way, the provincial EGP is related to the city’s EGP, but at the same time, it does not directly affect the city’s CO₂ emissions. Besides, there is horizontal competition for economic performance among local governments, so there is likely to be horizontal strategic interaction in setting their EDT. When the EDT of other cities in the province increase, the local government also tends to set a relatively high growth target. Therefore, other cities’ EGP correlates with the city’s EGP but does not directly affect the city’s CO₂ emissions. Based on the above logical analysis, our choice of instrumental variables is reasonable.

We employ the TSLS technique for the instrumental variable test, and Table 6 reports the results. Columns 1-2 display the test results for the instrumental variable of provincial EGP, while columns 3-4 show the test results for the instrumental variable of the mean EGP of other cities. Columns 2 and 4 indicate the significant positive effects of provincial EGP and other cities’ EGP on the city’s EGP, thus confirming the above theoretical analyses. Columns 1 and 3 also show that EGP positively impacts CO₂ emissions. This again demonstrates the robustness of the results in Table 2.

Table 6. Robustness check: Instrumental variable test.

Variable	Carbon emissions per capita (second-stage)	Pressure (first-stage)	Carbon emissions per capita (second-stage)	Pressure (first-stage)
Pressure	36.574***		21.243***
	(4.117)		(2.911)
IV: provincial pressure		0.588***
		(0.031)
IV: other cities’s pressure				0.688***
				(0.026)
lnpopulation	−0.183	−0.015**	−0.407	−0.011*
	(0.518)	(0.006)	(0.479)	(0.006)
lnGDP_per_capita	0.472***	−0.003**	0.401***	−0.002**
	(0.091)	(0.001)	(0.084)	(0.001)
Lngreen_patent	0.012	−0.001	−0.004	0.001
	(0.052)	(0.001)	(0.049)	(0.001)
Secondary	0.183	−0.072***	−1.020	−0.067***
	(0.993)	(0.012)	(0.901)	(0.011)
Tertiary	−2.103*	−0.002	−1.806	0.003
	(1.247)	(0.016)	(1.156)	(0.015)
Constant	3.972	0.164***	6.481**	0.117***
	(3.490)	(0.043)	(3.213)	(0.041)
City fixed effects	Yes	Yes	Yes	Yes
Time fixed effects	Yes	Yes	Yes	Yes
Obs.	3549	3549	3549	3549
R-squared	0.930	0.396	0.939	0.445

Robust Std. Errors are in brackets. *, **, and *** respectively denote the significance level of 10%, 5%, and 1%.

4.3. Further analyses

We further analyze the mediating mechanism of EGP on CO₂ emissions. Theoretically, in EGPs, local governments will use various approaches to promote economic growth. For example, endogenous economic growth theory suggests that technological progress is the fundamental driver of economic growth. Therefore, theoretically, EGP has an urgent need for technological progress. Also, technological progress contributes to CO₂ emissions reduction (Leitão et al., 2022). So, technological innovation theoretically mediators the nexus of EGP and CO₂ emissions. Economic growth likewise means industrial structure adjustment. Local governments change the industrial structure by reducing low value added and developing high value added industries or improving industries that can significantly contribute to economic development (Zhao & Tang, 2018) to realize the EDT. In the meantime, industrial structure affects CO₂ emissions significantly (Zhao et al., 2022; Huo et al., 2022). In general, carbon-intensive sectors such as steel, cement, manufacturing, mining, energy, and construction belong to the secondary industry; thus, the proportion of the secondary industry largely determines the CO₂ emissions. On the contrary, the tertiary industry, except transportation, mostly has relatively low CO₂ emissions. Therefore, industrial structure upgrading theoretically constitutes the second potential mediating variable. In addition, we examine the possible mediating effects of trade openness. Expanding foreign trade also contributes to economic growth (Yanikkaya, 2003) and is confirmed to significantly influence CO₂ emissions (Li & Haneklaus, 2022; Udeagha & Ngepah, 2022). Therefore, in theory, trade openness also can mediate the influence of EGP on CO₂ emissions.

Consequently, we construct the following mediating effect models based on model (1) to test the above three potential mediating mechanisms. (3) $$\mathit{mediato}{r}_{i,t}=\alpha +{\beta }_2{p}_{i,t}+\phi \mathit{co}{\mathit{n}}_{i,t}+{\gamma }_i+{\lambda }_t+{\epsilon }_{i,t},$$(3) (4) $$\mathit{carbo}{n}_{i,t}=\alpha +{\beta }_3{p}_{i,t}+{\beta }_{4}m\mathit{ediato}{r}_{i,t}+\phi \mathit{co}{\mathit{n}}_{i,t}+{\gamma }_i+{\lambda }_t+{\epsilon }_{i,t},$$(4) where mediator denotes the mediating variables including technological innovation, industrial structure upgrading, and trade openness. Technological innovation is expressed as the total number of patent granted since patents broadly contribute to economic growth. Industrial structure upgrading is measured as the ratio of the secondary industry to the tertiary industry. Finally, trade openness is measured as total import and export per capita. Based on the idea proposed by Baron and Kenny (1986), we use the bootstrap method to estimate the mediating effects (Preacher & Hayes, 2008). Conventionally, the number of bootstrap samples for bias corrected bootstrap confidence intervals is set to 5000, and the level of confidence for all confidence intervals in output is set to 95%. Table 7 shows the results of the mediating effect test.

Table 7. Mediating effects on the nexus of economic growth pressure and carbon emissions.

Variable	Explained variable
	Carbon emissions per capita			Industrial structure upgrading			lnpatent			lntrade
	Coefficient	Confidence intervals		Coefficient	Confidence intervals		Coefficient	Confidence intervals		Coefficient	Confidence intervals
Constant	−1.056	−8.890	6.776	−0.936***	−1.584	−0.288	−0.059	−1.779	1.660	1.458	−1.044	3.961
Pressure	4.432***	2.219	6.645	−0.102***	−0.144	−0.060	−0.871**	−1.378	−0.363	−0.488***	−0.750	−0.226
Industrial structure upgrading	−0.412	−1.685	0.862
Lnpatent	−0.305***	−0.461	−0.150
lntrade	−0.306***	−0.413	−0.199
lnGDP_per_capita	1.852***	1.389	2.315	0.149***	0.112	0.186	0.415***	0.317	0.513	0.747***	0.604	0.890
Lngreen_patent	0.119	−0.010	0.250	−0.009**	−0.018	−0.000	0.461***	0.437	0.485	0.114***	0.080	0.149
lnpopulation	−0.569	−1.476	0.337	0.151***	0.076	0.225	−0.001	−0.199	0.197	−0.556***	−0.845	−0.267
Secondary	−6.070***	−8.092	−4.048	−2.526***	−2.669	−2.384	0.624***	0.245	1.002	1.981***	1.430	2.531
Tertiary	1.237	−1.040	3.514	1.753***	1.575	1.931	0.711***	0.237	1.184	1.054***	0.365	1.743
Time fixed effects	Yes			Yes			Yes			Yes
Region fixed effects	Yes			Yes			Yes			Yes
Observations	3601			3601			3601			3601
R-squared	0.945			0.941			0.974			0.951
Indirect effect of economic growth pressure on carbon emissions per capita	Effect	Boot SE	BootLLCI	BootULCI
Total	0.457	0.131	0.185	0.729
Industrial structure upgrading	0.042	0.053	−0.108	0.192
Lnpatent	0.266	0.112	0.126	0.406
lntrade	0.149	0.046	0.068	0.231

*, **, and *** respectively denote the significance level of 10%, 5%, and 1%.

The results show that the combined mediating effect of the three variables is 6.296 and is statistically significant because the confidence interval does not contain 0. The mediating effect of industrial structure upgrading is insignificant as the confidence interval includes 0. However, the mediating effects of technological innovation and trade openness are significant, with 0.266 and 0.149, respectively. Therefore, technological innovation and trade openness significantly mediate the impact of EGP on CO₂ emissions. EGP affects CO₂ emissions through the channels of technological innovation and trade openness. Notably, we found an interesting phenomenon: EGP increases CO₂ emissions by reducing technological innovation. This suggests that EGP does not have a positive effect on technological progress.

In Chinese practice, there is more reliance on investment, especially infrastructure investment and real estate development, to drive the economy (Su et al., 2022); in contrast, the technology dividend is not significant. This has increased CO₂ emissions. Under the EGP, most economic agents are more cautious, such as the typical practice of reducing R&D investment. This strategy will lead to a decline in regional innovation performance. Especially in China with a relatively weak market environment, regional economic growth rarely relies directly on technological progress, and economic agents lack the motivation to invest in R&D. Therefore, even if the economic growth target increases, technological innovation will not significantly improve, and the innovation investment will be compressed.

Similarly, local governments prefer to rely more on infrastructure investment than foreign trade for economic growth because it is easier for them to finance infrastructure investments through their controlled municipal investment companies than through a foreign exchange. In addition, local banks are more willing to provide such credit when backed by the governments. This explains, to some extent, the high growth of local government debt during the growing local economy in China, which is significantly higher than the growth of GDP and fiscal revenue during the same period (Pan et al., 2017). This phenomenon is especially prominent in most regions of China, where trade openness is weak. The over-reliance on infrastructure investment leads to a decrease in trade openness, while both the increase in infrastructure investment and the decline in trade openness contribute to increasing CO₂ emissions. Notably, Table 7 shows that technological innovation and trade openness significantly reduce CO₂ emissions, so it is essential to play the negative mediating role of technological innovation and trade openness in the relationship between EGP and CO₂ emissions. That is, local governments should rely more on technological innovation and increased foreign trade to alleviate EGP.

We also examine the regional variability of the effect of EGP on CO₂ emissions. We divide the Chinese sample into four regions: East, Northeast, Central, and West, and investigate whether the relationship between EGP and CO₂ emissions varies across regions. The theoretical and practical background for this is that, due to differences in regional socioeconomic characteristics, local officials may adopt different approaches to achieve economic growth goals. For example, in the eastern coastal regions of China, import and export trade plays an important role in economic growth and local governments may increase foreign trade to achieve economic growth goals, while in the central and western regions of China, they may rely more on resource industries and infrastructure investment to achieve economic growth. In conclusion, regional differences in physical and human geography provide different options for local officials to mitigate EGP, resulting in different CO₂ emissions. We divide the overall sample into 86 cities in Eastern China, 32 cities in Northeast China, 79 cities in Central China, and 80 cities in Western China by referring to the administrative divisions of China. Table 8 reports the results of the regional heterogeneity test. Columns (1) to (4) show the regression results for Eastern China, Northeast China, Middle China, and Western China, respectively.

Table 8. Regional heterogeneity of economic growth pressure’s effects on carbon emissions.

Variables	Dependent variable: Carbon emissions per capita
	(1)	(2)	(3)	(4)
Pressure	−2.423**	0.739	1.678**	12.537***
	(1.237)	(0.938)	(0.763)	(3.287)
lnpopulation	−4.067***	−7.467***	−3.269***	11.472***
	(0.370)	(1.246)	(0.216)	(1.418)
lnGDP_per_capita	0.924***	0.450	1.348***	0.523
	(0.171)	(0.300)	(0.165)	(0.621)
Lngreen_patent	0.070	−0.077	0.088***	−0.004
	(0.045)	(0.085)	(0.031)	(0.152)
Secondary	−2.430***	−2.785***	−2.723***	−4.233
	(0.746)	(0.976)	(0.434)	(2.796)
Tertiary	−0.472	3.922***	1.372**	0.845
	(0.730)	(1.255)	(0.695)	(3.728)
Constant	22.904***	45.226***	12.131***	−59.369***
	(3.312)	(7.862)	(2.505)	(11.667)
City fixed effects	Yes	Yes	Yes	Yes
Time fixed effects	Yes	Yes	Yes	Yes
Obs.	1118	416	1027	1040
R-squared	0.969	0.966	0.982	0.949

Robust Std. Errors are in brackets. *, **, and *** respectively denote the significance level of 10%, 5%, and 1%.

Column (1) shows that in Eastern China, EGP exerts significant adverse effects on CO₂ emissions. However, it has a significant positive impact on CO₂ emissions in Central and Western China. Also, EGP in Western China has the greatest effect on CO₂ emissions. EGP’s effect on CO₂ emissions in Northeast China is also positive but statistically insignificant. In sum, Table 8 confirms the regional heterogeneity of the nexus of EGP and CO₂ emissions. Eastern China is more socio-economically developed, with unique location and human resource advantages, and is significantly better than the central and western regions in terms of both foreign trade and technological innovation. In addition, it is more market-oriented and has a more dynamic economy, which is more inclined to alleviate the EGP through technological innovation and trade openness. Since both technological innovation and trade openness help reduce CO₂ emissions (see Table 7), EGP considerably decreases CO₂ emissions in Eastern China.

However, the practices of Eastern China cannot be replicated in Central and Western China and Northeast China because of the different resource endowments and economic development bases. The central and western regions depend more on relatively carbon-intensive secondary industries than the eastern region, especially some resource industries. Although these regions are also promoting low-carbon transition, when facing more severe EGP, they will likely return to their reliance on traditional industries and perform poorly in technological innovation, foreign trade, and industrial structure upgrading, further increasing CO₂ emissions. This problem is particularly prominent in the western region. Because the western region is traditionally economically backward, local officials are more concerned with the political performance caused by economic development. The more backward technological reserve and the weaker economic base, especially in the external economy, coupled with the fact that the western region is a resource and fossil energy-intensive region in China, will limit the government’s decision making behavior to the development of some high-emission industries such as resource and fossil energy under the EGP.

4.4. Discussion

The article investigates the influence of EGP on CO₂ emissions for the first time. Although both economic growth and environmental governance are key performance measures in the political promotion tournament theory, the article provides evidence of local officials’ preference for economic growth. Extensive prior studies have examined the relationship between economic development and CO₂ emissions and concluded that economic development significantly increases CO₂ emissions (e.g. Mujtaba et al., 2022; Jia et al., 2022; Huo et al., 2022). Our study similarly confirms the positive effect of EGP on CO₂ emissions. This reaffirms the contradiction between economic growth and environmental protection from the perspective of political promotion competition. Therefore, under the existing or ongoing political evaluation system, local officials prefer to sacrifice environmental quality to relieve pressure on economic growth. Given the central role of EGP in the race for political promotion, our findings confirm the more serious challenges of the low-carbon transition in China and more developing countries and raise new requirements for the assessment mechanisms of officials’ political promotion in the context of decarbonization.

The article verifies several mediating factors of the nexus of EGP and CO₂ emissions. Existing studies have largely ignored the influence mechanism of EGP. Our study bridges this gap by proposing a mediating mechanism framework. This framework identifies several local governments’ behaviors to increase CO₂ emissions under EGP scenarios, which provides new theoretical guidance for understanding local government decision-making behavior and changing the drivers of local economic growth. For example, technological innovation has begun to shrink under the EGP, which contradicts the endogenous growth theory’s emphasis on technology-driven economies. This finding is also similar to the role of innovation in the negative impact of EGP on energy efficiency, as explained by Zhu and Lin (2022). Our results suggest that to relieve the pressure on economic growth, local governments invest less in technological innovation and rely on more extensive but short-term ways of driving economic growth.

Zhu and Lin (2022) suggested that industrial structure upgrading hinders the effect of EGP on energy efficiency; however, our results do not support the transmission mechanism of industrial structure upgrading in the effect of EGP on CO₂ emissions. The main possible reason for this difference is that Zhu and Lin (2022) only theoretically analyzed the possible role of industrial structure upgrading but did not empirically demonstrate it. Our results confirm the negative effect of industrial upgrading on CO₂ emissions, but the result is not statistically significant (see Table 7). Moreover, energy efficiency and CO₂ emissions are fundamentally different. In addition, we find a significant mediating mechanism of foreign trade in the effect of EGP on CO₂ emissions. The contraction of the export-oriented economy under the EGP confirms that local governments will become more conservative to preserve growth. This also implies that the GDP-only governing philosophy is not conducive to the sustainable development of a low-carbon economy, and a radical change is needed.

We confirm the regional heterogeneity of EGP’s effects on CO₂ emissions. We found that the effects of EGP on CO₂ emissions are strongly related to each region’s economic structure and resource endowment. This finding is similar to the heterogeneous effect of EGP found by Wang et al. (2021) and Zhu and Lin (2022) but differs significantly from them in terms of specifics. Wang et al. (2021) suggested that the effect of EGP on innovation is more significant in regions with higher marketization; Zhu and Lin (2022) suggested that the negative effect of EGP on energy efficiency is more significant in regions with higher EGP, while EGP positively affects energy efficiency in regions with lower EGP. On the contrary, we comprehensively examine differences in socioeconomic development levels through spatial grouping to identify heterogeneity of EGP-influenced CO₂ emissions. For example, EGP significantly reduces CO₂ emissions in the eastern Chinese region. This means that officials in export-oriented and technologically innovative regions may have the dual profits of economic development and CO₂ emissions mitigation to support their more stable promotion. On the other hand, the positive influence of EGP on CO₂ mitigation is much larger than the national average in the central, especially in the western region. In sum, our results suggest that the political promotion of officials requires a differentiated assessment mechanism in the context of different socio-economic characteristics.

An additional important concern is the CO₂ emissions during the COVID-19 pandemic. Many studies have confirmed the negative effects of COVID-19 on economic development (e.g. Altig et al., 2020; McKibbin & Fernando, 2021) and CO₂ emissions (e.g. Le Quéré et al., 2020). Notably, the primary reason for CO₂ emissions reduction is the restriction of many production and consumption activities. Therefore, during the COVID-19 or post-COVID-19 period, the status of CO₂ emissions in government performance assessment should be reconsidered in more significant pressure for economic recovery and growth. Consequently, local governments theoretically may further weaken the position of CO₂ emissions reduction in order to restore economic growth, especially in the foreseeable short term. This poses a new challenge to the trade-off between economic growth and CO₂ emissions reduction.

5. Conclusions, policy implications and limitations

5.1. Conclusions

Since the introduction of CO₂ emissions reduction targets in 2009, the Chinese government has been under pressure to develop the economy and reduce CO₂ emissions simultaneously. Under the political promotion tournament, economic growth and CO₂ emissions reduction have also become important performance indicators for local governments. However, these two indicators are in most cases contradictory in a developing country like China. This study investigates the impact of EGP on CO₂ emissions, based on panel data for China’s 277 cities from 2005 to 2017. This study respectively uses the ratio of the current year’s economic growth target and the previous year’s economic growth rate and their difference to indicate EGP. The results demonstrate that EGP significantly increases CO₂ emissions, which is confirmed via various robustness tests. Our results also suggest that EGP increases CO₂ emissions by reducing technological innovation and trade openness, but our hypothetical mediating effects of industrial structure upgrading do not hold. Moreover, EGP significantly reduces regional CO₂ emissions in Eastern China; in contrast, it still exhibits a significant positive effect in Central and Western China, especially in the west.

5.2. Policy implications

Examining the impact of EGP on CO₂ emissions is not only conducive to better understanding government strategies and behaviors in the process of economic and low-carbon governance, but also to a better understanding of the role of regional characteristics in these strategies and behaviors, as well as to constructing a more scientific assessment mechanism for political promotion. Under the GDP-centric assessment mechanism, local governments will likely ignore the economic laws and eventually undermine the low-carbon transition strategy, once again transforming into extensive economic development. Hence, there is a need to establish a multidimensional evaluation mechanism for officials’ promotion. To achieve its "double carbon" goal, China’s CO₂ emissions reduction must be assessed at least on the same level as economic growth. The concept of low-carbon development must be implemented in the economic activities of all actors, especially in the decision-making behavior of the government. In addition, the assessment mechanism of officials should be differentiated in different regions. Especially in China’s central and western regions, the weight of economic growth should be weakened, and the coordinated development of the economy and environment should be examined more. The importance of CO₂ emissions reduction should also be highlighted in government work reports and various government performances.

Another practical implication is economic structural reform. The eastern coastal regions have always dominated China’s foreign trade economy. However, the central and western regions also have the advantage of developing an export-oriented economy, especially under the promotion of the "One Belt One Road" initiative. Our results show that the external economy contributes to reducing CO₂ emissions; therefore, the relatively low-carbon export-oriented economic growth should be strongly promoted in central-western areas. Also, in these regions, it is important to perform the economic function of technological progress, promote the industrial upgrading of resource and energy-based economies, and strengthen the technological development and application of resource and energy industries to facilitate low-carbon development. Besides, under the background of the current construction of the Chinese national unified market, it is also necessary to increase the technical and foreign trade cooperation between different areas, such as expanding the talent and technical support from eastern China to central-western China as well as establishing an integrated product and foreign trade supply chain between the east, central and west.

5.3. Limitations

In addition to the above findings, some limitations still exist that should be further enhanced. Due to limited data availability, our sample is only available up to 2017 and thus does not include the most recent years. A continuous investigation including the COVID-19 period is still necessary in the future. The article only examines the linear effect of EGP on CO₂ emissions, but the non-linear effect deserves further exploration. Besides, the article only examines three possible mediating mechanisms including industrial structure upgrading, technological innovation, and trade openness. Theoretically, there may be more complex mechanisms linking EGP and CO₂ emissions, such as the demographic characteristics of government officials and regional policies. Examining additional mediating mechanisms will yield richer theoretical and practical insights that deserve future attention.

Additional information

Funding

This work was supported by the Improvement Project of Young and Middle-aged Teachers' Research Ability in Guangxi's Colleges under Grant 2020KY22018; the National Natural Science Foundation of China under Grant 71764027.