The Impact of Multi-Dimensional Urbanization on CO₂ Emissions: Empirical Evidence from Jiangsu, China, at the County Level

Abstract

Understanding the underlying mechanism of how various dimensions of urbanization affect CO₂ emissions could be helpful for achieving the goal of low-carbon cities in China. However, previous research has seldomly examined this relationship granularly in economically developed regions at the micro level, nor did they examine the mediating effects of economic development, industrial structure, and coal consumption. Using the panel dataset of 80 counties from 2002 to 2021 at the county level in Jiangsu, China, this study investigates the direct and indirect effects of population, economic, and land urbanizations on CO₂ emissions in Jiangsu province and examines the regional heterogeneity. The findings indicate that population and economic urbanization have positive impacts on CO₂ emissions, whereas land urbanization has insignificant effects. This finding is supported by various robustness tests. Population and economic urbanizations are found to have significantly positive impacts on CO₂ emissions in the southern and northern Jiangsu regions, whereas none of the three dimensions are significant in the middle Jiangsu region. Economic urbanization contributes the most to CO₂ emissions in southern Jiangsu. In addition, our results indicate that multi-dimensional urbanizations affect CO₂ emissions through the mediating roles of economic development, industrial structure, and coal consumption. Our analysis shed some insights into the nuanced relationship between multi-dimensional urbanization and carbon emissions, which could contribute to sustainable urban transformation.

1. Introduction

Global warming since the mid-20th century has mainly been caused by the greenhouse gases that were emitted by human activities [1,2]. The primary component of greenhouse gases, carbon dioxide (CO₂), accounted for 77% of the greenhouse gas emissions in 2004 [3,4,5]. Among the global CO₂ emissions in 2021, approximately 27% came from China, which made it one of the largest polluters [6]. As a responsible nation to make a nationally determined contribution, China has committed to reaching a carbon peak by 2030 and carbon neutrality by 2060. Therefore, China is under pressure to fulfill its carbon reduction commitment by reducing CO₂ emissions.

Urbanization refers to the process of population moving from rural to urban areas and the associated economic and social changes [7,8]. Due to migration to cities, 75–80% of global anthropogenic greenhouse gas emissions are attributable to urbanization [4,9,10]. Therefore, urbanization is a major driver of CO₂ emissions and environmental degradation. To reduce the environmental burden, the Chinese government has proposed the concept of new-type urbanization in the National New-type Urbanization Plan (2014–2020). Subsequently, researchers have extended the concept of urbanization to include multiple dimensions such as population, environment, land urbanization, economics, and equity [11,12].

Numerous studies have examined the connection between population urbanization and CO₂ emissions [13,14]. One strand of the published literature has identified that population urbanization has a favorable effect on CO₂ emissions [15,16]. Another strand of the literature has found that population urbanization could reduce CO₂ emissions since public infrastructures are more effectively used [17,18,19]. However, some studies even revealed an inverted U-shaped curve between population urbanization and CO₂ emissions [20,21]. The variations in the findings may be due to variations in the data, geographic research areas, and research procedures in different contexts.

Recently, there has been a focus on how multi-dimensions of urbanization contribute to CO₂ emissions and some studies have found a direct link between population, land, and economic urbanizations and CO₂ emissions [14,18,22]. In certain situations, there are some differences in the magnitudes of population, land, and economic urbanizations on CO₂ emissions [11,22]. For example, population urbanization has a greater effect on CO₂ emissions in China [11], but economic urbanization has the highest influence on CO₂ emissions in the southern developed region of China [22]. The disparities could be caused by the size of the urban population, with economic urbanization having a major impact in less populated cities.

The Stochastic Impacts by Regression on Population, Affluence, and Technology (STIRPAT) model has been used in most studies delving into the connection between urbanization and CO₂ emissions [13,15,20,23]. Panel estimation approaches or time series approaches may be used to estimate this model, depending on the features of the dataset. If the dataset is stationary, standard panel regressions such as pooled ordinary least squares (POLS), fixed effects (FE), and random effects (RE) could be applied [13,20]. Otherwise, dynamic approaches such as the Vector Error Correction Model (VECM) and the Auto-Regressive Distributed Lag (ARDL) model could be used to estimate [15,23]. The spatial models are also used in several investigations [14,24].

Jiangsu province is located in eastern China with a gross domestic product (GDP) of $1.6 billion in 2021, ranking second in China [25]. The percentage of people living the cities has grew from 37.4% in 2002 to 73.9% in 2021, with total CO₂ emissions has increased from 276 million metric tons in 2002 to 823 million metric tons in 2021 [26]. Jiangsu province’s 14th Five-Year Plan (2021–2025) for Ecological Environment Protection has specified that CO₂ emissions per unit of industrial value added should decrease by 20% by 2025, and Jiangsu province should be built into a pioneer for the construction of beautiful China by 2035 [27]. Jiangsu province is striving to play the leading role in promoting high-quality economic development and top-notch ecological and environmental conservation. Therefore, CO₂ emission reductions should be one of the development goals in Jiangsu province.

Prior research has examined the connection between urbanizations and CO₂ emissions at the national, provincial, and city levels [11,13,14,15,28,29,30]. However, none of the studies examines the relationship at the county level, which is a crucial administrative division in China when it comes to the implementation of policies. Furthermore, the indirect link between urbanization and CO₂ emissions in China has rarely been investigated. Therefore, this study uses the STIRPAT model to investigate the effects of multi-dimensional urbanizations on CO₂ emissions in Jiangsu province at the county level from 2002 to 2021. Considering the large regional variations within the province, the investigation in different regions could also provide some valuable insights. Furthermore, the mechanism by which urbanization affects CO₂ emissions is examined. We would expect to provide some references for understanding the emission mechanism and provide feasible strategies for legislators.

2. Literature Review

2.1. Impact of Urbanization on CO₂ Emissions

The effects between urbanization and CO₂ emissions across countries and studies are presented in

Table 1. Published studies on the impact of urbanization on CO₂ emissions.

Effect	Region (Reference)
Positive	Developing countries (Parikh and Shukla [33]); 137 countries (York et al. [41]); 86 countries (Cole and Neumayer [42]); European members (Liddle and Lung [34]); 99 nations (Poumanyvong and Kaneko [13]); China (Sheng and Guo [15]; Liu et al. [35]; Sun [36]; Wang [37]; Yang et al. [43]); 14 MENA countries (Al-Mulali and Ozturk [38]); Africa (Hussain et al. [16]); India (Franco et al. [39]); Turkey (Ozatac et al. [44]); G7 countries (Ahmed et al. [19]); Pakistan (Sufyanullah et al. [40]); 23 developing countries (Sikder et al. [28])
Negative	China (Zhao and Chen [45]); Pakistan (Danish et al. [46]); MENA region (Charfeddine and Mrabet [32]; Tawfeeq [24])
Mixed	Various countries (Fan et al. [47]); seven regions (Al-mulali et al. [48]); 73 countries (Li and Lin [49]); China (Xu and Lin [29]; Wang et al. [50]; SSEA countries (Behera and Dash [51]); Developing countries (Martínez-Zarzoso and Maruotti [52]); Indonesia (Ahmed et al. [19])
Inverted U-shaped curve	China (He et al. [20]); Malaysia (Bekhet and Othman [53]); 144 countries (Zhang et al. [54])
insignificant	China (Ji et al. [55]); 16 merging countries (Sadorsky [56]); Qatar (Charfeddine, [31])

, which exhibits a mixed effect, inverted U-shaped curve effect, negative effect, positive effect, and insignificant effect [13,19,20,21,28,31,32]. First, it is discovered that urbanization increases CO₂ emissions in both developing countries and developed countries (Table 1). Parikh and Shukla [33] found a positive relationship between urbanization and greenhouse gas emissions by arguing that the movement of people from rural to urban regions raises energy consumption demand, which in turn raises CO₂ emissions. Particularly, Liddle and Lung [34] find that urbanization, a measure of access to the power grid in developed countries, has a large positive impact on energy consumption and carbon emissions in the residential sector. Numerous extant studies have examined the effects of urbanization on carbon emissions in China [15,35,36,37]. Some studies identified a positive relationship, while others detected a bidirectional causal association between urbanization and CO₂ emissions. Furthermore, Sheng and Guo [15] discover a positive relationship both in the short run and in the long run. Similar findings have been reported by other studies conducted in 14 countries of the Middle East and North Africa, Africa, India, Turkey, and other developing countries [16,19,28,38,39,40].

Sometimes, the correlation between urbanization and carbon dioxide emissions could be negative. This finding is supported by Goodland and Anhang [57] by arguing that urbanization resulted in a significant decrease in the number of cows with the highest carbon emissions, contributing at least 51% of all livestock emissions worldwide. In addition, this negative relationship could be in the short term or the long term [45,46]. Charfeddine and Mrabet [32] find that urbanization benefits the environment in the Middle East and North African (MENA) countries by arguing that urbanization fosters the growth of tertiary industry and the innovation of green technology. Similar results are found in the United States [24]. Depending on the stage of economic development and income level, the impact of urbanization on carbon dioxide emissions may also be complex [47,48,49,50,51,52].

The Kuznets curve theory has also been extended to urbanization [58,59]. The environmental Kuznets curve (EKC) theory has been used in certain research to show the inverted U-shaped relationship between urbanization and CO₂ emissions. He et al. [20] divided 29 Chinese provinces into three zones and discovered an inverted U-shaped association between urbanization and CO₂ emissions in relatively economically developed and less developed regions. In Malaysia, urbanization and CO₂ emissions show a nonlinear inverted U-shaped relationship, with a positive short-term impact and a negative association above the threshold level [53]. Similarly, Zhang et al. [54] establish this link between urbanization and CO₂ emissions, with 73.8% as the turning point, using the panel data of 141 countries. Lastly, a few studies have discovered a negligible correlation between urbanization and CO₂ emissions [55,56].

2.2. Impact of Multi-Dimensional Urbanization on CO₂ Emissions

The literature on the effects of multiple dimensions of urbanization on CO₂ emissions is presented in

Table 2. Published studies on the impact of multi-dimensions of urbanization on CO₂ emissions.

Reference	Region	Urbanization Dimension	Findings
Xu et al. [22]	Pearl River Delta, China	Population; economic; land	Minimal;
Wang et al. [12]	China	Population; environment; space; economy; equity	Paradoxical effects
Zhou et al. [30]	Yangtze River Delta, China	Economic; demographic; spatial; social	Positive; two opposing effects; positive; negative
Chen et al. [11]	China	Population; economic; land	Direct: all positive Indirect: negative; positive; negative
Yao et al. [14]	351 Prefecture-level cities in China	Population; economic; land	Positive; positive in small, medium, and large cities, but negative in megacities; negative in small cities and megacities, but positive in medium and large cities

. The dimensions of urbanization could be different depending on the context and research designs. Xu et al. [22] found that population urbanization has the least effect on carbon emissions, land urbanization has the second-greatest impact, and economic urbanization has the largest. Wang et al. [12] find paradoxical effects between population, environment, space, economy, and equity urbanization and CO₂ emissions in China. Zhou et al. [30] discovered that while economic and spatial urbanizations have positive effects on CO₂ emissions in the Yangtze River Delta, China, demographic urbanization has an opposing effect, and social urbanization has a negative effect. Furthermore, the relations between economic urbanization and land urbanization and CO₂ emissions are inversely U-shaped. Similar studies conducted throughout all of China reveal that three aspects of urbanization have a favorable impact on CO₂ emissions [11,14]. On the other hand, Chen et al. [11] discovered that while economic urbanization has a favorable impact on CO₂ emissions, population and land urbanizations have negative consequences. Yao et al. [14] categorize cities into four groups—small-, medium-, and large-size cities, and megacities—based on the official Chinese city classification and discover different outcomes in each of these cities. Based on the above-mentioned analysis, we propose the following hypotheses:

H1. 

The population urbanization positively affects CO₂ emissions in Jiangsu province.

H2. 

The economic urbanization positively affects CO₂ emissions in Jiangsu province.

H3. 

The land urbanization positively affects CO₂ emissions in Jiangsu province.

2.3. Indirect Impact of Economic Development, Industrial Structure, and Coal Consumption on CO₂ Emissions

Direct causal effects, mediated causal effects, and moderated causal effects are the three commonly assumed causal hypotheses [60,61]. The direct causal effect suggests that the independent variables cause the dependent variable, which is straightforward [62]. The mediating effect suggests that the independent variables have a causal effect on the mediator, and subsequently, the mediator has a causal effect on the dependent variable [63]. Therefore, an indirect effect between the independent and dependent variables is identified through the mediator.

The Authur Lewis’s theory posits that the population migration from the rural to the urban has moved the labor surplus from the traditional economy with low productivity to the modern economy with high productivity [64,65,66]. As a result, economic output and employment in the modern economy sectors are created, and per capita GDP is increased. The environmental Kuznets curve suggests an inverted U relationship between economic development and environmental degradation [67,68]. In the early stages of development, the CO₂ emissions caused by intensive economic activities outweigh the CO₂ emission reduction due to technological improvement, thus leading to increased CO₂ emissions. In the latter stage of development, the carbon reduction due to technological innovations exceeds the emitted CO₂, leading to decreased CO₂ emissions.

The industrial structure is commonly expressed by the proportion of first, second, and tertiary industries’ outputs. The rapid development of urbanization promotes the upgrade of industrial structures by affecting the supply, demand, and allocation of resources at the regional level [69]. In addition, urbanization promotes the transformation from the first industry to the second and tertiary industries, which ultimately facilitates industrial structure upgrading [70,71]. The association between urbanization and industrialization results in increased demand for energy consumption, thus leading to more CO₂ emissions [69,72].

The urbanization process is usually accompanied by energy consumption since individuals in the city are prone to consume more products than those in the rural area [73]. It was reported that one-third of global energy is consumed by residential sectors [74]. In addition, urbanization promotes the production structure change from the agricultural sector to the industrial and tertiary sectors, thus leading to an increase in energy consumption [7]. Coal is the most carbon-intensive energy since coal combustion releases 1.08 tons of carbon for each ton of oil equivalent, natural gas discharges 0.64 tons, and oil emits 0.84 tons [75]. The reserve-to-production ratio for coal is 162 years, while that for oil is 40.6 years and for gas is 65.1 years. The large reserve volume of coal and relatively cheap price have given coal a competitive advantage [75]. Therefore, urbanization leads to more CO₂ emissions from coal consumption.

Hence, we proposed our hypotheses below:

H4. 

Multi-dimensional urbanizations have an impact on CO₂ emissions through the mediating role of economic development.

H5. 

Multi-dimensional urbanizations have an impact on CO₂ emissions through the mediating role of industrial structure.

H6. 

Multi-dimensional urbanizations have an impact on CO₂ emissions through the mediating role of coal consumption.

Based on the reviewed literature above, several points regarding how urbanization affects CO₂ emissions could be drawn. Firstly, the concept of urbanization has been expanded to encompass population, economic, land, and social viewpoints. Secondly, there is conflicting evidence in the available empirical literature regarding the relationship between urbanization and CO₂ emissions. Third, the effect of urbanization on CO₂ emissions is contingent upon various factors, including the degree of economic development (income levels), economic structure, energy structure, and the distinction between rural and urban areas. Fourth, the mediating roles of economic development, industrial structure, and coal consumption in the relationship between urbanization and CO₂ emissions have seldom been studied. Therefore, we proposed our conceptual framework in Figure 1 below.

3. Methodology and Data

3.1. Basic Model

In order to measure how much human activity affects the environment, Ehrlich and Holdren [76] established the IPAT model ($I=PAT$), wherein ($I$) denotes the environmental impact that is dictated by population size ($P$), affluence ($A$), and technology ($T$). However, the IPAT model’s fundamental premise, that the elasticities of population, affluence, and technology should add up to one, is too stringent to be fulfilled [77]. Thus, the IPAT model is developed into the STIRPAT model as follows:

$$I_{it}=\alpha \times P_{it}^{\theta }\times A_{it}^{\beta }\times T_{it}^{\delta }\times e_{it}$$

(1)

where $i$ stands for the analysis object, $t$ represents the time period, $\alpha$ is the equation’s constant term, $\theta$, $\beta$, and $\delta$ stands for elasticities of population, affluence, and technology, $e$ represents the error term. Considering 74.4% of the global greenhouse emissions from anthropogenic activities come from CO₂ emissions [5], the environmental impacts of $I_{it}$ could be replaced with CO₂ emissions (${CE}_{it}$) in this research. By taking the logarithmic transformation of Equation (1), the STIRPAT model has the expression below in Equation (2).

$${lnCE}_{it}=ln\alpha +\theta ln{P}_{it}+\beta {lnA}_{it}+\delta {lnT}_{it}+ln{e}_{it}$$

(2)

Therefore, we apply the STIRPAT model in Equation (2) in this study and add population urbanization ($PU$), economic urbanization ($EU$), and land urbanization ($LU$) as additional variables. $P_{it}$ is proxied with a total population in county $i$ in time $t$, whereas $A_{it}$ is proxied with the per capita GDP. $T_{it}$ is proxied with industry structure, which is the proportion of industrial added value by GDP. In addition, coal consumption is also included in Equation (3). There exists a strong collinearity between coal consumption and population, so we only include coal consumption in the model. Therefore, the STRIPAT model specification takes the form of Equation (3) below:

$${lnCE}_{it}=ln\alpha +{\alpha }_{1}lnP{U}_{it}+{\alpha }_{2}lnE{U}_{it}+{\alpha }_{3}lnL{U}_{it}+\beta {lnGDP}_{it}+\delta {lnInd}_{it}+\gamma lnCoa{l}_{it}+{ϵ}_{it}$$

(3)

When the panel dataset is stationary, standard panel data estimates such as POLS, FE, and RE could be performed. Otherwise, the panel dataset should be analyzed by using the panel ARDL model or the VECM model should be applied. We have tested the stationarity of the panel dataset, and the results suggest a stationary panel dataset. Therefore, Equation (3) is estimated using POLS, FE, and RE in the following analysis.

3.2. Mediating Effect Model

Baron and Kenny [78] first proposed the three-path mediating model, but various tests must be conducted to validate the effectiveness of mediating variables and endogeneity. Later, Kenny [79] proposed that the third step could be eliminated by theoretical explanations to avoid the problem of endogeneity. Based on the two-path mediating effect model, our mediating effect model is developed to test Hypotheses 4, 5, and 6. The first path of the model is the same as Equation (3). The second path of the model is specified as follows:

$${lnM}_{it}={\eta }_{1}lnP{U}_{it}+{\eta }_{2}lnE{U}_{it}+{\eta }_{3}lnL{U}_{it}+{\eta }_{4}lnGDP+{\eta }_{5}lnIn{d}_{it}+{\eta }_{6}lnCoa{l}_{it}+{ϵ}_{it}$$

(4)

where $M_{it}$ represents the mediator variables that include economic development, industrial structure, and coal consumption. When one of the mediator variables is the dependent variable in Equation (4), the other two mediators turn into control variables.

3.3. Data

There are 92 counties in Jiangsu province in total. Due to the changes in administrative divisions across years and the availability of data, 80 counties were selected. To conduct a comparative analysis, we divided the whole Jiangsu province into the southern region (Suzhou, Nanjing, Wuxi, Changzhou, Zhenjiang), middle region (Nantong, Yangzhou, Taizhou), and northern region (Suqian, Huaian, Liangyungang, Yancheng, Xuzhou) based on the administrative division. The distribution of the three regions is presented in Figure 2 below.

Since data before 2002 were unavailable, this article compiles county-level data from 2002 to 2021 in Jiangsu province, China. The CO₂ emission database is retrieved from the Emissions Database for Global Atmospheric Research (EDGAR) as gridmaps at 0.1 × 0.1 degree resolution at the global level and then processed by ArcGIS to obtain the CO₂ emissions at the county level. The dataset for other variables except coal consumption comes from the Statistical Yearbooks of Municipalities. The dataset for coal consumption at the municipal level comes from the Statistical Yearbook of Jiangsu Province and is then calculated based on the proportion of GDP in each county. The variable descriptions, units, and measurements are listed in

Table 3. Variables, definitions, units, and measurements (2002–2021).

Variable	Description	Unit	Measurement
${CE}_{it}$	Annual CO2 emission	Million tons	Total CO2 emissions
${PU}_{it}$	Population urbanization	%	Urban population/total population
${EU}_{it}$	Economic urbanization	%	Secondary and tertiary output/GDP
${LU}_{it}$	Land urbanization	%	Urban land area/total land area
${GDP}_{it}$	Per capita GDP	Yuan/person	Total GDP/total population
${Ind}_{it}$	Industrial structure	%	Industrial added value/GDP
${Coal}_{it}$	Coal consumption	10,000 tce	Total coal consumption

Note: tce represents tons coal equivalent.

.

Population urbanization: The urbanization of the population is caused by the household registration policy in China, with rural households mainly working on agricultural productions and urban households on industrial and tertiary industries. With the implementation of the opening and reform policy in 1978, rural households could migrate freely to the city to earn higher incomes by working in the industrial or tertiary sectors. Urbanization of population is proxied by the percentage of the population living in urban areas relative to the overall population in this paper.

Economic urbanization: According to Chen et al. [11], economic urbanization is the fundamental process of modernizing the industrial and non-agricultural economic structures. Zhou et al. [30] use the proportion of secondary industry output in GDP as the indicator for economic urbanization, Chen et al. [11] express it by using the proportion of tertiary industry, and Yao et al. [14] choose the proportion of secondary and tertiary industries. In this paper, the percentage of secondary and tertiary output in GDP serves as the proxy for economic urbanization.

Land urbanization: The transfer of agricultural land uses to non-agricultural uses such as industrial, commercial, residential, and infrastructure leads to land urbanization. The expansion of land area has been suggested to be an important factor in terrestrial CO₂ emissions [7]. This article uses the ratio of urban land area to the total land area to indicate land urbanization.

Other variables include per capita GDP, industrial structure, and coal consumption. Per capita GDP is a measure of economic development that is computed by dividing GDP by population. The percentage of industrial added value in GDP is known as industry structure. Due to its growing energy consumption, China’s industrial sector now accounts for the majority of its CO₂ emissions. Thus, the transformation of industrial structure from high- to low-carbon industries could contribute to CO₂ emissions reduction in China. The amount of coal consumption is selected because, according to the Jiangsu Statistical Yearbook (2022), coal made up approximately 76% of the province’s energy consumption in 2020.

The summary statistics of the variables used in this research are presented in

Table 4. Summary statistics (2002–2021).

Variable	Obs.	Mean	Standard Deviation	Minimum	Maximum
${CE}_{it}$	1600	7.6221	31.7137	0.3055	1205.99
${PU}_{it}$	1600	0.5235	0.2447	0.022	1.000
${EU}_{it}$	1600	0.8911	0.0966	0.4241	1.000
${LU}_{it}$	1600	0.5188	0.2280	0.0105	1.000
${GDP}_{it}$	1600	62,500.56	4997.37	2289.218	257,033.1
${Ind}_{it}$	1600	0.3958	0.1294	0.00154	1.000
${Coal}_{it}$	1600	301.7173	396.0197	0.3490	4953.447

. There are 80 counties for 20 years, with 1600 observations in total. The annual CO₂ emissions ranged from 0.3055 million tons to 1205.99 million tons, which suggests great variations within the study area during the period.

The annual CO₂ emissions and three dimensions of urbanization in Jiangsu province are presented in Figure 3. As illustrated in Figure 3, the CO₂ emissions in Jiangsu province have increased from 276 million metric tons in 2002 to 823 million metric tons in 2021. The population urbanization has increased from 0.37 in 2002 to 0.81 in 2021; economic urbanization increased from 0.79 to 0.92; land urbanization increased from 0.55 to 0.75. It seems obvious that rapid CO₂ emissions are accompanied by an increase in population, economic, and land urbanizations.

The distributions of CO₂ emissions in Jiangsu province at the county level are presented in Figure 4 with (4a), (4b), (4c), (4d) denoting years of 2005, 2010, 2015, 2020. It could be seen that the CO₂ emissions in Jiangsu province have continuously increased throughout the years. Furthermore, counties with greater CO₂ emissions are mainly located in the southern and northern Jiangsu regions.

4. Results and Discussion

The estimation results, robustness test, heterogeneity analysis, and mechanism analysis are presented in this section. To compare different models, various tests are conducted. The Breusch–Pagan (BP) test is developed to test for heteroskedasticity with the null hypothesis of homoskedasticity and a normally distributed error term [80]. The Durbin–Wu–Hausman test is a statistical test to evaluate the consistency of an estimator [81]. The null hypothesis of the Hausman test is that the coefficients of fixed effects and random effects models are consistent but random effects models are more efficient, whereas the alternative is that fixed effects models are consistent but random effects estimates are not [81].

4.1. Relationship between Urbanization and CO₂ Emissions

The empirical estimation results with POLS, FE, RE are shown in

Table 5. Estimation results with POLS, FE, and RE for Jiangsu province.

Variable	POLS	FE	RE
Constant	12.7662 *** (0.3635)	12.5919 *** (0.1088)	12.5897 *** (0.1575)
${lnPU}_{it}$	0.1793 *** (0.0397)	0.0276 ** (0.0127)	0.0287 ** (0.0127)
${lnEU}_{it}$	−0.0329 (0.2074)	0.1823 *** (0.0646)	0.1841 *** (0.0646)
${lnLU}_{it}$	0.2253 *** (0.0533)	0.0277 (0.0258)	0.03119 (0.0257)
${lnGDP}_{it}$	0.0139 (0.0427)	0.1685 *** (0.0157)	0.1660 *** (0.0157)
${lnInd}_{it}$	0.1820 *** (0.0448)	0.0641 *** (0.0202)	0.0668 *** (0.0202)
${lnCoal}_{it}$	0.4883 *** (0.0436)	0.1421 *** (0.0239)	0.1487 *** (0.0238)
F-statistic	261.22 ***
Breusch–Pagan	12,572 ***
Hausman	5.82

Note: *** p < 0.01; ** p < 0.05.

. The results of the F-statistic in Table 5 reject the null hypothesis and support the conclusion that the FE estimate is better. The BP test also rejects the null hypothesis and suggests that the RE estimate is better. The Hausman test failed to reject the null hypothesis, so the RE estimate is a better choice. Therefore, the results from the RE estimate are used for the following analysis. Results in Table 5 show that the estimated coefficients of population urbanization, economic urbanization, per capita GDP, industrial structure, and coal consumption are statistically significant at least at the 5% significance level, but not for land urbanization. These results support Hypotheses 1 and 3 but reject Hypothesis 2. All significant variables have positive effects on CO₂ emissions, with economic urbanization, per capita GDP, and coal consumption having the greatest impacts. With all other factors holding constant, the elasticity estimate of CO₂ emissions to population urbanization is 0.0287, meaning that for each 1% increase in population urbanization, CO₂ emissions grow by 0.0287%. The elasticity of economic urbanization is 0.1841, suggesting a rise in CO₂ emissions of 0.1841% for each 1% increase in economic urbanization. The elasticity to per capita GDP, industrial structure, and coal consumption is 0.1660, 0.0668, and 0.1487, respectively, indicating the CO₂ emissions increase by 0.166%, 0.0668%, and 0.1487% for each 1% increase in these variables.

To comprehend the environmental effects of urbanization, it is crucial to examine how various aspects of urbanization affect CO₂ emissions in Jiangsu province at the county level. The empirical findings are consistent with the studies by Zhou et al. [30] for the Yangtze River Delta, China, Chen et al. [11] for China, and Yao et al. [14] for 351 prefecture-level cities in China, indicating that CO₂ emissions in Jiangsu province are positively impacted by population urbanization. Population migration to cities increases urban population and carbon emissions, according to the positive elasticity estimate of population urbanization. Due to the increased demand for infrastructure, transportation, and personal resource consumption brought about by the population migration from distant rural to urban regions, city dwellers consume more energy than their rural counterparts [13,57].

Along with the population urbanization process, the total CO₂ emissions in Jiangsu province have increased nearly threefold from 2002 to 2021. The 14th Five-Year Plan of New-Type Urbanization Development in Jiangsu Province (2021–2025) specified that the population urbanization rate should reach 75% by 2025, thus, CO₂ emissions that were caused by population urbanization are expected to remain stable in the future. Therefore, Jiangsu province should advocate policies to improve the quality of urbanization to construct low-carbon cities by advocating sustainable consumption and conducting ecological civilization education.

However, this finding is contrary to some studies on how China’s urbanization affects CO₂ emissions [20,29,45]. Zhao and Chen [45] explain that urbanization largely decreases the usage of livestock, which accounts for 51% of global emissions, and this effect could mitigate environmental pollution in the short run. In contrast to our research, He et al. [20] discovered that in the majority of China’s areas, the elasticity of emission urbanization is positive for low urbanization and negative for high urbanization, revealing an inverted U relationship. They contend that China has surpassed the tipping point and that rising urbanization rates result in lower CO₂ emissions.

Economic urbanization is the main cause of CO₂ emissions in Jiangsu. According to this finding, Jiangsu may see an increase in carbon emissions as a result of increased urbanization, which may also raise the proportion of secondary and tertiary industries in the state’s economy. This result is consistent with that of Yao et al. [14], but it differs from Chen et al. [11], who found that population urbanization has the biggest influence. According to Yao et al. [14], economic activity has a greater impact on CO₂ emissions than land urbanization and population urbanization. Chen et al. [11] contend that economic urbanization has greater real effects on society despite its negligible economic impact, and they recommend that governments give both economic development and carbon emission reduction extra attention. In order to create a low-carbon economy, governments should thus give high-quality economic development more consideration and offer financial and technological support.

Similar to earlier research, the percentage of secondary and tertiary production is also found to be strongly correlated with CO₂ emissions in Jiangsu province [11,14,30]. The growth of tertiary and secondary sectors results in increased energy consumption and carbon emissions. In contrast to previous research, this study found no evidence of a meaningful correlation between Jiangsu province’s CO₂ emissions and the percentage of metropolitan regions. Jiangsu province’s land urbanization rate rose sharply from 55% in 2002 to 75% in 2007 and has stayed steady ever since. Consequently, it is plausible that there was a significant erosion of agricultural land in the vicinity of 2007; these fields were set aside for commercial, industrial, and residential purposes.

The correlation between CO₂ emissions and per capita GDP is positive, indicating that Jiangsu province’s economic expansion is coming at the expense of environmental deterioration. While some earlier research has found an inverted U-shaped relationship between economic development and CO₂ emissions, other studies have found a positive relationship [82,83]. The Kuznets curve explains how the degree of economic development affects the consequences of environmental problems. The CO₂ emissions in Figure 1 show an increasing trend, supporting the theory that Jiangsu province’s economic development will rise gradually. Since the tipping point has not been reached, policies and regulations for reducing carbon emissions should be made.

Jiangsu province’s industrial structure has a positive effect on CO₂ emissions, indicating that the region’s energy-intensive sectors are the source of carbon emissions. This result is consistent with other research showing that the industrial structure has a significant influence on both energy use and greenhouse gas emissions [30,40,56]. They clarify that carbon emissions in cities could not be offset by increased energy efficiency or emission reduction technologies in China. Governments should therefore establish policies to replace traditional energy with renewable energy sources like nuclear, solar, and wind energy, in addition to increasing energy consumption efficiency.

4.2. Robustness Test

The robustness of the basic model is tested in five ways below. First, all variables are winsorized at the 1% and 99% quantiles to remove the outliers and extreme values. Second, the control variables (CV), including per capita GDP, industrial structure, and coal consumption, lagged with one period. Then, population urbanization, economic urbanization, and land urbanization are treated with one-period lags individually. The results of robustness tests are presented in

Table 6. Robustness tests.

Variable	Winsorizing	Lagged CV	Lagged PU	Lagged EU	Lagged LU
Constant	13.2527 *** (0.4441)	13.5575 *** (0.4428)	13.1375 *** (0.4771)	12.9746 *** (0.4818)	13.1851 *** (0.4762)
${lnPU}_{it}$	0.3591 *** (0.0637)	0.3271 *** (0.0622)	0.2936 *** (0.0626)	0.3424 *** (0.0624)	0.3344 *** (0.0627)
${lnEU}_{it}$	−0.1726 (0.3799)	0.0208 (0.3769)	−0.0689 (0.3915)	−0.3348 (0.3536)	−0.0777 (0.3898)
${lnLU}_{it}$	0.1818 *** (0.0593)	0.1833 *** (0.0577)	0.1724 *** (0.00584)	0.1818 *** (0.0572)	0.1584 *** (0.0602)
${lnGDP}_{it}$	−0.0441 (0.0484)	−0.0630 (0.0488)	−0.0212 (0.0504)	−0.0068 (0.0502)	−0.0270 (0.0504)
${lnInd}_{it}$	0.2008 *** (0.0538)	0.1690 *** (0.0491)	0.1689 *** (0.0488)	0.1788 *** (0.0481)	0.1724 *** (0.0487)
${lnCoal}_{it}$	0.5341 *** (0.0461)	0.5156 *** (0.0452)	0.4991 *** (0.0455)	0.5026 *** (0.0454)	0.5039 *** (0.0453)

Note: *** p < 0.01.

below. The results indicate that the positive effects of population urbanization and land urbanization still hold, which indicates that the results from the basic model are robust. The robustness tests also show that per capita GDP is not significant, but the significance of industrial structure and coal consumption is consistent with these in the basic model, which also supports the robustness of the basic model estimation.

4.3. Heterogeneity Analysis

The heterogeneity results for Jiangsu province are presented in

Table 7. Heterogeneity estimation results for CO₂ emissions in Jiangsu province.

Variable	Southern Jiangsu	Middle Jiangsu	Northern Jiangsu
Constant	12.3343 *** (0.3337)	12.2471 *** (0.2415)	13.0037 *** (0.2285)
${lnPU}_{it}$	0.0736 ** (0.0352)	0.0131 (0.0165)	0.0350 ** (0.0161)
${lnEU}_{it}$	0.5661 ** (0.2448)	0.2581 (0.2265)	0.1453 ** (0.0725)
${lnLU}_{it}$	0.1344 (0.0603)	0.0026 (0.0219)	−0.0479 (0.0359)
${lnGDP}_{it}$	0.2752 *** (0.0388)	0.1393 *** (0.0246)	0.1053 *** (0.0226)
${lnInd}_{it}$	0.0802 ** (0.0327)	0.0150 (0.0543)	0.1337 *** (0.0361)
${lnCoal}_{it}$	0.0630 (0.0507)	0.1595 *** (0.0347)	0.1817 *** (0.0370)
F-statistic	245.82 ***	713.29 ***	173.46 ***
Breusch-Pagan	4217.77 ***	3120.09 ***	4611.23 ***
Hausman	4.8	1.03	1.35

Note: *** p < 0.01; ** p < 0.05.

. The results for the southern Jiangsu region in the second column show that the coefficient estimates for population urbanization, economic urbanization, per capita GDP, and industrial structure are statistically significant at the 5% significance level, whereas these for land urbanization and coal consumption are insignificant. Economic urbanization has the greatest impact on CO₂ emissions, with a 0.5661% increase in CO₂ emissions for each 1% increase in economic urbanization. The elasticity of CO₂ emissions to per capita GDP is 0.2752, suggesting that CO₂ emissions rise by 0.2752% for each 1% increase in per capita GDP. The elasticities of population urbanization and industrial structure are 0.0736 and 0.0802, respectively, indicating that CO₂ emissions increase by 0.0736% and 0.0802%.

The empirical results for the middle Jiangsu region in the third column in Table 7 show that only per capita GDP and coal consumption are statistically significant at the 5% significance level. For each 1% increase in per capita GDP, CO₂ emissions in middle Jiangsu province increase by 0.1393%. For each 1% increase in 0.1595 in coal consumption, CO₂ emissions in the middle Jiangsu province increase by 0.1595%. The results for the northern Jiangsu region in the fourth column demonstrate that, at the 5% significance level, CO₂ emissions are strongly correlated with economic urbanization, population urbanization, per capita GDP, industrial structure, and coal use. In northern Jiangsu province, industrial structure ranks third, economic urbanization ranks second, and coal consumption ranks first when it comes to CO₂ emissions. According to the findings, for every 1% increase in coal usage in the northern Jiangsu province, CO₂ emissions rise by 0.1817%.

The comparison of elasticity estimates and their rankings in three regions of Jiangsu province are presented in

Table 8. Comparison of estimation results for three regions.

Variable	Southern Jiangsu		Middle Jiangsu		Northern Jiangsu
	Elasticity	Rank	Elasticity	Rank	Elasticity	Rank
${lnPU}_{it}$	0.0736 **	4	0.0131		0.0350 **	5
${lnEU}_{it}$	0.5661 **	1	0.2581		0.1453 **	2
${lnLU}_{it}$	0.1344		0.0026		−0.0479
${lnGDP}_{it}$	0.2752 ***	2	0.1393 ***	2	0.1053 ***	4
${lnInd}_{it}$	0.0802 **	3	0.0150		0.1337 ***	3
${lnCoal}_{it}$	0.0630		0.1595 ***	1	0.1817 ***	1

Note: *** p < 0.01; ** p < 0.05.

. The ranking of elasticity coefficients in southern Jiangsu is economic urbanization, per capita GDP, industrial structure, and population urbanization; that in middle Jiangsu is coal consumption and per capita GDP; that in northern Jiangsu is coal consumption, industrial structure, economic urbanization, per capita GDP, and population urbanization. As a result, the effects of economic and population urbanization, per capita GDP, industrial structure, and coal usage on CO₂ emissions are heterogeneous across regions. Depending on the leading factors in the three regions, policymakers should have differentiated policy incentives to mitigate CO₂ emissions.

The per capita GDPs of the southern, middle, and northern Jiangsu regions in 2021 will be 174,604 Yuan, 141,203 Yuan, and 8933 Yuan, respectively, so they can also be classified into high-, middle-, and low-income regions. While the effects of coal consumption and industrial structure are more pronounced in low-income regions, the effects of population urbanization, economic urbanization, and per capita GDP are found to be larger in high-income regions than in low- and middle-income regions. Our results agree with the research of [13,49]. Their findings provide credence to the urban environmental transition theory by indicating that urbanization in megalopolises like Beijing and Shanghai may result in higher CO₂ emissions. According to the urban environmental transition theory, if proper infrastructural support is not provided, urbanization may result in environmental problems [13]. In our instance, it is probable that more public and private infrastructure would be built in developed regions to accommodate the growing urban population and economies, leading to higher energy consumption and carbon emissions.

Based on the urban environment transition theory, environmental impacts are inevitable products of the early stages of urbanization, and these impacts are more obvious in countries with high incomes. Jiangsu province is the second-largest economy in China, with per capita GDP ranking after Beijing and Shanghai in 2021 [25]. On the one hand, private and public infrastructures are more demanded by the urban population in high-income regions, and large amounts of CO₂ are emitted during the construction, operation, and maintenance stages. On the other hand, the consumption pattern and lifestyle in the cities are more resource-intensive, which results in CO₂ emissions. To achieve the primary objective of high-quality development, governments in Jiangsu province should develop a low-carbon economy by optimizing industrial structures, improving energy consumption efficiency, and promoting technological innovation.

These urbanization-related findings, however, conflict with findings from other studies [20,39]. Furthermore, He et al. [20] explain that it is plausible that developed regions have decreased their coal usage by substituting clean energy for coal. It is probable that the energy consumption patterns in the three Jiangsu provincial regions haven’t been successful in converting fossil fuels into clean energy in the developed regions. Our findings counter the compact city hypothesis, which addresses the advantages of urban compaction for the environment. By reducing private travel, increasing urban public infrastructure, and lowering energy consumption and CO₂ emissions, this theory highlights the advantages of economies of scale [13,84,85].

4.4. Mechanism Analysis

The estimated results of the mechanism analysis with the random effect model are presented in

Table 9. Estimated results of the mechanism analysis.

Variable	${\mathit{l}\mathit{n}\mathit{G}\mathit{D}\mathit{P}}_{\mathit{i}\mathit{t}}$	${\mathit{l}\mathit{n}\mathit{G}\mathit{D}\mathit{P}}_{\mathit{i}\mathit{t}}$	${\mathit{l}\mathit{n}\mathit{I}\mathit{n}\mathit{d}}_{\mathit{i}\mathit{t}}$	${\mathit{l}\mathit{n}\mathit{I}\mathit{n}\mathit{d}}_{\mathit{i}\mathit{t}}$	$\mathit{l}\mathit{n}\mathit{C}\mathit{o}\mathit{a}{\mathit{l}}_{\mathit{i}\mathit{t}}$	$\mathit{l}\mathit{n}\mathit{C}\mathit{o}\mathit{a}{\mathit{l}}_{\mathit{i}\mathit{t}}$
${lnPU}_{it}$	0.0055 (0.0220)	0.0487 *** (0.0182)	0.0675 ** (0.0276)	0.0466 * (0.0271)	0.0974 *** (0.0277)	0.0911 *** (0.0225)
${lnEU}_{it}$	1.0325 *** (0.1175)	1.0404 *** (0.1023)	1.9053 *** (0.1471)	1.8651 *** (0.1498)	0.0657 (0.1479)	0.4990 *** (0.1304)
${lnLU}_{it}$	0.0898 *** (0.0231)	0.0964 *** (0.0114	0.1045 *** (0.0289)	0.0986 *** (0.0286)	0.0171 (0.0291)	0.0639 *** (0.0238)
${lnGDP}_{it}$				0.0253 (0.0384)		0.6944 *** (0.0265)
${lnInd}_{it}$		0.0114 (0.0174)				0.1489 *** (0.0212)
${lnCoal}_{it}$		0.4526 *** (0.0173)		0.2146 *** (0.0305)

Note: *** p < 0.01; ** p < 0.05; * p < 0.10.

. The second and third columns display that multi-dimensional urbanizations affect CO₂ emissions through economic development, which supports Hypothesis 4. The estimated elasticities of economic urbanization and land urbanization to per capita GDP are 1.0325 and 0.0898, respectively, and are statistically significant at the 1% significance level. When the control variables are added in the third column, population urbanization turns from insignificant to significant, with an estimated coefficient of 0.0487. The coefficient of coal consumption is 0.4526, which is also significant at the 1% significance level. These results indicate that population, economic, and land urbanizations and coal consumption contribute to economic development. This finding agrees with previous studies of the effect of urbanization on economic development. The urbanization process enables the division and specialization of labor, forms cost advantages for producers, and finally achieves scale economies [86]. Economic growth demands energy consumption and leads to carbon emissions; as a result, economic development accounted for 73% of the global CO₂ emissions from 1997 to 2015 [87].

The fourth and fifth columns validate Hypothesis 5 by identifying that multi-dimensional urbanizations affect CO₂ emissions through industrial structure upgrading. The coefficients of population urbanization, economic urbanization, and land urbanization are 0.0675, 1.9053, and 0.1045, respectively, and are significant at least at the 5% significance level. When control variables are added to the model, the magnitudes and significance of these variables are similar. The estimated coefficient of coal consumption is 0.2146 and is significant at the 1% significance level. Economic urbanization is the major driver of industrial structure. These results indicate that urbanization promotes industrial structure upgrading, which is consistent with previous studies. The urbanization process facilitates the migration of population, economy, and land from agricultural sectors to industrial and tertiary sectors, thus promoting industrial structure upgrading [69,71]. The industrialization stage demands more energy consumption and leads to more carbon emissions.

The sixth and seventh columns indicate that multi-dimensional urbanizations affect CO₂ emissions through coal consumption, which supports Hypothesis 6. The coefficient of population urbanization is 0.0974 at the 1% significance level, whereas other variables are statistically insignificant. Taking other control variables into account, all variables are significant at the 1% significance level. The coefficients of population, economic, and land urbanizations are 0.0911, 0.4990, and 0.0639, respectively, while the coefficients of economic development and industrial structure are 0.6944 and 0.1489. This result demonstrates that multi-dimensional urbanizations increase CO₂ emissions through the mediating role of coal consumption. During the population, economic, and land urbanizations, coal is consumed more to provide energy for various sectors, which increases carbon emissions [73].

5. Conclusions

This study investigates the direct and indirect effects of various dimensions of urbanization on CO₂ emissions at the county level in Jiangsu province by using a balanced panel dataset of 80 counties from 2002 to 2021. The robustness tests are conducted to validate the empirical results. Furthermore, depending on the administrative region and economic performance, the whole of Jiangsu province was divided into the southern, middle, and northern regions to examine the regional heterogeneity. Finally, a mediating effect model is applied to examine the mechanism through which urbanizations affect CO₂ emissions. This study is the first research conducted at the county level in China to shed light on the connection between urbanization and CO₂ emissions from the micro perspective. The findings of this study are expected to enrich existing literature and provide some reference to policymakers in Jiangsu province about the environmental impacts of rapid urbanizations.

The first finding of this research is that economic and population urbanizations have positive effects on CO₂ emissions in Jiangsu province, while land urbanization has no discernible impact. The positive correlation between urbanizations and environmental degradation found in this study lends support to the urban environmental transition theory. Therefore, throughout the urbanization stage, it is important to emphasize both environmental and economic developments. Moreover, the CO₂ emissions in Jiangsu province are significantly positively impacted by coal usage, industrial structure, and per capita GDP. These results are validated by various robustness tests.

The second conclusion is that the impact of urbanizations on CO₂ emissions in Jiangsu province exhibits significant regional heterogeneity. Urbanization due to economic and population growth has varying effects on CO₂ emissions; the highest effects are observed in the high-income southern Jiangsu regions, the lowest in the low-income northern Jiangsu, and nonexistent in the central Jiangsu. While coal consumption is the primary element influencing CO₂ emissions in the northern Jiangsu region, economic urbanization is the main driver in the southern and central Jiangsu regions. Depending on the major driver of environmental pollution across three regions, policymakers should have differentiated regulations to manage resource demands and allocation, and finally achieve the sustainability of urban areas.

The third conclusion is that, besides the direct effects, multi-dimensional urbanizations affect CO₂ emissions indirectly through economic development, industrial structure, and coal consumption. The migration of the rural population to urban areas, the transformation of agricultural sectors to industrial and tertiary sectors, and land conversion from agricultural uses to industrial uses have promoted economic development, thus leading to more carbon emissions. The urbanizations have also facilitated industrial structure upgrades by increasing the percentage of industrial value added to GDP, which increases carbon emissions. The rising population in urban areas demands more energy consumption from coal combustion, which drives up the emission of CO₂.

The analysis conducted in this study shows that environmental degradation is an evitable stage of urbanization in Jiangsu province, so various mitigation strategies should be implemented. To achieve CO₂ reductions, county governments should invest more in technological innovations to increase energy usage efficiency. On the other hand, local governments could develop renewable energies by exploiting natural resources such as sunlight, wind, and water. Furthermore, local governments could promote industrial transformation by encouraging the transformation from industrial industry to tertiary industry. Under these efforts, sustainable urban transformation could possibly be achieved.