Impact of Economic Agglomeration on Carbon Emission Intensity and Its Spatial Spillover Effect: A Case Study of Guangdong Province, China

Abstract

Social and economic growth in developing countries has heightened the awareness of environmental challenges, with carbon emissions emerging as a particularly pressing concern. However, the impact of economic development on carbon emission intensity has rarely been considered from the perspective of economic agglomeration, and the relationships and mechanisms between the two remain poorly understood. We analyzed the impact of economic agglomeration on carbon emission intensity and its spatial spillover effect in Guangdong Province, the most economically advantaged province of China, based on a spatial weight matrix generated using geographic proximity, exploratory spatial data analysis (ESDA), and the spatial Durbin model. Between 2000 and 2019, economic agglomeration and carbon emission intensity in Guangdong Province exhibited persistent upward trajectories, whereas between 2016 and 2019, carbon emission intensity gradually approached zero. Further, 80% of the province’s economic output was concentrated in the Pearl River Delta region. Strong spatial autocorrelation was observed between economic agglomeration and carbon emission intensity in the cities, and the economic agglomeration of the province had a parabolic influence on carbon emission intensity. Carbon emission intensity peaked at an economic agglomeration level of 1.2416 × 10⁹ yuan/km² and then gradually decreased. The spatial spillover effect of the openness degree on carbon emission intensity was positive, while GDP per capita and industrial structure had negative effects. Further, the economic agglomeration effects of Guangdong Province increased the carbon emission intensity of major cities and smaller neighboring cities. The stacking effect of economic agglomeration between cities also affected the carbon emission intensity of neighboring cities in the region. During the period of rapid urban development, industrial development and population agglomeration increased resource and energy consumption, and positive externalities such as the scale effect and knowledge spillover were not well reflected, resulting in greater overall negative environmental externalities relative to positive environmental externalities.

1. Introduction

Industrialization and urbanization are major accomplishments of human civilization that have promoted rapid economic development; however, they have also brought about significant challenges, such as a simultaneous increase in carbon emissions, which severely threatens sustainable regional development [1,2]. The relationship between economic development and carbon dioxide (CO₂) levels has become a popular research topic [3]. According to the latest assessment report of the Intergovernmental Panel on Climate Change (IPCC), increases in global average temperature have been primarily attributed to increases in CO₂ gas caused by human activities, such as fossil fuel use and tropical deforestation [4]. In 2018, the IPCC formed a consensus on carbon peak goals, carbon neutrality, and 1.5 °C warming targets [5]. The relationship between CO₂ emissions and economic development is the essence of carbon neutrality and CO₂ emission peak goals [6,7]. Therefore, to promote the coupling, coordination, and balance between the economy and environment, further investigations are required to determine the spatiotemporal evolution characteristics of regional economic agglomeration and carbon emission intensity, reveal the influence of economic agglomeration on carbon emission intensity, and identify the spatial spillover effects on neighboring areas and their impact mechanisms.

Economic agglomeration represents a concentration of economic activities in a certain geographical space [8,9], and it links economic activities and affects the geographical spatial distribution of carbon emission quantity and intensity [10]. Economic agglomeration may exacerbate and intensify CO₂ based on the larger production scale [11]. Moreover, it is conducive to the positive effects of technological innovation and knowledge spillover, and improves the utilization efficiency of input factors, which effectively reduces carbon emissions [12,13]. Research has mainly focused on the impact of economic agglomeration externalities on carbon emissions. Due to the heterogeneity of economic development at different spatial scales, different levels of economic agglomeration have significantly different effects on carbon emissions [14]. Previous research on the relationship between economic development and carbon emissions has mainly yielded three major findings: (1) A one-way linear relationship exists between the economy and carbon emissions, with externalities having both positive and negative impacts. High-quality development of modern society promotes innovation of production efficiency, which has a positive effect on the inhibition of carbon emissions. However, increases in economic activities promote increases in carbon emissions [15]. Moreover, CO₂ has spatial fluidity and spatial spillover effects on neighboring areas. (2) Economic agglomeration is a dynamic and complex cyclical process that manifests as different external characteristics at different stages of the life cycle. This suggests a nonlinear relationship between economic factors and carbon emissions [3,6]. The results also showed that economic agglomeration and carbon emissions have an “inverted N-shaped” relationship at the county level. Thus, economic agglomeration must exceed a certain threshold before it leads to significant reductions in emissions. Moreover, economic agglomeration has a spillover effect on carbon emission intensity; that is, economic agglomeration impacts carbon emissions in and around the region [10]. The nonlinear relationship between the two has been reported to be similar to the “inverted U-shape” of a parabola [16]. In the early stage of economic development, the economic structure, dominated by secondary industry (manufacturing industry), causes a large increase in carbon emissions; however, with development, the economic structure transforms from secondary to tertiary industry (service industry). Moreover, under the dual effects of energy conservation and emissions reduction, carbon emissions gradually decrease after peaking [17]. Energy intensity may strengthen the relationship between economic expansion and carbon emissions [18]. Researchers have also reported an “inverted U-shaped” nonlinear relationship between industrial agglomeration and green economic efficiency as well as environmental pollution [19]. (3) A two-way relationship occurs between the economy and CO₂, with economic agglomeration both promoting and inhibiting carbon emissions [20]. In addition, carbon emissions are not static but gaseous and present spatial fluidity and trans-regional transmission characteristics. CO₂ levels in a region pose environmental problems and impact surrounding areas, and they are affected by surrounding areas [21]. The formulation and implementation of different policy measures also have significant impacts on economic agglomeration and carbon emissions. Since 2000, China has explicitly set binding energy-saving targets in its various plans. At the end of 2014, China first pledged to reach its carbon peak by 2030, which poses a challenge to China’s future high-quality economic development [22]. In terms of research methods, the relationship between CO₂ and economic and social factors is often studied based on IPAT, STIRPAT, and spatial econometric models [23,24] and the spatial spillover effects of economic factors and CO₂ levels [25,26]. The mediating effect model, panel smooth transition regression model, Tapio decoupling model [27,28], nonlinear autoregressive distributed lag (NARDL) model [29], and other quantitative methods have also been applied to analyze its influence. The convergence model and other methods have been used to simulate future trends in carbon emission intensity [30].

Although research on the impact of economic development on carbon emissions is more in-depth and specific, the perspectives on this topic are diverse. Spatiotemporal differences in the impact of economic development on the amount of carbon emissions have mainly been studied from different dimensions of economic development, with a focus on the relationships between economic development and total carbon emissions. Thus, the impact of economic development on carbon emission intensity has rarely been considered from the perspective of economic agglomeration, and the relationships and mechanisms between the two remain poorly understood.

In this study, Guangdong Province was selected as the study area because it is an economically advantaged province in China with a high level of urbanization, dense population, and developed economy, and an analysis was performed on the spatiotemporal distribution characteristics of the carbon emission intensity and economic agglomeration levels in the province from 2000 to 2019. The spatial Durbin model was used to analyze the spillover effect of the economic agglomeration level of cities in Guangdong Province and the influencing factors of carbon emission intensity. Finally, targeted policy recommendations were proposed to achieve CO₂ emission peaks and carbon neutrality in economically developed regions of developing countries.

The innovation of this study lies in two aspects: first, it shifts the focus from solely examining carbon emissions to also considering their intensity; second, it investigates how economic agglomeration influences spatial spillover effects on carbon emission intensity. Economic growth often leads to spatial interconnections that promote the collective development of neighboring regions, thereby creating an economic development corridor. Does this spatial interconnectedness affect carbon emission intensity? To answer this question, the cumulative impact of economic agglomeration on adjacent cities’ carbon emission intensities is analyzed by exploring the spatial spillover effects between neighboring urban areas.

2. Overview of the Study Area and Research Methods

2.1. Overview of the Study Area

Guangdong Province is located in the southernmost part of mainland China at 20°13′ N–25°31′ N and 109°39′ E–117°19′ E and contains the Pearl River Estuary, which is bordered on the east, west, and southwest sides by Hong Kong, Macao, and the Qiongzhou Strait, respectively. This province has a long coastline; its sea area is located at the key point for maritime communication in Southeast Asia and it represents a shipping hub in the South China Sea. A superior geographical location is an important condition for the economic growth of this province.

The province has a land area of 179,800 km² and is divided into four areas: eastern Guangdong, western Guangdong, northern Guangdong, and the Pearl River Delta. There are 21 prefecture-level cities under the jurisdiction of the province, and Guangzhou is the provincial capital (Figure 1). This province encompasses tropical and subtropical regions, with the Tropic of Cancer crossing the central region. Thus, this province experiences high temperatures, abundant precipitation, and superior natural conditions, which have laid a good foundation for land resources. In general, the topography is high in the north and low in the south, with a decline from the mountains of northern Guangdong to the southern coast. Thus, the geomorphological pattern is dominated by moderate mountains in the north, low mountains and hills in the middle, and plain platforms in the south.

According to statistics for Guangdong Province in 2021, GDP was 12.437 trillion yuan, with the value-added of secondary and tertiary industries accounting for 5021.919 billion yuan (40.38%) and 6914.682 billion yuan (55.60%), respectively. With 1.87% of the land area and 10.87% of the GDP of China, this province has a strong economy, and the Pearl River Estuary represents the southern gate of China’s foreign economy. At the forefront of the reform and opening up period, Guangdong developed rapidly based on its large foreign population and policy advantages. Urbanization in Guangdong has surpassed that of all provinces and cities in the country; however, the imbalance between the economic and urbanization development of various regions is becoming increasingly prominent. In 2021, the Pearl River Delta region accounted for 80.88% of the GDP of the province, whereas the remaining regions accounted for less than 20%, with obvious regional differences. The economic agglomeration in Guangdong Province presents obvious spatial characteristics. Founded in 1984, the Guangzhou Development Zone was among the first state-level economic and technological development zones, and it implemented the new “four in one” management model. In addition, there are various types of industrial parks with highly concentrated industries, such as the Guangzhou Nansha Economic and Technological Development Zone, Nanhai Park of the Foshan High-tech Industrial Development Zone, Jiangmen Pengjiang Industrial Transfer Industrial Park, and Guangzhou Huadu Economic Development Zone.

2.2. Research Methods

2.2.1. Spatial Weight Matrix Based on Geographic Proximity

The spatial weight matrix is a method used for quantifying the position of spatial units, the characteristics of spatial structure, and the relationship between spatial units; thus, it is a method of quantifying the relative spatial position of one spatial unit I of a region on another spatial unit J and the spatial interaction of the region [31]. Spatial weight matrices can be divided into several types: geographical proximity-based, spatial location-based, and socioeconomic structure-based matrices. Based on the actual situation in the study area, we selected the more commonly used spatial weight matrix based on Queen proximity.

$${\omega }_{ij}=\{\begin{array}{c}1, b_i{\displaystyle \cap }{b}_j\ne 0\\ 0, b_i{\displaystyle \cap }{b}_j=0\end{array}$$

(1)

When the boundary length of the intersection of space unit i and space unit j is greater than 0, space units i and j can be considered adjacent, and the units in row i and column j in the space weight matrix are denoted as 1. Otherwise, it is 0.

2.2.2. Exploratory Spatial Data Analysis

By describing and analyzing the spatial distribution pattern of objects, ESDA can identify spatial agglomeration and differentiation and reveal the spatial dependence between the objects [32,33]. At the core is the measurement of spatial autocorrelation, which includes global and local spatial autocorrelation (LSA) analyses.

The global Moran’s I index is a global spatial autocorrelation analysis method that can reveal the spatial correlation between economic agglomeration levels and carbon emission intensity in the process of urban development.

$$I={\displaystyle \frac{{{\displaystyle \sum }}_{i=1}^n{{\displaystyle \sum }}_{j=1}^n{W}_{ij}\left(Y_i-\overline{Y}\right)\left(Y_j-\overline{Y}\right)}{S^2{{\displaystyle \sum }}_{i=1}^n{{\displaystyle \sum }}_{j=1}^n{W}_{ij}}}$$

(2)

where I is the global Moran’s I index to be calculated, and its value presents a range of [−1, 1]. When I is greater than 0, then the city’s economic agglomeration level or carbon emission intensity tends to be concentrated. When I is less than 0, then the spatial autocorrelation is negative and the overall correlation index of the study area presents high and low spatial agglomeration distribution values adjacent to each other. When the value is 0, then there is no spatial autocorrelation. In addition, n indicates the number of cities studied; Y_i and Y_j represent the economic agglomeration levels or carbon emission intensities of cities i and j, respectively; $\overline{Y}$ is the mean of Y; S² is the variance of Y; and W_ij is the spatial weight matrix. The local Moran’s index can be used to analyze the strength of spatial dependence in the vicinity of region (city) i in more depth, and is calculated as follows:

$$I_i={\displaystyle \frac{n\left(x_i-\overline{x}\right){{\displaystyle \sum }}_{j=1}^n{w}_{ij}\left(x_j-\overline{x}\right)}{{{\displaystyle \sum }}_{i=1}^n{\left(x_i-\overline{x}\right)}^2}}$$

(3)

A positive number indicates that the high/low value of region (city) i is concentrated with high/low values in the surrounding area; and a negative number indicates that the high/low value of region (city) i is concentrated with low values (high values) in the surrounding area.

2.2.3. Spatial Durbin Model

The spatial Durbin model integrates the common effects of the spatial lag terms of the dependent and explanatory variables on the dependent variable, and it can better estimate the spillover effect of different cities (observation units) and the spatial spillover effect based on panel data [34]. Based on previous research [35,36] and a comprehensive analysis of the study area conditions, the quadratic term economic agglomeration was introduced to construct a spatial Durbin model of the impact of economic agglomeration on carbon emission intensity in Guangdong Province.

$$ce{i}_{it}={\beta }_0+\rho Wce{i}_{it}+{\beta }_{1}a{e}_{it}+{\beta }_{2}a{e}_{it}^2+{\beta }_{3}o{l}_{it}+{\beta }_{4}pgd{p}_{it}+{\beta }_{5}i{s}_{it}+{\beta }_{6}t{i}_{it}+\gamma X_{it}+{\epsilon }_{it}$$

(4)

where cei represents the carbon emission intensity and is the dependent variable; ae represents economic agglomeration and is the explanatory variable; ol represents the openness levels and is the control variable; pgdp represents the per capita gross domestic product and represents the ratio of the annual total import and export value of each city to the city’s GDP; is represents industrial structure and is the ratio of the annual added value of the city’s secondary industry to the gross regional product; ti represents industrial technology innovation and is the ratio of the research and experimental development expenditure of industrial enterprises to the city’s GDP; $\rho$ is the spatial regression correlation coefficient, that is, the direction and degree of influence of carbon emission intensity spillover in prefecture-level cities; W is the spatial weight matrix based on geographic proximity constructed above; $Wce{i}_{it}$ is the lagging term of the carbon emission intensity of each city; ${\epsilon }_{it}$ is a random distractor; X is the product of the independent variable and the spatial weight matrix W; and $\gamma$ is the estimated coefficient of each product. The subscripts i and t are the prefecture-level cities and years of Guangdong Province, respectively.

2.3. Data Source

The panel data of 21 prefecture-level cities in Guangdong Province from 2000 to 2019 were used as the analysis samples. The data were derived from the Guangdong Statistical Yearbook, statistical yearbooks of various cities, and statistical bulletin of national economic and social development, in which the economic variables of all years were converted to constant prices in 2000 to eliminate the impact of price changes in each year on the research results. Due to changes in the data indicators in the yearbook, the indicators of “research and experimental development funds for industrial enterprises” in various cities from 2000 to 2008 used the indicators of “research and experimental development funds for large industrial enterprises”. From 2009 to 2019, this indicator was expressed by the indicator of “internal expenditure of research and experimental development funds of industrial enterprises above designated size”. Carbon emissions data were obtained from the China Carbon Accounting Database (https://www.ceads.net.cn/, accessed on 10 March 2024) [37].

3. Results

3.1. Evolutionary Characteristics of Economic Agglomeration and Carbon Emission Intensity in Guangdong Province

3.1.1. Spatiotemporal Evolution Characteristics of Economic Agglomeration Level in Guangdong Province

The ratio of the output value of secondary and tertiary industries to the area of urban administrative areas was used as an indicator to measure economic agglomeration levels. From 2000 to 2019, the economic agglomeration levels in Guangdong Province showed a continuous upward trend, increasing from 0.0556 × 10⁸ to 0.3794 × 10⁸ yuan/km². The level and time of economic agglomeration were linearly fitted according to the characteristics of the curve, and the coefficient of the primary term was 0.018, indicating that the average annual growth rate of the economic agglomeration level in Guangdong Province was approximately 0.018 × 10⁸ yuan/km². Significant differences were observed in the level of economic agglomeration in the four regions (Figure 2 and Figure 3). The level of economic agglomeration in the Pearl River Delta region was much higher than the average level in the province, which was 1.0299 × 10⁸ yuan/km² in 2019. However, the economic agglomeration levels of eastern, western, and northern Guangdong were lower than the provincial average, especially in northern Guangdong, which was the lowest in 2019 (0.0429 × 10⁸ yuan/km²). From the perspective of the average annual growth rate of economic agglomeration, the growth rate of the Pearl River Delta region was much greater than that of the other regions, with an average annual growth rate of 4.66%; eastern Guangdong followed by 1.13%; however, the growth rates in western and northern Guangdong were slow, with average annual growth rates of only 0.50% and 0.19%, respectively. The difference in the level of economic agglomeration between the Pearl River Delta region and other regions has become increasingly obvious.

From the perspective of various cities (Figure 3), Shenzhen had the highest level of economic agglomeration in 2000, with a value of 1.087 × 10⁸ yuan/km², which was much larger than that of Guangzhou (0.331 × 10⁸ yuan/km²), and the third and fourth levels were observed in Dongguan (0.323 × 10⁸ yuan/km²) and Foshan (0.260 × 10⁸ yuan/km²), respectively. Driven by large cities, the economic agglomeration levels in adjacent regions have also significantly improved. Heyuan, Qingyuan, Meizhou, and Shaoguan, the four cities in northern Guangdong, had the lowest economic agglomeration levels at 0.0036 × 10⁸, 0.0051 × 10⁸, 0.0078 × 10⁸, and 0.0081 × 10⁸ yuan/km², respectively. In 2019, Shenzhen’s economic agglomeration level was much higher at 8.878 × 10⁸ yuan/km². Led by Shenzhen, Dongguan rose to second place with a value of 2.525 × 10⁸ yuan/km², followed by Guangzhou and Foshan. The characteristics of economic agglomeration in the Pearl River Delta region are becoming increasingly obvious: The top six cities with the highest levels of economic agglomeration in the province are all distributed in the Pearl River Delta region, while the cities with the lowest levels of economic agglomeration in the province are distributed in northern Guangdong. From 2000 to 2019, the economic agglomeration levels in 21 cities showed a continuous upward trend, although the growth rate had obvious regional differences. Heyuan and Qingyuan in northern Guangdong had the fastest average annual growth rate of economic agglomeration at 53.10% and 46.47%, respectively, followed by Shenzhen, Dongguan, and Zhaoqing in the Pearl River Delta region at 37.71%, 35.89%, and 35.56%, respectively. Shantou and Chaozhou in eastern Guangdong had the slowest annual growth rates in the province, with 16.36% and 17.58%, respectively.

3.1.2. Spatiotemporal Evolution Characteristics of Carbon Emission Intensity in Guangdong Province

From 2000 to 2019, the carbon emission intensity of Guangdong Province continued to increase (Figure 4) from 0.1191 × 10⁴ to 0.3685 × 10⁴ t/km². However, the carbon emission intensity curve from 2016 to 2019 tended to be flat, the growth rate gradually tended to 0, and an obvious growth inflection point was not observed. There were obvious differences in the carbon emission intensity of each region. The carbon emission intensity in the Pearl River Delta region was significantly higher than that of the other regions, and the stage characteristics were obvious. From 2000 to 2007, a remarkable growth trend emerged, and the carbon emission intensity increased from 0.2498 × 10⁴ to 0.5036 × 10⁴ t/km², with an average annual growth rate of 362.59 t/km². The periods from 2007 to 2009, 2011 to 2013, and 2016 to 2019 were stable, during which the carbon emission intensity did not increase but rather decreased slightly from 0.5036 × 10⁴ to 0.5031 × 10⁴ t/km², 0.5911 × 10⁴ to 0.5834 × 10⁴ t/km², and 0.6854 × 10⁴ to 0.6714 × 10⁴ t/km², respectively. The range of carbon emission intensity changes in eastern Guangdong was exceptional, and the stage was also obvious. Overall, the emission intensity increased from 0.0894 × 10⁴ to 0.5408 × 10⁴ t/km². The period from 2000 to 2011 was a stage of rapid increase, with a total increase of 4124.411 t/km², while 2011 to 2016 was a stable period in which the carbon emission intensity decreased slightly by 161.539 t/km².

After another significant increase in carbon emission intensity from 2016 to 2018, the level decreased to that of the previous year in 2019. The carbon emission intensity levels of western and northern Guangdong were similar, and the overall change was relatively flat. However, the carbon emission intensity of northern Guangdong was higher than that of western Guangdong before 2010. After 2010, western Guangdong surpassed northern Guangdong, and the carbon emission intensity of western Guangdong increased slightly in 2016.

From 2000 to 2019, the carbon emission intensity of all cities in the province showed a gradually increasing trend (Figure 5). In 2000, the carbon emission intensity of 21 cities was less than 1 × 10⁴ t/km². In 2005, the carbon emission intensities of four cities in the Pearl River Delta exceeded 1 × 10⁴ t/km². In 2010, the city of Shantou in eastern Guangdong was added to this list, while the city of Zhuhai in the Pearl River Delta region was added in 2015. Five cities in the Pearl River Delta region have high carbon emission zones. In 2000, there were 10 cities in the province with carbon emission intensities less than 0.1 × 10⁴ t/km²; in 2010, only Zhaoqing and Heyuan presented such values; and after 2015, only Heyuan City presented such values. In 2000 and 2019, Dongguan had the highest carbon emission intensity in the province and was the city with the largest increase in carbon emission intensity, from 0.9061 × 10⁴ to 2.5650 × 10⁴ t/km². Cities along the Pearl River Estuary have always been high-value carbon emission intensity clusters, and the spatial distribution of carbon emission intensity in the Shenzhen-Dongguan urban belt and Guangzhou-Foshan urban belt was evident.

3.2. Spatial Autocorrelation Analysis of Economic Agglomeration and Carbon Emission Intensity

Economic agglomeration is a geographical phenomenon with obvious spatial characteristics, while CO₂ disperses based on the atmosphere. Therefore, economic agglomeration and carbon emission intensity characteristics were analyzed from a spatial perspective. Based on Stata16, we conducted a multicollinearity test on the research panel data and showed that the tolerance (Tol) of the explanatory variable and each control variable was between 0.266 and 0.671, while the variance expansion factor (VIF) of each variable was between 1.49 and 3.76. Both indices reached the standards of Tol ≥ 0.1 and VIF ≤ 5, which are commonly used in academia to judge collinearity, indicating no multicollinearity problem between the explanatory and control variables of the data in this panel, which can be used for further spatial analysis.

Moran’s I index was applied using Stata16 to measure the global spatial autocorrelation between economic agglomeration level and carbon emission intensity in Guangdong Province (Figure 6 and Figure 7). From 2000 to 2019, Moran’s I index of economic agglomeration showed a trend of initially increasing, decreasing, and gradually tending toward stability.

Moran’s I index increased from 0.164 to 0.227 from 2000 to 2008, decreased to 0.166 year by year from 2008 to 2016, and tended to be stable without obvious fluctuations from 2016 to 2019. From the perspective of carbon emission intensity, the change in the global autocorrelated Moran’s I index from 2000 to 2019 showed a fluctuating parabolic trend, increasing from 0.072 in 2000 to 0.364 in 2009, and the volatility decreased to 0.256 in 2018. The spatial autocorrelation between economic agglomeration and carbon emission intensity in each year was significant at the 95% confidence level, indicating that the economic agglomeration and carbon emission intensity of the 21 prefecture-level cities in Guangdong Province had strong spatial autocorrelation characteristics of high- and high-value agglomeration and low- and low-value agglomeration.

To further study the spatial correlation between each city, the local Moran’s I index analysis method was applied based on GeoDa1.20 software to analyze the LSA of each city, and the scatter plot of economic agglomeration level and carbon emission intensity was obtained (Figure 8). From 2000 to 2019, a significant spatial autocorrelation was observed between the level of economic agglomeration and carbon emission intensity in Guangdong Province, and there was spatial correlation, with high–high and low–low value agglomerations. Among them, the local Moran’s I index of economic agglomeration in Guangdong increased from 0.162 to 0.212 and then gradually decreased to 0.166 from 2000 to 2019, and the LSA showed a significant parabolic change trend. Overall, in 2000, 2010, and 2019, cities with high–high and low–low economic level agglomerations accounted for 71.43%, 76.19%, and 80.95% of the 21 cities in Guangdong, respectively, of which, 4 cities had high–high economic agglomerations and 11 had low–low agglomerations in 2000. By 2010, the number of cities with low–low agglomerations had increased to 12, and by 2019, the number of cities with low–low agglomerations had increased to 13. Shantou and Jiangmen were newly added to the low–low agglomeration group. In addition, the local Moran’s I index of carbon emission intensity in Guangdong increased from the lowest value of 0.265 to a peak of 0.338 from 2000 to 2019, and then fluctuated between the highest and lowest values. The number of cities with high–high and low–low agglomerations of carbon emission intensity in the province increased from 14 to 15, among which, Zhaoqing and Jiangmen changed from original high–low agglomeration cities to low–low agglomeration cities, while Zhongshan changed from an original high–high agglomeration city to a high–low agglomeration city. Among them, the number of cities with low–low agglomeration carbon emission intensities increased from 10 in 2000 to 12 in 2019, and the number of cities with high–high agglomeration increased from 4 in 2000 to 5 in 2005, and then decreased to 3 in 2015.

3.3. Spatial Spillover Effects of Economic Agglomeration on Carbon Emission Intensity

3.3.1. Spatial Econometric Regression Analysis

To determine the specific form of the spatial econometric model used, the Lagrange multiplier (LM) test, Wald test, likelihood ratio (LR) test, and Hausman test were performed on panel data based on Stata16. First, the LM test showed that the data in this panel were more suitable for the spatial lag model than for the spatial error model. However, the LM test is a preliminary screening test for spatial econometric models, which does not completely represent the test results. Furthermore, the Wald test showed that the chi-square test values of the spatial Durbin model, when degenerated into a spatial lag model and a spatial error model, were 78.21 and 74.05. Using the LR test, the chi-square test values of the spatial Durbin model, when degraded into a spatial lag model and a spatial error model, were 71.41 and 68.23, respectively. The results of the Wald’s and LR tests showed that the spatial Durbin model, when degenerated into a spatial lag model or a spatial error model, should be rejected. Finally, the test statistic of the Hausman test was 125.87, indicating that the fixed-effect spatial Durbin model should be used for the analysis. To determine the fixed object of the fixed-effects model to achieve the best regression effect,

Table 1. Regression results of fixed effects based on spatial Durbin model.

Variables and Symbols	Time Fixation	Individual Fixation	Two-Way Fixation
	cei	cei	cei
Economic agglomeration (ae)	0.6065 ***	0.4954 ***	0.6189 ***
	(0.0637)	(0.0599)	(0.0623)
Economic agglomeration quadratic terms (ae2)	−0.0466 ***	−0.0399 ***	−0.0481 ***
	(0.0056)	(0.0055)	(0.0055)
Openness level (ol)	−0.0622	0.0679	−0.3760
	(0.2629)	(0.2629)	(0.2771)
Per capita GDP (pgdp)	−0.0632 ***	0.0426 ***	−0.0686 ***
	(0.0211)	(0.0129)	(0.0208)
Industrial structure (is)	0.1920	0.1739	−0.3452 *
	(0.1692)	(0.1686)	(0.2092)
Industry technical innovation (ti)	−1.0944	0.1511	−0.6299
	(2.3521)	(2.3837)	(2.3603)
Constant terms (cons)	−40.5824 ***	0.0914	0.3953 ***
	(6.5369)	(0.0752)	(0.1004)
Total sample size (N)	420	420	420
Goodness of fit (R2)	0.5383	0.5371	0.5958

*** and * indicate that the variables pass the significance test of 1% and * 10%, respectively.

shows the regression results of the spatial Durbin model under three fixed effects: individual fixations, time fixations, and two-way fixations.

The goodness-of-fit R² values of the time-fixed, individual-fixed, and bidirectional fixed-mode models were 0.5383, 0.5371, and 0.5958, respectively, and the fitting conditions of the three fixed models were basically the same. A comparison of the significance levels of the coefficients of each variable revealed that the coefficients of the explanatory variables fitted by the three models were all significant at the 1% level. The two-way fixed model had the best fitting effect on the coefficients of the control variables, followed by the time-fixed model. Considering the significance of the regression coefficients of goodness of fit R² and standard error in the three fixed-effects models, the spatial Durbin model with two-way fixed effects was selected as the fitting regression result. According to the regression results of the two-way fixed effects, the primary and quadratic coefficients of economic agglomeration (ae) were 0.4954 and −0.0399, respectively, both of which were significant at the 1% level. The results show that the level of economic agglomeration in Guangdong Province first had a positive impact on its carbon emission intensity (cei), and this then transformed into a negative nonlinear impact relationship, creating an inverted “U”-shaped function curve. The “inflection point” value of the parabola was 12.4160 × 10⁸ yuan/km².

In terms of control variables, the spatial Durbin model with two-way fixed effects revealed that the fitting coefficient of the total effect of openness level was −0.3760, indicating that the carbon emission intensity of cities will gradually decrease with an increase in openness to the outside world. The fitting coefficient of per capita gross domestic product was −0.0686, indicating a negative relationship between per capita GDP and carbon emission intensity. For every 1 unit increase in per capita GDP, carbon intensity decreased by 0.0686 units.

The fitting coefficient of the industrial structure was −0.3452 and passed the significance test of 10%, indicating that the city’s carbon emission intensity decreased by 0.3452 units. The fitting coefficient of industrial technical innovation is −0.6299, which is negatively correlated, indicating that improving the scientific and technological innovation of urban industrial enterprises can reduce their carbon emission intensity.

3.3.2. Analysis of Spatial Spillover Effects

Based on the characteristics of the spatial Durbin model, the total spatial spillover effect of the spatial Durbin model constructed using the Stata16 pair was decomposed into direct and indirect effects (

Table 2. Decomposition results of total spatial spillover effects in spatial Durbin model.

Variable	Variable Symbol	Direct Effects	Indirect Effects	Total Effect
Economic agglomeration	ae	0.4347 *** (0.06612)	0.3577 ** (0.1425)	0.7924 *** (0.1346)
Economic agglomeration quadratic terms	Ae2	−0.0316 *** (0.0056)	−0.0055 (0.0124)	−0.0371 *** (0.0126)
Opening level	ol	−0.3182 (0.2627)	2.2755 *** (0.5321)	1.9573 *** (0.5102)
Per capita GDP	pgdp	−0.0361 * (0.0212)	−0.1375 *** (0.0368)	−0.1737 *** (0.0359)
Industrial structure	is	−0.3510 * (0.1895)	−0.7383 ** (0.3335)	−1.0893 *** (0.3352)
Industry technical innovation	ti	−0.3856 (2.2964)	−0.3223 (4.3982)	−3.6089 (4.5188)

***, **, and * indicate that the variables pass the significance test of 1%, 5%, and * 10%, respectively.

).

For every 1 unit increase in the economic agglomeration levels in Guangdong Province, the carbon emission intensity of the city increased by 0.7924 units, of which the direct and indirect effects provide 0.4347 and 0.3577 units of the total effect, respectively, and the total effect of economic agglomeration was significant at the 1% level. The direct and indirect effects were significant at the 1% and 5% levels, respectively, indicating that the carbon emission intensity of cities in Guangdong Province was directly affected by the level of economic agglomeration in the province and the indirect effects of the level of economic agglomeration in other neighboring cities. However, direct effects remained more influential than indirect effects. From the economic agglomeration quadratic terms, the carbon emission intensity of the city was reduced by 0.0371 units, with direct and indirect effects providing 0.0316 and 0.0055 units of inhibition on the total effect, respectively. Both the direct and total effects were significant at the 1% level, while the indirect effect was too low to pass the significance test. Therefore, the carbon emission intensity of cities in Guangdong Province was affected by the level of economic agglomeration of their own cities and neighboring cities and the stacking effect of economic agglomeration.

From the perspective of control variables, the total spatial spillover effect of the openness level on carbon emission intensity was positive while the per capita GDP, industrial structure, and industrial innovation were negative. Although the total spatial spillover effect value of industrial innovation was large, it did not pass the significance test and thus was not considered. Specifically, for every 1 unit increase in openness level, the city’s carbon emission intensity increased by 1.9573 units, which was mainly manifested in the impact of indirect effects, with direct effects failing to pass the significance test. The spatial spillover effect of industrial structure on carbon emission intensity could be described as the superimposed negative impact of direct and indirect effects. For every 1 unit increase in per capita GDP, the carbon emission intensity was reduced by 0.1737 units and indirect effects had the greatest impact and contributed 0.137 units.

4. Discussion

4.1. Influencing Mechanism of the Spatiotemporal Heterogeneity of Economic Agglomeration and Carbon Emission Intensity in Guangdong Province

Economic and industrial agglomerations are intensive, efficient, and optimized development modes. Areas with a high level of urbanization and a more developed economy are often large-scale agglomerations of population, industry, and economy. From 2000 to 2019, the economic agglomeration levels in Guangdong Province continued to rise, and the economic agglomeration levels of the Pearl River Delta region among the four major regions were higher than those of the other three regions. After nearly 20 years of development, the economic agglomeration levels were significantly higher than those of other regions, consistent with the characteristics of GDP (

Table 3. Comparative characteristics of economic agglomeration and GDP levels in the four major regions and cities in Guangdong Province.

Year	Index	The Pearl River Delta	Eastern Guangdong	Western Guangdong	Northern Guangdong	Guangzhou	Shenzhen	Heyuan	Qingyuan
2000	Economic agglomeration	0.1454	0.0570	0.0180	0.0062	0.3308	1.0873	0.0036	0.0051
	GDP/100 million yuan	8471.28	1067.61	951.37	756.20	2505.58	2219.20	87.22	157.92
2010	Economic agglomeration	0.5318	0.1430	0.0601	0.0244	1.1295	3.9392	0.0192	0.0308
	GDP/100 million yuan	38,028.65	3107.09	3487.27	2934.93	10,640.67	10,069.06	444.03	867.13
2019	Economic agglomeration	1.0299	0.2715	0.1132	0.0429	2.1406	8.8779	0.0403	0.0498
	GDP/100 million yuan	86,899.05	6957.09	7609.24	6205.69	23,628.60	26,927.09	1080.03	1698.22

). Since the reform and opening up in 1978, Shenzhen’s economy has developed rapidly, and its GDP has only been exceeded by that of Shanghai and Beijing in recent years, which have the leading positions of the national economy; therefore, the economic agglomeration level of Shenzhen in 2000–2019 was the highest in the province, with an average annual growth rate of 37.71%. Before 2012, the GDP of Guangzhou was higher than that of Shenzhen; however, since 2012, Shenzhen has become the strongest economic city in the province. Guangzhou is the provincial capital city and has been the economic center of South China since the Qin Dynasty, while Shenzhen’s economic development started late due to the gradual rise of the reform and opening-up policy in 1978. Therefore, the industrial spatial layout of the city is in line with the concept of modern urban development, which completely differs from the developmental characteristics of Guangzhou. Therefore, before 2012, although Guangzhou’s economic level was higher than that of Shenzhen, Shenzhen’s economic agglomeration levels were always higher. In addition, Heyuan and Qingyuan are the two cities with the highest average annual growth rate of economic agglomeration levels in the province, mainly due to the relatively late start of economic development in northern Guangdong and the relatively low starting point of economic development and economic agglomeration levels. Even with the fastest growth rate in the province, the economic agglomeration levels of the two cities in 2019 ranked second and fourth from the bottom of Guangdong Province and were far from the levels of the Pearl River Delta region in 2000, and their economic levels also showed the same characteristics. As for the spatial correlation of economic agglomeration levels, they initially increased and then weakened since 2000, peaking in 2008. Moran’s I index was closer to 1, and the agglomeration trend was more obvious, indicating that the trend of economic agglomeration in Guangdong Province began to weaken after reaching its peak in 2008. This was likely related to the main function positioning of the four major regions of Guangdong Province.

The urbanization and economic levels of the Pearl River Delta region are greater than those of Guangdong Province, and the economy has gradually transformed from high-speed to high-quality development, from the agglomeration of the industrial economy in the past to fulfilling the function of a scientific and technological innovation center or highland, carrying a centralized layout of new quality productivity [38]. The future development of Guangdong Province should focus on gradually narrowing regional differences in economic development and promoting balanced development between regions by narrowing the economic gap between the Pearl River Delta region and eastern, western, and northern regions. This will require the overall stability of the spatial pattern and local adjustments. Moreover, to change the economic agglomeration trend, certain industrial development functions should be gradually transferred from the Pearl River Delta to the east, west, and north of Guangdong.

With economic development, the carbon emission intensity of Guangdong Province increased from 2000 to 2019. However, from 2016 to 2019, the growth rate of carbon emission intensity gradually tended to zero since the carbon emissions of energy consumption and industrial production in Guangdong Province accounted for a relatively high proportion, and the total carbon emissions were still large. However, the growth rate of carbon emission intensity has gradually flattened owing to improvements in the energy consumption structure, an increase in the proportion of clean energy consumption, and the acceleration of industrial upgrading [39]. According to the Statistical Yearbook of Guangdong Province [40], from 2000 to 2019, the proportion of raw coal in the province’s primary energy consumption decreased from 52.2% to 34.2%, and the proportion of natural gas increased from 0.2% to 8.7%. The share of primary power and other energy sources increased from 12.6% to 31.2%, and the proportion of clean energy consumption increased significantly. The global spatial autocorrelation characteristics of carbon emission intensity in the four regions of Guangdong Province were mainly affected by changes in carbon emission intensity. Before 2009, the carbon emission intensities of the four regions exhibited a relatively steady upward trend. However, since 2009, the carbon emission intensity of the Pearl River Delta region has increased but gradually flattened. The carbon emission intensity in eastern and western Guangdong maintained a certain growth trend, with the highest growth rate in eastern Guangdong, which began to exceed the provincial average. Moreover, the spatial autocorrelation of carbon emission intensity in the province showed a fluctuating and decreasing trend.

4.2. Influencing Mechanism of Economic Agglomeration on the Spatial Spillover Effect of Carbon Emission Intensity

The impact of economic agglomeration in Guangdong Province on the spatial spillover effect of carbon emission intensity affected the city and carbon emission intensity of neighboring cities, and the superposition effect of inter-city economic agglomeration further affected the carbon emission intensity of neighboring cities in the region. This is consistent with previous research results [41]. When the city is in a stage of rapid economic growth and its economic level is below 12.416 × 10⁸ yuan/km² (Table 1), increasing the proportion of added value of secondary and tertiary industries will continue to increase its carbon emission intensity, and the upward trend will be rapid and then slow down. At this stage of development, the migrant population continues to increase, the industry remains in a period of intensified agglomeration, a large number of various industrial parks and industrial parks are rapidly increasing to improve production capacity, and the use of resources and energy is increasing, resulting in a continuous increase in CO₂ emissions and carbon emissions [42,43]. Further, the energy utilization technology of cities has not yet made effective breakthroughs, and the positive externalities, such as the scale effect and knowledge spillover brought about by population and industrial agglomeration, have not been well reflected, resulting in negative environmental externalities being greater than positive externalities. In the study, when Guangdong’s economic agglomeration level is greater than 1241.60 × 10⁶ yuan/km², its carbon emission intensity will gradually decrease as the city’s economic agglomeration level continues to increase, and the downward trend will be slow at first, and then accelerate. This is because when urban economic development reaches a certain stage, positive externalities, such as scale effects and knowledge spillover, begin to appear. Technological progress, improvement in energy efficiency, energy structure optimization, industrial structure upgrading, and spatial agglomeration are more conducive to the unified treatment of negative environmental externalities, resulting in a gradual decline in total carbon emissions and carbon emission intensity. The city will transition to a low-carbon economy [42]. In addition, carbon emissions from urban construction land are mainly contributed by industry [44]; hence, there is an obvious negative correlation between the proportion of secondary industries and carbon emission intensity. Improvements in industrial innovation will be conducive to reducing carbon emission intensity, and the scientific and technological innovation of industrial enterprises will upgrade the industry and effectively improve production efficiency and energy utilization, which is conducive to carbon emission reduction. With an increase in the city’s openness to the outside world, the carbon emission intensity gradually decreases. This is because cities with a high degree of openness to the outside world are more susceptible to the influence of advanced low-carbon technologies from abroad and conducive to reducing carbon emissions by optimizing technologies and adjusting industrial structure. The main reason for the negative impact of per capita GDP and carbon emission intensity is that cities with higher per capita GDP correspond to cities with higher economic development levels, developed economies, advanced technology, industrial structures dominated by tertiary industries, strict environmental policies, and low-carbon urban development. Thus, a sustainable development path between humans and the environment can be explored in these cities. A previous study conducted a comparative analysis of the spatial effects of energy efficiency in 99 countries around the world and revealed obvious spatial spillover effects in all of them [45]. In addition to the influence of economic factors, the quality of national policies is important for promoting energy efficiency, with the direct positive effect of policies on energy efficiency capable of overcoming insignificant indirect negative effects. Such changes have a spatial spillover effect and will affect surrounding countries. A study on the evolution pattern of the spatial network structure of CO₂ and the economy revealed that economically developed regions are usually in the core position and play a controlling and guiding role in the common development of other regions [46]. This finding is similar to that in the Pearl River Delta region, which represents in the core of the network and influences the development of east and northwest Guangdong.

4.3. Uncertainty Analysis

The main driver of China’s carbon emissions is economic output; however, these emissions are also affected by other factors, such as population-scale effects and the energy mix [45,47]. In addition, CO₂ is affected by natural factors, such as topography and atmospheric circulation. Its spatial spillover effect is evident and aggravates the complexity and uncertainty of the factors influencing carbon emissions.

A possible reason for the increase in the number of cities with high–high and low–low economic agglomerations and carbon emission intensities in Guangdong Province is that some industries in the Pearl River Delta region have gradually shifted to the east, west, and north of Guangdong, which has enhanced the spatial correlation of carbon emission intensity in some cities. However, some cities in the Pearl River Delta region have gradually shown a declining trend in carbon emission intensity. For example, Guangzhou in 2016 and Shenzhen, Zhuhai, Foshan, and Huizhou in 2018 had the maximum carbon emission intensities during the study period. However, further research is needed to determine whether the carbon intensity peak is reached. Further, the impact mechanism of carbon emission intensity in some cities remains unclear, for example, according to the local Moran index analysis of economic agglomeration and carbon emission intensity in Guangdong Province. In 2000 and 2005, the economic agglomeration levels and carbon emission intensity in Zhongshan showed high–high agglomeration, whereas in 2010, 2015, and 2019, the spatial agglomeration relationship became insignificant, indicating that the growth in economic agglomeration levels and carbon emission intensity in Zhongshan were not spatially compatible. At different stages of urbanization and economic development, the relationship between the economy and carbon emissions is significantly different. For example, in the early stages of urbanization, there is an obvious positive relationship between the two, with the amount and intensity of carbon emissions increasing with economic development. However, in the later stages of urbanization, economic development and CO₂ are decoupled, and the relationship between the two gradually separates from the impact relationship [44]. This relationship varies under the influence of cities and different policy measures in different countries [48].

5. Policy Recommendations

Based on an analysis of the impact of economic agglomeration on carbon emission intensity in Guangdong Province, the overall economic agglomeration levels and carbon emission intensity from 2000 to 2019 showed an increasing trend and obvious spatial differences. The Pearl River Delta region is also a core area of carbon emissions because it is the main core of the province and even the country’s economy. The spatial spillover effect of economic agglomeration on carbon emission intensity in the province was obvious. Therefore, based on the research results and mechanism analysis, targeted countermeasures and specific measures for carbon emission reduction are proposed to achieve peak CO₂ emission and carbon neutrality goals in Guangdong Province.

(1) Local governments must weigh the benefits of economic development against the disadvantages of carbon emission growth and propose low-carbon economic development goals. According to the fitting results of the spatial Durbin model, when the economic agglomeration level of a city is approximately 12.4160 × 10⁸ yuan/km², the carbon emission intensity level will reach a peak. According to the current situation and development trends (Figure 2), the economic agglomeration levels of each area and city in Guangdong Province present a certain distance from the “inflection point” of 12.4160 × 10⁸ yuan/km². Therefore, the government should clarify its position according to current planning documents, drive the entire Pearl River Delta region based on Guangzhou and Shenzhen, and drive the high-quality economic development of the eastern, western, and northern regions of Guangdong with the Pearl River Delta region. Such changes will improve the quality and growth rate, optimize the industrial structure, realize technological upgrades, and reveal the coordination relationship between economic development and the ecological environment. Moreover, the following represent important directions for low-carbon urban development in the future: optimize the allocation of land use, formulate feasible urban land-use planning and rational land-use policies, implement effective management policies and measures for urban forests, control the urbanization development rate, protect basic farmland, and restrict the non-agricultural transfer of agricultural land and expansion of construction land [39].

(2) Foreign trade and technological exchanges can be increased appropriately to ensure that the domestic economic circulation pattern is not subjected to major impacts. According to the fitting results of the spatial Durbin model to the control variables, the openness of cities to the outside world can have a negative impact on their own carbon emission intensity. Therefore, it is important to strengthen foreign trade and technology exchanges, such as actively holding international trade activities, including CIIE and CIFTIS. It is also important to improve domestic energy efficiency by promoting the transformation of energy structures by actively introducing and learning from foreign technological experience [49], reducing the carbon peak as much as possible, and successfully achieving the “dual carbon” and sustainable development goals.

(3) Special funds should be allocated to help enterprises improve energy efficiency and reasonably guide enterprises to invest in their research funds to improve energy utilization methods and energy efficiency ratios. The decomposition of the effect of the spatial econometric model on industrial innovation shows that industrial innovation is negatively correlated with carbon emission intensity. The increase in scientific research funds for industrial enterprises will help reduce carbon emission intensity. Therefore, enterprises and institutions, especially industrial enterprises, should be guided to invest more in optimizing their own energy efficiency ratios through policies such as financial subsidies and special tax incentives.

6. Conclusions

From 2000 to 2019, the overall economic agglomeration levels of Guangdong Province continued to rise from 0.0556 × 10⁸ to 0.3794 × 10⁸ yuan/km². The Pearl River Delta region represents the economic engine of the province and accounts for more than 80% of the province’s GDP; thus, its economic agglomeration level is obviously the highest and shows the fastest growth rate. However, the economic agglomeration level of the east and northwest of Guangdong Province is backward; thus, the gap between these regions and the Pearl River Delta continues to widen. Although the carbon emission intensity of the province continued to rise, the growth rate continued to decrease. Moreover, the overall change trend was similar to that of economic agglomeration levels, among which the carbon emission intensity level of the Pearl River Delta was the highest and that of eastern Guangdong showed the fastest increase. This indicates that the eastern Guangdong region still presented a relatively extensive economic development mode. In addition, its economic growth was highly dependent on energy.

Strong spatial autocorrelation was observed in the level of economic agglomeration and carbon emission intensity in Guangdong Province from 2000 to 2019, with obvious high–high and low–low agglomerations. Currently, industrial land use in Guangdong Province is primarily organized through clustered development within specialized parks, which leverages economies of scale and concentration to reduce unit costs. Additionally, these clusters offer advantages in centralized pollution treatment and enhanced energy efficiency. However, the spatial concentration of industrial land use also results in a higher density of carbon emissions.

Economic agglomeration has a nonlinear parabolic effect on carbon emission intensity, which peaks when the economic agglomeration level reaches 12.4160 × 10⁸ yuan/km². The decomposition results of the spatial spillover effect showed that carbon emission intensity was affected by variables such as the economic agglomeration levels of neighboring cities, among which the openness of neighboring cities can positively increase the carbon emission intensity. In addition, increases in the proportion of industrial output value, per capita GDP, and industrial innovation level of neighboring cities can reduce the carbon emission intensity level of the city.

To achieve peak CO₂ emissions and carbon neutrality as soon as possible, local governments should weigh the relationship between economic development and carbon emissions, improve the quality and growth rate of the social economy, optimize the industrial structure, strengthen foreign trade and technology exchanges, and allocate special funds to help enterprises improve their energy efficiency and achieve technological upgrades. Territorial spatial planning should be implemented in a unified manner across the province, and policies and measures such as optimizing the land-use structure, controlling the speed of urban development, and effectively protecting basic farmland represent important future directions for low-carbon urban development.