This study explores the factors driving CO2 emissions related to energy use in Fujian Province from 2000 to 2019, with an emphasis on long-term trends, short-term fluctuations, and spatial disparities. Utilizing annual data on CO2 emissions and various influencing factors from multiple cities within Fujian Province, we examine the factors driving long-term changes in CO2 emissions. To analyze short-term emission trajectories, we employ a temporal decomposition model, while spatial decomposition techniques are used to assess the variability in emission drivers across 9 prefecture-level cities over different years. Our findings reveal an inverted U-shaped relationship between CO2 emissions and urbanization over the 20-year study period. Furthermore, short-term fluctuations indicate a gradual reduction in the impact of urbanization on the increase in CO2 emissions within the industrial, transportation, and household sectors in Fujian Province. Additionally, economic development, measured as per capita gross domestic product, is shown to significantly influence CO2 emissions. Efforts to reduce energy intensity, which refers to the amount of energy consumed per unit of economic output, in both the industrial and household sectors are identified as potential strategies for emission reduction. The variability in CO2 emissions among cities is primarily attributed to differences in energy intensity and population sizes. These insights are critical for formulating policies aimed at promoting low-carbon development, reducing carbon emissions, and enhancing sustainability throughout Fujian Province.
1. Introduction
The surge in industrialization in China and other parts of the world has precipitated a notable increase in energy consumption and CO2 emissions (Wang et al., 2016a; Wang et al., 2016c). The International Energy Agency (2019) estimated that global energy-related CO2 emissions will reach 33.6 billion tons in 2019, with China contributing 9.9 billion tons (29.5%), making it the paramount emitter of CO2 globally. This has highlighted the importance of addressing CO2 emissions within China. In response, China has committed to attaining a peak in CO2 emissions before 2030 and to strive for carbon neutrality by 2060, with various provinces targeting specific emission reduction objectives. For example, Fujian Province, situated along China’s southeastern coast, promulgated its 13th Five-Year Plan (FYP) aimed at curbing CO2 emissions and fostering energy conservation (Zheng, 2017). A specific goal was developed to realize a 40%–45% reduction in CO2 emissions per unit of economic output by 2020 relative to 2005 levels, with each city being allocated precise reduction benchmarks based on this metric. An assertive push toward low-carbon industrial advancement was initiated to encourage the upgrading of an industrial structure (Central Government of the People’s Republic of China, 2021). Consequently, carbon emission discussions have increased at both the national and regional levels.
In this study, Fujian Province was chosen as the study object as it is strategically located in a significant economic zone along the southeastern coast of China. Fujian was identified as an inaugural national zone for showcasing best practices for promoting a healthy ecological environment. The goal is to build a societal model characterized by efficient resource use, a robust ecological environment, and harmonious coexistence between humans and nature through targeted planning and implementation. Furthermore, several cities within the province, such as Xiamen, have been incorporated into China’s low-carbon pilot cities initiative (National Development and Reform Commission of People’s Republic of China, 2010). Given these designations, Fujian Province is poised to exemplify leadership in carbon emission reduction endeavors. However, the trajectory of rapid industrialization and urbanization in the province underscores an escalating tension between environmental stewardship and economic advancement. In 2019, the per capita gross domestic product (GDP) of Fujian Province increased to 107,139 yuan, ranking fifth among all Chinese provinces. Notably, manufacturing and construction emerged as the predominant sector, accounting for 48.6% of the economic activity (National Bureau of Statistics, 2020). For example, in the same year, Fujian was positioned 15th among all provinces in China with regard to the total energy consumption, and the city experienced an annual average energy consumption growth rate of approximately 4.5%, surpassing the national mean of approximately 3.3% from 2000 to 2019. This trend underscores the accelerated pace of energy consumption in Fujian compared to the national average. Alarmingly, in 2019, fossil fuels constituted 75.1% of the total primary energy production in the province, significantly eclipsing the proportion of renewable energy sources. This augmented energy consumption trajectory would produce severe environmental ramifications, with escalating CO2 emissions being a primary concern. Without judicious oversight, Fujian Province may encounter substantial hurdles in attaining its peak carbon emissions and carbon neutrality objectives.
There are notable divergences in industrial development levels between the distinct subregions of Fujian Province, which invariably lead to significant disparities in energy consumption. For example, in 2019, large-scale industrial enterprises defined as establishments with an annual revenue surpassing 20 million RMB in Quanzhou exhibited a fivefold higher energy consumption than their counterparts in Xiamen. This pronounced discrepancy in energy consumption between the two cities translates to significant spatial variances in CO2 emissions levels. However, when emission reduction strategies are designed at a broader regional scale, they often overlook the unique differences and underlying causes of CO2 emissions across various subregions. As a result, these broad regional policies may not be as effective as approaches that are specifically tailored to address the distinct characteristics of each subregion.
Therefore, it is imperative to rigorously investigate the driving factors of CO2 emissions and to understand the spatial variability of emissions in Fujian Province. This would greatly assist the government in devising a subsequent action plan tailored to the region’s distinct CO2 emissions landscape for effective emission reductions. Moreover, an in-depth examination of Fujian Province can furnish a valuable framework for strategizing research and formulating CO2 emission policies in other regions (e.g., Guangdong Province and Jiangsu Province) that, while relatively developed overall, possess internal developmental heterogeneity.
In light of the aforementioned objective, this study undertakes a holistic examination of the determinants that underpin CO2 emissions, encompassing long-term trends, short-term fluctuations, and spatial variations. Understanding long-term trends requires a meticulous analysis of relationships and impacts over time and are shaped by variables such as economic evolution, urbanization rates, industrial configurations, and technological progressions. Conversely, short-term fluctuations, such as those occurring over 5 years, in carbon emissions are typically driven by immediate factors such as sudden policy changes, rapid technological advancements, and short-lived economic events. It is crucial to acknowledge that other determinations, including policy shifts and technological advancements, may be affected by spatial variations. The study of spatial variations can explain the factors that lead to spatial differences in carbon emissions, such as geographical locales, urbanization schema, and industrial structures. To our knowledge, this study is the first endeavor to combine these 3 facets to understand the influence of CO2 emissions at the urban scale.
2. Literature review
In contemporary research on the determinants of CO2 emissions, there are 3 research directions. The first direction is to delve into the enduring long-term relationships between CO2 emissions and various socioeconomic parameters including economic growth, population dynamics, industrial composition, energy consumption, and technological advancements. Broadly, two fundamental types of associations have been posited between CO2 emissions and socioeconomic factors. The first entails a conjectured, nearly monotonic linear relationship with economic growth. The second constitutes an inverted U-shaped, curvilinear relationship: the Environmental Kuznets Curve (EKC) hypothesis, which was initially advanced by Grossman and Krueger (1995). According to this notion, economic growth can result in heightened environmental degradation. Environmental degradation tends to diminish with economic growth once a certain level of economic development is achieved, though alternative relationships, such as exponential associations, may also exist. As economies progress, several plausible rationales underlie the potential reduction in environmental harm. Economic growth often accompanies an increase in affluence, which, in turn, can heighten awareness of environmental challenges and augment the demand for sustainable solutions. Moreover, advanced stages of economic development typically result in the resources to invest in cleaner technologies and enforce rigorous environmental regulations. These advancements can substantially contribute to the mitigation of environmental damage. This conceptual backdrop underpins our exploration of the interplay between diverse socioeconomic variables and CO2 emissions using the theoretical framework of the EKC.
For example, Dong et al. (2018) scrutinized the dynamic causal relationship between per capita CO2 emissions and other environmental and economic variables in China using the EKC framework. Pata (2018) conducted an analysis of the long-term, dynamic linkages between CO2 emissions and socioeconomic factors to assess the validity of the EKC hypothesis in Turkey from 1971 to 2014. Juan et al. (2021) presented empirical evidence supporting the EKC hypothesis within the Association of Brazil, Russia, India, China, and South Africa (BRICS) economies (excluding Russia) from 1980 to 2018. The Stochastic Impacts by Regression on Population, Affluence, and Technology (STIRPAT) model, an evolution of the IPAT model first proposed by Dietz and Rosa (1997), has been commonly employed to corroborate the presence of the EKC hypothesis. This model provides a comprehensive theoretical and analytical framework that encompasses the influence of economic growth, population dynamics, technological progress, and urbanization on the environment (Wang et al., 2016a; Wang et al., 2016b; Wang et al., 2017). For example, Wang et al. (2017) applied the STIRPAT model to affirm the existence of the EKC hypothesis in relation to income/urbanization and CO2 emissions within the manufacturing sector in China. Nevertheless, while inquiries into the long-term correlation between socioeconomic variables and CO2 emissions can shed light on whether CO2 emissions decline as socioeconomic development reaches a particular stage, they generally fall short of providing an in-depth analysis of the immediate effects of socioeconomic shifts or policy implementations on CO2 emissions.
Hence, the second direction is the study of short-term influences on CO2 emissions. Decomposition analysis techniques, such as structural and index decomposition, are typically used to study the drivers of CO2 emissions in the short term (Dong et al., 2020; Wang et al., 2021). Structural decomposition analysis is primarily employed in national-scale research. In contrast, index decomposition analysis has been nearly universally adopted to analyze energy savings and CO2 emission reductions for one or multiple sectors in a region. The logarithmic mean Divisia index (LMDI) is one of the most popular index decomposition analysis methods (Ang, 2005; Cahill and Gallachóir, 2010) because of its many advantages, including the absence of decomposition residuals, the aggregation of subsector effects to the same value as the total effects, and the ability to decompose and process datasets containing zero and negative values (Ang, 2004). The LMDI model was proposed by Ang et al. (1998), and it has been extensively used in investigating temporal and spatial variations in energy consumption or CO2 emissions. The temporal LMDI model has been applied to examine the drivers of CO2 emission changes over time on a global scale (Xiao et al., 2019), as well as the country level (Gao et al., 2019; Chen and Lin, 2020; Quan et al., 2020), regional level (Lin and Du, 2014; Wang et al., 2018; Xue et al., 2019), city level (Li, et al., 2019a; Cui et al., 2020), and sector level (Geng et al., 2014; Zhu et al., 2017; Shigetomi et al., 2018; Quan et al., 2020; Zheng et al., 2020). Although previous analyses have comprehensively analyzed the driving factors in short-term change, they did not consider the spatial differences of the impact factors. Xiao et al. (2019) used the temporal LMDI model to analyze the differences and similarities between countries, provinces (Chen and Yang, 2015; Feng et al., 2020; Li et al., 2020), and cities (Zhu et al., 2017); however, the quantitative relationships between drivers could not be defined.
Therefore, the third perspective considers the spatial difference of the drivers of CO2 emissions. Ang et al. (2015) subsequently expanded the temporal LMDI method and proposed a spatial LMDI approach. This approach is static and only valid for the year of the research (Ang et al., 2016). The spatial LMDI has also been used to study CO2 emissions, mercury emissions (Li, et al., 2019b), and water intensities (Yao et al., 2019). Many studies have applied this approach to study spatial differences in the influencing factors of CO2 emissions between regions. Furthermore, researchers have conducted spatial decomposition studies of CO2 emissions at different scales, such as at the national (Roman-Colladoa and Morales-Carrion, 2018) and regional scales (Li et al., 2017; Chen et al., 2019), and for different sectors, such as the power sector (Liu et al., 2017) and the household sector (Shi et al., 2019). Decomposition analyses of the temporal and spatial dimensions of CO2 emissions are typically performed at the national and provincial scales. Moreover, the driving factors of CO2 emissions are commonly categorized into 4 groups: energy structure, economic scale, energy intensity, and population effects. In the context of rapid global urbanization, these have become essential factors that affect CO2 emissions. According to Wang et al. (2016a), for every 1% increase in the urban population, CO2 emissions will increase by 0.20% in Southeast Asian countries. Hence, urbanization should also be included in the study of impact factor indicators.
The comprehensive identification of affecting factors of CO2 emissions is a precondition for developing an action plan to realize carbon peaking and carbon neutrality targets. The influencing factors of CO2 emissions have been extensively studied in terms of long-term trends, short-term fluctuations, and spatial variations. However, very few of these studies have reached broad conclusions by considering all the aspects together, and this has led to incomplete research results. In addition, the spatial differences of driving factors have been studied at the national and provincial levels by other researchers. However, the variability at the city scale has not yet been systematically analyzed. The study of spatial differences in the driving factors of CO2 emissions at the city scale would enable local units to formulate effective regional emissions reduction policies. Hence, this topic requires further study.
In this study, we first calculate the CO2 emissions in 9 sub-provincial-level cities in Fujian Province. Second, we adopt a sub-provincial-level CO2 emissions dataset (2000–2019) covering all sectors to explore the long-term relationship between CO2 emissions and urbanization. Third, we use a temporal LMDI model to study the factors that affect CO2 emission changes in the short-term from the industrial sector, transportation sector, and household sector at the provincial level. A comparison between the multiple sectors offers robust proof of the driving factors and helps discern the difference in the factors across various sectors. Finally, we apply the spatial LMDI model to analyze the spatial dynamic of CO2 emissions of 9 prefecture-level cities from the industrial, transportation, and household sectors.
3. Methodology and data
3.1. Analysis framework
The aim of this study was to utilize a comprehensive conceptual analytical framework, which is depicted in Figure 1, to compare the influencing factors of carbon emissions across different temporal and spatial scales and various industries in Fujian Province. First, we analyzed the long-term relationships between population, energy intensity, economic scale, urbanization rate, and carbon emissions. Second, we examined the short-term fluctuations and spatial disparities in carbon emissions within different industries across Fujian Province and its cities, considering the effects of the energy structure, energy intensity, industrial composition, economic scale, urbanization, and population. Furthermore, we assessed the carbon emission reduction potentials of different industries and cities in Fujian Province in order to provide targeted policy recommendations for facilitating low-carbon development.
3.2. Model
3.2.1. Calculation of CO2 emissions
We used the guidelines of the Intergovernmental Panel on Climate Change (2006) to calculate the CO2 emissions from fossil fuel consumption.
where i represents different types of fossil fuels; j represents different sectors; Cij is the CO2 emissions from the combustion of fossil fuel i in sector j; Eij is the consumption of fossil fuel i in sector j; and EFi represents the emission factor.
3.2.2. STIRPAT model
The theoretical and analytical framework employed in this study utilizes the STIRPAT model to examine the EKC hypothesis, specifically exploring the relationship between CO2 emissions and urbanization.
where I represents the environmental change; P, A, and T are the population, affluence, and technology, respectively; a, b, c, and d denote the estimated coefficients of the explanatory variables; ∊ represents the random error; and the subscript i denotes the panel unit.1
In our analysis, we adopted the STIRPAT model as the theoretical and analytical framework, drawing inspiration from York et al. (2003). To expand this model, we introduced a quadratic term for urbanization, thus employing a multivariate polynomial regression model. This approach allowed us to examine the possible EKC relationship between urbanization and its environmental impact. The incorporation of a negative coefficient for the quadratic term suggests a non-linear relationship, indicating that with increasing urbanization, CO2 emissions may initially rise, but will eventually begin to decline. To ensure the meaningful interpretation of the estimated coefficients, we transformed all variables in the model into the natural logarithmic forms. This transformation allowed us to interpret the coefficients as the sensitivity of the dependent variable to changes in the independent variable and facilitated a more comprehensive analysis of the influence of urbanization on carbon emissions.
The improved STIRPAT models described by Equation 3 provided us with a robust framework to analyze and evaluate the impact of urbanization on carbon emissions.
where Cit is the CO2 emissions from city i in year t; P denotes the total population; A is the GDP per capita; Ei is the energy intensity, measuring the energy consumed per unit of GDP output; and UR is the urbanization rate, denoting the proportion of the urban populace relative to the total populace.
To overcome the limitations of traditional parametric modeling methods in identifying the distinct form of relationship between variables (Yatchew, 1998), we integrated a semi-parametric panel fixed-effects regression methodology into our study (Baltagi and Li, 2002). This method offers increased flexibility compared to purely parametric models and can mitigate potential issues of dimensionality. Our aim was to combine parametric and nonparametric methods to boost the precision and resilience of our models. We also applied parametric and semi-parametric panel fixed effects regression models to estimate Equations 3 and 4, primarily for the model’s urbanization variables. The regression results were used to determine whether there was an inverted U-shaped curve relationship between CO2 emissions and urbanization. The equation is as follows:
The function f () represents the semi-parametric component to be estimated. Since the urbanization variable is entered nonlinearly into the model, the functional form, f (), in the model is uncertain.
3.2.3. Decomposition model
We then utilized the LMDI method that incorporates the Kaya identity to conduct a comprehensive decomposition analysis of CO2 emissions. This method allowed us to quantitatively assess the contribution of each factor to the CO2 emissions and compare their relative influences. The Kaya identity provides a useful framework for understanding the drivers of CO2 emissions. By using this approach, we were able to assess the relative importance of different factors in driving CO2 emissions, thereby gaining valuable insights into emission dynamics.
(1) Kaya identity
According to the Kaya identity, we decomposed the driving factors of CO2 emissions into 7 factors:
where subscript i represents 1 of 9 cities in Fujian Province (i = 1, 2, 3, …9); j refers to the different sectors type (j = 1, 2, …7), where j = 1 corresponds to the primary sectors that include farming, forestry, animal husbandry, fishing, and water conservation, j = 2 is the industry sector, j = 3 is the construction sector, j = 4 is the transportation sector, j = 5 is the wholesale, retail trade, and hotel, and catering services sector, j = 6 is the household sector, and j = 7 is other sectors; k denotes the type of fossil fuels (k = 1, 2, …17), as detailed in Table 1; Cijk is the CO2 emissions with the unit of Mt (million tons); Eijk is the energy consumption with the unit of million tons of the coal equivalent (Mtce); Eij is the total energy consumption in sector j with the unit of Mtce; Gij is the out value of sector j measured in million Chinese yuan (CNY); Gi is the total GDP measured in million CNY; Uri is the urban population size measured in million people; Pi is the total population size with the unit of million people; Qi is the CO2 emissions coefficient; Es is the energy structure, Ei is the energy intensity, Is is the industrial structure; G is the economic development level; Ur is urbanization; and P is the population.
Types of fossil fuels
Order . | Energy Type . | Order . | Energy Type . | Order . | Energy Type . |
---|---|---|---|---|---|
1 | Raw coal | 7 | Other gas | 13 | Fuel oil |
2 | Cleaned coal | 8 | Other coking products | 14 | LPG |
3 | Other washed coal | 9 | Crude oil | 15 | Refinery gas |
4 | Briquettes | 10 | Gasoline | 16 | Other petroleum products |
5 | Coke | 11 | Kerosene | 17 | Natural gas |
6 | Coke oven gas | 12 | Diesel oil |
Order . | Energy Type . | Order . | Energy Type . | Order . | Energy Type . |
---|---|---|---|---|---|
1 | Raw coal | 7 | Other gas | 13 | Fuel oil |
2 | Cleaned coal | 8 | Other coking products | 14 | LPG |
3 | Other washed coal | 9 | Crude oil | 15 | Refinery gas |
4 | Briquettes | 10 | Gasoline | 16 | Other petroleum products |
5 | Coke | 11 | Kerosene | 17 | Natural gas |
6 | Coke oven gas | 12 | Diesel oil |
(2) Temporal LMDI model
According to the temporal LMDI model (Ang, 2005), we decomposed the factors that influence CO2 emission changes in region i from the base year 0 to the target year t into 7: the CO2 emissions coefficient effect (Qi), the fossil energy structure effect (Es), the energy intensity effect (Ei), the industrial structure effect (Is), the economic development level effect (G), the urbanization effect (Ur), and the population effect (P), as shown in Equations (A1–A8) in the supplementary material.
(3) Spatial LMDI model
Regarding the selection of the spatial exponential decomposition model, contrasting with the bilateral-regional and radial-regional models, the multiregional model uses the mean of all the research units during the study period as the reference region, thereby avoiding the complex computational burden of the bilateral-regional model and the subjectivity of the radial-regional model when selecting reference objects (Ang et al., 2015; Ang et al., 2016). For this reason, the multiregional model (Ang et al., 2016) was adopted in our study. The spatial LMDI model was applied to investigate spatial differences in the driving factors of CO2 emissions and is described as follows:
where the benchmark, u, is calculated using the average from the 9 cities in Fujian Province. The various influencing factors were calculated using the Equations (A9–A16) in the supplementary material.
(4) Relative contribution rate
We converted the absolute contribution rate of different effects to a relative contribution rate using the ratio of each effect to the sum of its absolute value (Equation 8). This approach facilitated a comparison and analysis of the time trend in the contribution of each effect.
where I(△Ck) represents the relative contribution rate of each effect; and ΔCk represents the value of each effect.
3.3. Data sources
The GDP figures for the various economic sectors and the population statistics for the 9 cities within Fujian Province were sourced from the Fujian statistical yearbooks from 2000 through 2019. Economic data were adjusted to constant 2,000 prices for consistency. To calculate the inventory of the energy consumption and carbon dioxide (CO2) emissions for the 9 cities in Fujian Province, we adhered to the methodology outlined in the work of Shan et al. (2017). Due to the unavailability of energy balance sheets within the statistical yearbooks of the 9 cities, we relied upon the provincial-level energy balance sheets to estimate the energy consumption at the municipal level. For ease of reference, Table 2 provides a list of the notations employed in this study.
Notations used in this article
Notation . | Definition . | Notation . | Definition . |
---|---|---|---|
ICEs | CO2 emissions from the industrial sector | FZ | Fuzhou |
TCEs | CO2 emissions from the transportation sector | LY | Longyan |
HCEs | CO2 emissions from the household sector | NP | Nangpin |
Qi | Carbon emissions coefficient effect | ND | Ningde |
Es | Fossil energy structure effect | PT | Putian |
Ei | Energy intensity effect | QZ | Quanzhou |
Is | Industrial structure effect | SM | Sanming |
G | Economic development level effect | XM | Xiamen |
Ur | Urbanization effect | ZZ | Zhangzhou |
P | Population effect |
Notation . | Definition . | Notation . | Definition . |
---|---|---|---|
ICEs | CO2 emissions from the industrial sector | FZ | Fuzhou |
TCEs | CO2 emissions from the transportation sector | LY | Longyan |
HCEs | CO2 emissions from the household sector | NP | Nangpin |
Qi | Carbon emissions coefficient effect | ND | Ningde |
Es | Fossil energy structure effect | PT | Putian |
Ei | Energy intensity effect | QZ | Quanzhou |
Is | Industrial structure effect | SM | Sanming |
G | Economic development level effect | XM | Xiamen |
Ur | Urbanization effect | ZZ | Zhangzhou |
P | Population effect |
4. Results and discussion
4.1. Long-term trends
4.1.1. Data and variable
We used an unbalanced panel data2 comprising data from 2000 to 2019 to examine the presence of an inverted U-shaped relationship between urbanization and CO2 emissions in this study. Descriptive statistics for all the primary variables are presented in Table 3. With the exception of urbanization, variables were converted into their natural logarithmic forms to facilitate analysis. Table 2 provides a succinct overview of the distribution and variability of the variables under consideration. This summary offers useful insights into their central tendencies, variability, and range, thereby enhancing our understanding of the variables’ characteristics and their potential associations with CO2 emissions.
Descriptive statistics of variables
Variable . | Definition . | Mean . | Std. Dev. . | Min . | Max . |
---|---|---|---|---|---|
lnC | Natural logarithm of CO2 emissions (kilogram) | 23.7 | 0.9 | 21.3 | 25.9 |
lnP | Natural logarithm total population | 8.3 | 0.4 | 7.6 | 9.1 |
lnEi | Natural logarithm of energy use (kce) per 1,000 GDP (constant 2,000 price) | 5.2 | 0.8 | 3.8 | 7.1 |
lnA | Natural logarithm (ln) of measured GDP per capita in thousands (at constant 2,000 price) | 3.3 | 0.7 | 1.9 | 4.9 |
UR | The share of the urban population in the total population (%) | 55.0 | 11.0 | 29.0 | 89.0 |
UR2 | The squared term of the urbanization rate | 0.3 | 0.1 | 0.1 | 0.8 |
Variable . | Definition . | Mean . | Std. Dev. . | Min . | Max . |
---|---|---|---|---|---|
lnC | Natural logarithm of CO2 emissions (kilogram) | 23.7 | 0.9 | 21.3 | 25.9 |
lnP | Natural logarithm total population | 8.3 | 0.4 | 7.6 | 9.1 |
lnEi | Natural logarithm of energy use (kce) per 1,000 GDP (constant 2,000 price) | 5.2 | 0.8 | 3.8 | 7.1 |
lnA | Natural logarithm (ln) of measured GDP per capita in thousands (at constant 2,000 price) | 3.3 | 0.7 | 1.9 | 4.9 |
UR | The share of the urban population in the total population (%) | 55.0 | 11.0 | 29.0 | 89.0 |
UR2 | The squared term of the urbanization rate | 0.3 | 0.1 | 0.1 | 0.8 |
4.1.2. Urbanization–CO2 emission models
Table 4 shows the outcomes of the urbanization–CO2 emissions models according to the EKC hypothesis. The parametric fixed effects regression model, detailed in the first column, incorporates “city dummies” and “year dummies” to mitigate city-specific heterogeneity and temporal variations. The analysis revealed a robust and positive correlation between CO2 emissions and the following key variables: population, energy intensity, and economic growth, each at a 1% significant level. A 1% uptick in these variables corresponded to increases in CO2 emissions of 0.989%, 0.867%, and 0.913%, respectively. Notably, while the urbanization variable and its quadratic term did not show statistical significance, their coefficient signs aligned with the EKC hypothesis, suggesting a potential inverted U-shaped relationship between urbanization and CO2 emissions.
Estimation results of the urbanization–CO2 emission modelsa
. | Parametric Model . | Semi-parametric Model . | ||
---|---|---|---|---|
Variables . | Coefficient (Std. Dev.) . | t-Statistic [P value] . | Coefficient (Std. Dev.) . | t-Statistic [P value] . |
Constant | 7.978b (0.378) | 21.1 [<0.001] | ||
lnP | 0.989b (0.032) | 31.2 [<0.001] | 1.050b (0.072) | 14.7 [<0.001] |
lnEi | 0.867b (0.020) | 42.4 [<0.001] | 0.843b (0.030) | 28.5 [<0.001] |
lnA | 0.913b (0.034) | 27.1 [<0.001] | 0.881b (0.029) | 30.8 [<0.001] |
UR | 0.326 (0.522) | 0.6 [0.550] | ||
UR2 | −0.151 (0.308) | −0.5 [0.638] | ||
City dummies | Yes | Yes | ||
Year dummies | No | No | ||
Adjusted R2 | 0.996 | 0.985 | ||
Obs. | 137 | 114 |
. | Parametric Model . | Semi-parametric Model . | ||
---|---|---|---|---|
Variables . | Coefficient (Std. Dev.) . | t-Statistic [P value] . | Coefficient (Std. Dev.) . | t-Statistic [P value] . |
Constant | 7.978b (0.378) | 21.1 [<0.001] | ||
lnP | 0.989b (0.032) | 31.2 [<0.001] | 1.050b (0.072) | 14.7 [<0.001] |
lnEi | 0.867b (0.020) | 42.4 [<0.001] | 0.843b (0.030) | 28.5 [<0.001] |
lnA | 0.913b (0.034) | 27.1 [<0.001] | 0.881b (0.029) | 30.8 [<0.001] |
UR | 0.326 (0.522) | 0.6 [0.550] | ||
UR2 | −0.151 (0.308) | −0.5 [0.638] | ||
City dummies | Yes | Yes | ||
Year dummies | No | No | ||
Adjusted R2 | 0.996 | 0.985 | ||
Obs. | 137 | 114 |
aCluster-robust standard errors are in parentheses.
bDenotes P value <0.01.
The second column focuses on the control variables, and we used a semi-parametric panel fixed effects model. This approach further substantiated the significant impact of population, the energy intensity, and the per capita GDP on CO2 emissions. Figure 2 shows a nonparametric fitting that illustrates the nuanced dynamics between urbanization and CO2 emissions. The adjusted CO2 data in Figure 2 isolate the specific impact of urbanization on emissions from other factors, clarifying its independent contribution. This result aligns with an inverse U-shaped curve, initially rising before plateauing, indicative of the EKC hypothesis. Furthermore, this trend implies an initial surge in CO2 emissions that is concurrent with early urbanization phases characterized by rapid, unregulated growth and increased consumption. However, upon reaching a critical threshold of urban development, a decline in emissions was observed. This was attributed to advancements in industrial processes, technological innovation, and urban environmental infrastructure, steering economic growth toward a sustainable, low-carbon pathway.
A partial fit of the relationship between CO2 emissions and urbanization. The curves represent the semi-parametric components of the fitted values of the urbanization–CO2 emissions relationship, and the shaded portion represents the 95% confidence band, where each point represents a panel unit.
A partial fit of the relationship between CO2 emissions and urbanization. The curves represent the semi-parametric components of the fitted values of the urbanization–CO2 emissions relationship, and the shaded portion represents the 95% confidence band, where each point represents a panel unit.
Our findings are consistent with the EKC hypothesis within the urbanization–CO2 emissions context, as shown by similar studies in Organization for Economic Co-operation and Development countries (Wang et al., 2015). The initial phase of urbanization, marked by unstructured growth and elevated consumption, contributed to a rise in CO2 emissions. Over time, as the urban areas developed and adopted modern industrial practices and enhanced environmental strategies, a shift toward sustainable, low-carbon development was observed, underpinning the inverse U-shaped relationship postulated by the EKC hypothesis.
4.2. Short-term fluctuations
4.2.1. CO2 emissions in Fujian Province
During the period from 2000 to 2019, which encompasses China’s 10th to 13th Five-Year Plans (FYPs), significant fluctuations in CO2 emissions were observed in Fujian Province. This study provides a detailed analysis of these changes, correlating them with the respective FYP timelines.
Figure 3 shows the sector-wise distribution of CO2 emissions in Fujian Province. Notably, there was a substantial increase in total emissions from 124.2 Mt in 2000 to 459.1 Mt in 2019, representing an average annual growth rate of 7.12%. This growth, however, was not uniform across the study period. The initial phase (2000–2005) had a moderate increase in emissions at an average yearly rate of 6.48%, followed by a rapid surge (10.65% annually) between 2005 and 2010. The period from 2010 to 2015 marked a deceleration in this trend, with the growth rate dropping to 5.21% per annum. The final phase (2015–2019) was characterized by variable growth, with an average annual rate of 6.0%.
Energy consumption structure (%) and total CO2 emissions from 7 sectors in Fujian Province (Million tons: Mt). The colored bars each represent the percentage of energy consumption by different sectors relative to the total annual consumption. The black line chart with blue dots indicates the trend in carbon emissions over time.
Energy consumption structure (%) and total CO2 emissions from 7 sectors in Fujian Province (Million tons: Mt). The colored bars each represent the percentage of energy consumption by different sectors relative to the total annual consumption. The black line chart with blue dots indicates the trend in carbon emissions over time.
The observed CO2 emission dynamics are reflective of the evolving economic landscape in Fujian Province that is influenced significantly by energy conservation and emission reduction initiatives. Despite these efforts, the trajectory of emissions underscores the need for more robust measures to effectively curtail emissions. The industrial sector remains the predominant contributor to the region’s CO2 emissions, followed by the transportation sector, which showed minimal fluctuations ranging between 3.01% and 8.38% of the total emissions from 2000 to 2019. The household sector emerged as the third-largest emitter, accounting for 5.56% of the total in 2019. The contributions of other sectors were marginal and are thus not a primary focus of this study.
Figure 4 presents an analysis of the energy intensity across the industrial, transportation, and household sectors for the same period. The energy intensity, defined as the energy consumption to economic output ratio, had significant sectoral differences. The industrial sector was characterized by the highest energy intensity, followed by the transportation sector. Notably, the industrial sector experienced a marked decrease in the energy intensity, averaging an annual decline of 5.27% from 2000 to 2019. In contrast, the household sector showed a more moderate decrease, with an average annual reduction rate of 3.19%. The transportation sector, however, exhibited an increase in the energy intensity, with an average annual growth rate of 3.15%. These variances in the energy intensity trends across sectors are crucial for understanding their respective impacts on CO2 emissions and form a key component of our analysis.
Changes in the energy intensities of the industrial, transportation, and household sectors. Figure (A) illustrates the trend of energy intensity in the industrial sector, depicted by the ratio of the energy consumption (tons of standard coal: tce) to economic output (Chinese Yuan: CNY) within the industrial sector over time. Figure (B) represents the trend of energy intensity in the transportation sector, indicated by the ratio of the energy consumption (tce) to economic output (CNY) within the transportation sector over time. Figure (C) depicts the trend of energy intensity in the household sector, characterized by the ratio of the energy consumption (tce) to GDP (CNY) within the household sector over time.
Changes in the energy intensities of the industrial, transportation, and household sectors. Figure (A) illustrates the trend of energy intensity in the industrial sector, depicted by the ratio of the energy consumption (tons of standard coal: tce) to economic output (Chinese Yuan: CNY) within the industrial sector over time. Figure (B) represents the trend of energy intensity in the transportation sector, indicated by the ratio of the energy consumption (tce) to economic output (CNY) within the transportation sector over time. Figure (C) depicts the trend of energy intensity in the household sector, characterized by the ratio of the energy consumption (tce) to GDP (CNY) within the household sector over time.
4.2.2. Temporal decomposition analysis of CO2 emissions
Figure 5 shows the results of the temporal decomposition analysis focused on discerning the factors that influence variances in CO2 emissions from the industrial sector (ICEs), CO2 emissions from the transportation sector, and CO2 emissions from the household sector (HCEs) from 2000–2019. The cumulative effect of these changes shows the annual fluctuations in carbon emissions. This analysis, grounded in Equations A1–A8 from the supplenemtary material, culminates in Table 5, which summarizes the relative impact rates of each factor.
Temporal decomposition of CO2 emissions in Fujian Province from 2000 to 2019. (A) Industry sector, (B) transportation sector, and (C) household sector. This figure illustrates the annual carbon emissions per sector, with gray bars indicating total emissions and adjacent colored bars indicating contributions from various factors. A positive value in a colored bar signifies an emission increase due to the corresponding factor, while a negative value denotes a reduction.
Temporal decomposition of CO2 emissions in Fujian Province from 2000 to 2019. (A) Industry sector, (B) transportation sector, and (C) household sector. This figure illustrates the annual carbon emissions per sector, with gray bars indicating total emissions and adjacent colored bars indicating contributions from various factors. A positive value in a colored bar signifies an emission increase due to the corresponding factor, while a negative value denotes a reduction.
Relative contribution rates (%) of each driver to changes in the CO2 emissions (Mt) of the different sectors (2000–2019)
Sector . | Period . | △C (Mt) . | Qi (%) . | Es (%) . | Ei (%) . | Is (%) . | G (%) . | Ur (%) . | P (%) . |
---|---|---|---|---|---|---|---|---|---|
Industry sector | 10th FYPs | 27.6 | 0.2 | 1.0 | −37.1 | 10.1 | 29.4 | 17.6 | 4.6 |
11th FYPs | 95.6 | 0.5 | 0.8 | −25.8 | 12.3 | 44.6 | 12.8 | 3.2 | |
12th FYPs | 78.9 | −0.4 | 0.1 | −33.4 | 8.1 | 38.5 | 11.1 | 8.3 | |
13th FYPs | 74.2 | −0.2 | 0.1 | −13.0 | −9.4 | 49.7 | 18.0 | 9.6 | |
2000–2019 | 276.4 | 0.1 | 1.2 | −30.1 | 9.4 | 38.9 | 14.6 | 5.9 | |
Transportation sector | 10th FYPs | 5.7 | 0.7 | 6.4 | 43.0 | 2.6 | 26.9 | 16.2 | 4.2 |
11th FYPs | 7.8 | 0.0 | 4.4 | −13.8 | 3.7 | 57.4 | 16.5 | 4.2 | |
12th FYPs | 1.5 | 0.0 | −1.2 | −30.6 | −14.0 | 36.1 | 10.4 | 7.8 | |
13th FYPs | 7.9 | 0.0 | 0.5 | 25.8 | −17.5 | 36.2 | 13.1 | 7.0 | |
2000–2019 | 24.2 | 0.6 | 14.1 | 7.2 | −6.2 | 47.6 | 17.3 | 7.0 | |
Household sector | 10th FYPs | 5.1 | 0.3 | 2.2 | 17.8 | 0.0 | 45.4 | 27.2 | 7.1 |
11th FYPs | 4.5 | 0.3 | −1.0 | −31.5 | 0.0 | 49.5 | 14.2 | 3.6 | |
12th FYPs | 3.1 | 0.0 | 2.4 | −40.6 | 0.0 | 37.9 | 10.9 | 8.2 | |
13th FYPs | 6.2 | 0.0 | 1.7 | −8.2 | 0.0 | 58.0 | 21.0 | 11.2 | |
2000–2019 | 18.9 | 0.4 | 6.6 | −21.3 | 0.0 | 47.1 | 17.6 | 7.1 |
Sector . | Period . | △C (Mt) . | Qi (%) . | Es (%) . | Ei (%) . | Is (%) . | G (%) . | Ur (%) . | P (%) . |
---|---|---|---|---|---|---|---|---|---|
Industry sector | 10th FYPs | 27.6 | 0.2 | 1.0 | −37.1 | 10.1 | 29.4 | 17.6 | 4.6 |
11th FYPs | 95.6 | 0.5 | 0.8 | −25.8 | 12.3 | 44.6 | 12.8 | 3.2 | |
12th FYPs | 78.9 | −0.4 | 0.1 | −33.4 | 8.1 | 38.5 | 11.1 | 8.3 | |
13th FYPs | 74.2 | −0.2 | 0.1 | −13.0 | −9.4 | 49.7 | 18.0 | 9.6 | |
2000–2019 | 276.4 | 0.1 | 1.2 | −30.1 | 9.4 | 38.9 | 14.6 | 5.9 | |
Transportation sector | 10th FYPs | 5.7 | 0.7 | 6.4 | 43.0 | 2.6 | 26.9 | 16.2 | 4.2 |
11th FYPs | 7.8 | 0.0 | 4.4 | −13.8 | 3.7 | 57.4 | 16.5 | 4.2 | |
12th FYPs | 1.5 | 0.0 | −1.2 | −30.6 | −14.0 | 36.1 | 10.4 | 7.8 | |
13th FYPs | 7.9 | 0.0 | 0.5 | 25.8 | −17.5 | 36.2 | 13.1 | 7.0 | |
2000–2019 | 24.2 | 0.6 | 14.1 | 7.2 | −6.2 | 47.6 | 17.3 | 7.0 | |
Household sector | 10th FYPs | 5.1 | 0.3 | 2.2 | 17.8 | 0.0 | 45.4 | 27.2 | 7.1 |
11th FYPs | 4.5 | 0.3 | −1.0 | −31.5 | 0.0 | 49.5 | 14.2 | 3.6 | |
12th FYPs | 3.1 | 0.0 | 2.4 | −40.6 | 0.0 | 37.9 | 10.9 | 8.2 | |
13th FYPs | 6.2 | 0.0 | 1.7 | −8.2 | 0.0 | 58.0 | 21.0 | 11.2 | |
2000–2019 | 18.9 | 0.4 | 6.6 | −21.3 | 0.0 | 47.1 | 17.6 | 7.1 |
This analysis offers important insights into the factors that influence CO2 emissions over time. It enables policymakers and researchers to comprehend the dynamics that affect emissions and devise strategies for effective mitigation. Significantly, the economic growth (G) of the industrial, transportation, and household sectors contributed 269 Mt, 17 Mt, and 15.5 Mt to CO2 emissions, respectively, with corresponding relative contribution rates of 38.9%, 47.6%, and 47.1%, respectively. These findings align with previous research (Li et al., 2017; Chang et al., 2019; Chen et al., 2019; Shi et al., 2019; Xue et al., 2019; Quan et al., 2020), which identified economic development level as the primary factor driving increased CO2 emissions in these sectors in Fujian Province.
Interestingly, the energy intensity effect (Ei) emerged as the chief mitigating factor for CO2 emissions, except in the transportation sector. This decline in the energy intensity was attributed to technological advancements and enhanced research and development efficiencies. These advancements, which are supported by studies such as Lin and Wang (2021) and Zhang et al. (2020), highlight the critical role of scientific and technological progress in curbing CO2 emissions. The urbanization rate (Ur) was identified as the second-largest factor increasing CO2 emissions. It is driven by infrastructure development, industrial growth, and population expansion. In addition, population (P) consistently showed a positive effect on emissions in all sectors and periods.
G exhibited short-term fluctuations characterized by an initial increase, subsequent decrease, and a final increase. Notably, a peak in the relative contribution rate of G was observed during the 11th FYP period (2005–2010) in Fujian Province, coinciding with significant shifts in its economic development. The average annual GDP growth rates in Fujian Province for the 10th (2000–2005), 11th (2005–2010), and 12th (2010–2015) FYPs were 10.0%, 14.7%, and 11.2%, respectively. However, the highest relative contribution rate of G was recorded during the 13th FYP (2015–2019), particularly in the industrial and household sectors, despite the mitigating influence of Ei. While Ei played a crucial role in restraining the growth of CO2 emissions from 2000 to 2019, its impact varied across different sectors and timeframes. For example, in the industrial sector, the relative contribution rate of Ei decreased from −33.4% to −13% between the 12th and 13th FYPs. This trend indicates that early technological advancements had a more pronounced effect on reducing CO2 emissions, but the rate of improvement in the energy intensity slowed during the later stages, highlighting the need for continued investment in technological innovation for effective CO2 emission reductions. These findings illustrate the complexities in managing the energy intensity and emphasize the continuous efforts required to promote sustainable development and decarbonization strategies in Fujian Province.
The relative contribution of Is in the industrial and household sectors in Fujian Province declined from 10.1% and 2.6% during the 10th FYP period to −9.4% and −17.5% in the 13th FYP, respectively. This trend suggests that industrial restructuring policies implemented during the 12th and 13th FYPs (Fujian Provincial People’s Government, 2016) have effectively contained the growth of CO2 emissions. These policies consisted of diverse measures that included the development of advanced manufacturing, the promotion of modern service industries, the acceleration of the internet economy, the establishment of a marine economy, and the enhancement of service sector capabilities. As a result, there was a shift from energy-intensive industries to more sustainable sectors, contributing to the mitigation of the growth of CO2 emissions in the region. However, it is notable that despite these efforts, the absolute growth of CO2 emissions has not been fully curtailed, as evidenced in Figure 3. The rise in CO2 emissions since 2016 can be attributed to factors such as economic growth, changes in energy consumption patterns, and other influences. While industrial restructuring policies have mitigated the growth of CO2 emissions, they have not led to an overall reduction in emissions. This fact underscores the need for ongoing enhancements in policy execution to achieve substantial CO2 emission reductions in Fujian Province.
In an analysis of the 3 sectors, the relative contribution rate of Es in the transportation sector surpassed those of the industrial and household sectors. Figure 6 illustrates the energy consumption, energy intensity, and energy structure in Fujian Province’s transportation sector from 2000 to 2009. The data show a gradual increase in the total energy consumption in the transportation sector, followed by a sharp rise from 2016 onward. The energy intensity pattern mirrored this trend. Additionally, the energy structure evolved from 3 categories of fossil fuel consumption in 2000 to 9 by 2009, indicating a growing complexity in the energy structure. These observations highlight the critical need for comprehensive strategies to address the challenges within the transportation sector. Key areas include enhancing the energy efficiency, exploring alternative cleaner energy sources, and prioritizing technological innovations. Promoting renewable energy utilization and developing policies to optimize the sector’s energy structure are imperative. Moreover, the positive impact of Ei on the transportation CO2 emissions (TCEs) from 2000 to 2019 emphasizes the importance of improving the energy efficiency in the transportation sector. Although progress has been made, there remains substantial scope for further advancements to ensure a sustainable and eco-friendly transportation system.
Energy consumption and energy structure of the transportation sector in Fujian Province from 2000 to 2009. Figure (A) shows the energy consumption (million tons of standard coal: Mtce) of the transportation sector over time, while Figure (B) depicts the energy structure of the transportation sector, with differently colored bars representing the proportion of various energy types relative to the total energy consumption for the respective year (%).
Energy consumption and energy structure of the transportation sector in Fujian Province from 2000 to 2009. Figure (A) shows the energy consumption (million tons of standard coal: Mtce) of the transportation sector over time, while Figure (B) depicts the energy structure of the transportation sector, with differently colored bars representing the proportion of various energy types relative to the total energy consumption for the respective year (%).
In the analysis of the domestic consumption sector, the output was measured in terms of the GDP without distinguishing between industrial structures. Consequently, the Is values in the household sector were uniformly zero.
4.3. Spatial variations
4.3.1. Spatial characteristics of CO2 emissions
This study examined the spatial dynamics of CO2 emissions across 9 prefecture-level cities in Fujian Province by analyzing data from the years 2000, 2005, 2010, 2015, and 2019. The CO2 emissions data are detailed in Table S1. A general upward trend in CO2 emissions was observed in most cities, with Longyan being the exception, where a decline was noted from 2010 to 2015. The average annual growth rates of CO2 emissions varied significantly across the cities, ranging from 15.09% in Ningde to 1.98% in Nangping. The spatial disparity in emissions was pronounced, exemplified by Quanzhou’s 2019 emissions of 178.7 Mt, which starkly contrasted with Nanping’s 9.36 Mt. This necessitates a thorough analysis of the spatial variations in CO2 emissions to inform targeted emissions reduction strategies.
4.3.2. Spatial decomposition analysis of CO2 emissions
We employed the spatial LMDI model to investigate the spatial disparities in CO2 emissions and to assess the potential for emission reductions in various areas. The spatial decompositions of the industrial, transportation, and household CO2 emissions (ICEs, TCEs, and HCEs) for the 9 cities were calculated using Equations A9–A16, with detailed results shown in Tables S2, S3, and S4 (supplementary material). The regional average CO2 emissions were derived from the arithmetic mean of the emissions from the 9 cities of Fujian Province.
The spatial decomposition results, shown in Figure 7, indicated that there was increasing divergence over time for CO2 emissions and their driving factors. The primary contributor to spatial disparity in CO2 emissions was ICEs. Key factors that influenced these disparities included the economic development level (G), urbanization rate (Ur), energy structure (Es), industrial structure (Is), and energy intensity (Ei). Of these, Ei and P (population growth) were significant in all 3 sectors, with G being notably influential in driving spatial differences in emissions.
Spatial decomposition of factors that influence CO2 emissions for the 9 prefecture-level cities in Fujian Province for 2000, 2005, 2010, 2015, and 2019. Figures (A), (B), and (C) represent the decomposition results of the spatial index of CO2 emissions (Mt) from the industrial, transportation, and household sector, respectively. Different colored bars denote the contribution values of various effects, where positive values indicate a factor’s contribution to carbon emissions surpassing the regional average, while negative values denote a contribution lower than the regional average.
Spatial decomposition of factors that influence CO2 emissions for the 9 prefecture-level cities in Fujian Province for 2000, 2005, 2010, 2015, and 2019. Figures (A), (B), and (C) represent the decomposition results of the spatial index of CO2 emissions (Mt) from the industrial, transportation, and household sector, respectively. Different colored bars denote the contribution values of various effects, where positive values indicate a factor’s contribution to carbon emissions surpassing the regional average, while negative values denote a contribution lower than the regional average.
In terms of driving increases in CO2 emissions, population growth (P) was particularly impactful in economically flourishing cities such as Quanzhou, Fuzhou, and Zhangzhou. Conversely, Ei emerged as a major restraining factor, with Xiamen and Fuzhou demonstrating higher energy efficiencies than other regions. The contributions of Ei in cities such as Quanzhou and Zhangzhou were primarily positive, indicating lower energy efficiencies than the regional average. Addressing these inefficiencies through advanced technology, energy-efficient practices, and policy interventions is crucial for enhancing energy utilization efficiency.
The analysis revealed that the level of economic development (G) is a critical determinant of CO2 emissions, with higher economic development correlating with increased emissions. This underscores the importance of integrating sustainable development approaches in economically advanced regions to mitigate environmental impacts.
Urbanization (Ur) exerted a more pronounced effect on TCEs and HCEs than ICEs, highlighting its significant role in the transportation and household sectors. Higher urbanization levels, which were seen in cities such as Quanzhou, Fuzhou, and Xiamen, led to increased mobility and living standards, consequently elevating CO2 emissions in these sectors.
The industrial structure (Is) had a notable negative impact on TCEs in cities such as Quanzhou and Zhangzhou, suggesting that adjustments in the industrial structure can effectively curb emissions in these areas. Finally, the influence of the energy structure (Es) on regional emissions disparities was relatively minor, and this was attributed to the minimal variation in energy structures across the regions.
4.3.3. Performance ranking
In this analysis, we systematically evaluated the impact of various regional drivers on ICEs and presented them in descending order of effect (see Table S5 in the supplemental materials). This ranking not only allows for the assessment of energy efficiencies but also highlights differences in economic development levels. Specifically, a higher Ei value correlated with a greater energy efficiency, whereas a lower G value indicated advanced economic development. Significantly, lower rankings in this context suggested a greater potential for emission reductions, contingent upon the implementation of appropriate strategies.
The data, as depicted in Figure 5 and Table 5, offer several critical insights for emission reductions that are of substantial relevance for policymaking. For example, the analysis of Quanzhou revealed a consistent decrease in the ranking of variables, such as △C, Es, Ei, G, and P, over the study period. This trend highlighted Quanzhou’s substantial emissions reduction potential, achievable by enhancing the energy efficiency and industrial restructuring. Consequently, local authorities should prioritize policies that boost the energy efficiencies across sectors, promote renewable energy adoption, and facilitate a shift to sustainable industrial practices.
In contrast, Sanming’s scenario presents a dichotomy: despite economic growth, the persistently low Ei ranking indicates suboptimal improvements in the energy efficiency relative to economic expansion. To redress this imbalance, Sanming’s administration needs to elevate the energy efficiency in various sectors, integrate energy conservation into building designs, advocate for energy-efficient industrial technologies, and develop sustainable transportation infrastructures.
In Xiamen, a contrasting dynamic was observed, where a lower ranking in Ur coupled with a higher ranking in △C indicated an inverse correlation between urbanization and CO2 emissions, supporting the EKC hypothesis. Accordingly, Xiamen’s policy efforts should be directed toward fostering sustainable urbanization. This includes promoting energy-efficient building designs, implementing effective transportation systems, and integrating renewable energy into the urban framework.
In summary, our findings elucidated the distinct challenges and opportunities for managing CO2 emissions across various cities in Fujian Province. Governments are advised to leverage these insights to craft policies and initiatives. Focus areas should include enhancing energy efficiencies, encouraging the widespread use of clean and renewable energy sources, promoting sustainable urban development, and establishing regulations and incentives for environmentally friendly practices.
5. Conclusions and policy implications
5.1. Conclusions
This investigation comprehensively examined the determinants of CO2 emissions in Fujian Province, emphasizing long-term trends, transient variances, and spatial disparities. Our findings offer useful guidance for crafting CO2 reduction strategies and may serve as a pertinent benchmark for similarly situated developed regions. By using the STIRPAT model within an unbalanced panel data framework from 2000 to 2019, we examined the long-term factors influencing CO2 emissions in Fujian Province. Additionally, temporal and spatial LMDI models facilitated an assessment of the short-term oscillations and regional differences in emissions across the industrial, transportation, and residential sectors. We categorized the impact factors into 7 distinct clusters: CO2 emission coefficients, fossil energy structure, energy intensity, industrial configuration, income, urbanization, and population dynamics. Our principal conclusions are as follows.
For long-term trends, population growth, energy intensity, and economic advancement markedly influenced CO2 emissions, exerting a positive effect. This indicates that Fujian Province’s current technological and economic milieu predominantly drives CO2 emissions upward. Furthermore, our results corroborated the EKC hypothesis in the context of urbanization and CO2 emissions, suggesting that maintaining an intermediate to high urbanization level may facilitate industrial modernization, ecological enhancement, and consequent CO2 mitigation.
In the realm of short-term fluctuations, economic growth emerged as the most potent factor that augmented CO2 emissions in the industrial, transportation, and residential sectors from 2000 to 2019. The level of economic development in Fujian Province has not yet reached a threshold that would counteract CO2 emissions, highlighting the necessity for sector-specific regulatory interventions. In contrast, the energy intensity effect played a pivotal role in reducing CO2 emission growth in most cities, with the exception of the transportation sector. The introduction of diverse policies has advanced technological capabilities in Fujian Province, effectively restraining CO2 emissions. Moreover, urbanization was identified as a significant contributor to increased CO2 emissions across all sectors, predominantly due to the province’s relatively nascent urbanization stage. Urban expansion, characterized by heightened infrastructural development and energy demands, has led to escalated emissions.
Our spatial analysis revealed escalating disparities in CO2 emissions across these regions over time. High-emission cities, such as Quanzhou, Fuzhou, and Zhangzhou, offer substantial scope for emission reductions compared to their counterparts. Variations in technological levels across the regions were instrumental in these disparities. Hence, targeted initiatives aimed at technological advancement in regions that exhibit positive energy intensity effects could diminish regional emission discrepancies and contribute to the province’s overall emission reduction.
In summary, this study accentuated the critical need for intensified CO2 emission reduction efforts in Fujian Province. Thoughtfully formulated policies and measures that consist of technological advancements, judicious urban development, and sector-specific regulations are imperative for steering the province toward its carbon emission reduction goals. These salient findings provide a valuable reference for policymakers and stakeholders, outlining a roadmap for devising comprehensive low-carbon development strategies and advancing sustainability endeavors.
5.2. Policy implications
To align with China’s objective of peak carbon emissions by 2030, national and regional strategies have been deployed. However, our analysis revealed that Fujian Province has lagged behind in meeting these benchmarks, and this was exacerbated by the growing spatial disparities in CO2 emissions. This underscores the need for amplified efforts toward emission reductions. Based on our results, we recommend specific strategies for a transition toward low-carbon development.
Most importantly, it is crucial for urban planners and policymakers to foster an advanced level of urbanization and guide intelligent urban growth. This entails the construction of energy-efficient buildings, the enhancement of public infrastructure and transportation systems, and stricter enforcement of regulations governing the use of energy-intensive devices.
In addition, leveraging technological innovation can significantly decelerate the rise in CO2 emissions. During the 14th FYP period, it is imperative for policymakers to devise practical initiatives for further augmenting energy efficiency. This includes the development of fuel-efficient transportation infrastructure (roads, aviation, and maritime), energy-saving buildings and lighting, and efficient cooking technologies. Attention should also be given to the regional diversity in energy usage, prioritizing cities with higher energy consumption, such as Quanzhou and Sanming, for technological advancements.
Finally, it is vital to adopt differentiated and collaborative policies across different regions to reduce emissions that balance efficiency and equity. In rapidly developing cities such as Nanping, Ningde, and Longyan, policies should focus on decreasing the energy intensity, particularly as these areas that are likely to experience increased energy demands concurrent with economic growth. The introduction of energy-efficient technologies in these regions is thus paramount. Additionally, more developed cities, such as Fuzhou and Xiamen, should offer financial or technical support to regions with high energy usage or those still developing. These cities should also lead by example in increasing the adoption of renewable energy sources, such as electric vehicles, setting a precedent for others to follow. Cities with heavy industrial bases, including Quanzhou, Zhangzhou, and Sanming, need to prioritize industrial modernization and foster the growth of service and high-tech industries while phasing out pollution-intensive sectors.
Data accessibility statement
The datasets used in this study were sourced from the official portals of the Statistics Bureaus across the 9 cities in Fujian Province. In addition, we used the Fujian Statistical Yearbook and China Energy Statistical Yearbook. The precise datasets and variables examined are delineated herein, accompanied by the respective years and tables from which they were acquired:
The output value of sector and population metrics across the 9 cities in Fujian Province were extracted from the Fujian Statistical Yearbooks. Specific data are presented in Table S16 and Table S17 of the supplementary materials.
Due to the absence of energy balance sheets within the city-specific statistical yearbooks, the energy consumption at the urban scale was inferred from the provincial energy balance sheet and industrial energy consumption across the 9 cities in Fujian Province. Energy consumption data for Fujian Province from 2000 to 2019 were obtained from the CEADs database (https://www.ceads.net.cn/).
The industrial energy consumption metrics for Fuzhou, Longyan, Nanping, Ningde, Putian, Quanzhou, Sanming, Xiamen, and Zhangzhou for the period from 2000 to 2019 were obtained from their respective official statistical portals, specifically in Tables S6, S8, S10, and S12.
The methodology proposed by Shan et al. (2017) was used to compute the energy consumption and CO2 emissions inventories for the aforementioned cities in Fujian Province (http://creativecommons.org/licenses/by/4.0/). The municipal estimates for the energy consumption and carbon emissions are provided in the supplementary material (Tables S6–S15).
All data used in this analysis are publicly accessible via the respective websites and were amassed in adherence to the guidelines and policies set forth by each Statistics Bureau.
Supplemental files
The supplemental files for this article can be found as follows:
Supplementary Material.docx
Acknowledgments
We acknowledge the use of the Grammarly software (Grammarly Inc., USA) to assist with the refinement of grammar and syntax in this manuscript. While Grammarly helped improve the clarity and readability of the text, all the scientific content, ideas, and conclusions presented are entirely the authors’ own.
Funding
This work was supported by the Key Projects of Natural Science Foundation of Fujian Province in China [Grant number 2021J02030], the Social Science Foundation of Fujian Province in China [Grant number FJ2021B042], and the Research Funds for the Fujian Province Public-interest Scientific Institution in China [Grant number 2021R1002004].
Competing interests
The authors declare no competing interests.
Author contributions
Methodology, writing—original draft preparation, formal analysis, and visualization: YH.
Writing—review and editing: WL.
Conceptualization, supervision, and writing—review and editing: YW.
Writing—original draft preparation, formal analysis, and visualization: LZ.
Data curation and writing—review and editing: YH.
Data collection and processing: HL.
Investigations and writing—review and editing: FW, WtL.
Writing—review and editing: RS.
Read and approved the final manuscript: All authors.
Notes
Panel data refers to the data collected by different observation units at different time points and describes the changes of multiple observation units over time. Panel unit denotes a subset or specific element that constitutes the panel data, such as the data of one or more observational units at a particular time point.
Unbalanced panel data means that the observed values of individuals and time in the panel data do not completely match, that is, the data of some individuals at certain time points are missing.
References
How to cite this article: Huang, Y, Lin, W, Wang, Y, Luo, H, Zhu, L, He, Y, Wang, F, Lai, W-t, Shi, R. 2024. Driving factors of CO2 emissions in southeast China: Comparative study of long-term trends, short-term fluctuations, and spatial variations. Elementa: Science of the Anthropocene 12(1). DOI: https://doi.org/10.1525/elementa.2023.00028
Domain Editor-in-Chief: Detlev Helmig, Boulder AIR LLC, Boulder, CO, USA
Associate Editor: Lori Bruhwiler, Global Monitoring Division, National Oceanic & Atmospheric Administration Earth System Research Laboratory, Boulder, CO, USA
Knowledge Domain: Atmospheric Science