Analysis of the Social and Economic Factors Influencing PM2.5 Emissions at the City Level in China

Abstract

Respirable suspended particles (PM2.5) are one of the key components of haze, which not only causes a variety of lung, intestinal, and vascular diseases, but also affects cognitive levels. China is facing the challenge of severe PM2.5 concentrations, especially in urban areas with a high population density. Understanding the key factors that influence PM2.5 concentrations is fundamental for the adoption of targeted measures. Therefore, this study used the Logarithmic Mean Divisia Index (LMDI) method to identify the key factors influencing PM2.5 concentrations in 236 cities in northeastern, western, central, and eastern China. The findings were as follows. The emission intensity (EI) played an important suppressing role on PM2.5 concentrations in all cities from 2011–2020. The energy intensity (EnI) inhibited PM2.5 concentrations in 157 cities; the economic output (EO) stimulated PM2.5 concentrations in some less economically developed regions; and population (P) spurred PM2.5 concentrations in135 cities, mainly concentrated in developed eastern cities. This study provides a whole picture of the key factors influencing PM2.5 concentrations in Chinese cities, and the findings can act as the scientific basis and guidance for Chinese city authorities in formulating policies toward PM2.5 concentration reduction.

1. Introduction

Over the past three decades of reform and opening up, China’s development has made remarkable achievements across the world. However, environmental pollution problems caused by economic growth, such as ecological imbalance and threats to public health, are gradually becoming key constraints to sustainable economic growth and steady social development [1]. The Global Environment Outlook (VI) report, initiated by the United Nations Environment Programme (UNEP) and co-authored by scientists from more than 70 countries, showed that environmental damage and pollution problems of perceived causes account for a quarter of premature deaths and diseases worldwide [2]. Among them, air pollution problems caused 6 to 7 million premature deaths each year, making it a major environmental factor affecting human health. Fine particulate matter (PM2.5) is an important component of haze and it has been established that PM2.5 exposure not only causes damage to the lungs, intestines, blood vessels and metabolism [3,4,5,6,7,8], but also affects cognitive levels [9]. According to the World Health Organization’s recommendations [10], the annual average PM2.5 should not exceed 10 μg/m³. In 2020, only about 3% or fewer Chinese cities met this standard, a figure that was roughly the same as a decade ago. However, there is still a considerable margin between reality and the standard, and it is undeniable that China has made progress in the management of PM2.5 in the period 2011–2020, with a significant reduction in overall annual average PM2.5 concentrations (Figure 1).

Given the serious negative impacts of PM2.5 on human health, a growing volume of literature has explored the potential drivers of PM2.5 concentration growth from different perspectives. While many studies have revealed the mechanisms affecting PM2.5 concentrations from a natural science perspective [11], it must be clear that PM2.5 pollution is by no means only a natural phenomenon. Instead, is formed by the interaction between the natural climate and human socio-economic activities (e.g., population growth [12], economic development [13], energy consumption [14], green innovative technologies [15], urbanization levels [16], etc.). Therefore, analyzing the main drivers for altering PM2.5 concentrations from an economic and social perspective and making practical recommendations for policy development can help improve air quality and curb the harmful effects of haze on human health.

2. Literature Review

To date, there has been a large body of research in academia on the drivers of air pollution, with carbon emissions [17,18,19,20,21,22,23,24,25] accounting for the majority of studies, and some targeting other pollutants such as nitrogen oxides [26,27,28]. When it comes to PM2.5 concentrations, as mentioned before, the focus of research has fallen on meteorological science and the health effects over the years, while research on the socio-economic drivers that have far-reaching implications for PM2.5 prevention and control has only gained attention in recent years (

Table 1. Literature assessing the socio-economic driving factors of PM2.5 pollution and research methods.

Authors	Period	Regions	Methods
Xu et al. [13]	2006–2017	30 Chinese provinces	Logarithmic mean Divisia index (LMDI)
Wang et al. [29]	2015	All over China	Geographically weighted regression modeling
Mi et al. [30]	2015–2018	Middle Yellow River urban agglomerations	Geographically and temporally weighted regression
Zhang et al. [14]	2014–2016	152 Chinese cities	LMDI
Dong et al. [31]	2000–2014	30 Chinese provinces	LMDI
Fu et al. [32]	1998–2016	All over China	Geographically and temporally weighted regression
Ji et al. [16]	2001–2010	79 developing countries	Stochastic Impacts by Regression on Population, Affluence and Technology (STIRPAT)
Yan et al. [33]	2010–2016	273 Chinese cities	Panel quantile regression
Liu et al. [34]	1998–2015	287 Chinese cities	Geographically and temporally weighted regression
Luo et al. [12]	1999–2011	12 Chinese regions	STIRPAT
Wu et al. [35]	2016	13 Chinese cities	Geographical detector method
Xu et al. [36]	2001–2015	29 Chinese provinces	Panel quantile regression
Wang et al. [37]	2000–2014	G20 countries	Panel quantile regression
Liu et al. [38]	2009–2018	108 Chinese cities	STIRPAT
Gan et al. [39]	2000–2016	287 Chinese cities	Generalized three-stage least squares (GS3SLS) method
Sun et al. [40]	2006–2020	30 Chinese provinces	An expanded IDA–PDA model
Chen et al. [41]	2005–2015	Chinese industrial sectors	Refined Laspeyres index decomposition analysis (RLI)

).

The studies on the socio-economic drivers of PM2.5 can be divided into four categories according to the scope of the study, from large to small: national level, provincial level, prefectural level, and street level studies. At the national level, the ratio of renewable energy consumption, gross domestic product (GDP) per capita, carbon dioxide emissions per capita, urban population ratio, and fossil fuel consumption ratio are the main contributors to global PM2.5 pollution [32]. Income, urbanization, and service sector levels also contribute to PM2.5 [16]. Wang et al.’s [37] study of G20 countries revealed that democracy, political globalisation, and urbanisation influenced PM2.5 concentrations. An empirical study on industrial industries [42] found that the industrial development effect was the main reason for the increase in industrial PM2.5 emissions, while the decrease in industrial PM2.5 emissions was mainly due to the energy intensity effect, followed by the coal pollution intensity and energy mixing effects.

At the provincial level, a study by Sun et al. [40] of 30 Chinese provinces found that technological progress contributed the most to the reduction in the PM2.5 intensity in China, but the potential for emission reduction was lower in developed regions and higher in heavy industrial provinces. Specifically, there was a spatial heterogeneity in the energy technology gap, energy-oriented technology efficiency, energy-biased technology, and output-oriented technology gap [13]. In line with findings at the city level, at the province level, the emission intensity and energy intensity were PM2.5 contributors. To lower PM2.5 concentrations, investment size dominates, economic development, population, and increasing GDP per capita were the main positive drivers [31,43], and urbanization and the share of the secondary and tertiary sectors had an impact [12]. In addition, a study by Jia et al. [44] for coal consumption concluded that PM2.5 from coal consumption in 2060 will achieve the WHO air quality guidelines.

Refined to the urban scale, Wang et al. [45] used a geographically weighted regression model to analyze the joint influence of human activities and natural conditions on PM2.5 in Chinese cities. Yu et al. and Zhang et al. [14,46] found that the output scale effect and energy use effect were the main reasons for the increase in PM2.5 emissions, while the emission intensity effect and emission factor effect were the main reasons for the decrease in PM2.5 emissions. There was heterogeneity in the effects of several factors, such as the economic growth, population, urbanization process, urban form, industrialization, and scale of direct investment on PM2.5 concentrations in different cities [33,36,37,38,39,47,48,49]. In general, green technology innovation, increased R&D [35], improved management, and an optimized production scale can reduce PM2.5 emissions [15,50], while improving urban green spaces, parks, urban sanitation, and increasing investment in environmental facilities can also reduce local and neighboring PM2.5 concentrations [51]. If we consider the health risks of PM2.5, industrial structure is the determining factor [52], and the effects of the urbanization rate and energy consumption per unit of GDP are spatially heterogeneous [53]. Wang et al. [29] found that the spatial heterogeneity of the PM2.5 concentration levels was significantly higher in cities in eastern China than in western cities, and there was a strong correlation with urban floor area and patch density. Moreover, the main drivers of PM2.5 concentrations in cities in northwestern China were different from those in southeastern China [34]. In addition, Mi et al. [30] found that PM2.5 concentration changes in the middle reaches of the Yellow River urban agglomeration exhibited strong seasonality and obvious spatial aggregation characteristics, allowing for targeted regulatory policies for different regions.

Urban streets were the most microscopic scale in the current research. A study by Zeng et al. [54] on street-level PM2.5 concentrations in Shenzhen found that a denser street-level air pollutant monitoring network could improve the accuracy and robustness of spatial and temporal predictions for PM2.5 concentrations within the city. Zhou et al. [53], on the other hand, explored the effect of the built environment on street PM2.5 at different times of the day using a shorter time scale.

Summarizing the existing literature, we found research gaps in the past studies. Most of the literature explored the factors influencing PM2.5 concentrations from a general perspective but did not accurately reveal the individual influencing factors for each city. The management of PM2.5 is national in scope and each city has the responsibility to manage it, while it is essential to tailor PM2.5 management strategies to their own situations. Thus, our study analyzes the socio-economic factors of PM2.5 for 236 cities in China between 2011 and 2020.

The review of the literature above suggests that there was a positive correlation between appropriate policies and the effectiveness of PM2.5 control. This paper firstly analyzed the factors influencing PM2.5 concentrations on a city-by-city basis over long time scales and explored the mechanisms by which these factors act in different cities. These factors will provide a valid and targeted scientific basis for local governments to formulate policies.

The paper is organized as follows. The literature review is followed by Section 3, a brief description of the research methodology and data sources. Section 4 contains the discussion, which presents the results, and Section 5 provides the conclusions and policy recommendations.

3. Methods and Data

3.1. Decomposition Methods

Factor decomposition is one of the main analytical methods in the field of energy and environment. There are usually two types of decomposition analysis, SDA and IDA [55]. Compared with SDA, IDA originated from energy system analysis, which means that IDA is more closely related to energy system research. IDA is more popular among scholars because of its simplicity, flexibility, low data consumption, easy access, and easy production of time series. This study uses the LMDI method from the partitioned index algorithm, which allowed for complete factorization without leaving residuals during the analysis [56,57]. Although the method has the limitations of negative and zero values, the data in this study did not involve negative values and the probability of zero values occurring was small. For zero values, Ang et al. proposed two solutions, the SV method and the AL method [58], which could be chosen to solve the problem, according to practicality. In addition, Ang provided guidelines for implementing the LMDI decomposition method. Additive decomposition and multiplicative decomposition are the two forms of LMDI [57]. Compared to the latter, the additive decomposition is easier to use and interpret. In summary, the LMDI method in IDA based on Kaya’s constant equation was used in this study (

Table 2. Definitions of the variable symbols.

Symbol	Implication
PM	PM2.5 concentration, μg/m3
E	Energy consumption (electricity consumption of the whole society, billion kWh)
GDP	Gross national product, billion yuan
P	Population, million people
EI	Emission intensity (PM2.5 concentration/energy consumption)
EnI	Energy intensity (energy consumption/gross national product)
EO	Economic output (gross national product/population)
i	Different cities
CEI	Contribution rate of the emission intensity
CEnI	Contribution rate of the energy intensity
CEO	Contribution rate of the economic output
CP	Contribution rate of the population

).

The IDA model is now widely used in academia. Chontanawat et al. [18], Cansino et al. [23], and Ortega-Ruiz [24] used the LMDI approach to explore the drivers of CO₂ emissions in Thailand, Spain, and India, respectively. Quan et al. [19] used the LMDI decomposition model to analyze the factors influencing carbon emissions in China’s logistics industry in terms of five aspects (carbon emission factors, energy intensity, energy structure, economic level, and population size). They found that economic growth was the main factor increasing carbon emissions in the logistics industry, followed by the population size and energy structure. The energy intensity played a suppressive role in carbon emissions. Yang et al. [22] had similar findings in their study of carbon emissions across China. The LMDI method has also been widely used in studies targeting the drivers of carbon emissions in industries such as electricity [17,25], transportation [21], and construction [20]. In addition to carbon emissions, the LMDI method has been applied reasonably well in the analysis of drivers in several other areas (e.g., NOx [27,28], industrial energy consumption and emission reduction [59,60,61], municipal solid waste [62], power generation variance [63], etc.).

According to Ang et al. [56] and Zhang et al. [14], the PM2.5 concentration (PM) was decomposed as follows.

$$\mathrm{PM}=\frac{PM}{E} \times \frac{E}{GDP} \times \frac{GDP}{P}\times \mathrm{P}.$$

(1)

According to Equation (1), the factors affecting the change in the average PM2.5 concentration were decomposed into the emission intensity (EI), energy intensity (EnI), economic output (EO), and population size (P), and the above four factors together led to a change in the average PM2.5 concentrations in each place from 2011 to 2020.

The starting year of the study period was considered as the base period (t = 0), and the reasons for the change in PM2.5 concentrations in the i-th city after the study period (t = T) were examined. Referring to Equation (1), the effect of each factor can be expressed as follows.

ΔPM_i = PM_i,t PM_i,0 = ΔEI_i + ΔEnI_i + ΔEO_i + ΔP_i.

(2)

The formulae for calculating each impact factor are as follows.

$$\mathsf{\Delta }\mathrm{EI}{\mathrm{i}}_{ }=\frac{PMi,t-PMi,0}{lnPMi,t-lnPMi,0} \times \ln \left(\frac{EIi,t}{EIi,0}\right);$$

(3)

$$\mathsf{\Delta }\mathrm{EnI}{\mathrm{i}}_{ }=\frac{PMi,t-PMi,0}{lnPMi,t-lnPMi,0} \times \ln \left(\frac{EnIi,t}{EnIi,0}\right);$$

(4)

$$\mathsf{\Delta }\mathrm{EO}{\mathrm{i}}_{ }=\frac{PMi,t-PMi,0}{lnPMi,t-lnPMi,0} \times \ln \left(\frac{EOi,t}{EOi,0}\right);$$

(5)

$$\mathsf{\Delta }\mathrm{P}{\mathrm{i}}_{ }=\frac{PMi,t-PMi,0}{lnPMi,t-lnPMi,0} \times \ln \left(\frac{Pi,t}{Pi,0}\right).$$

(6)

In order to provide a clear indication of the degree of influence for each influencing factor on the change in PM2.5 concentrations, the contribution margin (Ci) was introduced as a measure indicator. The formula for calculating the contribution margin of each factor for a particular city/region is as follows.

$${\mathrm{C}}_{\mathrm{i}}=\frac{\mathsf{\Delta }\mathrm{E}\mathrm{I}\mathrm{i}}{\mathsf{\Delta }\mathrm{P}\mathrm{M}\mathrm{i}}+\frac{\mathsf{\Delta }\mathrm{E}\mathrm{n}\mathrm{I}\mathrm{i}}{\mathsf{\Delta }\mathrm{P}\mathrm{M}\mathrm{i}}+\frac{\mathsf{\Delta }\mathrm{E}\mathrm{O}\mathrm{i}}{\mathsf{\Delta }\mathrm{P}\mathrm{M}\mathrm{i}}+\frac{\mathsf{\Delta }\mathrm{P}\mathrm{i}}{\mathsf{\Delta }\mathrm{P}\mathrm{M}\mathrm{i}}+{\mathrm{C}}_{\mathrm{EIi}}+{\mathrm{C}}_{\mathrm{EnIi}}+{\mathrm{C}}_{\mathrm{EOi}}+{\mathrm{C}}_{\mathrm{Pi}}.$$

(7)

Similarly, the contribution of each influence factor can be calculated at a larger level (where I indicates the number of cities included in each sample at that study scale).

$$\mathrm{C}{\mathrm{x}}_{ }=\frac{{\sum }_{i=1}^{i=I}\mathsf{\Delta }\mathrm{E}\mathrm{I}\mathrm{i}}{{\sum }_{i=1}^{i=I}\mathsf{\Delta }\mathrm{P}\mathrm{M}\mathrm{i}}+\frac{{\sum }_{i=1}^{i=I}\mathsf{\Delta }\mathrm{E}\mathrm{n}\mathrm{I}\mathrm{i}}{{\sum }_{i=1}^{i=I}\mathsf{\Delta }\mathrm{P}\mathrm{M}\mathrm{i}}+\frac{{\sum }_{i=1}^{i=I}\mathsf{\Delta }\mathrm{E}\mathrm{O}\mathrm{i}}{{\sum }_{i=1}^{i=I}\mathsf{\Delta }\mathrm{P}\mathrm{M}\mathrm{i}}+\frac{{\sum }_{i=1}^{i=I}\mathsf{\Delta }\mathrm{P}\mathrm{i}}{{\sum }_{i=1}^{i=I}\mathsf{\Delta }\mathrm{P}\mathrm{M}\mathrm{i}}={\mathrm{C}}_{\mathrm{EIx}}+{\mathrm{C}}_{\mathrm{EnIx}}+{\mathrm{C}}_{\mathrm{EOx}}+{\mathrm{C}}_{\mathrm{Px}}$$

(8)

where C_x is the national level contribution, which can also be broken down into the contribution from the emissions intensity (C_EIx), energy intensity (C_EnIx), economic output (C_EOx), and population size (C_Px).

3.2. Data Sources

In this paper, prefectural cities were selected as the study area and 236 cities were retained for further analysis over the period 2011–2020 after excluding those with incomplete data. The PM2.5 concentration data in this study were collected from Aaron van Donkelaar et al. [64]. Based on the LMDI method, the factors affecting PM2.5 concentrations could be broken down into EI, EnI, EO, and P. The data on GDP (2011 as base year), electricity consumption, and population for the 236 individual cities were obtained from the statistical yearbooks of each region. Due to the large amount of data, Figure 1 shows the annual average PM2.5 concentrations for all the cities in the country in 2011 and 2020.

For a more visual representation of the calculated results, the data for the five variables (PM, EI, EnI, EO and P) for the study period are presented in Figure 2.

4. Results and Discussion

According to Equations (1) and (2), our research first decomposed the causes of the nationwide changes in PM2.5 concentrations into four factors: the emission intensity (EI), energy intensity (EnI), economic output (EO), and population effect (P). Figure 3 shows the PM2.5 concentrations and their influencing factors in China from 2014 to 2020. The changes in PM2.5 concentrations are presented in the form of a line graph, with the contribution values of the four decomposition factors represented in the bar graph, which together disclose the reasons for the variation in the line graph. From 2014 to 2019, China’s PM2.5 concentrations showed a decreasing trend each year on average, with a small rebound in 2020 of roughly the same level as in 2019.

As shown in Figure 4, the emission intensity (EI) was the primary inhibiting factor for PM2.5 concentrations in China from 2014 to 2020, indicating that China’s energy saving and emission reduction policies (more detailed information provided later in the paper) achieved significant results. It is worth noting that the inhibiting effect of the EI on PM2.5 concentrations has weakened year by year, or has reached a critical mass, and it is necessary to find a new policy force. The impact of the energy intensity (EnI) on PM2.5 concentrations was relatively less significant and had a greater potential for emission reduction. China’s total GDP and population size maintained an increase between 2014–2020. Therefore, the two factors, economic output (EO) and population size (P), played a part in raising PM2.5 concentrations each year. This finding was in line with popular perception. The increase in population inevitably caused an increase in the demand for housing, transportation, and electricity, which led to a direct increase in the occurrence of haze. The overly high population density in certain areas was also not conducive to the diffusion of atmospheric particulate substances, thus indirectly contributing to the severity of haze. China has still not phased out coal-based energy structure, and economic development needs to come at the cost of environmental pollution. However, it is encouraging to see that the contribution of the EO to the increase in PM2.5 concentrations has diminished year by year, indicating that China is gradually decoupling economic development from rising PM2.5 concentrations.

To provide stronger policy guidance, our research discussed in detail the factors that influenced PM2.5 concentrations in 236 cities across the country during 2011–2020, and grouped them into four regions according to the criteria of the National Bureau of Statistics [65], namely the eastern region, the central region, the western region, and the northeastern region. Similarly, we decomposed the variations in PM2.5 concentrations in the 236 cities into four factors: the emission intensity (EI), energy intensity (EnI), economic output (EO), and population size (P), and calculated the contribution of each factor separately. The calculation results are shown in Appendix B. Each of these factors will be analyzed in the following sections.

4.1. Emission Intensity (EI)

On a long time scale (2011–2020), the EI remained the largest contributor to the reduction in PM2.5 concentrations at the national level, with a contribution of −224% (Appendix A). PM2.5 became known to the general public in 2011 when the Ministry of Environmental Protection (now renamed the Ministry of Ecology and Environment) promulgated the Weight Method for the Determination of PM10 and PM2.5 in Ambient Air. This was the first time that the Chinese government fixed the measurement method for PM2.5 in the form of a policy, laying the foundation for subsequent monitoring and precise emission reduction. Up to now, China has formed a more comprehensive policy framework for the management of PM2.5 (

Table 3. A review of the existing effective policy documents related to PM2.5 emission reduction.

Date of Issue	Policy Documents
2011	Determination of PM10 and PM2.5 in Ambient Air Weight Method
2013	Technical Policy on Integrated Prevention and Control of Fine Particulate Matter Pollution in Ambient Air
	Technical Guidance on Source Analysis of Atmospheric Particulate Matter (Trial)
	Technical Specification for the Installation and Acceptance of Continuous Automatic Ambient Air Particulate Matter (PM10 and PM2.5) Monitoring Systems
	Technical Requirements and Test Methods for Continuous Automatic Ambient Air Particulate Matter (PM10 and PM2.5) Monitoring Systems
	Technical Requirements and Test Methods for Ambient Air Particulate Matter (PM10 and PM2.5) Samplers
	Technical Specification for Manual Monitoring Methods for Ambient Air Particulate Matter (PM10 and PM2.5) (Weight Method)
2014	Technical Guidance for the Development of Primary Source Emission Inventories of Atmospheric Respirable Particulates (Trial)
2018	Technical Specification for the Investigation of Ambient Air Particle Matter (PM2.5) Infiltration Coefficient in Civil Buildings
	Determination of Low Concentration Particulate Matter in Exhaust from Stationary Sources Weight Method
	Method for the Determination of Particulate Matter in Exhaust from Stationary Sources and Sampling of Gaseous Pollutants
	Technical Specification for Continuous Monitoring of Flue Gas (SO2, NOX, Particulate Matter) Emissions from Stationary Sources
	Technical Requirements and Test Methods for Continuous Monitoring Systems for Flue Gas (SO2, NOX, Particulate Matter) Emissions from Stationary Sources
2021	Technical Requirements and Test Methods for Continuous Automatic Ambient Air Particulate Matter (PM10 and PM2.5) Monitoring Systems
	Guidelines for Quality Assessment of Automated Ambient Air Particulate Matter (PM10, PM25) Monitoring
	Work Plan for the “One City, One Policy” Resident Follow-up Study on the Synergistic Prevention and Control of Fine Particulate Matter and Ozone Pollution

) and has scientifically optimized emission patterns in terms of the measurement methods and detection and prevention technologies.

The EI values of the 236 cities in this study were all negative, indicating that the emission intensities had varying degrees for reducing the average PM2.5 concentrations at the city level. Among them, Jiayuguan, Urumqi, Baise, and Hohhot had the most significant EI effect on reducing the average urban PM2.5 concentrations, reaching −3090.06%, −858.94%, −621.08%, and −617.03%, respectively (Appendix A). The commonality of the above cities was that GDP growth was not dominated by the secondary industry, and economic development was driven more by industries such as tourism and trade, with less reliance on fossil fuels. Jiayuguan’s tourism growth rate remained above 20% year round. However, since 2019, Jiayuguan’s industrial growth rate was several times higher than before, and the pillar industries were high-pollution and high-energy-consuming industries, such as chemical materials and chemical products manufacturing, non-metallic mineral products, and electricity production and supply. It is expected that the effect of the EI on the reduction in the average PM2.5 concentrations in Jiayuguan will weaken in the future. Similarly, Urumqi is an economic structure with tertiary industry as the mainstay, where the service sector and foreign trade maintained an optimistic growth rate during the study period, while the growth rate of above-scale industry was maintained at a low level, and even negative growth in individual years. In the future, Urumqi will focus on the development of new energy, new materials, intelligent equipment manufacturing, and digital economy-based high-tech industries, and EI will continue to play an important role in PM2.5 emission reduction. Hohhot is also a tertiary industry-dominated city, and due to its unique agricultural and animal husbandry advantages, the local industry mainly carried out the extension, supplementation, and enhancement of the industry chain. In the future, Hohhot will continue to complete its industrial transformation under the guidance of the concept of green and low-carbon development, and develop green agricultural savings processing, clean energy, new materials, and other industries in accordance with local conditions, which will not add pressure to the PM2.5 emission reduction work. Baise’s share of the secondary industry was the highest among the four cities and was roughly equal to that of the tertiary industry. The aluminum industry is the traditional leading industry in Baise, and this once highly polluting industry has now formed a whole industrial chain of bauxite–alumina–electrolytic aluminum–aluminum deep processing-recycled aluminum-supporting industries, which has moved towards high-level, green, and intelligent development. In the future, Baise will accelerate the adjustment of energy structure and promote industrial transformation through the construction of several renewable energy projects. Baise’s development experience provided heavy industrial cities with proven emission reduction ideas, namely, through the introduction of advanced technology and talent, complete industrial chain upgrading and support the creation of new energy and new materials industry to achieve high-quality development.

When the 236 cities studied were divided into geographical regions, the contribution of the EI to PM2.5 concentrations was ranked as follows (the plus and minus signs are only used as a means of distinguishing whether the effect was positive or negative): western region (−356.14%) > eastern region (−233.56%) > northeastern region (−223.41%) > central region (−194.35%). The EI in the western region had the greatest impact on PM2.5 concentrations. Although the level of economic development in the western region was not yet sufficient and the production technology was backward, the volume was smaller. Most of the cities in the western region focused on the development of tourism, trade, etc., so the mitigating effect of the EI on PM2.5 concentrations was far ahead. The weakest effect of the EI on reducing PM2.5 concentrations in the central region could be explained by the fact that the central cities had large-scale heavy industry development and relied mainly on fossil energy sources such as coal and oil. In this regard, the Chinese government should appropriately guide the exchange and cooperation between local governments, especially to strengthen the communication of new energy-saving and emission reduction technologies between cities in the east and west, so that economically backward cities do not destroy the environment in exchange for development. At the same time, the Chinese government should continue to limit the use of fossil energy sources, such as coal, and increase the promotion of new energy sources and investment in related technologies.

4.2. Energy Intensity (EnI)

The EnI is the ratio of electricity consumption of the whole society to GDP in the same region over a period of time. The smaller this value, the more efficient the energy utilization. At the national level, the energy intensity (EnI) was a positive factor for reducing PM2.5 concentrations, but its contribution was only −6.07% (Appendix A). The Chinese government has attempted to reduce the energy intensity through the promulgation of action guidelines for a number of industries However, this series of measures has not significantly reduced the energy intensity. Compared with developed countries, China’s energy efficiency remains at a relatively low level. Therefore, the Chinese government should continue to increase its investment in cleaner production technologies, actively promote technical exchanges with other countries, and limit the consumption of fossil fuels, such as coal, by raising taxes and forcing the development of renewable energy technologies.

Figure 5 shows the trend in China’s energy consumption elasticity coefficient over the period 2011–2020. The energy consumption elasticity coefficient refers to how many percentage points of energy consumption are needed to achieve one percentage point of GDP growth. Ideally, the elasticity coefficient of energy consumption was close to zero or even negative, which meant there was zero or negative growth in energy consumption while maintaining a certain economic growth rate; a goal that some developed countries have already achieved. As shown in Figure 6, China’s energy consumption elasticity index has climbed since 2016, reaching 1 in 2020, exceeding the level of a decade ago and suggesting a continued deterioration in energy efficiency. Figure 6 shows a comparison of the energy intensity (in AJ/billion yuan) in China and the world’s major developed countries from 2010 to 2020. Although China’s energy intensity in 2020 was 41.5% lower compared to 2010, there was still a large gap compared to developed countries, such as the United States, Japan, and Germany.

Of the 236 cities in this study, 81 cities (34.18%) had a positive EnI, indicating that the EnI in these cities contributed to the increase in PM2.5 concentrations during the study period. The city with the strongest contribution was Jiayuguan City in Gansu Province, with a contribution rate of 1394.72% (Appendix A). This city had a backward level of economic development, with a large growth rate of industry in recent years, and was dominated by highly polluting industries, with its production technology and equipment needing improvement, and inefficient energy use. In terms of the percentage of cities with a positive EnI, the western region was as high as 59.2%, followed by the northeast at 40.9%, while the eastern and central regions were not very different, at 28.2% and 19.8%, respectively. This result could be explained by the fact that the western and northeastern regions were underdeveloped economically, which vigorously developed high-energy-consuming industries to increase the GDP growth rate. Additionally, the economic development mode was rough, whereas most of the eastern and central regions completed the transformation from high-speed growth to high-quality growth, with more funds invested in energy-saving and emission reduction technology research and development, and equipment innovation. With a high percentage of renewable energy use, those cities gradually achieved a decoupling between environmental pollution and economic growth. Taking the new installations of wind power in 2021 as an example, the central and eastern regions accounted for approx. 61%, while the “Three North” regions (northwestern region, northeastern region, and northern China) accounted for only 39% (source: China Electric Power News). In this regard, the eastern and central regions should continue to invest in green technologies, and the government should work to break down technical barriers, strengthen exchanges and cooperation between the eastern and western regions, accelerate the promotion of advanced technologies, and reduce the energy intensity in less developed regions.

4.3. Economic Output (EO)

The economic output (EO) is expressed as GDP/population, namely GDP per capita. At the national level, the EO contributed 124.92% to PM2.5 concentrations, which was the most significant contributor to the elevated PM2.5 concentrations. This finding was in line with the general perception that GDP per capita increased significantly between 2011 and 2020, which implied an increase in the demand for consumer energy products. The data also showed that China has not yet completed the decoupling of economic development and PM2.5 emissions, since economic growth relies on increasing inputs of labor, scale investment, and other means of production. In the future, the development model should be gradually transformed from a crude to an intensive one to realize high-quality and green development.

At the city level, the EO contributed to the increase in PM2.5 concentrations in the 236 cities to varying degrees. Among them, the EO had a significant contributing effect on the increase in PM2.5 concentrations in Jiayuguan City and Qiqihar City, which were 1045.27% and 1020.81%, respectively (Appendix A). It further confirmed the previous statement that Jiayuguan City had a crude economic development method, so the growth of GDP needs to be at the expense of emitting a lot of PM2.5. Qiqihar City is located in Heilongjiang Province in northeastern China and is mainly based on agriculture and industry. In 2020, for example, Qiqihar’s three industrial structure was 31.7:22.4:45.9, with agriculture and industry dominating the economic development. Qiqihar should promote the structural reform of the agricultural supply side and industrial upgrading and continue to develop the advantages of agricultural production. In terms of industry, the government should promote the reform of state-owned enterprises, actively introduce strategic investors, eliminate backward production capacity by mergers and acquisitions, and introduce new capital to inject vitality and competitiveness into the old heavy industrial enterprises. The government should also build a scientific research and innovation platform that relies on large-scale enterprises, realize the introduction of advanced talents and high-level manufacturing technology, enhance its awareness of enterprise services, and simplify the approval procedures to create a healthy and favorable business environment.

The contribution of the EO to PM2.5 concentrations was ranked as Northeast (215.08%) > West (170.60%) > Central (128.85%) > East (119.51%) (Figure 7). This was consistent with the findings of Zhang et al. [14], where a poorer level of economic development led to a greater degree of influence of the EO on PM2.5 concentrations. The eastern region had a high level of economic development, and most cities had already realized high-quality economic development. The concept of sustainable development was deeply rooted in people’s minds, and the government’s attention and investment in environmental protection were more comprehensive while focusing on development, which resulted in a low energy consumption and a high use of renewable energy in the eastern region. Northeastern and western China had a lower level of economic development compared to eastern and central China, where the focus was on economic development, which posed a potential threat to the increase in PM2.5 concentrations. At the same time, overly rapid industrialization and urbanization have led to a sharp increase in the energy consumption demand, contributing to higher PM2.5 concentrations. In addition, governments and investors in economically weaker regions pay too much attention to short-term interests, and do not pay enough attention to or invest enough money in green technologies and environmental protection equipment, which directly leads to the fact that the relevant talents and investment returns are not as expected, discouraging the enthusiasm of the talents and funds. The lack of relevant technologies makes it even more difficult for that region to move forward on the road of green development, forming a vicious circle. Therefore, the government should take the initiative to break the deadlock, increase the introduction of advanced technology and capital investment according to local conditions, and at the same time strengthen the promotion of the concept of sustainable development to truly integrate green development into the existing industrial base.

It is worth noting that among the cities in the Northeast, the impact of the EO on PM2.5 concentrations in Shenyang was much lower than the average of the region. The reason for this was that Shenyang city makes use of its natural advantages of low temperatures, good geological conditions, and the talent power of research institutes, such as the Shenyang Branch of the Chinese Academy of Sciences, forming an industrial chain cluster with cloud computing, big data, and artificial intelligence as the main body, which form a basis for seizing the commanding heights of China’s artificial intelligence in the future. As the first batch of industrial transformation and upgrading demonstration zones, Shenyang city has vigorously created a good business environment to provide protection for the revitalization of enterprises. At the same time, it comprehensively promotes innovation and reform experiments to support the rapid development of industries using the power of science and technology and it carries out in-depth reforms of state-owned capital and state-owned enterprises, which revitalizes the vitality of local enterprises. On this basis, Shenyang city continues to expand its openness to the outside world and carry out in-depth regional cooperation. In 2018, the National Development and Reform Commission formally issued the Implementation Plan for Counterpart Cooperation between Beijing Municipality and Shenyang Municipality. The two cities of Beijing and Shenyang, guided by the government, carried out full-field, multi-level cooperation in a market-oriented manner, made use of to their respective advantages to promote the upstream and downstream integration of the industrial chain, and worked together to expand the market space for mutual benefit and win–win results. In addition, under the 150 preferential policies, the cumulative number of enterprises registered in Shenyang by 2020 exceeded 18,000, with a registered capital of 220 billion RMB. All these initiatives have become an important booster for the high-quality development of Shenyang’s economy in the post-industrial era. The experience of Shenyang’s transformation provides important insights into the high-quality development of other industrial cities in the Northeast.

4.4. Population (P)

In this study, the year-end resident population of the region was used as an indicator of the population size (P). As shown in Appendix A, the P was a positive factor contributing to the increase in PM2.5 concentrations at the China-wide level, but the impact was not significant, with a contribution rate of only 4.7%. This result could be explained by the increase in the population in China during the study period. The population increase contributed to air pollution in two ways. First, a larger population had a higher demand for transportation, electricity, and housing, which generated a higher demand for energy consumption and directly led to more severe haze. On the other hand, the excessive population density was not conducive to the diffusion of PM2.5, which weakened the self-purification ability of the natural environment and indirectly decreased the air quality.

The population factor P increased PM2.5 concentrations in 134 cities, approx. 57% of the sample in this study (Appendix A), due to the increase in the population in these cities during the study period. By region, the order of the percentage of cities with positive ΔP was east (82.4%) > west (49.0%) > center (45.7%) > northeast (18.2%), which was the opposite of the order of the average EO contribution. The livability in the eastern cities and the balanced and sufficient development also brought a lot of jobs and entrepreneurial opportunities, which created a siphon effect of talents, resulting in a situation where the population density of the eastern cities far exceeded that of the other regions. Based on these facts, city managers in the east and other regions should adopt different strategies. Eastern cities should strengthen the publicity of energy saving and environmental protection, raise the public’s environmental awareness, and advocate for a green and low-carbon social trend. Specifically, the government can control the growth of the energy demand by restricting the use and purchase of important energy-consuming products, such as fuel cars; implement a gradual settlement policy to direct some industries to neighboring cities to avoid a rapid increase in the city’s population; and urge local enterprises to raise their environmental awareness through a combination of “tax + compensation”. Managers of cities in other regions (especially in the Northeast) should take the initiative to seek opportunities for cooperation with developed cities in the eastern region, bring in capital and eliminate backward production capacities through mergers and acquisitions, and develop industries with special characteristics according to local conditions. At the same time, they should implement friendly settlement policies and improve the supporting infrastructure, such as hospitals, schools, and parks, in order to slow the rate of the brain drain.

5. Summary and Policy Implications

This paper analyzed the factors leading to changes in PM2.5 concentrations in 236 cities in China from 2011–2020 from a social and economic perspective. The main findings are as follows.

• The emission intensity (EI) is an important factor for reducing PM2.5 concentrations. The energy intensity (EnI) decreased in 66% of the cities, but contributed to higher PM2.5 concentrations in the other 34% of the cities. The economic output (EO) was the most important contributing factor, especially in less economically developed regions. The population size (P) showed a contributing effect in 57% of the cities, mainly in the eastern developed cities.
• The emission intensity was an important contributor for lowering PM2.5 concentrations, while the economic output was an important contributor for raising PM2.5 concentrations. In the western region, both the energy intensity and population size were factors that raised PM2.5 concentrations. In the central region, the energy intensity and emission intensity together contributed to lower PM2.5 concentrations, and the main raising factor was the economic output, with the population size playing a weaker role in raising PM2.5 concentrations. The northeastern region was the only region that the population size suppressed PM2.5 concentrations, but the percentage of cities in which the energy intensity contributed to PM2.5 was as high as 41%. In the East region, the energy intensity was positive in only 28% of cities, and the population size contributed to PM2.5 concentrations in 82% of cities.
• From a China-wide perspective, the emission intensity continued to be the most important factor for inhibiting PM2.5 concentrations, while the economic output was the most important factor for increasing PM2.5 concentrations. The energy intensity and population size had a slight inhibitory and promotional effect, respectively. Refined to the inter-annual changes, the inhibitory effect of the emission intensity on PM2.5 concentrations weakened year by year, and its abatement capacity was fully exploited. While GDP grew year by year, the facilitating effect of the economic output decreased year by year, suggesting that China is decoupling economic development and PM2.5 emissions.

The sample of this study involved a wide spatial scope and a long time span and was specifically analyzed for the economic and social factors of each city to provide valuable references for policy formulation. Based on the above findings, the author proposes policy recommendations in the following four main areas.

• The emission intensity (EI) was the main factor for reducing PM2.5 concentrations, but the attenuating effect was not significant in some cities. The Chinese government should continue to impose restrictions on coal consumption and promote the use of renewable energy and the popularization of related technologies.
• The role of the energy intensity (EnI) varied considerably between cities. The cities where the energy intensity is still a contributing factor should adjust their industrial structure, upgrade their industrial chains, invest more in cleaner production technologies and equipment, and improve their energy use efficiency.
• The economic output (EO), as an important contributing factor, is a subject that deserves to be taken seriously across the country. Among them, the economic output of the cities in the western and northeastern regions significantly contributed to PM2.5 emissions. While continuing to make economic development their top priority, these cities should strengthen exchanges and cooperation with developed cities, actively learn from development experiences, introduce advanced technologies, and move toward a path of high-quality development.
• The population size (P) increased PM2.5 concentrations mainly in the eastern region, especially in economically developed cities. In this regard, the administrators of eastern cities should plan urban expansion to avoid unreasonable growth in energy consumption demand due to rapid population expansion, and raise the environmental awareness of citizens and enterprises. The administrators of less developed cities in the western region should revitalize local characteristic industries according to local conditions, create more employment and entrepreneurial opportunities, improve urban supporting facilities, and implement friendly settlement policies to retain and attract more talent.
• The limitations of this study were that the analysis was restricted to Chinese cities, and the discussion did not take the physical environment characteristics into account due to concerns about the length of the article. In future studies, we will conduct a wider and deeper scope of research as the data availability permits.