High-Quality Development and Decoupling Economic Growth from Air Pollution: Evidence from Daily Electricity Consumption in Fujian

Abstract

In the context of growing challenges associated with pollution prevention and control, developing more efficient technologies and precise policy measures to address the bottleneck period is imperative. This study utilized daily electricity consumption data from nine prefecture-level cities in Fujian Province from January 2019 to June 2024 to develop a high-quality development index (HQDI) and empirically investigate how HQDI affects the decoupling of economic growth from pollution emissions. Results suggest that HQDI can significantly promote decoupling, with innovation, openness, and sharing playing positive roles, while brown industries’ development and the electricity capacity installation of small and micro-enterprises hinder these processes. Moreover, extreme high temperatures exert a significant negative impact on decoupling, whereas increased market concentration fosters decoupling. Policy recommendations include prioritizing innovation, green technologies, and energy efficiency (particularly for SMEs); addressing climate resilience; and expanding HQDI to include factors like digital technologies for sustainable growth in Fujian and similar regions.

1. Introduction

In recent years, the Chinese government has increased its regulation of environmental issues related to air, water, and soil because of the advancement of pollution control efforts. Despite the substantial progress achieved in pollution control during the 13th and 14th Five-Year Plans, the potential for further emission reductions has diminished, while the financial and operational costs of environmental governance have escalated, resulting in a bottleneck in environmental management. In light of these challenges, there is an urgent need to implement more efficient policies and technological support to address the increasingly complex environmental challenges that have emerged. Concurrently, the conventional developmental paradigm, typified by elevated pollution and energy consumption, is no longer a viable option. In response, the government has proposed a “high-quality development” strategy based on five key principles: innovation, greenness, coordination, openness, and sharing. Furthermore, policies such as the “14th Five-Year Plan for Ecological and Environmental Protection” and the “Action Plan for Carbon Peaking and Carbon Neutrality” have been introduced to promote the coordinated development of the economy and the environment.

Fujian, a pivotal province situated along China’s southeastern coast, is a prime example of China’s broader national policies aimed at promoting high-quality development and achieving the decoupling of economic growth from environmental degradation. In 2023, Fujian’s GDP reached 5.44 trillion yuan, constituting 4.31% of the national total. The province has undergone a significant transformation from a manufacturing-based economy to a more service-oriented structure, with the tertiary sector now accounting for 49.99%, followed by the secondary sector at 44.09%, and the primary sector at 5.92%. This structural shift aligns with China’s broader goal of transitioning to a more sustainable, consumption-driven economy. Fujian’s strategic positioning within the “Belt and Road Initiative” and the Maritime Silk Road has fostered the development of traditional industries, such as electronics and machinery, and emerging sectors like information technology and environmental protection. This economic evolution is directly influenced by national policies promoting innovation, technological advancement, and sustainable development. Furthermore, Fujian has consistently demonstrated its commitment to environmental sustainability, achieving a forest coverage rate of 65.12% for 45 consecutive years, aligning with national environmental goals.

Moreover, Fujian’s proactive approach to environmental sustainability reflects the central government’s push for cleaner, greener industrial practices. Notably, the province has implemented ultra-low emission measures, including the phase-out of 90% of small coal-fired boilers, to improve air quality. These efforts have had significant results, with 98.5% of days across nine cities classified as having good air quality. This is a direct manifestation of the high-quality development strategy promoted by national policies prioritizing a balance between economic growth and environmental protection. Through its combination of economic transition, technological innovation, and environmental commitment, Fujian serves as a representative case for the broader trends and challenges of high-quality development and decoupling in China. The province’s ability to harmonize economic and ecological goals makes it an ideal region for examining the implementation and impacts of national policies, thus providing valuable insights into the feasibility and challenges of achieving regional-level sustainable development.

Although it may be assumed that an increase in pollution emissions would impede high-quality development, studies have confirmed this negative effect [1,2]. However, it has also been demonstrated that at specific stages of development, increased pollution emissions reflect increased resource consumption, which may facilitate improvements in industrialization and production efficiency, thus promoting high-quality economic development [3]. Chen et al., 2024 [4] examined the impact of high-quality development on economic growth and carbon emissions in 26 cities of the Yangtze River Delta. The findings indicate that promoting economic growth in regions with medium- or high-quality economic development exhibits a nonlinear relationship. Furthermore, high-quality development plays a pivotal role in reducing environmental pollution and carbon emissions. Nevertheless, the study did not elucidate how high-quality development specifically affects the decoupling effect of economic growth from pollution emissions. The extant literature has primarily concentrated on the impact of factors such as technological advancement and green development on economic growth or pollution emissions [5,6]. Although these studies have elucidated the role of certain components of high-quality development, they have yet to comprehensively answer how high-quality development influences pollution emissions and economic growth. Furthermore, the impact of high-quality development on the decoupling effect of economic growth and pollution emissions as well as the specific mechanism of action have not been sufficiently explored.

Considering the prevailing context, the present study investigated the factors influencing the decoupling effect between economic growth and pollution emissions, focusing on nine prefecture-level cities in Fujian Province. To achieve this, this study first employed the input–output analysis method to calculate the direct electricity consumption coefficients of various industries using the input–output table of Fujian Province. In addition, it combined data on industrial electricity consumption and electricity prices from January 2019 to June 2024 to calculate the total regional output. Further, a Tapio decoupling model was employed to construct a decoupling index utilizing the pollution emission data from these cities. Subsequently, this paper proposes a high-quality development index (HQDI) for each city. This index was developed utilizing the entropy method, incorporating industrial electricity consumption and business expansion data. Finally, this study empirically investigated the impact of the HQDI and its dimensions on the decoupling effect, revealing the mechanisms through which HQDI influences pollution decoupling. The results show that most observations achieved strong or weak decoupling, indicating significant progress in environmental governance. The empirical results suggest that HQDI positively influences decoupling effects, with innovation, openness, and sharing driving progress, while brown industries and SMEs capacity installation hinder these efforts. Moreover, extremely high temperatures have a notably negative impact, while increased market concentration contributes positively to decoupling.

This study makes several significant contributions, as outlined below. First, to address the dearth of monthly industry data, this study employed an innovative methodology that combines input–output tables and big data on electricity to estimate regional output and the scale of the digital economy. On the one hand, the input–output table elucidates the technical and economic interconnections between sectors. Conversely, as a “barometer” of economic activity, the big data on electric power directly correlate with both economic development and environmental sustainability. These data can reflect economic dynamics with greater precision and in real time, thereby compensating for the limitations of traditional data and enhancing insights into the dynamics of economic structure changes.

Secondly, by incorporating electricity consumption data, we can capture not only traditional economic growth indicators but also measure the activity of small and micro-enterprises, the structure of industries, the development of high-tech sectors, and the scale of digital industrialization. These factors are not typically available through conventional macroeconomic indicators and offer a more comprehensive view of high-quality development and innovation. Thus, these factors enhance our ability to understand the decoupling relationship between economic growth and pollution, which may be overlooked when relying solely on traditional measures such as GDP.

Thirdly, given that most research focuses on traditional macroeconomic factors such as energy consumption intensity, industrial structure, and population size, this study explored the dynamic relationship between economic growth and the decoupling of pollution emissions by refining industry-level data. The monthly year-on-year growth of pollution emissions and GDP were calculated using daily electricity consumption data and the Tapio model, eliminating seasonality and economic cycle effects and accurately capturing the influence of changes in industrial structure on the decoupling effect.

Fourthly, given the representative nature of Fujian Province with respect to economic and environmental governance, this study investigated the impact of HQDI on the decoupling effect at the prefecture level. This analysis reveals how high-quality development promotes the coordination of economy and environment, providing theoretical support and empirical evidence for other regions to achieve sustainable economic development.

Finally, the methodology proposed in this study applies to regions outside China with similar characteristics if relevant data, such as regional electricity consumption and price, are available. The core input–output analysis method, supported by global databases like EORA, WIOD, and EXIOBASE, can be utilized to extend this approach to other regions, contributing to the broader literature.

The remainder of this study is organized as follows. Section 2 is the literature review, which provides an overview of the measurement of the decoupling effect, the measurement of HQDI, and the impact of HQDI on the decoupling effect and proposes the hypothesis. Section 3 presents the empirical research design, including data and variable definitions, the measurement of key variables (including regional GDP, Tapio decoupling index, and HQDI), the descriptive statistics, and the model specification. Section 4 presents the results and discussion, which analyzes the distribution of the decoupling status of nine prefecture-level cities and discusses the influence of HQDI and its sub-indicators on the decoupling effect. The final section summarizes the conclusions and offers an outlook on future research. The research framework of this study is shown in Figure 1.

2. Literature Review

2.1. Measurement of the Decoupling from Air Pollution

The relationship between economic growth and pollution emissions frequently manifests as a complex dynamic. Grossman and Krueger (1995) [7] initially proposed an inverted U-shaped relationship between economic growth and pollution emissions known as the environmental Kuznets curve (EKC). This theory has garnered support from numerous studies indicating that during the nascent stages of economic development, there is an increase in pollution emissions concomitant with economic growth. However, as the economy attains a certain level of development, pollution emissions gradually decline in conjunction with ongoing economic growth [8,9]. Furthermore, with technological progress, policy interventions, and industrial restructuring, economic growth and pollution emissions may achieve decoupling, i.e., the separation between economic growth and negative environmental impacts (e.g., resource consumption, pollution emissions, etc.). However, Lazăr et al., 2019 [10] identified a complex, nonlinear relationship between GDP and CO₂ emissions in Central and Eastern European countries, which can take various forms, including N-shaped, inverted N, U-shaped, inverted U, and monotonic or even with no statistical link. Mughal et al., 2022 [11] further confirmed the inverted-U-shaped environmental Kuznets curve (EKC) relationship between economic growth and CO₂ emissions in selected South Asian economies.

When studying the correlation between transportation CO₂ emissions and GDP in 15 EU countries, Tapio (2005) [12] employed the elasticity coefficients of carbon emissions and economic growth to construct the Tapio decoupling model, which includes three types of states: decoupling, coupling, and negative decoupling. According to the elastic relationship between GDP and carbon emissions, the decoupling state is further subdivided into eight types. When the elastic relationship between GDP and carbon emissions is coupled, if both grow positively, it is considered expansionary coupling; otherwise, it is implicit coupling. Conversely, when GDP and carbon emissions are decoupled, if GDP grows and carbon emissions decrease, it is strong decoupling. If both increase, it is weak decoupling; otherwise, it is implicit decoupling. If the elastic relationship indicates negative decoupling, in which GDP is in recession and carbon emissions increase, it is considered strong negative decoupling. If both increase, it is expansionary negative decoupling; otherwise, it is weak negative decoupling.

The Tapio decoupling model addresses the effect of base period selection on decoupling results and refines the delineation of decoupled states by introducing the elasticity coefficient [13]. Consequently, scholars have increasingly adopted the model to analyze the relationship between economic growth and energy consumption or pollution emissions. For instance, Naqvi and Zwickl (2017) [14] utilized the Tapio model to investigate the decoupling status between economic growth and pollution emissions in EU countries across. Similarly, Shuai et al., 2019 [15] employed the Tapio model to investigate 133 countries, focusing on the decoupling status of economic growth and total carbon emissions, carbon emission intensity, and positive per capita carbon emissions. Luo et al., 2021 [16] utilized the Tapio model to analyze the decoupling status of economic growth and resources and the environment in the Central Plains Urban Agglomeration in China.

Furthermore, several studies have investigated the factors influencing the decoupling effect using models such as the logarithmic mean Divisia index (LMDI) and structural decomposition analysis (SDA). For instance, Engo (2018) [17] utilized the Tapio and LMDI methods, based on an extended Kaya identity, to assess the decoupling of economic growth and CO₂ emissions in Cameroon from 1990 to 2015. Song et al., 2019 [18] developed a two-dimensional decoupling theory between economic development and CO₂ emissions and investigated the turning points of the impact of per capita GDP on carbon emissions in China and the United States. Yang et al., 2021 [19] combined the Tapio method and LMDI to investigate the differences in the decoupling index and explore the key drivers of decoupling and their contributions to the decoupling index across six continents and major regions during the period 2001–2017. Dong et al., 2021 [20] employed the Tapio model and the C-D-Kaya equation to investigate the factors influencing the emergence of new and original decoupling indicators. Khan and Majeed (2023) [21] analyzed Pakistan’s decoupling from 1980 to 2018, finding expensive negative decoupling (END) driven by carbon intensity and urbanization. Yuan et al., 2024 [22] adopted the Tapio model to examine the nexus between economic growth and resource–environmental pressure in 110 cities situated within the Yangtze River Economic Belt (YREB). Riveros and Shahbaz (2024) [23] applied the TAPIO model, Kaya identity, and LMDI to study Colombia’s decoupling from 1975 to 2021, focusing on economic structure and energy consumption. Jia et al., 2024 [24] also combined LMDI and the Tapio model to evaluate the carbon emissions and sinks in the Yellow River Basin. In recent years, China has demonstrated a pattern of either weak or strong decoupling of economic growth and pollution emissions across a range of industrial sectors [13,25].

The Tapio model effectively characterizes the relationship between economic growth and pollution emissions, particularly by categorizing different decoupling states. However, it is limited to state analysis and does not capture the causal relationship between economic growth and emissions. To overcome this, it should be combined with other models for a more comprehensive analysis. Additionally, the model is highly reliant on data accuracy and completeness. Many countries and regions face challenges related to incomplete or inaccurate data collection, which can impact the model’s reliability. Furthermore, the applicability of the Tapio model may vary depending on regional differences in data quality and availability, which can restrict its use in diverse contexts.

2.2. High-Quality Development and the Decoupling of Economic Growth from Air Pollution

In recent years, China has implemented a high-quality development strategy aimed at achieving sustainable economic growth by enhancing economic quality, optimizing industrial structures, and strengthening innovation capabilities. This strategy is underpinned by five core principles: innovation, green development, coordination, openness, and sharing, which serve as key dimensions for evaluating development quality. Existing studies employ diverse methodologies and indicators to assess high-quality development. Some use single indicators such as total factor productivity (TFP) or green total factor productivity (GTFP) [2,26], while others adopt multidimensional frameworks. For instance, Peng et al., 2021 [27] used factor analysis at the prefecture level, focusing on economic scale, structure, efficiency, and welfare. Pan et al., 2021 [28] constructed an HQDI incorporating five domains: economic growth, innovation efficiency, environmental impact, ecological services, and livelihoods. Luo et al., 2023 [29] assessed enterprise-level development using six dimensions: innovation, greenness, openness, sharing, efficiency, and risk prevention. Chen et al., 2024 [4] extended these principles by adding indicators related to growth momentum and deepened reforms, while Han and Cao (2024) [30] constructed a framework to the marine economy, focusing on innovation, coordination, green development, openness, and sharing. These studies contribute to developing robust evaluation systems for high-quality development, and the entropy approach has been employed to construct HQDI, providing robust tools for assessing regional economic development.

High-quality development impacts economic growth and pollution emissions in multiple ways. Innovation is a key driver, enhancing productivity by increasing total factor productivity (TFP) [31,32]. It boosts economic growth through improved competitiveness and structural transformations. Wang et al., 2020 [33] found that technological progress in China’s high-tech industries helps reduce embodied carbon emissions, particularly in the electrical and optical equipment sector. Chen et al., 2022 [5] argued that technological innovation promotes green technology adoption, driving economic growth and improving environmental performance. The digital economy, incorporating technologies like big data, AI, and IoT, plays an important role in both economic growth and environmental sustainability. It improves productivity by optimizing production processes and creating new economic activities. Ma and Zhu (2022) [34] demonstrated that the digital economy can enhance green development by improving the efficiency of resource use and facilitating the spread of green technologies across industries. Xu et al., 2022 [35] and Zhang and Ran (2023) [36] also revealed that the digital economy not only improves industrial structures but also contributes to pollution reduction. Moreover, the digital economy also emerges as a crucial driver of technological innovation, particularly in improving productivity [28] and contributing to reductions in carbon emissions and pollution [37].

Green development is another crucial aspect of high-quality development. It emphasizes environmental protection through the adoption of cleaner technologies and sustainable practices. Research consistently shows that green development contributes to reducing pollution emissions while promoting economic growth. For instance, Peng et al., 2021 [27] showed that green innovation, especially in industries such as renewable energy, produces positive spillover effects both spatially and across sectors, thereby improving overall environmental quality. Ren et al., 2022 [38] also highlighted that green investment boosts energy conservation and emission reduction, upgrading industrial structures and fostering technological innovation. Ma et al., 2022 [6] showed that green technology adoption reduces carbon emissions while fostering economic growth, aligning economic performance with environmental sustainability. On the contrary, the development of brown industries will increase pollution emissions and promote economic growth through economies of scale. At the same time, it will lead to an increase in pollution emissions because these industries may ignore environmental protection.

Coordinated regional development plays a vital role in sustainable development. Studies such as those by Xiao et al., 2022 [39] have revealed that coordinated regional development, such as that in the Beijing–Tianjin–Hebei area, significantly impacts pollution reduction. The coordinated efforts lead to more efficient resource distribution, better industrial structures, and the adoption of cleaner technologies, which help reduce emissions. Furthermore, trade openness and foreign direct investment (FDI) are found to have negative associations with carbon emissions, as they often facilitate the transfer of cleaner technologies and enhance institutional quality, further promoting environmental sustainability [40]. Sharing, by optimizing industrial structures and ensuring resource allocation, can also promote economic growth and pollution reduction.

While existing research offers valuable insights, several gaps remain. Few studies have comprehensively examined how technological innovation, digital economy, green development, and coordination interact synergistically to promote sustainable development. This highlights the need for further exploration of the mechanisms that underpin the relationship between high-quality development and decoupling. Based on the literature, we propose the following hypotheses to explore the role of high-quality development in decoupling economic growth from pollution emissions:

Hypothesis 1:

High-quality development contributes to achieving the decoupling effect.

Hypothesis 2:

All dimensions of high-quality development contribute to decoupling effect, while the development of brown industries may hinder the decoupling effect.

3. Empirical Research Design

3.1. Data and Variable Definitions

This study adopted a monthly frequency for analysis, with all monthly indicators derived from daily data. This approach mitigates the noise inherent in daily data, which can compromise the accuracy of estimation results and the effectiveness of identification. By utilizing monthly data, the analysis better captures overall trends, minimizes the impact of short-term fluctuations, and enhances the robustness and explanatory power of the model estimates. The dataset consists primarily of monthly regional production and pollution emission data from January 2019 to June 2024. Daily power consumption and electricity price data, used to calculate the output value of prefecture-level cities in Fujian Province, were sourced from the Marketing Service Center of State Grid Fujian Electric Power Co., Ltd., Fuzhou, China, covering 133 sub-sector industries and 72 medium-category industries. The direct electricity consumption coefficients necessary for regional output calculations were derived from the Fujian Province input–output analysis based on the China Regional Input–Output Tables compiled by the Department of National Economic Accounting of the National Bureau of Statistics. These tables are developed using uniform techniques, with provincial statistical offices generating regional tables accordingly. Details on the calculation of prefecture-level city output values are provided in Section 3.2.1.

The World Air Quality Historical Database (WAQHD) serves as the primary source of air pollution emission data for prefecture-level cities. This comprehensive database provides air quality information for over 130 countries, encompassing more than 250,000 monitoring stations in 2000 cities. The WAQHD dataset encompasses a wide range of pollutants, including PM_2.5, PM₁₀, SO₂, NOx, CO, and O₃. For regions with minimal missing data, linear interpolation was used to address gaps, while in Zhangzhou, where data gaps are more significant, multiple linear regression was employed to estimate missing pollution emissions by regressing available data on emissions in other regions and local meteorological variables (e.g., maximum and minimum temperatures). The model-generated values were then used to fill out missing data.

The conversion of pollution emission values, originally expressed in units such as milligrams per cubic meter or parts per million, into pollutant-specific AQIs was conducted in accordance with the standards established by the United States Environmental Protection Agency (U.S. EPA). Monthly emissions are calculated by averaging the daily AQI values, thereby reflecting average monthly air quality. The primary focus of this study was the examination of the decoupling effect of economic growth and PM_2.5 emissions, focusing on robustness tests centered on SO₂ and PM₁₀ (refer to Section 3.2.2). The core explanatory variable, the regional high-quality development index (HQDI), is defined in Section 3.2.3. Control variables include monthly extreme high temperatures (number of days with maximum temperatures exceeding 30 °C), as such conditions increase electricity demand, particularly for cooling, leading to higher energy consumption and emissions without corresponding output increases. The original daily temperature data were sourced from the Tianqihoubao website (http://tianqihoubao.com/weather/province.aspx?id=350000 (accessed on 30 June 2024)). The impact of market structure on the decoupling relationship is also controlled by industry concentration, measured by the Herfindahl–Hirschman index (HHI), which is calculated based on the industry value added. Variable definitions are detailed in

Table 1. Variable definitions.

Type	Variable	Definition
Explained variable	Decoupling from PM2.5	A dummy variable equal to 1 if the decoupling state of economic growth from PM2.5 is SND and WD; equal to 0 otherwise.
Explanatory variable	High-quality development index	The logarithm of a composite index is calculated using the entropy method, including five major development concepts: innovation, green, coordination, openness, and sharing.
Controls	Extreme high temperature	Total number of days with a maximum temperature of 30 °C or higher per month.
	Herfindahl–Hirschman index	The sum of the squares of the output value of the prefecture-level city industry to the total regional output value; takes the logarithm form.
Others	Decoupling from SO2	A dummy variable equal to 1 if the decoupling state of economic growth from SO2 is SND and WD; equal to 0 otherwise.
	Decoupling from PM10	A dummy variable equal to 1 if the decoupling state of economic growth from PM10 is SND and WD; equal to 0 otherwise.
	Innovation	Regional innovation is measured from the two aspects of digital economic development (including digitization and industrial digitalization) and high-tech industry development; takes the logarithm form.
	Green	Regional green development is measured by the electricity consumption of green and brown industries; takes the logarithm form.
	Coordination	Regional coordinated development is measured based on the difference in electricity consumption between regions and industries; takes the logarithm form.
	Openness	Regional openness is measured by the electricity consumption of “transportation, warehousing and postal services”; takes the logarithm form.
	Sharing	Regional sharing level is measured by the market entity activity and the difference in urban and rural electricity consumption; takes the logarithm form.

.

3.2. Measurement of Key Variables

3.2.1. Regional GDP

This paper estimates the direct electricity consumption coefficients for each industry based on the input–output table of Fujian Province from previous years. The industry output value is then calculated by integrating electricity consumption data and electricity price data. The calculation process is as follows.

First is the calculation of the direct electricity consumption coefficient:

$$a_j={\displaystyle \frac{X_j}{{Value\_add}_j}}.$$

(1)

where $X_j$ is the direct consumption of industry $j$ to the electricity industry. ${Value\_add}_j$ is the added value of industry $j$, and $a_j$ is the direct consumption coefficient of industry $j$ to the electricity industry (calculated by value added), which reflects the direct consumption of the unit value added of industry $j$ to the electricity industry. In the absence of a regional input–output table for non-tabulated years, we apply linear interpolation to estimate missing coefficients between 2002 and 2017, following the approach of Pan et al., 2011 [41]. Given that AR models are highly effective in capturing temporal dependencies, we adopt this model to the historical coefficients from 2002 to 2017 to predict the coefficients for 2018–2024.

Next, the added value of the industry is estimated by combining data on the electricity consumption of the industry and electricity prices in the provinces. Since industrial output and electricity industry inputs are measured in monetary terms, electricity consumption (in kWh) is converted to monetary terms by multiplying it by the electricity price (yuan/kWh). Dividing these monetary inputs by the direct electricity consumption factor yields the industry’s added value:

$${Value\_add}_t^j={\displaystyle \frac{{Elec}_t^j}{a_t^j}}\times P.$$

(2)

where ${Elec}_t^j$ is the electricity consumption of industry $j$ in period $t$, $a_t^j$ is the direct electricity consumption coefficient of industry $j$ in period $t$, and $P$ is the average monthly selling price of electricity. Although the input–output table provides data for 42 industries, this paper extends the analysis to 72 medium-sized industries by assuming consistency in direct electricity consumption coefficients between sub-industries and larger industry groups. The total regional output ($Output$) is obtained by aggregating the output values of these 72 industries.

$${Output}_t={\sum }_{j=1}^{72}{Value\_add}_t^j.$$

(3)

In our sample period, global financial crises and the COVID-19 pandemic may have introduced potential biases in the data. To address this, we compare the GDP data from the National Bureau of Statistics of China with our estimated GDP for Fujian Province to evaluate the accuracy of our models. Both datasets are adjusted to constant prices, and the results are presented in Figure 2. As Figure 2 shows, the GDP shows a certain degree of seasonality, with GDP increasing slightly from the first to the third quarter, while the fourth quarter sees the highest level of fluctuations. The estimated GDP values follow a similar pattern, with both trends closely aligned. Moreover, the estimated GDP is slightly lower than the actual GDP, with the difference being less than 2%. This consistency further supports the validity of using electricity consumption and direct consumption coefficients to forecast industry value-added trends in Fujian Province.

3.2.2. Tapio Decoupling Index

Economic growth and pollution emissions often exhibit seasonal fluctuations, such as the Chinese New Year effect, increased winter heating pollution, and summer tourism-related economic activity. To mitigate these effects and ensure accurate estimates of monthly decoupling elasticity, this paper adopts a month-to-month analysis to smooth seasonal variations. Additionally, a year-on-year analysis is employed to capture long-term economic trends and reduce the impact of short-term fluctuations. The specific formula for the decoupling index is as follows:

$${\mathrm{D}\mathrm{E}\mathrm{C}}_{m,y}={\displaystyle \frac{\raisebox{1ex}{${∆Emission}_{m,y}$}\!\left/ \!\raisebox{-1ex}{${Emission}_{m,y-1}$}\right.}{\raisebox{1ex}{${∆Output}_{m,y}$}\!\left/ \!\raisebox{-1ex}{${Output}_{m,y-1}$}\right.}}={\displaystyle \frac{\raisebox{1ex}{${(Emission}_{m,y}-{Emission}_{m,y-1})$}\!\left/ \!\raisebox{-1ex}{${Emission}_{m,y-1}$}\right.}{\raisebox{1ex}{${(Output}_{m,y}-{Output}_{m,y-1})$}\!\left/ \!\raisebox{-1ex}{${Output}_{m,y-1}$}\right.}}.$$

(4)

Here, the subscripts $m$ and $y$ represent month and year, respectively. DEC is the decoupling elasticity coefficient, which represents the dynamic decoupling state between economic growth rate and air pollution. $Emission$ is pollution emission, including PM_2.5, SO₂, and PM₁₀. $Output$ is the total regional output. The eight states of decoupling effects are shown in

Table 2. Decoupling states of economic growth from air pollution.

Decoupling Type	Decoupling State	ΔEmission	ΔOutput	DEC
Decoupling	Strong decoupling	<0	>0	(−∞, 0)
	Weak decoupling	≥0	≥0	[0, 0.8)
	Recession decoupling	<0	<0	(1.2, +∞)
Negative decoupling	Weak negative decoupling	≤0	≤0	[0, 0.8)
	Strong negative decoupling	>0	<0	(−∞, 0)
	Expansive negative decoupling	>0	>0	(1.2, +∞)
Connection	Recession connection	<0	<0	[0.8, 1.2]
	Expansive connection	>0	>0	[0.8, 1.2]

, according to [12,20].

Under positive economic growth (i.e., $∆Output\ge 0$), strong and weak decoupling are considered optimal, indicating that economic growth coincides with reduced or minimally increased emissions, aligning with the goal of coordinated economic and environmental development. To facilitate empirical analysis, we define a dummy variable of decoupling from one pollution; for example, decoupling from PM_2.5 is set to 1 if the decoupling state of PM_2.5 is strong or weak decoupling, and decoupling from PM_2.5 is set to 0 for all other states. The same definition is applied to decoupling from SO₂ and decoupling from PM₁₀. This classification supports a clearer interpretation of the regression results when assessing the impact of high-quality development on decoupling.

3.2.3. High-Quality Development Index

This section constructs a regional HQDI based on electricity consumption data from key industries, SMEs expansions, urban and rural household electricity use, and public infrastructure sectors. The index is aligned with the five development principles—innovation, green growth, coordination, openness, and sharing—to analyze its impact on the decoupling of economic growth and pollution emissions.

Specifically, regional innovation is measured by digital economy development and high-tech industry growth. Green development is assessed using electricity consumption in green and brown industries. Coordination is measured through regional and industry-level electricity consumption disparities. Openness is evaluated using electricity consumption in transportation, warehousing, and postal services to construct a logistics index. Sharing is assessed from two perspectives: market activity and urban–rural electricity disparities. Market activity is captured by changes in SMEs expansions, reflecting electricity accessibility and economic vitality. Urban–rural differences in household electricity consumption and public infrastructure installation highlight disparities in energy access and living standards, underscoring the need for balanced development. Details of HQDI are presented in Appendix A,

Table A1. Content and definition of indicators of high-quality development system.

Connotation	Indicator	Definition of Indicator	Attribute
Innovation	Scale of industry digitalization	Calculated based on electricity consumption, direct electricity consumption coefficients, and the direct consumption coefficients of the digital economy.	+
	Scale of digital industrialization	Calculated based on electricity consumption in the digital economy sector and the direct electricity consumption coefficient.	+
	High-tech industry development	Electricity consumption of high-tech industries. Defined according to the classification of high-tech industries (manufacturing) (2017), which includes six major categories: pharmaceutical manufacturing, aerospace manufacturing, electronic and communication equipment manufacturing, computer and office equipment manufacturing, medical instruments and equipment manufacturing, and information chemicals manufacturing. This excludes other high-tech industries outside the digital economy sector.	+
Green	Green development	Monthly electricity consumption of green industries. Defined according to the green industry directory (2019) and electricity consumption data.	+
	Brown development	Monthly electricity consumption of brown industries. The six energy-intensive industries designated by the National Bureau of Statistics are defined as brown industries.	−
Coordination	Industrial coordination	Difference in the rate of industry electricity consumption growth. The standard deviation of the year-on-year growth rate of industry electricity consumption is divided by the average year-on-year growth rate.	−
	Regional coordination	Difference in regional electricity consumption growth rate. The standard deviation of the regional electricity consumption growth rate is divided by the average regional electricity consumption growth rate.	−
Openness	Logistics index	Electricity consumption in the transportation, warehousing, and postal services industry. Includes electricity consumption in several sub-sectors of the logistics industry: railway, road, water, air, and pipeline transportation as well as multimodal transportation, transport agency services, loading and unloading, warehousing, and postal services.	+
Sharing	Capacity expansions of SMEs	The number of capacity expansion enterprises with high-voltage non-residential capacity expansion, low-voltage non-residential new installations, high-voltage new installations, and high-voltage capacity expansion.	+
	Capacity reductions in SMEs	The number of capacity reduction enterprises with low-voltage account cancellations, high-voltage permanent capacity reductions, and high-voltage account cancellations.	−
	Residential electricity consumption difference	The standard deviation of per capita electricity consumption of urban and rural residents divided by the mean.	−
	Public infrastructure capacity expansion difference	The standard deviation of the net growth of capacity expansion projects in the public infrastructure sector (electricity connection projects) per capita in urban and rural areas divided by the mean. Net growth is defined as the number of enterprises with high-voltage and low-voltage capacity expansions minus the number of enterprises with capacity reductions.	−

Note: + and − represent that the attribute of indicator is positive and negative, respectively.

.

This paper follows Han and Cao (2024) [11] and Chen et al., 2024 [4] and utilizes the entropy method to construct an HQDI for prefecture-level cities, leveraging its advantage in assigning objective weights based on the characteristics of the data.

First, the indicators are standardized, and the standardization equations for positive (or negative) indicators are, respectively, as follows:

$$X_{nt}^{\prime }={\displaystyle \frac{X_{nt}-\mathrm{m}\mathrm{i}\mathrm{n}\left(X_{nt}\right)}{\mathrm{m}\mathrm{a}\mathrm{x}\left(X_{nt}\right)-\mathrm{m}\mathrm{i}\mathrm{n}\left(X_{nt}\right)}}\mathrm{}(\mathrm{o}\mathrm{r}\mathrm{}{\displaystyle \frac{\max \left(X_{nt}\right)-X_{nt}}{\mathrm{m}\mathrm{a}\mathrm{x}\left(X_{nt}\right)-\mathrm{m}\mathrm{i}\mathrm{n}\left(X_{nt}\right)}}),$$

(5)

where the subscripts $n$ and $t$ denote indicator and time, respectively. $X^{\prime }$ is the indicator normalized value, which lies between 0 and 1. Considering the seasonality of electricity consumption data, we use the Hodrick–Prescott filter to extract the long-term trend items for each indicator. Since there are only 54 observations in each prefecture-level city, the smooth parameter is set to 4. The $X^{\prime }$ below is based on these extracted trend items.

Next, the indicator weights are determined based on the information entropy of each indicator ($W_n$). The formula is as follows:

$$W_n={(1-e}_n)/{\sum }_{n=1}^N{(1-e}_n).$$

(6)

Among them, ${(1-e}_n)$ is the information entropy redundancy of the n-th indicator. $e_n$ is the information entropy of the n-th indicator ($e_n={\displaystyle \frac{-1}{ln\left(T\right)}}{\sum }_{t=1}^T({\omega }_{nt}\times ln({\omega }_{nt}\left)\right)$), where ${\omega }_{nt}$ is the weight of the n-th indicator in the t-th month, and ${\omega }_{nt}={\displaystyle \frac{X_{nt}^{\prime }}{{\sum }_{t=1}^T{X}_{nt}^{\prime }}}$.

Finally, the indicators are weighted and summed using the indicator weights and multiplied by 100 to obtain the HQDI, which is a value between 0 and 100.

$${\mathrm{H}\mathrm{Q}\mathrm{D}\mathrm{I}}_t={\sum }_{n=1}^N{W}_n\times X_{nt}^{\prime }\times 100$$

(7)

3.3. Descriptive Statistics

Table 3. Descriptive statistics of variables.

Variable	Observations	Mean	Standard Deviation	Min.	Median	Max.
Decoupling from PM2.5	486	0.523	0.500	0.000	1.000	1.000
High-quality development index	486	57.870	12.518	14.709	58.819	89.075
Logarithm of high-quality development index	486	4.029	0.256	2.688	4.074	4.489
Extreme high temperature	486	9.492	11.231	0.000	3.000	31.000
Herfindahl–Hirschman index	486	7.335	0.274	6.798	7.290	8.042
Decoupling from SO2	486	0.484	0.500	0.000	0.000	1.000
Decoupling from PM10	486	0.508	0.500	0.000	1.000	1.000
Innovation	486	2.327	1.463	−10.607	2.585	3.206
Green	486	2.148	0.911	−1.439	2.430	3.149
Coordination	486	2.790	0.417	−0.270	2.894	2.920
Openness	486	1.084	1.847	−11.759	1.437	2.057
Sharing	486	2.467	0.309	1.501	2.495	3.144

presents the descriptive statistics for all variables. The mean probability of achieving strong decoupling (SD) or weak decoupling (WD) between economic growth and pollutants (PM_2.5, SO₂, and PM₁₀) is approximately 0.52, indicating that over half of the observations during the sample period exhibited decoupling. This finding suggests that pollutant emissions are increasingly decoupled from economic growth, reflecting significant progress in environmental governance. The HQDI has a mean of 4.03 and a standard deviation of 0.26, implying relatively small regional differences and balanced performance in promoting high-quality development across regions. The average number of extreme high-temperature days is 9.49, with a standard deviation of 11.23, highlighting strong seasonal variability. The low standard deviation in regional industry concentration suggests relatively uniform market structures and competitive conditions across regions.

Among the HQDI components, while overall scores are similar, there is substantial variability in innovation and openness, with standard deviations of 1.46 and 1.85 and coefficients of variation of 0.58 and 1.59, respectively. This indicates significant regional disparities in these areas. In contrast, the indicators for coordinated and shared development show smaller variations, reflecting more balanced progress across regions.

3.4. Model Specification

To empirically analyze the impact of the high-quality development index on the decoupling effect between economic growth and pollution emissions, this paper constructs the following econometric model for testing.

$${DDEC}_{i,my}=\alpha +\beta \mathrm{L}\mathrm{n}{\mathrm{H}\mathrm{Q}\mathrm{D}\mathrm{I}}_{i,my}+\gamma X_{i,my}+{\lambda }_y+{\mu }_i+{\epsilon }_{i,my}.$$

(8)

where the subscripts $i$, $m$, and $y$ denote city, month, and year, respectively. The decoupling index ($DDEC$) is defined in the Section 3.2.2. The baseline regression analysis focuses on the decoupling relationship between economic growth and PM_2.5. Additionally, the impact of HQDI on the decoupling effect of economic growth from SO₂ and PM₁₀ is analyzed. The logarithm of the high-quality development index of prefecture-level cities is ($\mathrm{L}\mathrm{n}\mathrm{H}\mathrm{Q}\mathrm{D}\mathrm{I}$). $X$ is the control variables, including the city’s extremely high temperature and the Herfindahl–Hirschman index, which measures the concentration of city industries. $\lambda$ and $\mu$ represent year fixed effects and city fixed effects, respectively. $\epsilon$ is an error term capturing the effects of unobservable omitted variables on the city’s realization of the decoupling effect of economic growth and pollution emission. $\alpha$, $\beta$, and $\gamma$ are the regression coefficients. A positive $\beta$ indicates that high-quality development facilitates the decoupling effect between economic growth and pollution emissions; otherwise, it suppresses the decoupling effect. As the explanatory variables are calculated on a monthly year-on-year basis, the influence of seasonal fluctuations is eliminated, making it unnecessary to control for the monthly fixed effect and the interaction fixed effect between year and month in the model. The technical roadmap including step-by-step indicator construction and analysis methods is shown in Figure 3.

4. Results and Discussion

4.1. Decoupling State of Nine Cities

Figure 4 illustrates the distribution of the decoupling states across nine prefecture-level cities in Fujian Province from January 2020 to June 2024, highlighting significant regional disparities. Despite these differences, most cities exhibited strong decoupling (SD), with 174 out of 486 observations (35.80%), followed by weak decoupling (WD) with 80 observations (16.46%). These results suggest that most regions in Fujian have made notable progress in decoupling economic growth from pollution emissions.

Among the cities, Ningde, Zhangzhou, and Longyan showed the highest proportion of SD (24 observations each), reflecting a successful balance between economic growth and PM_2.5 control. In contrast, Xiamen and Quanzhou reported lower SD proportions (15 and 13, respectively), likely due to a higher dependence on pollution-intensive economic activities. For WD, Fuzhou and Zhangzhou led with 13 and 11 observations, respectively, while Xiamen had the highest proportion (13) of extended connectivity (END), reflecting varied environment–economy dynamics across development stages.

Other decoupling states also displayed regional variation. Fuzhou and Quanzhou showed a higher prevalence of weak negative decoupling (WND), suggesting more pronounced pollution increases alongside economic growth. Meanwhile, Sanming, Longyan, and Nanping exhibited higher rates of relative decoupling (RD) and strong negative decoupling (SND), indicating continued reliance on pollution-intensive economic activities. Overall, these findings highlight the diverse challenges and complexities faced by Fujian’s cities in achieving sustainable decoupling of economic growth from environmental degradation.

4.2. Impact of HQDI on the Decoupling of Economic Growth from Air Pollution

Table 4. Regression results of HQDI on decoupling from PM_2.5.

Variable	(1)	(2)	(3)
	Decoupling from PM2.5
High-quality development index	0.351 ***	0.243 **	0.258 ***
	(0.092)	(0.095)	(0.089)
Extreme high temperature		−0.006 **	−0.009 ***
		(0.002)	(0.003)
Herfindahl–Hirschman index			0.501 **
			(0.202)
Year fixed effects	YES	YES	YES
City fixed effects	YES	YES	YES
Observations	486	486	486
R2	0.032	0.325	0.333

Note: *** and ** represent 1% and 5% significant levels, respectively. All columns control for year fixed effects and city fixed effects. The values in parentheses represent robust standard errors.

presents the impact of HQDI on the decoupling of economic growth and PM_2.5 emissions. A stepwise regression approach is shown, beginning with HQDI alone in column (1) and progressively adding control variables such as extreme high-temperature days and the Herfindahl–Hirschman index (HHI) in columns (2) and (3). All models control for industry and year fixed effects. The results consistently demonstrate a significant positive effect of HQDI on the decoupling effect, even after the inclusion of controls. This supports Hypothesis 1 of this study that high-quality development contributes to achieving the decoupling effects.

In column (3), the benchmark results show that a 1% increase in HQDI increases the likelihood of achieving decoupling between economic growth and PM_2.5 by 26%, highlighting both statistical and economic significance. Specifically, compared with the average likelihood of decoupling of 0.52, this 26% increase equates to 50% higher odds of achieving decoupling, signaling a considerable shift toward more sustainable economic growth. For policymakers, these findings suggest that implementing strategies to enhance HQDI, such as investing in innovation and green industries, can significantly accelerate decoupling, resulting in more sustainable economic development. Industry stakeholders can note that aligning their business models with environmental sustainability goals contributes to reducing pollution and positions them to better meet future regulatory requirements and maintain long-term competitiveness.

The regression coefficients for the control variables reveal additional insights. Extreme high temperatures can negatively impact decoupling because they often lead to increased energy demand, particularly for cooling, which can strain energy systems and exacerbate environmental pollution. These weather extremes tend to heighten reliance on energy-intensive industries, especially in regions where cooling technologies are still largely dependent on non-renewable energy sources. This added strain makes it more challenging to decouple economic growth and pollution emissions, as it could increase energy consumption and pollution during peak temperature periods.

Conversely, the increased market concentration (HHI) can foster decoupling because larger, more concentrated industries often have more resources to invest in energy-efficient technologies and adopt environmentally friendly practices. In such industries, the potential for economies of scale may lead to more effective implementation of green technologies and stricter adherence to environmental regulations, thus reducing pollution intensity. The higher the market concentration, the more these firms can leverage innovations and technological advancements that contribute to decoupling economic growth from pollution.

4.3. Instrumental Variable Analysis of the Impact of HQDI on the Decoupling of Economic Growth from Air Pollution

As we found, HQDI can increase the likelihood that cities will achieve a decoupling state. However, it is necessary to discuss the validity of the adopted econometric model. The econometric model presented in this paper Equation (8) may be subject to two potential issues. First, due to difficulties obtaining specific monthly economic indicators for prefecture-level cities, such as energy consumption structure, industrial structure, and policy variables, both the HQDI and decoupling effects could be influenced simultaneously. As a result, endogeneity problems may arise from the omitted variables. To address this, we control for both year and city fixed effects, which helps mitigate the impact of omitted variable bias on our estimation results.

Second, while HQDI influences the decoupling effect, the decoupling effect itself captures an optimal state where economic growth occurs alongside reduced pollution emissions. This relationship may, in turn, affect HQDI, potentially leading to reverse causality and endogeneity concerns. To mitigate these concerns, we adopted instrumental variables, specifically the lagged term of the HQDI and the average HQDI of cities other than the focal city, as commonly practiced in the literature. The rationale for utilizing two instrumental variables is that when there is only one endogenous variable, the exogeneity of the instrumental variables can be tested using the Hansen J statistic. The results of the instrumental variable regression are presented in

Table 5. Results of instrumental variable regression.

Variable	(1)	(2)
	First Stage	Second Stage
	High-Quality Development Index	Decoupling from PM2.5
Lagged term of the high-quality development index	0.238 ***
	(0.052)
Average high-quality development index of cities other than the focal city	0.781 ***
	(0.094)
High-quality development index		0.479 ***
		(0.131)
Kleibergen–Paap rk Wald F statistic	102.229	-
Hansen J statistic	-	0.388
Controls	YES	YES
Year fixed effects	YES	YES
City fixed effects	YES	YES
Observations	477	477

Note: *** represents 1% significant level. All columns control for year fixed effects and city fixed effects. The values in parentheses represent robust standard errors.

.

The results in Table 5 suggest that the two instrumental variables significantly affect HQDI in the first stage at the 1% confidence level. Moreover, the Kleibergen–Paap rk Wald F statistic is 102.229, well above 10, indicating no issue of a weak instrumental variable. In the second stage, the p-value of the Hansen J statistic is 0.388, indicating that we cannot reject the null hypothesis of exogeneity for all instrumental variables, confirming their relevance and exogeneity. Regarding the regression coefficients, the impact of HQDI on decoupling from PM_2.5 remains significantly positive at the 1% level, confirming the robustness of our results. The HQDI coefficient of 0.479 is 1.86 times higher than the benchmark regression value of 0.258, suggesting that potential endogeneity may lead to an underestimation of HQDI’s effect on decoupling. This underestimation is within an acceptable range, further supporting the robustness of our findings.

4.4. Robustness Test of the Impact of HQDI on the Decoupling of Economic Growth from Air Pollution

4.4.1. Robustness Test: Explained Variable Replacement, Outlier Removal, Model Adjustment, and Control Variable Modification

To verify the robustness of the benchmark results, this study conducted several robustness tests, including replacing the explained variable, involving outlier removal, adjusting the econometric model, and modifying the control variables. First, we redefined the decoupling from SO₂ and decoupling from PM₁₀ to reveal whether the impact of HQDI on the decoupling effect from different pollutants still held. The results are shown in columns (1) and (2) of

Table 6. Results of robustness test: Explained variable replacement, outlier removal, model adjustment, and control variable modification.

Variable	(1)	(2)	(3)	(4)	(5)	(6)
	Replace Explained Variable		Delete Zhangzhou	Probit Model	Change Control Variable	Nonlinear Analysis
	Decoupling from SO2	Decoupling from PM10	Decoupling from PM2.5
High-quality development index	0.437 ***	0.415 ***	0.249 ***	0.958 ***	0.253 ***	−3.231 ***
	(0.096)	(0.097)	(0.092)	(0.362)	(0.090)	(1.070)
Extreme high temperature	−0.008 ***	−0.008 ***	−0.010 ***	−0.029 ***	−0.008 ***	−0.010 ***
	(0.003)	(0.003)	(0.003)	(0.008)	(0.002)	(0.003)
Herfindahl–Hirschman index	0.134	0.601 ***	0.543 **	1.578 **		0.427 **
	(0.211)	(0.202)	(0.211)	(0.688)		(0.195)
CR4					1.046 **
					(0.448)
Square of high-quality development index						0.471 ***
						(0.149)
Year fixed effects	YES	YES	YES	YES	YES	YES
City fixed effects	YES	YES	YES	YES	YES	YES
Observations	486	486	432	486	486	486
R2	0.285	0.319	0.361	-	0.331	0.344

Note: *** and ** represent 1% and 5% significant levels, respectively. All columns control for year fixed effects and city fixed effects. The values in parentheses represent robust standard errors.

. Second, to ensure the prediction method did not bias the results, Zhangzhou was excluded from the analysis due to significant gaps in its pollution emission data, which were estimated using data from other cities and local temperature information in the baseline analysis; the results are reported in column (3) of Table 6. Third, given that the explanatory variables are binary indicators of strong or weak decoupling between economic growth and pollution emissions, a probit model was adopted to assess whether the findings are sensitive to the choice of regression model. The results of this analysis are presented in column (4) of Table 6. Then, the measurement of industry concentration was replaced with the market share percentage of the top four industries to provide a more precise representation of regional industry concentration, as shown in column (5) of Table 6. Finally, we added nonlinear research into HQDI on the decoupling effect, as shown in column (6) of Table 6.

The results in the first two columns in Table 6 suggest that HQDI positively impacts the decoupling from SO₂ and PM₁₀ at the 1% confidence level, confirming the robustness of HQDI on the decoupling effect from different pollutants. Moreover, the magnitude of coefficients reveals that a 1% increase in HQDI can increase the probabilities of decoupling from SO₂ and PM₁₀ by 44% and 42%, respectively. These findings underscore the role of HQDI in promoting sustainable development driven by innovation, green development, and technological progress. Fujian Province has actively embraced high-quality development principles, adjusting its industrial and energy structures to support decoupling efforts. Moreover, the results in columns (3)–(5) in Table 6 indicate that HQDI has a significant positive impact on the decoupling effect of PM_2.5 at the 1% confidence level, thereby validating the reliability of the benchmark findings.

4.4.2. Robustness Test: Sensitivity Analysis of Extreme High Temperature

In the benchmark regression, we defined extreme high-temperature days as the total number of days per month with a maximum temperature of 30 °C or higher. This definition assumes that when temperatures exceed 30 °C, industrial equipment consumes additional energy for cooling, leading to higher overall energy consumption, which could influence the decoupling effect. The results indeed show that an increase in extreme high-temperature days reduces the decoupling effect between economic growth and pollution emissions. However, due to the difficulty in defining a precise threshold for extreme high temperatures, we conducted a robustness check by redefining extreme high-temperature days as those between 28 °C and 35 °C. The results are presented in

Table 7. Results of robustness test: Sensitivity analysis of extreme high temperature.

Variable	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
	Decoupling from PM2.5
High-quality development index	0.211 **	0.233 **	0.258 ***	0.244 ***	0.235 ***	0.231 ***	0.208 **	0.177 **
	(0.091)	(0.090)	(0.089)	(0.087)	(0.085)	(0.084)	(0.082)	(0.080)
Extreme high temperature (28 °C)	−0.006 **
	(0.002)
Herfindahl–Hirschman index	0.327 *	0.403 **	0.501 **	0.501 **	0.510 **	0.528 **	0.475 **	0.387 *
	(0.193)	(0.198)	(0.202)	(0.205)	(0.206)	(0.208)	(0.204)	(0.200)
Extreme high temperature (29 °C)		−0.007 ***
		(0.002)
Extreme high temperature (30 °C)			−0.009 ***
			(0.003)
Extreme high temperature (31 °C)				−0.009 ***
				(0.003)
Extreme high temperature (32 °C)					−0.009 ***
					(0.003)
Extreme high temperature (33 °C)						−0.011 ***
						(0.003)
Extreme high temperature (34 °C)							−0.012 ***
							(0.003)
Extreme high temperature (35 °C)								−0.013 ***
								(0.005)
Year fixed effects	YES	YES	YES	YES	YES	YES	YES	YES
City fixed effects	YES	YES	YES	YES	YES	YES	YES	YES
Observations	486	486	486	486	486	486	486	486
R2	0.324	0.327	0.333	0.331	0.330	0.331	0.329	0.326

Note: ***, ** and * represent 1%, 5% and 10% significant levels, respectively. All columns control for year fixed effects and city fixed effects. The values in parentheses represent robust standard errors.

.

The results of this robustness test, shown in Table 7, reveal that even with this modified definition, HQDI continues to significantly increase the decoupling effect. Additionally, the negative impacts of extreme high-temperature days and the Herfindahl–Hirschman index on decoupling remain statistically significant, consistent with the benchmark results. This sensitivity analysis indicates that the conclusions drawn in this study are robust and not sensitive to the specific choice of extreme high-temperature threshold. Regardless of how extreme high-temperature days are defined, HQDI has a robust and positive effect on decoupling, while extreme temperatures remain a significant factor hindering the decoupling of economic growth and pollution emissions.

4.5. Impact of HQDI Sub-Items on the Decoupling Effect

The findings of the preceding analysis reveal that HQDI plays a crucial role in the decoupling of economic growth and pollution emissions. To further explore the underlying drivers, HQDI was disaggregated into five dimensions, and their impacts on decoupling were analyzed. The results are presented in

Table 8. Impact of HQDI sub-items on decoupling from PM_2.5.

Variable	(1)	(2)	(3)	(4)	(5)
	Decoupling from PM2.5
Innovation	0.023 ***
	(0.009)
Green		0.003
		(0.022)
Coordination			0.035
			(0.039)
Openness				0.010 **
				(0.005)
Sharing					0.132 *
					(0.074)
Year fixed effects	YES	YES	YES	YES	YES
City fixed effects	YES	YES	YES	YES	YES
Observations	486	486	486	486	486
R2	0.328	0.324	0.325	0.325	0.328

Note: ***, ** and * represent 1%, 5% and 10% significant levels, respectively. All columns control for year fixed effects and city fixed effects. The values in parentheses represent robust standard errors.

.

The results in Table 8 reveal that innovation, openness, and sharing significantly enhance the decoupling relationship, suggesting that technological progress, global integration, and social sharing foster economic growth while reducing emissions. In contrast, green development and coordinated development show positive but statistically insignificant effects. For green development, including both green and brown industries dilutes the positive impact of environmentally friendly sectors, as high-energy-consuming brown industries offset gains. The coordinated development indicator reflects disparities in industry and regional electricity consumption and lacks significance. This finding suggests that resource distribution and regional development imbalances have not effectively contributed to decoupling economic growth from pollution emissions.

This study also analyzes how sub-indicators in the three dimensions—innovation, green development, and coordinated development—impact decoupling. The results, detailed in Appendix A,

Table A2. Impact of all HDQI sub-items on decoupling from PM_2.5 (1).

Variable	(1)	(2)	(3)	(4)	(5)
	Decoupling from PM2.5
Scale of industry digitalization	0.017 ***
	(0.006)
Scale of digital industrialization		0.017 ***
		(0.006)
High-tech industry development			0.020 ***
			(0.006)
Green development				0.004
				(0.006)
Brown development					−0.020 **
					(0.010)
Year fixed effects	YES	YES	YES	YES	YES
City fixed effects	YES	YES	YES	YES	YES
Observations	486	486	486	486	486
R2	0.328	0.328	0.329	0.325	0.329

Note: *** and ** represent 1% and 5% significant levels, respectively. All columns control for year fixed effects and city fixed effects. The values in parentheses represent robust standard errors.

and

Table A3. Impact of all HDQI sub-items on decoupling from PM_2.5 (2).

Variable	(1)	(2)	(3)	(4)	(5)	(6)
	Decoupling from PM2.5
Regional coordination	0.003
	(0.007)
Industrial coordination		−0.007
		(0.008)
Capacity expansions of SMEs			0.018 ***
			(0.005)
Capacity reductions of SMEs				−0.018 **
				(0.009)
Residential electricity consumption difference					0.013
					(0.010)
Public infrastructure capacity expansion difference						0.004
						(0.008)
Year fixed effects	YES	YES	YES	YES	YES	YES
City fixed effects	YES	YES	YES	YES	YES	YES
Observations	486	486	486	486	486	486
R2	0.325	0.325	0.328	0.329	0.327	0.325

Note: *** and ** represent 1% and 5% significant levels, respectively. All columns control for year fixed effects and city fixed effects. The values in parentheses represent robust standard errors.

, demonstrate that all three dimensions of innovation, including digital economy development (digital industrialization and industrial digitization) and high-technology industry growth, significantly promote decoupling. By introducing clean technologies and more efficient production processes, innovation helps reduce pollution emissions while driving economic growth, fostering a decoupling effect.

In contrast, the development of brown industries, as part of the green indicators, shows a significant negative impact. High-polluting, energy-intensive industries tend to have low production efficiency, and the cost of adopting cleaner energy or technologies is often prohibitive. These industries focus on immediate profits while neglecting environmental concerns. Consequently, an increase in the electricity consumption of brown industries signifies greater output but also leads to higher pollution emissions, hindering decoupling. On the other hand, green industries typically utilize clean energy or environmental technologies, meaning their growth contributes to lower pollution emissions, thus promoting decoupling. However, as the results show, the effect of green industries on decoupling is limited, likely due to their smaller share of total electricity consumption and relatively modest contribution to broader economic and environmental dynamics.

Regarding coordinated development (regional and industrial coordination), we find that while reduced differences between regions and industries could improve resource allocation, energy use, industrial structure, and technological innovation, it does not always enhance decoupling. The absence of a clear shift toward cleaner industries or cleaner energy technologies means that coordination, without an emphasis on sustainability, may inadvertently lead to more pollution-intensive industrial activities. Therefore, if coordination leads to an industrial structure that favors high-polluting and high-emission sectors, it could limit the potential for decoupling.

Openness, as indicated by the growth of the logistics industry, shows a positive influence on decoupling. An expansion of the logistics sector drives economic growth, but it also promotes smart logistics and efficient systems that reduce energy consumption and pollution. Moreover, logistical improvements lead to resource optimization, lower transportation costs, and reduced emissions, collectively contributing to the decoupling of economic growth from pollution.

As for shared development, specifically the disparities in urban and rural electricity consumption and public infrastructure differences, we observe no significant impact on decoupling. The primary reason is that residential electricity consumption in urban and rural areas mainly supports daily living, with limited direct influence on economic growth and emissions. Similarly, infrastructure disparities reflect regional imbalances but do not directly affect the decoupling process. Additionally, the expansion of SMEs positively affects decoupling by driving market dynamism, promoting technological innovation, and improving resource efficiency. In contrast, a reduction in SME capacity negatively impacts decoupling, likely because it diminishes economic activity and investment in environmental technologies. This, in turn, hinders decoupling efforts and further verifies the promotion of decoupling brought by expanding SMEs. All the results above support the hypothesis 2 in this study that all dimensions of high-quality development contribute to the decoupling effect, while the development of brown industries may hinder the decoupling effect.

5. Conclusions and Outlook for Future Research

This study conducted a comprehensive investigation into the influence of high-quality development on the decoupling effect between economic growth and pollution emissions across nine prefecture-level cities in Fujian Province. By employing the input–output analysis method, this study calculated the direct electricity consumption coefficients of various industries and determined the total regional output using industrial electricity consumption and electricity price data. This study constructed a decoupling index using the Tapio decoupling method, while the entropy method was utilized to build an HQDI grounded in the five principles of innovation, green, coordination, openness, and sharing. The findings reveal that HQDI significantly contributes to achieving decoupling, with innovation, openness, and sharing playing crucial roles. On the other hand, brown industry development and the installation of electricity capacity by SMEs negatively impact the decoupling process. Extreme temperatures also exert a substantial adverse effect, while increased market concentration supports decoupling.

Based on the study’s findings, several policy recommendations are proposed to promote sustainable economic growth while decoupling it from pollution emissions. Firstly, innovation and openness should be incentivized through R&D support and industry–university collaborations, fostering the adoption of cleaner technologies. Additionally, stronger policies to support green industries and environmentally friendly technologies are essential, as brown industries hinder decoupling. Energy efficiency standards for SMEs should be implemented, alongside financial and technical assistance, to reduce their energy consumption. To address the negative impacts of extreme weather, climate-resilient infrastructure and increased use of renewable energy should be prioritized. Furthermore, promoting market concentration can enhance energy efficiency and sustainability, though regulatory oversight is necessary to maintain competition. Expanding renewable energy usage and refining the HQDI to include emerging factors like digital technologies and circular economies will also support decoupling. These integrated policy measures aim to guide Fujian Province and similar regions toward a balanced and sustainable development trajectory.

The decision to focus on Fujian Province in this study stems primarily from the availability of extensive electricity consumption data, which provide a solid foundation for our analysis. Additionally, Fujian’s industrial structure and significance within the broader national context make it a representative case for understanding high-quality growth strategies and decoupling processes. However, this focus on a single province is also a limitation of the study, as the findings may not be fully generalizable to regions with different industrial profiles or energy dynamics. Fortunately, the proposed methodology will be applied to other regions or countries once electricity consumption and price data are available. Future research could expand the geographical scope and incorporate additional variables, such as renewable energy usage, innovation capabilities, and policy interventions, to offer a more holistic understanding of decoupling dynamics across diverse regions. Moreover, investigating the impact of green technologies and government policies on decoupling could provide valuable insights into how sustainable economic growth can be further promoted, shedding light on the complex relationship between economic development and environmental sustainability.