The Impact of Energy Transition on China’s Urban GTFP: Based on the Threshold Effects of Technology and Policy

Abstract

This research focuses on the impact of energy transition on China’s urban green total factor productivity (GTFP). We first constructed a comprehensive index system to measure the energy transition. Adopting this index, we analyzed the impact of energy transition on urban GTFP in China, identifying a significant threshold effect. Furthermore, we investigated the heterogeneity of this effect across different regions and city types (categorized by their resource endowments). By constructing benchmark regression and panel threshold models, the paper conducts studies by taking energy technology innovation and energy policy reform as threshold variables. The results show that energy transition is capable of promoting the improvement of GTFP in Chinese cities; its impact exhibits a nonlinear threshold effect, with energy technology innovation and energy policy reform showing a single threshold characteristic, and the promoting effect intensifies after crossing the threshold; there exists heterogeneity at the regional and city-type levels, with cities in the eastern, central, western, and northeastern regions, as well as different resource-based cities, exhibiting varying performances. Based on the results, in order to promote energy transition and urban sustainable development, it is necessary to formulate policies reasonably and strengthen technological innovation.

1. Introduction

There exists a close relationship between economic development and energy consumption. For a long time, fossil fuels such as coal, oil, and natural gas have served as the primary driving force for economic growth, supporting the rapid development of the global industrial system. The energy structure dominated by fossil fuels has also brought about serious environmental issues, for instance, greenhouse gas emissions, air pollution, and water resource pollution [1,2]. With the issue of global climate change becoming increasingly severe, governments worldwide have begun to realize the necessity of transforming energy consumption patterns and seeking clean and renewable energy alternatives to mitigate the impact of climate change and ensure the sustainable development of the ecological environment [3,4,5].

As a vital indicator for measuring the capacity of sustainable economic development, GTFP comprehensively considered multiple aspects such as economic growth, resource utilization, and environmental protection [6]. Under the background of energy transition, the significance of GTFP has become increasingly prominent. On the one hand, energy transition helps to enhance GTFP by improving energy efficiency, reducing energy consumption, and minimizing environmental pollution. On the other hand, the improvement of GTFP can also provide power and support for energy transition, promoting economic development towards a greener and more sustainable direction [7]. Therefore, studying the relationship between energy transition and GTFP has important theoretical and practical significance for promoting sustainable economic development, as well as achieving environmental protection and energy security.

In the technology dimension, energy transition drives the improvement of GTFP through technological innovation, which has a significant promoting effect on sustainable economic development. Technological innovation is a key driving force for energy transition. It can reduce the cost of renewable energy and improve its economy, as well as promote the efficient operation and flexible scheduling of energy systems. For instance, in the field of solar energy, with the advancement of photovoltaic technology and the decrease in costs, solar power generation has become one of the most competitive renewable energy sources in the world. According to the report of the International Energy Agency, by 2030, solar power is expected to become one of the world’s largest sources of electricity. Furthermore, in terms of energy storage technology, the rapid development of new energy storage technologies like lithium-ion batteries, sodium-ion batteries, and flow batteries provides strong support for the efficient utilization and storage of renewable energy [8].

In the dimension of policy, the government promotes energy transition and the improvement of GTFP by formulating a series of incentive measures and regulatory policies. For instance, financial subsidies and preferential taxation can reduce the investment costs and market risks of renewable energy and stimulate the investment enthusiasm of enterprises and individuals. The carbon emissions trading system can constrain the carbon emission behavior of enterprises through market mechanisms, promoting their development towards low-carbon and environmental protection. The renewable energy quota system can ensure that the proportion of renewable energy in energy consumption continues to increase, promoting the optimization and upgrading of the energy structure. These policies can not only provide powerful institutional guarantees and market environment for the development of renewable energy but also drive the continuous improvement of GTFP, which is of great significance to sustainable economic development and environmental protection [9].

Against the backdrop of global climate change and China’s dual carbon goals, energy transition has become an important pathway for promoting urban sustainable development. However, the energy transition process is complex, and focusing solely on individual dimensions such as energy structure or technology makes it difficult to fully uncover the intrinsic mechanisms through which energy transition affects GTFP. Although existing research has increasingly examined the relationship between energy transition and green economic development, most studies remain focused at the national or provincial level, with limited attention devoted to the systematic construction of a comprehensive evaluation framework for energy transition at the city level. This research gap hinders a nuanced understanding of the diverse transition trajectories and dynamic patterns across cities. To mend this gap, this study first systematically constructed a comprehensive evaluation system for energy transition at the scale of 284 prefecture level and above cities in China, providing a more refined perspective for understanding the diverse transformation paths at the urban level. Secondly, we not only verified the direct impact of energy transition on GTFP but also made a core contribution by adopting a panel threshold model to test and confirm the existence of a single threshold effect for both “energy technology innovation” and “energy policy reform”, revealing the key nonlinear mechanism of “significantly enhancing the promotion effect after crossing the threshold”. Finally, we further revealed the heterogeneity of the threshold effect from the dual dimensions of “regional geography” and “resource attributes”, clarifying that cities at different development stages and types have discrepancies in core paths (technologies or policies) in driving GTFP improvement through energy transition. This provides a precise theoretical basis for formulating differentiated policies.

The remainder of this paper is organized as follows. We present a literature review and the research hypotheses in Section 2. In Section 3, we discuss our materials and methods. Section 4 reports our analysis results, and Section 5 presents our conclusions and offers a discussion on the limitations of our results.

2. Literature Review and Research Hypotheses

2.1. The Effect of Energy Transition

The energy transition is a key element of the sustainable transition, and the correlation mechanism and impact path between energy transition and green total factor productivity (GTFP) have become a core issue in academic research; relevant empirical studies have revealed multidimensional transmission logic and heterogeneity characteristics. From the perspective of impact pathways—energy structure, energy technology, and energy policies are the main transmission channels—based on 78 economies in the world, Wang’s (2024) research shows that the promotion effect of energy structure optimization on GTFP exhibits nonlinear characteristics, with per capita GDP and carbon intensity as thresholds [10]. When the critical values are crossed, its effect changes from negative to positive; by adopting data from Chinese prefecture level cities, Zeng et al. (2025) further confirmed that energy consumption transformation not only enhances local GTFP, but also effectively promotes economic sustainable development and drives surrounding cities through spatial spillover effects, and the conclusion remains robust in different spatial econometric model tests [11]. As another core intermediary, the effects of technological innovation have been verified at multiple levels. The study of Wan (2025) on heavily polluting industries found that green technology innovation significantly enhances enterprise GTFP, which is conducive to energy sustainable development, with practical green patents playing a prominent role in promoting GTFP, while non practical patents have inhibitory effects [12]. Ren et al. (2025) and Zhang et al. (2025) further pointed out that green technology innovation improves the total factor productivity of enterprises through compensation effects, and innovation achievements that lag behind by one period still have a significant positive impact on enterprise performance [13,14]. The studies by Wu et al. (2025) and Ji et al. (2025) supplemented the boundary conditions of innovation mechanisms, confirming that increasing R&D expenditure can enhance the role of green innovation in total factor productivity, while equity balance can further amplify the effect [15,16]. The PVAR model constructed by Hu et al. (2023) also verified the direct driving effect of green technology innovation on manufacturing GTFP, laying a foundation for ecological sustainable development [17]. In addition, Liu et al. (2025) pointed out that under the background of the inhibitory effect of metal mineral mining on GTFP, the development of renewable energy and technological innovation are key methods to effectively offset the negative impact, further highlighting the importance of technological innovation in energy transition [18].

Policy tools and external regulations play an important regulatory role in the relationship between energy transition and GTFP. As a typical market mechanism, the carbon trading policy has been widely validated for its effectiveness. Based on the synthetic control method, Dong et al. (2025) found that carbon trading policies significantly increased the GTFP of the power industry. In 2019, the actual values in pilot areas such as Hubei and Beijing were 3.55–21.90% higher than the synthetic values [19]. Wang et al. (2025) and Zhang et al. (2024) carried out analysis from the perspectives of the micro mechanisms and transmission channels, confirming that carbon trading policies affect GTFP through promoting green and sustainable innovation in enterprises, improving financing efficiency, and optimizing energy structure, industrial structure, and outward direct investment, thereby advancing economic and ecological sustainable development [20,21]. The transmission of energy structure and industrial structure is mainly achieved through changes in green technology. The role of environmental regulation exhibits heterogeneity. Research by Gao et al. (2020) showed that economic-incentive and public-participation regulations can enhance GTFP by optimizing energy consumption structure, and their indirect spatial spillover effects show negative and positive characteristics, respectively [22]. Command-and-control and economic-incentive regulations all have a double threshold effect, and their impact on GTFP varies in stages with changes in regulatory intensity. From an international comparative perspective, Ahmed et al. (2019) found a bidirectional causal relationship between renewable energy consumption and economic growth in 30 emerging and developing countries, but its impact on GTFP varies regionally. The transmission effect is significant in South Asia, Asia, and some African countries, while the effect is weaker in Latin American and Caribbean countries due to the initial stage of renewable energy consumption [23]. This echoes the differential characteristics of China’s provincial GTFP revealed by Gao et al. (2021): the provincial GML productivity (including GTFP connotation), considering carbon emissions, shows regional heterogeneity driven by faster growth and technological progress in eastern provinces [24]. In addition, Yuan et al. (2025) pointed out that the positive impact of digitization and intelligentization on the GTFP of the Yangtze River Economic Belt can be moderated and strengthened by technological innovation, serving the goal of economic sustainable development, further enriching the research on moderating variables in the relationship between energy transition and GTFP [25]. Meng et al.’s (2024) research on the impact of green finance on GTFP provides a supplementary perspective for understanding the funding support mechanism for energy transition [26].

In summary, existing research, from global analysis to research at the level of Chinese cities and enterprises, consistently indicates a positive correlation between energy transition and green development outcomes. The supplementary research on digitalization and green finance further strengthens the potential path for energy transformation to promote economic and environmental sustainability. Based on the solid foundation of the above theoretical and empirical research, the paper proposes the first core hypothesis:

Hypothesis 1.

Energy transition has a promoting effect on GTFP in Chinese cities.

2.2. The Threshold Effect of Energy Technological Innovation

In the process of energy transition, technological innovation undoubtedly plays a core driving role. The research in this field not only focuses on the latest breakthroughs in cutting-edge technologies such as solar cells, wind power generation, and energy storage, but also deeply analyzes how these technologies fundamentally change the economy, efficiency, and reliability of renewable energy [27], thereby having a profound impact on the structure, operational logic, and market mechanisms of energy systems. Chai et al. (2024) emphasized in their global inventory that the close integration of energy transition and technological innovation is the mainstream trend in the current development of the global energy field [28]. This trend is not only reflected in the direct contribution of technological progress to cost reduction but also in how it reshapes the patterns of energy production and consumption.

The impact of energy technology innovation on urban green total factor productivity (GTFP) is not a simple linear relationship, as technological innovation has the characteristics of high cost and long return cycle [29]. Consequently, there may also be a significant “threshold effect” in its effectiveness release. Energy technology innovation will affect the cost of new energy use and the benefits of enterprises, thereby creating a threshold effect on GTFP. When technological innovation is at a low level, the conversion rate of new energy technologies is low, resulting in limited energy output, while higher operating and maintenance costs keep energy innovation investment high. At this point, the marginal cost of energy transition far exceeds the marginal benefit [30]. In this situation, promoting the energy transition will result in a large amount of capital being trapped in inefficient energy facilities and unable to flow towards areas that can improve production efficiency, leading to factor mismatches. This will not have a significant promoting effect on GTFP and may even inhibit its growth. After energy technology innovation crosses the threshold, the application of high-efficiency energy conversion efficiency and low-cost energy storage technology can make the cost of new energy lower than traditional energy [31], gradually increasing the benefits of energy transition beyond costs. From the perspective of factor allocation, the capital previously occupied by inefficient energy facilities can be released, and these resources will flow towards efficient fields such as high-end manufacturing and digital economy, improving total factor productivity. In terms of energy efficiency, new energy technologies can improve energy utilization [32], ensuring energy sustainable development, increase output per unit of energy consumption, reduce energy waste, and effectively promote the growth of local GTFP, realizing ecological sustainable development.

Based on the above analysis, the section proposes the following hypothesis.

Hypothesis 2.

The impact of energy transition on green total factor productivity in Chinese cities is influenced by the threshold effect of energy technology innovation. Which is, under different levels of energy technology innovation, energy transition has shown varying intensity of impact on the green total factor productivity of Chinese cities.

2.3. Threshold Effect of Energy Policy

In the study of the impact of energy transition on urban GTFP, the role of energy policies is not simply a linear relationship but may exhibit significant nonlinear characteristics. This is because the regulatory effect of energy policies on energy transition is influenced by various factors such as the degree of policy perfection, implementation intensity, and the relationship with the market and enterprises. At different policy levels, the direction and intensity of their impact on the relationship between energy transition and GTFP may vary. The threshold effects may be generated through environmental protection systems, market demand regulation, enterprise incentives, and other aspects.

Energy policies can serve as threshold variables in the impact of energy transition on GTFP by improving environmental protection systems. When the intensity of energy policies does not cross the threshold, the impact on energy transition is relatively weak. Due to the relatively vague regulations on pollutant emission standards and weak enforcement of low-intensity policies, enterprises lack emission reduction constraints in the process of energy transition, which makes them tend to continue adopting traditional energy for production. This has hindered the progress of the energy transition and failed to truly promote the transformation of energy structure towards low carbon. The extensive use of high-carbon energy not only causes serious environmental pollution [33], damaging ecological sustainable development, but also leads to waste of production factors due to low energy utilization efficiency, thereby suppressing the growth of GTFP. And when the intensity of energy policies crosses the threshold, it can effectively regulate the direction of enterprise energy transition. Some types of energy policies can have clear quantitative standards for pollutant emissions, and strict enforcement measures such as fines or production shutdowns can be used to force enterprises to reduce the use of traditional energy in production activities, increase investment in clean energy and environmental protection transformation of production processes, and achieve energy-saving and emission reduction effects [34,35]. In this process, by adopting new energy technologies and optimizing their energy consumption structure, enterprises are capable of reducing pollution emissions while improving energy utilization efficiency, achieving multiple goals of energy transition, environmental protection, and improving production efficiency, thereby promoting GTFP growth and facilitating sustainable economic and energy development.

Based on the above analysis, the section proposes the following hypothesis:

Hypothesis 3.

The impact of energy transition on green total factor productivity in Chinese cities is influenced by the threshold effect of energy technology policies. Which is, after the energy policy intensity reaches a certain level, it can enhance the promoting effect of energy transition on the green total factor productivity of Chinese cities.

2.4. Theoretical Framework

Based on the literature review and research hypotheses mentioned in the above literature, this study constructs a comprehensive theoretical analysis framework. This framework clearly outlines the core path of energy transition affecting urban green total factor productivity and the research logic of this paper. The mechanism transmission path is shown in Figure 1:

By changing the cost of using new energy and the efficiency of enterprises, energy technology innovation can increase the marginal benefits of the energy transition beyond the marginal cost after reaching a specific threshold, releasing a strong driving force for GTFP. Energy policy reform can effectively regulate the direction of transition and improve energy efficiency by improving environmental protection systems, regulating market demand, and motivating corporate behavior. After crossing the intensity threshold, it can prominently strengthen the positive impact of energy transition on GTFP.

In summary, the core objective of this study is to empirically test the direct impact of the energy transition on urban GTFP and to reveal the dual threshold mechanism driven by technology and policy behind it.

3. Materials and Methods

3.1. Measurement Model

To examine the impact of the energy transition index on the GTFP of Chinese cities, the paper establishes a benchmark regression model as follows:

$${\mathit{GTFP}}_{\mathrm{it}} = {\alpha }_0 + {\alpha }_1{\mathit{ETI}}_{\mathit{it}} + \sum \delta \mathit{Controls} + {\mu }_i + {\lambda }_i + {\epsilon }_i$$

(1)

${\mathit{GTFP}}_{\mathit{it}}$ represents the green total factor productivity of the $\mathrm{i}$-th city in the $\mathrm{t}$-th year, ${\mathrm{ETI}}_{\mathrm{it}}$ represents the comprehensive index of energy transition in the $\mathrm{i}$-th city in the $\mathrm{t}$-th year, ${\mathsf{\alpha }}_0$ represents the intercept term, ${\mathsf{\alpha }}_1$ represents the regression coefficient of the explanatory variable, $\mathrm{Controls}$ represents the various control variables that affect the green total factor productivity of Chinese cities, ${\mathsf{\mu }}_{\mathrm{i}}$ and ${\mathsf{\lambda }}_{\mathrm{i}}$ represent the regional fixed effect and time fixed effect, respectively, and ${\mathsf{\epsilon }}_{\mathrm{i}}$ is the residual term.

The threshold model based on energy transition is set as follows:

$${\mathit{GTFP}}_{\mathit{it}} = \sum _{k = 1}^{n + 1}{\alpha }_k{\mathit{ETI}}_{\mathit{it}}·I\left(d_{k\text{-}1} < m_{\mathit{it}} \le d_k\right) + {\beta \mathit{Z}}_{\mathit{it}} + {\epsilon }_{\mathit{it}}$$

(2)

Among them, ${\mathrm{GTFP}}_{\mathrm{it}}$ represents the green total factor productivity of city i in the $\mathrm{t}$-th period, ${\mathrm{ETI}}_{\mathrm{it}}$ represents the energy transition index of city i in the $\mathrm{t}$-th period, ${\mathrm{m}}_{\mathrm{it}}$ is the threshold variable (adopting energy technology innovation, energy policy reform, and energy structure innovation for threshold testing in sequence), and ${\mathrm{d}}_{\mathrm{k}}$ represents k threshold values; $\mathrm{I}\left(\blacksquare \right)$ represents the indicator function. $\mathrm{I}(·)$ represents the exponential function, ${\mathrm{Z}}_{\mathrm{it}}$ represents the control variable, and ${\mathsf{\epsilon }}_{\mathrm{it}}$ represents the random error term.

3.2. Variable Selection

3.2.1. Core Explanatory Variable

At present, there is no unified standard for measuring indicators related to energy transition paths, while studies on energy transition paths can be roughly divided into three categories: energy structure, energy technology, and energy policies. Based on which, this study comprehensively considers the availability and development of indicators for different paths of energy transition and adopts scientific and systematic measurement methods to construct a multidimensional energy transition measurement system. The indicator system includes the current mainstream paths of energy transition, namely, energy structure optimization, energy technology innovation, and energy policy reform.

Energy structure optimization: As an important path to achieve coordinated development of the energy industry and ensure energy security, energy structure optimization is an effective strategy to reduce emission intensity and promote sustainable urban economic development [36,37]. The core of the current energy transition is clean and low-carbon substitution. Natural gas and electricity are clean fuels, and an increase in their consumption proportion can significantly reduce dependence on high carbon-emitting energy sources such as coal, reduce pollutants and carbon emissions, and meet the “dual carbon” goal. At the same time, they have strong stability and are suitable for the consumption of renewable energy. The change in their proportion directly reflects the degree of transition of the energy system from high-carbon to low-carbon and from traditional to clean. Both are intuitive and key indicators for measuring the optimization process of regional energy structure. Given the limited availability of data on total energy consumption at the urban level, this paper refers to the research methods of scholars such as Mao (2024) and Wang (2018) to construct an energy structure optimization proxy variable based on the ratio of the sum of natural gas and electricity consumption to total energy consumption [38,39].

Energy technology innovation: Energy technology innovation is one of the core driving forces for energy transition and high-quality economic development [40]. The study draws on the study frameworks of scholars such as Ye (2018) and Fan (2020) and uses the number of energy technology innovation patents as a measurement indicator [41,42].

Energy policy reform: Energy policy reform refers to the systematic adjustment of national energy strategies and regulations to promote the transformation of energy structure towards low-carbon and sustainable direction. The paper draws on the method of Zeng and Tong (2018), adopting the number of energy-transition-related laws and regulations in each city as a standard to measure energy policy reform [43].

Table 1. Index system for calculating the comprehensive index of energy transition.

First-Level Indicator	Second-Level Indicators	Measurement Standards
Energy Transition Index	Energy Structure Optimization	Natural gas and electricity consumption/total energy consumption
	Energy Technology Innovation	Number of energy-saving and emission reduction patents
		Number of renewable energy patents
		Number of patents related to new energy
	Energy Policy Reform	Number of energy-saving and emission reduction policies
		New energy support policies
		Number of laws and regulations related to renewable energy

is an explanation of the energy transition indicator system.

The paper calculates the weights of energy transition in three dimensions: energy structure, energy technology, and energy policy by adopting the entropy weight method and the CRITIC method, respectively. The energy transition index of each city in China is calculated through linear weighted average, and the linear weighted average formula is:

$${\mathit{ETI}}_{\mathit{it}} = \sum _{j = 1}^n{\omega }_j·x_{\mathit{ijt}}$$

(3)

In the formula, ETI is the comprehensive index of energy transition, ${\mathsf{\omega }}_{\mathrm{j}}$ is the average weight calculated by energy structure, energy technology, and energy policy c, and ${\mathrm{x}}_{\mathrm{ijt}}$ represents the index of the i-th city in the t-th year. Based on the above method, a comprehensive evaluation score for the energy transition index of 284 cities in China was obtained.

The weights of energy transition indicators in different dimensions are shown in

Table 2. Weight table of energy transition indicators.

Indicator	Metric	Entropy Weight	CRITIC Weight	Overall Weight
Energy Technology Innovation	Number of energy technology-related patents	0.6321	0.1567	0.3944
Energy Policy Reform	Number of relevant energy policies	0.2743	0.2152	0.2448
Energy Structure Optimization	Share of electricity and natural gas consumption	0.0936	0.6280	0.3608

below:

The entropy weight method focuses on the degree of dispersion of indicator data and is an objective weighting method. The CRITIC method considers both the conflict between indicators and the strength of comparison, which is more objective. Combining the two methods (taking the average weight) can balance the information content of the data itself and the inherent correlation between indicators, thereby avoiding the bias that may be caused by a single method and making the constructed energy transition index (ETI) more robust and reliable.

To more intuitively present the distribution characteristics of the core explanatory variables, Figure 2 displays the frequency distributions of energy technology innovation (a), energy policy reform (b), and energy structure optimization (c) across the sample cities.

3.2.2. The Explained Variable

The GTFP measurement indicators involved in the study include input indicators and output indicators. Input indicators mainly include labor input, land input, capital input, and energy input, while output indicators include expected output and unexpected output. The expected output is measured by the actual annual GDP of the city, while the unexpected output includes sudden emissions from the urban environment, including industrial wastewater emissions, industrial exhaust emissions, and industrial smoke (dust) emissions. The specific input–output data is shown in

Table 3. Detailed input and output indicators.

	Indicator Selection	Indicator Implication	Unit
Input Indicator	Labor Input	Urban year-end employment number	Thousand(s) people
	Land Input	Urban built-up area	Square kilometer(s)
	Capital Input	Capital stock of each city	Thousand(s) yuan
	Energy Input	Urban energy consumption	Thousand(s) ton(s) of standard coal
Expected Output	Actual GDP	GDP of each city	Thousand(s) yuan
Unexpected Output	Waste Gas	Industrial waste gas emission volume	Ton(s)
	Wastewater	Industrial wastewater emission volume	Thousand(s) ton(s)
	Smoke (powder) dust Smoke (dust)	Industrial smoke (powder) dust emissions volume	Ton(s)

Note: Industrial waste gas emissions, as a comprehensive indicator, indirectly reflect the intensity of non-CO₂ greenhouse gas emissions, including methane.

.

This study employs the Slack-Based Measure–Global Malmquist–Luenberger (SBM-GML) model to measure green total factor productivity (GTFP). The Slack-Based Measure (SBM) model, proposed by Tone (2001), is a non-radial, slack-oriented efficiency evaluation approach [44]. By characterizing the slack discrepancies between inputs and outputs, it overcomes the limitations of the radial assumption inherent in traditional Data Envelopment Analysis (DEA) models and more precisely captures the actual efficiency of decision-making units (DMUs). Tone (2002) [45] subsequently extended his work to propose the super-efficiency SBM model. Leveraging the computational logic of temporarily removing the unit under evaluation from the production frontier. This model enables the differentiated ranking of efficient units (a super-efficiency score > 1 signifies a relative efficiency advantage).

This formula of SBM is:

$$\rho = min{\displaystyle \frac{(1/m){\sum }_{i=1}^m \frac{x}{X_{ik}}}{1/\left(r_1 + r_2\right)\left({\sum }_{s=1}^{r1} \frac{{\bar{y}}_s^d}{y_{sk}^d} + {\sum }_{q=1}^{r2} \frac{{\bar{y}}^u}{y_{qk}^u}\right)}}$$

$$s\left\{\begin{array}{ll}{\sum }_{j=1,j\ne k}^n x_{ij}{\lambda }_j ⩽{ x}^{-}& (i=\mathrm{1,2},\dots ,m)\\ {\sum }_{j=1,j\ne k}^n y_{sj} ⩾{ y}_{\mathrm{sk} }^d& \left(s=\mathrm{1,2},\dots ,r_1\right)\\ {\sum }_{j=1,j\ne k}^n y_{qk}^u ⩽{ y}^u& \left(q=\mathrm{1,2},\dots ,r_2\right)\end{array}\right.$$

(4)

The Malmquist–Luenberger (ML) index is an index designed to measure temporal changes in production efficiency, particularly in contexts where undesirable outputs (e.g., environmental pollution) are incorporated [46]. Built upon the Malmquist Index and the Luenberger Index, its core insight lies in the integration of the concept of undesirable outputs into traditional production efficiency analysis. Oh (2010) extended the ML index by incorporating the global production possibility set, thereby developing the Global Malmquist–Luenberger (GML) index [47]. The GML index offers a more comprehensive assessment of temporal changes in production efficiency. In the context of GTFP research, it facilitates inter-temporal comparisons and addresses the issue of infeasible solutions associated with the ML index in scenarios involving non-transitivity and linear programming.

This formula of GML is:

$$\begin{array}{l}{GML}^{t,t+1}\left(x^t,y^t,z^t,x^{t+1},y^{t+1},z^{t+1}\right)\\ ={\displaystyle \frac{1+D^g\left(x^t,y^t,z^t\right)}{1+D^g\left(x^{t+1},y^{t+1},z^{t+1}\right)}}\\ ={\displaystyle \frac{1+D^t\left(x^t,y^t,z^t\right)}{1+D^{t+1}\left(x^{t+1},y^{t+1},z^{t+1}\right)}}\\ \times \left[{\displaystyle \frac{1+D^g\left(x^t,y^t,z^t\right)}{1+D^t\left(x^t,y^t,z^t\right)}}\times {\displaystyle \frac{1+D^{t+1}\left(x^{t+1},y^{t+1},z^{t+1}\right)}{1+D^g\left(x^{t+1},y^{t+1},z^{t+1}\right)}}\right]\\ ={GEC}^{t,t+1}\times {GTC}^{t,t+1}\end{array}$$

(5)

Given that temporal variation must be considered in the measurement and analysis of production efficiency, the super-efficiency SBM model is limited in that it can only assess and estimate efficiency values at individual time points, failing to accurately capture productivity changes driven by temporal variation. These limitations constrain the measurement and analysis of GTFP; consequently, a growing number of scholars have adopted the SBM-GML model for investigating production efficiency [48,49,50].

3.2.3. The Data Source

The study sample covers 284 cities at the prefecture level and above, except Hong Kong, Macao, Taiwan, and the Xizang Autonomous Region. The data is sourced from authoritative official channels such as the China Urban Statistical Yearbook, China Energy Statistical Yearbook, “Legal Star” database, and the National Bureau of Statistics to ensure the accuracy and authority of the data. Considering the availability and integrity of data, the study sample covers 284 cities at the prefecture level and above, except Hong Kong, Macao, Taiwan, and the Xizang Autonomous Region. The detailed table of relevant indicators is shown in

Table 4. Detailed list of relevant indicators.

Variable	Abbreviation	Type
Green total factor productivity of Chinese cities	GTFP	Explained variable
Energy Transition Index	ETI	Explanatory variable
Energy Technology Innovation	TI	Threshold variable
Energy Policy Reform	EPR
Energy structure optimization	ESO
Degree of government intervention	DGI	Controlling Variable
Opening up level	OL
Marketization level	ML
Population density	PD

:

Table 5. Descriptive statistics of the data.

	Variable	Sample Capacity	Average Value	Standard Deviation	Least Value	Crest Value
Explained variable	GTFP	3124	1.006	0.0574	0.3734	1.7154
Core explanatory variables	ETI	3124	0.0825	0.0748	0.0001	0.7533
Threshold variable	TI	3124	33.54	142.4	0	2590
	EPR	3124	19.31	32.25	0	456
Controlled variable	DGI	3124	0.2026	0.1022	0.0439	0.9155
	OL	3124	0.0169	0.0184	0.0000	0.2287
	ML	3124	1.3989	1.7693	0.0127	76.7012
	PD	3124	5.7273	0.9492	0.6831	7.8816

provides data descriptive statistics, interpolation supplement, and individual missing data.

4. Empirical Result

4.1. The Direct Impact of Energy Transition on Green Total Factor Productivity in Chinese Cities

4.1.1. Analysis of the Benchmark Regression Results

Based on the panel data of 284 cities at the prefecture level and above in China from 2011 to 2021, the section empirically analyzed the impact of the ETI on China’s urban GTFP, adopting the mixed OLS model, the random effect model, and the fixed effect model.

Table 6. Regression results of benchmark model.

	(1)	(2)	(3)	(4)
Variable	OLS	OLS	RE	FE
ETI	0.174 ***	0.147 ***	0.145 ***	0.161 ***
	(0.0395)	(0.0409)	(0.0461)	(0.0524)
DGI		−0.036 ***	−0.033 ***	−0.043
		(0.0097)	(0.0089)	(0.0297)
OL		0.0131 ***	0.0132 ***	0.0128 **
		(0.0029)	(0.0029)	(0.0051)
ML		0.000051	0.000685	0.00119 *
		(0.0001)	(0.0009)	(0.0007)
PD		0.842 ***	−0.000548	−0.000452
		(0.0315)	(0.0004)	(0.0004)
Constant	0.973 ***	0.147 ***	0.842 ***	0.840 ***
	(0.0037)	(0.0409)	(0.0308)	(0.0537)
N	3124	3124	3124	3124
R2	0.115	0.186	0.098	0.135
Individual Value	284	284	284	284
Fixed Time	N	N	N	Y
Fixed Individual	N	N	Y	Y

Note: ***, **, * represent significance at levels of 1%, 5%, and 10%, respectively. Robust standard errors clustered at the city level are reported in parentheses.

shows the estimation results of the benchmark regression. Among them, models (1) and (2) are the mixed OLS regression results of panel data. Model (1) studied the relationship between ETI and China’s urban GTFP, based on which, model (2) added control variables. From the regression results, it can be said that the coefficients of the ETI in model (1) are 0.174, and all have passed the significance level test with a significance of 1%. Consequently, it can be seen that energy transition has a significant promoting effect on the improvement of GTFP in Chinese cities; the promoting effect remains robust even after adding controlling variables.

Models (3) and (4) show the results of random effects and fixed effects regression, respectively. In the random effects model (3), the coefficient of the ETI is 0.145, and it is significant at the 1% significance level, indicating that after controlling for individual fixed effects, energy transition still has a significant positive impact on the GTFP of Chinese cities. The results of the fixed effects model (4) also support this conclusion, with a coefficient of 0.161 for the ETI, which is also significant at the 1% significance level. The analysis results of model (4) reveal that the coefficient of the ETI is 0.161, which means that for every 1 unit increase in energy transition, the GTFP of Chinese cities increases by 0.161 units. The analysis results reveal a green growth path with energy transition as the core and multiple factors working together, providing empirical support for strengthening ETI investment, improving policy tools, and deepening market-oriented reforms. These findings from the benchmark regression provide strong evidence in support of Hypothesis 1.

4.1.2. Robustness Tests

To ensure the reliability of the benchmark regression results, this section separately used the entropy weight method and the CRITIC method to recalculate the energy transition indices (ETI-Entropy and ETI-CRITIC) and replace the original composite index (ETI) with it for regression under a fixed effects model. The regression results are shown in the following table.

As shown in

Table 7. Regression results with different weights.

	(1)	(2)	(3)
Variable	GTFP-Overall	GTFP-Entropy	GTFP-CRITIC
ETI	0.161 ***	0.431 ***	0.0921 ***
	(0.0524)	(0.1586)	(0.0162)
DGI	−0.043	−0.0461	0.0463 *
	(0.0297)	(0.0248)	(0.0264)
OL	0.0128 **	−0.0323	−0.0324
	(0.0051)	(0.0492)	(0.0554)
ML	0.00119 *	−0.0000965	−0.000552
	(0.0007)	(0.0003)	(0.0004)
PD	−0.000452	0.00130	0.00135
	(0.0004)	(0.0010)	(0.0012)
Constant	0.840 ***	0.974 ***	0.978 ***
	(0.0537)	0.0081	0.0078
N	3124	3124	3124
R2	0.135	0.129	0.102
Individual Value	284	284	284
Fixed Time	Y	Y	Y
Fixed Individual	Y	Y	Y

Note: ***, **, * represent significance at levels of 1%, 5%, and 10%, respectively. Robust standard errors clustered at the city level are reported in parentheses.

, whether adopting the entropy weight method (Model 2) or the CRITIC method (Model 3) to calculate the energy transition index, the regression coefficients are significantly positive at the 1% level. It indicates that the promotion effect of energy transition on urban GTFP is robust and not affected by specific weight calculation methods, further verifying the reliability of the benchmark regression conclusion in this study.

4.1.3. Endogeneity Tests

To reduce the issue of endogeneity, this paper employs Two-Stage Least Squares (2SLS) for re-estimation, selecting the mean value of the energy transition index of other cities within the same province as the instrumental variable (IV) for the energy transition index of the target city. This IV satisfies the relevance requirement because cities within the same province are highly correlated in terms of policy environment and technology diffusion. Meanwhile, it theoretically meets the exclusion restriction, as the energy transition level of other cities affects the target city’s transition decisions rather than directly acting on its GTFP.

The estimated results are presented in

Table 8. Regression results with endogeneity tests.

	First-Stage	Second-Stage	First-Stage	Second-Stage
Variables	ETI	GTFP	ETI	GTFP
ETI		0.205 ***		0.196 ***
		(0.0717)		(0.0723)
ETI_mean	0.772 ***		0.766 ***
	(0.0381)		(0.0384)
PD			−0.0003	0.0015
			(0.0018)	(0.0015)
ML			0.0004	0.0002
			(0.0009)	(0.0006)
OL			0.048	−0.014
			(0.0291)	(0.0452)
DGI			−0.027	−0.046 **
			(0.0225)	(0.0216)
Observations	3124	3124	3124	3124
Individual Value	284	284	284	284
Fixed time	YES	YES	YES	YES
Fixed Individual	YES	YES	YES	YES
Kleibergen–Paap rk LM statistic			49.309 (p = 0.0000)
Kleibergen–Paap rk Wald F statistic			740.305
Hansen J statistic			0.000

Note: ***, ** represent significance at levels of 1%, and 5%, respectively. Robust standard errors clustered at the city level are reported in parentheses.

. The first-stage regression shows that the IV is significantly and positively correlated with the endogenous variable at the 1% level. The Kleibergen–Paap rk LM test strongly rejects the null hypothesis that the IV is unidentifiable (p = 0.0000), and the Kleibergen–Paap rk Wald F statistic is 740.305, which is much higher than the critical value at the 10% bias level proposed by Stock-Yogo test [51], indicating that no weak instrumental variable problem. In the second-stage results, the coefficient of the energy transition index (ETI) remains significantly positive at the 1% level. This suggests that after controlling for endogeneity, the promotional effect of energy transition on GTFP remains robust, enhancing the credibility of the conclusions in this paper.

4.2. Threshold Effect Test of Energy Transition on Green Total Factor Productivity in Chinese Cities

4.2.1. Threshold Effect Test of Energy Technology Innovation

From the results in

Table 9. Threshold test results for energy technology innovation.

Independent Variable	Threshold Variable	Threshold Number	N	F Value	p Value	BS Times	Threshold Value	95% Confidence Interval
ETI	TI	1	3124	42.29	0.0003	3000	33	[29.5, 34]
ETI	TI	2	3124	9.19	0.1160	3000	33, 3	[29.5, 34], [1, 4]
ETI	TI	3	3124	3.41	0.7520	3000	33, 3, 37	[29.5, 34], [1, 4], [36, 38]

, it is clear that the threshold test results adopting the quantity of energy technology innovation as the threshold variable indicate that there is a single threshold effect on energy technology innovation in the model of the impact of energy transition on GTFP in Chinese cities. Therefore, the subsequent threshold effect analysis will be estimated adopting a single threshold model for energy technology innovation. Meanwhile, Table 6 presents the estimated threshold values for the number of energy technology innovations. According to the table, the threshold for the number of energy technology innovations is 33, with a 95% confidence interval of [29.5, 34].

To accurately analyze the threshold effect of the quantity of energy technology innovation in the mechanism of the impact of energy transition on China’s urban GTFP, the section has drawn an LR test chart for energy technology (Figure 3). By observation, it can be noted that when the number of energy policies is in the range of [29.5, 34], the LR statistic remains below the critical value of 7.35, indicating a significant single threshold effect of energy technology in the analysis of the impact of energy transition on GTFP in Chinese cities.

4.2.2. Threshold Effect Test of Energy Policy Reform

From the results in

Table 10. Threshold test results for energy policy reform.

Independent Variable	Threshold Variable	Threshold Number	N	F Value	p Value	BS Times	Threshold Value	95% Confidence Interval
ETI	EPR	1	3124	44.02	0.0000	3000	21	[19.5, 22]
ETI	EPR	2	3124	8.10	0.2050	3000	21, 48	[19.5, 22], [43.5, 49]
ETI	EPR	3	3124	2.84	0.7927	3000	21, 48, 40	[19, 22], [43.5, 49], [39, 46]

, it can be seen that for the threshold test with the quantity of energy technology innovation as the threshold variable, the F-value of a single threshold value is 45.56, which passed the 1% level significance test. While both the dual and triple thresholds did not pass the significance test. According to the test results, it is obvious that in the model of the impact of energy transition on GTFP in Chinese cities, there is only a single threshold effect in the number of energy policies. Thus, the subsequent threshold regression estimation results are also only estimated using the single threshold model of energy policy reform. Meanwhile, as shown in the table, the threshold for the number of energy policies is 21, with a 95% confidence interval of [19.5, 22].

In order to more intuitively conduct the research to the threshold effect of energy policy quantity in the analysis of the impact of energy transition on urban GTFP in China, a likelihood ratio (LR) test figure was drawn (as shown in Figure 4). The horizontal axis represents the candidate threshold for the number of energy policies, the vertical axis represents the likelihood ratio statistic, and the horizontal dashed line corresponds to the critical value of the chi-square distribution at the 5% significance level of 7.35. From the figure, it can be seen that when in the range from 19.2 to 22, the LR statistic remains below the critical value, indicating a significant single threshold for the number of energy policies.

4.2.3. Analysis of the Result of the Threshold Effect of Energy Transition on Green Total Factor Productivity in Chinese Cities

Table 11. Threshold regression estimation results.

Variable	(1)	(2)	(3)	(4)
	GTFP	GTFP	GTFP	GTFP
ETI (TI < 33)	0.128 ***	0.0861 ***
	(0.0277)	(0.0315)
ETI (TI ≥ 33)	0.337 ***	0.292 ***
	(0.0321)	(0.0360)
ETI (EPR < 21)			0.119 ***	0.0653 **
			(0.0282)	(0.0325)
ETI (EPR ≥ 21)			0.307 ***	0.258 ***
			(0.0290)	(0.0325)
Controls	NO	Yes	NO	Yes
Constant	0.991 ***	0.856 ***	0.827 ***	0.990 ***
	(0.0023)	(0.0501)	(0.0500)	(0.0022)
N	3124	3124	3124	3124
R2	0.138	0.141	0.172	0.168
Individual Number	284	284	284	284

Note: ***, ** represent significance at levels of 1%, and 5%, respectively.

shows the estimation results of the threshold models for the quantity of energy technology innovation and energy policy quantity. Model (1) is the threshold regression result of the quantity of energy technology innovation without controlling variables. Model (2) is the threshold regression result of the impact of energy technology innovation on the GTFP of Chinese cities on energy transition after controlling variables are added. According to the results, obviously, in the model without controlling variables (Model (1)), when in the low-tech innovation interval (TI < 33), the coefficient of energy transition on GTFP in Chinese cities is 0.128, which is significant at the 1% level. It reflected that when energy technology innovation is at a low level, energy transition still has a significant promoting effect on GTFP; when energy technology innovation reaches a high-level range (TI ≥ 33), the impact coefficient of energy transition on urban GTFP increases to 0.337, which is 160% higher than the low-technology range, indicating that when technological capability innovation accumulates to a critical value, it can significantly amplify the promoting effect of energy transition on GTFP [52,53]. This clear single threshold effect provides strong empirical evidence to support Hypothesis 2.

Models (3) and (4) are the threshold effect test results of the impact of energy transition on urban GTFP when the EPR is regarded as the threshold variable, with a threshold value defined as 21. Model (3) indicates the threshold regression results without controlling variables, while Model (4) shows the threshold regression results with controlling variables. As shown in the table, when the number of policies does not reach the threshold value (EPR < 21), the impact coefficients of ETI on GTFP are 0.119 (Model 3) and 0.0653 (Model 4), respectively, both passing the significance test at the 1% level; and when EPR crosses 21, the coefficients increase to 0.307 (Model 3) and 0.258 (Model 4), with no significant change. This confirms the “synergy effect” hypothesis of energy policy tools [54,55,56]: a single or small number of policies are difficult to form institutional synergy, and when the number of policies accumulates to a critical scale (EPR ≥ 21), policy tools such as subsidies, regulation, and demonstration create a more favorable institutional environment for energy transition through complementary synergy, thereby amplifying the promoting effect of ETI on GTFP. Similarly, when studying the impact of China’s energy policies on green innovation. Some scholars pointed out that the collaborative allocation of policy tools needs to be based on a certain quantity and scale in order to break through the “policy island” dilemma [57], which is consistent with the logic of the EPR threshold effect in the paper. Thus, Hypothesis 3 is supported by the single threshold effect of energy policy reform.

Overall, the promotion effect of energy transition on urban green total factor productivity significantly relies on the “threshold crossing” of energy technology innovation and policy tools. This not only verifies the adaptability of Hansen’s (1999) [58] threshold regression model to nonlinear relationships but also presents the key role thresholds of technological innovation accumulation and policy tool synergy from the empirical level of Chinese cities.

4.3. Heterogeneity Analysis of Threshold Effect

4.3.1. Analysis of Regional Heterogeneity

This chapter divides Chinese cities into four major regions based on geographical location: eastern, central, western, and northeastern.

Table 12. Threshold test results by region.

City Region	Independent Variable	Variable	N	F Value	p Value	Threshold Value	95% Confidence Interval
East	ETI	TI	946	2.17	0.8320	\	\
		EPR		6.56	0.4920	\	\
Middle	ETI	TI	880	6.14	0.1040	\	\
		EPR		35.53	0.0020	32	[-, -]
West	ETI	TI	924	6.16	0.1120	\	\
		EPR		9.83	0.0320	12	[9, 13]
Northeast	ETI	TI	374	21.14	0.0060	13	[9.5, 15]
		EPR		20.87	0.0400	13	[8.5, 14]

presents the threshold effect test results of energy technology and energy policies in different regions. It can be seen from the table that there are significant regional differences in the threshold effect when energy technology innovation and energy policy quantity are regarded as threshold variables in different regions.

In the eastern region, by analyzing the F-values and corresponding p-values of TI and EPR, the research results show that the threshold effect of technology and policy is not significant in the relationship between energy transition and GTFP. It may be due to the fact that the eastern region has entered a stage of innovation driven and diversified collaborative development, with profound technological accumulation and mature policy system, making it difficult for the editorial effect of technological innovation and policy reform to present new promoting effects. Urban GTFP may rely more on factors such as resource allocation and the proportion of green industries, which fundamentally weaken the single threshold effect of technology and policies. In the central region, the single threshold effect of EPR passed the significance test at the 1% level, while TI did not pass the threshold test. Possibly, the reason is the central region is in an accelerated stage of energy transition, and policies play a key role in guiding resource concentration and breaking path dependence. The number of policies can release stronger institutional efficiency after crossing the critical value. However, due to the constraints of innovation investment and R&D platforms, technological innovation is still in the stage of “catching up-digestion”, with insufficient accumulation, making it difficult to trigger significant threshold effects, and its contribution to GTFP is mostly gradual.

In the western region, EPR’ s single threshold test passed the significance test at the 1% level, while TI observed a threshold effect. This may be because although the western region has unique energy endowments, its economic foundation is relatively weak, and policies have become a key factor in promoting transformation. After crossing the threshold, they can play a role in integrating resources and attracting investment. Due to limitations in talent and research and development capabilities, technological innovation may not have a clear threshold effect, and its promoting effect will take longer to manifest. In the Northeast region, both TI and EPR have passed the single threshold effect test, and the threshold effect is significant. This is because, as an old industrial base, the Northeast region has a relatively serious phenomenon of energy industry path locking. Breakthroughs in technological innovation can reshape production paradigms, and crossing the critical value of policy quantity can help break through institutional constraints. The significance of the dual threshold effect reflects that the synergistic effect of energy technology and energy policies can effectively promote the positive impact of energy transition on GTFP.

Table 13. Regression results of threshold by region.

	(1)	(2)	(3)	(4)
	Central	West	Northeast
ETI (TI < Threshold)			0.185 *
			(0.103)
ETI (TI ≥ Threshold)			0.730 ***
			(0.141)
ETI (TI < Threshold)	−0.0646	−0.0395		0.167
	(0.0698)	(0.0348)		(0.115)
ETI (TI ≥ Threshold)	0.301 ***	0.149 ***		0.987 ***
	(0.0761)	(0.0351)		(0.195)
Controlling Variable	YES	YES	YES	YES
Constant	0.825 ***	0.764 ***	1.003 ***	0.818 ***
	(0.0708)	(0.0681)	(0.226)	(0.221)
N	880	924	374	374
R2	0.181	0.129	0.177	0.147
Individual Number	80	84	34	34

Note: ***, * represent significance at levels of 1%, and 10%, respectively.

reveals the regional heterogeneity regression results of the impact of ETI on urban GTFP when TI and EPR are used as threshold variables. For the western region, EPR has a significant threshold effect. When EPR is less than the threshold (32), the impact coefficient of energy transition on urban GTFP is −0.0646. At this point, the impact of energy transition on GTFP is negative, while it did not pass the significance test and has no practical significance. However, when ERP crosses the threshold value (32), the coefficient of energy transition on urban GTFP changes from negative to positive, and passes the significance test at the 1% level, that is, for every 1 unit increase in ETI, the local GTFP will increase by 0.301 units. This indicates that as the intensity of energy policies increases, the effect of ERI on GTFP shifts from inhibition to significant promotion.

Model (2) also regards EPR as the threshold variable to regress the impact of ETI on GTFP in western cities. From the results, it can be seen that the regression trends in the western and central regions are roughly consistent, that is, when EPR does not cross the threshold value, ETI has a inhibitory effect on GTFP (not passing the significance test), and when EPR reaches or exceeds the threshold value (12), the coefficient of ETI changes from negative to positive. It is worth noting that although the trend of ETI’ s impact on GTFP is the same when EPR is used as the threshold variable in the central and western regions, the threshold values of EPR are different (32 in the central region and 12 in the western region), which may be due to the differences in policy demand intensity and supply capacity in different regions. Due to the high dependence on traditional industries in the central region, the demand for energy transition is complex and requires more policy synergy. In contrast, the economic foundation in the western region is weak, the industrial structure is simple, and the required policy intensity is relatively weak.

Models (3) and (4) show the impact of ETI on local urban GTFP in the Northeast region when energy technology innovation and energy policy reform are adopted as threshold variables. Combining the threshold effect test regression results, clearly, the threshold values for TI and EPR in the northeastern region are both 13. When TI and EPR are less than the threshold value (13), the impact coefficients of ETI on urban GTFP are 0.185 and 0.167, respectively. When TI and EPR reach or cross the threshold, the impact coefficient of ETI on local GTFP increases to 0.730 and 0.987, both passing the 1% significance test. This may be due to the fact that most cities in Northeast China are old industrial clusters, with a strong degree of technological path locking in the energy industry. Advanced technological innovation levels help to break traditional path locking and optimize production methods. After reaching a certain intensity, energy policies can break the constraints of traditional systems, such as the assessment of state-owned enterprise transformation. For cities in Northeast China, energy technology innovation and energy policy reform complement each other, using a dual path of “technology innovation-policy reform” to break the traditional energy model and have a significant threshold effect on ETI promoting urban GTFP.

4.3.2. Heterogeneity Analysis of Resource-Based Cities

According to the notice of the State Council on issuing the National Sustainable Development Plan for Resource-Based Cities (2013–2020), the prefecture level and above cities in China are divided into non-resource cities, resource growing cities, resource mature cities, resource-declining cities, and resource-regenerative cities according to their resource maturity, and panel threshold models are used to test the threshold effects of energy transition in different resource-based cities.

Table 14. Threshold test results and confidence intervals.

City Type	Independent Variable	Threshold Variable	N	F Value	p Value	Threshold Value	95% Confidence Interval
Resource-Growing Cities	ETI	TI	154	0.11	0.744	\	\
		EPR		0.49	0.633	\	\
Resource-Mature Cities	ETI	TI	682	8.82	0.048	10	[7.5, 11]
		EPR		7.68	0.210	\	\
Resource-declining Cities	ETI	TI	253	1.19	0.74	\	\
		EPR		12.7	0.04	14	[12, 15]
Resource-Regenerative Cities	ETI	TI	165	8.63	0.124	\	\
		EPR		21.83	0.004	46	[19.5, 23.5]
Non-Resource Cities	ETI	TI	1870	30.84	0.000	81	[66.5, 82]
		EPR		24.86	0.014	44	[41, 45]

presents the threshold effect test results for cities with different resource maturity levels. From the test results, it is obvious that there are significant differences in the threshold effects of energy technology innovation and energy policy reform for different resource-based cities. In resource growing cities, the threshold effects of TI and EPR are not significant, which may be due to the fact that the energy transition of growing cities is newly emerging, the investment in technological innovation is relatively low, and the policy system is relatively incomplete. Small changes in TI and EPR are difficult to significantly constrain the GTFP of local cities. In resource-mature cities, the threshold effect of TI is significant while EPR is not, which may be due to the good industrial foundation for energy transition in mature cities, leading to technological-innovation breakthroughs that can drive local energy transition. However, the policy system is relatively complete, making it difficult for new policies to have new impacts. The EPR threshold effect of resource-declining cities is significant, while TI is not significant. This may be due to the fact that resource-declining cities have reduced resource reserves, reduced scale of related industries, and lack of relevant transformation motivation. While technological innovation is constrained by factors such as resource endowment, talent, and facilities, making it difficult to exert its effectiveness. Relevant policies can inject momentum into energy transition through fiscal subsidies, planning guidance, and other means. The EPR threshold effect is significant in resource-regenerative cities, while TI is not significant. This may be because resource-regenerative cities have long relied on the extraction and processing of traditional energy, resulting in a relatively single industrial structure and weak technological-innovation capabilities. In the early stages of transition, due to the lack of high-level research institutions and insufficient R&D investment in cities, technological innovation is difficult to form constraints. Simultaneously, there are compatibility issues between new industries and traditional energy systems, which require a large amount of policy coordination and regulation. High-threshold energy policies can gradually promote the upgrading and transformation of the energy system through subsidies, optimizing industrial structure, and other means, injecting momentum into urban green development, which also makes energy-policy reform play a significant threshold role in driving the improvement of urban green total factor productivity.

Compared to the differentiation of resource-based cities dominated by a single threshold, non-resource-based cities TI and EPR have shown significant threshold effects. Compared to resource-based cities, non-resource-based cities naturally have diversified industrial systems because they do not rely on single resource development. In the process of energy transition, technological innovation has become a core element in breaking through energy efficiency bottlenecks. On the one hand, energy technology innovation can rapidly spread among diverse industries, driving the coordinated improvement of energy efficiency across the entire industry. On the other hand, technological innovation can improve the adaptability of energy systems to industrial demand, enhance energy utilization efficiency, and promote GTFP growth. However, energy policy reform plays a crucial role in integrating resources and coordinating systems in the process of energy transition. By establishing cross-departmental coordination mechanisms and market-oriented incentive mechanisms, EPR can effectively eliminate barriers to factor flow and guide innovative factors such as capital and technology to cluster towards green industries. When the policy intensity crosses a specific threshold, this resource integration capability will significantly amplify the promoting effect of energy transition on GTFP.

Table 15. Threshold regression results for resource-based cities.

	(1)	(2)	(3)	(4)
	Resource Mature Cities	Resource-Declining Cities	Resource-Regenerative Cities	Non-Resource Cities
ETI (TI < threshold)	0.120			0.101 **
	(0.0800)			(0.0441)
ETI (TI ≥ threshold)	0.404 ***			0.330 ***
	(0.116)			(0.0440)
ETI (EPR < threshold)		0.0560	0.0473		0.114 ***
		(0.0577)	(0.0798)		(0.0440)
ETI (EPR ≥ threshold)		0.196 ***	0.588 ***		0.329 ***
		(0.0664)	(0.135)		(0.0452)
Controlling Variable	Yes	Yes	Yes	Yes	Yes
Constant Term	1.033 ***	0.998 ***	0.530 ***	0.840 ***	0.822 ***
	(0.109)	(0.00601)	(0.191)	(0.0680)	(0.0682)
N	682	253	165	1870	1870
R2	0.123	0.156	0.262	0.155	0.152
Individual Number	62	23	15	170	170

Note: ***, ** represent significance at levels of 1%, and 5%, respectively.

shows the regression results of the impact of TI and EPR on urban GTFP in different resource-based cities. The regression results show that when regarding resource mature cities as the research object, TI exhibits a significant threshold effect. When TI is less than the threshold value (10), the impact coefficient of ETI on urban GTFP is 0.12, but it does not pass the significance test, and the economic significance is not significant. When TI reaches or crosses the threshold value, the impact coefficient of ETI on urban GTFP increases to 0.404, passing the significance test at the 1% level. This indicates that in resource mature cities, when the number of energy technology innovations reaches a certain level of quantity, it will significantly promote the increase in ETI on GTFP. In the threshold regression results of resource declining cities and resource regenerating cities, EPR shows a significant threshold effect, but the threshold values are different (threshold value 14 for declining cities and threshold value 46 for regenerating cities). This may be due to the shrinkage of industries in declining cities, where a small amount of energy policies can activate transformation. The industries in regenerative cities are diverse and difficult to integrate, requiring complex policies to promote coordinated development. In the threshold regression results of resource-declining cities and resource-regenerative cities, EPR shows a significant threshold effect, but the threshold values are different (threshold value 14 for resource-declining cities and threshold value 46 for resource-regenerative cities). This may be due to the shrinkage of industries in resource-declining cities, where a small amount of energy policies can activate transformation. The industries in resource-regenerative cities are diverse and difficult to integrate, requiring complex policies to promote coordinated development.

In the threshold regression analysis of non-resource cities, when TI is less than the threshold value, ETI has a positive promoting effect on urban GTFP. After TI reaches the threshold, the coefficient increases (from 0.101 to 0.330) and passes the significance test. When EPR is used as the threshold variable and the threshold (44) is not reached, the coefficient of influence of ETI on GTFP is 0.114, which has a promoting effect. When EPR reaches or exceeds the threshold, the coefficient of ETI significantly increases and passes the significance test, showing a dual threshold mechanism synergistic amplification effect. These empirical results validate the differences in energy transition paths between resource-based cities and non-resource-based cities.

5. Conclusions and Policy Recommendations

5.1. Conclusions

In this research, we study the impact of energy transition on GTFP in Chinese cities. By constructing benchmark regression models and panel threshold models, energy technology innovation, energy policy reform, and energy structure innovation are used as threshold variables to analyze the threshold effects of different dimensions of energy transition on China’s urban GTFP. Through empirical testing and analysis, the following conclusions are drawn.

Firstly, from the results of the benchmark regression model, it can be seen that energy transition has a promoting effect on the urban GTFP, thereby advancing the sustainable development of Chinese cities.

Secondly, the impact of energy transition on China’s urban green total factor productivity (GTFP) is not a simple linear relationship but rather exhibits significant nonlinear threshold effects. Overall, energy technology innovation and energy policy reform are key threshold variables that affect this relationship, both of which exhibit single threshold characteristics. When the corresponding threshold is crossed, the promoting effect of energy transition on GTFP is significantly enhanced, creating sustained impetus for sustainable economic development.

Thirdly, the heterogeneity of threshold effects is significantly reflected at the regional and city-type levels. From a regional perspective, eastern cities have not shown significant technological or policy threshold effects due to their deep technological accumulation and mature policy systems. The central and western cities primarily rely on the threshold crossing of energy policies. After the policy intensity reaches a critical value, the promoting effect of energy transition on GTFP shifts from weak to strong. Due to the strong dependence on traditional industrial paths, there are significant barriers to technological innovation and policy reform in cities in Northeast China. The synergistic breakthrough of the two can effectively unleash the efficiency of transformation. From the perspective of urban resource attributes, resource-growing cities have not yet exhibited a significant threshold effect, while resource-mature cities show a prominent threshold effect on technological innovation. Resource-declining and resource-regenerative cities rely more on crossing policy thresholds, while non-resource cities have dual thresholds of technology and policy. Technological breakthroughs and policy synergy can jointly amplify the green benefits of energy transition. These heterogeneous characteristics reveal the differentiated demands for technology and policies in different development stages and types of cities during the energy transition process, helping different types of cities accurately achieve sustainable development goals.

5.2. Policy Recommendations

This paper proposes the following ideas to enable all cities to seize this historical opportunity and promote the high-quality development of regional economy by virtue of energy transition.

• Accelerate the process of optimizing the energy structure: strongly promote the large-scale development and utilization of renewable energy such as wind energy, photovoltaic power generation, and hydrogen energy, and gradually reduce the proportion of fossil energy consumption. Focusing on key core technology research and development breakthroughs in the energy field: establishing a national level energy technology innovation special project, focusing on supporting the research and development of technologies such as efficient energy storage, smart grid, green hydrogen production, carbon capture, utilization and storage (CCUS), and breaking through technological bottleneck. Improve the energy policy system: formulate long-term and stable energy transformation and development plans, reasonably formulate renewable energy grid prices, improve peak valley time of use electricity prices and residential tiered electricity prices policies, and guide rational energy consumption.
• For the Northeast region where both technology and policy threshold effects are significant, a comprehensive reform pilot strategy driven by both technology and policy will be adopted. Including the establishment of national level technology projects to break path dependence, while supporting comprehensive policy packages to overcome institutional lock-in. For the central and western regions where policy thresholds are key constraints, the focus should be on building a collaborative policy tool combination (such as fiscal subsidies and market incentives) to ensure that policy intensity crosses the critical threshold. For the eastern region where the threshold effect is not significant, policies should shift towards promoting the upgrading of green industries and optimizing resource allocation efficiency.
• For resource mature cities with significant technological thresholds, their advantages should be utilized by guiding research and development investment and supporting green technology innovation. For resource-depleted and regenerative-cities that rely more on crossing policy thresholds, policy interventions should provide precise support (such as transition funds, strategic planning) to activate new momentum. For non-resource cities with significant dual thresholds for technology policies, energy-transition-related majors can be added to universities, vocational skills training can be carried out, and composite energy technology talents with both technical research and development capabilities and practical application capabilities can be cultivated to improve the city’s energy technology innovation capabilities. Simultaneously, the government needs to strengthen energy regulation, establish a unified and efficient energy regulatory platform, strengthen the supervision of the entire process of energy production, transmission, and consumption, and crack down on illegal and irregular behaviors in the energy field in order to improve the energy policy system.

5.3. Contribution and Limitations

The contributions of this paper are mainly reflected in three aspects. Methodologically, it combines the entropy weight method and the CRITIC method to calculate the weights of the ETI, avoiding the bias of a single weight method and providing a more robust methodological reference for the quantitative measurement of urban energy transition, as well as a more reliable tool for evaluating energy-driven factors of urban sustainable development. Theoretically, it verifies the single threshold effects of energy technology innovation and energy policy reform on the “energy transition–GTFP” relationship, reveals the mechanism by which the promotion effect is significantly amplified after both break through the thresholds, and refines heterogeneity from the dual perspectives of “regional geography” and “resource attribute”, filling the theoretical gap in the threshold conditions for energy transition to affect GTFP in cities at different development stages and deepening our understanding of the inherent laws of energy transition for sustainable economic development. Practically, our findings provide a basis for differentiated policy-making: eastern regions need to focus on green industry upgrading; central and western regions on strengthening policy coordination; and northeastern regions on the dual drive of technology and policy. Different resource-based cities have clarified the key points of technology or policy breakthroughs, avoiding the inefficiency of “one-size-fits-all” policies and providing precise guidance for various regions to formulate adaptive policies and promote urban sustainable development.

This study has limitations in three aspects. At the data level, energy structure optimization is only represented by the ratio of “natural gas + electricity consumption to total energy consumption”, without including segmented data of renewable energy such as wind power and photovoltaics. Simultaneously, limited by the public availability of energy data at the municipal level, this study used annual data and did not use data at a higher frequency (such as quarterly) or segmented to the industry level for shorter period dynamic analysis. To a certain extent, this limits the accurate capture of short-term policy effects and internal adjustments in industrial structure. In addition, the sample does not include the Tibet Autonomous Region, which may underestimate the green effect of energy transition and lead to regional bias; energy policies are only quantified by quantity, without distinguishing types such as command-and-control and market-incentive types, making it difficult to accurately reflect the actual effectiveness of policies. At the model-mechanism level, it does not consider the spatial spillover effect of energy transition between cities. Although we have analyzed regional heterogeneity, we have not yet used spatial econometric models (such as the spatial Durbin model) to capture spatial interaction effects between neighboring cities, such as spillover mechanisms of technological innovation diffusion and policy synergy. Furthermore, the study fails to deeply analyze the transmission chain of “energy transition-intermediate path-GTFP” and the interactive threshold of technology and policy, resulting in a weak mechanism explanation. At the perspective level, it does not distinguish differences in urban scale and administrative level, and the sample does not cover the latest data after the proposal of the “dual carbon” goal in 2020, making it impossible to capture long-term dynamic trends and affecting the timeliness and accuracy of conclusions.