Can China’s Regional Industrial Chain Innovation and Reform Policy Make the Impossible Triangle of Energy Attainable? A Causal Inference Study on the Effect of Improving Industrial Chain Resilience

Abstract

This study used a double machine learning model (based on the random forest algorithm) and spatial Durbin DIDs model to conduct quasi-natural experiments. The results are as follows: (1) innovation and reform policy regarding regional industrial chains as well as their resilience can significantly and positively address the development of China’s impossible triangle coupling of energy; (2) implementing the innovation and reform policy for regional industrial chains in other regions can have a significant positive spatial transmission effect on the impossible triangle coupling coordinated development of energy in the region; (3) regional industrial chain resilience can produce a significant positive mediating effect between the innovation and reform policy of regional industrial chains and the safety, reliability, and economic feasibility of green and clean energy systems; (4) under the counterfactual framework, the mechanism path “innovation and reform policy of the regional industry chain→regional industry chain resilience→coordination degree of impossible triangle coupling of energy” has significantly positive direct and indirect effects in both the treatment group and the control group. However, “innovation and reform policy of the regional industrial chain→regional industrial chain resilience→the energy sector’s impossible triangle coupling coordination degree” and “innovation and reform policy of the regional industrial chain→leading power of the regional industrial chain→the energy sector’s impossible triangle coupling coordination degree” have significantly positive direct and indirect effects in the treatment group, but only the direct effect is significant in the control group.

1. Introduction

In the context of global energy and environmental governance, the sustainable use and efficiency of energy have garnered widespread attention. The concept of the “impossible triangle” in macroeconomics, which posits that central bank independence, a fixed exchange rate, and free cross-border capital flow cannot coexist, is mirrored in energy economic management through the “trinity paradox”. This introduces a global problem known as the “impossible triangle” of energy: the challenge of simultaneously ensuring energy security, eliminating energy poverty while achieving accessibility, and protecting the ecological environment [1].

China, as a major emerging market and large consumer of energy, is particularly constrained by this “impossible triangle”. The 2020 World Energy Triad Index ranks China only 55th worldwide, reflecting a disparity with its economic status. Data from China’s National Bureau of Statistics for 2023 show rising primary energy consumption and a growing dependence on external energy, with oil import dependence surpassing 70%, posing risks to energy security. Environmental reports indicate that despite improvements in some pollution indicators, air quality in many areas remains substandard. Data from the power grid in summer 2023 revealed a peak demand shortfall and frequent power outages, underscoring the system’s vulnerability [2].

Since 2013, China’s “Innovative Province” policy has aimed to drive regional high-quality development and connect technological innovation with industrial development. This policy has allowed selected regions to make significant advancements in various policy mechanisms, creating a high innovation level and value-added industrial chains [3]. The central aim of this policy is to construct an ecosystem of innovation that supports high-quality industrial chain development. Addressing the “impossible triangle” of energy involves overcoming outdated industrial structures and ensuring sustainable and stable energy evolution. On the supply side, China faces challenges due to its heavy reliance on imported fossil fuels and technological bottlenecks in new energy sectors. On the demand side, the country’s industrial focus has historically led to intensive energy consumption and pollution, which contrasts with green development goals. Future technology sectors also place additional demands on the energy system [4].

Given these challenges, there is a pressing need for extensive industrial chain upgrades and adjustments. This research explores the impact of China’s regional industrial chain innovation and reform policies on the impossible triangle of energy, offering valuable insights for other developing and emerging economies facing similar energy challenges.

This study aims to provide a comprehensive analysis of China’s “impossible triangle” of energy—the challenge of simultaneously achieving an energy system that is safe and reliable, economically feasible, and environmentally sustainable. By integrating the macrocoupling of these dimensions, the research transcends traditional single-facet energy studies and approaches the systemic challenges within China’s energy framework holistically. This research innovatively incorporates the dynamics of new regional industrial chain innovations, resilience, and the impossible triangle of energy into a single framework. Such integration aids scholars and policymakers in clarifying the roles of innovative policies and industrial transformations in addressing energy challenges. Methodologically, the study moves beyond the commonly used difference-in-differences analysis by employing spatial difference-in-differences and double machine learning models to provide more unbiased and robust insights into the effects of energy system reform. This comprehensive approach not only enhances the understanding of interlinked energy issues but also sets a precedent for policy-driven research in energy economics.

2. Literature Review

2.1. Literature Review of Policy Impact on Energy

A segment of the literature has contributed to exploring the impact of governmental policy tools on the green, clean, and sustainable development of energy systems. Dogan et al. (2022) studied data from 25 environmentally friendly countries from 1994 to 2018, demonstrating that environmental taxes and the development and utilization of renewable resources are significant pathways to achieving CO₂ emission reductions [5]. Meng and Yu (2023) found that increasing the quotas for renewable energy and the price of carbon taxes negatively impacts thermal power generation enterprises, heightening their willingness to invest in renewable energy and significantly affecting emission reductions, thus promoting a green and clean energy system [6]. Appelbaum (2021) reached a different conclusion, stating that, unlike capital markets, governments often do not assess policy risk based on default probabilities [7]. Hence, businesses might exploit high government debt to reduce carbon tax costs, which in turn weakens the effectiveness of carbon tax policies, leading to increased carbon emissions and hindering sustainable development. Furthermore, Marconi D (2010) noted that unilateral environmental tax policies could accelerate technological improvements and shorten the time to mitigate environmental pollution [8]. Another segment of the literature focuses on how regional policy reforms impact mechanisms ensuring energy supply security. The “Energy Supply Security” (2000) report highlighted that EU policy reforms encouraged energy market liberalization and competition, aimed at enhancing supply diversity and reliability through market mechanisms. Additionally, the report emphasized the importance of policies in ensuring the transparency of the energy supply chain and system flexibility, as well as in managing strategic reserves during the market liberalization process [9]. González-López (2018) specifically analyzed how energy policy reforms in Mexico enhanced domestic oil and gas production capacity, attracted foreign investment, and facilitated technological progress [10]. The study also assessed how these reforms practically affected energy production, distribution, and consumption through the MuSIASEM model, revealing the critical role of policy in balancing economic growth with energy supply security. Research shows that regional policy reforms, by influencing energy production, distribution, and consumption patterns, play a key role in ensuring energy supply security and promoting sustainable development of the energy system. These studies provide insights into the complex factors that policymakers need to consider when designing and implementing energy policies. Additionally, studies addressing the economic viability of energy systems influenced by policy offer diverse perspectives. Mar and Bakken (1981) used classical control theory to construct an energy–economic model, analyzing how economic development affects energy consumption and the changes in the structure of primary energy consumption [11]. Steg et al. (2018) posited that energy demand is directly influenced by individual and household preferences and behaviors and also depends on the acceptability and effectiveness of technology, strategies, and policies [12]. Wang et al (2019) further underlined the importance not only of reducing the total energy consumption and emission intensity but also of adjusting energy consumption structures through China’s carbon emission trading policy [13]. This structural adjustment aids in achieving environmental benefits and is economically feasible. In conclusion, these studies show that the impact of energy policies on the economy is multifaceted, including structural adjustments, individual behaviors, and market incentives. Policymakers should thoroughly consider these factors to ensure that economic development is promoted while aligning with environmental sustainability goals.

2.2. Literature Review of Industrial Chain Resilience and Energy

In terms of the economic aspects of energy, Mourougane A and Wise M (2005) examined the regulatory reforms in New Zealand’s natural gas industry and found that industrial chain resilience is crucial for enhancing market efficiency and competitiveness [14]. Through these reforms, New Zealand established a more resilient and adaptive energy market, beneficial for the effective allocation of resources and the long-term viability of the energy economy. Similarly, Mitchell C and Connor P (2004) in its analysis of the EU’s natural gas market, highlighted the importance of industrial chain resilience in managing supply disruptions and market fluctuations [15]. They emphasized that a resilient industrial chain could more effectively withstand external shocks and ensure the stability of the energy supply. Tamer and Fuat (2013) in their evaluation of the reforms in the Turkish natural gas market, also confirmed a similar view, noting that industrial chain resilience is a key factor in the success of market reforms, especially in the face of political and economic uncertainties [16]. Another segment of the literature discusses the role of innovation-enabled resilience in the industrial chain on the green cleanliness of energy. The widespread application of renewable energy is a significant aspect of enhancing industrial chain resilience.

Moreover, Jeremy Rifkin (2011) emphasized that integrating energy production and storage facilities in each building can transform energy consumption patterns, thereby reducing dependence on centralized energy infrastructure, enhancing the system’s resistance to external shocks, and achieving green, clean energy production [17]. According to Frank W. Geels et al. (2017), policy support is a crucial factor in promoting the widespread application of renewable energy technologies [18]. Many countries support renewable energy projects through subsidies, tax incentives, and regulatory measures. This policy drive not only enhances the resilience of the industrial chain but also enables the energy system to transition more rapidly from traditional fossil fuels to cleaner energy options. Further, the literature discusses the impact of innovative empowerment under the resilience of the industrial chain on the green cleanliness of energy. Jeremy Rifkin (2011) posited that transforming buildings into active units of energy production and storage can advance the transformation of energy patterns and enhance resilience at urban and regional levels [17]. Additionally, Frank W. Geels et al. (2017) stressed that policy support is vital for the application of these technologies and the realization of energy transition, as these technologies can improve the adaptability and resilience of the energy system [18]. James F. Crow discussed the importance of energy diversification in reducing environmental and supply risks, proposing that a diversified energy mix can enhance the resilience of the industrial chain [19]. Darren B. et al. (2017) further noted that distributed energy systems and smart energy routers play a crucial role in ensuring the continuity and reliability of energy supply [20]. These technologies can help the energy system remain operational in the face of natural disasters or human attacks, thereby enhancing the resilience of the entire industrial chain. This review provides a comprehensive perspective, exploring the impact of industrial chain resilience on the economic feasibility, technological innovation, and security of energy and highlights the key findings of various studies.

2.3. Literature Critique

This study, based on an in-depth analysis of the existing literature, has identified several key research gaps and proposes corresponding research directions to fill these gaps: (1) Limitations in considering core research subjects: the current literature extensively focuses on environmental and energy policies, especially the impact of incentives such as carbon taxes on energy conservation and the transition to clean energy. However, there is limited research on how regional innovation policies and industrial chain transformation policies can comprehensively enhance the security, economic viability, and environmental sustainability of the energy system. Specifically, there is a significant lack of literature conducting causal inference analysis that incorporates the challenge of the “impossible triangle” of energy (i.e., balancing safety, economic, and environmental goals). (2) Limitations in the research perspective: existing studies have not integrated regional industrial chain innovation, industrial chain resilience, and energy policy goals into a single analytical framework. This leads to a lack of deep understanding of how innovative policies can optimize industrial chain structures, enhance resilience, and address the multidimensional challenges of the energy system. (3) Limitations in research methods: many studies have used traditional difference-in-differences models to estimate policy effects, but this method often overlooks spatial dimensions, potentially leading to biased assessments of policy impacts. Specifically, when considering regional interactions and the spatial distribution of industrial chains, current methods fail to adequately capture the regional heterogeneity of policy impacts. The contributions of this paper are as follows: (1) In terms of core research subjects, it includes the macro-coupling of safety, economic viability, and environmental cleanliness of the energy system to explore China’s energy “impossible triangle” issue, rather than merely studying one aspect of the energy system. (2) In terms of research perspective, this paper integrates the three elements of new regional industrial chain innovation, industrial chain resilience, and the impossible triangle of energy into a research framework, which helps scholars and policymakers clarify the role of industrial innovation policies and industrial transformation in addressing the impossible triangle of energy. (3) In terms of research methods, this paper moves beyond the commonly used difference-in-differences and other general analysis methods previously applied to energy system reforms, adopting spatial difference-in-differences models and double machine learning models for a more unbiased analysis.

3. Literature Review and Mechanism Analysis

3.1. Mechanism Analysis of the Impact of Innovation and Reform Policies of the Industrial Chain in the Region on the Coordinated Development of the Energy Sector’s Impossible Triangle

As the Chinese government advances in innovating regional innovation systems and mechanisms, it nears the complete innovation and reform of the industrial chain. By guiding the development of regional pilot policies, it fosters sustainable innovation in both industrial and supply chains. These new regional industrial chain innovation and reform policies profoundly influence the optimization of regional industrial structures, elevate the status of value chains, and enhance comprehensive innovation capacity. As fundamental elements in regional development, the economic feasibility, safety, reliability, and environmental sustainability of energy systems are crucial to the development of industrial and supply chains. Consequently, these new policies significantly impact the energy system’s “impossible triangle” by reshaping the region’s original industrial and supply chains in terms of development mechanisms, organizational forms, and business models.

From the perspective of the energy system’s supply side, these innovation and reform policies can significantly boost the innovative development and sustainable supply of the energy system. The Chinese government has recognized that structural risks in the energy system and the environmental impact of energy use are major impediments to regional development. It is thus imperative to address the “energy curse” stemming from the energy system’s impossible triangle by focusing on the supply side. Energy production and supply constitute a vast industrial chain and system that require technological innovation for support [18].

Energy production technology is among the most promising and competitive fields in industrial science and technology today. The Chinese government and the market consider the technological innovation of photovoltaic power generation, wind power, intelligent energy storage, and UHV transmission as critical core technologies that must be independently mastered. Therefore, in the context of the new regional industrial chain’s innovation and reform policies, local governments prioritize supporting the innovation and development of the energy supply-side industrial chain [21]. For instance, in 2017, Jiangsu Province issued the “Decision on Enhancing Independent Innovation Capacity and Building Innovative Provinces”. This document centers on the innovation and reform policy of the new regional industrial chain, emphasizing breakthroughs in key technologies in energy, resources, and environmental sectors to provide scientific and technological support for building a resource-conserving, environmentally friendly society [22]. The document encourages the development and application of new, clean energy technologies and energy-saving technologies to optimize the energy mix and improve energy conservation and consumption reduction. It advocates for establishing a technological development model for a circular economy in key industries and regions and increasing the supporting capacity of natural resources [23].

Local authorities are poised to strengthen the frameworks that safeguard intellectual property while fostering greater collaboration between military and civilian technological sectors [24]. They are committed to launching robust funds and platforms dedicated to sparking a systemic revolution in energy innovation. Such initiatives are designed to dismantle barriers within renewable sectors like photovoltaic and wind energy, with the double aims of slashing production costs and guaranteeing a steady stream of clean power [25].

To galvanize the energy sector with top-tier talent, an attractive suite of policies, including incentives and subsidies, will be implemented. Tailored training programs in innovation and entrepreneurship are slated to create an irresistible allure for experts, with the goal of transforming the sector into a vibrant hub that entices the brightest minds. The overarching ambition is to strike a delicate balance between safety, economic viability, and ecological integrity [26].

H1. 

The innovation and reform policy of the regional industrial chain can significantly improve the impossible triangle synergistic development of the energy sector.

H2. 

The implementation of regional industrial chain innovation and reform policies in other regions will have a positive spatial transmission effect on the impossible coordinated development of triangular energy in this region.

3.2. Mechanism Analysis of the Impact of Regional Industrial Chain Resilience on the Impossible Triangle Coordinated Development of the Energy Sector

Industrial chain resilience is regarded as the innate capacity and capability of the chain to withstand fluctuations in the external economic cycle through a complex industrial network, a redundant industrial structure, and a high level of industrial innovation. It enables the chain to rapidly recover from external shocks or to gain a superior position in the value chain by leveraging technological competitive advantages. Specifically, industrial chain resilience encompasses three aspects: resistance, recoverability, and leading power [27].

Firstly, industrial chain resistance refers to the chain’s ability to mitigate risks in the face of economic downturns, a decline in external resource and technology supply, and end-of-life cycles. Industrial chain recoverability is the capacity to swiftly adjust internal elements, entities, structures, and scales by utilizing existing technical pathways and reserves after significant impacts or disruptions [28]. This adaptability not only hedges against risks but also promptly establishes a self-healing mechanism to restore the industrial chain to a healthy state. Leading power, on the other hand, concentrates on the chain’s innovation capabilities, value creation, and dominance over competing chains [29]. The development of this leading power is chiefly demonstrated through sustained and pioneering innovation activities, which rely on continual industrial technological advancements to break free from existing technological path dependencies and achieve a high position in the value chain [30]. The mitigation of the Chinese energy system’s “impossible triangle” of safety and reliability, economic feasibility, and environmental sustainability is inextricably linked to the high-quality development and deep reinforcement of industrial chain resilience. For example, the industrial chain’s resilience is crucial for enhancing the sustainable development capacities of both the supply and demand sides of the energy system. From the supply perspective, strengthening the resilience of the energy production industry chain can ensure diversified and reliable energy sources, particularly when nonrenewable energy sources like oil, coal, and natural gas face extraction, import, and nontechnical pressures. A robust energy production chain can transition to alternative energies, providing reliable, safe, and sustainable energy. From the demand side, in the event of an energy input crisis, industrial chains with strong resistance can adapt by adjusting their industrial layouts and supply chain linkages to tackle the challenges of energy security, reliability, and economic viability [31]. For instance, the Chinese government’s “East and West Computing” project, which is designed to shift the supply chain of data elements from resource-scarce but economically advanced eastern China to the energy-abundant but less developed western regions, illustrates this approach. In the era of rapid development driven by massive data integration technologies such as big data and cloud computing, relocating the data element supply chain to energy-rich areas not only bolsters the entire industrial chain’s resistance to energy supply risks but is also pivotal for addressing the “energy curse” in less developed regions and achieving balanced regional development [32].

Industrial chain resilience profoundly impacts the collaborative development of an energy system that is “safe and reliable, economically feasible, and green and clean”. This resilience manifests in the chain’s robustness and its agility in adapting to disruptions, such as interruptions in energy supply [33]. When an industrial chain faces an energy supply interruption, it can leverage its complex, redundant structures and supply network to find suitable alternatives, thus maintaining collaborative development despite challenges [34]. In practice, industrial chain resilience brings about a transformative effect on the energy sector’s “impossible triangle”, ensuring safety and reliability, economic feasibility, and environmental sustainability through innovation and redefined value creation. Consider, for instance, Shanxi, a significant coal-producing province in China. The region historically propelled its growth through an industrial chain focused on coal exploration, mining, cleaning, and utilization [35]. However, with the central government’s directives to fulfill “double carbon” objectives, these traditional industries have undergone substantial disruptions. In response, local authorities have crafted systematic plans to sustain the development of the local industrial chain, reorienting it towards the trifecta of “safe and reliable, economically feasible, and green and clean” energy use [36].

This strategic shift involves channeling large-scale industrial activities towards green electricity and green hydrogen, replacing traditional energy sources and requiring swift augmentation of the province’s green energy supply capacity [37]. Initiatives to develop circular economy industrial parks and new energy functional zones are underway, aiming to establish a hub for comprehensive resource utilization [21]. The region is also promoting the clean use of coal and the green transformation of sectors such as coking, chemicals, and steel, bolstered by preferential tax measures, financial incentives, and state-backed special funds to encourage green, low-carbon growth. Furthermore, integrating into the national carbon trading market is a strategic move to promote the adoption of low-carbon and zero-carbon energy solutions [38]. The industrial chain’s leadership role in the energy system’s coordinated development is both foundational and revolutionary. Ongoing disruptive innovation is transforming the industrial chain’s pattern, prompting an all-encompassing reconfiguration of its structure, format, and modality [38]. The acceleration of “Industry 4.0” and the assimilation of emerging technologies such as big data, artificial intelligence, and the industrial metaverse into conventional manufacturing are shifting the entire industrial and supply chain towards a paradigm of digital intelligence, lean management, and precise monitoring [39]. These radical changes not only cement the industrial chain’s dominance in the international market but also reduce unnecessary energy consumption. The integration of high-value, tech-intensive industrial chains can transform regional and national industrial landscapes, relegating traditional energy-intensive industries to less developed regions and nations, thus elevating the region’s status in the global value chain. Moreover, enhancing the innovation and supply capabilities of the new energy sector, including photovoltaics, wind power, and hydrogen energy, enables regions to break free from the instability and high costs of traditional energy supplies while simultaneously pursuing greener practices and emissions reductions.

Therefore, the following hypothesis is put forward:

H3. 

Regional industrial chain resilience can significantly promote theenergy sector’s impossible triangle coupling coordination degree.

3.3. Mediating Effect of Industrial Chain Resilience

Regional industrial chain resilience is crucial to solving the “impossible triangle” of energy. It effectively reconciles the relationship between safety, reliability, economic viability, and environmental sustainability through technological innovation promotion, efficiency optimization, and flexibility in supply chain management. Specifically, industry chain resilience works through the following mechanisms: firstly, by incentivizing continued technological innovation and applications, such as smart grids and renewable energy technologies, to strengthen the adaptive capacity of energy systems and enhance responsiveness to environmental changes without sacrificing safety and reliability [40]; secondly, by optimizing the energy consumption structure, improving energy efficiency, and reducing environmental impacts through process improvement and material upgrading to enhance economic benefits; and finally, by bolstering industrial chain resilience so that market fluctuations and supply disruptions can be managed effectively, mitigating economic risks through flexible supply chain management and a diversified energy structure, and ensuring the stability of energy supply [41]. This coordination mechanism not only enhances the inherent capacity of the regional industrial chain but also provides theoretical support for the formulation of energy policies and promotes the coordinated development of energy system objectives, indicating that industrial chain resilience is a key strategy for alleviating the challenge of the energy sector’s impossible triangle. As a key regulatory mechanism, industrial chain resilience plays an important role in the coordinated development of the regional industrial chain and the “impossible triangle” of energy, that is, safety, reliability, economic feasibility, and green cleanliness. It enhances the resilience and autonomy of energy systems by increasing supply chain diversity and redundancy design, thereby maintaining safety and reliability [42]. At the same time, industrial chain resilience promotes technological innovation and management optimization, enhances the efficiency and economy of energy use, and reduces costs, which helps to achieve the coordination of economic and environmental objectives without sacrificing safety. In addition, industry chain resilience supports green technologies and sustainable business models, enabling companies to reduce their environmental footprints while maintaining economic efficiency [43]. This multidirectional adjustment capability means that industrial chain resilience can act as a bridge to coordinate regional industrial chain development and solve the challenge of the energy sector’s impossible triangle, achieving long-term sustainable development in a dynamic global economic environment. By strengthening industrial chain resilience, we can build a more resilient economic system that can maintain balanced development in the face of evolving market and technological challenges and achieve the integration of security and the economic and environmental goals of the energy system [44].

3.3.1. The Transmission Path for the Safety and Reliability of the Energy System

The safety and reliability of the energy system are hinged on stable supply, autonomous technological development, and rapid crisis response. This requires ensuring a stable supply of energy locally and continuing to obtain resources from external markets through competitive advantages in the industrial chain. In China, reliance on domestic energy and policies such as “West to East coal” and “West to East gas” can keep the energy system running for the time being. With the growth of future demand, the addition of new infrastructure such as computing facilities may break the existing energy balance and rely on imports or the development of new energy channels [45]. The innovation and change policy of the regional industrial chain can stabilize the energy supply chain by enhancing the resilience of the industrial chain, forming a complex redundant structure and a sensitive market response mechanism, and improving the adaptability to market fluctuations [46]. At the same time, China should improve its position in the global value chain, enhance the bargaining power of the domestic market, and ensure a sustainable supply of energy [47]. In addition, independent energy technology development is closely related to industry chain innovation, and regional industry chain policies provide policy support for the research and development of new energy technologies, such as “hydrogen Jilin” in Jilin Province and other projects, to promote new energy applications [48]. The ability to quickly perceive and respond to energy crises depends on the technological development and optimization of the energy demand-side industrial chain, and an integrated demand response (DR) can quickly establish crisis early warning and response plans, improve the risk management and resilience of the system, and ensure the long-term security and reliability of the energy system [49].

3.3.2. Transmission Paths for the Economic Viability of the Energy System

By promoting technological progress and management innovation, regional industrial chain innovation policies enhance the resilience of energy supply chains so that they can effectively respond to external shocks and reduce systemic risks. For example, Shandong Province has used its abundant offshore oilfield resources to improve oil recovery through long-term water flooding and low-salinity water flooding, demonstrating the direct impact of policy on improving innovation and economic efficiency in the energy supply chain [50]. In addition, the innovation and change policy also enhances the control of the industrial chain over the energy flow by reorganizing the industrial organization structure, introducing advanced manufacturing technology and lean management, to optimize energy consumption and improve energy efficiency, providing a solid foundation for the economic operation of the energy system. With the improvement of the efficiency of the industrial chain and the optimization of energy consumption, the financial pressure and price volatility risk of the energy system are reduced, thus improving the overall economic viability [51]. This process not only reflects the positive impact of industrial chain innovation on the economic system but also highlights its central role in building a sustainable energy system, providing a new theoretical perspective for the interaction between regional economic development and energy consumption [52].

3.3.3. The Transmission Path of Green and Clean Energy System

Regional industrial chain innovation and reform policy have significantly enhanced the green and clean development of the energy system by improving industrial chain resilience [53]. Through technological development, organizational optimization, and business model innovation, the policy has not only strengthened the industry chain’s ability to withstand shocks but has also improved its ability to respond to environmental changes. The improvement of industry chain resilience on the energy supply side focuses on the optimization of energy efficiency and the recycling of resources, which directly affects the operation mode and environmental impact of the energy system. For example, the opening of services to foreign investment has reduced pollution emissions from manufacturing by optimizing the input of resources and increasing productivity. At the same time, the increased resilience of the industrial chain has improved the environmental performance of the industrial chain by introducing efficient production technologies and energy management systems, optimizing energy consumption, and reducing pollution emissions. This resilience also enables the industry chain to maintain operational stability and efficiency through flexible supply chain management during resource price fluctuations or supply chain disruptions, creating conditions for the adoption of clean and sustainable energy technologies. Therefore, regional industrial chain innovation and change policy not only improve the resilience of the industrial chain itself but also indirectly solve the problem of the green and clean development of the energy system.

Based on the above analysis, this paper puts forward the following hypothesis:

H4. 

Industrial chain resilience plays a positive mediating effect between the innovation and reform policy of the regional industrial chain and the coordinated development of an energy system that is safe and reliable,economically feasible, and green and clean, including the following internal mechanisms:

H4a. 

The positive transmission mechanism of “innovation and reform policy of the regional industrial chain→regional industrial chain resilience→safety and reliability of the energy system”;

H4b. 

The positive transmission mechanism of “innovation and reform policy of the regional industrial chain→regional industrial chain resilience→economic feasibility of the energy system”;

H4c. 

The positive transmission mechanism of “innovation and reform policy of the regional industrial chain→regional industrial chain resilience→green and clean energy system”.

4. Research Design

4.1. Explained Variable: Impossible Triangle Coordinated Development Index of Energy (Coupling Coordination Degree, Energy)

The impossible triangle of energy considers that the three dimensions of “safety and reliability, economic feasibly-green and clean” of the energy system cannot be simultaneously satisfied and coordinated, which leads to the “triad paradox” in the energy system. To measure the synergistic development of the impossible triangle of energy (namely, the synergistic development level of the three levels of safety and reliability, economic feasibility, and greenness and cleanness), it is necessary to measure the safety and reliability, economic feasibility, and greenness and cleanness dimensions of the energy system, respectively. Therefore, this paper refers to the research of Wang Yi et al. (2023) [54] to construct the index system shown in

Table 1. Evaluation index system of the “safe and reliable—economically feasible—green and clean” impossible energy triangle collaborative development index in China’s provinces.

First-Level Indicators	Secondary Indicators	Indicator Interpretation and Calculation Method	Direction	Data Sources
Safe and reliable (Safety)	Energy production elasticity coefficient	Total energy production growth rate/GDP growth rate	+	China’s Tertiary Industry Statistical Yearbook
	Energy abundance	Total energy production/resident population	+	China Fixed Asset Investment Statistical Yearbook
	Energy industry drive levels	Reflects energy industry investment intensity	+	The China Energy Statistical Yearbook
	Degree of energy self-sufficiency	Total energy production/total energy consumption	+	China Fixed Asset Investment Statistical Yearbook
	Energy Production Diversification Index	Refer to the Shannon–Wiener Index calculation method	+	---
	Level of energy technology	Number of invention patents of industrial enterprises	+	China Statistical Yearbook
Economically viable (Economy)	Intensity of energy	Regional GDP/total energy consumption	+	The China Energy Statistical Yearbook
	Natural gas penetration	Natural gas using population/regional population	+	The Ministry of Science and Technology of China
	Electricity consumption per capita in rural areas	Rural electricity consumption/rural population	+	China Statistical Yearbook
	Energy prices	Producer price index/Consumer price index	+	China’s economic boom monthly report
	Ability to pay for electricity/%	Per capita household electricity consumption/per capita disposable income	+	China Statistical Yearbook
Green Clean (Green)	The extent of sulfur dioxide emissions	Sulfur dioxide emissions/resident population	+	China Statistical Yearbook
	Level of solid emissions	General industrial solids production/resident population	+	China Statistical Yearbook
	Forest cover/	Total area of forest land/total area of evaluation unit	+	China Land and Resources Statistical Yearbook
	Carbon intensity	Carbon dioxide emissions/GDP	+	China Statistical Yearbook
	Level of industrial pollution control	Investment in industrial pollution control has been completed	+	China Statistical Yearbook
	Level of renewable energy consumption	Renewable energy consumption/total energy consumption	+	China Statistical Yearbook

. In Table 1, the safety and reliability of the energy system are evaluated by six secondary indicators, the economic feasibility is evaluated by five secondary indicators, and the green cleanliness is evaluated by six secondary indicators. The index system is established based on the development status of China’s energy system, and it is designed to fully evaluate China’s impossible energy triangle on the basis of ensuring the availability of data. The time window is from 2009 to 2021 (13 years in total).

After constructing the index system and collecting the data, this paper uses the entropy method to evaluate the safety and reliability, economic feasibility, and greenness and cleanness of China’s provincial energy system and obtain the safety and reliability index (Safety), economic feasibility index (Economy), and green cleanliness index (Green) of China’s provincial energy system.

After obtaining the three analysis systems of safety and reliability, economic feasibility, and greenness and cleanness, this paper evaluates the impossible triangle collaborative development index of energy in China based on the coupling coordination degree model. The calculation process of the coupling coordination degree model integrates two aspects of coupling degree and coordination degree between systems. The coupling degree is used to evaluate the interaction relationship between systems, while the coupling coordination degree evaluates the mutual promotion relationship between systems based on coupling degree. The method of calculation is as follows:

$$\begin{gathered} {\mathrm{C}}_{\mathrm{i}\mathrm{t}}=3^3\sqrt{{{\mathrm{S}\mathrm{a}\mathrm{f}\mathrm{e}\mathrm{t}\mathrm{y}}_{\mathrm{t}\mathrm{i}\mathrm{j}}·{\mathrm{E}\mathrm{c}\mathrm{o}\mathrm{n}\mathrm{o}\mathrm{m}\mathrm{y}}_{\mathrm{t}\mathrm{i}\mathrm{j}}·\mathrm{G}\mathrm{r}\mathrm{e}\mathrm{e}\mathrm{n}}_{\mathrm{t}\mathrm{i}\mathrm{j}}}/({\mathrm{S}\mathrm{a}\mathrm{f}\mathrm{e}\mathrm{t}\mathrm{y}}_{\mathrm{t}\mathrm{i}\mathrm{j}}+{\mathrm{E}\mathrm{c}\mathrm{o}\mathrm{n}\mathrm{o}\mathrm{m}\mathrm{y}}_{\mathrm{t}\mathrm{i}\mathrm{j}}+{\mathrm{G}\mathrm{r}\mathrm{e}\mathrm{e}\mathrm{n}}_{\mathrm{t}\mathrm{i}\mathrm{j}}) \\ {\mathrm{T}}_{\mathrm{i}\mathrm{t}}=\mathsf{\alpha }{\mathrm{S}\mathrm{a}\mathrm{f}\mathrm{e}\mathrm{t}\mathrm{y}}_{\mathrm{t}\mathrm{i}\mathrm{j}}+\mathsf{\beta }{\mathrm{E}\mathrm{c}\mathrm{o}\mathrm{n}\mathrm{o}\mathrm{m}\mathrm{y}}_{\mathrm{t}\mathrm{i}\mathrm{j}}+\mathsf{\gamma }{\mathrm{G}\mathrm{r}\mathrm{e}\mathrm{e}\mathrm{n}}_{\mathrm{t}\mathrm{i}\mathrm{j}} \\ {\mathrm{E}\mathrm{n}\mathrm{e}\mathrm{r}\mathrm{g}\mathrm{y}}_{\mathrm{i}\mathrm{t}}=\sqrt{{{\mathrm{T}}_{\mathrm{i}\mathrm{t}}·\mathrm{C}}_{\mathrm{i}\mathrm{t}}} \end{gathered}$$

where i represents region and t represents time; ${\mathrm{C}}_{\mathrm{i}\mathrm{t}}$ represents the coupling degree, the coupling coordination degree among the security and reliability system ($\mathrm{S}\mathrm{a}\mathrm{f}\mathrm{e}\mathrm{t}\mathrm{y}$), economic feasibility system, ($\mathrm{E}\mathrm{c}\mathrm{o}\mathrm{n}\mathrm{o}\mathrm{m}\mathrm{y}$) and green and clean system ($\mathrm{G}\mathrm{r}\mathrm{e}\mathrm{e}\mathrm{n}$) in the energy system, as well as the weight of each system, which is obtained according to the entropy method in this paper. $\mathsf{\alpha },\mathsf{\beta },\mathsf{\gamma }$, The coupling and coordination degrees of the three systems can be used to represent the explained variable impossible energy triangle coordinated development index $\mathrm{E}\mathrm{n}\mathrm{e}\mathrm{r}\mathrm{g}\mathrm{y}$.

4.2. Policy Disposal Variable: Policy Disposal Variable of Innovation and Reform of New Regional Industrial Chain (DIDs)

The policy disposal variable in this paper is the new regional industrial chain innovation and reform policy that the Chinese central government has successively implemented in China’s provincial administrative division since 2013. The main purpose of this policy is to forge a resilient and innovation-driven industrial chain at the regional level in China. Therefore, this policy is called the “innovative pilot province” construction policy in China’s official documents.

The Chinese provinces selected for this policy pilot successively include Jiangsu (2013), Anhui (2013), Shaanxi (2013), Zhejiang (2013), Hubei (2016), Guangdong (2016), Fujian (2016), Sichuan (2017), Shandong (2017), Hunan (2018), and Jilin (2019). When determining the policy disposal variable (DIDs) of the new regional industrial chain innovation and reform, this paper adopts the dummy variable, that is, the selected pilot provinces are the disposal group, and the value is changed to 1. The regions that have not been selected into the pilot group are the control group, and the value of DIDs is 0.

4.3. Core Explanatory Variable: Resilience of Industrial Chain (Chain)

In this study, industrial chain resilience is an important core explanatory variable in the process of studying the mechanism of the new regional industrial chain innovation and reform policy on the impossible triangle coordinated development of China’s energy, and it is also the mechanism variable identified in this paper. According to the previous mechanism analysis, this paper focuses the research on the resilience of China’s industrial chain on the three aspects of resistance, resilience, and leading power of the industrial chain.

Most of the data in

Table 2. Evaluation index system of China’s provincial industrial chain resilience (Chain).

First-Level Indicators	Secondary Indicators	Indicator Interpretation and Calculation Method	Direction of Travel	Data Sources
Industrial chain resistance	Industrial chain communication foundation	Mobile phone penetration rate	+	Statistical Bulletin of the People’s Republic of China
		Number of Internet broadband access ports for 10,000 people	+	The China Statistical Yearbook
		Length of long-distance optical cable line per unit area	+	The China Statistical Yearbook
	Industrial chain logistics foundation	Highway mileage per square kilometer	+	The China Statistical Yearbook
		Rail miles per square kilometer	+	The China Statistical Yearbook
		Goods turnover	+	The China Statistical Yearbook
	Human capital	Total labor productivity	+	The China Statistical Yearbook
	Investment in innovation	Full-time equivalent of R&D personnel in industrial enterprises above designated size	+	The China Statistical Yearbook
		R&D expenditure of industrial enterprises above designated size/GDP	+	The China Statistical Yearbook
	Innovation output	New product sales income/main business income of industrial enterprises above designated size	+	The China Industrial Statistical Yearbook
		Number of invention patents granted	+	The China Statistical Yearbook
		Technology market turnover/GDP	+	The China Statistical Yearbook
	Industry structure	The industrial structure will be upgraded	+	The China Statistical Yearbook
		Rationalization of the industrial structure	+	The China Statistical Yearbook
Industry chain resilience	Innovative talent potential	Number of college students	+	The China Marine Statistical Yearbook
	Innovative technology flow	Technology market Regional amount of technology inflow	+	China Statistical Yearbook of Science and Technology
		Technology market Technology export geographical amount	+	China Statistical Yearbook of Science and Technology
	Innovate the flow of money	Government investment in science and technology	+	The China Statistical Yearbook
		Enterprise technology investment	+	The China Statistical Yearbook
		Foreign investment in technology	+	The China Statistical Yearbook
	Innovative forces emerge	Number of graduates from sci-tech business incubators	+	The China Torch Statistical Yearbook
		Number of graduated enterprises in university science and technology parks	+	The China Torch Statistical Yearbook
	Financial synergy	Balance of various bank loans	+	The China Statistical Yearbook
		Outstanding bank loans/GDP	+	The China Statistical Yearbook
	Industrial linkage	EG index of co-agglomeration between manufacturing and producer services	+	The China Statistical Yearbook
	Government regulation	Per capita fiscal expenditure by region	+	The China Statistical Yearbook
	Industry benefits	The corresponding cost of 100 yuan of revenue of industrial enterprises above designated size	+	The China Industrial Statistical Yearbook
		Profit margin on total assets of industrial enterprises above designated size	+	The China Industrial Statistical Yearbook
		Profit margin on operating income of industrial enterprises above designated size	+	The China Industrial Statistical Yearbook
Industrial chain leadership	Digital leadership	Number of computers per 100 people	+	The China Statistical Yearbook
		Several websites per hundred businesses	+	The China Statistical Yearbook
		Proportion of enterprises with e-commerce transaction activity	+	The China Statistical Yearbook
		E-commerce sales/GDP	+	The China Statistical Yearbook
		Telecom manufacturing main revenue/GDP	+	The China Statistical Yearbook
	High-end leadership	High-tech industry main business/GDP	+	The China Statistical Yearbook
		Total book value of machines/total number of employees of listed companies	+	The China Statistical Yearbook
		Intelligent transformation of listed enterprises	+	The China Statistical Yearbook
	Chain control	Number of listed companies	+	The China Statistical Yearbook
		Number of state-controlled enterprises	+	The China Statistical Yearbook
		Complexity of China’s provincial product export	+	The China Statistical Yearbook

come from public statistical materials, such as China Statistical Yearbook, China Information Yearbook, China Electronic Information Industry Statistical Yearbook, China Science and Technology Statistical Yearbook, China Torch Statistical Yearbook, China Tertiary Industry Statistical Yearbook, China Population and Employment Statistical Yearbook, China High-tech Industry Statistical Yearbook, and China Education Statistical Yearbook. In addition, we refer to the method of Xu and Jiang (2015) [55] to measure the optimization of industrial structure. The inverse of the Tael index for the rationalization of industrial structure is proposed by [56]. Gan et al. (2011) [49] is used as the rationalization of industrial structure. The machine book value of listed companies is derived from the Wind database. For the intelligent transformation of listed companies, Python 3.12.1 is used to crawl the word frequency of 186 special words related to digital intelligence, such as “artificial intelligence”, “smart wear”, and “deep learning” in the annual reports of A-share listed companies in China, and the annual reports of listed companies are divided into words and the total word frequency is counted. The word frequency and total word frequency of special words are summarized into regional panel data according to the province of the listed company, and the ratio of the latter two is the measurement index of the intelligent transformation of the listed company.

As for the calculation of the export complexity of Chinese provincial products, this paper refers to the [57] classical measurement method of Hausman et al. (2007) [58] $\mathrm{j}$, measures and evaluates the comprehensive technical complexity of 22 types of Chinese customs export commodities. The formula for calculating the technical complexity of commodities is as follows:

$${\mathrm{P}\mathrm{R}\mathrm{O}\mathrm{D}\mathrm{Y}}_{\mathrm{j}}=\sum _{\mathrm{i}=1}^{\mathrm{n}}\frac{\left({\mathrm{x}}_{\mathrm{i}\mathrm{j}}/{\mathrm{x}}_{\mathrm{i}}\right)}{{\sum }_{\mathrm{j}=1}^{\mathrm{n}}\left({\mathrm{x}}_{\mathrm{i}\mathrm{j}}/{\mathrm{x}}_{\mathrm{i}}\right)}{\mathrm{y}}_{\mathrm{i}}$$

where ${\mathrm{y}}_{\mathrm{i}}$ represents the per capita GDP of the region $\mathrm{i}$, ${\mathrm{x}}_{\mathrm{i}\mathrm{j}}$ represents the export volume $\mathrm{j}$ of the region $\mathrm{i}$, ${\mathrm{x}}_{\mathrm{i}}$ represents the total export volume of the region i, and ${\mathrm{P}\mathrm{R}\mathrm{O}\mathrm{D}\mathrm{Y}}_{\mathrm{j}}$ is the technological complexity of the commodity j, and on this basis, we can measure the comprehensive export technological complexity ${\mathrm{e}\mathrm{x}\mathrm{p}\mathrm{y}}_{\mathrm{i}}$ of the products of each province in China:

$${\mathrm{e}\mathrm{x}\mathrm{p}\mathrm{y}}_{\mathrm{i}}=\sum _{\mathrm{j}=1}^{\mathrm{n}}\left({\mathrm{x}}_{\mathrm{i}\mathrm{j}}/{\mathrm{x}}_{\mathrm{i}}\right){\mathrm{P}\mathrm{R}\mathrm{O}\mathrm{D}\mathrm{Y}}_{\mathrm{j}}$$

After constructing the index system and acquiring the data, this paper employs the entropy method to conduct a comprehensive evaluation of the index system, ultimately obtaining panel data for 30 provinces (excluding Tibet) in Mainland China from 2009 to 2021.

4.4. Control Variables

In this paper, the strength of intellectual property protection (Patent), government size (Gov), two-way FDI technology spillover (FDII), OFDI reverse technology spillover (OFDII)), and marketization index (Market) are selected as the control variables that have an impact on the impossible triangle of energy. The data form is panel data, which involve 30 provinces in the Chinese mainland (except Xizang) from 2009 to 2021.

4.4.1. Strength of Intellectual Property Protection (Patent)

The evaluation is carried out in three aspects: the efficiency of intellectual property procedure (the number of patent grants in various regions of China/the total number of patent grants nationwide), the strength of intellectual property legislation (the number of local intellectual-property-related legislations/the total number of legislations), and the degree of judicial completion of intellectual property (the total number of patent dispute cases concluded in various regions in the current year/the number of patent dispute cases filed nationwide). The relevant data come from the China Statistical Yearbook for Science and Technology, the China Law Yearbook, and the website of Peking University Talkie, a comprehensive database of Chinese legislative documents.

After obtaining the data, this paper uses the entropy method to comprehensively evaluate the efficiency of intellectual property procedure, the strength of intellectual property legislation, and the degree of judicial completion of intellectual property as the strength of intellectual property protection (Patent).

4.4.2. Government Size (Gov)

Government size (Gov) is the ratio of local government general public budget expenditure to regional GDP in the current year.

4.4.3. OFDI Reverse Technology Spillover (OFDII)

The measurement steps of OFDI reverse technology spillover (OFDII) are as follows:

(1)

Firstly, 20 countries and regions including Malaysia, Japan, Thailand, Singapore, Indonesia, Vietnam, Macao, Hong Kong, Ireland, Germany, Russia, France, the Netherlands, Luxembourg, Sweden, Switzerland, the United Kingdom, Canada, the United States, and Australia are measured as the input of R&D capital.

$$\mathrm{R}\&\mathrm{d} \mathrm{c}\mathrm{a}\mathrm{p}\mathrm{i}\mathrm{t}\mathrm{a}\mathrm{l} \mathrm{i}\mathrm{n}\mathrm{p}\mathrm{u}\mathrm{t}=\frac{\mathrm{A}\mathrm{n}\mathrm{n}\mathrm{u}\mathrm{a}\mathrm{l} \mathrm{G}\mathrm{D}\mathrm{P} \mathrm{b}\mathrm{y} \mathrm{c}\mathrm{o}\mathrm{u}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{y}\times {\mathrm{R}\&\mathrm{d}}_{\mathrm{G}\mathrm{D}\mathrm{P}}}{100}$$

${\mathrm{R}\&\mathrm{d}}_{\mathrm{G}\mathrm{D}\mathrm{P}}$ represent $\mathrm{R}\&\mathrm{d} \mathrm{e}\mathrm{x}\mathrm{p}\mathrm{e}\mathrm{n}\mathrm{d}\mathrm{i}\mathrm{t}\mathrm{u}\mathrm{r}\mathrm{e} \mathrm{a}\mathrm{s} \mathrm{a} \mathrm{p}\mathrm{e}\mathrm{r}\mathrm{c}\mathrm{e}\mathrm{n}\mathrm{t}\mathrm{a}\mathrm{g}\mathrm{e} \mathrm{o}\mathrm{f} \mathrm{G}\mathrm{D}\mathrm{P} \mathrm{e}\mathrm{x}\mathrm{p}\mathrm{e}\mathrm{n}\mathrm{d}\mathrm{i}\mathrm{t}\mathrm{u}\mathrm{r}\mathrm{e}$.

(2)

Secondly, the deflator of consumer price level (DCPI) is calculated.

$${\mathrm{D}\mathrm{C}\mathrm{P}\mathrm{I}}_{\mathrm{t}}=\frac{\mathrm{A}\mathrm{n}\mathrm{n}\mathrm{u}\mathrm{a}\mathrm{l} \mathrm{i}\mathrm{n}\mathrm{f}\mathrm{l}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n} \mathrm{a}\mathrm{s} \mathrm{m}\mathrm{e}\mathrm{a}\mathrm{s}\mathrm{u}\mathrm{r}\mathrm{e}\mathrm{d} \mathrm{b}\mathrm{y} \mathrm{t}\mathrm{h}\mathrm{e} \mathrm{C}\mathrm{P}\mathrm{I}}{\mathrm{B}\mathrm{a}\mathrm{s}\mathrm{e} \mathrm{y}\mathrm{e}\mathrm{a}\mathrm{r}=100}{\mathrm{D}\mathrm{C}\mathrm{P}\mathrm{I}}_{\mathrm{t}-1}$$

(3)

The constant price R&D expenditure of each country is measured as follows:

$$\begin{gathered} \mathrm{R}\&\mathrm{d} \mathrm{e}\mathrm{x}\mathrm{p}\mathrm{e}\mathrm{n}\mathrm{d}\mathrm{i}\mathrm{t}\mathrm{u}\mathrm{r}\mathrm{e} \mathrm{b}\mathrm{y} \mathrm{c}\mathrm{o}\mathrm{u}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{y}(\mathrm{b}\mathrm{a}\mathrm{s}\mathrm{e} \mathrm{y}\mathrm{e}\mathrm{a}\mathrm{r} \mathrm{c}\mathrm{o}\mathrm{n}\mathrm{s}\mathrm{t}\mathrm{a}\mathrm{n}\mathrm{t} \mathrm{p}\mathrm{r}\mathrm{i}\mathrm{c}\mathrm{e} \mathrm{b}\mathrm{a}\mathrm{s}\mathrm{i}\mathrm{s}) \\ =\frac{\mathrm{R}\&\mathrm{d} \mathrm{i}\mathrm{n}\mathrm{v}\mathrm{e}\mathrm{s}\mathrm{t}\mathrm{m}\mathrm{e}\mathrm{n}\mathrm{t} \mathrm{b}\mathrm{y} \mathrm{c}\mathrm{o}\mathrm{u}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{i}\mathrm{e}\mathrm{s}}{\mathrm{C}\mathrm{o}\mathrm{n}\mathrm{s}\mathrm{u}\mathrm{m}\mathrm{e}\mathrm{r} \mathrm{p}\mathrm{r}\mathrm{i}\mathrm{c}\mathrm{e} \mathrm{i}\mathrm{n}\mathrm{d}\mathrm{e}\mathrm{x}}\times 100 \end{gathered}$$

(4)

The R&D capital stock of each country is measured as follows:

$$\begin{gathered} {\mathrm{R}\&\mathrm{d} \mathrm{c}\mathrm{a}\mathrm{p}\mathrm{i}\mathrm{t}\mathrm{a}\mathrm{l} \mathrm{s}\mathrm{t}\mathrm{o}\mathrm{c}\mathrm{k}}_{\mathrm{b}\mathrm{a}\mathrm{s}\mathrm{e}}=\frac{\mathrm{B}\mathrm{a}\mathrm{s}\mathrm{e} \mathrm{y}\mathrm{e}\mathrm{a}\mathrm{r} \mathrm{R}\&\mathrm{d} \mathrm{e}\mathrm{x}\mathrm{p}\mathrm{e}\mathrm{n}\mathrm{d}\mathrm{i}\mathrm{t}\mathrm{u}\mathrm{r}\mathrm{e} \mathrm{b}\mathrm{y} \mathrm{c}\mathrm{o}\mathrm{u}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{y}}{\sum \frac{\mathrm{T}\mathrm{h}\mathrm{e} \mathrm{g}\mathrm{r}\mathrm{o}\mathrm{w}\mathrm{t}\mathrm{h} \mathrm{r}\mathrm{a}\mathrm{t}\mathrm{e} \mathrm{o}\mathrm{v}\mathrm{e}\mathrm{r} \mathrm{t}\mathrm{h}\mathrm{e} \mathrm{p}\mathrm{r}\mathrm{e}\mathrm{v}\mathrm{i}\mathrm{o}\mathrm{u}\mathrm{s} \mathrm{y}\mathrm{e}\mathrm{a}\mathrm{r}}{\mathrm{G}\mathrm{r}\mathrm{o}\mathrm{w}\mathrm{t}\mathrm{h} \mathrm{c}\mathrm{y}\mathrm{c}\mathrm{l}\mathrm{e}}+\mathrm{R}\mathrm{a}\mathrm{t}\mathrm{e} \mathrm{o}\mathrm{f} \mathrm{d}\mathrm{e}\mathrm{p}\mathrm{r}\mathrm{e}\mathrm{c}\mathrm{i}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}} \\ {\mathrm{R}\&\mathrm{d} \mathrm{c}\mathrm{a}\mathrm{p}\mathrm{i}\mathrm{t}\mathrm{a}\mathrm{l} \mathrm{s}\mathrm{t}\mathrm{o}\mathrm{c}\mathrm{k}}_{\mathrm{b}\mathrm{a}\mathrm{s}\mathrm{e}}=\left(1-\mathrm{D}\mathrm{e}\mathrm{p}\mathrm{r}\mathrm{e}\mathrm{c}\mathrm{i}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n} \mathrm{r}\mathrm{a}\mathrm{t}\mathrm{e}\right)\times \mathrm{R}\&\mathrm{d} \mathrm{c}\mathrm{a}\mathrm{p}\mathrm{i}\mathrm{t}\mathrm{a}\mathrm{l} \mathrm{s}\mathrm{t}\mathrm{o}\mathrm{c}\mathrm{k} \mathrm{i}\mathrm{n} \mathrm{t}\mathrm{h}\mathrm{e} \mathrm{p}\mathrm{r}\mathrm{e}\mathrm{v}\mathrm{i}\mathrm{o}\mathrm{u}\mathrm{s} \mathrm{p}\mathrm{e}\mathrm{r}\mathrm{i}\mathrm{o}\mathrm{d}\mathrm{+}\mathrm{R}\mathrm{\&}\mathrm{d} \mathrm{e}\mathrm{x}\mathrm{p}\mathrm{e}\mathrm{n}\mathrm{d}\mathrm{i}\mathrm{t}\mathrm{u}\mathrm{r}\mathrm{e} \\ {\mathrm{R}\&\mathrm{d} \mathrm{c}\mathrm{a}\mathrm{p}\mathrm{i}\mathrm{t}\mathrm{a}\mathrm{l} \mathrm{s}\mathrm{t}\mathrm{o}\mathrm{c}\mathrm{k}}_{\mathrm{b}\mathrm{a}\mathrm{s}\mathrm{e}} \mathrm{r}\mathrm{e}\mathrm{p}\mathrm{r}\mathrm{e}\mathrm{s}\mathrm{e}\mathrm{n}\mathrm{t}\mathrm{s} \mathrm{R}\&\mathrm{d} \mathrm{c}\mathrm{a}\mathrm{p}\mathrm{i}\mathrm{t}\mathrm{a}\mathrm{l} \mathrm{s}\mathrm{t}\mathrm{o}\mathrm{c}\mathrm{k} \mathrm{b}\mathrm{y} \mathrm{c}\mathrm{o}\mathrm{u}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{y} \mathrm{i}\mathrm{n} \mathrm{b}\mathrm{a}\mathrm{s}\mathrm{e} \mathrm{y}\mathrm{e}\mathrm{a}\mathrm{r} \end{gathered}$$

(5)

The measure is based on R&D capital spillovers obtained through OFDI in the base year:

$${\mathrm{R}\&\mathrm{d} \mathrm{c}\mathrm{a}\mathrm{p}\mathrm{i}\mathrm{t}\mathrm{a}\mathrm{l} \mathrm{s}\mathrm{p}\mathrm{i}\mathrm{l}\mathrm{l}\mathrm{o}\mathrm{v}\mathrm{e}\mathrm{r} \mathrm{t}\mathrm{h}\mathrm{r}\mathrm{o}\mathrm{u}\mathrm{g}\mathrm{h} \mathrm{O}\mathrm{F}\mathrm{D}\mathrm{I}}_{\mathrm{b}\mathrm{a}\mathrm{s}\mathrm{e} \mathrm{y}\mathrm{e}\mathrm{a}\mathrm{r}}=\sum \frac{{\mathrm{C}\mathrm{F}\mathrm{I}\mathrm{S}}_{\mathrm{T}}}{\mathrm{G}\mathrm{D}\mathrm{P} \mathrm{o}\mathrm{f} \mathrm{h}\mathrm{o}\mathrm{s}\mathrm{t} \mathrm{c}\mathrm{o}\mathrm{u}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{y} \mathrm{d}\mathrm{u}\mathrm{r}\mathrm{i}\mathrm{n}\mathrm{g} \mathrm{p}\mathrm{e}\mathrm{r}\mathrm{i}\mathrm{o}\mathrm{d} \mathrm{t}}\times {\mathrm{S}}_{\mathrm{O}\mathrm{F}\mathrm{D}\mathrm{I},\mathrm{t}}$$

${\mathrm{S}}_{\mathrm{O}\mathrm{F}\mathrm{D}\mathrm{I},\mathrm{t}}$ represent $\mathrm{T}\mathrm{o}\mathrm{t}\mathrm{a}\mathrm{l} \mathrm{s}\mathrm{p}\mathrm{i}\mathrm{l}\mathrm{l}\mathrm{o}\mathrm{v}\mathrm{e}\mathrm{r} \mathrm{o}\mathrm{f} \mathrm{f}\mathrm{o}\mathrm{r}\mathrm{e}\mathrm{i}\mathrm{g}\mathrm{n} \mathrm{R}\&\mathrm{D} \mathrm{c}\mathrm{a}\mathrm{p}\mathrm{i}\mathrm{t}\mathrm{a}\mathrm{l} \mathrm{o}\mathrm{b}\mathrm{t}\mathrm{a}\mathrm{i}\mathrm{n}\mathrm{e}\mathrm{d} \mathrm{b}\mathrm{y} \mathrm{t}\mathrm{h}\mathrm{e} \mathrm{h}\mathrm{o}\mathrm{s}\mathrm{t} \mathrm{c}\mathrm{o}\mathrm{u}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{y}$ $\mathrm{t}\mathrm{h}\mathrm{r}\mathrm{o}\mathrm{u}\mathrm{g}\mathrm{h} \mathrm{O}\mathrm{F}\mathrm{D}\mathrm{I} \mathrm{i}\mathrm{n} \mathrm{p}\mathrm{e}\mathrm{r}\mathrm{i}\mathrm{o}\mathrm{d} \mathrm{t}(\mathrm{c}\mathrm{u}\mathrm{m}\mathrm{u}\mathrm{l}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{v}\mathrm{e} \mathrm{o}\mathrm{f} \mathrm{R}\&\mathrm{D} \mathrm{c}\mathrm{a}\mathrm{p}\mathrm{i}\mathrm{t}\mathrm{a}\mathrm{l} \mathrm{s}\mathrm{t}\mathrm{o}\mathrm{c}\mathrm{k})$.

${\mathrm{C}\mathrm{F}\mathrm{I}\mathrm{S}}_{\mathrm{T}}$ represent $\mathrm{C}\mathrm{h}\mathrm{i}\mathrm{n}\mathrm{a}’\mathrm{s} \mathrm{f}\mathrm{o}\mathrm{r}\mathrm{e}\mathrm{i}\mathrm{g}\mathrm{n} \mathrm{i}\mathrm{n}\mathrm{v}\mathrm{e}\mathrm{s}\mathrm{t}\mathrm{m}\mathrm{e}\mathrm{n}\mathrm{t} \mathrm{s}\mathrm{t}\mathrm{o}\mathrm{c}\mathrm{k} \mathrm{i}\mathrm{n} \mathrm{h}\mathrm{o}\mathrm{s}\mathrm{t} \mathrm{c}\mathrm{o}\mathrm{u}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{i}\mathrm{e}\mathrm{s} \mathrm{d}\mathrm{u}\mathrm{r}\mathrm{i}\mathrm{n}\mathrm{g} \mathrm{t} \mathrm{p}\mathrm{e}\mathrm{r}\mathrm{i}\mathrm{o}\mathrm{d}$.

(6)

The reverse spillover R&D capital stock OFDII obtained by Chinese provinces is measured as follows:

$$\mathrm{O}\mathrm{F}\mathrm{I}\mathrm{I}=\mathrm{R}\&\mathrm{d} \mathrm{c}\mathrm{a}\mathrm{p}\mathrm{i}\mathrm{t}\mathrm{a}\mathrm{l} \mathrm{s}\mathrm{t}\mathrm{o}\mathrm{c}\mathrm{k}\times \frac{\mathrm{T}\mathrm{h}\mathrm{e} \mathrm{f}\mathrm{o}\mathrm{r}\mathrm{e}\mathrm{i}\mathrm{g}\mathrm{n} \mathrm{i}\mathrm{n}\mathrm{v}\mathrm{e}\mathrm{s}\mathrm{t}\mathrm{m}\mathrm{e}\mathrm{n}\mathrm{t} \mathrm{s}\mathrm{t}\mathrm{o}\mathrm{c}\mathrm{k} \mathrm{o}\mathrm{f} \mathrm{e}\mathrm{a}\mathrm{c}\mathrm{h} \mathrm{p}\mathrm{r}\mathrm{o}\mathrm{v}\mathrm{i}\mathrm{n}\mathrm{c}\mathrm{e} \mathrm{i}\mathrm{n} \mathrm{t} \mathrm{p}\mathrm{e}\mathrm{r}\mathrm{i}\mathrm{o}\mathrm{d}}{\mathrm{N}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}\mathrm{a}\mathrm{l} \mathrm{f}\mathrm{o}\mathrm{r}\mathrm{e}\mathrm{i}\mathrm{g}\mathrm{n} \mathrm{i}\mathrm{n}\mathrm{v}\mathrm{e}\mathrm{s}\mathrm{t}\mathrm{m}\mathrm{e}\mathrm{n}\mathrm{t} \mathrm{s}\mathrm{t}\mathrm{o}\mathrm{c}\mathrm{k} \mathrm{i}\mathrm{n} \mathrm{t} \mathrm{p}\mathrm{e}\mathrm{r}\mathrm{i}\mathrm{o}\mathrm{d}}$$

The data are from the World Bank, China’s Outward Foreign Direct Investment Bulletin, and the IFS database.

4.4.4. Technology Spillover from FDI (FDII)

The above method is adopted to measure the R&D capital stock ${\mathrm{S}}_{\mathrm{j}\mathrm{t}}$ of each country, so as to measure the technology spillover ${\mathrm{F}\mathrm{D}\mathrm{I}\mathrm{I}}_{\mathrm{i}\mathrm{t}}^{\mathrm{f}\mathrm{d}\mathrm{i}}$ obtained by each province in China from the FDI channel:

$${\mathrm{F}\mathrm{D}\mathrm{I}\mathrm{I}}_{\mathrm{i}\mathrm{t}}^{\mathrm{f}\mathrm{d}\mathrm{i}}=\frac{{\mathrm{F}\mathrm{D}\mathrm{I}}_{\mathrm{i}\mathrm{t}}}{\sum _{\mathrm{i}}{\mathrm{F}\mathrm{D}\mathrm{I}}_{\mathrm{i}\mathrm{t}}}\sum _{\mathrm{j}}\frac{{\mathrm{F}\mathrm{D}\mathrm{I}}_{\mathrm{j}\mathrm{t}}·{\mathrm{S}}_{\mathrm{j}\mathrm{t}}}{{\mathrm{K}}_{\mathrm{j}\mathrm{t}}}$$

where ${\mathrm{S}}_{\mathrm{i}\mathrm{t}}^{\mathrm{f}\mathrm{d}\mathrm{i}}$ is the technology spillover obtained through FDI channel, is the province in China, is the source country of FDI.$\mathrm{i}\mathrm{j}$.

4.4.5. Marketization Index (Market)

The marketization index (Market) comes directly from the Annual China Marketization Index Report jointly compiled by Chinese economists Fan Gang and Wang Xiaolu.

4.5. Spatial Weight Matrix

The role of the spatial weight matrix is to reflect the spatial correlation between regions and assist the spatial DIDs model to evaluate the spatial spillover effect of variables. In the research horizon of this paper, the impossible triangle of energy, the innovation and reform policy of regional industrial chain, and the resilience of industrial chain are all closely related to the economic activities between regions, so the variables are likely to form spatial spillover effect based on the spatial communication between regions.

The spatial weight matrix of economic distance is generally set according to the inverse of the gap between the per capita income levels of two provinces. The smaller the income gap between two provinces is, the larger the weight is; otherwise, the smaller the weight is. This paper constructs the weight matrix of economic distance by substituting the mean value of real GDP per capita for per capita income. The specific elements of the matrix are constructed as follows:

$${\mathrm{W}}_{\mathrm{i}\mathrm{j}}=\left\{\begin{array}{c}\frac{1}{|{\mathrm{x}}_{\mathrm{i}}-{\mathrm{x}}_{\mathrm{j}}|}\\ 0 ,\mathrm{i}\ne \mathrm{j}\end{array}\right.,\mathrm{i}=\mathrm{j}$$

In the above formula, ${\mathrm{W}}_{\mathrm{i}\mathrm{j}}$ is the spatial weight matrix and the corresponding matrix elements of regions $\mathrm{i} \mathrm{a}\mathrm{n}\mathrm{d} \mathrm{j}$, ${\mathrm{x}}_{\mathrm{i}}$ and ${\mathrm{x}}_{\mathrm{j}}$ is the average annual per capita GDP of $\mathrm{i}$ and $\mathrm{j}$ regions, respectively.

4.6. Spatial and Temporal Distribution of Core Variables

As shown in Figure 1, in 2009, the chain was higher in eastern coastal areas such as Xinjiang, Beijing, Tianjin, Shandong, Jiangsu, and Shanghai, while it was lower in western regions such as Tibet, Qinghai, Gansu, and Guizhou; in 2021, the chain was higher in eastern coastal areas such as Xinjiang, Beijing, Tianjin, Shandong, Jiangsu and Shanghai, while it was lower in western regions such as Tibet, Qinghai, Gansu, and Guizhou. The overall distribution did not change greatly: the eastern part was higher and the western part was lower. Meanwhile, in 2009, energy in 2009 was higher in Inner Mongolia, Shanxi, Liaoning, Guangdong, and Hong Kong, lower in Hunan, Guangxi, Henan, and Gansu, higher in eastern coastal areas and central areas, and higher in Xinjiang. The distribution of energy in 2021 is almost identical to that of 2009.

4.7. Model Construction

4.7.1. Spatial Difference in Differences Model

The key assumption of the applicability of the DIDs method is the hypothesis of the stability of individual treatment, that is, the treatment effect of a certain individual does not change with the treatment of other individuals. Compared with the difference-in-differences model, the spatial difference-in-differences model can not only fully control the possible spatial correlation between variables, that is, control the part of the treatment effect on an individual that changes with the treatment of other individuals, but also the potential missing variables with spatial influence. Moreover, it can further decompose the average treatment effect from three aspects of direct effect, spatial spillover effect and total effect through parameter test. Referring to the derivation process of Dubé et al. [59], three types of spatial difference-in-differences models are constructed.

Spatial lagged difference-in-differences model (SAR-DIDs):

Model 1:

$${Energy}_{it}=\rho W{Energy}_{it}+{\alpha }_1{DID}_{it-1}+{\alpha }_2{x}_{it-1}+{\mu }_i+{\epsilon }_{it}$$

(1)

Spatial error difference in differences model (SEM-DIDs):

Model 2:

$$\begin{gathered} {Energy}_{it}={\alpha }_1{DID}_{it-1}+{\alpha }_2{x}_{it-1}+{\mu }_i+{\mu }_{it} \\ {\mu }_{it}=\lambda W{\mu }_{it}+{\epsilon }_{it} \end{gathered}$$

(2)

Spatial Durbin Difference in Differences Model (SDM-DIDs):

Model 3:

$$\begin{gathered} {Energy}_{it}=\rho W{Energy}_{it}+{\alpha }_1{DID}_{it-1}+{\beta }_{1}W{DID}_{it-1}+{\alpha }_2{x}_{it-1} \\ +\beta W{x}_{it-1}+{\mu }_{i}z+{\epsilon }_{it} \end{gathered}$$

(3)

Model 1 is the spatial lag model, and the spatial regression coefficient of the explained variable is $\rho$; Model 2 represents the spatial error model, with λ representing the spatial error regression coefficient; and Model 3 is the spatial Dubin model, where $\rho$ and ${\beta }_1$ and represent the spatial regression coefficient of the explained variable and the explained variable, respectively. The spatial weight matrix is represented by W, ${\epsilon }_{it}$ and ${\mu }_{it}$ as random disturbance terms, and ${\mu }_i$ as the individual, time-fixed effect. Where, the variable symbol is the same as Model 1. In the spatial weight matrix, the economic distance weight matrix $W_e$ is selected to construct the spatial relationship of 31 provinces in China, and its elements are constructed with the reciprocal of the absolute difference between the annual mean GDP of region i and region j during the sample period. The combination of spatial weight matrix and different variables ($W_{Policy}$, $W_{\mu }$ and $W_{Energy}$) represents the spatial influence of the average effect of the variable on the coordination degree of energy impossibility triangle coupling in the city.

4.7.2. Double Machine Learning Model

This paper aims to explore the effect of industry chain resilience on energy impossibility triangle coupling coordination degree based on the strategy of “new regional industry chain innovation and change”. In this paper, a dual machine learning model is used to evaluate the policy effect of the strategy of “new regional industry chain innovation and change” to overcome the limitations of the traditional causal inference model. This model improves the accuracy and robustness of estimates through efficient variable selection and handling of nonlinear relationships (Knittel and Stolper 2021) [60]. Based on this, in order to verify hypothesis H1 and H2, and to initially explore the joint effects of regional industry chain innovation and change policies and regional industry chain resilience on the impossible triangular coordinated development of energy (the regulatory effects of policies on regional industry chain), a partially linear dual machine learning model is constructed as follows:

Model 4$:$

$$Energy={\theta }_{0}DID+f\left(x\right)+ϵ,E(\left.ϵ\right|DID,X)=0$$

(4)

$$DID=g\left(x\right)+\phi ,E\left(\left.\phi \right|X\right)=0$$

(5)

Model 5:

$$Energy={\theta }_{1}Chain+f\left(x\right)+ϵ,E(\left.ϵ\right|DID,X)=0$$

(6)

$$DID=g\left(x\right)+\phi ,E\left(\left.\phi \right|X\right)=0$$

(7)

Model 6:

$$Energy={\theta }_{2}DID\ast Chain+f\left(x\right)+ϵ,E(\left.ϵ\right|DID,X)=0$$

(8)

$$DID\ast Chain=g\left(x\right)+\phi ,E\left(\left.\phi \right|X\right)=0$$

(9)

In Model 4, if the coefficient ${\theta }_0$ is significantly positive, it indicates that the new regional innovation and reform policy of the industrial chain can promote the comprehensive improvement of the energy economy, security, and environmental sustainability, and then enhance the coupling and coordination degree of the impossible energy triangle. Verifying H1, in Model5, if the coefficient ${\theta }_1$ is significantly positive, shows that improving industrial chain toughness can significantly promote energy systems in the economic efficiency, security stability, and environmental sustainability of comprehensive improvement. Verifying H3, in Model 6, if the coefficient of ${\theta }_2$ is significantly positive, shows that the new regional industrial chain innovation change policy can enhance industrial chain toughness in improving supply chain efficiency, reducing costs, diversifying products, and strengthening cooperation sharing positive effects.

$Energy$ is the explained variable affected by the experiment; $DID$ represents whether the sample $X$ is subjected to experimental intervention, usually 0, 1 dummy variables or is a covariate that represents an individual characteristic that has not been experimentally interfered with, usually a high-dimensional vector. Usually, we directly construct a linear regression model, using X and DIDs to regression Energy, and estimate the coefficient of DID, which assumes that we already know the distribution of X. In fact, there may be collinearity in the higher-dimensional X, which is nonlinear with Energy, and there will be bias if it is simply estimated by a linear model. At this point, we apply a machine learning nonlinear model to estimate the distribution of X, where f (x) and g (x) are machine learning models. Specifically, we define $l_0\left(X\right)=E\left(\left.Energy\right|X\right)$, that is, $l_0\left(X\right)=f\left(x\right)+g\left(x\right)$, can derive $Y-l_0\left(X\right)$ = ${\theta }_0\phi +ϵ$. Therefore, the estimation ${\theta }_0$ needs $Y-l_0\left(X\right)$ = $Y-E\left(\left.Energy\right|X\right),$ and $\phi =DID-E\left(\left.DID\right|x\right)$, that is, the estimation $E\left(\left.Energy\right|X\right)$ and $E\left(\left.DID\right|x\right)$. Machine learning methods such as Lasso, ANN (artificial neural network), and random forest can be used to fit these two expectation functions.

4.7.3. Stepwise Regression Method Based on Double Machine Learning Model

The innovation and reform policy of the regional industrial chain aims to enhance the resilience of the industrial chain through technological innovation, resource allocation optimization, and innovation ecosystem cultivation. As a positive intermediary of the coupling and coordination between the industrial chain and the impossible triangle of energy (namely economic feasibility, environmental adaptability, and supply security), this strengthening of resilience is crucial to enhance the overall coordinated development of the energy system. The enhanced resilience of the industrial chain reduces production costs, improves production efficiency, enhances market adaptability, and promotes sustainable development, thus indirectly enhancing the economic feasibility of the energy system. At the same time, the stability of the industrial chain is also directly related to the security and reliability of energy supply, which ensures the continuous and stable operation of the energy system by reducing the vulnerability to production disruptions and raw material supply fluctuations. Therefore, the innovation and transformation of China’s regional industrial chain is not only a strategic measure to promote industrial upgrading but also a key factor in ensuring the coordinated development and sustainability of the energy system. Referring to Wen Zhonglin et al. [61], this paper uses the mediating factor test method, and the specific research model is as follows:

Model 7:

$$\{\begin{array}{l}\mathit{Energy} (\mathit{Safety},\mathit{Economy},\mathit{Green}{)}_{i,t+1}={\theta }_{0}DI{D}_{i,t}+g(X_{i,t})\\ U_{i,t},E(U_{i,t}|X_{i,t},Polic{y}_{i,t})=0\\ Chai{n}_{i,t}={\theta }_{1}DI{D}_{i,t}+g\left(X_{i,t}\right)+\\ U_{i,t},E(U_{i,t}|X_{i,t},DI{D}_{i,t})=0\\ \mathit{Energy} (\mathit{Safety},\mathit{Economy},\mathit{Green}{)}_{i,t+1}={\theta }_{0}DI{D}_{i,t}+{\theta }_2{ \mathit{Chain}}_{i,t}+g(X_{\mathrm{i},\mathrm{t}})+\\ U_{i,t},E(U_{i,t}|X_{i,t},DI{D}_{i,t})=0\\ DI{D}_{i,t}=m(X_{i,t})+V_{i,t},E\left(V_{i,t}\right|X_{i,t})=0\end{array}$$

(10)

In Model 10, If the coefficient ${\theta }_0$ is significantly positive, it indicates that the new regional innovation and reform policies of the industrial chain can jointly promote the economy, security, and green development of the energy system. ${\theta }_1$ represents the validity of DIDs on Chain; if it is positive, it represents the fact that the new regional industrial chain innovation and reform policy has a positive enhancement effect on the resilience of the industrial chain. If ${\theta }_2$ is positive, it means that the resilience of the industrial chain can jointly promote the economy, safety, and greenness of the energy system, which well verifies hypothesis H4.

Among them, Chain is the specific measurement index of the industrial chain toughness of enterprise i in the t + 1 year, and it serves as the intermediary variable of the industrial chain toughness. The greater the value of Chain, the greater the toughness of the enterprise industrial chain, and the more willingness to invest in innovation and research and development. This paper also uses the PLR model for estimation, and other relevant control variables remain the same as in the previous section.

4.7.4. Counterfactual Framework Model

Based on the double machine learning model, this paper examines the effect of “new regional industrial chain innovation and reform policy” on mechanism variables. Different from the traditional mediating effect analysis, the causal mediating effect analysis believes that the mediating factor $M$ depends on the disposal state $d$, while the explained variable depends on both the mediating factor $M$ and the disposal state $\mathrm{d}$. Therefore, the direct effect $\theta$ is determined by the state of individual mediating factors $M\left(d\right)$ on the basis of the change of disposal status, while the indirect effect $\delta$ is determined by the state of individual mediating factors $\mathrm{d}$ on the basis of the change in the state of mediating factors, which to a certain extent indicates that the mediating factors $M$ and the initial disposal status $d$ are “separated” from each other (counterfactual framework), and the two can be expressed as follows:

Model 8:

$$\{\begin{array}{l}\theta \left(1\right)=E\left(Y\right(1,M\left(1\right)-Y(0,M(1\left)\right)\left)\right)\\ \theta \left(0\right)=E\left(Y\right(1,M\left(0\right)-Y(0,M(0\left)\right)\left)\right)\\ \delta \left(0\right)=E\left(Y\right(1,M\left(1\right)-Y(1,M(0\left)\right)\left)\right)\\ \delta \left(0\right)=E\left(Y\right(0,M\left(1\right)-Y(0,M(0\left)\right)\left)\right)\end{array}$$

(11)

In addition to the different expressions of direct effect and indirect effect from traditional mediating effect analysis, there are also differences in interpretation. This paper focuses on examining the mechanism path of mediating factors based on causal mediating effect analysis. Therefore, the indirect effect is taken as an example for special explanation. The so-called “disposal group indirect effect” and “control group indirect effect” in this paper are not only for the pilot or nonpilot cities of “new regional industrial chain innovation and reform”. They can be further explained as the change of explained variable caused by the change of mediating factors from the state without disposal to the state after disposal when a city accepts or does not accept disposal at the beginning. Moreover, due to the different initial disposal statuses of different individuals, the indirect effect should also be different. Therefore, it may be more appropriate to name them as “disposal time connection effect” and “nondisposal time connection effect”. Of course, there are also studies that call them “total indirect effect” and “natural indirect effect”. In order to avoid readers’ doubts about the description in this paper, we explain it here.

5. Empirical Analysis

5.1. Spatial Dependence Test (Moran’s I Test)

In this paper, the global Moran’s I is used to investigate the spatial dependence of variables, and the panel data of 30 provinces (excluding Tibet) in China from 2009 to 2021 are selected to test the global Moran’s I of the coordination degree of energy impossible triangular coupling based on the spatial weight matrix of economic distance. The results are shown in

Table 3. Global Moran’s I of the coordination degree of impossible triangular coupling (Energy) in China’s provinces from 2009 to 2021.

Year	Moran’s I	Significance Level (p)	Year	Moran’s I	Significance Level (p)
2009	0.094 ***	0.000	2016	0.119 ***	0.000
2010	0.063 ***	0.000	2017	0.106 ***	0.000
2011	0.065 ***	0.000	2018	0.118 ***	0.000
2012	0.083 ***	0.000	2019	0.124 ***	0.000
2013	0.087 ***	0.000	2020	0.109 ***	0.000
2014	0.111 ***	0.000	2021	0.103 ***	0.000
2015	0.119 ***	0.000

Note: *** indicate significance at the 1% levels, respectively.

. The global Moran’s I of the impossible energy triangle coupling coordination degree (Energy) from 2009 to 2021 passed the significance test and was positive, indicating that there is a significant positive correlation between the energy of China’s provinces, and it is reasonable to establish the spatial difference in differences model to analyze the impossible energy triangle.

At the same time, in order to explore the spatial distribution characteristics of the impossible triangle coupling coordination degree of energy, this paper draws the corresponding local Moran scatter diagram (Figure 2). As shown in Figure 2, the local Moran scatter diagram of the coordination degree of impossible triangular coupling of Energy in several provinces is distributed in the first and third quadrants, indicating that there are “high–high” aggregations and “low–low” aggregations of energy in geography.

5.2. Applicability Test and Model Selection of Spatial Econometric Models

Three groups of spatial difference-in-differences (DIDs) models based on the spatial autoregressive model (SAR), spatial error model (SEM), and spatial Durbin model (SDM) were constructed to analyze the spatial effect in the previous section. To select the appropriate model for empirical analysis, a series of tests must be conducted, and the results are presented in

Table 4. Results of applicability test.

Inspection	The Value of the Statistic	Significance Level (p)
LM (of Lag)	32.838 ***	0.000
Robust LM (of Lag)	5.217 **	0.022
LM (of Error)	148.137 ***	0.000
Robust LM (of Error)	120.516 ***	0.000
Hausman Test (of Re)	777.06 ***	0.0000
Wald Test (SDM or SAR)	28.63 ***	0.0002
Wald Test (SDM or SEM)	29.82 ***	0.0001
LR Test (SDM or SAR)	30.17 ***	0.0001
LR Test (SDM or SEM)	45.66 ***	0.0000

Note: **, and *** indicate significance at the 5%, and 1% levels, respectively.

. Initially, the pretests for spatial lag and spatial error terms indicate the inclusion of spatial effects in the model is superior to the ordinary least squares estimation, suggesting the use of the spatial Durbin DIDs model (SDM-DIDs) for this paper. However, the chi-square statistic from the Hausman test for random effects is 777.206, passing the 1% level test and indicating that the model should be based on random effects. Subsequent post hoc tests of Wald and LR demonstrated that SDM would not degenerate into SAR or SEM. Based on these findings, this paper ultimately selects the spatial Durbin DIDs model as the model for empirical analysis.

5.3. Analysis of Parameter Estimation Results

In this paper, the spatial Durbin DIDs model based on random effects is used to estimate the parameters, and the local effect, neighborhood effect, and total effect are considered simultaneously. The estimated results are shown in

Table 5. Regression results.

	Energy Impossible Triangulation Coupling Coordination Degree			Energy Impossible Triangulation Coupling Coordination Degree
	Local Effect	Neighborhood Effect	Total Effect	Local Effect	Neighborhood Effect	Total Effect
DID	0.0345 ***	0.211 ***	0.245 ***	0.0122	0.184	0.196
	(6.31)	(3.13)	(3.54)	(1.21)	(1.50)	(1.56)
Chain	0.0769 *	0.626	0.549	0.00625	0.673	0.667
	(1.72)	(1.43)	(1.21)	(0.12)	(1.32)	(1.26)
DID × Chain				0.0932 **	0.0171	0.0760
				(2.56)	(0.04)	(0.16)
Legal environment	0.00761	0.0325	0.0401	0.00835	0.00406	0.00430
	(0.36)	(0.20)	(0.24)	(0.39)	(0.02)	(0.02)
ofdi reverse technology spillover 1	0.0000218 ***	0.0000878	0.0000660	0.0000141 *	0.0000390	0.0000249
	(3.08)	(1.12)	(0.83)	(1.96)	(0.53)	(0.34)
fdi technology spillover 1	0.00319 **	0.0460 **	0.0492 **	0.00433 ***	0.0568 **	0.0611 **
	(2.55)	(2.03)	(2.09)	(3.10)	(2.35)	(2.43)
Size of government 2	0.00315	1.325 **	1.328 **	0.00219	1.289 **	1.287 **
	(0.07)	(2.42)	(2.34)	(0.05)	(2.07)	(2.00)
Marketization index x	0.00883	0.000684	0.00952	0.00857	0.00200	0.00657
	(3.56)	(0.04)	(0.56)	(3.30)	(0.13)	(0.40)
Spatial
rho	0.547 ***			0.549 ***
	(5.72)			(5.67)
Variance
lgt_theta	2.231			2.217
	(14.21)			(13.99)
sigma2_e	0.000409 ***			0.000402 ***
	(12.69)			(12.64)
N	390			390
R2	0.379			0.406

Note: *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively. ( ) represents standard errors.

. The results are shown in Table 5. The estimated values of Spatial rho are 0.547 and 0.549, respectively, which are significantly positive at the level of 1%.

At the same time, the coefficients of the local effect and the neighborhood effect of DIDs are 0.0345 and 0.211, respectively, which pass the significance test of 1%. The local effect of Chain is positive, coefficient = 0.0769, p < 0.1, but the neighborhood effect is not significant, which verifies the mechanism hypothesis H3.

In addition, the local effect coefficient of the interaction term DIDs × Chain is 0.0932, which passes the significance test of 5%, indicating that the innovation and reform policy of the regional industrial chain can positively regulate the positive impact of the resilience of regional industrial chain on the coordination degree of impossible triangular coupling of regional energy. This preliminarily verified the joint effect of the innovation and reform policy of the regional industrial chain and the resilience of the regional industrial chain on the impossible triangle of regional energy. These data together validate H2.

5.4. Parallel Trend Test

In this study, the fixed-effect regression model and the addition of appropriate control variables are used to reduce the endogeneity problem that may be caused by missing variables, but the possible violation of the parallel trend hypothesis cannot be avoided. Therefore, referring to the practice of Beck et al., the following model is set for parallel trend test:

$$\begin{gathered} {Energy}_{it}=\alpha +{\theta }_1{Policy}_{i,t-5}+{\theta }_2{Policy}_{i,t-4}+\cdots + \\ {\theta }_9{Policy}_{i,t+4}+{\theta }_{10}{Policy}_{i,t+5}+\cdots +x_{it-1}+{\mu }_i+{\epsilon }_{it} \end{gathered}$$

where represents the dummy variables ${Policy}_{i,t\pm n}$ of n years before and after the implementation of the policy. If the coefficient is not significant, it indicates that the treatment group and the control group selected by Policy have parallel trends: ${Policy}_{i,t-n}$. If the coefficient is significant, it indicates that the impact of the ${Policy}_{i,t+n}$ innovation and reform policy of the new regional industrial chain is obvious. It can be seen from Figure 3 that before the implementation of the innovation and reform strategy of the regional industrial chain, there was no significant difference in the evolution trend of urban form between the treatment group and the control group. However, in the regions that implement the innovation and reform strategy of the regional industrial chain, the coordination degree of regional impossible energy triangle coupling increases significantly after the implementation of the policy.

5.5. Empirical Analysis Based on Double Machine Learning Model

5.5.1. Benchmark Regression

In this paper, the double machine learning model based on a random forest algorithm is used to estimate the parameters, and the sample split ratio is set as 1:4. The results of Models 1–3 are reported in

Table 6. Benchmark regression relationship test.

	(1)	(2)	(3)
	Energy	Energy	Energy
DID	0.0455 ***		0.0455 ***
	(5.74)		(5.74)
Chain		0.239 ***	0.239 ***
		(4.37)	(4.37)
DID × Chain			0.551 **
			(2.34)
_cons	0.000876	0.000688	0.00167
	(0.49)	(0.38)	(1.03)
Control variables	Yes	Yes	Yes
Fixed area	Yes	Yes	Yes
Fixed time period	Yes	Yes	Yes
N	390	390	390
R2	-	-	-

Note: **, and *** indicate significance at the 5%, and 1% levels, respectively. ( ) represents standard errors.

. The results of Model 1 show that the coefficient of DIDs is 0.0455, which passes the significance test of 1%, indicating that the innovation and reform policy of the new regional industrial chain can improve the coordination degree of impossible triangular coupling of energy. The results prove that the innovation policy of the regional industrial chain can stimulate technological innovation and improve efficiency. The results show that the innovation policy of the regional industrial chain can significantly reduce the cost of energy production and use and enhance the competitiveness of the energy economy. At the same time, optimizing supply chains and strengthening infrastructure have improved the security and resilience of energy systems and reduced the risk of supply disruptions. In addition, the policy has promoted the research development, and application of green and clean technologies, and the transformation to sustainable energy, realizing the coupling and coordination of the impossible triangle of energy on the whole and promoting the sustainable development of the energy sector. The mechanism hypothesis H1 is verified.

The results of Model 2 show that when the Chain Increases by one unit, Energy will increase by 0.239 units under the condition that other conditions remain unchanged. The results pass the significance test of 1%, indicating that Chain can significantly promote the level of Energy. This result proves that improving the resilience of the industrial chain (resistance, resilience, and leading force) has a significantly positive effect on the coordination of the impossible triangle of energy and promotes the comprehensive improvement of the energy system in terms of economic efficiency, security and stability, and environmental sustainability, which verifies hypothesis H3.

The results of Model 3 show that the coefficient of DIDs × Chain is 0.551, which is significantly positive at the level of 5%, indicating that the innovation and reform policy of the new regional industrial chain can positively regulate the positive effect of the resilience of regional industrial chain on the coordination degree of the impossible triangle coupling of energy. The innovation and reform policy of the industrial chain can strengthen the positive effects of the industrial chain, such as improving supply chain efficiency, reducing costs, diversifying products, and strengthening cooperation and sharing. These efforts promote the economy, security, and greenness of the energy system.

5.5.2. Analysis of Mediating Effect Mechanism and Path

Employing a double machine learning model, this paper applies stepwise regression to validate the mechanism pathway “Policy→Chain→Energy” and utilizes the Sobel, Aroian, and Goodman tests to examine the mediation effects. The results reveal that the mechanisms “DIDs→Chain→Safety” and “DIDs→Chain→Green” exhibit significant partial mediation effects, with intermediary proportions of 9.8% and 19.4%, respectively, as delineated in

Table 7. Analysis of causal mediating effect.

Path of Mediation	Dependent Variable	DID	Mediating Variables	Covariates	Fixed Area	Fixed Time	Intermediary Percentage	Sobel (Z Statistic)	Aroian (Z Statistic)	Goodman (Z-Statistic)
DIDs→Chain→Energy	Energy	0.0455 ***		Yes	Yes	Yes	13.8%	2.250 **	2.206 **	2.295 **
		(5.74)
	Chain	0.0287 ***		Yes	Yes	Yes
		(2.65)
	Energy	0.0389 ***	0.219 ***	Yes	Yes	Yes
		(4.92)	(4.25)
DIDs→Chain→Safety	Safety	0.0263 ***		Yes	Yes	Yes	9.8%	1.940 *	1.879 *	2.007 **
		(3.53)
	Chain	0.0287 ***		Yes	Yes	Yes
		(2.65)
	Safety	0.0236 ***	0.0900 ***	Yes	Yes	Yes
		(3.06)	(2.84)
DIDs→Chain→Economy	Economy	0.0218 ***		Yes	Yes	Yes	1.5% ***	2.325 **	2.288 **	2.365 **
		(3.42)
	Chain	0.0287 ***		Yes	Yes	Yes
		(2.65)
	Economy	0.0151 ***	0.220 ***	Yes	Yes	Yes
		(3.02)	(4.83)
DIDs→Chain→Green	Green	0.0129 ***		Yes	Yes	Yes	19.4%	1.780 ***	1.714 ***	1.854 ***
		(2.67)
	Chain	0.0287 ***		Yes	Yes	Yes
		(2.65)
	Green	0.0182 ***	0.0868 ***	Yes	Yes	Yes
		(3.17)	(2.40)

Note: *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively. ( ) represents standard errors.

. The Sobel, Aroian, and Goodman tests substantiate these findings. These results imply that the industrial chain plays a substantial positive mediating role between the difference-in-differences policy indicator (DIDs) and the safety and green cleanliness dimensions of the energy system. The innovation and reform of the new regional industrial chain significantly and partially mediate this relationship by enhancing the resilience of the industrial chain. The increased resilience strengthens supply chain stability, bolsters infrastructure robustness, improves crisis management capabilities, and encourages the adoption of innovative technologies, thereby elevating the security and reliability of energy systems. Furthermore, a resilient industrial chain promotes clean technology research and development, improves supply chain sustainability, emphasizes environmental responsibility, and fosters innovation-driven green development, leading to a transformative shift towards a green and clean energy system.

Coincidentally, the mediating proportion of the path DIDs→Chain→Economy is relatively low (1.5%), but Sobel, Aroian, and Goodman all pass the tests, indicating that although Chain can play an intermediary effect between DIDs and Economy, the direct effect of DIDs on Economy is limited. Although the structural adjustment and efficiency improvement of the industrial chain are important ways to show the influence of policies, DIDs still have a direct effect on the Economy, indicating that policies also have a direct impact on economic performance. As a bridge between policy and economic performance, industrial chain adjustment and innovation play a crucial role in the transformation of policy into economic power. This effect implies that although the direct impact of policies may be limited by institutional factors and market dynamics, the optimization and innovation of the industrial chain are still important channels to activate economic vitality. Policymakers expect that through the integration and intelligent upgrading of the industrial chain to promote technological innovation, improve the efficiency of resource allocation, and strengthen the synergy between enterprises, so as to promote sustainable economic development. Therefore, although the industrial chain plays a partial intermediary role in transforming policy effects into economic outcomes, the direct positive impact of policies on the economy cannot be ignored. The above analysis verifies hypothesis H4.

5.5.3. Robustness Test

In order to verify the effectiveness of the model setting and the robustness of the research results, this paper conducts a robustness test on the results of Model X, and the specific data are shown in

Table 8. Robustness test.

Mediation Path	Dependent Variable	DIDs	Mediating Variables	Covariates	Fixed Area	Fixed Time	Proportion of Intermediaries	Sobel (Z Statistic)	Aroian (Z Statistic)	Goodman (Z-Statistic)
DIDs→Chain→Energy (excluding first year)	Energy	0.0592 ***		Yes	Yes	Yes	17.5%	2.975 ***	2.937 ***	3.016 ***
		(6.25)
	Chain	0.0466 ***		Yes	Yes	Yes
		(3.77)
	Energy	0.0488 ***	0.222 ***	Yes	Yes	Yes
		(5.23)	(4.84)
DIDs→Chain→Energy (sample split changed to 1:7)	Energy	0.0465 ***		Yes	Yes	Yes	12.8%	2.232 ***	2.183 ***	2.284 ***
		(5.53)
	Chain	0.0299 ***		Yes	Yes	Yes
		(2.76)
	Energy	0.0404 ***	0.199 ***	Yes	Yes	Yes
		(4.79)	(3.80)
DIDs→Chain→Energy (sample split changed to 1:3)	Energy	0.0527 ***		Yes	Yes	Yes	11.0%	2.248 **	2.201 **	2.298 ***
		(6.34)
	Chain	0.0291 ***		Yes	Yes	Yes
		(2.72)
	Energ	0.0460 ***	0.200 ***	Yes	Yes	Yes
		(5.60)	(3.98)
DIDs→Chain→Energy (algorithm changed from random forest to gradboost)	Energy	0.0513 ***		Yes	Yes	Yes	7.8%	1.821 *	1.756 *	1.893 *
		(5.63)
	Chain	0.0271 ***		Yes	Yes	Yes
		(2.63)
	Energy	0.0469 ***	0.148 **	Yes	Yes	Yes
		(5.06)	(2.53)
DIDs→Chain→Energy (algorithm changed from random forest to support vector machine (svm))	Energy	0.0592 ***		Yes	Yes	Yes	17.1%	4.035 ***	4.011 ***	4.060 ***
		(6.25)
	Chain	0.0466 ***		Yes	Yes	Yes
		(3.77)
	Energy	0.0488 ***	0.222 ***	Yes	Yes	Yes
		(5.23)	(4.84)

Note: *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively. ( ) represents standard errors.

:

(1)

The interference in the first year of the policy implementation is excluded. Since the policy is not implemented in the whole year in the first year of the policy implementation, the inclusion of the region in the disposal group in the first year of the policy implementation may cause interference with the parameter estimation results. Therefore, the sample data in the first year of the policy implementation of the region are eliminated in the robustness test, and the parameter estimation is reconducted.

(2)

The algorithm of the double machine learning model was changed from a random forest algorithm to a support vector machine and gradient boosting algorithm.

(3)

Adjust the sample segmentation ratio (1:4 as above) to 1:7 and 1:3.

After a series of tests, the results are basically consistent with the above, indicating that the research conclusion of this paper is robust.

5.6. Heterogeneity Analysis

5.6.1. Heterogeneity Analysis of Construction Period

China launched the pilot policy for the innovation and reform of regional industrial chains in 2013, targeting the country’s provincial-level administrative divisions. Prior to the rollout of this pilot policy, the Chinese central government had already implemented a prepilot policy focused on the innovation and reform of regional industrial chains at the city level. It was only after observing some initial successes at the city level that the central government escalated the policy to the provincial level. Building on this context and referencing Yan and Zhu (2023), this paper classifies provinces with two or more cities under their jurisdiction that were selected for the city-level pilot policy as mature regions (provinces) [60]. This classification suggests that these areas were well-prepared prior to the provincial pilot policy, had developed systematic and operational experience conducive to continuing promotion, and were relatively advanced in their construction period. Other areas are designated as budding regions. This paper conducts a subsample heterogeneity test for the mechanism pathway “DIDs→Chain→Energy”. The estimated results of the double machine learning parameters for the two groups are presented in

Table 9. Heterogeneity analysis of construction Period.

Path of Mediation	Dependent Variable	Policy	Mediating Variable	Covariates	Fixed Area	Fixed Time	Intermediary Percentage	Sobel (Z Statistic)	Aroian (Z Statistic)	Goodman (Z-Statistic)
DIDs→Chain→Energy (budding provinces)	Energy	0.0224		Yes	Yes	Yes	-	-	-	-
		(1.11)
	Chain	0.0308 *		Yes	Yes	Yes
		(1.84)
	Energy	0.00715	0.999 ***	Yes	Yes	Yes
		(0.54)	(6.17)
DIDs→Chain→Energy (mature provinces)	Energy	0.0467 ***		Yes	Yes	Yes	12.5%	2.010 **	2.052 **	2.151 **
		(5.42)
	Chain	0.0425 ***		Yes	Yes	Yes
		(3.87)
	Energy	0.0410 ***	0.138 **	Yes	Yes	Yes
		(4.69)	(2.50)

Note: *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively. ( ) represents standard errors.

.

Table 9 indicates that in budding provinces, while the innovation and reform policy of the industrial chain significantly enhances the resilience of the industrial chain (coefficient = 0.0308, p < 0.1), it does not have a significant overall or direct effect on the degree of coordination within the “impossible triangle” of energy coupling. This suggests that in budding provinces, even though the policy substantially boosts the mediating variable of industrial chain resilience, this improvement does not effectively translate into the coordinated development of the energy system via the anticipated mechanism pathway. A possible explanation is that the industrial chains in budding provinces have a relatively short development period, and the corresponding policies and mechanisms are not yet fully matured, preventing structural reforms within the industrial chain from significantly impacting the coupling and coordination of the energy system in the short term. Conversely, the parameter estimation results from the mature regions indicate that the resilience of the industrial chain exerts a partial positive mediating effect of 12.5% on the relationship between the innovation and reform policy of the regional industrial chain and the “impossible triangle” of energy.

5.6.2. Heterogeneity Analysis of Geographical Areas

From the perspective of the macroregional development level, the economic and social development process of the eastern and central regions of China is quite different from that of the western region. The demand for energy in the eastern and central regions is higher than that in the western region, but the reserves of fossil energy are not as rich as that in the western region, and the development conditions of solar energy, wind energy, and other renewable energy are not as superior as those in the western region. Therefore, it is necessary to divide the research samples for China into samples from the eastern and central regions and samples from the western regions to test the possible geographical heterogeneity. The results of parameter estimation are shown in

Table 10. Heterogeneity analysis of geographical Areas.

Mediation Path	Dependent Variable	Policy	Mediating Variable	Covariates	Fixed Area	Fixed Time	Intermediary Percentage	Sobel (Z Statistic)	Aroian (Z Statistic)	Goodman (Z-Statistic)
DIDs→Chain→Energy (Eastern and Central regions)	Energy	0.0347 ***		Yes	Yes	Yes	11.5%	1.806 *	1.745 *	1.873 *
		(4.21)
	Chain	0.0255 **		Yes	Yes	Yes
		(2.26)
	Energy	0.0317 ***	0.157 ***	Yes	Yes	Yes
		(3.91)	(3.01)
DIDs→Chain→Energy (Western Region)	Energy	0.0667 ***		Yes	Yes	Yes	42.4%	2.970 ***	2.929 ***	3.013 ***
		(4.89)
	Chain	0.0586 ***		Yes	Yes	Yes
		(4.16)
	Energy	0.0361 **	0.482 ***	Yes	Yes	Yes
		(2.37)	(4.24)

Note: *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively. ( ) represents standard errors.

.

Table 10 shows that the mechanism path “DIDs→Chain→Energy” in the eastern, central, and western regions is a significantly positive mediating effect path, accounting for 11.5% and 42.4%, respectively. The results show that in the case of subsamples, the mechanism path of “DIDs→Chain→Energy” is established in both the eastern and central regions and the western regions. Moreover, from the analysis of the proportion of intermediaries, the resilience of the industrial chain in the western region plays a greater role in the mechanism effect.

5.7. Causal Mediating Effect Analysis: Counterfactual Framework

This paper uses the causal mediating effect (Model 8) analysis under the counterfactual framework and further explores the “Policy→Chain→Energy” based on the double machine learning model. The model results are shown in Table 7.

According to

Table 11. Analysis of causal mediating effect (resilience of industrial chain).

Intermediary Path	Specific Indicators	Total	Dir.Treat	Dir.Control	Indir.Treat	Indir.Control	Y (0, M (0))
DIDs→Chain→Energy	Effect	0.090	0.087	0.073	0.018	0.004	0.347
	P-val	0.000	0.000	0.000	0.000	0.102	0.000
DIDs→Resistance→Energy	Effect	0.090	0.087	0.073	0.018	0.004	0.347
	P-val	0.000	0.000	0.000	0.000	0.045	0.000
DIDs→Resilience→Energy	Effect	0.090	0.088	0.078	0.012	0.002	0.347
	P-val	0.000	0.000	0.000	0.001	0.237	0.000
DIDs→Innovation→Energy	Effect	0.090	0.088	0.080	0.010	0.003	0.347
	P-val	0.000	0.000	0.000	0.001	0.166	0.000

, the direct effect of the innovation and reform policy of the industrial chain in the disposal group on the coordination degree of the impossible triangular coupling of energy is 0.087, and the direct effect in the control group is 0.073, both significant at the level of 1%. This shows that in both the disposal group and the control group, DIDs can directly improve Energy (without the mediating variable of Chain). However, the indirect effect of the path “DIDs→Chain→Energy” in the treatment group is 0.018, which is also significant at 1% level, while the indirect effect of the path “DIDs→Chain→Energy” in the control group is not significant. This shows that in the disposal group, the mechanism path “DIDs→Chain→Energy” can operate smoothly and effectively. However, when the policy is promoted to the nonpilot areas (the control group), the resilience of the regional industrial chain cannot play a mediating effect, and the innovation and reform policy of the regional industrial chain needs to rely on other ways to promote the coupling and coordinated development of the impossible triangle of energy. The possible reason is that the current innovation and reform policy of the regional industrial chain only forms the institutional and mechanism experience of driving the resilience of the industrial chain based on the implementation region, while the industrial chain structure of the nonpilot region is different, the development mode is diversified, and the anchor point of the policy role is heterogeneous, so the current industrial chain resilience cannot play a positive mediating effect on the coordination degree of the impossible triangle coupling between policy and energy.

In addition, this paper pays special attention to three key mechanism paths, that is, “DIDs→Resistance→Energy”,“DIDs→Resilience→Energy”,and“DIDs→Innovation→Energy”, which together constitute the framework of how the innovation and reform policy of regional industrial chains plays a role in the energy system by influencing the resilience of industrial chains. The specific form is still analyzed through the causal mediating effect, and the results are reported in Table 11.

Table 11 shows that the direct effect of DIDs on Energy is significantly positive in both the treatment group with policy intervention of innovation and reform of regional industrial chain and the control group without policy intervention, which indicates that the positive effect of innovation and reform policy of regional industrial chain on energy systems is universal and can be directly promoted. At the same time, Table 11 shows that the indirect effects of the three mechanism paths of “DIDs→Resistance→Energy”, “DIDs→Resilience→Energy”, and “DIDs→Innovation→Energy” in the disposal group are significantly positive, indicating that under the current policy mechanism. The three subdimensions of industrial chain resilience (resistance, resilience, and leading power of industrial chain) can all play a positive mediating effect between the innovation and reform policy of the regional industrial chain and the impossible triangular coupling coordination degree of energy. However, in the control group, the indirect effect of mechanism path “DIDs→Resilience→Energy” and “DIDs→Innovation→Energy” is not significant, while the indirect effect of “DIDs→Resistance→Energy” is significantly positive. This shows that once the innovation and reform policy of the regional industrial chain is promoted to the control group, only the resistance of the industrial chain can play the mechanism effect, while the resilience and leading power of the industrial chain cannot play the intermediary effect. The possible reason is that the improvement of industrial chain resistance depends more on the redundancy of the internal structure of the industrial chain and the improvement of market mechanisms. The possible reason is that the improvement of industrial chain resistance is more dependent on the internal structure of the industrial chain and the improvement of the market mechanism. Therefore, it is easy to form common institutional and mechanism experience through policy reform, while the resilience and leading force of the industrial chain emphasize the technological path change and disruptive technological breakthroughs of the industrial chain.

6. Conclusions

Based on theoretical analysis and empirical testing, under the research framework of “innovation and reform policy of the regional industrial chain→industrial chain resilience→impossible triangle of energy”, this paper discusses the synergistic effect of China’s pilot policy of innovation and reform of the regional industrial chain and the forging of regional industrial chain resilience on the “safe and reliable→economically feasible→green and clean” collaborative development of the energy system. The conclusions are as follows:

(1)

The innovation and reform policy of the regional industrial chain and regional industrial chain resilience can both play a significant positive role in promoting the impossible triangle coupling coordinated development of energy in China, and the implementation of innovation and reform policy of the regional industrial chain in other regions can have a significant positive spatial transmission effect on the impossible triangle coupling coordinated development of energy in the region.

(2)

The innovation and reform policy of the regional industrial chain can positively moderate the positive impact of regional industrial chain resilience on the impossible triangle coupling coordinated development of energy, and the resilience of the regional industrial chain can produce a significant positive mediating effect between the innovation and reform policy of the regional industrial chain and the impossible triangle coupling coordinated development of energy.

(3)

The innovation and reform policy of the regional industrial chain can indirectly promote the safety, reliability, economic feasibility, and greenness and cleanness of regional energy systems by promoting the resilience of regional industrial chains. However, the mechanism path of “innovation and reform policy of the regional industrial chain→regional industrial chain resilience→economic feasibility of the energy system” is a complete mediating effect path.

(4)

Under the counterfactual framework, the mechanism path of “innovation and reform policy of regional industry chain→regional industry chain resilience→coordination degree of impossible triangle coupling of energy” has significant positive direct effect and indirect effect in both the treatment group and the control group. However, “innovation and reform policy of the regional industrial chain→regional industrial chain resilience→the energy sector’s impossible triangle coupling coordination degree” and “innovation and reform policy of the regional industrial chain→leading power of the regional industrial chain→the energy sector’s impossible triangle coupling coordination degree” have significantly positive direct and indirect effects in the treatment group, but only the direct effect is significant in the control group.

(5)

In the subsample heterogeneity analysis, the mechanism path of “innovation and reform policy of the regional industrial chain→regional industrial chain resilience→the energy sector’s impossible triangle coupling coordination degree” can be established in mature regions but cannot operate effectively in budding regions. However, this path shows a significantly positive partial mediating effect path in the eastern, central, and western regions.

7. Policy Suggestions

Based on the above conclusions, this paper puts forward the following policy recommendations:

(1)

Implementing the development plan of the differentiated industrial chain: a customized policy framework should be designed according to the development levels and industrial characteristics of different regions. For regions with mature industrial chains, high-end manufacturing and intelligent development projects should be carried out to enhance the technological content and market competitiveness of the industrial chain. For regions with lagging industrial chains, enterprises should be encouraged and attracted to invest in technological improvement and energy efficiency improvement in key links through government guidance funds, tax breaks, project support, and other measures. At the same time, a linkage mechanism between industrial development and energy efficiency improvement should be established to ensure that industrial chain development and reduced energy consumption are pursued synchronously; finally, the regional industrial chain resilience and the improvement of energy coupling and the coordination degree should be enhanced.

(2)

The establishment of inter-regional energy Internet: the government should promote the interconnection of inter-regional energy networks and realize the optimal allocation and flexible scheduling of energy resources through the construction of trans-regional energy Internet. This will help balance the supply and demand of energy between regions and enhance the stability and resilience of the entire regional energy system. Specific measures include upgrading existing power grid infrastructure, upgrading traditional power grids to smart grids, and improving the efficiency and reliability of energy supply. In addition, cross-regional clean energy connectivity should be promoted, for example, by building cross-regional sharing platforms for wind, solar, and other renewable energy generation sources. To effectively implement this policy, special funds could be set up to support key technology research and development and infrastructure construction of the energy Internet. At the same time, corresponding technical standards and operational norms should be formulated to ensure that energy systems in different regions can work together safely and efficiently.

(3)

Strengthening regional policy coordination and integrated development strategy: the government needs to coordinate regional development strategy at a higher level, ensure the consistency and coordination of regional policies, and avoid local protectionism and market segmentation. Inter-regional policy dialogue and cooperation should be encouraged, trade barriers should be reduced through policy coordination, and regional integration should be promoted. This would not only contribute to the rational allocation of resources but would also reduce transaction costs and enhance market efficiency in the region as a whole. Macrocontrol and policy guidance should be strengthened to ensure that regional development plans are in line with national goals regarding energy and environmental protection while promoting the development of a green and low-carbon economy.

8. The Marginal Contribution and Future

This study is innovative in both content and methodology. It not only discusses the incentives in environmental and energy policies but also analyzes how regional innovation policies can comprehensively enhance the security, economic feasibility, and environmental sustainability of the energy system. By introducing a spatial difference model and a dual machine learning model, this paper provides a new analytical framework for the multidimensional challenges of energy systems. The following are the potential marginal contributions of this article:

(1)

Expansion of Research Topics:

This study not only focuses on the impacts of incentive measures in environmental and energy policies, such as the effects of carbon taxes on energy conservation and clean energy transitions, but also broadly explores how regional innovation policies and industrial chain transformation policies can comprehensively enhance the security, economic viability, and environmental sustainability of energy systems.

(2)

Integration of Research Perspectives:

By integrating regional industrial chain innovation, the resilience of industrial chains, and the energy “impossible trinity” (balancing security, economic, and environmental goals), this paper provides a framework for scholars and policymakers to deeply understand how to address the multidimensional challenges of energy systems through industrial innovation policies and transformations.

(3)

Innovation in Methodology:

This study introduces spatial difference models and double machine learning models, going beyond traditional difference models, to more accurately analyze the impacts of energy system reforms and handle the heterogeneity in regional interactions and spatial distribution of industrial chains.

In discussing the complex issues of sustainable development of energy systems, this study combines the perspective of macrodata analysis and industrial chain resilience and reveals how policies affect the “impossible triangle” of energy through these mechanisms—and the balance between security, economy and environment. In addition, we highlight the limitations of data acquisition and suggest that future research should further deepen our understanding of this issue by adopting more advanced technologies and considering emerging market-specific factors, thus proposing future perspectives:

(1)

The research data are limited by availability and mainly focus on the macrolevel. In the future, the microdata of enterprises can be obtained by means of network crawlers, theme analysis, and other means to explore the role of innovation and reform on the impossible energy triangle from the microperspective.

(2)

The article takes the resilience of the industrial chain as the mechanism factor of policy transmission to the energy system. At the same time, it can also focus on the role of finance, business environment, and innovation ecosystem in China and other emerging markets in the energy system in the future.

(3)

The impossible energy triangle is the current major sustainable development issue. According to this paper, only from the perspective of regional innovation policy can the energy ternary paradox’s possible path be alleviated. The future can be divergent from other more microscopic points of view analysis; at the same time, the sustainability of energy systems is a multidimensional problem, involving economic, environmental, social, and technical levels. Future studies should consider the interaction of these factors and explore more effective strategies for multidomain policy synergy.