New Perspectives on the Causes of Stagnation and Decline in the Sharing Economy: Application of the Hybrid Multi-Attribute Decision-Making Method

Abstract

Against the backdrop of global economic changes and rapid technological innovation, the sharing economy model is gradually transforming the operational mechanisms of traditional industries. However, some industries have experienced stagnation and recession during this transition, leading to market development constraints. The necessity of this study lies in filling the gap in the existing literature by conducting an in-depth analysis of the critical factors contributing to industrial stagnation and recession in the sharing economy. This study aims to provide concrete countermeasures for businesses and policymakers. The novelty of this research study lies in integrating multiple key variables affecting industrial development, including green production concepts, the circular economy, large-scale production, high-quality product demand driven by industrial automation, the sharing economy, and smart production. By employing multi-criterion decision-making methods, we quantitatively assess the impact of these factors more accurately. This study employs the Multi-Attribute Decision-Making (MADM) model, integrating the Decision-Making Trial and Evaluation Laboratory (DEMATEL) and the Analytic Network Process (ANP) to form D&ANP for analytical research. Highly automated industries are selected as the research subjects. The DEMATEL technique is used to construct the Influential Network Relationship Map (INRM), while the ANP concept is incorporated to develop the D&ANP model. Through the D&ANP method, influential weights are calculated and combined with industry-specific assessments of the suitability of potential causes (or attributes) contributing to economic stagnation and recession to determine the average performance values for each industry. These values are further compared with benchmark suitability performance values to distinguish ideal and non-ideal conditions across industries facing economic stagnation and recession. The analysis results indicate that different industries are influenced by varying factors, requiring strategic adjustments based on their unique development environments. Accordingly, this study provides industry-specific recommendations to optimize business models and resource allocation, mitigate the risks of economic stagnation and recession, and promote sustainable industrial development and economic recovery. The findings of this study not only contribute to empirical research on the impact of the sharing economy on industrial development but also serve as a decision-making reference for businesses. By offering strategic insights, enterprises can better respond to market dynamics, enhance competitiveness, and ensure long-term stable growth.

1. Introduction

The concept of ESG (environmental, social, and governance) can be traced back to 2004 and has become an important international standard for measuring corporate green and sustainable development. ESG evaluates whether corporate actions align with long-term development goals in terms of their impact on the environment, social interests, and governance practices. However, as global economic growth continues to slow down and stagnate, growth rates are declining, making it increasingly difficult to accurately identify the root causes of economic stagnation or recession [1].

With ongoing industrialization, social issues, climate change, and corporate governance have gradually become focal points of public concern [2]. Therefore, companies are increasingly being evaluated based on their environmental protection efforts, social responsibility practices, and governance structures. For instance, environmental sustainability indicators such as ISO 14001 and EMAS (Eco-Management and Audit Scheme) and carbon footprint assessment methods serve to strengthen the evaluation framework for sustainable industrial practices [3]. ISO 14001 focuses on environmental management systems, ensuring that businesses adopt a systematic approach to minimizing their environmental impact. EMAS further integrates environmental performance assessments and public reporting mechanisms, promoting corporate transparency and continuous improvement. Additionally, carbon footprint assessment methods are essential to quantifying greenhouse gas emissions, helping companies identify key emission sources and develop carbon reduction strategies. The integration of these standards enhances the measurement of environmental sustainability in highly automated industries, providing a comprehensive approach to industrial sustainability assessment. Therefore, businesses must consider both economic benefits and the long-term sustainability of the environment and cultural aspects in their decision-making processes [4].

As the global sustainability agenda progresses, an increasing number of international organizations, businesses, and investment institutions have begun adopting ESG standards. In 2015, the United Nations proposed the 2030 Sustainable Development Goals (SDGs) [5], further emphasizing the importance of ESG and its close connection with sustainable development [6].

The primary benefits of industrial automation technology include increased productivity, reduced unit costs, improved quality, the introduction of new technologies, reduced labor input, improved working environments, and enhanced precision. The economic forces driving these benefits are worth further exploration, as they can help address rising labor costs, wages, environmental pollution, and other production expenses while improving product quality and enhancing international competitiveness [7].

In light of this, this study aims to explore, from the perspective of different industries, the rise of green production concepts, the promotion of the circular economy, the demand for large-scale production and high-quality products driven by industrial automation, the emergence of the sharing economy, and the advancement of smart production. By analyzing these dimensions and attributes, this study seeks to examine their impact on economic stagnation and recession, with the goals of identifying the root causes of these issues and proposing viable solutions.

With the widespread adoption of the circular economy concept, consumer purchasing behavior is increasingly influenced by environmental awareness, emphasizing the reuse value of products. This shift has led to a decline in demand for new products and a supply–demand imbalance. The phenomenon of market oversupply not only weakens the operational performance of many manufacturing enterprises but has also become one of the key causes of economic stagnation and recession [8].

Currently, the global economy faces severe imbalances and oversupply issues, with most people’s living standards stagnating and economic growth showing no significant improvement [9]. Mass production and consumption have become defining characteristics of 21st-century economic development. However, while increased production and consumption may stimulate short-term economic growth, the relentless pursuit of economic expansion has long overlooked environmental and ecological sustainability, leading to environmental degradation and resource depletion.

In recent years, as environmental awareness has risen, governments worldwide have introduced more comprehensive environmental protection regulations. In response, businesses have increasingly adopted the green supply chain to ensure that their products comply with environmental standards [10].

The existing natural resources and environmental capacity can no longer sustain the traditional linear economic development model, leading to the emergence of the circular economy (CE) concept [11]. The core idea of the circular economy is to reduce resource waste and environmental impact by promoting resource reuse, remanufacturing, and material circulation, thereby achieving both economic and ecological benefits. This approach aims to drive sustainable development for both the economy and the environment.

The sharing economy is an economic model based on resource sharing and collaborative consumption, where digital platforms connect idle resources or services with users in need, improving resource utilization efficiency and reducing transaction costs. This model typically relies on online platforms, social networks, and data analytics to facilitate supply–demand matching, allowing individuals or businesses to share goods, services, knowledge, or technology rather than simply owning or purchasing them [12].

The sharing economy is characterized by four key attributes [13]: network collaboration, social commerce, the promotion of online sharing, and consumer ideology. As noted in [14], “Industry 4.0” is fundamentally transforming the global manufacturing landscape. Smart factories not only optimize traditional manufacturing processes by enhancing production efficiency but also give rise to new business models. The integration of the Internet of Things (IoT) and information and communication technology (ICT) enables comprehensive technological applications across supply chains and production networks, forming a highly intelligent and integrated ecosystem.

This transformation raises a critical question: does Industry 4.0 technology play a significant role in improving sustainable performance? The importance of green supply chain collaboration, circular economy practices, technological readiness, and environmental resilience deserves in-depth exploration.

Traditionally, Multi-Attribute Decision-Making (MADM) models have been widely applied across various fields to evaluate and compare multi-criterion decision problems. However, existing studies primarily focus on static industries or single-attribute decision analysis, for example, Enhancing DVNN-WCSM Technique for Double-Valued Neutrosophic Multiple Attribute Decision-Making in Digital Economy: A Case Study on Enhancing the Quality of Development of Henan’s Cultural and Tourism Industry. Under the drive of the digital economy, the high-quality development evaluation of Henan’s cultural and tourism industry falls under the category of Multi-Attribute Decision Making (MADM). This study proposes the Double-Valued Neutrosophic Number CSM (DVNN-CSM) and Weighted CSM (DVNN-WCSM) techniques, integrating Double-Valued Neutrosophic Sets (DVNSs) for evaluation to enhance accuracy and effectiveness [15]. Another example is Enhanced Decision-Making Technique for Innovation Capability Evaluation in the Core Industries of Digital Economy under Double-Valued Neutrosophic. The evaluation of innovation capability in the core industries of the digital economy is a Multi-Attribute Decision-Making (MADM) problem. In recent years, the TODIM and TOPSIS methods have been applied to address this issue. The following study utilizes Double-Valued Neutrosophic Sets (DVNSs) to represent fuzzy data and proposes the DVNN-TODIM-TOPSIS approach, which is validated through a numerical case study in innovation capability evaluation [16]. Sets Advancing sustainability in the automotive sector: A critical analysis of environmental, social, and governance (ESG) performance indicators. Although previous studies have explored the evaluation and ranking of ESG KPIs in the automotive industry, new methods are still needed to determine the priority and interrelationships of key factors. This study identifies ESG indicators and applies Fuzzy DEMATEL and Fuzzy TOPSIS to analyze the prioritization and causal relationships of ESG KPIs in the automotive sector, addressing existing research gaps. These cases mostly adopt Multi-Attribute Decision-Making (MADM) models such as Fuzzy DEMATEL. Therefore, this study utilizes D&ANP (DEMATEL and ANP) as its research methodology [1].

Despite these advancements, research on highly automated industries within the sharing economy—particularly addressing their unique dynamic variables, inter-industry dependencies, and feedback mechanisms—remains insufficient. This gap serves as the primary motivation for this study.

To address these challenges, this research study adopts the MADM approach and further integrates the D&ANP (DEMATEL and ANP) method, aiming to overcome the limitations of traditional decision models in handling multi-layered dependencies and complex influence relationships.

This study utilizes the MADM framework to systematically identify key dimensions and attributes influencing economic stagnation and recession. The D&ANP method is applied to evaluate the interrelationships and significance of these dimensions and attributes, ultimately establishing a more precise and decision-oriented analytical model. Additionally, this study focuses on highly automated industries, including A: the 3C industry; B: the machinery and equipment industry; C: the optoelectronics industry; and D: the metal manufacturing industry. Factors such as technological advancements, labor market changes, smart production, and the impact of the sharing economy must be considered holistically.

In the real world, when addressing the aforementioned issues, evaluation systems often exhibit complex interdependencies. Traditional AHP or standalone ANP methods struggle to fully capture the interactions among these factors [17]. While DEMATEL excels at identifying the influence relationships between evaluation systems and inferring causal relationships, it is unable to effectively determine their relative importance [18].

Furthermore, in highly automated industries, factors such as technological advancements, labor market changes, smart production, and the impact of the sharing economy must be considered holistically. Therefore, this study employs a hybrid MADM model to provide a more objective analysis of industry adaptability and strategic decision-making recommendations. The findings aim to offer both theoretical and practical insights to support the development of highly automated industries in the sharing economy environment.

Research Objectives

• Identify key factors of economic stagnation and recession:
  • Utilize the DEMATEL and ANP (D&ANP) method to identify 5 key dimensions and 14 attributes influencing economic stagnation and recession.
  • Analyze the interrelationships and significance of these factors and integrating the two methods to generate D&ANP influential weights, thereby establishing a more comprehensive decision-making analysis framework.
• Assess industry suitability and perform a comparative analysis:
  • Apply D&ANP influential weights in conjunction with expert evaluations of the suitability of potential causes (or attributes) contributing to economic stagnation and recession.
  • Calculate the average performance values of attribute suitability and compare them with benchmark suitability performance values to distinguish ideal and non-ideal conditions across different industries in economic stagnation and recession scenarios.
• Optimize industry strategies and provide improvement recommendations:
  • Based on the study’s findings on non-ideal industry conditions, propose specific improvement recommendations to help industries adjust business models and resource allocation.

Finally, this study aims to reduce economic recession risks and foster sustainable industry development and economic recovery.

2. Literature Review

2.1. Rise of Green Production Concept

Due to environmental issues related to greenhouse gas and hazardous substance regulations, global industries have been significantly impacted, creating an urgent need for a response [19]. Green production management, based on green manufacturing theories and production line management techniques, involves the comprehensive participation of suppliers, manufacturers, retailers, and consumers. Its core objective is to minimize environmental impact throughout a product’s entire lifecycle—from raw material sourcing, processing, packaging, transportation, and usage to disposal—while maximizing resource utilization efficiency. Green production management strives to balance economic and environmental benefits, promoting sustainable economic and environmental development. Green supply chain management covers supply, production, sales, and reverse logistics, emphasizing transportation, packaging, storage, and waste management. By integrating eco-friendly technologies, it enhances resource efficiency and reduces pollution [20]. The key focus is on innovative operational models, optimizing resource utilization, reducing waste, and achieving a balance between economic demands and sustainability goals [21]. The development of corporate environmental awareness can be integrated into business operations through Total Quality Management (TQM), embedding sustainability into enterprise strategies. Therefore, managing open innovation and green strategies is crucial to environmental sustainability. Studies show that ecological policies, government regulations, and financial assistance effectively drive green innovation [22].

In terms of corporate environmental awareness, companies can examine the entire product lifecycle—from research and development, production, distribution, and usage to disposal—to assess the amount of waste generated. By conducting environmental impact assessments, businesses can implement sustainability principles to reduce resource waste, increase material reuse, enhance product safety, minimize waste production, and ensure the proper recycling and disposal of industrial waste. Green production management, based on green manufacturing theories and production line management techniques, involves suppliers, manufacturers, retailers, and consumers in a holistic approach. Its core objective is to minimize environmental impact throughout a product’s entire lifecycle—from raw material extraction, processing, packaging, transportation, and usage to disposal—while maximizing resource efficiency. Green production management strives to balance economic and environmental benefits, ensuring an optimal approach to sustainability. Sustainable development is built on three pillars: social, economic, and environmental factors. Both the industrial and service sectors are directly and indirectly influenced by these factors, emphasizing their interconnections and providing insights for businesses to drive sustainability initiatives. Sustainability is also shaped by social capital, including human and natural resources, while competitiveness and brand value can be demonstrated through long-term sustainable business practices that lead to economic success [23].

2.2. Circular Economy

The circular economy (CE) is an economic model centered on material circulation and flow. It emphasizes transforming the traditional resource-dependent, linear-growth economy into a sustainable development economy based on ecological resource cycles throughout the entire process of resource input, corporate production, product consumption, and waste management. Additionally, it underscores the importance of regulating and prioritizing policies and laws for national circular economy development [24].

This model integrates human activities, natural resources, and technological innovation, striving to minimize resource waste and environmental impact while balancing economic and ecological benefits. Green entrepreneurs focus on renewable energy, sustainable agriculture, and eco-friendly products, driving the circular economy through innovation and mitigating climate change. Sustainable strategies enable these businesses to stand out in the market, attract customers, and gain a competitive advantage. Therefore, the engagement of supporting enterprises, consumers, and SMEs with the circular economy is essential. Producers, investors, distributors, consumers, and recyclers should be offered relevant incentives to ensure the fair distribution of benefits and costs across the entire value chain [25]. Policies and climate agreements, such as the Paris Agreement, further support the growth of this movement [26]. Under the trend of sustainable resource utilization and circular economy principles, advanced countries like Germany and Japan have made building a circular society a key pillar of sustainable development, which is gradually becoming a global consensus. Waste that cannot be reduced but can be reused in production and consumption should be reintegrated into the material cycle, while only non-recyclable waste should undergo final harmless disposal. To reduce waste generation and promote sustainable resource use, consumers have increasingly adopted a circular economy mindset, emphasizing product reuse over constant consumption. This shift has led to a decline in demand for new products and an imbalance in new-product supply. As a result, overproduction has emerged, reducing corporate profitability and contributing to economic stagnation. Additionally, the environmental protection movement primarily focuses on the ecological consequences of economic activities, ensuring that economic development does not degrade environmental quality or deplete the world’s natural resources [27].

2.3. Industrial Automation, Mass Production, and High-Quality Demands

The primary benefits of automation technology include increasing productivity, reducing unit costs, improving quality, introducing new technologies, reducing labor input, enhancing working environments, minimizing occupational hazards, and overcoming human limitations by replacing heavy, monotonous, and repetitive tasks while enhancing precision. Industry 4.0 is driven by technological advancements, utilizing machine learning to achieve interconnected equipment and improve production efficiency. However, an excessive emphasis on economic benefits and large-scale expansion often overlooks the role of workers, leading to controversies over labor cost reduction strategies.

Therefore, the adoption of automation technology enhances production efficiency and reduces costs, with productivity increasing alongside scale expansion [28].

The rapid advancement of artificial intelligence (AI), particularly the emergence of large language models (e.g., ChatGPT), has transformed humanoid robots from mechanical entities into emotionally interactive intelligent objects. In 1928, W.H. Richards created the first humanoid robot, marking a new chapter in robotics history. By the late 20th century, humanoid robots such as Honda’s ASIMO and SoftBank’s NAO demonstrated significant advancements in motion control and human–robot interaction, although they were initially designed for entertainment purposes [29].Despite efficiency gains, the widespread introduction of industrial robots has objectively reduced employment opportunities, triggering worker and labor union opposition and intensifying workplace tensions [30]. The large-scale automation of industries and high-quality product demands have enhanced material quality and extended product lifespan [31]. However, this has led to a persistent oversupply, as high-quality products are less likely to fail, reducing consumers’ need to repurchase new products. Consequently, inventory levels have increased, which in turn has negatively impacted the operational performance of many manufacturing enterprises, further contributing to economic stagnation.

2.4. The Rise of the Sharing Economy

The sharing economy, also known as the collaborative economy, primarily operates on a renting rather than owning model, where access to usage rights replaces ownership. The sharing economy is no longer just a conceptual idea but has been deeply integrated into daily life, even directly impacting traditional businesses. At its core, the sharing economy enables the full reuse of any resource or idle asset, maximizing the efficiency of unused resources and creating greater market value.

The sharing economy, through Internet of Things (IoT) connectivity and platform networks, transforms idle societal inventory resources into new supply. Sharing platforms enable individuals to rent out services under suitable conditions, simply by accessing a website or downloading an app. As proponents of the sharing economy advocate, access is better than ownership [32]. For example, personal resources such as homes, vehicles, capital, knowledge, experience, and skills can be matched with demand on a large scale, while also reducing transaction costs. The sharing economy is characterized by network collaboration, social commerce, online sharing advocacy, and shifts in consumer mindset. Network collaboration allows people to overcome geographical and temporal barriers, fostering cooperation, interaction, and resource exchange among users. Through online trading platforms, idle resources—whether skills, space, or material assets—can be monetized and exchanged [33].

The sharing economy is characterized by the interchangeability of usage rights and ownership, allowing both parties to exchange resources through interactive platforms. Users can simultaneously take on the roles of producers and consumers, depending on the transaction context. The sharing economy operates through diverse supply and demand interactions, where usage rights are transferred via sharing mechanisms. However, without internet connectivity, effective resource matching becomes difficult. Consequently, the rapid growth of the sharing economy model has led to widespread adoption, while all industries face the challenge of a shrinking labor force [34].

Transportation and technological advancements in the sharing economy are key to achieving sustainable growth. A study [35] proposed an intelligent shuttle solution, exploring influencing factors through a literature review to provide insights for businesses implementing smart transportation and logistics solutions, thereby contributing to the development of sustainable business networks. From a macroeconomic perspective, the sharing economy and traditional economic models will coexist in the long term. However, the rapid rise of the sharing economy has outpaced regulatory adjustments, leading to concerns about market fairness and rising competition and market segmentation pressures for traditional businesses. Regardless of business size, companies should proactively understand the collaborative consumption model and its impact on traditional vertical operations, including how the sharing economy reshapes market demand, threatens or enhances conventional business activities, and presents regulatory challenges. Establishing a comprehensive regulatory framework will be crucial to fostering harmonious integration between the sharing economy and traditional industries.

The sharing economy replaces ownership with access, promoting rental over purchase and effectively reducing long-term demand for physical goods while significantly improving resource and idle asset utilization. However, as the sharing economy expands, product demand decreases, and rapid supply results in inventory accumulation. This supply–demand imbalance not only increases inventory costs for businesses but also compresses profit margins for traditional manufacturers, exerting operational pressure and emerging as a potential factor contributing to economic stagnation and recession.

2.5. Smart Production

Industry 4.0 is a high-tech strategic initiative proposed by the German government to enhance the computerization, digitalization, and intelligence of Germany’s manufacturing industry. It aims to establish smart factories that are adaptive, resource-efficient, and ergonomically optimized while integrating customers and business partners into commercial and value-added processes. The technological foundation of Industry 4.0 lies in transforming production processes through the Internet of Things (IoT) and internet-based services. As a technological leader, Germany seeks to leverage Industry 4.0’s smart production model to increase competitiveness and secure its advantageous position in industrial development. Product sales forecasting and demand analysis utilize big data to evaluate current demand fluctuations and market composition. Factors such as regional demand distribution, market shifts, and product category popularity are assessed to optimize product strategies and distribution planning. Furthermore, the Industrial Internet of Things (IIoT) in modern manufacturing enables big data applications across industrial production lines. This allows manufacturers to analyze the entire production process, gaining insights into equipment diagnostics, usage analysis, energy consumption monitoring, and quality incident assessments [36].

The core concept of Industry 4.0 is not merely about replacing human labor with robots but rather emphasizing human–machine collaboration, paving the way for an intelligent production model. As global manufacturing technologies rapidly evolve, Industry 5.0 is gradually emerging as the central direction for modern industrial development. Industry 5.0 advances smart manufacturing technologies, enabling intelligent and automated production while significantly influencing the sustainable development of the manufacturing sector. However, in the context of sustainability, the transformation of smart manufacturing technologies still faces challenges and uncertainties in key driving factors. Through smart products, diversified data can be collected to support the development of new products and services, making production processes and business models more adaptive. This facilitates energy conservation, reduces excess inventory, and enhances overall operational efficiency [37].

Although Industry 4.0 emphasizes productivity and efficiency, its objectives do not fully align with the principles of sustainable development, particularly in terms of social and environmental sustainability. Moreover, the increasing challenges of global climate change, resource shortages, and the COVID-19 pandemic have highlighted the limitations of Industry 4.0. To drive industrial transformation and build a more resilient economic system, Industry 5.0 has emerged as a new paradigm [38]. In future smart factories, every production unit will be interconnected through IoT technology, even integrating real-time raw material data from upstream supply chains. This will allow business teams to accurately monitor raw material availability and respond swiftly to changes. Whether handling urgent orders or last-minute modifications, smart Industrial IoT systems can issue production commands in real time, enabling fully automated and unmanned rapid production while maximizing the balance between production and resource utilization [39].

Enterprises can monitor production lines in real time, seize business opportunities accurately, and achieve seamless order-to-delivery integration through value creation networks.

Smart factories also integrate product and production system lifecycle management, reducing unnecessary waste, minimizing inventory pressure, and shortening lead times for customized products. At the same time, Industry 4.0 integrates all relevant technologies, sales models, and product experiences, building an adaptive, resource-efficient, and human-centered intelligent production system that creates new value for the future of manufacturing [40]. As the core technology for manufacturing transformation and upgrading, smart manufacturing combines information technology, advanced manufacturing, automation, and artificial intelligence, while Industry 5.0 further emphasizes human–machine collaboration and sustainability. However, amid the rapid advancement of information technology and the guiding principles of sustainability, achieving sustainable development and transformation in the manufacturing sector has become a major focus for both nations and enterprises [41]. Human–machine relationships will follow the “5C Path”—Coexistence, Cooperation, Collaboration, Care, and Co-evolution—with Care and Co-evolution at its core, fostering human-centered manufacturing that ensures manufacturing excellence and human well-being.

2.6. Methodologies in Practice

The DEMATEL method was introduced in 1973 at the Battelle Memorial Institute in Geneva. At the time, DEMATEL was used to study complex and challenging issues, such as racial conflicts, hunger, environmental protection, and energy problems [42]. It was introduced to solve intricate and entangled problems by converting complex qualitative issues into a quantitative framework, deriving direct and indirect relationships [43,44]. It effectively analyzes complex causal structures, examines the influence of dimensions and attributes, and uses matrix operations to identify causal relationships and influence intensities. This approach explores the complexities between corporate decision making and problem solving [45].

The AHP is primarily applied to decision-making problems under uncertainty and those involving multiple evaluation factors. Owing to its continued application, refinement, and validation, the AHP has garnered significant attention from scholars and has contributed substantially to decision making [46]. However, a key limitation of the AHP is its assumption of independence among elements within each hierarchy level, which can oversimplify problems, distort problem structures, and affect decision quality [47]. To address this limitation, the ANP was introduced, incorporating interdependence and feedback relationships into the decision-making process, thereby overcoming the shortcomings of AHP [48]. D&ANP is a hybrid multi-criterion decision-making model that combines DEMATEL and ANP. D&ANP adopts the network structure of the ANP, derived from the linear structure of the AHP, and integrates the feedback concept to address interdependencies among multiple criteria [45,48]. This allows for a more comprehensive evaluation of complex decision-making scenarios and improves decision quality.

3. Materials and Methods

The primary analytical tools used in this study are the DEMATEL technique and the ANP method. By referring to the previous literature and consulting with experts and scholars, this study conducted interviews with 92 senior executives from highly automated industries across four major sectors (as shown in

Table 1. Background of senior executives interviewed in highly automated industries across four major sectors.

Service Unit	Job Title	Work Experience	Service Department	Number of Interviewees
3C I	Product Manager	10~20 years	Product Planning Department	25
MEI	Product Manager	10~20 years	Business Department	23
OI	Product Manager	10~20 years	Business Department	20
MMI	Product Manager	10~20 years	Product Planning Department	24

). From these interviews, five key dimensions and fourteen attributes contributing to economic stagnation or recession were identified, and their relevant implications were outlined (as shown in

Table 2. Relevant implications of the five key dimensions and fourteen attributes.

Dimension	Attribute	Definition	References
W1: Rise of Green Production Concepts (%)	W11: Enhancement of Organizational Operations for Environmental Awareness (%)	The green supply chain covers supply, production, sales, and recycling, optimizing transportation, packaging, storage, and waste management.	[20]
	W12: Emergence of Environmental Issues (%)	Greenhouse gas and hazardous substance regulations impact global industries, driving an urgent increase in environmental protection demands.	[19]
	W13: Corporate Sustainable Management (%)	Enterprises implement sustainable development through quality management, driving open innovation and green strategies to promote environmental protection.	[22]
W2: Emergence of Environmental Issues as Hot Topics (%)	W21: Strengthening of Policies and Regulations (%)	Strengthen policies and regulations to promote national circular economy development.	[24]
	W22: Incentives for Technology Research and Development Policies (%)	Support the circular economy, incentivize the entire industry chain, ensure fair profit distribution, and promote business and consumer participation.	[25]
	W23: Strong Promotion of the Environmental Protection Movement (%)	The environmental movement focuses on economic impact, promotes sustainable development, and preserves environmental quality and natural resources.	[27]
W3: Large-Scale Industrial Automation (%)	W31: Rapid and Mass Production Through Equipment Automation (%)	Automation enhances efficiency, reduces costs, and boosts productivity through scale expansion.	[28]
	W32: High-Quality Products Extend Product Lifespan (%)	Automated production enhances quality, optimizes materials, and extends product lifespan.	[31]
W4: Rise of the Sharing Economy (%)	W41: Reduction in Industry-Owned Production Equipment Due to Repeated Equipment Rentals (%)	Sharing platforms facilitate service rentals, enhance convenience, and emphasize access over ownership.	[32]
	W42: Full Utilization of Idle Items (%)	Online platforms release idle resources, enabling transactions of skills, space, and assets to enhance value utilization.	[33]
	W43: Reduction in Workforce for Companies	The sharing economy matches supply and demand through the internet, facilitating usage rights transfer, driving model growth, and addressing labor reduction challenges.	[34]
W5: Smart Production (%)	W51: Accurate Estimation of Market Demand and Equipment Utilization Rate (%)	Big data optimize supply chain and product strategies, enhancing production efficiency, energy management, and quality control.	[36]
	W52: Flexible Adaptation to Market Changes and Diverse Customer Demands (%)	Smart products collect data, optimize R&D and operations, enhance flexibility, conserve energy, and reduce inventory.	[37]
	W53: Enhancement of Smart Production (%)	Smart IoT enables automated production, enhances efficiency, optimizes resource utilization, and maximizes unmanned manufacturing.	[39]

). This study further explores the interrelationships and weighted influence among these dimensions and attributes to determine the most significant factors affecting economic stagnation or recession, as illustrated in Figure 1. The D&ANP method will be applied to analyze empirical data and evaluate the feasibility and validity of this research framework. Finally, case studies from four highly automated industries will demonstrate how the D&ANP method can be used to measure and assess potential causes of economic stagnation or recession.

3.1. Data Analysis Method

In the following, we outline the methodology of our analysis.

Step 1. Calculate the initial average influence matrix using scores from the DEMATEL questionnaire. Based on experts’ opinions, a scale ranging from 0 to 4 was used to evaluate the relationships among the dimensions or attributes that influenced each other. The scoring criteria are as follows: “0” indicates “no influence”, “1” indicates “low influence”, “2” indicates “moderate influence”, “3” indicates “high influence”, and “4” indicates “very high influence.” A directed graph illustrating the contextual relationships among the elements of the system is shown in Figure 2. For example, an arrow from b to a indicates that b influences a with an influence score of 3. Equation (1) represents the average direct influence matrix.

$$A=\left[\begin{array}{ccccc}{a}_{11}& \cdots & a_{1j}& \cdots & a_{1n}\\ ⋮& & ⋮& & ⋮\\ a_{i1}& \cdots & a_{ij}& \cdots & a_{in}\\ ⋮& & ⋮& & ⋮\\ a_{n1}& \cdots & a_{nj}& \cdots & a_{nn}\end{array}\right]$$

(1)

$$\mathit{R}=s\mathit{A}$$

(2)

Step 2. Normalize the initial direct influence matrix. Based on the average influence matrix A, the normalized direct influence matrix is obtained by using Equations (2) and (3):

$$s=\min \left\{1/\underset{i}{\max }{\displaystyle \sum _{j=1}^n{a}_{ij}},1/\underset{j}{\max }{\displaystyle \sum _{i=1}^n{a}_{ij}}\right\} i, j\in \{1, 2, \dots , n\}$$

(3)

Step 3. Obtain the total influence matrix. Since direct influences have a continuous diminishing effect, they follow the increasing powers of R, for example, ${\mathit{R}}^{\mathbf{2}},{\mathit{R}}^3,\dots ,{\mathit{R}}^k$, and its convergence limit, $\underset{k\to \mathrm{\infty }}{\lim }{\mathit{R}}^k={\left[0\right]}_{n\times n}$, will converge to 0, where R = ${\left[r_{ij}\right]}_{n\times n}$, $0\le r_{ij}<1$, $0<\sum _{j=1}^n{r}_{ij}, \sum _{i=1}^n{r}_{ij}\le 1,$ and the sum of only one row or one column equals 1. Once the normalized direct influence matrix R is obtained, the total influence matrix T can be calculated by using Equation (4), where I represents the identity matrix. The calculation of the total influence matrix is as follows:

$$\begin{array}{l}\hspace{1em}\mathit{T}=\mathit{R}+{\mathit{R}}^2+{\mathit{R}}^3+\dots +{\mathit{R}}^k=\mathit{R} (\mathit{I}+\mathit{R}+{\mathit{R}}^2+\dots +{\mathit{R}}^{k-1})[( \mathit{I}-\mathit{R} ) ( \mathit{I}-\mathit{R} {)}^{-1}]\\ =\mathit{R} ( \mathit{I}-{\mathit{R}}^k) ( \mathit{I}-\mathit{R} {)}^{-1}\end{array}$$

(4)

where R = ${\left[r_{ij}\right]}_{n\times n}$, $0\le r_{ij}<1$, $0<\sum _{j=1}^n{r}_{ij}, \sum _{i=1}^n{r}_{ij}\le 1$. Assuming that at least one row or column sums to one (but not all), $\underset{k\to \infty }{\lim }{\mathit{R}}^k={[0]}_{n\times n}$. In this stage, Equations (5)–(7) are applied, where the row and column sums are represented as column vectors ${\left[r_{i.}\right]}_{n\times 1}=\left(r_{1.},\dots ,r_{i.}.\dots ,r_{n.}\right){\left[c_{.j}\right]}_{1\times n}=\left(c_{.1},\dots ,c_{.j},\dots ,c_{.n}\right)$.

Let $i=j$ and $=j\in \left\{1,2,\dots ,n\right\}$. Then, by adding r to c, it is treated as a horizontal axis vector $(\mathit{r}+\mathit{c})$, indicating the importance of the dimension. Similarly, by subtracting r from c, it is treated as a vertical axis vector $(\mathit{r}-\mathit{c})$. This separates the dimensions into a cause group and an affected group. Generally, when $(\mathit{r}-\mathit{c})$ is positive, the dimension is part of the cause group. Conversely, if $(\mathit{r}-\mathit{c})$ is negative, the dimension is part of the affected group, providing a valuable method for decision making [41,42].

$${\mathit{T}=\left[t_{ij}\right]}_{n\times n},i, j=1, 2, \dots , n,$$

(5)

$${\mathit{r}=\left[{\sum }_{j=1}^n{t}_{ij}\right]}_{n\times 1}={\left[r_{i.}\right]}_{n\times 1}={\left(r_{1.}, \dots , r_{i.}, \dots , r_{n.}\right)}^{\prime }, \mathrm{and}$$

(6)

$$\mathit{c}=\left[{\sum }_{i=1}^n{t}_{ij}\right]={\left[c_{.j}\right]}_{1\times n}=\left(c_{.1}, \dots , { c}_{.j}, \dots , c_{.n}\right)$$

(7)

where r_i. represents the row sum of the i^th row in matrix T, indicating the total direct influence that dimension i exerts on other dimensions. Similarly, c_.j represents the column sum of the jth column in matrix T, indicating the total direct influence received by other dimensions from dimension j. In addition, when i = j (i.e., the sum of row and column combinations), (r_i. + c_.j) provides an indicator of the intensity of influence given and received. If (r_i. − c_.j) is positive, then dimension i is influencing other dimensions. If (r_i. − c_.j) is negative, then dimension i is being influenced by other dimensions [41,42].

Step 4. Set the threshold value. A threshold is set to filter out dimensions with smaller influences in matrix T, making it necessary to separate these dimensions. Based on matrix T, each dimension provides information about how dimension i influences dimension j in practice. If all information from matrix T is transferred, the mapping would become too complex and unsuitable for decision making. To reduce complexity, decision makers set a threshold for the impact level: only dimensions with influence values above the threshold in the matrix can be selected. The threshold can be determined through expert brainstorming. Once the threshold is established, the structural impact can be presented, as shown in Figure 3.

(8)

The ANP is an extension of the AHP [48], designed to overcome the issues of interdependence and feedback between dimensions and attributes. Although both the AHP and the ANP derive priority scales through pairwise comparisons of dimensions or attributes, there are differences in the weights between them. The first distinction is that the AHP is a special case of the ANP. The ANP can handle dependencies within (internal dependence) and between (external dependence) clusters. The initial step of the ANP is to compare the attributes across the entire system to form a super-matrix through pairwise comparisons. This is performed by asking, “How important is one attribute relative to another in terms of our interest or preference?” Relative importance was determined by using a scale of 1 to 9, ranging from equal to extreme importance [49].

The general form of the super-matrix is as shown in Equation (9), where C_n represents the nth element in each cluster and $e_{{nm}_n}$ represents the nth element in each cluster ${m_n}^{th}$. The matrix W_ij is the principal eigenvector of the influence of the attributes compared between the jth and ith clusters. The form of the super-matrix depends on the diversity of its structure. For example, if the system’s structure is an unweighted super-matrix W, it contains local priorities derived from pairwise comparisons across the entire network, as shown in Equation (10):

$$\begin{gathered} W=\begin{array}{cc}& \begin{array}{cccc}{C}_1& C_2& & C_{\mathrm{n}}\\ e_{11}\dots {e_1}_{m_1}& e_{21}\dots {e_2}_{m_2}& \dots \dots & e_{n1}\dots {e_n}_{m_n}\end{array}\\ \begin{array}{cc}{C}_1& \begin{array}{c}{e}_{11}\\ e_{12}\\ ⋮\\ {e_1}_{m_1}\end{array}\\ C_2& \begin{array}{c}{e}_{21}\\ e_{22}\\ ⋮\\ {e_2}_{m_2}\end{array}\\ ⋮& ⋮\\ C_n& \begin{array}{c}{e}_{n1}\\ e_{n2}\\ ⋮\\ {e_n}_{m_n}\end{array}\end{array}& \left[\begin{array}{cccc} & & & \\ W_{11}& W_{12}& \dots & W_{1n}\\ & & & \\ & & & \\ & & & \\ W_{12}& W_{22}& \dots & W_{2n}\\ & & & \\ & & & \\ ⋮& ⋮& \begin{array}{c}\dots \\ \dots \\ \dots \end{array}& ⋮\\ & & & \\ & & & \\ W_{n1}& W_{n2}& \dots & W_{nn}\\ & & & \end{array}\right]\end{array} \end{gathered}$$

(9)

$$W=\begin{array}{cc}& \begin{array}{ccc}{D}_1 & D_2& D_3\end{array}\\ \begin{array}{c}{D}_1\\ D_2\\ D_3\end{array}& \left[\begin{array}{ccc}0& W_{12}& 0\\ W_{21}& 0& W_{23}\\ W_{31}& 0& W_{33}\end{array}\right]\end{array}$$

(10)

In this case, W₁₂ represents the matrix of weights for cluster 1 (or dimension 1) relative to cluster 2, W₂₃ represents the weights for cluster 2 relative to cluster 3, and W₃₁ shows the weights for cluster 3 relative to cluster 1. W₃₃ is represented as the internal dependencies and feedback within cluster 3 (see Figure 3).

The weighted super-matrix is obtained by multiplying it with the normalized matrix derived from DEMATEL. Traditional methods generate a weighted super-matrix by con-verting each row value into a single value, where each element in a row is divided by the number of clusters (or dimensions). Thus, each row is normalized to 1. By using this normalization method, we assumed that each cluster had an equal weight. However, the influence of each cluster (or dimension) on other clusters may differ, as described in Step 3 of Section 3.1. Therefore, it is unreasonable to assume equal weights for each cluster to derive a weighted super-matrix. We used the DEMATEL technique to address this issue [22,23]. Next, the total influence matrix T is shown in Equation (11).

$${\mathit{T}}_{\alpha }=\left[\begin{array}{ccccc}{t}_{11}^{\alpha }& \cdots & t_{1j}^{\alpha }& \cdots & t_{1n}^{\alpha }\\ ⋮& & ⋮& & ⋮\\ t_{i1}^{\alpha }& \cdots & t_{ij}^{\alpha }& \cdots & t_{in}^{\alpha }\\ ⋮& & ⋮& & ⋮\\ t_{n1}^{\alpha }& \cdots & t_{nj}^{\alpha }& \cdots & t_{nn}^{\alpha }\end{array}\right]$$

(11)

where t_ij, the total influence matrix T, must be normalized by dividing it by Equation (12):

$$d_i={\displaystyle \sum _{j=1}^n{t}_{ij}^{\alpha }}$$

(12)

Therefore, we can normalize the total influence matrix T and represent it as Equation (13):

$${\mathit{T}}_s=\left[\begin{array}{ccccc}{t}_{11}^{\alpha }/d_1& \cdots & t_{1j}^{\alpha }/d_1& \cdots & t_{1n}^{\alpha }/d_1\\ ⋮& & ⋮& & ⋮\\ t_{i1}^{\alpha }/d_2& \cdots & t_{ij}^{\alpha }/d_2& \cdots & t_{in}^{\alpha }/d_2\\ ⋮& & ⋮& & ⋮\\ t_{n1}^{\alpha }/d_3& \cdots & t_{nj}^{\alpha }/d_3& \cdots & t_{nn}^{\alpha }/d_3\end{array}\right]=\left[\begin{array}{ccccc}{t}_{11}^s& \cdots & t_{1j}^s& \cdots & t_{1n}^s\\ ⋮& & ⋮& & ⋮\\ t_{i1}^s& \cdots & t_{ij}^s& \cdots & t_{in}^s\\ ⋮& & ⋮& & ⋮\\ t_{n1}^s& \cdots & t_{nj}^s& \cdots & t_{nn}^s\end{array}\right]$$

(13)

We used the normalized total influence matrix (hereafter referred to as the “normalized matrix”) and the unweighted super-matrix W. The influence super-matrix ${\mathit{W}}_W$ is calculated by using Equation (14) [24].

$${\mathit{W}}_W=\left[\begin{array}{c}{t}_{11}^s\times {\mathit{W}}_{11} t_{21}^s\times {\mathit{W}}_{12} \cdots \cdots t_{n1}^s\times {\mathit{W}}_{1n} \\ t_{12}^s\times {\mathit{W}}_{21} t_{22}^s\times {\mathit{W}}_{22} ⋮ ⋮ \\ ⋮ \cdots t_{ji}^s\times {\mathit{W}}_{ij} \cdots t_{ni}^s\times {\mathit{W}}_{in}\\ ⋮ ⋮ ⋮ \\ t_{1n}^s\times {\mathit{W}}_{n1} t_{2n}^s\times {\mathit{W}}_{n2} \cdots \cdots t_{nn}^s\times {\mathit{W}}_{nn}\end{array}\right]$$

(14)

The normalization of the weighted super-matrix in Equation (14) is achieved by con-verting the “sum of all rows” into a single value, resulting in the normalized weighted super-matrix. At this point, ${\mathit{W}}_W$ becomes the normalized weighted super-matrix. This step is very similar to the concept of a Markov chain, ensuring that the sum of the probabilities for all states equals 1 [24]. ${\mathit{W}}_W$ is increased to a sufficiently large power k to limit the normalized weighted super-matrix, as shown in Equation (15), until the normalized weighted super-matrix reaches a stable state. Then, ${\mathit{W}}_W$ is converged to a long-term stable super-matrix to determine the overall influential weights.

$$\underset{k\to \infty }{\lim }{\left({{\mathit{W}}^{\prime }}_w\right)}^k$$

(15)

If the limiting super-matrix is not unique, for example, if there are N super-matrices, the average value of these matrices is obtained by adding the N super-matrices and dividing by N.

4. Analysis and Verification

4.1. Basic Information of DEMATEL Questionnaire

We used cluster sampling to proceed with our analysis. Survey participants were senior- and middle-level executives from four major industries: 3C, machinery and equipment, optoelectronics, and metal manufacturing. The questionnaires were completed under optimal conditions with no bias and from a neutral standpoint. A total of 200 questionnaires were distributed, and 92 valid responses were returned.

4.2. DEMATEL Analysis of Interdependence Between Dimensions

• Establish the Direct Influence Matrix A. The dimensions in the 92 collected questionnaires were calculated by using the geometric mean to create the direct influence matrix A (refer to Equation [1]), as shown in

  Table 3. Geometric mean direct influence matrix of dimensions, A.

  Dimension	W1	W2	W3	W4	W5	Row Sum
  W1	0.000	3.714	2.571	2.714	2.571	11.571
  W2	3.571	0.000	2.367	3.286	2.571	11.796
  W3	2.429	2.776	0.000	3.143	2.571	10.918
  W4	2.429	3.469	3.429	0.000	2.000	11.327
  W5	3.286	3.000	3.143	3.429	0.000	12.857
  Row Sum	11.714	12.959	11.510	12.571	9.714

  .
• Calculate the Normalized Direct Influence Matrix R. According to Equation (3), first, the maximum values of the column and row sums are 12.857 and 12.959, respectively (refer to Table 1). By taking 1/12.857 and 1/12.959, we obtain the values 0.078 and 0.077. Then, we multiply each value in Table 1 by the smaller value, 0.078 or 0.077, to obtain the normalized matrix R, as shown in

  Table 4. Normalized direct relationship matrix, R.

  Dimension	W1	W2	W3	W4	W5
  W1	0	0.289	0.198	0.209	0.198
  W2	0.276	0	0.183	0.254	0.198
  W3	0.187	0.214	0	0.243	0.198
  W4	0.187	0.268	0.265	0	0.154
  W5	0.254	0.231	0.243	0.265	0

  .
• Establish the Total Influence Relationship Matrix T. By using Equation (4) and Table 2, the total influence matrix T is calculated, as shown in

  Table 5. Total influence relationship matrix, T.

  Dimension	W1	W2	W3	W4	W5
  W1	1.650	2.020	1.784	1.917	1.576
  W2	1.889	1.824	1.799	1.970	1.596
  W3	1.726	1.885	1.543	1.855	1.507
  W4	1.767	1.964	1.791	1.703	1.513
  W5	1.995	2.142	1.959	2.107	1.535

  .
• Set the Threshold Value by Using the Total Influence Relationship Matrix T. From Table 5, the threshold is set according to Equation (8). To filter out dimensions with smaller influences in matrix T, we set the threshold $\mathsf{\alpha }={\sum }_{j=1}^5{\sum }_{\mathrm{i}=1}^5{t}_{ij}/25$ and obtain $\alpha$ = 1.14 (i.e., the average total influence of all dimensions) and the value $t_{W_{\mathrm{i}}{W}_{\mathrm{i}}}$. This is the normalized total influence matrix. By calculating with an Excel worksheet, as shown in Equation (13), the value 1.302 in

  Table 6. Total influence matrix, T, Tα, and normalized total influence matrix T.

  Dimension	W1	W2	W3	W4	W5
  W1	0.883$\to$0 ($t_{W_1{W}_1}=$ 0)	0.992$\to$0 ($t_{W_1{W}_2}=$ 0)	1.173 ($t_{W_1{W}_3}=$ 1.173)	1.185 ($t_{W_1{W}_4}=$ 1.185)	0.933$\to$0 ($t_{W_1{W}_5}=$ 0)
  W2	1302 ($t_{W_2{W}_1}=$ 1.302)	1.039$\to$0 ($t_{W_2{W}_2}=$ 0)	1.398 ($t_{W_2{W}_3}=$ 1.398)	1.459 ($t_{W_2{W}_4}=$ 1.459)	1.158 ($t_{W_1{W}_5}=$ 0)
  W3	1.204 ($t_{W_3{W}_1}=$ 1.204)	1.216 ($t_{W_3{W}_2}=$ 1.216)	1.143 ($t_{W_3{W}_3}=$ 1.143)	1.456 ($t_{W_3{W}_4}=$ 1.456)	1.154 ($t_{W_3{W}_5}$= 0)
  W4	1.017$\to$0 ($t_{W_4{W}_1}=$ 0)	0.953$\to$0 ($t_{W_4{W}_2}=$ 0)	1.060$\to$0 ($t_{W_4{W}_3}=$ 0)	0.990$\to$0 ($t_{W_4{W}_4}=$ 0)	0.937$\to$0 ($t_{W_4{W}_5}=$ 0)
  W5	1.159 ($t_{W_5{W}_1}=$ 1.159)	0.000$\to$0 ($t_{W_5{W}_2}=$ 0)	1.254 ($t_{W_5{W}_3}=$ 1.254)	1.403 ($t_{W_5{W}_4}=$ 1.403)	0.917$\to$0 ($t_{W_5{W}_5}=$ 0)

  represents the value of “W₂: Circular Economy” after pairwise comparison in the total influence matrix. The value 1.302, which is greater than the threshold, indicates that “W₂: Circular Economy” is positively correlated. The value 1.017 in the table represents the result of the pairwise comparison between “W₁: Rise of Green Production Concepts” and “W₄: Emergence of the Sharing Economy.” Since 1.017 is smaller than the threshold value of 1.14, the value in the total influence matrix becomes 0. This process continues accordingly.

4.3. Calculation of ANP Weights

After averaging the relative importance ratio data from the ANP attribute questionnaires of each industry and inputting all questionnaire data, the process moves to the calculation of the super-matrix. The super-matrix calculation process was divided into three stages. First, software was used to input the questionnaire data and establish network relationships between the attributes. Before calculating the unweighted super-matrix, each attribute must pass a consistency check. Finally, the eigenvalue of each pairwise comparison matrix of the attributes is used to form the unweighted super-matrix W for the attributes. In the consistency check, the inconsistency value for each attribute must be less than 0.1.

4.4. DEMATEL Combined with ANP to Form D&ANP

This section presents the normalized total influence matrix obtained by using the DEMATEL method (refer to

Table 7. D&ANP influence unweighted weight matrix, ${\mathit{W}}_W$.

Attribute	W11	W12	W13	W21	W22	W23	W31	W32	W41	W42	W43	W51	W52	W53
W11	0	0	0	0	0	0	0.08	0.05	0	0	0	0.07	0.03	0.04
W12	0	0	0	0	0	0	0.05	0.08	0	0	0	0.03	0.05	0.06
W13	0	0	0	0	0	0	0.07	0.07	0	0	0	0.1	0.12	0.1
W21	0	0	0	0	0	0	0.07	0.1	0.12	0	0	0	0	0
W22	0	0	0	0	0	0	0.1	0.06	0.05	0	0	0	0	0
W23	0	0	0	0	0	0	0.03	0.04	0.03	0	0	0	0	0
W31	0.31	0.13	0.63	0.11	0.05	0.11	0	0	0.09	0.09	0.14	0.07	0.1	0.08
W32	0.94	1.13	0.63	0.11	0.16	0.11	0	0	0.09	0.09	0.05	0.11	0.08	0.1
W41	0.95	1.14	0.63	0.14	0.12	0.06	0.09	0.11	0.09	0	0	0.14	0.14	0.14
W42	0.63	0.25	0.18	0.06	0.08	0.09	0.04	0.05	0	0	0	0.06	0.06	0.06
W43	0.39	0.39	0.42	0.04	0.03	0.08	0.06	0.03	0	0	0	0.04	0.04	0.04
W51	0	0	0	0.08	0.1	0.11	0.14	0.14	0.03	0.06	0.06	0	0	0
W52	0	0	0	0.05	0.06	0.05	0.04	0.06	0.08	0.13	0.04	0	0	0
W53	0	0	0	0.06	0.03	0.03	0.06	0.04	0.12	0.05	0.14	0	0	0

). By using the ANP software application, the eigenvalues of the pairwise comparison matrices of each attribute are used to form the unweighted super-matrix W for the attributes. Equation (14) is then used to calculate the influence unweighted super-matrix ${\mathit{W}}_W$ for D&ANP [24]. The results are shown in Table 6. According to Equations (11) to (13), after normalizing the influence unweighted matrix of D&ANP, the normalized new influential weight matrix is obtained. To obtain the limiting influential weight super-matrix, the influential weight super-matrix must be multiplied multiple times by itself until convergence is achieved. As $\underset{k\to \infty }{\lim }{\left({{\mathit{W}}^{\prime }}_w\right)}^k$, the limiting influential weight super-matrix is as shown in Table 6. For each dimension, the sum of the attribute weights can be used to obtain the influential weight of the D&ANP dimensions, as shown in the far-right column of

Table 8. Limiting influential weight super-matrix $\underset{k\to \infty }{\lim }{\left({{\mathit{W}}^{\prime }}_w\right)}^k$ and dimension weights.

Attribute	W11	W12	W13	W21	W22	W23	W31	W32	W41	W42	W43	W51	W52	W53	Dimension
W11	0.00	0.00	0.00	0.00	0.00	0.00	0.08	0.05	0.00	0.00	0.00	0.07	0.03	0.04	0.14
W12	0.00	0.00	0.00	0.00	0.00	0.00	0.05	0.08	0.00	0.00	0.00	0.03	0.05	0.06
W13	0.00	0.00	0.00	0.00	0.00	0.00	0.07	0.07	0.00	0.00	0.00	0.10	0.12	0.10
W21	0.00	0.00	0.00	0.00	0.00	0.00	0.07	0.10	0.12	0.00	0.00	0.00	0.00	0.00	0.11
W22	0.00	0.00	0.00	0.00	0.00	0.00	0.10	0.06	0.05	0.00	0.00	0.00	0.00	0.00
W23	0.00	0.00	0.00	0.00	0.00	0.00	0.03	0.04	0.03	0.00	0.00	0.00	0.00	0.00	0.26
W31	0.31	0.13	0.63	0.11	0.05	0.11	0.00	0.00	0.09	0.09	0.14	0.07	0.10	0.08
W32	0.94	1.13	0.63	0.11	0.16	0.11	0.00	0.00	0.09	0.09	0.05	0.11	0.08	0.10
W41	0.95	1.14	0.63	0.14	0.12	0.06	0.09	0.11	0.09	0.00	0.00	0.14	0.14	0.14	0.28
W42	0.63	0.25	0.18	0.06	0.08	0.09	0.04	0.05	0.00	0.00	0.00	0.06	0.06	0.06
W43	0.39	0.39	0.42	0.04	0.03	0.08	0.06	0.03	0.00	0.00	0.00	0.04	0.04	0.04
W51	0.00	0.00	0.00	0.08	0.10	0.11	0.14	0.14	0.03	0.06	0.06	0.00	0.00	0.00	0.23
W52	0.00	0.00	0.00	0.05	0.06	0.05	0.04	0.06	0.08	0.13	0.04	0.00	0.00	0.00
W53	0.00	0.00	0.00	0.06	0.03	0.03	0.06	0.04	0.12	0.05	0.14	0.00	0.00	0.00

.

4.5. Average Performance Value Analysis

We assessed the performance of the dimensions and attributes related to the causes of economic stagnation and recession by using scores ranging from 1 to 9. A score of “1” represents the lowest score in the evaluation criteria, while a score of “9” represents the highest score. The higher the score, the higher the suitability level.

The geometric mean of the suitability ratings from 92 senior- and middle-level executives across four major industries—the 3C, machinery and equipment, optoelectronics, and metal manufacturing industries—was multiplied by the D&ANP influential weights to obtain the average performance values (refer to

Table 9. Average performance values and rankings of attribute suitability for all questionnaires.

Attribute	D&ANP Weight	Average Suitability	Average Performance Value
W11	0.04 (11)	6.75 (12)	0.27 (12)
W12	0.04 (11)	6.75 (12)	0.27 (12)
W13	0.06 (7)	6.75 (12)	0.41 (8)
W21	0.05 (9)	8.25 (1)	0.41 (8)
W22	0.04 (11)	8.25 (1)	0.33 (11)
W23	0.02 (14)	8.00 (4)	0.16 (14)
W31	0.12 (3)	7.25 (9)	0.87 (3)
W32	0.14 (2)	7.75 (5)	1.09 (2)
W41	0.17 (1)	8.25 (3)	1.40 (1)
W42	0.06 (7)	7.50 (6)	0.45 (7)
W43	0.05 (9)	7.25 (9)	0.36 (10)
W51	0.08 (4)	7.25 (9)	0.58 (5)
W52	0.07 (6)	7.50 (6)	0.53 (6)
W53	0.08 (4)	7.50 (6)	0.60 (4)
Summation	1.00	105.00	7.72

). The average performance values obtained for the suitability of the attributes related to the causes of economic stagnation or recession were compared with the overall average performance values from all the questionnaires, which were set as benchmarks (see Table 9).

The average suitability of the attributes related to the causes of economic stagnation or recession was calculated and compared with the overall average performance values from all questionnaires (see Table 9), distinguishing ideal and non-ideal conditions across the four major industries (see

Table 10. Average performance rankings of attribute suitability by industry and ideal/non-ideal states.

Attribute	MMI	OI	MEI	3CI
W11	Ideal	Ideal	Non-ideal	Non-ideal
W12	Ideal	Ideal	Non-ideal	Non-ideal
W13	Ideal	Non-ideal	Non-ideal	Ideal
W21	Ideal	Non-ideal	Non-ideal	Non-ideal
W22	Ideal	Ideal	Non-ideal	Non-ideal
W23	Ideal	Non-ideal	Non-ideal	Non-ideal
W31	Ideal	Ideal	Non-ideal	Non-ideal
W32	Ideal	Ideal	Ideal	Non-ideal
W41	Ideal	Non-ideal	Ideal	Non-ideal
W42	Ideal	Ideal	Ideal	Non-ideal
W43	Ideal	Non-ideal	Non-ideal	Non-ideal
W51	Ideal	Non-ideal	Non-ideal	Non-ideal
W52	Ideal	Non-ideal	Non-ideal	Ideal
W53	Ideal	Ideal	Non-ideal	Non-ideal

). The major causes of economic stagnation and recession were identified by organizing and ranking the D&ANP weights, suitability, and average performance values.

As shown in Table 9 and Figure 4, “W₄₁: Reduced Ownership of Production Equipment in Industries” ranks first in both D&ANP weight and performance value but ranks third in suitability. This indicates that there is a gap between execution and suitability that must be strengthened. However, “W₄₃: Reduction in Workforce in Companies or Organizations” has rankings close to tenth in weight, suitability, and performance value, suggesting that this attribute is well matched in both execution and potential impact. The subsequent analyses followed this pattern.

In today’s fast-changing economy, traditional industries face significant challenges. This study explores economic stagnation and recession causes, offering solutions. Through the literature review, key concepts such as “green production”, “circular economy”, and “industrial automation” were established. Five major dimensions, including “Sharing Economy” and “Smart Production”, were analyzed, leading to 14 evaluation attributes.

Traditional ANP calculations overlooked interdependencies among dimensions, so this study integrated DEMATEL with the ANP to refine weighting by using D&ANP.

Findings show that “Sharing Economy” has the highest influence (0.28), while “Circular Economy” has the lowest (0.11). Among the attributes, “Reduced Ownership of Production Equipment” (0.17) and “High-Quality Products Extend Lifespan” (0.14) ranked the highest. In metal manufacturing, “Environmental Awareness” scored 0.32, exceeding the 0.27 benchmark, indicating strong performance. However, in machinery industries, it scored 0.24, below the benchmark, requiring improvement.

The global economy faces wealth gaps, regional disparities, and geopolitical uncertainties. Environmental concerns and economic protectionism disrupt supply–demand balance, worsening economic challenges. To address this, industries must innovate, embracing green production, circular economies, and smart production. However, smart production relies on real-time market data, which remain controlled by private and state-owned enterprises. Policymakers should enhance data-sharing regulations, enabling industries to analyze demand and adopt flexible business strategies, fostering economic transformation.

Traditional production models are passive and lack flexibility. Germany’s Industry 4.0, introduced in 2013, used automation and digitalization to enhance adaptability. In 2019, the European Parliament introduced Industry 5.0, emphasizing human–machine collaboration, sustainability, and social responsibility. By integrating AI and robotics, Industry 5.0 balances technology with human creativity. Companies must embrace these changes to bridge the supply–demand gap, drive innovation, and achieve sustainable economic growth.

4.6. Discussion

This study identifies the metal manufacturing industry as the most ideal and the machinery and equipment industry as the least ideal based on Table 10. The metal industry excels in environmental protection, policy support, automation, and resource utilization, benefiting from strong regulatory compliance and technological incentives. Automation and high-quality products enhance competitiveness, while efficient resource use reduces costs. Accurate market forecasting and adaptability further strengthen its position.

In contrast, the machinery industry struggles with environmental compliance, policy support, and sustainability. High emissions and resource consumption make it difficult to meet stricter regulations. Lagging automation and inefficient production limit competitiveness, while poor resource utilization increases costs. Weak market adaptability further hampers long-term growth.

Overall, the metal industry thrives due to technology and flexibility, whereas the machinery industry faces environmental and regulatory challenges.

The machinery and optoelectronics industries face different economic challenges, affecting their response strategies. The machinery industry is highly influenced by economic fluctuations, with demand rising and falling based on investment cycles, whereas the optoelectronics industry, driven by technological advancements, has greater market flexibility and adaptability. However, the optoelectronics industry requires high R&D investment due to rapid technological changes, while the machinery industry experiences more stable technological development, focusing on automation and smart manufacturing. Environmental regulations have a greater impact on the optoelectronics industry due to high energy consumption and material pollution, whereas the machinery industry mainly addresses carbon emissions and energy efficiency improvements. In terms of supply chains, the machinery industry has diverse raw material sources and lower risks, whereas the optoelectronics industry, with its highly concentrated technology, is more vulnerable to trade wars and geopolitical tensions. Regarding capital requirements, the optoelectronics industry faces high equipment and R&D costs, while the machinery industry prioritizes automation investments. Overall, the machinery industry should enhance smart manufacturing and market flexibility, while the optoelectronics industry must optimize R&D and supply chain strategies to strengthen competitiveness.

5. Conclusions

This study aims to explore the causes of stagnation and decline in the sharing economy, analyze industrial response strategies in the face of economic changes, and utilize the D&ANP (DEMATEL and ANP) method to assess the relative importance of influencing factors, identifying key elements affecting industrial development. This research study focuses on five major dimensions: Rise of Green Production Concepts, Promotion of the Circular Economy, Development of Industrial Automation and High-Quality Demand, Impact of the Sharing Economy, and Application of Smart Production. In addition, this study integrates expert interviews and secondary data analysis to ensure comprehensive insights and practical applicability. Through weight analysis, the study evaluates the impact of each dimension on economic stagnation. highlighting the dynamic interactions among these factors and offering strategic recommendations for enterprises and policymakers to foster resilience and long-term competitiveness in a rapidly evolving global market.

The findings indicate that the development of the sharing economy has the most significant impact on economic stagnation, followed by industrial automation and smart production, while the influence of the circular economy is relatively lower. In terms of individual attributes, reducing the ownership of production equipment has the greatest effect on business operations, and extending product lifespan through high-quality products is another critical factor. This suggests that while the sharing economy and automation technologies enhance resource efficiency, they may also lead to reduced market demand and profit compression, further intensifying the risk of economic stagnation. Moreover, industries exhibit different levels of adaptability to economic transitions. The metal manufacturing sector demonstrates stronger performance in environmental awareness and organizational operations, indicating relative success in transitioning toward a green economy. In contrast, the machinery and equipment sector still requires improvements in environmental sustainability strategies. In addition, the findings reveal that sectors with proactive investment in digital transformation and cross-industry collaboration are better positioned to mitigate stagnation risks. This study also underscores the need for continuous workforce upskilling, policy support, and infrastructure development to foster industrial agility and long-term stability.

Ideally, enterprises should fully utilize smart production technologies and big data analytics to mitigate economic risks arising from supply–demand imbalances and enhance industry competitiveness. Additionally, policymakers should strengthen regulations on the sharing economy and improve data flow mechanisms to enable more precise market demand predictions, preventing supply–demand mismatches caused by information asymmetry. Moreover, fostering collaboration among government, academia, and industry is essential to building comprehensive innovation ecosystems that support sustainable industrial transformation. The contribution of this study lies in establishing an impact–weight model, providing decision makers with concrete strategies for improvement, such as optimizing production models, enhancing data sharing, and adjusting industrial policies. Furthermore, this study emphasizes the importance of developing early-warning systems based on real-time data monitoring, enabling companies to rapidly adjust strategies in response to external shocks and market volatility. These measures aim to promote sustainable economic development and ensure that businesses maintain a stable competitive edge in an ever-changing economic landscape.

This study explored the impact of business operations on economic shocks. How-ever, owing to the limited availability of the relevant literature and data, we must consider the following: First, this study focuses only on analyzing the potential influencing factors and industry performance of four highly automated industries in the context of economic stagnation and recession. It does not cover how foreign industries respond to economic stagnation and recession. Second, the data used in this study are based solely on four major industries; therefore, they cannot comprehensively reflect the performance status of other industries.

This study proposes the following recommendations and future directions:

• The current state of the economic environment

The global economy faces challenges such as income inequality, regional imbalances, and geopolitical uncertainties. Rising environmental awareness, economic protectionism, and leadership shifts further disrupt supply–demand dynamics, contributing to stagnation.

2.

Industries must accelerate corporate innovation and adapt to environmental changes

To adapt, industries must innovate themselves by adopting the circular economy, green production, and automation. Smart production requires real-time market data for accurate demand analysis, yet many of these data remain inaccessible under private or state control. Governments should enhance data-sharing policies to help industries respond flexibly. By leveraging data analytics, businesses can develop adaptive strategies, foster new models, and enhance resilience.

3.

Moving towards the development of smart production

Traditional manufacturing lacks flexibility, making smart production essential. Industry 4.0 improves efficiency through automation and AI, while Industry 5.0 prioritizes human–machine collaboration and sustainability. Future production must integrate technology with social values, optimizing processes, conserving resources, and addressing labor and environmental concerns. Businesses embracing agile, data-driven production will bridge supply–demand gaps, drive transformation, and sustain competitiveness.