Unveiling Capability Structures for Resilient Supply Chains in Cruise Shipbuilding: A Hybrid DEMATEL-ISM-MICMAC Approach

Abstract

The cruise shipbuilding industry faces significant disruptions stemming from escalating trade frictions and regional conflicts which threaten its operational and economic sustainability. Enhancing supply chain resilience is thus crucial for sustainable development. This study identifies critical resilience factors and examines their interrelationships within growth-stage cruise shipbuilding supply chains. Fuzzy Decision-Making Trial and Evaluation Laboratory (DEMATEL), Interpretive Structural Modeling (ISM), and Cross-Impact Matrix Multiplication Applied to Classification (MICMAC) analysis are integrated to explore causal linkages, hierarchical structures, and driver-dependence dynamics. The analysis reveals that customized demand responsiveness, learning organization, specialized industrial clusters, and inter-industry collaboration are fundamental causal drivers. In contrast, knowledge stock, risk culture, and final-assembly orchestration serve as critical mediators. Based on these findings, we propose distinct resource-contingent strategic pathways for managers. This study provides an actionable framework for building resilience, offering critical guidance for securing the sustainable development of the cruise shipbuilding industry amid uncertainty.

1. Introduction

Cruise shipbuilding constitutes an integrated and high-value industrial system that signals the technological and organizational maturity of a nation’s shipbuilding sector [1]. Given the imminent imbalance between global supply and demand, advancing domestic cruise construction has become a strategic priority for emerging economies like China [2]. However, this pursuit of industrial development is fraught with sustainability challenges. The heavy reliance on geographically dispersed, European-centric support industries leaves growth-stage cruise shipbuilding supply chains (GCSSCs) acutely vulnerable to disruptions [3]. Compounded by demand volatility, information asymmetries, and geopolitical tensions, these structural risks jeopardize operational continuity and lead to significant economic losses, resource waste, and social instability due to project delays and shutdowns, thereby undermining the industry’s sustainable development [4,5].

Unlike other manufacturing sectors (e.g., automotive, electronics), GCSSCs are characterized by high modularity, long lead times, geographically concentrated high-tier suppliers, and strong dependence on European-centric support industries [1,6]. These distinctive features render conventional resilience tactics (e.g., inventory buffering) less effective and demand a context-specific, hierarchical understanding of resilience capabilities. While DEMATEL–ISM–MICMAC has been applied in other supply chain contexts, its integration with CAS theory and a four-capability resilience framework provides a novel lens to capture the dynamic, multi-level interactions unique to GCSSCs—a gap not yet addressed in extant literature. Strengthening GCSSC resilience is, therefore, not merely a tactical goal but a fundamental prerequisite for achieving long-term economic and operational sustainability.

The GCSSC’s distinctive features—high modularity, strong component specificity, and extensive foreign dependency—make conventional resilience tactics such as inventory buffering less effective [6,7]. Hence, contextualized resilience frameworks are urgently needed. Yet, existing research has predominantly focused on downstream cruise operations, leaving the resilience of upstream supply systems critically underexplored from a sustainability perspective [8,9,10,11]. Furthermore, most studies examine factors in isolation, overlooking their complex interdependencies and thus failing to provide a holistic understanding of how to build a resilient and sustainable supply ecosystem [7,12,13]. The literature also prioritizes production efficiency, while the strategic configuration of capabilities for systemic resilience remains overlooked.

To address these gaps, this study investigates:

(1) Which factors are critical for building resilience in the GCSSC?

(2) How do these factors interact and propagate to collectively influence overall supply chain resilience?

Thus, the main objective of this study is to develop a structured, hierarchical understanding of how resilience capabilities are configured and interact within GCSSCs, and to derive actionable strategic pathways for building such resilience.

To achieve this objective, we first conceptualize GCSSC resilience as an emergent property arising from dynamic interactions among multiple actors, guided by complex adaptive systems (CAS) theory [14]. We integrate engineering, ecological, and evolutionary resilience perspectives to delineate a portfolio of four capabilities: predictive, resistant, restorative, and transformative. To decode the complex interdependencies among the factors underlying these capabilities, we then employ a hybrid fuzzy DEMATEL–ISM–MICMAC framework. This integrated methodology is particularly suited to mapping the causal and hierarchical relationships that underpin a sustainable resilience structure.

By pursuing this objective, this study aims to contribute to the discourse on sustainable operations in three ways. First, by developing a comprehensive resilience capability framework tailored to GCSSC, we extend the theory of sustainable supply chains to complex, project-based manufacturing. Second, by unveiling the interaction mechanisms and identifying critical investment levers, we clarify how resources can be allocated most effectively to build resilience. Third, by proposing resource-contingent pathways for resilience enhancement, we provide actionable guidance for managers to navigate disruptions while advancing the economic sustainability and competitive longevity of their firms and the broader industrial ecosystem.

The remainder of the paper is organized as follows. Section 2 reviews the literature. Section 3 describes the conceptual framework, factor identification, and fuzzy DEMATEL–ISM–MICMAC methodology. Section 4 reports the empirical application and results. Section 5 discusses the findings and implications. Section 6 concludes with limitations and future research directions.

2. Literature Review

2.1. Complex Adaptive Systems Theory and Sustainability

The concept of Complex Adaptive Systems (CAS) is derived from complexity theory and was initially applied to living systems [15]. CAS theory explains how order emerges within systems that are complex and nonlinear, emphasizing how systems evolve through interactions among agents and make adaptive adjustments through agent-environment interactions [16,17]. The application of CAS theory in supply chain management reveals that supply chains exhibit typical CAS features such as distributed control, self-organization, and emergent behaviors [18], then is expanded in a robust CAS framework centered on three critical elements: internal mechanisms, the environment, and co-evolution [19].

Internal mechanisms encompass agents, self-organization, emergence, connectivity, heterogeneity, and dimensionality [19]. Agents—individual firms or groups—can take meaningful action during disruptions, often cooperating through alliances [15]. Self-organization indicates that system behaviors arise from the parallel actions of multiple agents without central coordination. Through this process, new structures or properties emerge unexpectedly, a phenomenon known as emergence [14,18]. Key characteristics like connectivity (interdependence among agents), heterogeneity (diversity of units and actors), and dimensionality (degree of agent autonomy) are essential for understanding a system’s adaptive capacity [15,19,20].

The environment in a CAS consists of external agents and dynamics with which the system interacts. Crucially, the environment is not static but evolves dynamically as system actors respond to external conditions [19]. This interaction leads to co-evolution, a process where systems and their environments mutually reshape each other through feedback loops, often involving both competitive and cooperative relationships [16,18,21].

From a sustainability perspective, CAS theory provides a vital lens for understanding how supply chains can maintain their viability and performance—a core aspect of economic and operational sustainability—amidst turbulent environments. The theory’s emphasis on adaptation, self-organization, and co-evolution aligns directly with the principles of sustainable systems that must continuously adjust to survive and thrive. Therefore, CAS offers a theoretically grounded framework to capture the interdependent and dynamic nature of resilience in GCSSC, moving beyond linear models that are inadequate for fostering long-term sustainable operations.

2.2. Resilience Capabilities of the Growth-Stage Cruise Shipbuilding Supply Chains

Within the CAS framework, a GCSSC is composed of heterogeneous agents whose operations depend on complex interactions and continuous adaptation [14,19]. This complexity necessitates a multi-faceted approach to resilience. Synthesizing three key resilience perspectives—engineering (focusing on return to efficiency) [22], ecological (emphasizing stability domains and buffering capacity) [23], and evolutionary (stressing adaptation and transformation)—we delineate four interrelated capabilities that form a full resilience cycle [13,22,24]. These capabilities are interdependent and path-dependent, consistent with CAS principles, and collectively form the conceptual foundation for identifying GCSSC resilience factors with a focus on long-term sustainability.

Predictive Capability (preparation): The ability to anticipate and prepare for disruptions, rooted in engineering resilience’s emphasis on forecasting [22]. This capability is crucial for preventing resource waste and promoting efficient operations.

Resistant Capability (resistance): The ability to withstand disruptions with minimal performance degradation. This capability, central to both engineering and ecological resilience, ensures system stability and continuity under stress, thereby safeguarding economic resources [21].

Restorative Capability (recovery): The ability to recover system functionality after a disruption. While engineering resilience highlights speed, ecological resilience emphasizes reorganization [23]. Efficient recovery is fundamental to minimizing economic and environmental losses.

Transformative Capability (transformation): The ability to adapt and emerge stronger from major disruptions. Grounded in evolutionary resilience, it involves structural change, knowledge renewal, and innovation [13,24]. This capability is paramount for a supply chain to evolve towards more sustainable and robust configurations in a changing world.

Together, these capabilities form a full cycle—preparation, resistance, recovery, transformation—consistent with CAS: they are interdependent, path-dependent, and multilevel. We adopt this four-capability structure as the conceptual foundation to identify GCSSC resilience factors.

2.3. Factors Shaping Supply Chain Resilience

Research on resilience antecedents spans cross-industry frameworks and industry-specific studies. Cross-industry studies identify foundational factors such as leadership, collaboration, agility, redundancy, and business continuity planning [25,26]. Furthermore, learning mechanisms, social capital, and adaptive governance are emphasized as enablers that translate resources into resilient performance [24]. Industry-specific studies, particularly in shipbuilding, highlight contextual risks and capabilities, such as production errors in military projects [27] or the role of Industry 4.0 technologies in enhancing competitiveness [28]. Notably, a growing body of literature in outlets like Sustainability explores the nexus of digitalization, resilience, and sustainability, examining how technologies like blockchain and big data can contribute to more transparent and sustainable supply chains [29].

Previous research on cruise shipbuilding has examined the industry from multiple perspectives, some of which directly engage with sustainability themes. From a sustainability and quality perspective, studies have assessed comprehensive sustainability indicators, lifecycle environmental impacts, and resource efficiency in cruise design and construction [30,31]. Other research has explored green shipbuilding practices and waste reduction strategies, aligning with the environmental pillar of sustainability. Furthermore, studies have proposed advanced quality-control systems to minimize defects and resource waste [13,30], and discussions on the role of digital transformation in enhancing both sustainability and operational performance are also evident [28,31]. From risk and flexibility perspectives, research highlights information asymmetries and proposes adaptive sourcing and logistics models to handle disruptions [3]. From organizational and operational perspectives, studies have examined lean principles and digital solutions to improve efficiency [7,31].

However, a significant gap remains, as synthesized in

Table 1. Summary of Key Literature on Supply Chain Resilience and Methodological Approaches.

Author(s) and Year	Context/Industry	Methodology	Key Findings	Gaps/Limitations	Positioning of This Study
Choi et al. (2001) [18]	General supply chains	Conceptual/CAS theory	Supply chains exhibit CAS properties: distributed control, self-organization, emergence.	Lacks empirical validation; no industry-specific resilience framework.	Adopts CAS as theoretical lens; extends to GCSSC with empirical validation.
Tukamuhabwa et al. (2015) [14]	General supply chains	Systematic literature review	Defines resilience, reviews theoretical foundations, identifies antecedents and outcomes.	Limited focus on project-based, high-value manufacturing and capability hierarchies.	Tailors’ resilience antecedents to GCSSC; structures them into hierarchical capabilities.
Adobor & McMullen (2018) [24]	Cross-industry	Conceptual/Dynamic model	Resilience is multi-dimensional; emphasizes learning, adaptation, and transformative capacity.	Lacks empirical validation and industry-specific contextualization.	Provides empirical, context-specific framework for GCSSC; integrates learning and transformation.
Nair & Reed-Tsochas (2019) [19]	Operations and SCM	Literature review/CAS framework	Proposes robust CAS framework with three elements: internal mechanisms, environment, co-evolution.	Limited focus on project-based, high-value manufacturing contexts.	Applies CAS framework to GCSSC; integrates resilience capabilities.
Könnölä et al. (2020) [30]	Cruise shipbuilding/Sustainability	Case study/Certification analysis	Identifies sustainability indicators and lifecycle impacts in cruise vessel design.	Limited focus on resilience and dynamic capability interactions.	Integrates sustainability with resilience; links green practices to restorative and transformative capabilities.
Wieland & Durach (2021) [21]	Cross-industry	Conceptual/Perspective analysis	Distinguishes engineering vs. ecological resilience; highlights adaptation and transformation.	Does not integrate multiple resilience perspectives into a unified capability model.	Synthesizes engineering, ecological, and evolutionary resilience into a four-capability framework.
Ivanov & Dolgui (2021) [32]	Industry 4.0/Digital twins	Conceptual/Simulation	Digital twins enhance disruption risk management and resilience in dynamic environments.	Focuses on digital tools, not on hierarchical factor interactions or capability structuring.	Positions digital technology as an enabling factor within a broader hierarchical resilience model.
Zhu et al. (2021) [3]	Shipbuilding material supply chain	Fuzzy DEMATEL-QFD	Flexible strategy design for material procurement under uncertainty.	Focus on design phase only; lacks system-level resilience modeling.	Extends to full GCSSC resilience with hybrid DEMATEL-ISM-MICMAC; integrates CAS theory.
Yaroson et al. (2021) [16]	Pharmaceutical SCM	CAS theory/Qualitative	CAS explains resilience in pharma supply chains through adaptation and self-organization.	Industry-specific; limited generalizability to project-based manufacturing.	Transfers CAS insights to GCSSC; contextualizes adaptation in cruise shipbuilding.
Centobelli et al. (2023) [28]	Shipbuilding/Digital SCM	Empirical survey	Digital technologies enhance resilience and competitiveness in shipbuilding supply chains.	Static analysis; lacks hierarchical interaction modeling and causal analysis.	Models dynamic, multi-level interactions via ISM; integrates digital tech as a transformative factor.
This study	Growth-stage Cruise Shipbuilding Supply Chains (GCSSC)	Hybrid fuzzy DEMATEL-ISM-MICMAC integrated with CAS theory	Identifies 20 resilience factors; reveals hierarchical and causal relationships; proposes resource-contingent pathways.	N/A	Provides an integrated, hierarchical, and actionable resilience framework specific to GCSSC, filling gaps in factor interaction modeling and strategic pathway development.

. While these studies provide valuable insights, they tend to be single-dimensional. They often overlook the complex interdependencies between the identified factors. Moreover, they fail to articulate how these factors interact dynamically across the supply network to build system-level resilience. The dynamic and emergent properties of resilience, which are central to CAS and crucial for long-term adaptive sustainability, are largely neglected. There is a lack of a structured understanding of how factors collectively configure to build the predictive, resistant, restorative, and transformative capabilities needed for a GCSSC to be both resilient and sustainable.

3. Methods

3.1. Research Procedure

This study follows a structured, multi-phase research procedure to ensure methodological rigor and clarity:

(1) Literature Review: Conducted to synthesize resilience perspectives, identify key capabilities, and review existing factor studies (Section 2).

(2) Conceptual Model Development: Built on CAS theory and the four resilience capabilities to map resilience onto GCSSC processes (Section 3.2.1).

(3) Factor Identification: Derived 20 GCSSC-specific resilience factors through structured literature synthesis and three rounds of expert consultation (Section 3.2.2).

(4) Fuzzy DEMATEL Analysis: Applied to quantify causal relationships among factors using expert judgments transformed via triangular fuzzy numbers (Section 3.3.2).

(5) ISM Analysis: Used to derive the hierarchical structure of factors based on reachability matrices (Section 3.3.3).

(6) MICMAC Analysis: Conducted to classify factors by driving and dependence power (Section 3.3.4).

These steps are integrated into a coherent hybrid framework, as visualized in Figure 1.

3.2. Framework and Factors

3.2.1. Conceptual Framework

The conceptual model (Figure 2) was developed based on the integration of CAS theory and the four resilience capabilities derived from the literature review. It visually maps these capabilities onto the key nodes and processes of the GCSSC to provide a structured basis for identifying resilience factors. The GCSSC is a tightly coupled network orchestrated around the final-assembly shipyard, integrating diverse actors such as shipowners, design firms, and multi-tier suppliers [3,6]. Its primary operational objective is to assure material availability—delivering the right materials in the right quantity, quality, and time to maintain schedule integrity [33]. In a capital- and resource-intensive industry, achieving this objective is fundamental to economic and operational sustainability, as failures directly lead to massive resource waste, financial losses, and project failures [4,5].

Compared with traditional shipbuilding, the GCSSC demands more precise delivery and stricter control of logistics nodes [1,6]. Consequently, disruptions at any key node—whether design changes, supplier bottlenecks, or logistics constraints—can propagate, causing schedule delays and cost overruns [2,34]. These disruptions severely undermine the economic and environmental sustainability of projects through inefficiencies and waste [4,32].

To address this, we conceptualize the GCSSC resilience not merely as risk mitigation but as a cornerstone of sustainable operations. Grounded in Complex Adaptive Systems (CAS) theory, we integrate four resilience capabilities—predictive, resistant, restorative, and transformative—that form a dynamic cycle essential for long-term viability. As visualized in Figure 2, our conceptual model maps these capabilities onto the key nodes and processes of the GCSSC, creating a foundational framework for identifying factors that build a resilient and sustainable supply chain. This model directly enables the subsequent analysis to pinpoint key investment levers for sustainable resilience.

Figure 2 illustrates how predictive, resistant, restorative, and transformative capabilities are mapped onto key GCSSC nodes (e.g., design, procurement, assembly, logistics). The model emphasizes the dynamic interactions between capabilities across disruption phases, providing a holistic view of resilience as an emergent property of the system.

3.2.2. Factors Shaping Cruise Shipbuilding Supply Chain Resilience

Based on the conceptual model, we identified twenty GCSSC-specific resilience factors through a structured literature review and expert consultation (three rounds with senior managers from final-assembly yards, first-line suppliers, and logistics providers). Factors were organized under four resilience capabilities: predictive capability (A), resistant capability (B), restorative capability (C), and transformative capability (D).

Table 2. GCSSC resilience factors.

Factors	Indicators	Description
Predictive capability (A)	Schedule deviations (A1)	Deviations between baseline and actual milestones arising from incremental owner confirmations, evolving technical requirements, or upstream supply delays.
	Modular management teams (A2)	Monitoring risks and facilitating communication among contractors and technical domains.
	Intelligence management (A3)	Timely collection and use of policy, market, and geopolitical signals, particularly for cross-border logistics corridors.
	Knowledge stock (A4)	Knowledge sharing, resource integration, acquisition capability, and experiential learning.
	Risk culture (A5)	Shared norms and incentives for early escalation and pre-emptive decision-making.
Resistant capability (B)	Logistics integration (B1)	Coordination of multimodal transport, out-of-gauge/heavy -lift handling, and differentiated warehousing conditions.
	Component and module manufacturing performance (B2)	Quality consistency, supply stability, and integration reliability for high-specificity components and functional modules.
	Schedule discipline (B3)	Milestone adherence for long-announced sail-away dates with limited slack.
	Strategic supplier stability (B4)	Stable ties or long-term cooperation agreements with high-end, low-substitutability suppliers of core equipment and technologies in concentrated global niches.
	Industry backups (B5)	Dual sourcing where feasible, inventory buffers, and contingency logistics channels to provide short-term flexibility.
Restorative capability (C)	Final-assembly orchestration (C1)	Strategic leadership by the final-assembly yard to coordinate upstream/downstream recovery and re-baselining.
	Inter-industry collaboration (C2)	Joint problem solving and decision rights spanning adjacent sectors (e.g., marine engineering, luxury fit-out).
	Social capital (C3)	External networks and relational trust that mobilize resources quickly under time pressure.
	Structural configuration (C4)	Balanced geographic dispersion and tier coordination that reduce bottlenecks and enable parallel recovery.
	Specialized industrial clusters (C5)	Regional co-location that accelerates knowledge/resource sharing.
Transformative capability (D)	Learning organization (D1)	Continuous learning and knowledge renewal that support capability reconstruction; purposeful knowledge inflows from related industries.
	Industrial autonomy (D2)	Localization and control over critical technologies and supply segments to reduce structural dependence.
	Information sharing (D3)	Transparent, timely multi-party data exchange improving visibility and joint responsiveness.
	Customized demand responsiveness (D4)	Capability to translate unique shipowner/passenger requirements into adaptive material planning, procurement, and logistics for specialized fixtures and systems.
	Digital technologies (D5)	Digitally enabled coordination (e.g., integrated PLM/ERP/MES, digital twins) to manage complexity and overcome localization bottlenecks.

reports factors, indicators, and their description.

Predictive capability (A). This capability is fundamental to preemptive resource management and waste minimization, key tenets of sustainable operations. Schedule deviations (A1) often arise from discrepancies between original plans and actual execution, caused by incremental confirmations from shipowners, evolving technical requirements, or unexpected supply delays [1,33]. A comprehensive tracking and forecasting mechanism are therefore crucial to mitigate these inefficiencies [2,3]. Modular management teams (A2) play a vital role in monitoring risks and facilitating communication among contractors and technical domains to maintain schedule stability [12]. Effective intelligence management (A3), which involves the timely collection and application of policy and market information, is essential for accurate risk prediction, particularly in cross-border logistics scenarios [32,34]. Furthermore, accumulated knowledge (A4) and a well-established risk culture (A5) collectively improve the organization’s ability to anticipate risks, support timely and informed decision-making, and contain the spread of disruption [35,36,37,38].

Resistant capability (B). This capability ensures operational continuity when supply chain members encounter unexpected disruptions. Given the wide variety of materials involved with specific storage requirements, transportation methods, and delivery timing, logistics integration capability (B1) coordinates multi-modal transport, warehousing and oversized cargo management [6,7]. And the manufacturing performance of key components and modules (B2), reflected in quality consistency and supply stability, directly affects the supply chain’s ability to withstand disruptions [2,3,13]. Strict construction schedules (B3) are essential to avoid the massive financial and reputational losses associated with missing fixed milestones and sailing dates [7,33]. Moreover, Maintaining stable relationships with high-end suppliers (B4) secures reliable access to critical technologies, while industrial backup systems (B5) provide flexibility. For substitutable products, establishing industrial backup systems through multi-source procurement, inventory buffers, and alternative logistics channels (B5) offers crucial flexibility, enabling the supply chain to absorb shocks and sustain operations during unexpected event [1,39].

Restorative capability (C). This capability accelerates the return to normalcy, minimizing the duration and impact of disruptions. The strategic leadership of the final-assembly plant (C1) is critical for orchestrating coherent recovery efforts across the network [1,6,40,41,42,43]. Inter-industry collaboration (C2) further enhances this capability by fostering joint risk response, supported by incentives and shared decision-making [3,31,41]. Meanwhile, social capital (C3), rooted in the relational trust, enables rapid mobilization of resources during crises [44]. A well-structured supply chain with balanced geographic distribution and coordinated hierarchies (C4), also facilitates to faster recovery by minimizing structural bottlenecks [45]. Finally, specialized industrial clusters (C5) promote regional synergies and facilitate resource and knowledge sharing [46]. While, countries in the growth phase of cruise shipbuilding often face challenges from scattered and immature domestic suppliers and lack of systematic clustering [1,2].

Transformative capability (D). This capability is central to evolving the GCSSC into a more sustainable and future-proof system. Learning organizations (D1) establish mechanisms for continuous learning and knowledge renewal, fostering a positive virtuous cycle of experience accumulation and capability reconstruction within the organization. At the same time, they actively absorb knowledge from related industries such as luxury hospitality, maritime engineering, and advanced manufacturing [1]. This internal-external knowledge integration enables organizations to more effectively adapt to market dynamics and technological evolution, thereby supporting the progression of GCSSC toward higher levels of maturity [47]. Industrial autonomy (D2) involves localizing critical supply chain segments and achieving control over core technologies. In the GCSSC, innovation-driven self-reliance is essential for reducing structural dependence on overseas suppliers [1,31]. Moreover, information sharing (D3) is critical in cruise shipbuilding’s international, multi-industry collaboration [44,46]. Transparent information flow improves supply chain visibility and enhances responsiveness to disruptions. The industry’s unique customized demand response capability (D4) captures specialized requirements from cruise lines seeking distinctive passenger experience. Shipowners specify vessel capability, route capabilities, and brand-specific amenities, while passenger preferences drive innovations in entertainment venues and stateroom configuration [1,28]. These dynamic and personalized demands necessitate highly adaptive material planning, procurement schedules, and logistics factors, directly influencing the sourcing of specialized materials and custom fixture. Finally, digital technology (D5) enables integration of predictive maintenance technologies, and smart ship capabilities. This allows supply chains to coordinate complex construction processes and overcome localization bottlenecks in specialized marine equipment manufacturing [31,32,48].

3.3. DEMATEL-ISM-MICMAC

We employ the triangular fuzzy numbers (TFNs)–based DEMATEL–ISM–MICMAC framework to (i) identify causal relationships among resilience factors (DEMATEL), (ii) derive their hierarchical structure (ISM), and (iii) classify factors by driving and dependence power (MICMAC). TFNs reduce subjectivity in expert judgments by capturing linguistic uncertainty. Figure 3 summarizes the research framework.

3.3.1. Data Collection

We used a group decision-making approach with 10 experts: five ship construction operations managers, three project management professionals, and two professors specializing in maritime supply chains. Their profiles are shown in

Table 3. Profiles of selected experts.

Expert	Type of Organization	Designation	Relevance Phase in GCSSC	Years of Experience
1	Cruise shipyard	Production manager	Material procurement, construction, assembly	12
2	Cruise shipyard	Quality assurance manager	Construction, testing, delivery	8
3	Cruise shipyard	Production manager	Design, procurement, construction coordination.	13
4	Maritime logistics services	Operations manager	Material transportation, delivery scheduling	15
5	Cruise company	Cruise operation manager	Design requirements, delivery, operations	13
6	Cruise shipyard	Project manager	Material sourcing, component delivery	8
7	Cruise company	Senior project coordinator	Multi-phase project management, delivery	9
8	Maritime consulting firm	Senior consultant	Project management, supply chain factor	10
9	University	Professor of naval architecture	Design, construction, manufacturing, supply chain optimization	13
10	University	Professor of supply chain management	Supply chain management, process optimization	15

. Selection criteria included role relevance to upstream supply and logistics and familiarity with cruise projects.

Experts rated the influence of factor i on factor j on a 0–4 linguistic scale: 0 = No influence, 1 = Low, 2 = Moderate, 3 = High, 4 = Very high. Linguistic terms were mapped to TFNs using a standard scheme in

Table 4. Fuzzy interpretation for expert’s score.

Numeric Score	Linguistic Term	TFN ($\mathit{l},\mathit{m},\mathit{n}$)
0	No influence	(0.00, 0.00, 0.25)
1	Low	(0.00, 0.25, 0.50)
2	Moderate	(0.25, 0.50, 0.75)
3	High	(0.50, 0.75, 1.00)
4	Very high	(0.75, 1.00, 1.00)

. TFNs could reduce expert subjectivity in two key ways. First, TFNs convert linguistic assessments (e.g., “High” influence) into a mathematical form that captures the spectrum of possible interpretations, thus modeling linguistic uncertainty. Second, they enable a more robust aggregation of group opinions. By compiling fuzzy assessments from all experts before defuzzification, the final direct-relation matrix Z synthesizes the panel’s collective judgment more accurately and democratically than a simple average of crisp scores.

3.3.2. Fuzzy DEMATEL

Fuzzy DEMATEL procedures quantifies inter-factor influences which begins with standardizing fuzzy expert judgments and proceeds to defuzzify them into a direct-relation matrix, following established fuzzy multi-criteria decision-making methodologies [49,50]. The specific steps are as follows.

Step 1: Fuzzy number standardization. Let N be the number of factors (N = 20). For each expert k, the fuzzy direct influence of i on j is a TFN ${\tilde{a}}_{ij}^k=(l_{ij}^k,m_{ij}^k,u_{ij}^k)$, with ${\tilde{a}}_{ii}^k=(0,0,0)$. The experts’ TFNs are obtained as ${\tilde{a}}_{ij}^k=(l_{ij}^k,m_{ij}^k,u_{ij}^k)$ and these fuzzy numbers are then standardized to [0, 1] using Equations (1)–(3).

$$x{{\displaystyle l}}_{ij}^k={\displaystyle \frac{l_{ij}^k-min(l_{ij}^k)}{{\mathrm{\Delta }}_{min}^{max}}}$$

(1)

$${{\displaystyle xm}}_{ij}^k={\displaystyle \frac{m_{ij}^k-min(l_{ij}^k)}{{\mathrm{\Delta }}_{min}^{max}}}$$

(2)

$${{\displaystyle xu}}_{ij}^k={\displaystyle \frac{u_{ij}^k-min(l_{ij}^k)}{{\mathrm{\Delta }}_{min}^{max}}}$$

(3)

where ${∆}_{min}^{max}=max\left(u_{ij}^k\right)-min\left(l_{ij}^k\right)$.

Step 2: Fuzzy number decomposition and defuzzification. According to Equations (4)–(8), the converting fuzzy data into crisp scores (CFCS) method is applied to obtain the crisp scores Z_ij, resulting in the crisp direct-relation matrix Z [49].

$${{\displaystyle xls}}_{ij}^k={\displaystyle \frac{x{m}_{ij}^k}{1+x{m}_{ij}^k-x{l}_{ij}^k}}$$

(4)

$${{\displaystyle xms}}_{ij}^k={\displaystyle \frac{x{u}_{ij}^k}{1+x{u}_{ij}^k-x{m}_{ij}^k}}$$

(5)

$${{\displaystyle x}}_{ij}^k={\displaystyle \frac{{{\displaystyle xls}}_{ij}^k·(1-{{\displaystyle xls}}_{ij}^k)+{{\displaystyle xms}}_{ij}^k·{{\displaystyle xms}}_{ij}^k}{1-{{\displaystyle xls}}_{ij}^k+{{\displaystyle xms}}_{ij}^k}}$$

(6)

$${{\displaystyle z}}_{ij}^k=min(l_{ij}^k)+x_{ij}^k·{\mathrm{\Delta }}_{min}^{max}$$

(7)

$$z_{ij}={\displaystyle \frac{1}{k}}\left(z_{ij}^1+z_{ij}^2+\cdots +z_{ij}^k\right)$$

(8)

Step 3: Normalization process to obtain the normalized influence matrix G. Normalize the direct-relation matrix Z to ensure that its spectral radius is less than 1, and obtain the normalized influence matrix G using the row-sum method via Equation (9).

$$G={\displaystyle \frac{z_{ij}}{\max \left({\displaystyle \sum _{j=1}^n{z}_{ij}}\right)}}$$

(9)

Step 4: Calculate the comprehensive impact matrix T. The comprehensive impact matrix T is calculated according to Equation (10).

$$T=G{(I-G)}^{-1}$$

(10)

where I is the identity matrix.

Step 5: Cause–effect metrics. Determine each factor’s degree of influence C_i, degree of being influenced R_i, centrality M_i, and causal degree D_i. A larger C_i indicates stronger influence; a larger R_i indicates higher susceptibility. M_i reflects the overall importance of a given factor. A positive D_i indicates a net cause, while a negative D_i indicates a net effect.

$$C_i={\displaystyle \sum _{j=1}^n{t}_{ij}},\hspace{1em}i=1,2,\dots ,n$$

(11)

$$R_i={\displaystyle \sum _{j=1}^n{t}_{ji}},\hspace{1em}i=1,2,\dots ,n$$

(12)

$$M_i=R_i+C_i$$

(13)

$$D_i=R_i-C_i$$

(14)

3.3.3. ISM Hierarchy Derivation

The hierarchical structure of factors is derived using ISM, a method widely applied in supply chain sustainability studies [51,52,53], which starts with the comprehensive impact matrix from DEMATEL. A threshold is then applied to obtain a binary adjacency matrix, which is processed to yield the reachability matrix [54]. Finally, level partitioning is conducted to establish the multi-level model, following established ISM procedures [55]. The specific steps are as follows.

Step 1: Calculate the overall impact matrix H based on the comprehensive impact matrix T and Equation (15).

$$H=T+I$$

(15)

Step 2: Set the threshold λ and calculate the adjacency matrix B. By setting the threshold λ via Equation (16), the overall impact matrix H is converted into the adjacency matrix B.

$$\lambda =u+v$$

(16)

where u is the mean value of all elements in T and v is their standard deviation. The adjacency matrix B is as follows.

$$b_{ij}=\left\{\begin{array}{ll}1,\hfill & h_{ij}\ge \lambda \hfill \\ 0,\hfill & h_{ij}<\lambda \hfill \end{array}\right.,\hspace{1em}(i=1,2,\dots ,n; j=1,2,\dots ,n),\hspace{1em}B={[b_{ij}]}_{n\times n}$$

(17)

Step 3: Calculate the reachability matrix M. Based on the adjacency matrix B, utilizing the operational properties of Boolean matrices, successive Boolean multiplication operations are performed on matrix B until the matrix reaches a convergent state, ultimately obtaining the reachability matrix M.

$$M={(B+I)}^{n+1}={(B+I)}^{n+2}\ne {(B+I)}^n\ne B+I$$

(18)

Step 4: Level partitioning. Compute the reachability set E_i, antecedent set S_i, and their intersection Q_i. The reachability set includes all factors that can be reached from a given factor, while the antecedent set includes all factors that can reach the given factor.

$$E(a_i)=\{a_i\mid m_{ij}=1\}$$

(19)

$$S(a_i)=\{a_i\mid m_{ji}=1\}$$

(20)

$$Q(a_i)=E(a_i)\cap S(a_i)$$

(21)

3.3.4. MICMAC Classification

Factors are classified based on their driving and dependence power via MICMAC analysis, a cross-impact matrix multiplication method often integrated with ISM in supply chain studies [51,52]. The driving and dependence forces are computed from the weighted total relation matrix. Factors are then plotted and categorized into four clusters (autonomous, dependent, linkage, independent) for comparative analysis.

Step 1: Calculate driving force W_i and dependent force Y_i. They are computed Based on the reachability matrix M (binary) or the total relation matrix T (weighted). Specifically, the former is the sum of the i-th row of T, and the latter is the sum of the i-th column of T.

Step 2: Classify factors into quadrants. According to the driving force and dependence force, factors are divided into four categories: autonomous cluster, dependent cluster, linkage cluster, independent cluster.

4. Results

4.1. Cause–Effect and Centrality Analysis of Fuzzy DEMATEL Results

The fuzzy DEMATEL analysis first identifies the causal roles and systemic importance of each factor. The cause–effect metrics (e.g., the affected degree, influence degree, centrality and causal degree), presented in

Table 5. Cause–effect metrics of GCSSC resilience factors.

Factor	R	C	M	D	Type
A1	1.881	0.654	2.535	−1.227	Result
A2	1.359	0.957	2.316	−0.402	Result
A3	1.503	0.793	2.296	−0.710	Result
A4	1.100	1.678	2.778	0.578	Cause
A5	1.562	1.796	3.358	0.234	Cause
B1	1.414	0.868	2.282	−0.546	Result
B2	2.225	0.665	2.890	−1.560	Result
B3	2.113	0.876	2.989	−1.237	Result
B4	1.507	1.039	2.546	−0.468	Result
B5	2.019	0.453	2.472	−1.566	Result
C1	0.946	1.771	2.717	0.825	Cause
C2	0.992	1.982	2.974	0.990	Cause
C3	0.761	1.550	2.311	0.789	Cause
C4	1.132	1.049	2.181	−0.083	Result
C5	0.617	1.990	2.607	1.373	Cause
D1	0.580	2.028	2.608	1.448	Cause
D2	1.222	1.336	2.558	0.114	Cause
D3	1.518	0.961	2.479	−0.557	Result
D4	0.461	2.283	2.744	1.822	Cause
D5	1.394	1.578	2.972	0.184	Result

and Figure 4, reveal the foundational drivers and key outcomes of the system.

The centrality analysis (Figure 4) highlights ten factors as the system’s most influential and susceptible elements. A key insight is the frequent appearance of transformative (D) and restorative (C) factors, suggesting that system-level coordination and renewal mechanisms are pivotal to GCSSC resilience. The ten most central factors are risk culture (A5), schedule discipline (B3), inter-industry collaboration (C2), digital technologies (D5), component and module manufacturing performance (B2), knowledge stock (A4), customized demand responsiveness (D4), final-assembly orchestration (C1), learning organization (D1), specialized industrial clusters (C5). These factors either influence many others, are strongly influenced by others, or both.

In terms of causal structure, D_i > 0 indicates that the factor is a cause factor that impacts other factors, while D_i < 0 suggests that the factor is a result factor affected by other factors. The strongest cause factors are customized demand responsiveness (D4), learning organization (D1), specialized industrial clusters (C5), inter-industry collaboration (C2), final-assembly orchestration (C1). These deep drivers primarily reside in the transformative (D) and restorative (C) capability sets, pointing to learning, orchestration, collaboration, clustering, and responsiveness as leverage points for system-wide improvement.

Conversely, the strongest result factors are industrial backups (B5), component and module manufacturing performance (B2), schedule discipline (B3), schedule deviations (A1), intelligence management (A3). These tend to be direct performance outcomes that materialize once drivers are in place. Practically, they serve well as KPIs to monitor resilience performance but are less suitable as first targets for investment without strengthening the upstream drivers.

Aggregating by capability, the mean centrality is higher for transformative, and restorative factors than for resistant and predictive factors in our sample, indicating that system-level coordination and renewal mechanisms sit at the core of GCSSC resilience. At the same time, many resistant and predictive factors exhibit negative D_i, consistent with their role as immediate outcomes that are themselves shaped by higher-level drivers. This centrality finding for factors like risk culture (A5) and knowledge stock (A4) aligns with prior research emphasizing cultural and intellectual capital as critical enablers of resilience in complex supply chains [36,47].

4.2. Hierarchy Analysis of ISM Results

Based on the causal results from DEMATEL, the ISM analysis elucidates the hierarchical structure and the transmission paths of influences. The threshold λ is set at 0.1, determined by applying the common “mean plus one standard deviation” criterion to the elements of matrix T. This approach in ISM studies to balance structural clarity with completeness [51,52]. The reachability matrix M was then obtained via Equation (18), as shown in

Table 6. Reachability matrix M.

	A1	A2	A3	A4	A5	B1	B2	…	C4	C5	D1	D2	D3	D4	D5
A1	1	0	0	0	0	0	1	…	0	0	0	0	0	0	0
A2	0	1	0	0	0	0	0	…	0	0	0	0	0	0	0
A3	0	0	1	0	0	0	0	…	0	0	0	0	0	0	0
A4	0	0	0	1	1	0	0	…	0	0	0	0	0	0	1
A5	1	1	1	0	1	1	1	…	0	0	0	1	1	0	0
B1	1	0	0	0	0	1	1	…	0	0	0	0	0	0	0
…
C4	0	0	0	0	0	0	0	…	1	0	0	0	0	0	0
C5	1	1	0	1	0	1	1	…	1	1	0	0	1	0	1
D1	0	0	0	1	1	0	1	…	0	0	1	0	1	0	1
D2	1	0	0	0	0	0	1	…	0	0	0	1	0	0	0
D3	0	0	0	0	0	0	0	…	0	0	0	0	1	0	0
D4	1	1	1	1	1	1	1	…	0	0	0	0	1	1	1
D5	1	0	1	0	0	0	1	…	0	0	0	0	0	0	1

.

Based on Equations (19) and (20) and M, we computed for each factor a_i its reachability set E, antecedent set S, and intersection Q = E ∩ S which were summarized in

Table 7. Reachability, antecedent, and intersection sets.

Factor	Reachability Set E	Antecedent Set S	Intersection Set Q
A1	1	1, 5, 6, 11, 15, 17, 19, 20	1
A2	2, 7	2, 5, 11, 12, 15, 19	2
A3	3	3, 5, 11, 19, 20	3
A4	4, 5, 7, 8, 20	4, 12, 15, 16, 19	4
A5	1, 2, 3, 5, 6, 7, 8, 9, 10, 17, 18	4, 5, 12, 13, 16, 19	5
B1	1, 6, 7	5, 6, 15, 19	6
B2	7	2, 4, 5, 6, 7, 9, 11, 12, 13, 15, 16, 17, 19, 20	7
B3	8	4, 5, 8, 11, 12, 13, 15, 16, 19, 20	8
B4	7, 9, 10	5, 9, 11, 12, 15, 16, 19	9
B5	10	5, 9, 10, 11, 12, 15, 16, 17, 18, 19, 20	10
C1	1, 2, 3, 7, 8, 9, 10, 11, 18, 20	11	11
C2	2, 4, 5, 7, 8, 9, 10, 12, 18, 20	12, 16, 19	12
C3	5, 7, 8, 13	13	13
C4	14	14, 15	14
C5	1, 2, 4, 6, 7, 8, 9, 10, 14, 15, 18, 20	15	15
D1	4, 5, 7, 8, 9, 10, 12, 16, 18, 20	16	16
D2	1, 7, 10, 17	5, 17	17
D3	10, 18	5, 11, 12, 15, 16, 18, 19	18
D4	1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 12, 18, 19, 20	19	19
D5	1, 3, 7, 8, 10, 20	4, 11, 12, 15, 16, 19, 20	20

.

The reachability matrix M (Table 6) and the derived reachability/antecedent sets (Table 7) form the basis for level partitioning. Factors with identical reachability and antecedent sets are assigned to the same hierarchical level, resulting in the multi-level model shown in Figure 5.

The resulting multi-level model in Figure 5 illustrates how effects propagate from foundational drivers to surface-level outcomes.

Foundational factors represent the fundamental factor that impacts the resilience of GCSSC, causing changes in the overall resilience by influencing all middle and outcome factors. Customized demand responsiveness (D4), learning organization (D1), specialized industrial clusters (C5), and inter-industry collaboration (C2) reside at the base of the hierarchy. The pivotal role of specialized industrial clusters (C5) as a foundational driver corroborates findings in maritime supply chain research, where geographic clustering has been shown to enhance knowledge spillovers and foster collective adaptability [46]. These factors exhibit broad reachability and comparatively fewer antecedents, indicating leverage for propagating improvements system-wide. Notably, transformative elements (D4, D1) underpin recovery elements (C5, C2), consistent with the DEMATEL cause–effect results.

Middle factors in the ISM model play a transmission role, which is influenced by foundational factors and also affects surface factors. Modular management teams (A2), knowledge stock (A4), risk culture (A5), logistics integration capability (B1), strategic supplier stability (B4), final-assembly orchestration (C1), social capital (C3), industrial autonomy (D2), information sharing (D3), and digital technologies (D5) occupy middle tiers. Notably, all four resilience capabilities are represented at this level. Restorative factors (C1, C3) and most predictive factors (A2, A4, A5) are close to the foundational level, underscoring their critical bridging role between foundational drivers and surface outcomes. Transformative and resistant factors, by contrast, are located nearer to the surface. Transformative factors demonstrate the cross-level distribution: D1 and D4 are at the foundational level as deep drivers, acting through predictive and restorative mechanisms to further enable D2, D3, and D5 at the middle tier, thereby forming a closed-loop pathway. Resistant factors (B1, B4) are concentrated near the surface, highlighting their role as proximate performance outcomes.

Surface factors represent the system’s ultimate performance outcomes. They are highly dependent, meaning their status is a direct consequence of the interactions and states of all underlying factors in the hierarchy. Schedule deviations (A1), intelligence management (A3), component and module manufacturing performance (B2), schedule discipline (B3), industrial backups (B5), and structural configuration (C4) are positioned toward the top. They are highly reachable from lower tiers and thus sensitive to changes in foundational and mediator factors. Resistant capability mainly appears here, reflecting its role as a proximate performance layer shaped by deeper governance and learning mechanisms.

4.3. Analysis of MICMAC Results

The primary motivation for comparing classifications from Figure 5 (ISM hierarchy) and Figure 6 and Figure 7 (MICMAC) was to enhance the validity of our structural interpretation and test the robustness of factor roles. The ISM model (Figure 5) depicts hierarchical levels based on binary reachability, while MICMAC classifies factors based on driving/dependence power. To achieve this, we performed MICMAC analysis using two input matrices: the binary reachability matrix M (resulting in Figure 6) and the weighted total relation matrix T (resulting in Figure 7). This dual-matrix approach serves to: (1) test whether factor classifications are consistent across different analytical representations (binary presence vs. weighted strength of influence), and (2) refine our understanding by revealing which factors have stable roles versus those whose classifications are sensitive to the measurement granularity.

The comparative results are presented below. Factors are classified into four clusters based on their driving and dependence power: Independent (high driving, low dependence), Dependent (low driving, high dependence), Autonomous (low driving, low dependence), and Linkage (high driving, high dependence).

Independent Cluster: Factors in this cluster exert a significant impact on others. Social capital (C3) is classified as an independent factor under the M-MICMAC, but falls into the autonomous cluster in the T-MICMAC. This indicates that while social capital appears to exhibit strong driving power and low dependence under binary M-MICMAC, its actual influence weakens once continuous weights are considered, lacking significant cross-node ripple effects. In contrast, learning organization (D1), customized demand responsiveness (D4), specialized industrial clusters (C5), final-assembly orchestration (C1), inter-industry collaboration (C2), knowledge stock (A4), and risk culture (A5) remain consistently categorized as independent factors in both analyses. These align with ISM base and mediator layers and DEMATEL cause factors, marking them as prime investment levers.

Dependent Cluster: Factors here are more influenced by others. Modular management teams (A2), intelligence management (A3), logistics integration capability (B1), strategic supplier stability (B4), industrial autonomy (D2), and information sharing (D3) are classified as dependent factors in the M-MICMAC but shift to the autonomous zone in the T-MICMAC. This shift indicates that while they exhibit strong dependency under binary thresholding, the actual strength of influence they receive from others is weak in the weighted MICMAC analysis. This suggests that these foundational elements are relatively stable, self-contained, whose autonomous characteristics are more accurately captured in the weighted network. In contrast, schedule deviations (A1), component and module manufacturing performance (B2), schedule discipline (B3) and industrial backups (B5) as dependent factors in both analyses. These variables are sensitive to upstream drivers and better monitored as performance outcomes or matured after foundational investments.

Autonomous Cluster: Factors here have limited systemic influence. Structural configuration (C4) and digital technology (D5) are classified as autonomous factors under both analyses. This suggests that these factors possess independent operational characteristics and can drive system resilience enhancement through internal optimization and autonomous innovation.

The linkage cluster does not show any scattered distribution under either analysis, indicating that no such influencing factor exists in the system.

5. Discussion and Implications

5.1. Discussion of Results

Our findings propose a hierarchical model of resilience capability formation for GCSSCs that posits that resilience emerges not from a flat list of factors, but through a structured, causal pathway: transformative and restorative capabilities act as the foundational engine for long-term adaptation and renewal. These foundational drivers empower a set of mediating transitional factors (encompassing governance, culture, and platforms), which in turn directly shape the proximal predictive and resistant outcomes that constitute day-to-day operational performance. This model moves beyond identifying what factors matter to explaining how and in what sequence resilience is systemically built, highlighting the path dependence and multi-level emergence central to CAS theory.

(1) Foundational Drivers: The Engine of Resilience

The analysis identifies four factors as the foundational drivers of resilience, positioned at the base of the ISM hierarchy. These include customized demand responsiveness, learning organization, specialized industrial clusters, and inter-industry collaboration. Consistent with DEMATEL, these show strong causal degree, and MICMAC places them in the independent quadrant. Their influence aligns with the GCSSC’s high customization, modular interdependence, and long qualification cycles. Customized demand responsiveness translates unique brand and passenger requirements into adaptive engineering and procurement, preventing late-stage design/fit-out shocks. Learning organization builds absorptive capacity and codified knowledge, enabling cross-chain integration and faster adaptation to unforeseen disruptions [47]. Specialized industrial clusters and inter-industry collaboration provide access to scarce skills, shared infrastructure, and cross-sector knowledge exchange, which is vital in a domain with concentrated upstream technologies [3,46]. From a managerial economics perspective, these factors represent strategic investment areas with high leverage effects for achieving both resilience and sustainability.

(2) Transitional Factors: The Transmission Mechanism

The transitional factors comprise modular management teams, knowledge stock, risk culture, logistics integration, strategic supplier stability, final-assembly orchestration, social capital, industrial autonomy, information sharing, and digital technologies. These factors transmit influence from deep drivers to proximal outcomes. Orchestration and governance (modular management teams; final-assembly orchestration) convert learning and collaboration into executable resequencing, prioritization, and assurance routines. Cultural and cognitive enablers (knowledge stock; risk culture) improve sensing and early escalation, raising predictive and recovery effectiveness. Platform enablers (information sharing; digital technologies) integrate PLM/ERP/MES and logistics visibility, amplifying sensing and coordination across nodes. These factors act as the crucial transmission belt that converts strategic investments in deep drivers into operational gains. Managers must ensure these mediating mechanisms are in place to realize the full economic value of their resilience investments.

(3) Proximal Outcomes: Key Performance Indicators

At the top of the hierarchy reside the direct influencing factors, which serve as key performance indicators of systemic resilience. These factors (e.g., schedule deviations, intelligence management, component and module manufacturing performance, schedule discipline, industrial backups, and structural configuration) are highly reachable from lower tiers and mostly fall in the dependent quadrant of MICMAC. They are observable performance levers. Resistant outcomes (component and module manufacturing performance, schedule discipline, industrial backups) improve when orchestration and cluster-enabled supplier networks. Predictive outcomes (schedule deviations, intelligence management) reflect enhanced sensing and planning fidelity following improvements in knowledge and digital platforms.

Via capability aggregation, transformative and restorative capabilities dominate the base and exhibit higher average centrality, indicating that structural renewal and coordination are core to GCSSC resilience. Predictive and resistant capabilities appear largely as proximal outcomes in our sample—highly consequential for performance but shaped by deeper governance, learning, and structural conditions.

The identification of customized demand responsiveness, learning organization, specialized clusters, and inter-industry collaboration as foundational drivers is theoretically meaningful because it aligns with CAS principles of co-evolution and emergent adaptation [19,21]. Unlike traditional resilience models that prioritize redundancy or agility in isolation, this configuration highlights how transformative and restorative capabilities serve as the engine for long-term resilience in project-based, high-value systems. This finding extends CAS theory to the context of GCSSCs by illustrating how system-level adaptation emerges from structured interactions between learning, collaboration, and localized clustering—a novel contribution to both resilience and sustainable operations literature.

5.2. Theoretical Implications

This study advances the GCSSC resilience theory in three ways. First, building on CAS and integrating engineering, ecological, and evolutionary resilience, we propose a four-capability framework mapped across disruption phases. The results show transformative and restorative capabilities as foundational layers that shape emergent performance in predictive and resistant layers—demonstrating path dependence and multilevel emergence central to CAS. Second, using fuzzy DEMATEL, we identify driver factors with high centrality—notably customized demand responsiveness, learning organization, specialized clusters, and inter-industry collaboration—then position them structurally via ISM and MICMAC, clarifying which factors are deep drivers, mediators, or proximal outcomes. Third, and most critically for decision-making, by linking factor roles to capability layers, we move beyond identifying what factors matter to explaining why and in what sequence they should be addressed. This provides a theoretically grounded resource allocation framework that helps managers prioritize investments under budget constraints to maximize the economic returns of resilience building.

5.3. Managerial Implications

Our findings offer an economically grounded rationale for tailoring resilience strategies to firm-specific resources and risk profiles. Firms must align their resilience-building pathway with their resource constraints and risk exposure to maximize cost-effectiveness.

For resource-constrained firms, a three-stage progressive strategy is recommended. Stage 1: investing in foundational drivers entails building learning mechanisms, codifying knowledge, fostering inter-industry collaboration, developing specialized clusters, and institutionalizing customized demand responsiveness through modular standards and configurable bill of materials. Stage 2: strengthening mediating capabilities requires leveraging knowledge stock, embedding risk culture, enhancing orchestration at the final-assembly level, and deploying integrated digital platforms to support information sharing, supplier stability, and selective localization. Stage 3: securing outcomes and monitoring performance involves improving logistics reliability, component quality, and schedule discipline, while establishing industrial backups where feasible. It also ensures the systematic monitoring of predictive indicators as early-warning signals and validation tools, thereby translating upstream investments into resilience outcomes. This phased approach allows managers to allocate scarce capital efficiently, building resilience in a way that minimizes upfront costs while establishing the necessary foundations for future capabilities.

For resource-sufficient firms, two paths are proposed: a value-creating strategy (transformative capability → restorative capability → predictive capability → resistant capability) to embed long-term advantage, and a risk-mitigating strategy (resistant capability → predictive capability → restorative capability → transformative capability) to secure operations under immediate pressure before upgrading upstream capabilities. This dual-path framework empowers senior management to make a conscious strategic trade-off between investing for long-term transformational advantage versus securing short-term operational continuity, based on their firm’s immediate risk landscape and strategic appetite.

6. Conclusions

This study developed a structured, hierarchical understanding of resilience capability formation in GCSSCs by addressing two research questions: (1) Which factors shape GCSSC resilience? (2) How do these factors interact and propagate to collectively influence overall supply chain resilience? Guided by Complex Adaptive Systems (CAS) theory, we integrated engineering, ecological, and evolutionary resilience perspectives into a four-capability framework and identified 20 GCSSC-specific resilience factors. An integrated fuzzy DEMATEL-ISM-MICMAC approach was applied to unravel their causal relationships, hierarchical structure, and driving-dependence dynamics.

The analysis reveals that customized demand responsiveness, learning organization, specialized industrial clusters, and inter-industry collaboration act as deep drivers with high causal power and centrality, forming the foundational set for GCSSC resilience. Knowledge stock, risk culture, and final-assembly orchestration function as cross-cutting mediators that transmit the influence of these drivers to proximal outcomes such as schedule deviations, component manufacturing performance, schedule discipline, and industrial backups—which serve as key performance indicators reflecting upstream resilience investments. At the capability level, transformative capability underpins restorative capability; restorative capability enhances predictive capability through improved coordination and knowledge sharing; and resistant capability manifests as the performance layer sensitive to upstream improvements.

Beyond the firm level strategies, our findings suggest that public investment should prioritize fostering specialized industrial clusters and incentivizing inter-industry collaboration, as these foundational drivers create positive externalities and a fertile ecosystem that enhance the collective resilience and international competitiveness, especially for emerging economies entering high-value maritime markets.

This study has three limitations that suggest future research directions: (1) the expert-dependent methodology invites large-scale quantitative validation (e.g., structural equation modeling); (2) the static analysis could be extended with dynamic simulation (e.g., system dynamics) to explore resilience evolution over time; (3) the GCSSC-specific focus calls for comparative studies across industries, potentially using methods like fsQCA to identify equifinal resilience configurations. Ultimately, building resilient GCSSCs is not only a tactical necessity but a strategic imperative for achieving sustainable industrial development in an era of uncertainty. This study provides a structured, evidence-based framework for enhancing resilience through targeted capability investment, paving the way toward more robust and sustainable cruise shipbuilding ecosystems.