Introduction

Currently, over 80% of global goods are transported by ships. However, in the past few years, the frequent occurrence of major international emergencies has brought great challenges to the security of the MSC1. In February 2020, the outbreak of COVID-19 led to 8.5% of global container congestion in ports, triggering a surge in shipping prices2. In March 2021, the Suez Canal was blocked for 6 days, and more than 1000 cargo ships were routed around the Cape of Good Hope in Africa3. In February 2022, the Russia-Ukraine conflict resulted in the sudden closure of Ukrainian ports, affecting the global security of energy and food supply chains. In December 2023, the Red Sea crisis caused about 82% of container ships on the Suez Canal route to be diverted, greatly increasing the cost of shipping from Asia to Europe. Given the increasing complexity of the global trade landscape, the resilience of the MSC, as a key factor in global supply chain security, is becoming increasingly prominent4.

To enhance the resilience of MSC, we need to address the following issue: “How can maritime-related enterprises use their resources to build a resilient MSC?” .When discussing the resilience of MSC, most existing scholars rely on statistical techniques that are based on the assumption of independent variables, unidirectional linear relationships and symmetrical causality5. These techniques analyze the marginal “net effect” of independent variables on dependent variables while controlling for other factors, and thus fail to explain the complex causal relationships, such as the interdependencies among independent variables6,7. In reality, the resilience of maritime supply chains is not influenced by a single factor alone, it is likely subject to the influence of multiple factors, with different elements interdependent and interconnected8. Additionally, it is important to recognize that different combinations of these elements can lead to the same outcomes, hence the pathways to achieve high resilience in maritime supply chains are not singular9.

The configurational perspective and qualitative comparative analysis (QCA) adopt a holistic analytical approach, treating the research object as a configuration of different combinations of conditional variables10. This approach integrates the strengths of case study and variable research methods, and through set analysis, it uncovers the relationship between element configurations and outcomes11. It is instrumental in addressing complex causal issues such as multiple concurrent causal relationships, causal asymmetry, and the equivalence of various solutions12. This perspective represents a new path in management research. Currently, configurational perspective and QCA have been widely applied across various fields of management, including strategic management, business model innovation, corporate governance, human resources, international business, and digital transformation10,11,12,13,14. However, its application in the resilience of MSChas not yet been explored.

To address the limitations of existing research, this paper proposes an analytical framework for enhancing the resilience of MSC based on the configurational perspective and QCA. The framework integrates SEM, Necessary NCA, and fsQCA to systematically investigate the pathways to improving MSC resilience. The reasons for employing this analytical framework are as follows: Firtsly, SEM is based on a structured model that incorporates a large number of items and constructs, which can be used to examine the impact of individual factors on MSC resilience15,16. However, this analytical approach is limited to examining the average influence of exogenous constructs on endogenous constructs17. Sencdly, fsQCA is a type of QCA method, which has the advantage of being able to handle fuzzy data18. This method is particularly useful for analyzing how different configurations of conditions generate specific outcomes for different groups of cases19,20. However, the fsQCA method can only qualitatively determine whether the antecedent conditions are necessary for the outcome21.Thirdly, NCA is a method that helps researchers identify the necessity of individual conditions, assess their effect size and statistical significance, and determine to what extent a predictor variable acts as a bottleneck to the outcome variable22,23. NCA can complement and expand methods such as SEM and fsQCA22,23.In this way, our study proposes a new analytical framework to verify the ‘causal complexity’ of MSC resilience and enriches the application of quantitative research in this area. Specifically, this paper reveals the multiple paths to achieve high resilience of MSC, which can help MSC members to choose the appropriate resilience management strategy based on their existing resources and situations.

The remaining part of this paper is organized as follows: Sect. 2 is literature reviews. Section 3 provides theoretical background and hypothesis development. Section 4 outlines the methodological framework of the expanded analytical process. Section 5 conducts empirical analysis. Section 6 presents the conclusion.

Literature reviews

Maritime supply chains

MSC is a network structure composed of upstream and downstream enterprises or departments providing maritime services to the ultimate users5. Fernando et al.6defined the maritime supply chain as a linked structure that integrates essential logistics functions such as transportation, storage, and information processing, in accordance with the physical flow of goods from their place of origin to destination. This process is collaboratively completed by related enterprises in the maritime, port, and terrestrial sectors, with shipping companies at its core7. MSC includes maritime equipment suppliers (shipbuilding companies and fuel suppliers etc.), maritime service suppliers (shipping companies or agents), maritime demand parties (shippers or their agents), as well as ports and customs8. The network structure of MSC is shown in Fig. 1.

Fig. 1
figure 1

Maritime supply chain network structure.

Full size image

Maritime supply chains resilience

The concept of supply chain resilience means the ability of a supply chain to respond to unforeseen disruptions and return to normal operation24,25. Ribeiro et al.26believed that a resilient supply chain should be capable of preparing for, responding to, and recovering from disruptions, while maintaining positive steady-state operations within acceptable costs and time frames. Rodrigue27pointed out that the resilience of a transport system is its capability to resume operations at a level similar to that before a disruption occurred. As the MSC has been exposed to risks and disasters due to its complexity, vulnerability, and uncertainty, the resilience-building is a critical approach to mitigating risks and keeping the competitiveness28. The definition of MSC resilience can be defined as the capacity of the MSC to promptly respond to the risks of international maritime disruptions, and to maintain or restore stable operations to its initial state4,5,9.

Paths to enhancing the resilience of MSC

Scholars have made preliminary explorations on how to improve the resilience of MSC. Lam and Bai5 developed an original quality function deployment approach to enhance MSC resilience. They found that contingency plans, monitoring and maintenance, and supply chain relationship management are the top three measures that enhance resilience. Gu et al.29 proposed an integrated method that combines the Best-Worst Method and fuzzy TODIM (Interactive and Multi-criteria Decision-Making, the acronym in Portuguese) approach to prioritize strategies for MSC resilience. Their findings demonstrated that the accelerated container turnover, human resource backup, and long-term shipping contracts emerge as paramount strategies for effectively improving resilience. Zavitsas et al.30 established a link between environmental and network resilience performance for MSC and provides a methodological framework to analyze the impact of various abatement options, regulatory and pricing strategies. Liu et al.9 concluded that robustness and adaptability are the two most important resilience goals, and the resilience strategy of collaboration and flexibility should be paid more attention in the post COVID-19 era. Liu et al.28 pointed out that “Government support”, “Resource sharing”, “Emerging technology”, “Maritime infrastructure”, and “Redundancy plan” are significant enablers in enhancing MSC resilience during pandemic, in which the most influential enabler is the redundancy plan.

The configurational perspective and qualitative comparative analysis

The configurational perspective and the QCA method, as an emerging research paradigm, are well-suited for analyzing the causal complexity of concurrent multi-factor issues and have been widely applied in management research10,11,12,13,14. Zheng et al.10 employed fsQCA to identify alternative configurations leading to high innovation performance from the perspective of knowledge management and innovation ecosystems. Alnoor et al.11 utilized a three-stage SEM-fsQCA-MCDM to identify the optimal configuration of marketing influencers on impulsive buying behavior. Linyu et al.12 employed the the panel data-based qualitative comparison method to analyze the driving factors that affect the green development of energy systems in various countries and the configuration effect on the vertical axis of time and its synergistic promotion path.

Hypothesis development

Current research on pathways to enhancing the resilience of supply chains most frequently adopts the Dynamic Capabilities Theory (DC)31,32,33,34,35,36and the Organizational Information Processing Theory (OIPT)37,38,39,40. The DC emphasizes the dynamic attributes of resource capabilities31,32. ‘Dynamic’ refers to the ability to continuously adjust and innovate in response to changes in the external environment, while ‘capabilities’ include the ability to build, integrate, and restructure for reuse33,34. Supply chain resilience precisely addresses the issue of supply chain disruption caused by external shocks in a dynamic and unpredictable environment35,36. According to the DC, combined with the increasingly complex international trade situation and decarbonization requirements faced by the maritime supply chain, this paper believes that agility, cooperation and carbon emission control are the key factors to improve the resiliences of the MSC.

The OIPT is a theoretical framework about how organizations process and use information37. The OIPT theory emphasizes that in the information age, organizations need to process and use information efficiently to adapt to the rapidly changing environment38. The core of OIPT theory is to understand how an organization optimizes its decisions and actions through information processing39. Digitalization can improve the transparency and circulation speed of upstream and downstream information in the MSC, help maritime related enterprises better cope with challenges in a complex and uncertain environment, and thus improve the resilience of the MSC40. Thus, according to OIPT, this paper believes that early warning ability and digitalization level are the key factors to improve the MSC resiliences .

Therefore, based on the theory of DC and OIPT, this paper posits hypotheses regarding the impact on MSC resiliences from five perspectives: early warning capability, agility, collaboration, digitalization level, and carbon emission control.

Early warning capability and MSC

Early warning capability refers to the ability to predict unexpected events and prepare in advance41. Alliger et al.42 believed that early warning capability can help organizations develop plans before the risk event. Mittal et al.43 viewed early warning capability as a market sensing ability, which reflects a company’s ability to perceive market opportunities and threats. Kanamoto et al.44pointed out that early warning capability can be reflected in two aspects: risk awareness and visibility. Risk awareness helps enterprises quickly identify maritime-related risks and take appropriate measures, thus minimizing the risk of interruption45. The visibility of the MSC will enhance its warning capability46. Therefore, the following hypothesis is proposed:

H1. Early warning capability positively affects MSC resilience.

Agility and MSC

MSC face risk shocks because they cannot respond timely to market demand changes or supply interruptions47. Agility enables enterprises to act more quickly, select appropriate plans and strategies to reduce the impact of interruptions, and thus achieve faster recovery48. Moreover, agility can compress response time, reduce interruption costs, and maintain the shipping industry’s competitive advantage, thereby improving the resilience of MSC38. The complexity of the MSC and the dynamic adjustment capabilities of the core enterprises affect the agility of the MSC43. Therefore, we propose the following hypothesis:

H2. Agility positively affects MSC resilience.

Collaboration and MSC

Collaboration refers to the ability to effectively collaborate with partners in the supply chain to quickly respond to disruptions49. Simatupang et al.50 believed that collaboration is “the glue that holds supply chain partners together in a crisis”. Wu et al.51 indicated that strengthening collaboration among supply chain members can enhance the response ability to supply chain disruptions, and thus improve supply chain resilience. Lai et al.52proved that information sharing will improve the collaboration of the MSC, thus mitigating the “bullwhip effect” of MSC. The information sharing, collaboration mechanisms, profit, and risk-sharing mechanisms impact the Collaboration of the MSC53. Based on these, we propose the following hypothesis:

H3. Collaboration positively affects MSC resilience.

Digitalization level and MSC

Bahrami and Shokouhyar39proved that supply chain resilience requires more support from data science than ever before. Firstly, digitalization can improve the organization’s capability to prepare for supply chain-related emergencies by gathering internal and external information54. Secondly, when a disruptive event occurs, digitalization can enable companies to collect a wealth of information from various sources throughout the supply chain to support quick decision-making and timely risk response55. Thirdly, prior experience, information, and knowledge of dealing with disruptive events contribute to the improvement of supply chain resilience56. Therefore, the following hypothesis is proposed:

H4. Digitalization level positively affects the resilience of MSC.

Carbon reduction and MSC

In 2023, the carbon emissions of the shipping industry exceeded 1 billion tons, accounting for 2–3% of the global total, and the shipping industry is undergoing a green and low-carbon transformation42. In 2018, the International Maritime Organization (IMO) established carbon emission reduction targets for the shipping industry: Compared to 2008, the total greenhouse gas emissions from international shipping should be reduced by at least 50% by 205062. On November 1, 2022, the International Convention for the Prevention of Pollution from Ships entered into force, compelling all ships to calculate their existing ship energy efficiency index58. On April 25, 2023, the European Council adopted five legislative proposals to address climate change, with shipping emissions being included in the European Emissions Trading System for the first time59. Therefore, the level of carbon emission control will become a key factor for the future MSC resilience60. Hence, we propose that:

H5. Carbon reduction positively affects MSC resilience.

The combination effect of these factors on MSC

As the MSC is an open complex system with numerous enterprises8, and the resilience of MSC is not necessarily affected by a single factor, having only a single capability is difficult to achieve high MSC resilience9. Based on configurational perspective10,11,12,13,14, this paper proposes hypothesis 6.

H6. MSC resilience is the result of the combination effect of multiple factors.

Integrating the above assumptions, this article proposes a conceptual model of factors influencing the resilience of MSC, as shown in Fig. 2:

Fig. 2
figure 2

Conceptual model of factors impacting the resilience of maritime supply chains.

Full size image

Methodology

Figure 3 presents the methodological framework of the study22. In stage 2, the SEM method is utilized to examine the impact of individual factors on MSC resilience. In stage 3, the NCA and fsQCA methods are employed to investigate the synergistic effects of multiple conditions on MSC resilience. All methods employed in this study were carried out in accordance with relevant guidelines and regulations. All experimental protocols in this sduty were approved by College of Navigation, Jimei University, and informed consent has been obtained from all participants in the questionnaire survey or their legal guardians.

Fig. 3
figure 3

Methodological framework.

Full size image

Structural equation model

The SEM includes four parts: the measurement model, the structural model and the model assumptions. In the measurement model, we can measure the internal structure of latent variables through manifest variables16. In this study, a total of six potential variables are considered, including early warning capability, agility, Collaboration, digitalization level, carbon reduction, and the resilience of MSC. The first five variables serve as exogenous variables, while the resilience of MSC is the endogenous variable. Because all six variable data in this study are obtained through a questionnaire survey, both the exogenous and endogenous variables are latent. The model is shown as follows17:

The measurement model:

$$\:y={\varLambda}_{y}\eta\:+\varepsilon$$
(1)
$$\:x={\varLambda}_{x}\xi\:+\delta\:$$
(2)

The structural model:

$$\:\eta\:=B\eta\:+\varGamma\:\xi\:+\zeta\:$$
(3)

Equation (1) is the measurement equation of the endogenous variable, y is a p×1 vector composed of p endogenous indicators, η is a m×1 vector composed of m endogenous latent variables (factors), \(\:{\varLambda\:}_{y}\) is the p×m factor loading matrix of y on η, and ε is a p×1 vector composed of p measurement errors. Equation (2) is the measurement equation of the exogenous variable, x is a q×1 vector composed of q exogenous indicators, ξ is a n×1 vector composed of n exogenous latent variables (factors),\(\:\:\:{\varLambda\:}_{x}\) is the q×n factor loading matrix of x on ξ, and δ is a q×1 vector composed of q measurement errors. In the structural Eq. (3), B is an m×m coefficient matrix that describes the influence between endogenous latent variables η; Γ is an m×n coefficient matrix that describes the influence of exogenous latent variables ξ on endogenous latent variables η; and ζ is a m×1 residual vector.

Based on the measurement model and the structural model, the covariance matrixes are shown as follows.

$$\:\sum\:_{xx}\left(\theta\:\right)={\varLambda\:}_{x}\Phi\:{\varLambda\:}_{x}^{{\prime\:}}+{\varTheta\:}_{\delta\:}$$
(4)
$$\mathop \sum \limits_{{yy}} \left( \theta \right) = \Lambda _{y} \overset{\lower0.5em\hbox{$\smash{\scriptscriptstyle\smile}$}}{B} \left( {\Gamma \Phi \Gamma ^{\prime} + \Psi } \right)\overset{\lower0.5em\hbox{$\smash{\scriptscriptstyle\smile}$}}{B} ^{'} \Lambda _{y} ^{'} + \Theta _{\varepsilon }$$
(5)
$$\:\sum\:_{yx}\left(\theta\:\right)={\varLambda\:}_{y}\stackrel{\sim}{B}\varGamma\:\Phi\:{\varLambda\:}_{x}^{{\prime\:}}$$
(6)

Equation (4) is the covariance matrix of x, Φ is the covariance matrix of the latent variable ξ, \(\:{\varTheta\:}_{\delta\:}\) is the covariance matrix of δ; Eq. (5) is the covariance matrix of y, Ψ is the covariance matrix of the residual term ζ, \(\:{\varTheta}_{\varepsilon}\) is the covariance matrix of ε; Eq. (6) is the covariance matrix of y and x; Eq. (7) is the covariance matrix of the (p + q)×1 vector composed of all indicators (\(\:{y}^{{\prime\:}}\),\(\:{\:x}^{{\prime\:}}\)).

$$\sum\:\left(\theta\:\right)=\left(\begin{array}{cc}{\varLambda\:}_{y}\overset{\lower0.5em\hbox{$\smash{\scriptscriptstyle\smile}$}}{B}(\varGamma\:\Phi\:{\varGamma\:}^{{\prime\:}}+\varPsi\:)\overset{\lower0.5em\hbox{$\smash{\scriptscriptstyle\smile}$}}{B}^{{\prime\:}}{{\varLambda\:}_{y}}^{{\prime\:}}+{\varTheta\:}_{\epsilon\:}&\:{\varLambda\:}_{y}\stackrel{\sim}{B}\varGamma\:\Phi\:{\varLambda\:}_{x}^{{\prime\:}}\\\:{\varLambda\:}_{x}\stackrel{\sim}{B}\varGamma\:\Phi\:{\varLambda\:}_{y}^{{\prime\:}}&\:{\varLambda\:}_{x}\Phi\:{\varLambda\:}_{x}^{{\prime\:}}+{\varTheta\:}_{\delta\:}\end{array}\right)$$
(7)

The model assumptions:

(1) The mean of the error terms ε and δ in the measurement equation is zero;

(2) The mean of the residual term ζ in the structural equation is zero;

(3) The error terms ε and δ are uncorrelated with the factors η and ξ, and ε and δ are uncorrelated;

(4) The residual term ζ is uncorrelated with ξ, ε, and δ.

Fuzzy set qualitative comparative analysis and necessary condition analysis

Sociologist Ragin (1987) introduced the Qualitative Comparative Analysis (QCA) method in 1987. This method employs set theory and Boolean algebra operations to identify necessary and sufficient conditions that affect outcomes, thereby determining which factors are crucial for the occurrence of results and further understanding the interactions among factors18. fsQCA is a type of QCA method, which can effectively analyze the interaction of more than 3 variables and is suitable for addressing configuration issues19,20. However, the fsQCA method can only qualitatively determine whether the antecedent conditions are necessary for the outcome22. The NCA method can measure the antecedent conditions that are necessary for the result at “a specific level”23. The integration of fsQCA with NCA exhibits a higher degree of scientific rigor22.

Empirical analysis

Questionnaire survey

Questionnaire design

This paper uses survey data as samples to examine the influence factors and their combination mechanisms on the resilience of the MSC. The questionnaire items use the Likert five-level scale measurement, where “1→5” respectively represents “completely disagree, a bit disagree, unclear, agree, completely agree”21. The specific measurement indicators and sources of each variable are shown in Table 1.

Table 1 Measurement items of influence factors on maritime supply chain resilience.
Full size table

Data collection

Beginning in May 2023 and ending in November 2023, with the help of the China Shipowners’ Association, this study distributed questionnaires to over a hundred MSC-related companies in China. The individuals who completed the questionnaire were experienced maritime industry staff and managers. We applied multiple questionnaire survey methods such as face-to-face interviews (distributing 20 copies), questionnaires online (distributing 150 copies), and email (distributing 75 copies), and ultimately 151 valid questionnaires were obtained, with a valid questionnaire recovery rate of 61.6%.

Sample descriptive analysis

We can find in Table 2 that the survey respondents cover various types of maritime-related companies in China, and the questionnaire compilers include senior, middle, and grassroots staff, which ensures the reliability of the questionnaire data.

Table 2 Descriptive analysis of samples.
Full size table

Single-factor impact analysis based on the SEM

Reliability and validity analysis

The reliability of the scale data is generally assessed by two indicators: the internal consistency coefficient (Cronbach’s α) and the composite reliability. The SPSS 26.0 software (Statistical Package for the Social Sciences 26.0, https://spss.mairuan.com/xiazai.html) is used for reliability analysis, and the results are shown in Table 3. The Cronbach’s α and composite reliability value of each variable all exceed 0.8, indicating that the scale used in this study has good reliability54.

Table 3 Reliability and validity analysis of scale. Note: diagonal values representing the average variance extracted (AVE) for each variable; the values between different variables represent the Pearson correlation coefficients; *** indicates p < 0.01.
Full size table

The validity analysis of the scale includes content validity, structural validity, convergent validity, and discriminant validity. The scales used in this study refer to internationally published and validated mature scales in international journals, which to a certain extent ensure the content validity of this study17. The results of the confirmatory factor analysis show that χ2/df = 2.476 < 3, RMSEA = 0.089 < 0.1, RMR = 0.042 < 0.05; GFI = 0.936, NFI = 0.949, CFI = 0.923, IFI = 0.924, all greater than 0.9, indicating the scale has good structural validity16. The factor loadings of all items are greater than 0.5, the average ex-traction variance (AVE) of each factor is greater than 0.5, and the composite reliability is greater than 0.8 for each factor, demonstrating that the scale has good convergent validity17. The square root of the AVE for each variable is greater than the correlation coefficient between that variable and other variables, testifying that the scale has good discriminant validity.

Common method bias test

In this study, the Harman single factor method is employed to test for common method bias in the questionnaire survey conducted17. From Table 4, it can be observed that there are five factors with eigenvalues greater than one, exceeding the single factor. Moreover, the variance explained by the largest factor is 38.351%, which is below 40%. Therefore, we can conclude that there is no severe common method bias present in the current study.

Table 4 Harman’s single factor test.
Full size table

Hypothesis test

Based on the good model fitting, this study conducted a SEM test using the SPSS 26.0 software. The hypothesis testing results are presented in Table 5, and the single factor impact analysis diagram is shown in Fig. 4.

Table 5 Results of hypothesis testing. Note: *** indicates p < 0.01, ** indicates p < 0.05, and * indicates p < 0.1.
Full size table
Fig. 4
figure 4

The single factor impact analysis diagram based on SEM.

Full size image

From Fig. 4; Table 5, it can be seen that the early warning capability (β = 0.43, p < 0.05), agility (β = 0.337, p < 0.1), collaboration (β = 0.46, p < 0.01), digital level (β = 0.372, p < 0.01), carbon reduction (β = 0.464, p < 0.01), all have a significant positive impact on the resilience of the maritime supply chain. Hypotheses H1-H5 are supported by the SEM.

Configuration analysis based on NCA and fsQCA

Variable selection and date calibration

Based on the previous theoretical analysis and empirical research results, this study selects early warning capability, agility, collaboration, digital level, and carbon emission control as conditional variables, and MSC resilience as the result variable. The fsQCA3.0 software (Fuzzy-Set Qualitative Comparative Analysis 3.0, https://sites.socsci.uci.edu/~cragin/fsQCA/software.shtml) is used to calibrate each variable data into fuzzy set membership degrees between 0 and 1. Firstly, the mean value of each variable corresponding to the item is calculated as the reflective value of the variable. Then, we set three calibration points for the 5-point Likert scale used in the questionnaire, where “5” is the fully belonged calibration point, “3” is the intersection point, and “1” is the fully not belong calibration point. Since the data system with a membership of 0.500 will not be included in the truth table analysis, a calculation of + 0.001 is performed18.

Necessary condition analysis

We conduct necessary condition analysis through the upper bound regression (UR) and upper bound envelope analysis (UE) methods, and the following conditions need to be met to determine the antecedent conditions as necessary conditions: the effect size (d) ≥ 0.1, and the Monte Carlo simulation permutation test (P Value) < 0.0162. Table 6 NCA necessary condition analysis and Fig. 5 NCA scatter plot were calculated using R4.3.1 software (R Project for Statistical Computing 4.3.1, https://www.r-project.org/). We can observe that: the effect size of collaboration (SY), calculated using both the UR and UE methods, is greater than 0.1, and the P value is significant. This suggests that SY is a necessary condition for MSC resilience. The effect size of the digitalization level (DL) calculated by the UR method, is greater than 0.1 and has a significant P value, indicating that DL is a necessary condition for the resilience of MSC to some degree. The same method was applied to calculate other variables, but they do not qualify as necessary conditions.

Table 6 Necessary condition analysis. Note: the effect size d ranges from 0 to 0.1, indicating a ‘low level’; d ranges from 0.1 to 0.3, indicating a ‘medium level’; and d is above 0.3, indicating a ‘high level’. The P-value represents the permutation test in NCA, with a P-value closer to 0 indicating a more significant impact. 
Full size table
Fig. 5
figure 5

Scatter plot from NCA analysis. (a) NCA polt: EW-MSCR (b) NCA polt: AG-MSCR. (c) NCA polt: CO-MSCR (d) NCA polt: DL-MSCR. (e) NCA polt: CR-MSCR.

Full size image

The NCA can not only determine whether a variable is a necessary condition but also analyze the effectiveness of the necessary conditions22. The effectiveness is also known as the bottleneck level, which refers to the minimum level value (%) that each antecedent condition needs to meet within its observed range. The results of the bottleneck level analysis are shown in Table 7. Based on the results, to achieve a 100% resilient MSC, the level of early warning capability, agility, collaboration, digitalization level, and carbon emissions control, need to reach 50%, 42.2%, 73%, 48.7%, and 46.9% respectively. In the fsQCA approach, antecedent conditions with a consistency level exceeding 0.9 are considered necessary conditions19. The fsQCA 3.0 software is used for the necessary condition testing. Table 8 reveals that, except for the consistency of collaboration and digitalization level surpassing 0.9, all other causal conditions do not qualify as necessary conditions for the resilience of MSC. The analysis results are consistent with the NCA method. In NCA, the necessary condition refers to the lower level of the antecedent condition. That is to say, the lower level of digitalization and collaboration needs to be paid attention in the process of improving the resilience of MSC. In fsQCA, the necessary condition refers to the achievement of a certain level of membership. In this article, we analyze the necessary conditions for achieving high resilience in MSC. Therefore, this study believes that the five conditions do not exist independently to produce high MSC resilience, and we need to conduct a multi-factor configuration analysis.

Table 7 Bottleneck level analysis (%) based on NCA .Note: using CR analysis, NN represents non-essential. 
Full size table
Table 8 Necessary condition analysis based on fSQCA.
Full size table

Configuration sufficiency analysis

The fsQCA software outputs three types of solutions: complex solutions, simple solutions, and intermediate solutions. Intermediate solutions are used for configuration analysis, supplemented by simple solutions to distinguish core and auxiliary conditions18. Core conditions are those that appear simultaneously in the intermediate and simple solutions. Auxiliary conditions are those that appear only in the intermediate solutions23. In conducting configuration analysis, this study sets the consistency threshold value at 0.8, the PRI value at 0.7, and the frequency threshold value at 10, retaining more than 80% of the cases23. We obtain four types of causal configurations that trigger high MSC resilience (In Table 9). Therefore, hypothesis H6 “The resilience of MSC is the result of a combination effect of multiple factors.” is verified.

Table 9 The results of configuration sufficiency analysis. Note: ● represent the presence of core conditions, ⊗ represent the absence of core conditions, ● represent the presence of auxiliary conditions ⊗ represent the absence of auxiliary conditions, and spaces represent conditions that can be present or absent.
Full size table

(1) The M1 path ‘collaboration + carbon reduction + digitization’. In this path, the core conditions are collaboration and carbon emissions control, with digital proficiency as an auxiliary condition. The consistency of this configuration is 0.9531, which can explain 83.1% of the sample enterprises. Configuration M1 indicates that digitalization can improve the collaboration efficiency among different members in the MSC, alleviate the information asymmetry between cross-border departments, reduce communication and human resource costs, and simultaneously enhance the risk perception ability in the international market, ultimately improve the resilience of the MSC. Besides, “carbon emissions” and “carbon capture” based on digitalization will directly reduce the absolute carbon emissions intensity of shipping industries, which is conducive to the green development of the MSC.

(2) The M2 path ‘early warning + collaboration’. This path includes two configuration conditions, M2a and M2b. Configuration M2a is a path with early warning capability and collaboration as core driving factors, and digitalization level as auxiliary factors. The consistency of M2a reaches 0.9588, which can explain 79.3% of the sample enterprises. Configuration M2b is a path with early warning capability, collaboration, and carbon emissions control as core conditions. The consistency of M2b is 0.9694, which can explain 77.3% of the sample cases. Configuration M2 demonstrates that it is necessary to establish an early warning mechanism alliance with the main chain enterprises (shipping companies or ports) taking the lead and supporting enterprises (import and export companies or freight forwarding agencies) playing a secondary role. Meanwhile, it is essential to build a dynamic database for MSC, which can comprehensively, dynamically, and timely reflect the development status and trends of international maritime transport. Lastly, this mechanism should promote the smooth transmission of risk information in MSC and enhance the monitoring capacity for key risks in maritime transportation.

(3) The M3 path ‘Agility + Early Warning’. This path focuses on agility and early warning capability as the core condition, with non-carbon emissions control as an auxiliary condition. The consistency of M3 is 0.9623, which can explain 25.3% of the sample cases. Configuration M3 emphasizes the importance of agility in the rssilience of MSC. The agility of MSC includes three aspects: digital agility, physical agility, and process agility. Digital agility refers to the ability of enterprises in MSC to use information flow to improve and accelerate decision-making. Physical agility refers to the ability to quickly adjust physical assets to minimize risks, such as maritime transportation capacity, routes, personnel, and equipment. Process agility refers to the ability of companies in MSC to quickly adjust and restructure business processes to adapt to changes in the international maritime market.

(4) The M4 path ‘Carbon Reduction’. This path takes carbon emission control as the core condition, and non-early warning capability and non-agility as auxiliary conditions. The consistency of this configuration reaches 0.9447, which can explain 24.5% of the sample companies. Configuration M4 manifests efforts should be made to develop new environmentally friendly vessels such as electric ships, fuel cell ships, and LNG ships, as well as low-carbon fuels such as methanol, ammonia, and hydrogen. On the other hand, energy-saving retrofits and carbon capture systems should be conducted on existing vessels to improve energy efficiency and reduce emissions. Furthermore, smart ship technologies should be introduced to reduce travel time and enhance shipping efficiency, thereby reducing carbon emissions.

In summary, Fig. 6 shows the conditional configurations affecting the high resilience of the MSC.

Fig. 6
figure 6

Conditional configurations affecting the high resilience of the maritime supply chain.

Full size image

Robustness analysis

This article increases the frequency threshold and original consistency threshold to conduct robustness analysis54,58. First, the frequency threshold is increased from 10 to 14. The resulting configurations M1 and M2b, are subsets of the original configuration. Second, the original consistency threshold is increased from 0.80 to 0.85. The results show that the consistency of the overall solution has improved, while the coverage of the overall solution has decreased, and all indicators have experienced only minor changes. Hence, the research conclusions of this paper are robust.

Conclusions

In the past few years, the increasing uncertainty and complexity of the international environment have led to an increased risk of disruption in global supply chains. This paper’s review has shown that there is no unified or absolute definition of resilience in MSC, and different approaches to understanding it are taken depending upon the focus and framework of reference adopted. This study designs and applies a new methodological framework based on the SEM-NCA-fsQCA model and uses the questionnaire survey data of the MSC enterprises in China to empirically analyze the key factors and different paths to achieve high resilience in MSC.

Theoretical implications

The findings reported in this paper demonstrate that the developed SEM-NAC-fsQCA methodology framework can be successfully and systematically applied to identify and compute the driving factors to achieve high resilience in MSC. Contrary to previous research, our study put forwards a new analytical framework to validate the ‘causal complexity’ of MSC resilience and enriches the application of configurational perspective and QCA in this area.

Firstly, few existing literatures consider carbon emission control as an influencing factor of MSC resilience. Our empirical research result of the SEM shows that the carbon emissions control has the largest and most significant impact on the resilience of the MSC, which is more aligned with the evolving demands of international shipping carbon emissions.

Secondly, through the dual necessity verification of the NCA method and fsQCA method, it was found that the five conditions do not exist independently to produce high resilience in MSC. Configuration M1 (collaboration + carbon reduction + digitization) has the highest original and unique coverage, which can be considered the core configuration. Carbon emissions control, early warning ability, and Collaboration are all core conditions in three of the paths, indicating that these three elements are key factors in improving the resilience of MSC.

Thirdly, the level of digitization appears the most times in the auxiliary conditions, which indicates that digitization can play a catalytic role in enhancing the resilience of the maritime supply chain by helping other preconditions. The role of the same element in different paths may be opposite, indicating that antecedent elements will produce different effects depending on the changes in companies or environments.

Managerial implications

The presence of synergistic effects among multiple conditions reflects the complexity of managing the resilience of the maritime supply chain. With these findings, we should be able to better comprehend the resilience enhancement mechanism of MSC, which will be useful not only for academic and practitioner communities but also for the policymakers in describing and evaluating the impacts of such policies in a more systematic fashion. Different types of maritime-related companies and departments can take targeted measures to improve the resilience of the maritime supply chain based on their existing factor endowment.

Firstly, the M1 path ‘collaboration + carbon reduction + digitization’ emphasizes that to improve the resilience of the MSC, we need to apply digital technology to achieve interconnection and shared interests among all parties in the MSC, and jointly respond to problems in any link or process, including solving the problem of maritime decarbonization.

Secondly, The M2 path ‘early warning + collaboration’ encourage organizations in MSC to use digital technologies such as big data, knowledge graphs, artificial intelligence, 5G, and blockchain to build a dynamic early warning mechanism alliance. The M3 path ‘Agility + Early Warning’ emphasizes that the ability to take rapid action after discovering significant risk events through the early warning system is crucial.

Thirdly, the M4 path ‘Carbon Reduction’ indicates that with the IMO’s ‘2023 Ship Greenhouse Gas Emission Reduction Strategy’ released and the implementation of the EU carbon tax policy, the international shipping industry is facing increasing pressure to decarbonize. In the future, enterprises in the MSC need to continuously invest and cooperate in Carbon emission reduction to ensure resilience.

Limitations and propose directions for future research

This study still has several limitations which provide avenues for future research. First, this study collects data through questionnaire surveys, which may be subject to some degree of subjectivity. In the future, further exploration can be conducted on how to use actual operational data of enterprises related to maritime transportation for empirical research. Second, this study doesn’t analyze the differences in resilience management strategies among enterprises of the maritime supply chain. Further research is needed to explore the heterogeneity of resilience management among different enterprises. Third, this study only analyzes data from a specific period of the maritime supply chain, without including cross-temporal case data. Future research can select longer periods and use methods such as linear growth QCA and multi-temporal multi-linear growth QCA to track complex dynamic changes.