Literature review on digital empowerment supply chain resilience

With the rapid advancement of digital technologies such as the Internet of Things, Big Data, and 5G, the global economy has undergone a digital transformation, and traditional supply chains have transitioned into the digital era. Supply chain resilience, as an emerging field in supply chain management, has garnered increasing attention. The term "resilience" has long been a subject of study in disciplines such as psychology and ecosystems. However, a unified explanation or definition of supply chain resilience has not yet been established. Christopher and Peck were among the first to define supply chain resilience as the ability of a supply chain system to restore to its initial state after an operational disruption, encompassing flexibility, adaptability, and agility [14]. The most comprehensive definition of supply chain resilience comes from Ponomarov and Holcomb, who define it as the ability of a supply chain to adapt to sudden unforeseen events and disruptions, maintain structural and functional stability, and recover from shocks [15]. They further categorize supply chain resilience into four dimensions: readiness, response, recovery, and growth. Specifically, building supply chain resilience requires capacity building in pre-crisis risk prevention, rapid response to disruptions, and quick recovery and growth after risks materialize [16].

From an enterprise perspective, supply chain resilience is a multi-dimensional capability, with its core objective being the sustainable development of companies. In the digital era, digital empowerment holds significant potential to integrate supply chain management [17], thereby enhancing both supply chain resilience and sustainable development [18]. For example, Wang et al. found that under the impetus of China’s “dual carbon” goals, some high-tech retail enterprises have achieved green transformation through digital technologies, thereby enhancing their supply chain resilience. By analyzing these enterprises’ cases, it was revealed that digital empowerment not only improved operational efficiency but also fostered greater environmental awareness, which positively impacted their financial performance [19]. Additionally, Xue et al. highlighted that within big data pilot zones, digital innovation has enabled enterprises to optimize energy utilization efficiency, thereby enhancing the sustainability of their supply chains. This practice not only improves enterprises’ ability to respond to unexpected events but also provides long-term development assurance [20]. While supply chain resilience has garnered widespread attention, there remains a critical need to explore how digital empowerment can reinforce supply chain resilience across various dimensions and advance its development. Drawing on existing research, this paper categorizes the relationship between digital empowerment and supply chain resilience into the following dimensions:

(1) Collaboration capability perspective: The collaboration capability of supply chain subjects participants enhance collaboration efficiency based on the digital supply chain service platform [21], fostering overall synergy between the upstream and downstream elements of the supply chain and thereby strengthening supply chain resilience [22]. Digitalized supply chains require enterprises to transition from stability to adaptability, from standardization to customization, and from isolation to collaboration [23]. For instance, blockchain, as a peer-to-peer distributed network, can effectively mitigates information asymmetry within the supply chain and promote collaboration among its participants [24]. Saglam et al. highlighted the critical role of the coordination among supply chain members to realize supply chain resilience, and from the three dimensions of the quality of communication, reciprocity, and relational commitment, they emphasized that the process and systematic collaboration among supply chain members is a very important part of supply chain resilience. Collaborative relationships at both the process and system levels among supply chain members are integral to the development of supply chain resilience [25].

(2) Operational capability perspective: One of the specific manifestations of supply chain disruption is operational disruption, which significantly impacts supply chain resilience [26]. Digital empowerment enables modern organizations to leverage hardware, software, and communication networks to facilitate the development of platform-based supply chains [27], and the emergence of platforms and services enhance enterprises’ operational capabilities across purchasing, production, and sales processes. Zhu et al. highlighted that digital empowerment allows modern organizations to enhance operational capabilities in purchasing, production, and sales through the use of hardware, software, and communication networks [28]. Additionally, Fu et al. found that information accessibility significantly impacts enterprise operations in a digitalized environment, emphasizing the importance of information flow in enhancing operational capabilities [2]. Brusset and Teller conducted a questionnaire survey on 177 managers based on the resource-based view based on the theory of dynamic capabilities, and concluded that supply chain resilience is fundamentally an operational capability that enables interrupted supply chains to recover and strengthens supply chain resilience [29]. Therefore, digital empowerment not only enhances the operational capabilities of individual enterprises but also promotes synergies across the entire supply chain, strengthening its ability to address disruption risks.

(3) Financial strength perspective: Supply chain resilience is significantly and positively associated with corporate financial performance in the event of risky shocks [30]. The emergence of digital supply chain platforms provides dual credit services to supply chain enterprises, improving accessibility and affordability, enhancing demand responsiveness, and mitigating the risk of supply chain disruptions. This also helps address the financing challenges faced by micro, small, and medium-sized enterprises (MSMEs) in the supply chain [31]. Chen et al. examined the relationship between corporate financial performance and environmental performance in the context of a low-carbon economy, highlighting that digital transformation can enhance financial flexibility, enabling enterprises to better manage risks [32]. Kaur et al. emphasized that digital empowerment enables supply chain participants to respond rapidly and accurately to demand, providing services that satisfy stakeholders and making these services accessible and affordable, thus improving the financial strength of supply chain participants [33]. Based on the theory of resource orchestration, Jamal et al. verified the positive relationship between supply chain resilience and enterprise financial performance through the analysis of a structural equation model of 289 French firms [34].

(4) Redundant resource perspective: During state of supply chain disruption, supply chain members holding a higher inventory share will have a more favorable cost advantage, mitigating the shock of supply chain disruption as much as possible [35]. Lin et al. observed that current optimization of redundant resources in enterprises focuses primarily on inventory, which significantly shortens product life cycles and reduces cash flow velocity. However, the digitization of traditional supply chains can transform traditional inventory control and management approaches, breaking down barriers between enterprises and modern logistics [36]. Leung et al., utilizing a digital twin framework, integrated machine learning and simulation to optimize inventory replenishment policies, thereby minimizing the cash conversion cycle of the supply chain [37]. A digital supply chain platform that integrates simulation, machine learning and optimization enable real-time monitoring and analysis of key metrics such as production and inventory, enhancing service efficiency and reducing total costs during risky shocks. In supply chain risk management, business operations such as safety stock, maintenance of multiple suppliers, and low capacity utilization can be achieved to mitigate the impact of supply chain disruptions by keeping resources in reserve, which in turn improves the ability to anticipate and adapt to risks [38].

(5) Human capital perspective: The relational and human capital possessed by supply chain managers facilitates the establishment of key prerequisites for supply chain resilience, such as visibility, responsiveness, and flexibility, which collectively in turn strengthens supply chain resilience [39]. Schrauf et al. argued that the effective development of digital supply chains depends on the awareness of supply chain participants and the digital competence of employees [40]. The digital competence of employees in supply chain enterprises is a key factor in driving the advancement of digital supply chain, and the digital competence of employees will be transformed into enterprise knowledge stock, which promotes the formation of supply chain resilience through mutual learning among supply chain participants [41].

Through the above research, it is evident that there are multiple perspectives in the existing research on supply chain resilience, indicating that supply chain resilience is a complex set of concepts and that digital empowerment of supply chain resilience is not isolated or unilateral but is rooted in multi-dimensional factors, including collaborative relationships, operational capabilities, financial strength, redundant resources, and human capital. In today’s industrial landscape characterized by division of labor and specialization, the impact and influence faced by the manufacturing supply chain will also show complexity and depth. Therefore, focusing solely on the digital empowerment of the supply chain is insufficient. Instead, a comprehensive multi-dimensional assessment of how to enhance the resilience of the manufacturing supply chain in the context of digitization is particularly important. However, to date, the existing literature has not considered the supply chain resilience of digitally-enabled manufacturing supply chains from a multi-dimensional perspective. This paper aims to measure supply chain resilience based on the aforementioned five dimensions in order to explore the ’black box’ mechanism of digitally-enabled manufacturing supply chain resilience. The specific research framework is shown in Fig 1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Research framework.

https://doi.org/10.1371/journal.pone.0316183.g001

Research hypotheses

The moderating role of environmental uncertainty.

In the era of digital economy, supply chains and the enterprises in the chain are operate in highly uncertain environments. The rules of international trade are undergoing significant changes, with FTAs, higher customs unions, and common markets increasingly dominating global trade regulations, making uncertainty the norm. According to Teece’s theory of dynamic capabilities, ’dynamic’ refers to the ability to adjust promptly to changes in the external environment, while ’capability’ denotes the capacity to build, integrate, and reorganize resources [57]. Supply chain resilience is the scenario of coping with external shocks in a volatile and unpredictable environment, therefore, environmental uncertainty in the process of digitally empowering supply chain resilience is bound to have a deep impression on the formation of supply chain resilience.

Environmental uncertainty can affect supply chain resilience at both micro and macro levels. At the micro level, supply chain uncertainty primarily consists of demand uncertainty, supply uncertainty and technological uncertainty. First, demand uncertainty can lead to unknown and unpredictable demand quantity and time in the supply chain, resulting in inventory surpluses or shortages [58]. In turbulent environments, digital supply chain platforms can foster cooperative relationships among supply chain enterprises, enabling them to access necessary resources from partners, maintain market competitiveness, and collectively resist supply chain risks [59]. Secondly, supply uncertainty can lead to production stoppages, quality problems, logistical obstacles, etc. Walker and Weber pointed out that due to the opportunistic behavior of suppliers, they will change suppliers, adjust production and product specifications in the face of supply uncertainty, which will lead to unnecessary cost changes [60]. Therefore, digital platforms enhance collaboration between enterprises, strengthening the supply chain’s capacity to withstand risks associated with supply uncertainty. Finally, technological uncertainty requires firms to be able to quickly predict and share information to maintain their innovation capability in a state of rapid technological change. In turbulent environments, stronger supply chain relationships, facilitated by digital empowerment, enable enterprises to share information efficiently and enhance supply chain resilience. From the macro level, environmental uncertainty is mainly manifested in economic policy uncertainty [61], governments frequently adjust economic and trade policies to stabilize growth and smooth fluctuations, making it difficult for enterprises to anticipate changes in the policy environment. So enhancing supply chain resilience with digital empowerment becomes an important means for enterprises to cope with environmental uncertainty [62]. Based on this, this paper proposes the third hypothesis.

H3: Environmental uncertainty has a positive moderating effect on supply chain resilience in digitally empowered manufacturing.

Based on the above theoretical derivation and research hypotheses, the following Fig 2 conceptual model is constructed:

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Construct model.

https://doi.org/10.1371/journal.pone.0316183.g002

Sample selection and data sources

Given the rapid development of digital technologies in China and their integration with the real economy primarily since 2012, as well as to ensure the availability of digital empowerment data, this study selects China’s A-share manufacturing firms listed from 2012 to 2020 as the initial research sample. This period avoids the significant disruptions caused by the COVID-19 pandemic in 2021–2022, which resulted in supply chain breaks and delays for many firms, creating unavoidable shocks to manufacturing supply chains from both market and policy perspectives. The data handling process follows these steps: first, companies labeled as ST, PT, or those insolvent are excluded; second, instances with significant missing key data are omitted; and third, to mitigate the effects of data outliers, a 1% bilateral Winsorization is applied to continuous variables. The primary data source is the CSMAR database.

The explanatory variables (supply chain resilience), explanatory variables (digital empowerment), mediating variables, moderating variables, and control variables are discussed below:

(1) Supply chain resilience (Scr): the above literature review and analysis indicate that this paper constructs dimensions through five categories: redundant resources, financial strength, collaborative relationships, operational capacity, and human capital. These dimensions encompass a total of seven quantitative indicators, specific indicators as detailed in Table 1, and are measured using factor analysis. The specific measurement process is described in the following section.

(a) Redundant resources: Following methodologies employed in prior research, the ratio of management expenses to operating income is selected as an indicator to measure an enterprise’s redundant resources [63].

(b) Financial strength: As a critical measure of an enterprise’s capacity, financial strength is included as a key dimension of supply chain resilience. This paper uses return on net assets and net profit margin on total assets as indicators to assess financial strength [64].

(c) Collaboration: Based on a comparison of existing studies, this paper selects supply chain concentration as the indicator of collaboration [65].

(d) Operating capacity: Drawing on prior research results, accounts payable turnover and accounts receivable turnover are chosen as indicators to evaluate operating capacity [66].

(e) Human capital: Existing studies commonly use the ratio of employees with bachelor’s degree or higher as the indicator of human capital [67].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Definition and measurement of variables.

https://doi.org/10.1371/journal.pone.0316183.t001

(2) Digital Empowerment (Digi): Digital empowerment is a theoretical concept that has emerged alongside the development of digital technologies. At its core, it refers to "empowerment through digital technologies." Most scholars define it as the use of digital technologies to optimize organizational structures, business models, and management systems, thereby enhancing organizations’ knowledge accumulation and skill levels [68]. Enterprise digital empowerment is a systematic process, and existing research on digitalization measures mainly approaches it from the following perspectives.

(a) Some studies use the proportion of intangible assets and IT investment to measure the density of enterprise informatization [69, 70]. However, the level of enterprise informatization investment does not necessarily reflect the level of its actual application [71].

(b) The other part of the literature uses a questionnaire survey to measure the degree of digital transformation of the enterprise by adopting the proportion of enterprise information technology personnel [72], yet the use of computers may just be a simple web application behavior. Unlike the above research literature, this paper draws on Wu Fei et al.’s study [73] and uses textual analysis to portray the digital empowerment of enterprises.

Since enterprises often outline digital initiatives as part of their strategic development, these are reflected in their annual reports through summaries and guidance. The vocabulary used in these reports reflects the company’s future strategies, business philosophy, and development trajectory. Moreover, the vocabulary usage in the annual report can reflect the future strategy of the enterprise as well as its business philosophy and development path. Therefore, it is feasible to use text analysis to count the word frequency in the annual reports of enterprises. The specific steps are as follows: First step, annual reports of A-share manufacturing firms from 2012 to 2020 are collected. Python is used to extract the ’business analysis’ sections from these reports. The second step, identify digital empowerment keywords by reviewing national digitization-related policies and research reports issued between 2012 and 2020, such as the "Special Action Plan for Digital Empowerment of SMEs," the "2020 Digital Transformation Trend Report," and the "China Digital Transformation Research Report (2022)," as well as government work reports and research publications by the China Academy of Information and Communication Technology. Building on the research of Wu Fei et al. [73], the scope of digital empowerment feature words is further expanded to form a keyword library, covering AI, Big Data, Cloud Computing, Blockchain, and digital applications. In the third step, the annual reports are textually analyzed. Based on the digital empowerment thesaurus obtained in the second step, Python software is used to capture the digital empowerment keywords in the annual reports. In the fourth step, Words with negative modifiers (e.g., ’no’ or ’none’) preceding keywords are excluded. Keyword frequencies are counted, categorized, and totaled. To address the right-skewed nature of the data, keyword frequencies are logarithmized to create the digital empowerment index.

(3) Mediating and moderating variables:

(a) Mediating variable: Innovation vitality, measured by the ratio of R&D investment to business revenue in the manufacturing industry.

(b) Moderating variable (Eu): Environmental uncertainty, which can be narrowly defined as institutional factors such as policies and laws, or more broadly as encompassing macroeconomic, institutional, and physical environments. In enterprise-level studies, environmental uncertainty is often defined as a comprehensive external environment that affects business operations and can lead to fluctuations in business performance. With reference to the method of Shen Huihui et al. [74], environmental uncertainty is measured using the industry-adjusted coefficient of variation of business revenue. Since environmental uncertainty originates from fluctuations in external environments that affect an enterprise’s business activities, it ultimately leads to fluctuations in sales revenue. Its specific calculation formula is as follows: (1) In Eq (1), Sale represents firm’s sales revenue, Year denotes the year variable (Year = 1 for observations in the forth year, Year = 2 for the third year, and so on, up to Year = 5). The residuals obtained from the OLS regression of Eq (1) are the abnormal sales revenues, which are obtained by dividing the standard deviation of the abnormal sales revenues obtained in the 5 years by the average of the sales revenues in the 5 years. unadjusted industry uncertainty. Finally, the industry-adjusted environmental uncertainty indicator is obtained by dividing the industry-adjusted environmental uncertainty by the median environmental uncertainty of the industry in which the sample is located.

(4) Control variables: In order to ensure the accuracy of the study, a total of five control variables are selected in this paper:

(a) Enterprise size (Size): The data in this paper is selected from the data of A-share listed companies. Measured by the natural logarithm of total assets at the end of the previous period, based on the existing digitized literature;

(b) Cash flow (Cashflow): To capture the trend of corporate cash in the future, measured by the balance of net profit and total assets;

(c) Growth(growth): Capturing the development status and development speed of the enterprise’s business capability, measured by the ratio of total assets in the current period to total assets in the previous period;

(d)Liability (Lev): Captures the ability of a company to conduct business activities through creditor funds and the level of corporate liabilities, measured by the ratio of total assets to total liabilities;

(e) Top 1 Shareholding (Top1): Reflects corporate governance by capturing the level of ownership concentration. It is measured by the ratio of shares held by the top shareholder to the firm’s total shares.

Treatment of supply chain resilience variables

Applicability and correlation test.

The applicability and correlation test is conducted using the KMO test, which compares the correlation coefficient and partial correlation coefficient between variables. The results of the KMO test and Bartlett’s test, obtained through SPSS software, are presented in Table 2. From Table 2, it can be observed that the KMO statistics are all greater than 0.5, and the P-values of Bartlett’s sphericity test are all 0.000. These results indicate that the selected variables in this study have strong correlations with each other and are suitable for factor analysis.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Results of KMO test and Bartlett test for cross-sectional data from 2012 to 2020.

https://doi.org/10.1371/journal.pone.0316183.t002

Factor analysis.

Since some original data points are missing, which could affect the results of factor analysis, this paper employs the linear interpolation method to address the missing values. Specifically, a linear function is constructed based on the neighboring samples of the missing values to ensure sufficient sample size. Additionally, a shrinking-tail treatment is applied to mitigate the impact of extreme values. The principal component method is used to extract the common factors, and the eigenvalues and variance contribution rates of these factors are presented in Table 3. The eigenvalues of the first three factors are all greater than 1, with a cumulative variance contribution rate of approximately 65%. Thus, the three extracted factors effectively capture the information of the secondary indicators.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Contribution rate of total variance in factor analysis of cross-section data from 2012 to 2020.

https://doi.org/10.1371/journal.pone.0316183.t003

Synthesis of the composite index.

The scores of the three factors (F1,F2,F3), and the coefficient matrix of the 2019 factor scores are presented in Table 4, and the factor scores of a sample company are calculated based on the coefficient matrix of the 2019 factor scores:

F1 = 0.895X1+0.897X2-0.660X3+0.311X4+0.253X5-0.071X6-0.151X7

F2 = 0.311X1+0.321X2+0.219X3-0.670X4-0.512X5-0.073X6+0.595X7

F3 = 0.103X1+0.080X2+0.197X3+0.188X4-0.006X5+0.954X6+0.153X7

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Factor score coefficients in 2019.

https://doi.org/10.1371/journal.pone.0316183.t004

Subsequently, based on the variance contribution rate of each public factor, the composite factor score (Scr) of supply chain resilience of the sample company is derived. Taking 2019 as an example, the Scr of a sample company is calculated as follows:

Using the same method, the composite index score of supply chain resilience (Scr) is calculated sequentially for each listed company across all years, providing a consistent measure of supply chain resilience.

Descriptive statistics

The results of the descriptive statistics are presented in Table 5, with a total of 14,163 samples obtained through data organization and matching. The mean value of supply chain resilience (Scr) is 7.3876 with a standard deviation is 6.894, indicating significant variation in supply chain resilience among individuals. For digital empowerment (Digi), the mean and standard deviation are 1.2419 and 1.275, respectively, with minimum and maximum values of 0 and 6.06. These results suggest a noticeable disparity in the level of digital empowerment among enterprises, reflecting an uneven integration of digitalization and the real economy. Additionally, the horizontal differences among the other variables are significant, and the variability of each control variable is large. All main variables fall within a reasonable range, highlighting the substantial gap in digital empowerment levels among listed companies. Notably, some enterprises have yet to enhance their supply chain resilience through digital empowerment. Therefore, testing the supply chain resilience of China’s manufacturing industry in the context of digital empowerment holds strong theoretical and practical significance.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 5. Descriptive statistics.

https://doi.org/10.1371/journal.pone.0316183.t005

Model setting

Based on the hypotheses proposed in this paper, the following regression model is constructed to examine the impact of digital empowerment on manufacturing supply chain resilience: (2) In Eq (2), the explanatory variable Scr represents supply chain resilience, as previously defined. The explanatory variable Digi denotes the digital transformation of enterprises, and the coefficient β measures the extent of Digi’s influence on supply chain resilience. Controls is the firm-level control variable. Finally, in order to control the impact of year and industry factors on supply chain toughness, this paper adds industry and year fixed effects in the equation.

To further explore the mechanisms through which digital empowerment influences manufacturing supply chain resilience, the following model is constructed based on Eq (2) using enterprise innovation vitality as the mediating variable. (3) (4) In this model, R&D serves as the mediating variable, representing innovation vitality, while other variables remain the same as in Eq (2). A significant regression coefficient β indicates that R&D contributes to the mediating mechanism of digital empowerment in enhancing supply chain resilience, although it may not be the sole mechanism.

In order to analyze whether environmental uncertainty moderates the relationship between digital empowerment and supply chain resilience, Eq (2) incorporates the interaction term between environmental uncertainty and digital empowerment: (5) Here, Eu denotes environmental uncertainty, and the coefficient of the interaction term represents the moderating effect of environmental uncertainty on the relationship between digital empowerment and supply chain resilience.

Benchmark regression

Table 6 presents the regression results for digital empowerment(Digi) and supply chain resilience (Scr). The regression coefficients of Digi, the primary focus of this paper, are all significantly positive across columns (1) (2) (3). In column (1), which includes control variables, the regression coefficient is 0.173 and significant at the 1% level. Column (2) incorporates industry and year fixed effects, resulting in a regression coefficient of 0.111, significant at the 5% level. In column (3), after controlling for both industry and year fixed effects, the regression coefficient increases to 0.224, remaining significant at the 1% level. These findings tentatively suggest that digital empowerment effectively enhances manufacturing supply chain resilience, supporting the validity of Hypothesis 1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 6. Empirical results of digital empowerment of manufacturing supply chain resilience.

https://doi.org/10.1371/journal.pone.0316183.t006

Mechanism path and moderating effect tests

Mechanism path test.

To verify hypothesis 2, this paper uses innovation vitality as the mechanism variable, and further explores the influence mechanism of digital empowerment on manufacturing supply resilience based on the steps of mediation effect test.

Column (3) of Table 7 indicates that digital empowerment can effectively promote the growth of manufacturing innovation vitality and improve the innovation level of supply chain enterprises, with a regression coefficient of 0.349, which is significant at the 1% level. In column (1), which excludes control variables, the results remain significant (coefficient 0.259, p<0.01). Column (4) reports the test results of the mediating effect of firms’ innovation vitality (coefficient 0.185, p<0.01). Similarly, in column (2), which excludes control variables, the coefficient is 0.141, also significant at the 1% level. These findings suggest that digital empowerment contributes to the resilience of manufacturing supply chains by promoting increased innovation vitality among firms in the supply chain, thereby supporting Hypothesis 2.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 7. Test results of mechanism path and regulatory effect.

https://doi.org/10.1371/journal.pone.0316183.t007

Endogeneity and robustness test

Troubleshooting endogeneity. Selecting appropriate instrumental variables is an effective way to solve the endogeneity problem. Drawing on the methodology of Huang Qunhui et al. [75], this study uses the number of landline telephones per million people in 1984 in historical cities is used as an instrumental variable for digital empowerment. On the one hand, dial-up access to telephone lines marked that the Internet began to gradually come into the public’s view, followed by ISDN, ADSL, and fiber-optic broadband access, and at the same time, contributed significantly to the continued development of the subsequent digital technology. On the other hand, along with the development of digital technology and the economy, the frequency of the use of traditional dial-up telephones has gradually declined in line with the exclusivity. In addition, since the original data of instrumental variables are cross-sectional data, this study adopts the approach of Nunn and Qian [76] by incorporating dynamic variables. Specifically, the amount of national Internet investment and the number of landline telephones per million people in the preceding year are used to construct interaction terms, enabling the instrumental variable for digital empowerment to vary over time.

From the results in column (1) (2) of Table 8, it can be seen that the F-test for the instrumental variables exceeds the critical value for the 10% level of the Stock-Yogo weak identification test. Additionally, the P-value of the LM statistic is 0.000, which significantly rejects the original hypothesis. These findings confirm the validity of the selected instrumental variables. Furthermore, the regression results for the impact of digital empowerment on manufacturing supply chain resilience remain significantly positive after addressing endogeneity concerns, thereby reaffirming the validity of Hypothesis 1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 8. Endogeneity and robustness tests.

https://doi.org/10.1371/journal.pone.0316183.t008

Heckman two-step method. To address the endogeneity caused by sample selection bias in the model, the Heckman two-step model is employed. In the first stage, a digital empowerment dummy variable (Digi_dum) is constructed using the annual median of the digital empowerment variable (Digi_m). Values greater than or equal to the median are assigned 1, while others are assigned 0. A Probit regression is then performed with the digital empowerment dummy variable, Digi_m, and the control variables from the baseline regression, and the inverse Mills ratio (Imr) is calculated. In the second stage, the Imr is incorporated into Eq (2) for regression. As reported in column (3) of Table 8, the estimated coefficient of Digi is 0.236, significantly positive at the 1% level. Additionally, the estimated coefficient of Imr passes the 10% significance level, indicating the presence of sample selection bias in this study. After adjusting for this bias, the research conclusions remain robust.

Controlling for the effect caused by industry year events. To address potential interference from industry- and year-specific events, regional-year interaction fixed effects are introduced in addition to the industry and year fixed effects. This approach accounts for region-specific time effects and mitigates the impact of special regional changes on the baseline regression. As reported in column (3) of Table 8, the coefficient of Digi is 0.240, which is significantly positive at the 1% level.

Replacement Variable Tests. To further ensure the robustness of the findings, the digital transformation index is employed as an alternative for the digital empowerment index, and the replacement variable is reintroduced into Eq (2) for further testing. The digital transformation index is derived from the CSMAR (Cathay Pacific) database, and is computed by assigning weights of 0.3472, 0.162, 0.0969, 0.0342, 0.2713, and 0.0884 to six key indicators: strategic leadership, technology-driven, organizational empowerment, environmental support, digital outcomes, and digital applications. These values are then aggregated to produce the final digital transformation index. As reported in column 4 of Table 8, the Digi coefficient is 0.0302, which remains significantly positive at the 1% level.

Quantile regression. Quantile regression was employed to exam in the differential impact of digital empowerment across varying levels of supply chain resilience. Quantile regression can capture the effects of the independent variable under different conditions, making it particularly well-suited for analyzing heterogeneous impacts at different levels. Based on Eq (2), Table 9 presents the results of quantile regression at the 10%, 30%, 50%, 70%, and 90% quantiles of supply chain resilience. The findings reveal that digital empowerment significantly promotes supply chain resilience at all levels, and the stronger the supply chain resilience, the greater the effect of digital empowerment on it.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 9. Quantile regression.

https://doi.org/10.1371/journal.pone.0316183.t009