Untangling the role of blockchain technology in shaping supply chain concentration from the contingency and configuration perspectives

Abstract

This study presents an empirical exploration of the relationship between blockchain technology (BT) and supply chain concentration (SCC) from contingency and configuration perspectives by using a panel dataset of 147 Chinese enterprises that applied blockchain technology from 2019 to 2023. Multiple regression analysis (MRA) results indicate that BT is essential for the two dimensions of SCC: supplier concentration (SC) and customer concentration (CC). Additionally, we find that R&D investment, listed years, and industry type strengthen the BT-SCC relationship. A fuzzy-set qualitative comparative analysis (fsQCA) further reveals that none of the aforementioned variables independently lead to a high level of SCC; rather, SCC requires a combination of these variables alongside BT adoption. These findings highlight alternative pathways through which different enterprises can achieve high levels of SCC by utilizing BT, a topic that has not been explored in prior research.

Introduction

The COVID-19 pandemic has highlighted the importance of supply chain concentration (SCC) as a viable strategy that firms can use to address the growing uncertainties prevalent in industries while improving their business performance¹. SCC denotes how dependent a firm is on a limited number of key partners in its supply network, encompassing both its upstream and downstream relationships. Specifically, it reflects the extent to which firms concentrate their transactions among a few major upstream suppliers and downstream customers, forming the two dimensions of SCC: supplier concentration (SC) and customer concentration (CC)². By leveraging SCC, companies can lower their operational costs, create customer value, and enhance their overall business performance². Nevertheless, many firms continue to face challenges with regard to effectively integrating with their key suppliers and customers³. Evidence has shown that SCC-related initiatives have a high failure rate, leading executives from Fortune 1000 companies to be confused about how to pursue a higher level of concentration with suppliers and customers. Thus, the underlying drivers of SCC need to be identified to achieve competitive supply chain advantages.

In fact, SCC is a complex phenomenon that can be influenced by a myriad of factors, including partnership⁴ and trust⁵. Among these factors, information technology (IT) has been widely recognized as one of the most critical enablers. Scholars have increasingly investigated the relationship between IT and SCC, and empirical evidence has substantiated the crucial role of IT in enhancing SCC^6,7. However, an alternative perspective has been proposed in the literature, which posits that an SCC can be realized only through supply chain coordination⁸. This coordination can be conceptualized in terms of trust, commitment, and interpersonal relationships, emphasizing intrafirm relationships and, in particular, the level of trust⁸. Nevertheless, the traditional IT architecture may pose challenges such as poor information transfer, opacity, asymmetry, and untraceability, presenting significant hurdles for enterprises’ attempts to bridge the trust gap with their upstream/downstream partners to achieve supply chain coordination and enhance their SCC⁹. Fortunately, the emergence of blockchain technology (BT) has resolved these issues, facilitating supply chain coordination and enhancing SCC.

BT has recently emerged as an innovative solution for IT in supply chains, which transforms traditional IT architectures^10–12. It enables a new service pattern according to which supply chain stakeholders can engage in transactions in a decentralized environment and collectively maintain distributed databases (ledgers) for recording transaction data^13,14. Compared with conventional IT, BT has several unique features, such as decentralization and negation of trust, security, traceability, and intelligence, making it highly compatible with supply chain business scenarios involving multiple parties¹⁵. These features ensure that blockchain data are tamper-proof and that their flow is mutually recognized¹⁶, enabling enterprises to share information and achieve full traceability with their upstream and downstream collaborators. Hence, BT can establish a “decentralized trust structure” among supply chain stakeholders, facilitating the maintenance of cooperative relationships as well as supply chain coordination^5,9,17.

Thus, some scholars have theoretically proposed that BT adoption improves concentration among major customers and suppliers¹⁶. However, other researchers have highlighted the potential dark side of adopting BT, which may hinder supply chain management and concentration¹⁸. These scholars have suggested that BT makes information about major customers and suppliers more transparent, which may cause these actors to focus on their own interests and be reluctant to concentrate business with focal firms¹⁹. However, most arguments regarding the relationship between BTand the supply chain in the aforementioned studies have been made at a theoretical level, entailing only qualitative descriptions and inferences^20,21. Accordingly, our first research aim is to investigate the BT-SCC relationship.

In the context of commerce, suggesting that BT is universally applicable would be an erroneous assumption^22,23. At present, academic discourse, as reflected in the study by Zhong, et al²⁴., postulates that the effectiveness of emerging technology applications is not a static value but instead intricately intertwined with the characteristics of the adopting firms²³. Regrettably, the academic literature has thus far largely overlooked the moderating effect of environmental factors, whether intrinsic or extrinsic to a firm, on the potency of BTadoption, and even less attention has been given to the distinctive adoption strategies employed by diverse firms within the sphere of blockchain technologies with a specific focus on supply chain management(SCM)^25,26. To identify the situational value of BT, this study incorporates the variables of R&D investment, market longevity, and industry type as potential moderators. In accordance with the literature, these variables are divided into internal and external firm factors, and their potential to moderate the BT-SCC relationship is scrutinized.

Fuzzy-set qualitative comparative analysis (fsQCA), a method that connects qualitative and quantitative research, uncovers the links between antecedent combinations and resulting outcomes^27,28. This approach, which has been employed in fields such as corporate governance and innovation management, reveals causal complexities that are often elusive with respect to standard quantitative methods²⁹. By complementing the findings of research models that were initially studied using regression analyses^29–32, fsQCA provides a nuanced understanding of intricate issues, overcoming the limitations of traditional hypothesis testing. Despite its strengths, the use of this approach to explore the mechanism of SCC formation is uncharted, and prior studies have relied mainly on cross-sectional data, thereby neglecting the relevance of longitudinal studies³³. Given the intricate interactions between BT and various intrafirm and extrafirm factors that influence an enterprise’s SCC in reality, this paper advocates an innovative longitudinal study method based on fsQCA. This approach aims to identify organizations’ aptitude for BT implementation as well as combinations that can foster high SCC levels³³.

This research meaningfully advances and makes several distinct contributions to the literature on BT and SCC. (1) Whereas prior research has primarily examined the overall relationship between BT adoption and supply chain performance, we open the black box by developing a dual-layered capability framework grounded in the hierarchical organizational capability perspective. This study among the first to clarifies how BT facilitates capability building and value creation through functional hierarchies, thereby revealing the underlying mechanism by which BT influences supplier and CC. (2) We explore the interplay between BT and contextual factors by identifying the moderating roles of intraorganizational factors (R&D investment and firm age) and external environmental factors (industry type) that shape the BT–SCC relationship. These findings specify the boundary conditions under which BT exerts a stronger or weaker effect on SCC. (3) By integrating longitudinal fsQCA with traditional regression methods, we uncover multiple equifinal configurations corresponding with high SCC level and propose two pathways: “BT-driven solution” and “intrafirm & extrafirm solution”. Together, these insights enrich the literature on digital supply chain management and offer actionable implications for managers seeking to leverage BT strategically to strengthen supplier and customer relationships.

Theoretical background

BT implementation

BT, which is the foundation of Bitcoin¹³, represents a comprehensive distributed system that securely logs a linear, immutable record of transactions among network participants^13,20. It acts similarly to a collaboratively maintained, distributed ledger⁹. Its distinguishing features are decentralization^15,17, negation of trust^13,20, security¹⁸, traceability³⁴, and intelligence^35,36.

Previous research has largely been conceptual and has exhibited a technical focus on specific BT issues such as operational mechanisms and algorithm optimization. Empirical studies have primarily investigated the factors affecting BT adoption, and the link between BT and business ability has often been overlooked. Since BT research remains in this nascent stage, challenges persist with respect to its application in the field of enterprise operations management¹⁶. The operation-related capability-building path supported by BT urgently needs to be explored using its unique features as a starting point.

BT and the supply chain

BT is being explored in the context of SCM as a potentially disruptive innovation^16,17. However, BT is still emergent, and only limited literature has investigated this topic, especially the BT–SCM connection. Extant BT-SCM research can be categorized into the following groups: (1) conceptual studies^21,37,38, (2) game theory-oriented research^14,39,40, (3) case studies on BT applications²³, and (4) empirical investigations into value creation through BT^16,41.

Most previous BT-SCM research has largely been conceptual, and BT’s value-creation mechanism and its interaction with SCC has received insufficient empirical attention.

Table 1, which illustrates recent “dual-chain” relationship empirical studies, shows that these studies have employed mainly questionnaires, with few using secondary data for quantitative analysis. Studies have overlooked the impact of BT on SCC, leaving the relationship between BT and SCC unclear.

Table 1.

Representative empirical studies in BT-SCM field.

Representative literature	Influenced SCM objectives								Data source
	Cost	Agility	Reliability	Risk	Sustainability	Transparency	Performance	Concentration
Wang, et al42 Jajja, et al43	✔ ✔	✔ ✔	✔	✔ ✔		✔	✔		Questionnaire Questionnaire
Di Vaio and Varriale20 Donkor, et al44 Munir, et al45				✔	✔ ✔		✔ ✔		Questionnaire Questionnaire Questionnaire
Aslam, et al41							✔		Questionnaire
Fosso Wamba, et al16						✔	✔		Questionnaire
Karamchandani, et al46			✔	✔			✔		Questionnaire
Orji, et al47							✔		Interview
Cong, et al18									/
Klöckner, et al48							✔		Secondary data
This study								✔	Secondary data

SCC

SCC reflects the density of a firm’s supplier and customer portfolios in terms of the relative importance of its major suppliers and customers^2,19. Thus, SCC is an important structural factor of the supply chain portfolio and is one of the core goals of SCM¹. High levels of SCC indicate that a firm distributes a high percentage of its purchases to a few major suppliers or sells a high percentage of its products to a few major customers. Concentrating business on a few major partners signals a firm’s long-term relationship commitment to these partners and increases mutual interdependence between a firm and its concentrated partners. This situation can further encourage a firm and its partners to abandon their short-term interests and instead actively promote the cultivation of strong exchange ties^2,19. To some extent, the SCC captures the extent to which the supplier and customer portfolio are characterized by embedded ties with major suppliers and customers. These embedded ties between external business processes and supply chain members, which are established through strategic cooperation, can help enterprises achieve efficient product, service, information, capital, and decision flows and create maximum value for customers efficiently and cost-effectively¹⁹. In this sense, SCC can be regarded as a higher-level functional capability associated with operation-related capabilities from the hierarchical organizational capability perspective.

With respect to the dimensions associated with SCC, most scholars have divided this notion into SC and CC^19,24. These terms denote the level of cooperation between firms and their principal suppliers and customers in the context of streamlining supply chain processes. Previous empirical studies on SCC have predominantly used questionnaires without providing a quantitative delineation of SC and CC. In this study, the SCC is measured quantitatively in two dimensions (SC and CC) using secondary data. This approach can, to a certain degree, overcome the limitations of existing studies and facilitate a more rigorous examination of the connection between BT implementation and the SCC.

Hierarchical organizational capability theory

Hierarchical organizational capability theory, which is rooted in the resource-based view (RBV)^49,50, has been an influential yet relatively understudied development in the RBV literature^49,50. According to this theory, organizational capabilities are divided into lower-level functional capabilities (also referred to as operational capabilities) and higher-level dynamic capabilities⁵¹. The former capabilities focus on functions such as logistics, marketing, and distribution, whereas the latter capabilities facilitate the adaptation of lower-level capabilities to dynamic environments⁵². Scholars have connected organizational resources and these capabilities to enterprise performance, positing that lower-level capabilities are cultivated through resource and expertise integration and then further incorporated into higher-level capabilities to promote value creation (enhanced organizational performance or competitive advantage achievement).

However, despite investigating the organizational capability hierarchy, most studies have broadly classified lower-level functional capabilities and higher-level dynamic capabilities separately, neglecting the hierarchical structure that can be observed within a range of dynamic and functional capabilities⁵⁰. Thus, these capabilities must be explored in greater depth, as this approach can enable us to obtain a more comprehensive understanding of the sources of capability-building and value creation in enterprises.

As the proposer of the organizational capability hierarchy, Grant⁵² conceptualized organizational capabilities in terms of firms’ ability to consistently perform specific productive tasks. In his seminal work, he highlighted the existence of capability levels within lower-level functional capabilities (Grant⁵². Task-level capabilities enable a firm to efficiently perform specific, repetitive tasks and can be integrated into higher-level functional capabilities, such as product development, marketing, and logistics^51,52. Grant classified organizational capabilities into five tiers: single-task, specialized, operation-related, broad-functional, and cross-functional capabilities. SCC falls under operation-related capabilities, while various information technologies are categorized as specialized capabilities.

Thus, building upon hierarchical organizational capability theory, configurational theory, and the conceptualizations of the hierarchy of capabilities provided in Grant’s research, this paper explores the capability-building process and the hierarchies present within functional capabilities. We propose a framework that highlights the mechanism underlying the influence of BT on SCC, including “BT implementation” (lower-level functional capabilities related to specialized capabilities) and “SCC” (higher-level functional capabilities related to operation-related capabilities). The previous conceptualizations of the hierarchy of capabilities are compared with those of the model developed in this study in Fig. 1.

Fig. 1.

A comparison of the proposed and existing capability hierarchies.

Theoretical model and hypotheses

This study is anchored in hierarchical organizational capability theory, transaction cost theory, and configurational theory. The latter has been employed by numerous scholars as a method to enhance findings that have primarily been analyzed using regression models^30,31. It offers a way to counter the potential oversimplification of hypothesis testing through regression methods^30,31. Considering the intricate dependencies between BT and a myriad of intra- and extrafirm factors, configurational theory could provide a fresh analytical perspective from which to investigate the correlation between BT and SCC in real-world settings. Consequently, our proposed research models of SCC, which are grounded in these theories, are visualized in Fig. 2.

Fig. 2.

Research model.

BT and enterprise SCC

SC represents a competency that enables firms to cultivate strategic alliances with their upstream suppliers, thereby achieving comprehensive and synchronized collaboration across operational and transactional processes with the goal of enhancing overall supply chain efficacy^53,54. Through its distinctive technical attributes, BT revolutionizes the traditional IT-based mode of information transfer within the supply chain⁵⁵. It establishes a “decentralized trust structure” between firms and their upstream suppliers, considerably decreasing transaction costs and effectively promoting SC.

Notably, current research has identified several unique features of BT that allow firms to share information with their upstream suppliers, thereby increasing the transparency and security of supply chain information^55,56. Throughout this top-down information flow, product information traceability is ensured without the involvement of third-party intermediaries. This paradigm shift in information transfer establishes a novel decentralized trust structure between firms and upstream suppliers, thus establishing “trust without trust”^5,57. Trust, as an important catalyst for the establishment of robust supply chain relationships and the enhancement of supply chain efficacy, can be significantly fostered by this new decentralized trust structure. Wamba and Queiroz⁵⁸ suggested that this “decentralized trust structure” can alleviate trust-related issues between firms and their suppliers, suppress opportunistic behaviors, and foster the establishment and maintenance of partnerships. Hence, the role of BT in increasing SC is crucial. According to hierarchical organizational capability theory, this situation represents a capability-building process in which “BT implementation” (a lower-level functional capability linked with specialized capabilities) is integrated into “SC” (a higher-level functional capability linked with operation-related capabilities).

Furthermore, similar to the ability of TCP/IP to reduce connectivity costs considerably¹⁸, scholars have noted that the decentralized technical nature of BT can significantly reduce transaction costs for BT-supported supply chain companies^17,59. From the perspective of transaction cost theory, transaction costs originate mainly from transaction uncertainty and the opportunistic behavior of parties toward the transaction¹⁷. This cost reduction facilitates the establishment of closer partnerships between firms and their upstream suppliers^17,18, subsequently bolstering SC.

Recently, several enterprises, including Walmart⁶⁰, have successfully enhanced SC by deploying BT in supply chain scenarios. Consequently, we posit that BT applications can enhance SC.

Consistent with these multiple streams of logic, we hypothesize the following:

Hypothesis 1

BT implementation has a positive effect on enterprise SC.

CC, as a concept, captures the ability of firms to establish robust strategic alliances with their downstream clients, facilitating synchronized collaboration across operational and transactional processes and subsequently enhancing the overall efficiency of the supply chain⁶¹.

The crux of blockchain as a distributed database lies in its unique encryption technology, consensus mechanism, and decentralized operational mode¹⁷, which ensure the immutability of supply chain data and establish a persistent, shareable, traceable, transparent, and secure database⁵⁷. These traits significantly heighten the transparency and security of product information¹⁶, thus bolstering the authenticity and traceability of products and amplifying the trust of downstream clients in upstream enterprises¹⁶.

Moreover, the concept of smart contracts, which are embedded with scripted code to facilitate the automated execution of transactions, mitigates the high transaction costs and risk of default associated with traditional contracts, thereby suppressing opportunistic behaviors⁴⁰. In line with transaction cost theory, such reductions can pave the way for more seamless partnerships between enterprises and their downstream clients, leading to enhanced CC¹⁷.

From the perspective of hierarchical organizational capability theory, this scenario involves a concrete capability-building process wherein “BT implementation” (a lower-level functional capability associated with specialized capabilities) is amalgamated into “CC” (a higher-level functional capability associated with operation-related capabilities). Several companies, such as IBM and Maersk, have attempted to employ BT in supply chain contexts to augment CC, leading to successful outcomes^60,62. Hence, the following hypothesis is proposed:

Hypothesis 2

BT has a positive effect on enterprise CC.

The moderating effect of listed years

Both tangible and intangible resource investments are needed for enhancing SCC and the adoption of BT and include research and development initiatives, the development of supply chain information systems, and the engagement of enterprise management¹⁶. Compared with nascent firms, mature enterprises with a long IPO history typically possess a robust set of resources, expertise, and technology, which effectively offers them a more potent resource base. Resource integration has frequently been emphasized as the foundation of SCC. Consequently, a relative advantage in terms of resource capabilities equips mature enterprises with the necessary capital and technical reserves to support their BT applications, rendering these applications more effective with regard to augmenting SCC. Some researchers have claimed that this resource advantage, in addition to external environmental factors, can shape the trajectory of an enterprise’s strategic evolution⁶³.

In our view, this resource advantage positively influences the use of BT to drive the implementation of SCC strategies. Specifically, for startups, the process of harnessing BT to achieve SCC faces substantial obstacles because of resource constraints. Conversely, enterprises with a more extended history on the stock market are likely to attain greater SCC upon the adoption of BT because of their comparative resource advantages. Thus, the following hypothesis is proposed:

Hypothesis 3

The impact of an enterprise’s implementation of BT on SC is positively moderated by the number of years for which it has been listed.

Hypothesis 4

The impact of an enterprise’s implementation of BT on CC is positively moderated by the number of years for which it has been listed.

The moderating effect of R&D

The journey from investing in to applying BT is founded on a robust resource base. Over time, researchers have explored the connection between R&D investment and technological advancement, and the majority of such studies have suggested that R&D investment provides solid resource support for technological exploration and application, thereby positively influencing the technology level of the enterprise^64,65. However, numerous scholars have emphasized that misguided R&D investment may not necessarily result in technological progress and may even impede the daily operations of an enterprise⁶⁶.

From an input–output perspective, this “R&D investment–technology level” paradox could be attributed to low technological innovation efficiency. Some scholars have suggested that a higher level of technological innovation efficiency and the resulting improved SC can be achieved only on the foundation of supply chain coordination⁸, which embodies commitment, trust, and interpersonal relationships among enterprises.

In this paper, we incorporate R&D investment into our research framework to investigate the existence of the “R&D investment-technology level” paradox within the blockchain context. We posit that BT, as a representative of next-generation IT, can enable enterprises to achieve full traceability and information sharing through key technical attributes such as decentralization, negation of trust, security, traceability, and intelligence. Supported by this technology, a decentralized trust structure can be established between enterprises and their upstream/downstream collaborators, leading to true supply chain coordination. As BT is still in a nascent stage, its initial exploration and application are characterized by high investment, long periods, substantial risk, and slow returns⁶⁷. Enterprises with higher R&D investments typically have greater resource reserves; thus, they exhibit an elevated level of technological innovation efficiency. Hence, the enhancement effect of the application of BT on SCC is likely to be stronger for such enterprises. Hence, we hypothesize the following:

Hypothesis 5

The impact of an enterprise’s BT implementation on SC is positively moderated by its R&D investment.

Hypothesis 6

The impact of an enterprise’s BT implementation on CC is positively moderated by its R&D investment.

Moderating effect of industry heterogeneity

Owing to the burgeoning service economy, supply chain integration is progressively being extended from conventional manufacturing to service sectors²⁵. The overall technical efficiency level of nonmanufacturing enterprises, which primarily includes service sectors such as advertising, software services, and health care, is on the rise. Studies have suggested that these nonmanufacturing entities generally exhibit higher technological efficiency levels than their manufacturing counterparts do, implying that the positive impact of IT, including BT, on SCM tends to be more potent for these entities⁶⁸.

Service-oriented nonmanufacturing companies typically exert weaker control over their upstream and downstream partners than do manufacturing firms do. Furthermore, their operations are not usually anchored in well-established SCM information systems. Consequently, nonmanufacturing enterprises frequently encounter challenges such as a lack of supply and sales information and unified system standards with upstream and downstream entities. This situation often results in a gap between these enterprises and manufacturing enterprises in terms of the closeness of the industrial and supply chains. This inherent deficit in SCC means that nonmanufacturing enterprises have greater potential for SCC enhancement; when these enterprises apply BT to SCM, they can more effectively facilitate information sharing and resource integration, leading to a more pronounced improvement in SCC²⁵. Accordingly, we propose the following hypothesis:

Hypothesis 7

The implementation of BT leads to greater enhancements in the SC among nonmanufacturing companies than among manufacturing enterprises.

Hypothesis 8

The implementation of BT leads to greater enhancements in CC among nonmanufacturing companies than among manufacturing enterprises.

Methodology

Data collection and sample test

Our research methodology is based on the foundation laid by Pan, et al⁶⁷., involving a manual search on Baidu, China’s leading search engine, to procure announcements and reports pertinent to the implementation of BT by Chinese A-share listed companies. Search terms included phrases such as “blockchain technology” and abbreviations of listed companies. Each of the A-share listed companies that implemented BT was examined individually.

The data collection and organization process featured three distinct phases. First, we curated a preliminary list of 207 sample companies by investigated whether the enterprises released announcements related to their engagement with BT. This list was expanded by including BT-associated enterprises from the “Blockchain Development Report for Chinese Listed Companies (2023–2024)” issued by the Interchain Pulse Institute, resulting in 271 eligible enterprises.

In the third phase, to further ensure the validity and credibility of our sample, we omitted 124 enterprises that had been listed for only a few years, those undergoing major asset restructuring, and those with an excessive amount of missing data. Enterprises that were labelled as ST/PT were also eliminated.

The final sample comprised a panel of 147 A-share listed enterprises and featured data spanning a 5-year period from 2019 to 2023. The descriptive statistics of our variables can be found in Table 2.

Table 2.

Descriptive statistics.

Construct	n	Mean	Std. Dev.	Min	Max
SC	735	3.442	0.593	2.010	4.605
CC	735	3.112	0.576	1.619	4.223
BT	735	0.456	0.498	0.000	1.000
Size	735	22.351	1.200	19.959	26.395
Lev	735	39.860	19.078	6.100	86.270
Age	735	10.410	6.106	1.000	25.000
Industry	735	0.305	0.461	0.000	1.000
R&D	735	18.470	1.289	15.671	21.838

Model construction

To verify the impacts of BT implementation on SC and CC, an empirical analytical model was constructed, the specific forms of which are shown in Models (1) and (2):

1

2

In Models (1) and (2), subscript i represents the enterprise, while subscript t represents the year. and reflect the SC and CC, respectively, of enterprise i in year t. , the core explanatory variable in these models, reflects BT implementation. is the regression parameter to be estimated, which reflects the influence of BT implementation on enterprises’ SC and CC. is an identically, independently and normally distributed error term. Furthermore, given the comparability and availability of the financial index of the listed enterprises, we included the total asset size (Size), the financial leverage ratio (Lev), and the dummy variable year as control variables. Linear ordinary least squares (OLS) were used as the basic regression method.

To test Hypotheses 3–6, we successively added the interaction terms BT and age (listed years) and the interaction terms BT and R&D to the model, thereby verifying the moderating effects of the enterprises’ year-to-market and R&D investment. The specific forms of these empirical analytical models are as follows. Next, to test Hypotheses 7 and 8, we conducted a group regression analysis based on whether the sample enterprises belonged to the manufacturing industry. Then, based on the coefficients and significance of the results thus obtained, we discussed the different impacts of the implementation of BT on the SSC of manufacturing and nonmanufacturing enterprises.

3

4

5

6

Variable definition and measurement

Independent variable: blockchain technology implementation ()

All the sample enterprises adopted BT and exhibited differences only in the year of implementation. In addition, the publicity of the BT project indicates that BT has a major impact on SCC. However, given the complexity of BT, the process of advancing from layout to actual application usually takes 1–2 years. Thus, this paper draws on the metric of “BT implementation” developed by Pan, et al⁶⁷. to define the years associated with BT implementation. Concretely, the year in which the enterprises issued the BT layout announcement and the subsequent two years were identified as implementation-related years, in which context the value of was 1, while the value of in other years was 0.

Dependent variables: supplier concentration () and customer concentration ()

In line with prior studies on SSC¹⁹, SC and CC are measured as the ratio of the total purchase (or sales) amount from the top five suppliers (or the top five customers) to the firm’s annual total purchase (or sales) amount. These quantitative variables provide standardized, objective measures of firms’ reliance on key partners (suppliers and customers) in the supply chain, respectively. On this basis, we logarithmically processed the data to construct indicators and .

Moderating variables

• Listed years (Age).

This indicator represents the number of years from IPO to year t, thereby reflecting the accumulation of capital, technology and experience to a certain extent.

• R&D investment (R&D).

“R&D investment” has been used in multiple previous studies to measure the amount of capital invested in product research and development²⁶. Thus, this indicator was measured in terms of the “total R&D expenditures” listed in the annual reports of the enterprises included in this paper.

Control variables

According to financial management theory, liquidity, profitability and security are basic principles to which enterprises must adhere with regard to SCM. Leverage is the ratio of equity capital to total assets on the balance sheet. It measures the operational security and liability risk of a certain enterprise, and it reflects an enterprise’s repayment ability. Furthermore, assets are a basic unit of supply chain operations, playing an important role in SCM. From the perspective of transaction cost theory, increasing asset size may affect costs and SCC. Hence, we included total asset size (Size), financial leverage ratios (Lev) and a dummy variable for year as control variables. The definitions of our variables are listed in Table 3.

Table 3.

Definition of main variables.

Type of variables	Name of variables	Notations	Definitions of notations
Dependent variables	Supplier concentration		Logarithm of “the ratio of top five suppliers’ total purchase amount to total annual purchase amount”
	Customer concentration	CC	Logarithm of “the ratio of top five customers’ total sales amount to total annual sales amount”
	Supply chain concentration	SCC	Logarithm of “the ratio of top five customers’ total sales amount to total annual sales amount” * “the ratio of top five suppliers’ total purchase amount to total annual purchase amount”
Independent variable	Blockchain technology implementation		The year in which enterprises issued the BT layout announcement and the following two years are identified as implementation-related years, where the value of is 1 and the value of in other years is 0.
Moderating variables	Listed-years		Number of years from the year of IPO to the year t
	R&D investment	R&D	Logarithm of the R&D investment
Control variables	Financial leverage ratios		(Total assets/total liabilities) * 100%
	Firm size		Logarithm of total asset

Regression analysis

In our model test, the hypotheses were tested using the OLS method as the basic regression method. OLS was deemed particularly appropriate for our research since it is a well-established method in IS/IT⁶⁹, and supply chain research has supported the quantitative and reliable estimation of the causal relationships among independent and dependent variables simultaneously⁶⁷.

Before the regression analysis was conducted, the Pearson’s correlation coefficients and variance inflation factors (VIFs) of our variables were calculated; these values are shown in Table 4. None of the correlation coefficients were greater than 0.5. Moreover, all the VIFs were between 1 and 5, i.e., below the recommended maximum threshold of 10, thus indicating low multicollinearity⁷⁰. In addition, to ensure the consistency and validity of the model estimation, the following treatments were applied to the sample data before the specific tests were conducted. (1) To prevent outliers from having significant effects on the test results, we winsorized the core continuous variables included in the model at the 1% level. (2) To exclude possible problems such as heteroskedasticity in the panel data, robust standard errors were used in the estimation.

Table 4.

Pearson’s correlation coefficients and VIFs of the variables.

Construct	SC	CC	BT	Asset	Lev	Age	Industry	R&D
SC	1
CC	0.252***	1
BT	0.135***	0.231***	1
Size	0.017	−0.098***	0.101***	1
Lev	−0.010	−0.069***	0.037	0.447***	1
Age	0.159***	−0.044	0.143***	0.392***	0.304***	1
Industry	0.073**	0.150***	−0.038	−0.068**	0.103***	−0.023	1
R&D	0.011	−0.028	−0.004	0.043	0.017	0.117***	−0.016	1
VIF			1.04	4.96	1.33	1.29	1.09	1.03

 ***, ** and * in the table indicate p < 0.01, p < 0.05 and p < 0.1, respectively

FsQCA

This paper focuses on SCC as an outcome of the systemic structural arrangement of our focal theoretical constructs—BT application, three intrafirm factors, and one extrafirm factor. Since conventional correlation-based quantitative methods cannot explain the causal complexity among the variables, it is infeasible to infer such systematic arrangements among multiple elements using this approach^30,33. Therefore, we conducted a fsQCA, a set-theoretic configurational approach that enables us to explore how multiple elements can be combined into multiple configurations to produce SCC. A particular advantage of the fsQCA approach is that it essentially achieves equifinality, which assumes that various systemic arrangements of constructs may lead equally to the same result^30,31. Hence, fsQCA allows us to analyze how enterprises can achieve SCC via multiple configurations. In recent years, as a qualitative‒quantitative approach, fsQCA has been viewed as a complementary method to traditional quantitative analysis and has been employed several times to complement the findings of research models that were initially analyzed using MRA and SEM^30–32. Combining these methods helps scholars overcome the oversimplified nature of the hypotheses tested using regression approaches and obtain sufficiently unique and new findings regarding the complex problems under analysis^30–32. Additionally, in IT/IS research, fsQCA has been increasingly used to capture the complex nature of various digital phenomena⁷¹. Thus, a fsQCA is used in this paper to investigate the mechanisms underlying enterprises’ achievements of high SCC levels during the application of BT that were not revealed using an MRA.

As a technique, fsQCA has also been the subject of development, notably in terms of its appropriate use with respect to panel data^33,72. This development specifically acknowledges the inherent structure of panel data and proposes a new suite of general descriptive measures for evaluating set-theoretic relationships for such panel data^33,72. In this context, the focus is on consistency-oriented developments (see Appendix B), namely, pooled consistency (POCONS), between consistency (BECONS), and within consistency (WICONS)⁷³. These measures offer a promising approach to the task of incorporating time into the fsQCA to facilitate “longitudinal set-theoretic research.” To our knowledge, this study represents the first attempt to employ these measures with a large enterprise dataset.

To perform the fsQCA, precalibration with regard to the considered conditions and outcomes is needed^30,31. This method calibrates (transforms) the considered conditions and outcomes, which were originally scored on their own scales, to fuzzy sets ranging from 0 to 1³⁰. Our calibration process followed the direct method proposed by Ragin²⁷, specifically the developed version described by Beynon, et al⁷⁴.. To ensure that no samples were removed as the fuzzy sets were calculated, we added 0.001 to the column conditions with a membership score of 0.5 in the fuzzy sets.

Results

Basic regression analysis

To test the influence of BT implementation on SCC, on the basis of the model construction and variable selection, we performed a comprehensive test of two aspects: SC and CC.

We identified the appropriate form for the panel model using the Hausman test. Afterward, we analyzed the impact of BT implementation on SCC on the basis of the random effect/fixed effect model. Specific regression outcomes are presented in Table 5. The data in Table 5 indicate that the BT regression coefficients were positive and significant at the 1% level, which reflects the fact that BT implementation had positive effects on both SC and CC. These conclusions remained robust in Models (2) and (3) after the addition of the moderating variables. Hence, Hypotheses 1 and 2 are supported.

Table 5.

Impact of BT implementation on SC/CC.

Variables	Impact on SC			Impact on CC
	(1)	(2)	(3)	(1)	(2)	(3)
BT	0.232*** (0.057)	0.153*** (0.045)	0.193*** (0.048)	0.269*** (0.046)	0.246*** (0.064)	0.272*** (0.045)
Age		0.004 (0.007)			−0.012 (0.021)
R&D			−0.029 (0.027)			−0.025 (0.024)
BT x Age		0.032*** (0.010)			0.026*** (0.008)
BT x R&D			0.085*** (0.033)			0.088*** (0.029)
Asset	−0.130* (0.078)	−0.055 (0.040)	−0.027 (0.038)	−0.037 (0.029)	0.005 (0.058)	−0.036 (0.029)
Lev	−0.002 (0.002)	−0.001 (0.002)	−0.001 (0.002)	−0.001 (0.002)	−0.001 (0.002)	−0.001 (0.002)
Year	Yes	Yes	Yes	Yes	Yes	Yes
_cons	6.305*** (1.730)	4.590*** (0.851)	4.509*** (0.888)	3.855*** (0.627)	3.019*** (1.223)	4.297*** (0.799)
N	735	735	735	735	735	735
R-squared	0.055	0.102	0.063	0.082	0.116	0.103
F	5.67***	-	-	-	7.58***	-
Wald chi2	-	27.38***	18.92**	35.35***	-	45.37***
Hausman	8.44*	10.41	8.64	1.68	10.87*	6.23

(1) ***, ** and * in the table indicate p < 0.01, p < 0.05 and p < 0.1,respectively

Moderation analysis

Moderating effects of listed years and R&D investment

Table 5 presents an analysis of how a company’s number of years in the market (Age) and R&D investment (R&D) affect the correlation between BT implementation and SC/CC. In Model (2), the regression coefficients for the interaction term (BT × age) indicate a significant and positive relationship at the 1% level, signifying that the duration of a company’s listing positively moderates the associations between BT implementation and SC/CC. These findings support Hypotheses 3 and 4 proposed by this study. Similarly, Model (3) examines the influence of R&D investment on the correlation between BT implementation and SC/CC. The regression coefficients for the interaction term (BT x R&D) are positive and statistically significant at the 1% level, indicating that R&D investment favorably moderates the relationships between BT implementation and SC/CC, thus affirming Hypotheses 5 and 6 proposed by this study.

Industry heterogeneity test

To investigate the possibility of industry differences in the relationship between BT implementation and SC/CC, we divided the sample enterprises on the basis of whether they were operating in the manufacturing industry and then performed a group regression. The specific regression outcomes are presented in Table 6. In Table 6, Column (1) shows the regression results for manufacturing enterprises, while Column (2) shows the regression results for nonmanufacturing enterprises. The results show that although the coefficient of BT was significantly positive in all groups of samples (coef. = 0.136, p < 0.1; coef. = 0.274, p < 0.01; coef. = 0.228, p < 0.01; coef. = 0.264, p < 0.01), the coefficients of BT were greater for nonmanufacturing firms. These findings indicate that the implementation of BT by nonmanufacturing enterprises enhanced their SC and CC to a greater extent. Hence, Hypotheses 7 and 8 proposed by this paper are supported.

Table 6.

Industry heterogeneity test of the relationship between BT implementation and SC/CC.

Variables	SC		CC
	Manufacturing	Non-Manufacturing	Manufacturing	Non-Manufacturing
BT	0.136*	0.274***	0.228***	0.264***
	(0.076)	(0.074)	(0.077)	(0.062)
Asset	−0.147	−0.130	0.049	−0.005*
	(0.082)	(0.099)	(0.096)	(0.062)
Lev	−0.003	−0.001	−0.002	−0.001
	(0.003)	(0.003)	(0.004)	(0.003)
Year	Yes	Yes	Yes	Yes
_cons	6.830***	6.213***	2.141	3.055**
	(1.748)	(2.228)	(2.110)	(1.368)
N	255	480	255	480
F	3.19**	5.00**	3.58**	6.48***
R2	0.035	0.068	0.086	0.084

***, ** and * in the table indicate p < 0.01, p < 0.05 and p < 0.1, respectively

Robustness tests for the regression analysis

Change the regression analysis method

To test the robustness of the regression analysis results reported above in further detail, we retested the link between BT implementation and SC/CC using maximum likelihood estimation⁶⁷; the results are presented in Table 7. Our conclusions were thereby preliminarily proven to be robust.

Table 7.

Robustness test on the relationship between BT implementation and SC/CC.

Construct	SC			CC
	Coef	Std.Err	t-value	Coef	Std.Err	t-value
BT	0.160***	0.041	3.91	0.281***	0.039	7.24
Asset	0.006	0.019	0.34	−0.052***	0.018	−2.91
Lev	−0.001	0.001	−0.53	−0.001	0.001	−0.79
Year	Yes	Yes	Yes	Yes	Yes	Yes
_cons	3.252***	0.405	8.02	4.12***	0.384	10.92
N	735	735	735	735	735	735
Log likelihood	−739.984			−694.431
LR chi2		15.87***			61.66***

***, ** and * in the table indicate p < 0.01, p < 0.05 and p < 0.1, respectively

Change the BT implementation measurements

Previous studies have revealed a time lag in the effectiveness of BT implementation⁶⁷; namely, it usually takes 1–2 years from BT implementation to real implementation. In the preceding analysis, the years in which the enterprises issued the BT layout announcement and the subsequent two years were identified as implementation-related years, in which context the value of BT_it was 1; in contrast, the value of BT_it in other years was 0. However, the time lag for the effects of BT implementation in reality can be much longer, potentially 3 years or more. Therefore, to examine the stability of the impact of BT implementation on SCC in further detail, we remeasured the BT implementation variable by identifying the year in which the enterprises issued a BT layout announcement and the subsequent three years as implementation-related years. The regression results shown in Tables 8 and 9 indicate that the main findings of this paper remain robust after the measurement of BT implementation is changed.

Table 8.

Robustness test on the relationship between BT implementation and SC/CC.

Variables	SC			CC
	(1)	(2)	(3)	(1)	(2)	(3)
BT	0.191*** (0.046)	0.157*** (0.042)	0.193*** (0.046)	0.247*** (0.044)	0.186*** (0.049)	0.227*** (0.045)
Age		0.006 (0.007)			0.011 (0.017)
R&D			−0.015 (0.026)			0.002 (0.026)
BT x Age		0.032*** (0.009)			0.027*** (0.007)
BT x R&D			0.0611** (0.033)			0.078** (0.031)
Asset	−0.022 (0.038)	−0.054 (0.040)	−0.023 (0.038)	−0.029 (0.029)	0.005 (0.056)	0.041 (0.051)
Lev	−0.001 (0.002)	−0.001 (0.002)	−0.001 (0.002)	−0.001 (0.002)	−0.001 (0.002)	−0.001 (0.002)
Year	Yes	Yes	Yes	Yes	Yes	Yes
_cons	3.891*** (0.821)	4.547*** (0.849)	4.184*** (0.889)	3.712*** (0.626)	2.949** (1.177)	2.096* (1.214)
N	735	735	735	735	735	735
R-squared	0.050	0.108	0.058	0.074	0.117	0.096
F	-	-	-	-	7.82***	8.29***
Wald chi2	17.46***	31.85***	18.33***	32.04***	-	-
Hausman	6.94	9.90	6.56	3.32	13.86**	17.17***

***, ** and * in the table indicate p < 0.01, p < 0.05 and p < 0.1, respectively

Table 9.

Industry heterogeneity test on the relationship between BT implementation and SC/CC.

Variables	SC		CC
	Manufacturing	Non-Manufacturing	Manufacturing	Non-Manufacturing
BA	0.126*	0.264***	0.194**	0.242***
	(0.072)	(0.066)	(0.075)	(0.058)
Asset	−0.140	−0.112	0.069	0.017
	(0.082)	(0.098)	(0.092)	(0.061)
Lev	−0.003	−0.001	−0.002	−0.001
	(0.003)	(0.003)	(0.004)	(0.003)
Year	Yes	Yes	Yes	Yes
_cons	6.677***	5.831***	1.712	2.588*
	(1.760)	(2.206)	(2.019)	(1.349)
N	255	480	255	480
F	2.90**	5.63***	2.87**	6.32***
R2	0.034	0.071	0.071	0.080

***, ** and * in the table indicate p < 0.01, p < 0.05 and p < 0.1, respectively

Reanalysis of the data using FsQCA

As part of this research, the initial fsQCA reanalysis employs a standard iterative method, including sufficiency and necessity analyses. These analyses are described as follows:

Necessity analysis

The necessary condition analysis associated with the fsQCA method is a separate step that involves examining whether one or more conditions are necessary or mostly necessary for a certain outcome to occur²⁷. Given the asymmetric characteristics of the fsQCA, both outcomes (high SCC and low SCC) were considered.

In light of the standard threshold of 0.9³³, Table 10 indicates that no single conditional variable, whether in relation to high-SCC or low-SCC outcomes, is necessary with regard to the associations between the cases (enterprise-year observations) and the SCC.

Table 10.

Analysis of necessity for SCC (High-SCC and Low-SCC).

Condition		Outcome-SCC
		High-SCC		Low-SCC
		Consistency	Coverage	Consistency	Coverage
Lev	High	0.648	0.576	0.640	0.711
	Low	0.675	0.600	0.618	0.687
Age	High	0.647	0.573	0.653	0.724
	Low	0.688	0.613	0.614	0.685
Size	High	0.608	0.576	0.655	0.777
	Low	0.764	0.639	0.642	0.672
BT3.0	High	0.342	0.442	0.345	0.557
	Low	0.658	0.445	0.655	0.555
Industry	High	0.363	0.529	0.259	0.471
	Low	0.637	0.407	0.741	0.593
R&D	High	0.640	0.583	0.656	0.747
	Low	0.722	0.627	0.633	0.687
Statistics	Min	0.342	0.407	0.259	0.471
	Max	0.764	0.639	0.741	0.777

Sufficiency analysis

The goal of a sufficiency analysis is to identify the various combinations of conditions that satisfy the specific sufficiency criteria for the occurrence of a certain outcome^27,28. For the sufficiency condition to hold for an enterprise–year observation, the membership scores associated with outcomes should always be higher than those associated with the combination of the conditions under consideration³³.

Following the application of fsQCA analysis, three distinctive solutions with varying degrees of simplification emerged: a complex solution, a parsimonious solution, and an intermediate solution. Each solution is characterized by differences in complexity, the capacity for revelatory insight, and generalizability. Although it is the most rigorous, the complex solution yields conclusions that are relatively intricate, thereby limiting its generalizability²⁸. Conversely, the parsimonious solution, although less rigorous, tends to offer excessively simplistic conclusions, which may not accurately mirror the actual situation²⁸, thereby inhibiting its revelatory potential. Striking a balance between these extremes, the intermediate solution merges theoretical insights with case analyses, yielding conclusions with enhanced revelatory capacity and generalizability²⁸. Consequently, the majority of scholars employing the fsQCA methodology choose the intermediate solution^27,30. Accordingly, only intermediate solutions are reported in this manuscript.

Sufficiency analyses were performed with respect to each outcome (High-TEA and Low-TEA) to identify various combinations of conditions, i.e., “causal recipes,” that connect the configurations with specific outcomes⁷¹. The results of our sufficiency analyses are presented in Table 11. Four (CHT1–CHT4) and six (CLT1–CLT6) causal recipes were identified to facilitate our understanding of the high-SCC and low-SCC conditions, respectively. These causal recipes incorporate core conditions based on a parsimonious solution derived from counterfactual analysis²⁸, which have relatively robust causal associations with the outcomes^27,28. Additionally, the causal recipes incorporate peripheral conditions that fall outside the parsimonious solutions and exhibit relatively weaker associations with the SCC than the core conditions do²⁸. Each configuration provides a comprehensive picture of the interactions among the conditions required to produce SCC and highlights the role of each condition as peripheral or core (or as absent, present, or “does not matter”)³³. Consequently, these findings elucidate the multiple roles played by BT applications, which prove to be essential for achieving SCC in certain configurations but might be irrelevant or even detrimental in other configurations.

Table 11.

Sufficiency analysis of Lev, Age, Size, BT3.0, Industry, and R&D conditions with High-SCC and Low-SCC outcomes.

Conditions	SCC
	High				Low
	C1	C2a	C2b	C3	C1	C2	C3	C4	C5	C6
Blockchain usage
BT3.0	⊗	⬤	⬤	⬤					⬤	⬤
Intra-firm
Age	⬤		⬤	⊗			⬤		⊗	⬤
Lev		⬤	⬤	⬤		⊗		⬤	⬤	⊗
R&D	⊗	⊗		⬤	⊗		⬤	⬤
Size	⊗	⊗	⊗	⬤	⬤	⬤		⊗
Extra-firm
Industry	⬤	⬤	⬤	⬤	⊗	⊗	⊗	⊗	⊗	⊗
Consistency	0.911	0.907	0.928	0.936	0.845	0.864	0.838	0.888	0.868	0.840
Raw coverage	0.098	0.062	0.056	0.041	0.339	0.319	0.350	0.239	0.097	0.119
Unique coverage	0.098	0.009	0.004	0.005	0.048	0.013	0.065	0.017	0.006	0.003
Solution consistency	0.909				0.812
Solution coverage	0.170				0.540

black circles (⬤) indicate presence; white circles (⊗) denote negation; blank spaces denote absence; small and large circles respectively denote peripheral and core conditions. *This analysis is based on the Intermediate Solution

Table 11 uses a circle notation adapted from Fiss²⁸. Each configuration exhibits a consistency greater than 0.8, thus corroborating the claim that the solutions reliably engender high SCC/low SCC^27,28. In addition to individual configuration consistency, Table 11 presents the aggregate solution consistency (0.909 and 0.812), reflecting the reliability with which all the configurations collectively give rise to a high-SCC/low-SCC situation²⁷. Another key measure, i.e., raw coverage, indicates the proportion of outcome-related cases that each configuration encompasses, while unique coverage illustrates the degree to which each configuration uniquely (i.e., without any overlap with other causal recipes) accounts for the high-SCC/low-SCC enterprise²⁷, thus paralleling the R-squared value in a MRA.

To analyze the four outcomes of fsQCA regarding the attainment of high SCC, we examine them on the basis of discernible patterns identified by comparing the divergences and parallels among these configurations. Specifically, we adopted Fiss²⁸ conceptual interpretations of peripheral and core conditions to structure the results, categorizing the combinations of conditions associated with the intermediate solution by reference to the core condition. Consequently, we derived three first-order or cross-type equifinal configurations (C1–C3), each of which encompassed a unique set of core conditions leading to the same outcome. Notably, the C2 type encompassed two second-order or within-type equifinal configurations (C2a and C2b), indicating variations within the C2 type. This finding implies that multiple recipes, although they may share the same core condition (i.e., the utilization ofBT, intrafirm elements, and extrafirm elements) and harbor differing peripheral conditions, can yield identical outcomes. In addition, the core conditions for the C2 and C3 types—BTutilization, intrafirm elements, and extrafirm elements—suggest that they can be viewed as similar first-order types. From our perspective, these configurations exemplify two primary patterns (i.e., intrafirm and extrafirm solutions as well as BT-centric solutions) by which high SCC levels can be attained. We propose a contingency framework that facilitates a more systematic integration of our key findings, effectively elucidating the interaction between BT application and high SCC, as illustrated in Fig. 3.

Fig. 3.

Configurational Solutions for Achieving High-SCC: From A Contingency Perspective.

• Pattern 1— “BT-driven solution” (configurational solutions to achieve high SCC levels with BT).

The “BT-driven solution”, as depicted in configurational structures C2a, C2b, and C3 (as referenced in the Table 11 and the Fig. 3), provides a method that firms can use to achieve superior supply chain collaboration (SCC) levels by integrating BT under both intrafirm and extrafirm conditions. This strategy is equally effective for small-scale and large-scale firms. In this section, the theoretical mechanisms underlying this approach with regard to these differing firm sizes are explored.

Our initial focus is on the configurational structure C4 because of the larger size of the cases examined in this context. As a firm’s size increases, so does its complexity, which consequently impacts SCC. Large-scale firms often feature several processes spanning various domains as well as business units that tend to be compartmentalized, thus posing a challenge to seamless information flow across the organization and its supply chain. This scenario makes cross-enterprise business unit integration a challenging task⁷¹. Newly listed large firms, which must confront the inherent complexities resulting from their substantial size, must urgently overcome these obstacles. Concurrently, high leverage, which is synonymous with reduced operational security and heightened liability risk, intensifies trust crises, making it challenging for these firms to cultivate close relationships with other entities both upstream and downstream in the supply chain.

Implementing BT provides a solution to these challenges, facilitating the attainment of superior SCC levels. The distinct BT attributes of decentralization, negation of trust, security, traceability, and intelligence guarantee that the chain data remain unaltered and mutually accepted⁹. Hence, BT-aided firms can achieve full traceability and share information with their upstream and downstream collaborators¹⁷, thereby overcoming the natural complexities associated with large-scale firms and establishing a “decentralized trust structure” among supply chain stakeholders. However, substantial research and development (R&D) investments are often prerequisites for this cutting-edge technology²⁶, making the “BT-driven solution” costly for large-scale firms.

The strategic configurations employed by small-scale enterprises to achieve high SCC levels differ from those employed by their larger counterparts. Configurations C2a and C2b suggest that the theoretical rationale that explains the role of BT implementation in mitigating trust crises due to high leverage remains equally valid for smaller enterprises. However, for small firms, R&D investment often plays a secondary role, largely because of the resource constraints they face. These firms typically lack the means to independently implement BT and R&D, compelling them to seek assistance from third-party entities that offer BT services at a lower cost.

The C2b configuration is similar to C2a; however, in this case, the firm’s age compensates for the lack of R&D investment. Specifically, long-established small enterprises have accumulated substantial resources in terms of capital, technology, and experience, thus providing them with a solid foundation for BT implementation. These firms can therefore mitigate the trust crisis arising from high leverage through BT implementation irrespective of their R&D investment.

• Pattern 2—“Intrafirm & extrafirm solution” (An intrafirm & extrafirm configuration solution for achieving high SCC levels).

Configuration C1, as displayed in Table 11 (in the column on the left) and Fig. 3 (in the column on the right), illustrates the potential for the attainment of high SCC through a judicious amalgamation of intrafirm and extrafirm conditions, hereinafter referred to as the “intrafirm & extrafirm solution.” This solution emphasizes the lack of necessity of BT implementation, highlighting the potential detriments of such technology for some small-scale enterprises in their attempts to achieve high SCC levels.

This configuration lends further weight to the key role of “age” within the high-SSC framework, reinforcing the applicability of the theoretical logic previously utilized in the discussion of age. Enterprises with a significant lifespan enjoy abundant resources in the form of capital, technology, and experience, and they command unassailable trust from their upstream/downstream collaborators. This inherent trust diminishes their reliance on BT for fostering such relationships, making it a nonessential factor in their pursuit of elevated SCC levels. Additionally, the significant costs associated with BT implementation may prove counterproductive for these smaller entities.

Furthermore, our examination of all four (CHT1–CHT4) and six (CLT1–CLT6) causal formulae, which aimed to explore high and low SCC levels, consistently highlights the pivotal role of industry type (manufacturing vs. nonmanufacturing) in determining an enterprise’s SCC level. This observation is in line with our prior analytical findings. Specifically, service-centric nonmanufacturing enterprises often face greater challenges than their manufacturing counterparts with respect to exerting control over supply chain nodes, predominantly because of insufficient backing from robust information systems.

These nonmanufacturing entities face more significant issues pertaining to inefficient supply and sales information flow as well as the lack of unified system standards with their collaborators. Therefore, inherent industry advantages and shortcomings inevitably influence an enterprise’s SCC. Consequently, the “industry” factor emerges as a crucial condition in configurations linked to high SCC levels (as displayed in Fig. 2), and its absence is a hallmark of configurations associated with low SCC levels.

Panel data breakdown of the FsQCA results

The analysis of the previous fsQCA results is extended in this section. We aim to use panel data to analyze our fsQCA results across different enterprises and years. In so doing, we employ the technique proposed by Guedes, et al⁷³. and Garcia-Castro, et al⁷⁵.; brief technical details concerning this technique are as follows.

All three measures considered here involve consistency formulae and are used to test multiple subsets of the enterprise–year dataset. POCONS, the first measure, includes all the observations. Thus, this measure corresponds to the previously reported consistency values that were related to each causal recipe (see also Table 11)^33,73,75. BECONS, the second measure, focuses on observations of a certain year^33,73,75. WICONS, the third measure, consists of certain enterprise observations^33,73,75. Analyses utilizing WICONS and BECONS are detailed for the causal recipes of the complex solutions that were related to each outcome (high and low SCC levels).

Table 11 shows the four causal recipes (CHT1–CHT4) associated with high SCC levels and the six causal recipes (CHT1–CHT6) associated with low SCC levels. The BECONS diagrams corresponding to these recipes are plotted in Figs. 4 and 5. For each causal recipe, a specific BECONS value was calculated for each year considered here.

Fig. 4.

BECONS values (2019–2023) for causal recipes CHT1 - CHT4 (High-SCC outcome).

Fig. 5.

BECONS values (2019–2023) for causal recipes CHT1 – CHT6 (Low-SCC outcome).

Our BECONS results for CHT1-CHT4 and CHT1-CHT6 are reported for each individual year included here, regardless of how many enterprise–year observations pertained to each year (see Figs. 4 and 5). The POCONS values corresponding to the recipes are displayed both numerically and as horizontal dashed lines.

The four BECONS diagrams for high SCC levels and the six BECONS diagrams for low SCC levels exhibit several similarities. Given the similarities in the causal recipes thus obtained, such observations can be easily understood. Figure 4 shows the specific years in which the BECONS values for CHT1-CHT4 differed: 2021 (CHT1 > CHT4 > CHT2 = CHT3) and 2023 (CHT4 > CHT3 > CHT2 > CHT1). These differences may be associated with the response to theBT–related policies introduced, in which context, especially in multiple low-SCC-level enterprises, the associated recipes fluctuated before several enterprises approached (or returned to) the recipes pertaining to the prior BT policies. The data in Fig. 5 indicate that CLT2 was highly consistent from 2019 to 2020. This consistency decreased over time until 2021 (as it does for CLT3 and CLT4). This trend was in contrast to that observed for CLT1.

The WICONS results focused on the individual enterprise level. The consistency results thus obtained reflect the consistency of a certain causal recipe for a given enterprise. The corresponding WICONS values for all 147 enterprises are presented in Figs. 6 and 7 (for CHT1-CHT4 and CHT1-CHT6).

Fig. 6.

WICONS values for all enterprises for causal recipes CHT1 – CHT4 (High-SCC outcome).

Fig. 7.

WICONS values for all enterprises for causal recipes CHT1 – CHT4 (Low-SCC outcome).

In each diagram of Fig. 6, the cross axis represents the enterprises’ rank of the WICONS values. This rank was ordered according to each of the four causal recipes; thus, a distinct ranking was obtained in each case. The POCONS values corresponding to the recipes are displayed both numerically and as horizontal dashed lines. In each diagram, for a specific enterprise, four WICONS values are presented (i.e., for CHT1–CHT4) to facilitate comparisons among causal recipes.

The first interesting discovery pertained to some inconsistent results (i.e., WICONS < 1 within the rank-ordered values) concerning approximately 40 enterprises for CHT1, approximately 55 enterprises for CHT2, approximately 50 enterprises for CHT3 and approximately 60 enterprises for CHT4. The results shown on the right-hand side of the diagram indicate that enterprises did not consistently adhere to a particular causal recipe across the years covered by the dataset containing 735 enterprise–year observations. Overall, this finding shows that approximately 100 enterprises exhibited strong consistency. However, the other enterprises exhibited varying degrees of inconsistency across the years included in our analysis.

As mentioned, Fig. 7 shows that approximately 50 enterprises exhibited strong consistency in terms of the causal recipes to which they are related. Alternatively, the other enterprises exhibited varying degrees of inconsistency across the years included in our analysis. Specifically, as in the case of our analyses of high SCC levels, for every low SCC level, 90 to 100 enterprises exhibited WICONS consistency. In summary, the technical WICONS and BECONS consistency measures reveal nuances at the enterprise and year levels, confirming the robustness of the configurations we obtained.

Discussion

Findings

This research explores how the application of BT affects enterprise SCC; highlights the moderating roles of R&D investment, listed years, and industry heterogeneity; and identifies combinations of intra/extrafirm factors that lead to various SCC levels. Through both MRA and fsQCA, all the hypotheses were supported.

MRA outcomes reveal that BT can serve as a valuable instrument to promote capability building among enterprises, positively influencing the two dimensions of SCC: SC and CC (H1 and H2, respectively). These findings emphasize the significant role that BT implementation (a high-level functional capability associated with specialized capabilities) plays in the attainment of SCC (a lower-level functional capability associated with operation-related capabilities). This inference is in line with our conceptualization of the hierarchy of capabilities and can be explained to some degree by transaction cost theory. The corresponding improvements in SC and CC could be attributed to the decentralized trust building and cost reduction inherent in BT, a claim that is in line with multiple recent studies.

The incorporation of R&D investment and listed years into the moderation analysis revealed their positive effects on the effects of BT on SC and CC. Additionally, the ability of BT implementation to enhance SC and CC was more notable for nonmanufacturing enterprises, thus emphasizing the inherent advantages and disadvantages of different industry types with regard to achieving SCC.

To improve our understanding of how various elements coalesce into multiple configurations to promote SCC, we categorized the indicators utilized for the MRA into three groups (BT implementation, three intrafirm factors, and one extrafirm factor) and conducted a fsQCA analysis on the same dataset. The fsQCA results indicated four (CHT1–CHT4) and six (CLT1–CLT6) causal recipes that could explain high and low SCC levels, respectively. These configurations, which were associated with high SCC levels and represented two primary patterns (the “intrafirm & extrafirm solution” and the “BT-driven solution”), identified pathways for achieving high SCC levels. A contingency framework is thus proposed to systematically integrate our key findings, effectively illuminating the relationships between BT implementation and high SCC levels.

On the basis of this proposed framework, we examined the underlying theoretical mechanisms for the two patterns within small and large enterprises. Overall, the fsQCA results offered intriguing findings, which refined and fortified the results of the MRA. For instance, in line with H1 and H2, the “BT-driven solution” suggested that BT implementation is a core condition for achieving high SCC levels. However, the “intrafirm & extrafirm solution” demonstrated that high SCC levels could be achieved even without BT implementation, a situation that is applicable to small manufacturing enterprises that have been listed for a relatively short time and that face high financial risk and low R&D investment.

Implications for theory and research

This research contributes significantly to the literature on BT and SCC by introducing several distinct advances. First, we enhance the hierarchical organizational capability perspective to explore the mechanism underlying the influence of BT on supplier and CC. By examining capability-building processes and the corresponding inherent functional capability hierarchies, we develop a dual-layered analytical framework. This approach provides deeper insights into the origin of capability-building and elucidates BT’s role in generating value with respect to SCM.

Second, this research highlights the context-specific value of BT by examining how intraorganizational elements (such as R&D and listed years) and external environmental influences (such as industry type) modulate the BT-SCC link. Our findings demonstrate how these factors influence firms’ relationships with concentrated suppliers and customers, thereby strengthening the interrelation between BT and SCC.

Third, we extend our research beyond the contingency perspective, which typically focuses on multiregression analysis (MRA), by incorporating a novel longitudinal study method called (fsQCA within the domain of SCM. Using this approach, we identify a series of equifinal configurations and two patterns associated with a high SCC level from a configurational perspective. Although the use of fsQCA itself is not groundbreaking, its application within this specific context provides valuable insights and facilitates the generation of a framework for practical recommendations. This framework empowers organizations of various sizes to develop unique and effective strategies for maximizing the benefits of SCC by adopting different BT strategies.

In conclusion, this research significantly advances the understanding of the relationship between BT and SCC by employing a hierarchical organizational capability perspective, examining context-specific influences, and incorporating a longitudinal fsQCA study. These contributions contribute to the existing knowledge in the field and provide organizations with actionable insights for leveraging BT to increase SCC.

Implications for practice

This research has valuable practical implications for industry stakeholders that are aiming to increase SCC through BT.

First, managers, particularly those from long-listed enterprises, should strategically integrate BT into supply chain operations to strengthen supplier and CC. Our findings indicate that moderate increases in R&D investment amplify the positive impact of BT on SCC. Accordingly, enterprises are advised to allocate dedicated innovation budgets to blockchain-related supply chain initiatives to ensure efficient implementation of BT.

Second, industry type should guide BT adoption strategies. Non-manufacturing enterprises, such as those in logistics, healthcare, or software services, are more likely to benefit from BT. Managers in these industries can prioritize BT for improving information transparency and traceability across service networks. Conversely, for manufacturing firms, BT implementation should be accompanied by comprehensive risk assessments to ensure successful digital integration.

Third, resource-constrained small manufacturing enterprises should adopt a staged or partnership-based BT implementation method. Instead of full internal development, they can collaborate with external blockchain service providers or industry consortia to lower implementation costs while maintaining trust and data traceability.

Fourth, the study’s configurational analysis identifies two viable strategic pathways to achieve high SCC: the “BT-driven solution” and the “intrafirm & extrafirm solution”. Managers should evaluate their firms’ internal conditions (R&D capacity, listing years, financial leverage) and external environment (industry type, policy support) before choosing which pathway to follow. In general, enterprises facing high transaction risk or low inter-firm trust should prioritize the technology-driven route, while those with mature, long-standing supply chain relationships may achieve high SCC through conventional coordination mechanisms without heavy BT investment.

Finally, the causal recipes and patterns that correspond to high and low SCC levels are not immutable and must be adjusted in a timely manner according to new policies. Therefore, managers should monitor national and regional BT-related policies and adjust their BT implement strategies accordingly. Specifically, they can establish an internal “policy intelligence” function or participating in blockchain-related industry alliances.

Limitations and directions for future research

This investigation, while comprehensive, has certain inherent limitations. First, the study’s data collection was limited to Chinese A-share enterprises employing BT. This choice inherently restricts the generalizability of the study because of the influence of environmental and cultural factors on trust, SCC, and the interaction between BT and SCC. Future research could therefore replicate this study in cross-country settings, such as emerging economies in Africa and Latin America or mature markets in Europe and North America, to examine the boundary conditions of BT’s impact.

Second, although multiple robustness checks were conducted to validate our findings, potential endogeneity and omitted variable issues cannot be completely addressed. Future study could employ instrumental variable approaches, difference-in-differences (DID) models, propensity score matching (PSM), or natural experiments to further strengthen causal inference.

Third, the paper identifies two distinct strategies for attaining high SCC levels: the “BT-driven solution” and the “intrafirm & extrafirm solution”. Enterprises may employ these patterns sequentially rather than concurrently, resulting in a transition over time. Future work could (1) longitudinally track firms that adopted these solutions and uncover how their strategies evolve over time; (2) explore how external policy interventions or supply chain disruptions (e.g., trade shocks, digital regulation) trigger transformation between the two strategies.

Conclusions

This paper explored how BT implementation affects the SCC of an enterprise; how R&D investment, listed years, and industry type moderate the relationship between BT and SCC; and the combinations of BT implementation with intrafirm and interfirm factors that lead to high/low SCC. On the basis of hierarchical organizational capability theory, we propose a set of hypotheses. The MRA results supported all of the hypotheses and revealed some interesting findings. In addition, the results of the fsQCA conducted using the same dataset complemented our MRA results, identifying several equifinal configurations and highlighting two patterns associated with a high SCC level. Moreover, the technical WICONS and BECONS consistency measures provided nuanced insights at the enterprise and year levels, confirming the robustness of the configurations we obtained. Overall, this study contributes to the literature by revealing the combinations of and links among the elements that lead to high SCC levels in the supply chain blockchain context. It also suggests ways in which practitioners and managers can achieve an SCC advantage. Hopefully, these contributions can encourage scholars to research in further detail how BT can be utilized to improve SCC and help businesses improve their SCM.

Author contributions

All authors participated in the study, and the research did not involve human experiments or human tissue samples. All methods were carried out in accordance with relevant guidelines and regulations. The specific division of labor is as follows: Jinxin Zhang and Sen Liu conceptualized the study, with Sen Liu also handling funding acquisition, software development alongside Xiaoyan Guo, and resource gathering. Jinxin Zhang was responsible for drafting the original manuscript, while Zhenyu Fan and Linguo Chen both contributed to reviewing and editing the manuscript. Methodologically, the framework was developed by Jinxin Zhang and Sen Liu. Supervision of the project was provided by Xuejian Yang. All authors reviewed the manuscript.

Funding

This work was supported by The National Natural Science Foundation of China (No.72372143); The Yunnan Fundamental Research Key Project (No. 202401AS070020); Yunnan Provincial Science and Technology Research Project of China National Tobacco Corporation (No.2025530000241028); The Yunnan Philosophy and Social Sciences Innovation Team under grant number 2025CX17; and the Yunnan Xingdian Talent Support Program.

Data availability

Data will be made available on request. Requests for access to the data can be directed to the author: ZHANG Jinxin, Lingnan University, at jinxinzhang@ln.hk.

Declarations

Competing interests

The authors declare no competing interests.

Ethical approval

All participants have given written informed consent to participate in the study. This paper is the authors’ own original work, which has not been previously published elsewhere. The paper properly credits the meaningful contributions of co-authors. All authors have been personally and actively involved in substantial work leading to the paper, and will take public responsibility for its content.