Blockchain technology and its impact on sustainable supply chain management in SMEs

Blockchain technology

Several tactics have been implemented to create trust and decrease knowledge asymmetries between manufacturers and consumers. The most prominent among manufacturers who pledge to adhere to certain predetermined provisions is expanding a common label. A common label has been used to assure customers of the particular values of the products they have purchased. The common label has a range of flaws and severe limitations. First, opportunistic methods are listed. Some manufacturers claim to follow the instructions of the common labels, but in fact, they do not follow them. Furthermore, rivals often employ fake tactics, whether by using deceptive labels or complex alternatives, such as tags and pictures from third countries to confuse or mislead customers. Many of these barriers have been overcome by the evolution of digital technologies (Dong et al., 2023). Blockchain technology is aimed at reinforcing trust and creating new dealings between manufacturers and consumers. In the post-pandemic era, blockchain and the Internet of Thing (IoT) are key components of this newest wave. The ability to offer digital identities to every product and component is at the heart of this modern digital revolution. As a result, markets will become more transparent in the post COVID-19 era. The Internet of Things would build marketplaces by finding and balancing supply and demand in real-time. These dynamic digital markets will rapidly extend this transition’s scope by building on the groundwork laid by gadgets and social networks. They will generate peer-to-peer economic models and encourage shared markets.

Blockchain was first used in the application of Bitcoin, the primary decentralized digital currency that removes the middle person from the transaction. Blockchain is also used in many domains due to its benefits in distributed ledger data management, and the possibility of inspection trials. Engineering management, decentralized governance, and health care management are among the sectors that have adopted it since 2014. Blockchain provides peer-to-peer data sharing, immutability, and computational logic. However, the precise method depends on consensus algorithms. There are three phases of blockchain applications. Blockchain 1.0 deals with cryptocurrencies, such as Bitcoin; Blockchain 2.0 incorporates smart contracts for selling processing; and Blockchain 3.0 provides an excessive level of control through an autonomous network based on smart contracts with certain guidelines. Everyone could join the network in permissionless blockchain, such as Ethereum and Bitcoin. It is primarily built on a consensus algorithm known as proof of work, in which every user can certify blocks. While in permissioned networks, such as Hyperledger Fabric, only members can validate the transactions. Every node generates blocks in the proof of authority and proof of stake algorithms. In addition, based on predefined guidelines, the consortium is a mixture of permissioned and permissionless blockchains. One latent use of blockchain is to eliminate the trust-related problems associated with technology, hence cutting costs in the transaction processes. Due to the dynamic structure of supply chains in the digital period, blockchain is intended to increase speed. It would be the primary data source, and connect all supply chain processes. Blockchain applications help bring transparency and increase traceability in the traditional supply chain system. Subsequently, in the post-pandemic era, blockchain adoption could increase cost efficiency, high product security, data sharing, teamwork planning, and operation integration (Dutta et al., 2020).

Blockchain based supply chain system

Malik et al. (2019) conducted research on supply chain challenges in terms of traceability and trust. They proposed a blockchain-based sustainable multi-layer trust management system to inspect exchanges between supply chain users, and vigorously assign trust scores based on these dealings. Subsequently, it lessens overhead regarding latency and throughput when related to a simple distributed ledger technology (DLT)-based supply chain model. Another study by Dobrovnik et al. (2018) explains potential applications for adoption, and provides a framework for classifying blockchain openings in the logistic industries. Their study aids managers in systematically evaluating where to start building organizational capabilities to deploy blockchain effectively. The study of Sahebi, Masoomi & Ghorbani (2020) thoroughly examines the barriers to blockchain technology (BCT) acceptance in the context of humanitarian sustainable supply chain management. The barriers were studied using an integrated methodology that included Fuzzy Delphi and the Best Worst Technique. Regulatory ambiguity, a lack of expertise and high sustainability costs were identified as major barriers. Their findings aid policymakers with important advice to improve their solutions. Kim & Laskowski (2018) discussed blockchain and its potential applications in the ontological engineering of supply chain provenance. Their study focused on potential effects to demonstrate techniques grounded on formal and informal ontology. Their research analyzed a traceability ontology and translated some of its representations into Ethereum-based smart contracts that performed a provenance trace and imposed traceability constraints. Another study by Omran et al. (2017) proposed a blockchain-based sustainable supply chain finance model, with automatic reconciliation and quality identified as the supply chain’s primary value drivers. Blockchain was created with these goals in mind. Their findings contributed to future advancements of suitable supply chain finance solutions based on blockchain revolutions. The study by Tian (2017) developed a hazard analysis blockchain-based sustainable traceability system for food tracing. They also unveiled BigchainDB, a novel model aimed at filling the gap in the distributed ledger system at scale. Their research finishes with a discussion of a use case and the challenges of implementing blockchain in future food supply chain traceability systems. The study by Nakasumi (2017) developed a blockchain-based approach to address sustainable supply chain difficulties, such as information asymmetry. Their research contributes to asymmetric information between the organization, which disrupts planning algorithms. The study of Korpela, Hallikas & Dahlberg (2017) presented a cloud-based solution to solve the gap between commerce readiness and current functionalities. Smart contracts are the most critical for revolutionizing the digital supply chain through blockchain integration, according to their results. The study of Toyoda et al. (2017) presented a blockchain-based product ownership management system for anti-counterfeit goods using radio-frequency identification (RFID). Their findings revealed that normally the expense of handling the ownership of a commodity with up to six transferals is less than US $1. The study by Abeyratne & Monfared (2016) offers a thorough analysis of the current status of blockchain, emphasizing its potential to significantly transform the manufacturing supply chain of the future. The authors investigate various applications of BCT, showcasing its considerable benefits for improving supply chain operations in manufacturing. They outline a progressive vision for manufacturing systems that leverage BCT, using the production of cardboard boxes as a concrete example to illustrate how this technology can be integrated into a global supply chain network. Furthermore, the article addresses the critical requirements and challenges that need to be tackled to successfully implement BCT in future manufacturing initiatives.

Technology adoption models

In the field of industry, technical advancements often play a critical role. Technological advancement also promotes information sharing. But technology is of no use before and until it is approved or used. Although the word adoption is used discretely, the masses should think of diffusion as adoption. The adoption of technology will enable diffusion. So, identifying the adoption of technology is of paramount importance. The study of Carr (1999) described the adoption of technology as the stage of selecting technology that a person or group can use. Adopting technology may also be characterized as the ability within a community of customers to use technology to their advantage. Recent findings have shown that technology implementation is not linked to technical features alone, but has developed as an even more difficult procedure, including users’ attitude and personality dimensions, and making several conditions simpler. Previous research has shown that technology adoption is not solely related to the characteristics of the technology itself, but is a complex process involving customer characteristics, trust, and various facilitating environments. To understand the dynamics influencing the decision to implement blockchain technology in SMEs, several conceptual models based on adoption theories have been developed. These models, such as the technology acceptance model and the theory of planned behavior, have gained popularity in the literature due to their success in determining IT implementation. However, it is important to note that these theories have mainly been developed and tested in Western countries, with limited research conducted in developing nations like Malaysia. Therefore, it is crucial to study the adoption of blockchain technology specifically in the context of SMEs in Malaysia. This research would shed light on the factors influencing adoption, and provide valuable insights for organizations considering the implementation of blockchain technology. The recent literature on blockchain technology by using technology adoption models is presented in Table 1.

Proposed model

The use of blockchain in the supply chain has been researched and written by some researchers. Many of the basic approaches that were deployed involved the technology acceptance model (Kamble, Gunasekaran & Arha, 2019), Technology-Organization-Environment framework (Clohessy, Acton & Rogers, 2019), and the unified theory of acceptance and use of technology (Queiroz & Wamba, 2019). To supplement these prior studies, the present study shows how the technology acceptance model can be integrated with the diffusion of innovation theory and theory of planned behavior for the sustainable supply chain management of SMEs. Accordingly, the adoption of blockchain in development is still nascent, and its market potential has yet to be grasped entirely. Ideally, the diffusion of an invention assessment does not have any apparent connection with technology acceptance model, but mutually share some primary factors. It was uncovered that the relative advantage in the diffusion of innovation theory is the same as the idea of the perceived usefulness in technology acceptance model. The complexity paradigm in the diffusion of innovation theory captures the perceived ease of use in the technology acceptance model, while the sign is reversed (Moore & Benbasat, 1991). In addition, in terms of complexity, technology acceptance model and diffusion of innovation theory indicate that individual purpose is partially dictated by how challenging it is to grasp or practice the technology. Moreover, the less difficult it is to use, the more likely it is for a person to consider it. Compatibility is associated with prior experience with a technology’s performance, whereas the desire to seek observation is related to the availability of appropriate opportunities for experience. These factors relate to previous experience with technology, or opportunities to experience the technology being considered. Compatibility and the ability to observe can be regarded as external variable quantity that directly affect the factors in the technology acceptance model. Finally, constructs from the theory of planned behavior, such as subjective norms and perceived behavior control, which incorporate user experience dimensions, when linked to technology acceptance model constructs, offer further insights into technology acceptance. The basic conceptual model is exhibited in Fig. 2.

Hypotheses development

Instrument development

A survey instrument was developed for PLS-SEM to test the correlation between constructs in the proposed model. The pilot 5-point Likert scale close-ended questionnaire was selected for the study (Croasmun & Ostrom, 2011). The five experts first checked the questionnaire’s content for clarification within Malaysian context. For the pretest, questionnaires were sent to 12 experts working in finance, logistics, and IT departments of SMEs in Malaysia. Their feedback has helped us to validate the instrument. The details of the construct measurements is presented in Table 2.

Data collection

An online survey method was applied to save time and get fast feedback from the respondents. Google Forms was used to get responses through email from the experts. The questionnaires were accessible online for a 16 week duration (mid-January 2023–mid-April 2023). In total, we obtained only 206 responses from experts working in the SMEs of Malaysia. The average completion period was approximately five minutes, close to the expected completion time based on the pilot testing. We have deleted 12 replies that were complete in less than one and a half minutes. Moreover, all responses were tested, and we uncovered that no answers to all items had the same grade. Due to missing data, further eight responses were withdrawn. Finally, for PLS-SEM, a total of 186 correct responses were considered. To determine the required sample size for testing our proposed model, we utilized the G* Power 3.1 software. We selected the setting for an F test with linear multiple regression, specifically the fixed model, where the R square increases. The parameters we used were as follows: an effect size (f square) of 0.15, an alpha error probability of 0.05, and a desired power of 0.95. Additionally, our model included a total of eight predictors. Based on these specifications, the calculated sample size required for testing our proposed model was determined to be 107. The study of Sideridis et al. (2014) has acclaimed that a small sample size is sufficient for PLS-SEM analysis. The sample size met the five observations per parameter norms. The demographic variables are presented in Table 3.

Method bias issues

We conducted a Harman’s single factor test (Chang, Witteloostuijn & Eden, 2010) to check the common method bias. Harman’s single factor test shows that 45% was the highest difference explained by a single variable. So, there is no multicollinearity problem in the data. Then, the variance inflation factor (VIF) test was used to determine the presence of excessively correlated factors before evaluating the model. The highest VIF score (3.705) was confirmed to be less than the standard value of 5 (Hair et al., 2019). The results show that this study does not have multicollinearity issues and that the measurement model is adequate. The VIF test results are presented in Table 4. Recursively from the structural model, endogeneity can be created. Consequently, we applied the Ramsey regression equation error test (Ramsey, 1969), and there was no endogeneity issue.

Results

The proposed model was tested by a two-step process for finding a relationship among constructs. In the first phase, we tested the constructs’ reliability and validity, and in the second phase, the structural equation modeling was analyzed.

Convergent validity

A convergent validity test is practical when the hypothetical factor created for the analysis strongly correlates with the items utilized to measure it. There must be a significant discrepancy among the items in a construct. Nine constructs were investigated in this study using the rules for determining validity. The significant level of all factor loadings must be evaluated, and the value of all measurements must be more than 0.70.

• After validating the loadings, the composite reliability test must be performed on all of the factors; each construct’s value must be greater than 0.70 (Hair et al., 2016).

• Variance average: after confirming the reliability of each construct, it must be extracted and tested; the standard for each construct is 0.50 and above (Fornell & Larcker, 1981).

• The discriminant validity analysis examines the degree to which measurement factors in a proposed model are dissimilar from each other (Fornell & Larcker, 1981).

All of the tests for the study were completed using SmartPLS. The loading value for measurement item CPX4 was lower than expected (0.70), indicating a problem with internal consistency. As a result, the CPX4 component was removed from the research. Table 5 shows the factor loading values for each item.

Table 5:

Outer loading.

	β	Mean	Standard deviation	T statistic	P values
ATT1 <- Attitude	0.827	0.829	0.023	35.669	0.000
ATT2 <- Attitude	0.847	0.849	0.022	39.309	0.000
ATT3 <- Attitude	0.849	0.847	0.022	38.826	0.000
ATT4 <- Attitude	0.846	0.845	0.021	40.330	0.000
BI1 <- Behavioral intention	0.925	0.925	0.011	85.098	0.000
BI2 <- Behavioral intention	0.926	0.926	0.010	89.784	0.000
CPT1 <- Compatibility	0.803	0.804	0.028	28.402	0.000
CPT2 <- Compatibility	0.739	0.734	0.039	18.910	0.000
CPT3 <- Compatibility	0.840	0.840	0.020	43.001	0.000
CPT4 <- Compatibility	0.778	0.778	0.035	22.147	0.000
CPX1 <- Complexity	0.846	0.845	0.032	26.036	0.000
CPX2 <- Complexity	0.867	0.864	0.027	31.991	0.000
CPX3 <- Complexity	0.758	0.750	0.047	15.979	0.000
CPX4 <- Complexity	0.499	0.482	0.103	4.860	0.000
PBC1 <- Perceived behavioral control	0.879	0.877	0.022	40.483	0.000
PBC2 <- Perceived behavioral control	0.918	0.917	0.011	80.349	0.000
PEOU1 <- Perceived ease of use	0.823	0.824	0.025	33.090	0.000
PEOU2 <- Perceived ease of use	0.822	0.822	0.024	34.405	0.000
PEOU3 <- Perceived ease of use	0.857	0.858	0.023	37.289	0.000
PEOU4 <- Perceived ease of use	0.827	0.827	0.023	35.197	0.000
PU1 <- Perceived usefulness	0.821	0.820	0.028	29.008	0.000
PU2 <- Perceived usefulness	0.849	0.850	0.022	39.369	0.000
PU3 <- Perceived usefulness	0.876	0.876	0.022	40.265	0.000
PU4 <- Perceived usefulness	0.841	0.842	0.020	41.788	0.000
RAD1 <- Relative advantage	0.748	0.743	0.051	14.534	0.000
RAD2 <- Relative advantage	0.781	0.782	0.036	21.501	0.000
RAD3 <- Relative advantage	0.750	0.745	0.044	16.912	0.000
RAD4 <- Relative advantage	0.806	0.806	0.031	25.577	0.000
SN1 <- Subjective norms	0.764	0.764	0.030	25.069	0.000
SN2 <- Subjective norms	0.835	0.835	0.028	30.112	0.000
SN3 <- Subjective norms	0.825	0.826	0.026	31.248	0.000
SN4 <- Subjective norms	0.739	0.737	0.036	20.610	0.000

DOI: 10.7717/peerj-cs.2466/table-5

After factor loadings, the composite reliability and average variance extracted were examined. Figure 4 depicts the measurement model. As shown in Table 6, all constructs were above the standard of 0.7, indicating a strong sign of measurement item consistency.

Table 6:

Composite reliability and validity.

Code	Cronbach alpha	rho_A	Composite reliability	Average variance extracted
ATT	0.864	0.864	0.907	0.710
CPT	0.800	0.804	0.870	0.626
CPX	0.768	0.768	0.866	0.684
BI	0.832	0.832	0.923	0.856
PBC	0.764	0.781	0.893	0.808
PEOU	0.759	0.767	0.892	0.805
PU	0.850	0.851	0.909	0.769
RAD	0.775	0.777	0.855	0.597
SN	0.801	0.804	0.870	0.627

DOI: 10.7717/peerj-cs.2466/table-6

The discriminant validity of each latent variable was determined by testing the square root of average variance extracted, as indicated in Table 7. According to the data, average variance extracted square root is better than the association between the constructs, indicating that all items met the validity requirements and may be utilized to evaluate the structural equation model.

Table 7:

Discriminant validity.

	ATT	CPT	CPX	BI	PBC	PEOU	PU	RAD	SN
ATT	0.843
CPT	0.708	0.791
CPX	−0.487	−0.456	0.827
BI	0.771	0.616	−0.504	0.925
PBC	0.741	0.647	−0.454	0.712	0.899
PEOU	0.735	0.571	−0.367	0.680	0.616	0.897
PU	0.793	0.576	−0.373	0.734	0.687	0.854	0.877
RAD	−0.432	−0.364	0.562	−0.451	−0.369	−0.444	−0.427	0.772
SN	0.751	0.632	−0.358	0.694	0.710	0.710	0.711	−0.373	0.792

DOI: 10.7717/peerj-cs.2466/table-7

Note:

The diagonals (bolded) are the square root of the AVE, while the off-diagonals are correlations.

Structural model

The significance assessment of the structural path was done using bootstrapping in the second phase of evaluating the structural equation modeling (SEM). The bootstrapping procedure involved testing several subsamples (5,000) from the original samples with replacements to check bootstrap faults. This led to the predicted T-values for measurement model significance testing. The bootstrapping procedure result estimates data normality, as shown in Fig. 5.

The findings confirm that the independent construct justified 66.6% of the variation on the intention to use. As a result, Table 8 shows the ultimate decision about hypothesis development.

Table 8:

Hypothesis test result.

	Code	β	Mean	STDEV	T	P	Decision
H1	Perceived ease of use -> Perceived usefulness	0.675	0.672	0.056	12.156	0.000	Accepted
H2	Perceived ease of use -> Attitude	0.214	0.214	0.083	2.574	0.010	Accepted
H3	Perceived usefulness -> Attitude	0.610	0.612	0.076	7.990	0.000	Accepted
H4	Attitude -> Intention to use	0.328	0.332	0.084	3.918	0.000	Accepted
H5	Perceived usefulness -> Intention to use	0.235	0.229	0.074	3.176	0.002	Accepted
H6	Compatibility -> Perceived usefulness	0.060	0.061	0.058	1.041	0.298	Rejected
H7	Compatibility -> Perceived ease of use	0.473	0.478	0.078	6.069	0.000	Accepted
H8	Complexity -> Perceived usefulness	−0.017	−0.016	0.050	0.341	0.733	Rejected
H9	Complexity -> Perceive ease of use	0.001	0.002	0.085	0.015	0.988	Rejected
H10	Relative advantage -> Perceived usefulness	−0.030	−0.030	0.048	0.633	0.527	Rejected
H11	Relative advantage-> Perceived ease of use	−0.272	−0.276	0.081	3.381	0.001	Accepted
H12	Subjective norms -> Perceived usefulness	0.177	0.180	0.067	2.653	0.008	Accepted
H13	Subject norms -> Intention to use	0.124	0.127	0.069	1.797	0.073	Rejected
H14	Perceived behavioral control -> Intention to use	0.220	0.219	0.071	3.098	0.002	Accepted

DOI: 10.7717/peerj-cs.2466/table-8

Furthermore, we applied the partial least square (PLS) predict, a holdout sample-based technique that creates case level estimates on an item utilizing the PLS predict with a 10-fold process to test for predictive power. According to Table 9, the PLS model’s majority of errors were lower than the linear model’s (LM), representing that our proposed model has modest predictive power.

Table 9:

PLS-predict.

	PLS		LM		PLS – LM
	RMSE	MAE	RMSE	MAE	RMSE	MAE	Q2_predict
BI1	0.920	0.783	0.933	0.765	0.013	−0.018	0.489
B12	0.880	0.703	0.885	0.705	0.005	0.002	0.487

DOI: 10.7717/peerj-cs.2466/table-9

Theoretical implications

Our study represents a significant contribution to the field of information systems adoption and sustainable SCM by proposing a model to investigate the acceptance of blockchain in post-COVID-19 sustainable SCM systems of SMEs in Malaysia. Unlike the predominantly conceptual nature of existing research on blockchain, our study stands out as an empirical analysis that integrates traditional adoption theories, including the technology acceptance model, diffusion of innovation theory, and theory of planned behavior. This integration allows us to gain a comprehensive understanding of the factors influencing the adoption of blockchain technology. Importantly, there is a lack of empirical research focusing on developing economies like Malaysia in this area. By utilizing constructs derived from these adoption theories, such as perceived ease of use, perceived usefulness, attitude, complexity, compatibility, relative advantage, subjective norms, and perceived behavior control, our study fills a gap in the literature and provides valuable insights into the acceptance of disruptive technologies in SMEs’ sustainable SCM systems. The findings of our proposed model contribute to the body of knowledge on innovation adoption and offer practical implications for academics and experts seeking to understand and implement blockchain technology in the post-COVID-19 sustainable SCM of SMEs.

Practical implications

The findings of our proposed model have significant practical implications, as they demonstrate a strong explanatory power, accounting for 66.6 percent of the variation in the intention to use BCT in SMEs post-COVID-19. These implications provide valuable insights and realistic recommendations for organizations seeking to address critical sustainable SCM issues through the adoption of blockchain technology. For SMEs in emerging economies considering the adoption of BCT, it is crucial to grasp the fundamental constructs of behavioral intention identified in our study. Additionally, BCT developers should not only focus on understanding the technology itself, but also develop BCT applications specifically designed for sustainable SCM systems. These applications can enable retailers to track stock in real-time, enhance replenishment efficiency, reduce product misplacement, and lower labor costs. Moreover, BCT implementation can assist administrations in organizing anti-counterfeit measures in the post-COVID-19 era. Based on the results of our study, the conceptual model demonstrates its descriptive power and provides a holistic approach for future research on information system implementation. For experts implementing BCT, our study offers valuable insights into developing innovative technological solutions. BCT implementation can promote transparency in the manufacturing and service industries, increase cost-effectiveness, centralize marketing strategies, enable organizations to swiftly obtain customer information, and contribute to gaining a competitive edge in the post-COVID-19 business landscape.

Conclusion and future work

This study explores the factors influencing the adoption of BCT in SMEs in the post-COVID-19 era. The study utilizes the unified framework of three traditional theories: technology acceptance model (TAM), diffusion of innovation (DOI), and theory of planned behavior (TPB). The study examines the relationship of perceived usefulness, perceived ease of use, attitude, relative advantage, compatibility, subjective norms, and perceived behavioral control on the intention to adopt BCT in SMEs for post-COVID-19. In response to Research Question 01, the results indicate that perceived usefulness, perceived ease of use, and attitude is significantly associated with the intention to accept BCT in SMEs. Additionally, relative advantage and compatibility have a significant relationship with perceived ease of use. Subjective norms is significantly associated with perceived usefulness, while perceived behavioral control is substantially associated with the intention to use BCT in SMEs in Malaysia. However, the complexity does not show a significant relationship with either perceived usefulness or perceived ease of use. In response to Research Question 02, the study also investigates the importance of TAM constructs (perceived usefulness, ease of use, and attitude) during BCT adoption in SMEs for sustainable SCM systems. The findings reveal that TAM construct perceived ease of use is most crucial in this context. However, compatibility, relative advantage, and subjective norms show no significant relationship with perceived usefulness. Moreover, subjective norms have a negligible relationship with intention to use blockchain technology.

While this study provides valuable insights, there are some limitations that open avenues for further research. Firstly, the study is based on the opinions of Malaysian SME practitioners, and perspectives from practitioners in other countries may differ. Secondly, the sample size is limited to Malaysia, so caution must be exercised when generalizing the findings to other developed countries like the US and UK in the post-COVID-19 context. Thirdly, the practical implementation of blockchain technology in developing economies like Malaysia is still limited, and further research is needed to understand post-adoption behavior. Fourthly, blockchain is not a standalone technology, and future research can explore its integration with AI and IoT to address security, privacy, and data integration concerns. Fifthly, local regulations play a significant role in technology adoption, and considering them will enhance the validity of future research in the post-COVID-19 era. Finally, other factors such as cost, traceability, self-control, perceived enjoyment, information quality, and system quality should be included in future research to gain a comprehensive understanding of blockchain adoption in sustainable SCM for post-COVID-19 SMEs. Overall, this analysis provides valuable insights into the critical success factors for BCT adoption and its impact on sustainable supply chain management in SMEs. Further research in these areas will contribute to a better understanding of the topic and its implications.