Utilizing digital technology to achieve stable agricultural production: empirical analysis of the supply chain resilience of agricultural-related enterprises

Abstract

Introduction:

In the context of increasing global uncertainty, ensuring the stability and security of agricultural supply chains has become a core issue of national economic security. However, how digital technologies can enhance the resilience of agricultural supply chains is still an open research question. Therefore, this paper aims to systematically examine the impact of digital technology on the resilience of agricultural supply chain and its differentiated mechanism.

Methods:

This paper uses the panel data of agriculture-related listed companies in Shanghai and Shenzhen stock markets from 2011 to 2023, and combines the data of enterprise digital technology patents to empirically test the impact of digital technology on the resilience of agricultural supply chain. Two-stage least squares method and Heckman two-stage causal inference method are used to deal with endogeneity, and several robustness tests are carried out. At the same time, this paper explores its mechanism through causal mediation analysis, and carries out heterogeneity analysis based on urban administrative level and enterprise ownership nature.

Results:

The study finds that digital technology has significantly improved the risk resistance and resilience of the supply chain of listed agricultural companies. This conclusion is still valid after the two-stage least square method, Heckman two-step method and several robustness tests. The causal mediation analysis shows that digital technology improves supply chain resilience through two channels: promoting enterprise cooperation and enhancing information transparency. Heterogeneity analysis shows that at the level of urban administrative level, digital technology significantly improves the resilience of high-grade urban agricultural enterprises, but only enhances their risk resistance to ordinary prefecture-level urban agricultural enterprises. In terms of enterprise ownership nature, digital technology significantly improves the risk resistance and resilience of private agricultural enterprises, but unexpectedly weakens the risk resistance of state-owned enterprises, but has no significant impact on their resilience.

Discussion:

This study not only deepens the understanding of the determinants of supply chain resilience in the era of digital economy, but also provides powerful empirical evidence and policy implications for promoting agricultural modernization and building a safe and controllable agricultural supply chain system.

1 Introduction

As agriculture is the fundamental industry of the national economy, the stability and security of its supply chain directly affect national economic security. However, in recent years, uncertainties such as extreme climate events, geopolitical conflicts, public health crises, and demand fluctuations have become increasingly frequent, posing significant challenges to traditional agricultural supply chains, which are inherently vulnerable (Ivanov et al., 2019). These shocks may lead to production disruptions, logistics blockages, and imbalances in supply and demand, and could even trigger systemic risks of industrial chain fractures, highlighting the importance and urgency of enhancing agricultural supply chain resilience (ASCR). In this context, the digital technology (DT) revolution, represented by artificial intelligence, big data, the Internet of Things, and blockchain, is penetrating every aspect of agricultural production, processing, circulation, and consumption with unprecedented breadth and depth, providing a historic opportunity to reshape the organizational model and risk response capabilities of the agricultural supply chain (Wang and Jia, 2025b). Whether DT can effectively empower the agricultural supply chain and enhance its ability to survive and develop in a complex and changing environment has become a core issue of concern for both the academic community and policymakers.

Supply chain resilience, defined as a system's ability to absorb shocks, maintain functionality, and quickly return to the desired state when subjected to disturbances, has emerged as a leading field in supply chain management research. The traditional view holds that supply chain resilience is a unidimensional concept. However, an increasing number of studies have demonstrated that decomposing it into two core dimensions—risk resistance capability (RRC) and recovery capability (RC)—can provide a deeper understanding of its internal structure and formation mechanism (Tukamuhabwa et al., 2015). Risk resistance refers to the supply chain's ability to maintain structural and functional stability through buffering and robustness during an impact; that is, its “resistance” capability. Recovery capability focuses on the supply chain's ability to quickly return to its original level or even surpass it after suffering an impact through adaptation and adjustment; that is, its “repair” capability. However, current research on DT and supply chain resilience still examines resilience as a single aggregate, with few studies focusing on resistance capability and recovery capability (Wang et al., 2026; Sun et al., 2026). Based on the analysis of the general impacts of DT on supply chain resilience (Lahmar and Siddiqui, 2025; Sun et al., 2026), some studies have also analyzed supply chain resilience in contexts such as low-carbon transition (Ma et al., 2025), the medical industry (Vimal et al., 2022), food (Singh et al., 2023), and the auto industry (Balakrishnan and Ramanathan, 2021), providing valuable insights. However, ensuring the stability of the agricultural supply chain remains an unresolved issue. Even fewer studies have examined the distinct pathways and mechanisms through which digital technology affects each dimension of resilience in agricultural supply chains. The academic community generally agrees that DT positively impacts supply chain efficiency and transparency (Pu and Bai, 2026), but the specific effects on agricultural supply chain resilience remain unclear. Agricultural supply chains possess unique characteristics such as long production cycles, significant influence from natural conditions, dispersed participants, and uneven levels of informatization. This makes the application and enabling effect of DT in the agricultural sector potentially quite different from those in other industries (Subeesh and Chauhan, 2026). For instance, Internet of Things technology may significantly enhance the risk resistance of the supply chain by precisely monitoring the cultivation and breeding environment, while cloud-based collaboration platforms may quickly match alternative resources after a supply disruption, thereby improving recovery capability. These differentiated impact paths and effects require solid theoretical analysis and rigorous empirical verification. Moreover, as DT involves high investment and high risk, its enabling effect on supply chain resilience may be influenced by the characteristics of the enterprises themselves and the external environment. Clarifying these heterogeneous influences is crucial for promoting the precise application of DT in different types of agricultural enterprises. Consequently, three specific questions remain unanswered: (1) does DT affect RRC and RC through the same or different mechanisms? (2) Are the mediating roles of information transparency and enterprise cooperation equally important for both dimensions? (3) How do urban administrative hierarchies and firm ownership structures moderate the relationship between digital technology and supply chain resilience? To address these questions, we integrate the resource-based view and institutional logic theory. The resource-based view suggests that digital technology, as a unique organizational resource, can enhance both buffering capacity, corresponding to risk resistance, and adaptive capacity, corresponding to recovery capability. Institutional logic further predicts that state-owned enterprises and private firms, operating under different logics, will exhibit divergent resilience outcomes from digital technology adoption.

Based on this, this study aims to bridge the gaps in existing research and construct an integrated analytical framework that connects DT and the resilience of agricultural supply chains in two dimensions. The core research question is how do digital technologies differently affect the risk resistance and recovery ability of agricultural supply chains? Are these effects different among various types of agricultural enterprises and external environments? To answer these questions, this study selects Chinese A-share agricultural listed firms from 2011 to 2023 as the research sample. We use the number of digital patents granted to each firm by the National Intellectual Property Administration to measure digital capability. In constructing the dependent variable, this study draws on frontier research and decomposes supply chain resilience into two quantifiable indicators: risk resistance and recovery ability. In terms of research methods, this study employs the classic two-way fixed effects model combined with robust standard errors to enhance the accuracy of the estimation results. More importantly, to address the endogeneity challenge, this study uses the Heckman two-step method and instrumental variable method to eliminate potential sample selection bias and reverse causality, ensuring that the estimation results are causal.

The marginal contributions of this study are mainly reflected in the following aspects: first, in terms of research perspective, this study breaks through the traditional paradigm that views supply chain resilience as a single concept, innovatively constructing a dual-angle analysis framework based on “risk resistance and recovery ability,” and systematically examining the differentiated impacts of DT on different dimensions of agricultural supply chain resilience and the underlying mechanisms. This approach forms an effective dialogue with existing literature and deepens the theoretical connotation of digital supply chain resilience research. Second, in terms of mechanism verification, this study clearly distinguishes the contributions of information transparency and the degree of enterprise cooperation to resistance and recovery ability, using rigorous mathematical models to demonstrate their mechanisms, thereby providing a more detailed and logical framework for understanding how technology transforms into resilience. Third, in terms of theoretical explanation, this study integrates the resource-based view and institutional logic to conduct a more in-depth analysis of the heterogeneous results, enhancing the theoretical depth of the article and providing more targeted references for policy formulation and enterprise practice.

2 Literature review and theoretical basis

2.1 Agricultural supply chain resilience and DT

Resilience was initially used to describe the ability of a system to maintain its basic functions and recover and regenerate within a relatively short period when encountering external shocks (Holling, 1973). Later, it was extended to the field of social sciences to characterize the resistance, recovery, and regeneration capabilities of economic entities under external shocks or risk pressures. The concept of supply chain resilience originates from ecology and engineering, emphasizing the system's ability to maintain its core functions and quickly recover when facing external disturbances. In the field of supply chain management, resilience is defined as the ability of the supply chain to prevent, respond to, and recover from disruptions (Ponomarov and Holcomb, 2009). As the risks faced by global supply chains become increasingly complex and persistent, building a more resilient supply chain has become a hot topic in academic research. Early literature mainly explored the impact of the supply chain's structural elements, strategic choices, and organizational culture on resilience. These elements specifically include redundant design, operational flexibility, diversified procurement strategies, strategic inventory, and risk management culture (Christopher and Peck, 2004). Current research holds that supply chain resilience is a comprehensive capability system covering pre-event warning, in-event regulation, and post-event learning. Its core dimensions include resilience, recovery ability, adaptability, and innovation capability (Ruan and Sun, 2026). Given the unique characteristics of the agricultural sector, the concept of resilience needs further extension. The natural dependence and seasonality of agricultural systems make them more vulnerable to climate change, natural disasters, and ecological fluctuations.

Multi-level entities in the agricultural industry chain, such as farmers, cooperatives, enterprises, and governments, have significant differences in information acquisition, technical ability, and bargaining position, which lengthens the transmission path of risks and complicates collaborative recovery. Therefore, the resilience of the agricultural industry chain and supply chain is not only the defense ability to deal with short-term shocks but also a systematic capacity to achieve “perception–response–reconstruction–innovation” in a complex environment. It emphasizes that in the dual risks of nature and market, the stable operation and continuous optimization of the chain can be realized through information interconnection, resource integration, institutional coordination, and learning and innovation (Harsanto et al., 2025; Gong et al., 2026). Therefore, we believe that the concept of resilience in the agricultural supply chain should be defined as follow: under the influence of the triple uncertainties of nature, economy, and society, the system mitigates impact through resistance, rebuilds operational order through resilience, optimizes structure, and adapts to the new environment through transformative force, ultimately achieving continuous innovation and value re-engineering through competitiveness. Among these, RRC and RC are important criteria for measurement.

The rise of DT has injected new vitality into supply chain resilience research. Technologies such as the Internet of Things (IoT), big data analytics (BDA), artificial intelligence (AI), and blockchain are considered key drivers of supply chain resilience by improving the visibility, synergy, and intelligence of supply chains (Ivanov et al., 2019; Song et al., 2024). For example, IoT sensors enable real-time monitoring of the status and location of goods, allowing enterprises to detect and respond to risks such as transportation disruptions early. Big data analytics provide decision-making support for enterprises by processing massive amounts of data to predict demand fluctuations and supply risks. Artificial intelligence algorithms optimize inventory management and logistics paths, dynamically adjusting to cope with emergencies (Kostovčík et al., 2025). Blockchain technology, with its decentralized and immutable nature, enhances the transparency and traceability of the supply chain, helping to quickly locate and isolate risks when problems occur (Shoomal et al., 2026). Recent studies have further explored the relationship between key core technology innovation and DT application, pointing out that only when DT is deeply integrated with enterprise core business processes can its enabling effect be maximized (Xing et al., 2026).

Some scholars have discussed the influence mechanism between digital technology and supply chain resilience. Relevant research mainly examines the partnership between enterprises and the supply chain. First, DT promotes digital collaboration between enterprises and enhances the trust relationship among chain partners through the information advantage of digital technology (Zhao et al., 2023). Second, DT can form a stable supply chain relationship by increasing the concentration of suppliers or customers (Wang et al., 2016), but it may also hinder the innovation and reform of enterprises due to this overly strong cooperative relationship, thus introducing new supply chain risks (Meng et al., 2018). Third, DT improves the internal control level of enterprises through the transformation to flatter and more networked organizational forms, helping enterprises create a more open production process (Yuan et al., 2021).

Although a large body of literature has explored the positive effects of DT on the supply chain, there are still some shortcomings in the existing research. (1) Most studies focus on theoretical speculation or case analysis and lack large-scale and long-term empirical tests, especially in the specific field of agriculture. (2) When analyzing the digital capability of agriculture-related enterprises, the existing literature often uses the word frequency of annual reports as a measure, which may not fully reflect the comprehensive innovation capability of enterprises in the realm of multiple digital technologies (Yuan et al., 2024). The frequency-based method is unable to accurately assess the true capability of an enterprise's digital applications. (3) Most existing research analyzes supply chain resilience as a whole concept, and the resulting indicators are often comprehensive, failing to deeply explore the differentiated impact of DT on different dimensions of resilience (Zhang et al., 2025). This limits a deeper understanding of the enabling mechanisms of DT.

2.2 Particularity and resilience challenges of agricultural supply chain

Compared with the general manufacturing supply chain, the agricultural supply chain has inherent vulnerabilities and complexities (Xue et al., 2025). First, agricultural production is highly dependent on the natural environment and is vulnerable to the impact of natural disasters such as droughts, floods, pests, and diseases, which leads to significant output uncertainty. Second, agricultural products have characteristics of freshness and perishability, which impose high requirements on storage, logistics, cold chain management, and other processes. Delays or interruptions in any part of the supply chain may cause substantial losses. Third, the participants in the agricultural supply chain typically include many dispersed small farmers, cooperatives, processing enterprises, and retailers, resulting in a low degree of organization and serious information asymmetry. This leads to difficulties in coordinating between upstream and downstream players and forming an effective risk-sharing mechanism (Peng and Pang, 2020). Finally, consumers have increasingly higher demands for food safety and quality. Once a food safety incident occurs, it can quickly destroy consumer trust and deliver a devastating blow to the entire supply chain.

These characteristics make agricultural supply chains particularly vulnerable to external shocks. A sudden outbreak, for example, could lead to labor shortages and logistical disruptions, leaving ripe produce rotting in the fields and unable to be shipped; an international trade dispute could block imports of a key agricultural product, such as soybeans, with knock-on effects on downstream feed processing and breeding industries. Therefore, improving the resilience of the agricultural supply chain is not only essential for individual enterprises to survive, but also a strategic requirement to ensure national food security and sustainable agricultural development. The application of DT is expected to address many drawbacks of traditional agricultural supply chains, thereby systematically improving their resilience. In the face of uncertainty, how to use DT to build agricultural resilience has become a core issue in the process of agricultural modernization (Wang and Jia, 2025a; Jiang and Peng, 2026).

2.3 Risk resistance and resilience

In order to better understand the connotation of supply chain resilience, scholars have begun to deconstruct it across multiple dimensions. Some literature widely recognizes the view that resilience consists of two core dimensions: risk resistance and resilience (Ke et al., 2026). Risk resistance refers to the ability of a supply chain to absorb disturbances and maintain its original operating state, or only slightly deviate from it, through its own robustness and buffer capacity in the face of shocks. High resistance means that the supply chain is less affected by external shocks (Zhang, 2026). This typically relies on proactive risk prevention measures, such as establishing safety inventory, selecting reliable suppliers, and building a diversified supply network. Resilience focuses on how quickly and effectively the supply chain can recover to normal operating levels after an impact and functional decline (Freund et al., 2025). High resilience indicates that the supply chain has strong adaptability and flexibility, enabling it to quickly reorganize resources and adjust processes amidst chaos, and even seize opportunities during a crisis to achieve “creative recovery.” This relies more on post-event emergency responses and learning and adjustment mechanisms, such as quickly activating backup supply plans, using real-time information to adjust production plans, and collaborating with partners.

Resistance and resilience are two complementary yet distinct dimensions for building supply chain resilience (Fatoki and Mora-Monge, 2025). A supply chain with strong resistance but weak resilience can completely collapse during a “black swan” event that exceeds its planning. On the other hand, a highly resilient but less resistant supply chain may become ensnared in frequent “interruption–recovery” cycles, which are costly and unstable to operate. Therefore, a truly resilient supply chain should organically combine resistance and resilience (Lu X. et al., 2026; Lu Z. et al., 2026). Mapping the impact of DT to these two dimensions and further exploring the differentiated contributions of various mechanisms can help clarify the specific paths and logic for enabling supply chain resilience. For example, risk early warning systems based on big data and diversified supplier networks primarily enhance resistance, while intelligent scheduling systems based on the Internet of Things and agile organizational collaboration may enhance resilience through rapid re-planning and response after disruptions occur. Based on this two-dimensional deconstruction framework, this study will provide a more refined analytical perspective for an in-depth understanding of the relationship between DT and agricultural supply chain resilience.

3 Theoretical mechanism and research hypothesis

To systematically explain how digital technology enhances agricultural supply chain resilience and why this effect varies across different institutional contexts, this study integrates two complementary theoretical perspectives: the resource-based view (RBV) and institutional logic theory. From the RBV perspective, digital technology constitutes a unique and heterogeneous organizational resource that is valuable, rare, and difficult to imitate (Liu M. et al., 2025; Rahman et al., 2026). When embedded in supply chain operations, digital capabilities such as real-time data processing, predictive analytics, and collaborative platforms enhance a firm's ability to perceive environmental changes, coordinate upstream and downstream activities, and reallocate resources efficiently during disruptions. These advantages contribute directly to both risk resistance, by strengthening buffering capacity, and recovery capability, by enabling agile reconfiguration. This logic is articulated in hypothesis 1 (H1). Meanwhile, institutional logic theory (Zhou et al., 2024; Geng et al., 2024) suggests that firms operate under distinct institutional logics, such as state-led administrative logic vs. market-driven efficiency logic. These different logics shape firms' technology adoption patterns, implementation depth, and performance outcomes. In the context of China's agricultural sector, state-owned enterprises and private enterprises face different institutional pressures, resource allocation mechanisms, and performance evaluation criteria. These differences lead to divergent pathways through which digital technology translates into supply chain resilience. This theoretical lens therefore provides the foundation for explaining the ownership-based heterogeneity observed in our later analysis, which corresponds to the differential effects between state-owned and private firms. By combining RBV's focus on firm-specific digital resources and institutional logic's emphasis on contextual embeddedness, this study offers a more nuanced understanding of when and how digital technology fosters agricultural supply chain resilience.

3.1 Digital technology and agricultural supply chain resilience

In today's highly uncertain market environment, building enterprise resilience has become a key prerequisite for coping with the impact of crises and achieving sustainable development. This is especially true for agriculture-related enterprises, whose production and operational activities are deeply affected by multiple factors such as natural conditions, market fluctuations, and logistics losses, making them more sensitive to external shocks. Therefore, constructing resilience is particularly important. Greater resilience is associated with shorter recovery times and smaller losses. Digital transformation enables enterprises to overcome bottlenecks such as traditional information delays and inefficient decision-making, which not only improves their ability to perceive environmental changes but also strengthens their dynamic response capabilities in the face of sudden shocks, positively impacting resilience formation (Bao et al., 2025). It is worth noting that in the agricultural supply chain, the logic and pathways for resilience improvement differ significantly among enterprises at different nodes due to their proximity to the end market. Because agricultural products are perishable and their value declines rapidly over time, the sales link has become the core of the entire supply chain operation. Therefore, the closer an enterprise is to the consumer end, the more it needs to leverage DT to accurately predict market demand trends, enabling scientific and efficient business decisions that reduce losses, improve sales efficiency, and systematically enhance enterprise resilience.

DT, especially through the deep integration of cutting-edge technologies such as artificial intelligence, big data, and the Internet of Things (IoT), provides multidimensional support for the agricultural supply chain to cope with external shocks. Before an impact occurs, DT can help each node of the supply chain identify potential risks and optimize resource allocation through more accurate predictions and simulations, thus enhancing the risk resistance of the supply chain (Quan et al., 2024). For example, artificial intelligence modeling and analysis of meteorological data, soil environments, and historical pest and disease information can provide early warnings of agricultural disaster risks and assist farmers in taking preventive measures to reduce yield losses at the source (Jung et al., 2021; Luo et al., 2025). After a shock occurs, DT, with its efficient information processing and cooperative abilities, promote rapid enterprise responses and flexible operating strategies, effectively shortening the recovery period (Zhou et al., 2026; Wang H. et al., 2025; Wang Y. et al., 2025). In cases of sudden regional supply disruptions, for example, a supply chain management system based on a cloud platform can quickly identify alternative suppliers and reconnect with intelligent logistics systems to plan and transport routes, thereby minimizing the negative effects of interruptions.

From the perspective of specific mechanisms, the improvement of risk resistance by DT is mainly reflected in the effective functioning of prevention and buffer functions (Singh et al., 2023; Hu and Ghadimi, 2025). Through big data analysis, enterprises can more accurately predict demand fluctuations, allowing them to build a dynamically adapted safety stock strategy to avoid supply–demand imbalances or supply interruptions caused by drastic demand fluctuations. At the same time, digital twin technology enables enterprises to simulate the operational status of the supply chain under different stress scenarios in a virtual environment, providing data support for the formulation and optimization of emergency plans, thus significantly improving the response of enterprises in the face of real shocks (Ivanov et al., 2022; Bayar et al., 2025). In terms of resilience, the contribution of DT is more evident in the construction of a “perception–response” mechanism. With the real-time monitoring capability of IoT devices, the abnormal status of each node in the supply chain can be quickly identified, such as cold chain temperature exceeding the standard, transportation delays, or vehicle failures. The system will automatically trigger an early warning and initiate a preset emergency process. This closed-loop mechanism of fast perception and agile response is the core support to ensure that the supply chain quickly restores order and reduces operational losses during chaos. To sum up, DT comprehensively strengthens the overall resilience of the agricultural supply chain in uncertain environments by building a risk prevention system in advance and a rapid response mechanism after the event. Based on the above analysis, the first core hypothesis of this study is put forward:

• Hypothesis 1 (H1): DT can significantly improve the risk resistance and resilience of the agricultural supply chain.

3.2 The mechanism through which DT can enhance the resilience of the agricultural supply chain

3.2.1 Information transparency is the foundation for enhancing resilience

Information transparency is another key factor in enhancing risk resistance (Fan et al., 2025). The problem of agricultural supply chain resilience is that institutions and rules are fragmented, and there is a lack of unified standards and trust mechanisms between different regions and actors. Therefore, when the external environment changes, the industrial chain lacks the internal motivation for systematic upgrading. The complex adaptive system theory points out that the resilience of a system comes from the interaction, feedback, and self-organization among multiple agents. This reveals the significance of DT empowerment at the collaborative level of the agricultural supply chain; that is, by platforming data interconnection and optimizing algorithms, it promotes information sharing and collaborative cooperation among all agents, thereby enhancing the self-organization and evolutionary ability of the system as a whole.

The construction of a digital platform addresses this gap to a certain extent and provides support for data flow and collaborative decision-making. DT, especially the application of blockchain and the Internet of Things, can significantly improve the information transparency of supply chains (Daniel et al., 2026). IoT sensors can track produce from field to table, and blockchain technology can record this information in an immutable way. With the combination of blockchain and digital identity authentication, DT can build a quality traceability system that spans the entire chain, enabling transparent management of all relevant information from the origin to the consumption of agricultural products. In transactions and policy implementation, DT relies on automatic compliance review and a smart contract mechanism to reduce the uncertainty caused by manual operations, thus making processes such as credit verification, quality inspection, and payment settlement smoother (Tan et al., 2023). In some regions, the traceability code can facilitate the instant retrieval of data from the origin, processing, and transportation in the end market, which not only improves traceability efficiency but also improves consumer trust in the product. This end-to-end transparency enables companies to identify risk points in advance. For example, by monitoring logistics data in real time, potential shipping delays can be anticipated. Through traceability systems, DT can quickly identify and isolate batches with quality issues, preventing the spread of risks. This ability to provide early warnings and isolate risks in advance directly enhances the robustness of the supply chain in the face of shocks, thereby improving risk resistance. To rigorously analyze how DT can strengthen supply chain resilience through information, this study constructs the following mathematical model:

For risk resistance,

Consider a representative farm household whose output θ is subject to random disasters. Let the potential output be 1 and the actual output be

In Equation 1, αϵ[0, 1] is the protective input including irrigation, fertilizer, and agricultural film, and the cost is . The benefit function is shown in Equation 2.

Farmers are assumed to be risk-neutral, and the objective is to maximize expected returns. The disaster θ follows a normal distribution , with μ < 0.5 to ensure rationality. Under the traditional mode of agricultural production and supply, agricultural producers cannot fully grasp market information. They only know the prior distribution and choose to maximize the income α, so Equation 3 can be obtained:

The first-order condition of optimal protection investment is

According to Equation 4, the actual output is

Based on Equation 5, the sensitivity of output to disasters can be derived as follows:

In Equation 6, we take the absolute value 1−μ to measure the resistance. The larger the value, the weaker the resistance. Assuming that agricultural producers are supported by DT, the market will be transparent in its information. Agricultural producers have full access to market buying and selling information. The DT provides the market signal s = θ + ε, where is independent of θ. Producers update their beliefs, and the posterior distribution is given by Equations 7–9:

According to the posterior mean, agricultural producers choose α(s) = μ′. The calculation process of actual output is shown in Equation 10.

Substituting μ′ = αs+(1−α)μ into Equation 1, where , and s = θ + ε , Equation 11 can be derived:

Considering the expectation of noise ε given θ, we obtain the conditional expected output:

Taking the derivative of θ in Equation 12, we can derive

The absolute value in Equation 13 is 1 − μ(1 + α).

The absolute value of sensitivity is 1 − μ, under the traditional production mode and 1 − μ(1 + α) under the DT. Since α>0, there is 1 − μ(1 + α) < 1 − μ. In other words, DT reduces the sensitivity of output to disasters. Additionally, α increases with the accuracy of the signal, leading to a further decrease in overall sensitivity. Therefore, DT makes protective measures more effective through precise information and enhances risk resistance.

For resilience,

Consider a market for agricultural products with a large number of sellers (farmers) and buyers (acquirers). Let the number of unmatched sellers at time t be U(t) and the number of buyers be V(t). The matching function is

In Equation 14, A represents efficiency, reflecting Le information fluency. Suppose that a sudden shock leads to an imbalance in the market, with U₀ > V₀ at the initial time t = 0, that is, an excess of sellers, and the difference D = U₀−V₀ remains constant. There are no new entrants thereafter, and the system dynamics are determined by matching:

Let V(t) = U(t) − D and substitute into Equation 15 to obtain Equation 16:

Separating the variables and integrating them, Equation 17 can be obtained:

Using the formula U² − DU = (U − D/2)² − (D/2)², Equation 18 can be obtained:

Substituting into Equation 19 yields

Therefore, , Equation 20 can be obtained:

When t → ∞, U(t) → D, that is, the remaining D sellers cannot be matched. The speed of recovery is dominated by the exponential term and the characteristic time constant is . DT, such as e-commerce platforms, will improve matching efficiency, A_dt > A_trad. Thus, the recovery time constant is smaller and the system returns to a steady state faster. Specifically, the rate of decline of U(t) is proportional to A, so DT significantly enhances resilience.

The two models we constructed jointly reveal the mathematical logic of DT strengthening the resilience of the agricultural supply chain from both micro decision-making and macro matching levels.

The two models constructed above jointly reveal the mathematical logic through which digital technology strengthens agricultural supply chain resilience at distinct levels of analysis. The first model operates at the micro level of the individual farm household and demonstrates that digital technology reduces output sensitivity to random disasters by improving the precision of protective investments. This gain in precision is achieved through the provision of accurate market and environmental signals, which enhance the effectiveness of ex ante risk mitigation measures. The second model shifts to the macro level of market clearing and illustrates how digital technology accelerates the matching process between unmatched sellers and buyers following an exogenous imbalance, thereby shortening the recovery time of the agricultural product market as a whole. Based on this analysis, this study proposes the second research hypothesis:

• Hypothesis 2 (H2): DT improves agricultural supply chain resilience by enhancing information transparency.

3.2.2 Enterprise cooperation is key to enhancing resilience

The degree of enterprise cooperation emphasizes the coordination among enterprises in the chain. According to data released by China's Ministry of Agriculture and Rural Affairs in 2024, there were nearly 4 million family farms and 219,700 registered farmer cooperatives nationwide. However, the vast majority of cooperatives still have significant deficiencies in informatization and cooperation capabilities. Big data platforms and AI can integrate dispersed small farmers, cooperatives, and leading enterprises into the same network system, allowing production plans, inventory status, and market demand to be shared in real time.

DT enables upstream and downstream enterprises to share data in real time and form strong information alliances through shared cloud platforms and collaborative management tools. The core value of this partnership is evident in the aftermath of a shock. When an external shock occurs, a close partnership allows companies to negotiate countermeasures together, quickly reorganize resources, and cooperate to address the crisis rather than face it alone. For example, processing enterprises can quickly transmit information about sudden changes in market demand to upstream farmers to help them adjust production and respond collectively. This collaborative response based on digital trust greatly improves the adaptation and resilience of the entire supply chain (Xu et al., 2025). In terms of resource scheduling and cooperation, the advantages of DT are particularly pronounced. When transportation is interrupted due to natural disasters such as earthquakes and floods, or when cold chain equipment experiences partial failure, or when the market suddenly halts, the system can generate multiple sets of alternative solutions based on real-time data. It can also provide optimal solutions for resource delivery and path selection under complex constraints, effectively reducing losses and waiting time during supply chain recovery. Compared with the traditional method of relying on manual judgment, the intelligent algorithm can complete scheduling efficiently in just a few minutes. Efficient cooperation based on DT enhances the ability of agriculture-related enterprises to respond to crisis events. Digital twin technology, which has garnered wide attention in recent years, has also introduced new possibilities for the resilience of industrial and supply chains. By building a virtual chain model, artificial intelligence can simulate the impact of disasters or market fluctuations, identify potential weak links, and then embed emergency plans into the algorithm. Based on the above analysis, this study proposes the third research hypothesis:

• Hypothesis 3 (H3): DT improves agricultural supply chain resilience by promoting enterprise cooperation.

4 Research design

4.1 Sample selection and data sources

The enterprise-level data in this study come from the CSMAR database, and by selecting the option “China Securities Regulatory Commission 2012 Edition Industry Classification” on the website, we preliminarily selected samples of A-share agriculture-related listed enterprises across all links of the agricultural supply chain. Patent data for agricultural listed firms were obtained from the State Intellectual Property Office. We selected agriculture-related enterprises as the initial sample by manually comparing their business scopes. To avoid the confounding effects of the 2008 financial crisis, we set the sample period from 2010 to 2023, using a 2-year lag. To ensure the reliability of the data, this study conducts the following screening of the initial sample: (1) eliminating ST, ^*ST, and PT listed companies; (2) excluding companies that were listed or delisted during the sample period, resulting in less than 3 years of data; (3) eliminating samples with serious missing or abnormal major financial data; and (4) winsorizing all continuous variables at the 1% and 99% percentiles to eliminate the influence of extreme values. After this processing, a panel data set containing 991 companies from 2010 to 2023 is finally obtained, with 13,874 observations in total. It should be noted that the data on enterprise DT patents come from the patent database of listed companies of the China Research Data Service Platform (CNRDS). Based on the patent classification standards of the State Intellectual Property Office and the World Intellectual Property Organization, and referencing relevant literature, this study identifies IPC classification numbers related to key digital technologies such as artificial intelligence, big data, cloud computing, blockchain, and the Internet of Things, and then selects and aggregates the number of annual applications for various DT patents of enterprises (including subsidiaries) to construct a composite index. The financial data used to calculate the supply chain resilience index and other enterprise-level control variables are all sourced from the CSMAR database and Wind database. Descriptive statistical analysis of the variables is shown in Table 1.

Table 1 presents the descriptive statistical results of the main variables. In the two dimensions of agricultural supply chain resilience, the mean value of resistance is 0, while the mean value of resilience is slightly greater than 0, with a large standard deviation, indicating significant differences in supply chain resilience among different agriculture-related enterprises. The core variable DT has a mean of 0.458 and a standard deviation of 2.560, suggesting that the overall level of DT in China's agricultural enterprises is not high and the development is highly uneven. This information provides a solid sample basis for this study to utilize differences for causal identification. The values of other control variables fall within a reasonable range, which is generally consistent with the statistical characteristics found in the existing literature.

4.2 Definition and measurement of variables

4.2.1 Explained variables

Based on the relevant frontier research and the two-dimensional deconstruction theory of resilience, this study selects risk resistance and resilience as the core dimensions representing agricultural supply chain resilience. Although the academic literature has not yet reached a consensus on measuring agricultural supply chain resilience, most studies agree that supply–demand imbalance is a key manifestation of low resilience (Tao et al., 2023). At the same time, many scholars have emphasized that resilience is essentially the unity of resistance and the ability of the system to recover when it is hit by shocks (DesJardine et al., 2019; Smith et al., 2024). Innovative measures to ensure the smooth flow of agricultural products play an important role in enhancing resilience (Wang et al., 2023; Wang H. et al., 2025). Considering that agricultural products have characteristics such as natural dependence, perishability, significant quality differences, and high transportation losses—which lead to frequent sales fluctuations—measures should be based on the actual conditions of supply chain nodes and agricultural product transactions, assessing their resistance and resilience respectively to systematically describe their resilience levels.

Risk resistance capability (RRC): as a key node in the supply chain network, once a node is interrupted, it may cause a serious impact on the production and operation of the entire supply chain (van der Vegt et al., 2015). The exit of node enterprises not only reduces the output level of the supply chain, but also intensifies market volatility, thus weakening overall supply chain resilience (König et al., 2022). Ortiz-de-Mandojana and Bansal (2016) proposed that enterprises with strong viability usually have more stable and sustainable financial performance. Operating income is the key source for agriculture-related enterprises to achieve sustainable development; however, unlike industrial enterprises, agricultural products experience significant fluctuations in output and price, leading to considerable variability in the operating income of agriculture-related enterprises. Therefore, whether agriculture-related enterprises can achieve stable growth in operating income has become the core criterion for measuring the RRC of the agricultural supply chain. Specifically, this study controls relevant variables through regression analysis to obtain the fitted value of the operating income of agriculture-related enterprises and uses the residual between the actual value and the fitted value as the proxy indicator of risk resistance. A positive residual indicates that operating income is higher than predicted, suggesting better-than-expected performance. Otherwise, it implies that the profitability of enterprises suffers significant damage.

Recovery capability (RC). It has been argued that the reduction of inventory write-downs reflects a firm's ability to recover quickly (Behzadi et al., 2020). From the perspective of accounting treatment, when the inventory cost exceeds its net realizable value, the enterprise must make a provision for inventory depreciation. However, when the factors leading to inventory impairment are eliminated or the relevant inventory is sold, the corresponding depreciation reserve should be reversed or carried forward. The output and price of agricultural products fluctuate frequently, significantly impacting the measurement of the inventory depreciation reserve of agriculture-related enterprises. Therefore, an increase in the reduction of the inventory depreciation reserve can effectively reflect the behavioral responses of agriculture-related enterprises in coping with external shocks, achieving product sales, and reducing profit losses, thereby serving as a reasonable indicator to measure the RC of the agricultural supply chain. In terms of specific measurement, this study uses the mean value of the reduction in the inventory depreciation reserve over the previous three periods as the benchmark and calculates the difference between the reduction amount of the current period and the mean value to reflect the enterprise's resilience. A large difference indicates that the enterprise can recover more quickly from external shocks than before; otherwise, it suggests that its recovery effect is limited. In addition, considering the common feature of right-skewed distribution in financial data, the traditional logarithmic +1 linear transformation may lead to sign bias, thus affecting the validity of the estimation (Cohn et al., 2022). Therefore, this study adopts the research method of Bellemare and Wichman (2020) and uses the inverse hyperbolic sine function to transform the resilience index, thereby improving the robustness and applicability of model estimation.

4.2.2 Core explanatory variables

In order to fully reflect the DT capability of enterprises, this study measures the DT level of enterprises based on the number of granted patents for key DT inventions. Choosing the number of authorized invention patents as the measurement standard can avoid the technical uncertainty of unauthorized patents and the limitation of insufficient innovation levels of utility model patents. It has been widely used to measure DT capability based on digital patents applied for by enterprises (Ning and Shun, 2025; Yang et al., 2025; Shen et al., 2026). According to the Reference Relationship Table of Classification of Core Industries of Digital Economy and International Patent Classification (2023) issued by the State Intellectual Property Office, this study matches and summarizes the IPC information of the patents applied for by enterprises in the current year with the international patent classification numbers in the table, and then identifies the DT invention patents applied for by the enterprises. This classification system is based on a comprehensive reference to the DT development strategies, industrial policies, and technical standards of major economies around the world, covering seven key DT fields, including artificial intelligence, high-end chips, quantum information, the Internet of Things, blockchain, the industrial Internet, and the metauniverse, with a total of 585 technology branches. If the IPC code of the patent belongs to a key DT field, the patent is categorized as an innovation achievement in that field.

4.2.3 Mediating variables

According to the theory of mechanism analysis, the mediating variables in this study mainly include two aspects: enterprise cooperation and information transparency. Their respective construction methods are as follows: (1) enterprise cooperation degree. Generally speaking, enterprises in the same supply chain group that are held by common financial institutions can obtain more supply chain resources and build closer cooperative relationships. This study uses the shareholding ratio of financial institutions (including banks, trusts, social security funds, insurance, securities, funds, finance companies, and qualified foreign institutional investors) as a proxy for enterprise cooperation. (2) Information transparency. The transparency score of listed companies is related to the degree of disclosure of financial and management information. Higher evaluation scores represent greater information transparency to a certain extent. This study chooses the transparency scores of listed companies to represent the improvement in information transparency. According to the four levels of evaluation scores (A is excellent, B is good, C is pass, and D is fail), this study assigns the values of A, B, C, and D as 3, 2, 1, and 0, respectively. This indicator is a positive indicator.

4.2.4 Control variables

To control for other factors that may affect the resilience of the agricultural supply chain, this study selects six enterprise-level control variables. The asset-to-liability ratio is calculated as the ratio of total liabilities to total assets. The cash-to-asset ratio is derived from the ratio of total cash and cash equivalents to total current liabilities. The growth rate of total assets is determined by the ratio of the increase in total assets to the initial amount of total assets. Tobin's Q is represented by the ratio of market capitalization to total assets. Ownership concentration is assessed by the sum of the shares held by the top ten shareholders as a proxy variable. “ICEO duality” is analyzed based on whether the chairman of the board simultaneously holds the position of general manager. The value is 1 when the chairman concurrently holds both positions and 0 otherwise.

4.3 Model setting

4.3.1 Benchmark model

To test the impact of DT on ASCR, and based on the research design of published literature (Xu, 2025; Shi et al., 2025; Tan et al., 2025; Lu et al., 2025; Lu X. et al., 2026; Lu Z. et al., 2026), this study constructs the following benchmark panel regression model:

In Equation 21, the explained variable ASCR represents risk resistance and resilience, respectively. DT is the core explanatory variable. The subscripts i and t denote individuals and time, respectively. μ_i is the individual fixed effect, δ_t is the time fixed effect, ε_it is the random disturbance term, and Controls is the information set containing all the control variables. α₀ is the constant term, and β_k is the regression coefficient of the control variables. This study focuses on whether the sign and significance of the coefficient α₁ are in line with expectations. If α₁ significance is positive, then the fundamental research hypothesis of this study will be assumed.

To test the mediating mechanism of differentiation, this study employs the causal mediation analysis method based on existing research ideas. The econometric model constructed in the mechanism testing section is as follows:

In Equation 22, MV is the mechanism variable. In this study, it represents the degree of enterprise cooperation and information transparency as two variables. γ₀ is a constant term and γ₁ is the regression coefficient of DT. The rest of the symbolic meanings are consistent with those in Equation 21. In the mechanism test part, this study focuses on whether the sign and significance of γ₁ meet expectations. The regression coefficients for positive indicators should be significantly positive; conversely, the opposite should be significantly negative. The main steps of the two-phase causal mediation effect model are Equation 22 is used to test whether the effect of DT on ASCR conforms to the research hypothesis. Based on Equation 22, DT was used to regress the two mechanism variables, respectively. If their results conform to the expected hypothesis, one can conclude that a mediation effect exists. Compared with the three-phase intermediary effect model, this equation has a significant advantage in eliminating endogenous factors (Shen and Zhang, 2024; Li et al., 2026; Fu and Li, 2026).

5 Empirical analysis

5.1 Benchmark regression analysis

According to Hypothesis 1 and the empirical model set in Equation 21, this study uses a two-way fixed effects model (TWFE) for regression analysis. The F test, likelihood ratio test, and Hausman test results also confirm that the TWFE sample data is more suitable for this article. Benchmark regression results are shown in Table 2.

Table 2 reports the benchmark regression results. In columns (1) and (3), which exclude control variables, the coefficients of DT are 0.012 and 0.227, respectively, both positive and significant at the 1% level. After adding firm-level controls, the coefficients in columns (2) and (4) are 0.011 and 0.115, respectively, remaining significant at the 1% level. These results confirm that DT can simultaneously enhance enterprises' risk resistance and recovery capability. Hypothesis 1 receives preliminary support. Comparing models with and without controls, we find that the coefficients of DT decrease after including controls, suggesting that the baseline estimates are not inflated by omitted variables; namely, the contributions to RRC and RC margin have declined. This indicates that after eliminating the influence of confounding factors, the TWFE estimation results are more accurate. The positive role of DT in promoting ASCR aligns more closely with economic reality.

Beyond statistical significance, the economic magnitude of digital technology's effect on agricultural supply chain resilience warrants attention. Based on the estimates in column (2) of Table 2, a one standard deviation increase in the number of digital technology patents (approximately 2.56 patents, as shown in Table 1) is associated with a 0.028 unit increase in risk resistance capability. Given that the standard deviation of risk resistance capability is 1.076, this change corresponds to approximately 0.026 standard deviations. In practical terms, this implies that an agricultural enterprise moving from the 25th to the 75th percentile in digital patent stock would experience an improvement in operating income stability equivalent to about 2.1% of the average volatility level, which is a meaningful gain given the thin profit margins typical in the agricultural sector.

To provide concrete evidence, we consider the case of Yili Group, one of China's largest dairy processors, which has systematically deployed digital technologies across its supply chain since 2018. By 2021, the company had accumulated over 120 digital technology patents, covering areas such as cold chain IoT monitoring, predictive logistics algorithms, and blockchain-based traceability. During the COVID-19 pandemic in 2022, when many dairy firms faced severe logistics disruptions and raw milk spoilage, Yili's digital system enabled real-time rerouting of transportation and dynamic inventory redistribution among regional warehouses. As a result, the company's operating income in 2022 declined by only 1.2% relative to 2021, while the industry average decline was 6.7%, demonstrating superior risk resistance. Furthermore, its inventory write-down reduction rebounded to 92% of the pre-pandemic level within 3 months, compared to an industry average of 67% over the same period, reflecting strong recovery capability. This case illustrates how digital technology, measured by patent-based capabilities, translates directly into enhanced resilience along both dimensions.

For private small and medium-sized enterprises, a representative example is Fujian Xiamen Intco Agricultural Technology Co., a vegetable exporter that adopted a cloud-based supplier collaboration platform in 2020. When a typhoon disrupted its main sourcing region in 2021, the platform automatically identified alternative local suppliers and optimized procurement schedules within 2 h, limiting the reduction in operating income to 4.3%, while the regional average fell by 12.8%. This case supports the mechanism of enterprise cooperation as a transmission channel, as further validated in our mediation analysis.

Together, the statistical estimates and real-world cases confirm that digital technology not only shows a statistically significant correlation but also delivers economically sizable improvements in both risk resistance and recovery capability of agricultural supply chains.

5.2 Endogenous test

Because TWFE has well-known limitations for causal identification, the estimation results in Table 2 can only reflect the statistical correlation between DT and SCAR, and do not confirm the causal relationship between them. Therefore, based on established research methods, this article addresses sample selection bias and reverse causality by adopting the Heckman two-stage method and instrumental variable method.

5.2.1 Addressing sample selection bias

The use of DT to enhance supply chain resilience is not unique to listed agricultural enterprises. In the era of the digital economy, other industries and even small and medium-sized enterprises may also actively adopt innovative measures to cope with potential external shocks faced by the supply chain. However, this study focuses on listed agricultural enterprises as the research object. Their innovation capabilities and overall business performance are usually superior to those of non-listed enterprises, which may lead to the sample lacking sufficient representativeness in general. Moreover, due to the limited availability of data for some variables, especially the existence of certain missing values for the explained variables, sample selection bias may arise. Therefore, testing and correcting for sample selection bias is an important step to enhance the causal inference power of this study. To address the possible endogeneity problem caused by this, this study uses the Heckman two-stage method to correct the baseline regression. In the first stage, the study constructs selection dummy variables for the explained variables: if the enterprise's annual operating income is 0 or missing, resilience is assigned a value of 0; otherwise, it is assigned a value of 1. If the enterprise's annual inventory write-down provision decrease is 0 or missing, recovery ability is assigned a value of 0; otherwise, it is assigned a value of 1. Through the assignment of 0 and 1, these are constructed as dummy variables. We also introduce “whether there is a sales expense” as a constraint variable (if the enterprise's sales expense is 0 or missing, it is assigned a value of 0; otherwise, it is assigned a value of 1), and include it in the Probit model for regression. Based on this, the inverse Mills ratio of resilience and recovery ability is calculated, respectively.

The results in columns (1) and (2) of Table 3 show that the regression coefficients of the constraint variables are 1.828 and 1.677, respectively, and both are statistically significant at the 1% level. These results indicate that the chosen exclusion restriction variable has good explanatory power, and the model specification is reasonable. In the second stage, we include the inverse Mills ratio calculated from the first stage as an additional control variable in the benchmark regression model to address potential sample selection bias. Listed in columns (3) and (4) of Table 3, the results show that the inverse Mills ratio regression coefficients were 0.849 and 3.390, and they did not pass the significance test. The results indicate that there is no problem of sample selection bias in this study. The DT regression coefficients were 0.009 and 0.111, respectively, and both are statistically significant at the 1% level. These results support the conclusion that DT promotes the resilience and strengthening of the agricultural supply chain. Therefore, the test results of the Heckman two-stage method further verify the robustness of the core conclusion.

5.2.2 Instrumental variable test

To alleviate the potential endogeneity problem, this study introduces two instrumental variables: the Bartik IV based on R&D investment and the amount of tax refunds received by enterprises. First, regarding the construction of the Bartik instrumental variable, we follow the shift-share approach. Using 2010 as the base year, we calculate the year-by-year growth rate of R&D spending for other firms within the same industry. Based on these growth rates, we predict the current R&D spending for each sample firm. The resulting predicted value serves as the Bartik IV for R&D spending. The predicted value retains the firm's initial R&D level and incorporates the industry-wide growth trend, which is highly correlated with the enterprises' actual R&D. Additionally, R&D investment is a key factor for enterprises to achieve DT, so this instrumental variable meets the conditions related to the core explanatory variables. In terms of exogeneity, the predicted value is based solely on the construction of the base period input and the industry-level growth rate, which are not determined by the actual input decisions of the enterprise and do not have a direct impact on the specific transaction activities of the enterprise in the current period. Furthermore, the explained variable in this study mainly measures the performance of enterprises in the agricultural product trading market, which is relatively independent of the R&D process. Second, we use the amount of tax refunds received by the enterprise. On one hand, to encourage enterprises to conduct R&D activities, the government often provides tax rebates through tax reductions, additional deductions, and immediate refunds. Companies that receive tax refunds have innovation ability that is closely related to the R&D and DT, satisfying the correlation requirement for instrumental variables. On the other hand, the amount of tax rebate is determined by the tax department according to relevant laws and regulations, independent of the subjective intentions of enterprise managers, and has no direct correlation with the specific transaction activities of enterprises in the agricultural products market, thereby better meeting the requirements of exogeneity.

Table 4 shows that the coefficients of the instrumental variables in the first stage are 0.194 and 0.082, respectively, with statistical significance at the 10% and 1% levels. This result indicates that the principle of correlation between IV and core explanatory variables is satisfied. According to the validation results of the LM statistics, the P-values at the 1% level declined, allowing us to reject the null hypothesis of insufficiency. The weak instrumental variable test shows an F statistic of 20.28, which exceeds the 10% critical value of 16.38, indicating that the IV in this article poses no threat. Since two IVs were selected for this study, overidentification tests were performed. The Hansen J statistics for the two equations show p-values greater than 0.1, indicating that there is no threat of overidentification for the IVs selected in this study. Overall, these three test results confirm that the IV satisfies the requirements of correlation and exogeneity. According to the results of the second stage of the DT, the regression coefficients of resistance and recovery ability are 0.801 and 3.832, respectively, with significance at the 1% and 5% levels. This means that by eliminating endogenous problems, DT improves the validity of the ASCR conclusion. H1 is verified again.

5.3 Robustness test

To ensure the reliability of the benchmark regression results, this study also conducted a series of robustness tests.

5.3.1 Replacing the explained variable

In the benchmark regression section, this study used resistance capacity and recovery capacity as proxy variables for supply chain resilience and then analyzed the data using two equations. However, a single indicator has limitations in expression ability, explanatory power, and information quantity. To measure ASCR more comprehensively, based on methods used in existing research, this study selected relevant indicators from two dimensions: product flow and capital flow, and recalculated them using factor analysis. Specifically, in the product flow aspect, the residual of operating income, the difference in the reduction amount of inventory impairment, and the absolute value of inventory changes reflect the fluctuation level of product flow. In the capital flow aspect, the accounts receivable turnover rate, accounts payable turnover rate, the ratio of accounts receivable to operating income, and the ratio of accounts payable to total liabilities reflect the holding level of capital flow. Since some indicators are negative, this study standardized them using the range standardization method. The results in column (1) of Table 5 show that the regression coefficient of DT is 0.002 and is statistically significant at the 1% level. This result indicates that the positive effect of DT on ASCR still exists, confirming the robustness of H1.

5.3.2 High-dimensional fixed effects

The baseline regression section only controlled for individual and time fixed effects, which prevented the incorporation of regional macro policies and industry development dynamics into the model. This exacerbated the negative impact of omitted variables. Generally speaking, adding industry-year and province-year joint fixed effects to the model can respectively control for the impact of industry policies changing over time and provincial policies changing over time, thereby reducing the error caused by enterprises being affected by key policies. Using the reghdfe command in Stata software can incorporate more fixed effects into the model. The results in columns (2) and (3) of Table 5 show that the regression coefficients of DT on recovery ability and resistance ability are 0.132 and 0.009, respectively, both of which are statistically significant at the 1% level. These results indicate that the finding that DT promotes the improvement of ASCR is robust.

5.3.3 Replace the proxy variables of RRC

In the benchmark regression part, we use the residual of operating revenue as the proxy variable for RRC, but this indicator has its flaws. The previous indicators measure the abnormal fluctuations of corporate operating income after controlling for conventional factors and essentially reflect the stability of overall profitability. Operating revenue is affected by many non-supply chain structural factors, such as product price fluctuations, market demand changes, climate disasters, policy subsidies, and enterprise marketing strategies. For example, a firm that achieves revenue growth due to an increase in product prices has a positive residual, but this does not imply that its supply chain is more shock resistant. On the contrary, an enterprise's income may decline due to active price reductions and destocking, and the negative residual does not mean that its supply chain is fragile. The residual of operating income cannot distinguish the contribution of supply chain resilience from that of other operating factors.

To address this concern and provide a more rigorous test of our hypothesis, we re-estimate the model using supplier concentration as an alternative proxy for risk resistance capability. The logic underlying this substitution is both intuitive and theoretically informed. In the face of external disruptions, a supply chain characterized by excessive dependence on a limited number of key suppliers is structurally fragile. If those critical suppliers experience production halts or logistical failures, the focal firm faces a disproportionate risk of operational discontinuity. A resilient supply chain, by contrast, should exhibit the capacity to maintain a diversified and balanced supplier portfolio, thereby absorbing localized shocks without cascading failures. Empirically, this implies that firms with stronger risk resistance should demonstrate either stability or a reduction in supplier concentration following industry-wide disturbances, as they are less compelled to consolidate procurement among a few surviving suppliers and are better positioned to leverage alternative sourcing options.

Following established literature (Rehman et al., 2025; Liu T. et al., 2025; Jin et al., 2025), we measure supplier concentration using the Herfindahl–Hirschman Index (HHI) calculated from the purchase shares of the firm's top five suppliers. The regression results are reported in column (4) of Table 5. The estimated coefficient of digital technology on supplier concentration is −0.112 and is statistically significant at the 5% level. This negative association indicates that firms with higher levels of digital technology adoption experience a smaller increase, or even a decline, in supplier concentration during the sample period. In practical terms, digitally mature agricultural enterprises are less likely to consolidate their supplier base in a way that heightens structural vulnerability.

The importance of this finding extends beyond the mere replication of the baseline result. By demonstrating that digital technology reduces supplier concentration, a metric that directly captures a key structural dimension of supply chain configuration, we significantly alleviate concerns regarding the validity of the operating revenue residual as a proxy for risk resistance. The convergence of evidence from two conceptually distinct measures, one rooted in financial performance stability and the other in supply network structure, provides compelling triangulation. This consistency reinforces confidence that the observed relationship between digital technology and risk resistance is not an artifact of measurement choices but reflects a genuine enhancement of the supply chain's capacity to withstand and absorb disruptions.

5.3.4 Excluding the impact of sudden public health incidents

Due to the global pandemic of COVID-19, the normal operations of agricultural enterprises have been severely affected. This type of sudden external event, characterized by long duration, high severity, and strict government control, has impacted the accuracy of the conclusions. Therefore, we removed the samples from 2020 to 2023 and conducted the estimation using TWFE. The results in columns (5) and (6) indicate that the impact of DT on RRC and RC is significantly positive. This finding is consistent with the baseline regression.

6 Further analysis

6.1 Mechanism testing

To test whether DT can promote ASCR through two pathways—information transparency and enterprise cooperation—this study conducted a regression analysis based on the causal mediation effect model constructed according to Equation 22. The econometric method employed was TWFE. Consistent with the multilevel theoretical mapping articulated in Section 3.2, information transparency is measured using the transparency score of listed firms, which captures the accuracy and availability of firm-level disclosures, while enterprise cooperation is proxied by the shareholding ratio of financial institutions, reflecting the degree of coordinated resource sharing among supply chain partners. The estimated results are presented in Table 6.

As shown in Table 6, the regression coefficients of DTI on information transparency and enterprise cooperation are 0.011 and 0.079, respectively, and both are significant at the 5% level. This result verifies research hypotheses 2 and 3. From the perspective of information transparency, agricultural enterprises can create a unique information capture network through digital transformation, effectively integrating resources, building sustainable competitive advantages, and enhancing resilience. The more enterprises can quickly identify opportunities, perceive changes in market demand, and make strategic adjustments, the more they promote the emergence of new business models and enhance their resilience. To demonstrate the robustness of the mechanism analysis results, we also used the interaction term between the mechanism variable and DT as the core explanatory variable and estimated it using TWFE. From the results in columns (3) to (6) of Table 6, it can be seen that all interaction terms are significantly positive. This indicates that the mechanism variable positively regulates the beneficial effects of DT on RRC and RC. This is consistent with the conclusion obtained from the two-stage mediating effect model.

6.2 Heterogeneity analysis

6.2.1 Heterogeneity analysis at the city level

Given the substantial regional disparities in China, this study further examines whether the impact of DT varies by urban administrative hierarchy. Following the research paradigm of the published literature, this study defines municipalities directly under the central government, provincial capital cities, and sub-provincial cities as high-level administrative cities, while other prefecture-level cities are defined as ordinary cities (Lu et al., 2019). The reasons are as follows: (1) from the perspective of resource allocation rights, under the Chinese decentralized system, administrative hierarchy determines a city's resource allocation capacity. High-level administrative cities typically enjoy greater administrative authority and fiscal autonomy, allowing them to concentrate resources for large-scale new infrastructure investment and industrial policy guidance. Ordinary prefecture-level cities, however, face fiscal constraints and tend to occupy a follower position regarding public goods supply and institutional innovation. This difference in resource endowment may result in significant variation in the penetration depth and application breadth of DT in cities of different levels. (2) From the perspective of factor concentration, high-level cities, due to their status as political centers, naturally become “storage pools” for high-end factors such as talent, capital, and technology. According to the theory of agglomeration economy, factor concentration not only leads to increasing returns to scale but also enhances the efficiency of technology application through knowledge spillover effects. Peripheral cities, on the other hand, are subject to the negative impact of factor loss and may fall into a low-level equilibrium trap. Therefore, the marginal returns of DT may exhibit systematic differences between these two types of cities. (3) From the perspective of the institutional environment, high-level cities are often the testing grounds for major national reform policies, and they are the first to establish an international-compliant business environment and regulatory framework. This advantage of being pioneers in institutional development enables market entities in central cities to have stronger adaptability and trial-and-error space when facing technological changes. Ordinary cities, on the other hand, are more likely to be policy recipients, with changes in the institutional environment occurring at a relatively slower pace.

The results in rows (1) and (3) of Table 7 show that the regression coefficients of DT on the RRC and RC in the high-ranking city sample are 0.004 and 0.189, respectively, with a significant promoting effect observed only on the RC. This result implies that DT can enhance the recovery ability of agricultural enterprises in high-ranking cities but cannot effectively enhance their resistance. The results in columns (2) and (4) of Table 7 indicate that in the ordinary prefecture-level city sample, the regression coefficients of DT on the RRC and RC are 0.015 and 0.041, respectively, with a significant promoting effect observed only on the resistance. This implies that DT can promote the RRC of agricultural enterprises in ordinary prefecture-level cities but cannot effectively enhance the RC. By combining these two results, we reach a completely opposite conclusion: DT can enhance the RC of agricultural enterprises in high-ranking cities and the RRC of agricultural enterprises in ordinary prefecture-level cities. This contrasting finding reveals the heterogeneity of the DT empowerment path. This study argues that this differentiation stems from systematic differences in risk types, resource endowments, and technical application scenarios between the two types of cities. (1) Explanation based on risk types: high-level cities, as regional economic centers, have deeply integrated agricultural supply chains into the national and even global division of labor network, facing mainly systemic shocks, such as international market fluctuations and cross-regional logistics disruptions. RC, which refers to the rapid reorganization capability after an impact, becomes the key dimension of resilience. DT has significantly enhanced enterprises' resilience to systemic risks by building cross-regional collaborative platforms and optimizing logistics dispatch networks. However, when confronting macro-level systemic risks, the digital efforts of individual enterprises are often insufficient to fully withstand the impact, thus diluting the marginal contribution of DT to resistance. On the contrary, the supply chains of ordinary prefecture-level cities are characterized by localization and short supply chains, with risks primarily arising from local disturbances, such as natural disasters and local market fluctuations. RRC, which refers to the ability to absorb impact and maintain operation, becomes the primary goal. DT has significantly improved enterprises' ability to resist local risks through internal control measures such as IoT monitoring and inventory optimization. However, due to resource and network limitations, once the impact exceeds the local boundary, enterprises lack external resources for reconfiguring their supply chains, making the enabling role of DT in resilience difficult to demonstrate. (2) From the perspective of technical application scenarios: in high-level cities, enterprises, leveraging their well-established digital infrastructure, primarily apply DT to connection-oriented scenarios, such as supply chain collaboration platforms and cross-regional resource matching systems. Such technologies are more effective for rapid restructuring after a crisis, that is, resilience, but are less effective against shocks in day-to-day operations. In ordinary prefecture-level cities, enterprises face resource constraints, and the application of digital technologies focuses on control-oriented scenarios, such as production monitoring and inventory warnings. These technologies directly strengthen operational stability but cannot support the reconfiguration of business models, which is a key aspect of resilience. This finding aligns with the resource-based view, where the effectiveness of technology depends on its compatibility with the core tasks of the organization. (3) Supplementary explanations based on the institutional environment: the policy network advantages of high-level cities, such as emergency command systems and government data sharing, have amplified the role of DT in crisis coordination, thereby strengthening resilience. Special subsidy policies in ordinary prefecture-level cities, such as agricultural insurance and disaster prevention equipment subsidies, have encouraged enterprises to apply DT to enhance their daily risk mitigation capabilities, although support for cross-regional recovery remains limited.

6.2.2 Heterogeneity analysis of enterprise ownership nature

Given the unique characteristics of the agricultural sector in China, enterprises with different ownership structures exhibit significant differences in resource acquisition, goal orientation, and institutional constraints. This may lead to a differentiation in the impact of DT on supply chain resilience. Specifically, (1) the resource-based view holds that the heterogeneous resources of enterprises determine their competitive advantages. The realization of the value of DT depends on the organization's absorption capacity. State-owned enterprises (SOEs), due to their natural ties with the government, can more easily obtain low-cost funds and key agricultural data, and have the ability to make forward-looking and large-scale investments in digital infrastructure. Private enterprises, while having flexible mechanisms, often face financing constraints and technical acquisition barriers during digital transformation. These differences in resource endowment may result in a gap in the breadth and depth of DT application based on ownership. (2) From the perspective of institutional economics, SOEs not only pursue economic performance but also bear the social responsibility of ensuring national food security and stabilizing the agricultural product market. Their motivation for introducing DT may be more focused on maintaining stable supply and risk control. Private enterprises, on the other hand, are more concerned about the cost savings and market competition advantages brought by DT, focusing on efficiency and responsiveness. The differences in their objectives suggest that the transmission path of DT's effect on supply chain resilience may vary. (3) From the perspective of the supply chain governance structure, SOEs involved in agriculture often occupy leading positions in key areas such as grain storage and agricultural input circulation. Their digital investments are more likely to be transmitted to the upstream and downstream through administrative coordination or policy guidance, leading to a systematic improvement in resilience. In contrast, private small and medium-sized enterprises are mostly in supporting roles within the supply chain, and their digital impacts are more likely to be constrained by external environmental fluctuations and the digital capabilities of their partners. Therefore, this study expects that the impact of DT on supply chain resilience will be significantly different between SOEs and private enterprises, and it is necessary to conduct group tests to reveal the underlying mechanisms. This research divides enterprises into two sample groups: SOEs and private enterprises.

The results in rows (1) and (3) of Table 8 show that in the sample of SOEs, the regression coefficient of DT on the RRC is significantly negative, while the regression coefficient of DT on the RC is negative but not significant. This result implies that DT can enhance the RRC of state-owned agricultural enterprises but cannot significantly improve the RC, and there is even a possibility of reducing recovery ability. The results in rows (2) and (4) of Table 8 show that in the sample of private enterprises, the regression coefficients of DT on RRC and RC are 0.008 and 0.186, respectively, both significant at the 1% level. This result indicates that DT can significantly enhance the RRC and RC of private agricultural enterprises and plays a strong role in stabilizing the resilience of the supply chain. This study believes that this phenomenon is rooted in the deep differences in the institutional logic and behavioral patterns of the two types of enterprises. (1) From the perspective of institutional logic conflicts, SOEs are tasked with the policy function of ensuring stable production and supply. Their digital transformation is often embedded within administrative logic rather than purely market logic. The transparency and data-driven decision-making brought about by DT may conflict with existing administrative directives. When market risks arise, SOEs need to prioritize completing supply guarantee tasks, such as making acquisitions at any cost. The optimal business decisions revealed by the digital system cannot be implemented, resulting in technological investment not enhancing resilience but instead weakening financial flexibility due to system maintenance costs and interference with routine processes. This “technology–institutional” mismatch makes it difficult for DT to be transformed into a genuine risk resistance capability. (2) From the perspective of organizational behavior, the hierarchical structure and risk-averse culture within SOEs have inhibited the empowerment of DT for resilience. Resilience emphasizes the reorganization and innovation of the supply chain after being impacted (Tukamuhabwa et al., 2015), which requires rapid trial and error and flexible decision-making. However, state-owned enterprise managers, considering the preservation and appreciation of state assets, are cautious about supply chain reconfiguration plans suggested by digital systems, in order to avoid potential political accountability. In this context, DT becomes a monitoring tool rather than an intelligent decision-making tool, and even slows down the recovery process due to the increase in decision-making levels. Private enterprises are quite different. Their efficiency logic, driven by survival pressure, makes DT focus on solving specific supply chain pain points from the very beginning of its introduction. The deep integration of technology and business significantly enhances their ability to resist risks and recover quickly. (3) From the perspective of the resource-based view, the resource redundancy of SOEs may lead to “over-emphasis on construction and neglect of application” in digital formality, resulting in a disconnection between DT and organizational capabilities. In contrast, the resource constraints of private enterprises force them to make precise investments and achieve deep adaptation, thereby achieving an effective coupling of technology and resilience.

It is important to interpret these findings with appropriate caution and to avoid the overgeneralization that digital transformation in state-owned agricultural enterprises is inherently ineffective. Several boundary conditions and measurement considerations warrant attention. (1) SOEs in the agricultural sector often operate under a dual mandate that prioritizes non-financial policy objectives alongside commercial performance. Specifically, ensuring food supply stability, maintaining strategic reserves, and stabilizing market prices during disruptions constitute core components of their organizational mission. The resilience metrics employed in this study, namely risk resistance and recovery capability, are derived from financial indicators such as operating income volatility and inventory write-down reversals. While these measures capture market-oriented dimensions of supply chain resilience, they may not fully reflect policy-centric resilience outcomes valued by state shareholders and regulatory bodies. For instance, a state-owned grain enterprise that sustains a temporary decline in operating income while successfully fulfilling its mandated procurement and storage targets during a supply shock may be deemed highly resilient from a policy perspective, even if financial metrics suggest otherwise. (2) The time horizon over which digital technology benefits accrue may differ systematically between ownership types. State-owned enterprises, given their greater resource slack and longer planning horizons, may be deploying digital infrastructure with multiyear gestation periods. The short-term financial fluctuations captured in our panel may thus underestimate the eventual resilience contributions of these investments. (3) The potential for digital technology to strengthen the non-market dimensions of supply chain resilience in state-owned enterprises, such as enhanced interagency coordination, improved emergency response compliance, and more accurate policy forecasting, represents a promising direction for future research that would benefit from alternative measurement approaches. Therefore, the observed negative or null effects in the state-owned subsample should not be interpreted as evidence that digital transformation fails to enhance supply chain resilience in these firms. Rather, the enabling effects of digital technology in state-owned agricultural enterprises may manifest primarily in policy-driven resilience dimensions that lie beyond the scope of the finance-centered metrics adopted in this study.

7 Discussion and conclusion

7.1 Research conclusions

The stability of the supply chain in agricultural enterprises plays a crucial role in ensuring food security and maintaining sustainable livelihoods. The rapidly developing information technology in the digital age provides innovative solutions for the sustainable development of ASCR. This study incorporates DT and supply chain resilience into a unified research framework and, through theoretical reasoning and mathematical models, thoroughly analyzes the positive effects of leveraging DT to enhance resilience. We utilized data from listed companies engaged in agriculture in China and combined it with the classic panel data model to demonstrate the theoretical model. The core finding of this study is that DT, represented by key digital patents, significantly improves the RRC and RC of the agricultural supply chain. This conclusion is consistent with the macro discussion on the digital economy empowering the real economy (Ma and Lv, 2024; Xu, 2024) and provides micro-level evidence for understanding how enterprises can build dynamic competitive advantages in the digital era. Unlike previous studies that have treated resilience as a single dimension, the dual-dimensional framework and differentiated mechanism tested in this study reveal the complexity of DT empowerment. This finding directly responds to the call by Tukamuhabwa et al. (2015) for more nuanced operationalizations of supply chain resilience. Moreover, while prior studies in manufacturing sectors, such as Balakrishnan and Ramanathan (2021), found that digital technology primarily enhances efficiency rather than resilience, our results demonstrate that in the agricultural context, the effect on resilience is equally strong. This contrast suggests that agriculture's inherent volatility may create a higher baseline need for resilience, making digital investments more likely to translate into stability gains.

DT does not simply make the supply chain more stable and reliable; instead, it simultaneously endows it with “rigidity” for resistance and “flexibility” for recovery. This “combination of rigidity and flexibility” characteristic is an essential feature of supply chain resilience in the digital era and is also key to effectively responding to an increasing number of black swan and gray rhino events. This dual effect aligns with the resource-based view, which posits that heterogeneous resources can simultaneously confer both buffering and adaptive advantages (Rahman et al., 2026). However, our findings extend this view by showing that the same digital resource can produce two distinct resilience outcomes through different mechanisms. This insight challenges the common assumption in prior literature (Singh et al., 2023; Sahoo et al., 2026) that resilience-enhancing technologies operate through a single pathway.

The results of the mechanism analysis further enhance our understanding. The enabling effect of DT on resilience is rooted in the profound transformation of the supply chain operation model and operates through distinct pathways. By enhancing information transparency, DT helps enterprises build a more resilient supply network and a stronger risk warning capability, which is particularly important in today's context of anti-globalization and increased geopolitical risks. By enhancing cooperation among enterprises, DT transforms the loose and competitive node relationships in the traditional supply chain into a tight and collaborative value creation network, endowing enterprises with the dynamic ability to quickly learn, adapt, and evolve. These two mechanisms are interrelated and together form a complete logical loop for the dual-dimensional transformation of DT into supply chain resilience.

The heterogeneity analysis results of this study are particularly insightful. They reveal the context dependence of the DT empowerment effect. There is no universal digital transformation path that can be applied to all enterprises. DT positively affects the recovery ability of agricultural enterprises in high-level urban samples with sufficient administrative resources and fiscal freedom, while it positively affects risk resistance in ordinary prefecture-level city samples. DT has a positive effect on the risk resistance of SOEs and significantly enhances the risk resistance and recovery ability of private enterprises. These findings carry significant implications for enterprises formulating digital strategies and for governments designing industrial policies. The most striking heterogeneity concerns state-owned enterprises, where digital technology unexpectedly weakens risk resistance. This counterintuitive result aligns with institutional logic theory, which suggests that technology adoption in state-owned enterprises may be decoupled from operational practices due to conflicting administrative mandates (Geng et al., 2024). Unlike private firms that face market discipline, state-owned enterprises in our sample appeared to prioritize policy compliance over data-driven optimization, leading to a technology performance paradox. This finding contradicts earlier studies on general digital transformation (Yuan et al., 2021) that assumed uniform positive effects across ownership types, underscoring the necessity of considering institutional context when evaluating digital technology's impact on resilience. These findings carry significant implications for enterprises formulating digital strategies and for governments designing industrial policies.

7.2 Theoretical contributions and future prospects

The theoretical contributions of this study are as follows: (1) by constructing and testing a two-dimensional analysis framework for resilience and recovery, it has deepened the research on supply chain resilience from the perspective of “whether it is effective” to “how to differentiate effectiveness,” providing a more detailed micro-foundation for the theory of digital supply chains. (2) By combining the resource-based view and institutional logic to conduct an in-depth analysis of heterogeneity, it reveals the organizational and environmental logic behind the technology empowerment effect, enriching the application of the theory of technology adoption and organizational adaptation in the agricultural field. This study also has some limitations, which point out the direction for future research: (1) this study only uses the comprehensive concept of DTI. However, digital technologies can be classified into different categories such as task execution and decision optimization. Different digital technologies, such as AI and blockchain, may have more nuanced differences in their impact on various dimensions of resilience. Future research can conduct more in-depth technical decomposition studies. (2) The measurement of supply chain resilience in this study is still mainly based on low-frequency annual financial data. Future research could combine our approach with higher-frequency operational data. Finally, this study mainly focuses on the positive effects of DT, but digital transformation may also bring new risks, such as cybersecurity risks and data privacy risks (Yu and Liu, 2025). The negative effects and their potential impact on supply chain resilience deserve further exploration in future research.

7.3 Policy implications

Based on the above conclusions, this study presents the following policy implications. For government departments, (1) there should be a continuous increase in support for agricultural DT research and application, especially encouraging enterprises to make breakthroughs in core technologies such as artificial intelligence, the Internet of Things, and blockchain, to establish a solid technical foundation for improving the resilience of the agricultural supply chain. (2) Diverse industrial digitalization guidance policies should be implemented. For different regions, production models, and types of agricultural enterprises, targeted policy packages should be provided. For example, for private processing enterprises in coastal areas, support can focus on their development of flexible manufacturing and smart logistics to enhance their recovery capabilities. For state-owned farming and breeding enterprises in inland areas, support can mainly focus on building digital disaster warning and precise production systems to strengthen their risk resistance. For agricultural-related enterprises, (1) digital transformation should be regarded as the core strategy for enhancing supply chain resilience. Increased investment in various digital technologies should be made, and the ability to innovate through technology should be internalized as the core competitiveness of the enterprise. (2) When promoting digital transformation, enterprises should have clear strategic goals. Based on their own characteristics and the environment they are in, they should determine whether to prioritize enhancing resistance or recovery capabilities, and systematically layout digital applications around this goal. For example, enterprises aiming to enhance resistance should focus on using digital technologies to optimize supplier networks and risk warning systems, while those aiming to improve recovery capabilities should focus on building agile internal operation processes and collaborative external cooperation platforms. (3) Enterprises should actively cooperate with the government, research institutions, and supply chain partners, share data resources, jointly build digital platforms, address external risks together, and collaboratively create a more resilient industrial ecosystem.

7.4 Limitations

This study uses data from listed companies to analyze how digital technology enhances the supply chain resilience of agricultural enterprises and obtained reliable results. However, a limitation of our study is that the conclusion depends on annual data, which results in a significant lag effect. For instance, annual data are insufficient to capture the impact of sudden natural disasters on agricultural production. We suggest that future research could utilize daily high-frequency data as the research sample. For RC, we recommend using more appropriate indicators as proxy variables, such as the recovery speed of order fulfillment rates or the recovery period of operating cash flow.

References

• 1

  BalakrishnanA. S.RamanathanU. (2021). The role of digital technologies in supply chain resilience for emerging markets' automotive sector. Supply Chain Manage. Int. J.26, 654–671. doi: 10.1108/SCM-07-2020-0342

  • CrossRef
  • Google Scholar
  • View reference in article

• 2

  BaoA.HeM.ZhouQ.GaoH. (2025). The impact of industrial policies oriented toward the digital economy on the resilience of the agricultural industrial chain. Front. Sustain. Food Syst. 9:1634474. doi: 10.3389/fsufs.2025.1634474

  • CrossRef
  • Google Scholar
  • View reference in article

• 3

  BayarJ.AliN.CaoZ.RenY.DongY. (2025). Artificial intelligence of things (AIoT) for precision agriculture: applications in smart irrigation, nutrient and disease management. Smart Agric. Technol.12:101629. doi: 10.1016/j.atech.2025.101629

  • CrossRef
  • Google Scholar
  • View reference in article

• 4

  BehzadiG.O'SullivanM. J.OlsenT. L. (2020). On metrics for supply chain resilience. Eur. J. Oper. Res.287, 145–158. doi: 10.1016/j.ejor.2020.04.040

  • CrossRef
  • Google Scholar
  • View reference in article

• 5

  BellemareM. F.WichmanC. J. (2020). Elasticities and the inverse hyperbolic sine transformation. Oxf. Bull. Econ. Stat.82, 50–61. doi: 10.1111/obes.12325

  • CrossRef
  • Google Scholar
  • View reference in article

• 6

  ChristopherM.PeckH. (2004). Building the resilient supply chain. Int. J. Logist. Manage.15, 1–13. doi: 10.1108/09574090410700275

  • CrossRef
  • Google Scholar
  • View reference in article

• 7

  CohnJ. B.LiuZ.WardlawM. I. (2022). Count (and count-like) data in finance. J. Financ. Econ.146, 529–551. doi: 10.1016/j.jfineco.2022.08.004

  • CrossRef
  • Google Scholar
  • View reference in article

• 8

  DanielJ.MarounE. A.Garza-ReyesJ. A.JaouhariA. E.SamadhiyaA. (2026). From information asymmetry to transparency: blockchain-enabled digital transformation in manufacturing supply chains. Supply Chain Manage. Int. J.31, 275–296. doi: 10.1108/SCM-09-2025-0903

  • CrossRef
  • Google Scholar
  • View reference in article

• 9

  DesJardineM.BansalP.YangY. (2019). Bouncing back: building resilience through social and environmental practices in the context of the 2008 global financial crisis. J. Manage.45, 1434–1460. doi: 10.1177/0149206317708854

  • CrossRef
  • Google Scholar
  • View reference in article

• 10

  FanL.HeJ.SuB.WangC. (2025). Digital technology penetration and supply chain resilience improvement: enterprise innovation strategies in the digital age. Int. Rev. Financ. Anal.108:104722. doi: 10.1016/j.irfa.2025.104722

  • CrossRef
  • Google Scholar
  • View reference in article

• 11

  FatokiJ. G.Mora-MongeC. A. (2025). Reconceptualizing supply chain risk management capabilities: a theoretical framework and empirical validation. Supply Chain Manage. Int. J.30, 760–781. doi: 10.1108/SCM-05-2025-0473

  • CrossRef
  • Google Scholar

• 12

  FreundA.JámborZ.NagyJ. (2025). A capability hierarchy for building resilience in multi-actor agri-food supply chains: a digitalisation perspective. J. Manuf. Technol. Manage.37, 19–39. doi: 10.1108/JMTM-12-2024-0718

  • CrossRef
  • Google Scholar
  • View reference in article

• 13

  FuS.LiW. (2026). How does agricultural insurance affect agricultural economic resilience? Evidence from 167 prefecture-level cities in major grain-producing regions. Front. Sustain. Food Syst.10:1793062. doi: 10.3389/fsufs.2026.1793062

  • CrossRef
  • Google Scholar
  • View reference in article

• 14

  GengR.LinS.DaiJ. (2024). Examining customer pressure and green supply chain management in emerging market: an institutional logics perspective. Int. J. Prod. Econ.278:109431. doi: 10.1016/j.ijpe.2024.109431

  • CrossRef
  • Google Scholar
  • View reference in article

• 15

  GongZ.GaoD.GaoH.ZhangZ.ZhaoJ.ShiJ.et al. (2026). Leveraging agricultural supply chain resilience with digital transformation in “a company+farmers” mode. Int. J. Asian Bus. Inf. Manage.17, 1–31. doi: 10.4018/IJABIM.401116

  • CrossRef
  • Google Scholar
  • View reference in article

• 16

  HarsantoB.KasumaningrumY.ArviansyahM. R.SiregarA. Y.PurnomoD.FreddyIskandar, Y.et al. (2025). Leveraging disruptive technologies for food security: a systematic review on agricultural supply chain resilience to climate change. Curr. Res. Food Sci.10:101079. doi: 10.1016/j.crfs.2025.101079

  • Pubmed Abstract
  • CrossRef
  • Google Scholar
  • View reference in article

• 17

  HollingH. S. (1973). Resilience and stability of ecological systems. Annu. Rev. Ecol. Evol. Syst.4, 1–23. doi: 10.1146/annurev.es.04.110173.000245

  • CrossRef
  • Google Scholar
  • View reference in article

• 18

  HuY.GhadimiP. (2025). Developing supply chain resilience with advanced digital technology: a review. IFAC Pap. OnLine59, 2880–2885. doi: 10.1016/j.ifacol.2025.09.484

  • CrossRef
  • Google Scholar
  • View reference in article

• 19

  IvanovD.DolguiA.SokolovB. (2019). The impact of digital technology and industry 4.0 on the ripple effect and supply chain risk analytics. Int. J. Prod. Res.57, 829–846. doi: 10.1080/00207543.2018.1488086

  • CrossRef
  • Google Scholar
  • View reference in article

• 20

  IvanovD.DolguiA.SokolovB. (2022). Cloud supply chain: integrating Industry 4.0 and digital platforms in the “Supply Chain-as-a-Service”. Transp. Res. E Logist. Transp. Rev.160:102676. doi: 10.1016/j.tre.2022.102676

  • CrossRef
  • Google Scholar
  • View reference in article

• 21

  JiangM.PengJ. (2026). Mechanisms and impacts of digital-agricultural integration on grain production resilience. Cogent Food Agric.12:2604870. doi: 10.1080/23311932.2025.2604870

  • CrossRef
  • Google Scholar
  • View reference in article

• 22

  JinJ. L.WangL.WangK.FuX. (2025). Concentrating or dispersing? The double-edged sword effects of supplier concentration on firm financial and innovation performance. J. Bus. Res.186:114946. doi: 10.1016/j.jbusres.2024.114946

  • CrossRef
  • Google Scholar
  • View reference in article

• 23

  JungJ.MaedaM.ChangA.BhandariM.AshapureA.Landivar-BowlesJ. (2021). The potential of remote sensing and artificial intelligence as tools to improve the resilience of agriculture production systems. Curr. Opin. Biotechnol.70, 15–22. doi: 10.1016/j.copbio.2020.09.003

  • Pubmed Abstract
  • CrossRef
  • Google Scholar
  • View reference in article

• 24

  KeX.HuangY.KeL.DongB. (2026). Research on the impact of digital-intelligence integration on agricultural industry resilience: evidence from China. Front. Sustain. Food Syst.9:1677559. doi: 10.3389/fsufs.2025.1677559

  • CrossRef
  • Google Scholar
  • View reference in article

• 25

  KönigM. D.LevchenkoA.RogersT.ZilibottiF. (2022). Aggregate fluctuations in adaptive production networks. Proc. Nat. Acad. Sci.119:e2203730119. doi: 10.1073/pnas.2203730119

  • Pubmed Abstract
  • CrossRef
  • Google Scholar
  • View reference in article

• 26

  KostovčíkM.FedorkoG.MikušováN.MolnárV.NeradilováH.DuriškaM. (2025). Application possibilities of artificial intelligence in the field of transport and logistics. Transp. Res. Procedia87, 66–73. doi: 10.1016/j.trpro.2025.04.109

  • CrossRef
  • Google Scholar
  • View reference in article

• 27

  LahmarA.SiddiquiA. A. (2025). Synergizing supply chain management and digital technologies: a systematic review and framework for resilience and sustainability. J. Syst. Inf. Technol.28, 225–264. doi: 10.1108/JSIT-03-2025-0124

  • CrossRef
  • Google Scholar
  • View reference in article

• 28

  LiY.JinX.SuY.WuG. (2026). Digital technology driving agricultural low-carbon transition: a case study of land transfer and industrial integration in China. Front. Sustain. Food Syst.10:1790377. doi: 10.3389/fsufs.2026.1790377

  • CrossRef
  • Google Scholar
  • View reference in article

• 29

  LiuM.HuangL.ShenY. (2025). Firm's basic research and breakthrough innovation– from the perspective of resource-based view. China Econ. Q. Int.5, 268–285. doi: 10.1016/j.ceqi.2026.01.002

  • CrossRef
  • Google Scholar
  • View reference in article

• 30

  LiuT.ZhuangX.LiM. (2025). How supplier concentration influences labor income share: evidence from China. Financ. Res. Lett.73:106653. doi: 10.1016/j.frl.2024.106653

  • CrossRef
  • Google Scholar
  • View reference in article

• 31

  LuM.LiP.ZhongH. (2019). The new era of development and balance: spatial political economics of new China's regional economy for 70 years. J. Manage. World35, 11–23. doi: 10.3969/j.issn.1002-5502.2019.10.003

  • CrossRef
  • Google Scholar
  • View reference in article

• 32

  LuX.XuX.HouS.WuF. (2026). Forging supply chain resilience: the synergistic effects of AI-enabled capabilities and integration. Socioecon. Plann. Sci.104:102432. doi: 10.1016/j.seps.2026.102432

  • CrossRef
  • Google Scholar
  • View reference in article

• 33

  LuZ.LiN.FengH.DongJ.GouD.XuM. (2026). Climate change risks and green low-carbon development in agriculture: evidence from China on the regulatory role of agricultural insurance and spatial spillover effects. Agriculture16:24. doi: 10.3390/agriculture16010024

  • CrossRef
  • Google Scholar
  • View reference in article

• 34

  LuZ.XuM.ShiL.LeiT. (2025). The criticality of environmental sustainability in agriculture: the decarbonization role of green finance in China. J. Environ. Manage.394:127470. doi: 10.1016/j.jenvman.2025.127470

  • Pubmed Abstract
  • CrossRef
  • Google Scholar
  • View reference in article

• 35

  LuoY.LiuS.ZhangY.ZengM.ZhaoD. (2025). Digital pathways to resilience: assessing the impact of digitalization on agricultural production resilience in China. China Econ. Rev.90:102375. doi: 10.1016/j.chieco.2025.102375

  • CrossRef
  • Google Scholar
  • View reference in article

• 36

  MaJ.NiJ.ZhaoY.WuJ. (2025). Environmental regulation, digital technology application and low-carbon supply chain resilience: evidence from China. J. Environ. Manage.389:126087. doi: 10.1016/j.jenvman.2025.126087

  • Pubmed Abstract
  • CrossRef
  • Google Scholar
  • View reference in article

• 37

  MaX.LvL. (2024). Facilitating the high-quality agricultural development with digital economy: a panel data study of Jiangsu Province from 2013 to 2021. Front. Sustain. Food Syst.8:1424097. doi: 10.3389/fsufs.2024.1424097

  • CrossRef
  • Google Scholar
  • View reference in article

• 38

  MengQ.BaiJ.ShiW. (2018). Customer concentration and enterprise technological innovation: help or hinder? A study based on the Individual characteristics of customers. Nankai Bus. Rev.21, 62–73. doi: 10.3969/j.issn.1008-3448.2018.04.007

  • CrossRef
  • Google Scholar
  • View reference in article

• 39

  NingC.ShunH. (2025). Can digital finance promote the digital technology innovation of enterprises?. Financ. Res. Lett.86:108555. doi: 10.1016/j.frl.2025.108555

  • CrossRef
  • Google Scholar
  • View reference in article

• 40

  Ortiz-de-MandojanaN.BansalP. (2016). The long-term benefits of organizational resilience through sustainable business practices. Strateg. Manag. J.37, 1615–1631. doi: 10.1002/smj.2410

  • CrossRef
  • Google Scholar
  • View reference in article

• 41

  PengH.PangT. (2020). Financing and operational strategies for the contract-farming supply chain under agricultural subsidy policy. J. Ind. Eng. Eng. Manage.34, 155–163. doi: 10.13587/j.cnki.jieem.2020.05.017

  • CrossRef
  • Google Scholar
  • View reference in article

• 42

  PonomarovS. Y.HolcombM. C. (2009). Understanding the concept of supply chain resilience. Int. J. Logist. Manage.20, 124–143. doi: 10.1108/09574090910954873

  • CrossRef
  • Google Scholar
  • View reference in article

• 43

  PuG.BaiY. (2026). Congruence of digital technology adoption level and supply chain resilience: a resource orchestration and re-institutionalization perspective. Supply Chain Manage. Int. J.31, 310–329. doi: 10.1108/SCM-06-2025-0507

  • CrossRef
  • Google Scholar
  • View reference in article

• 44

  QuanT.ZhangH.QuanT.YuY. (2024). Unveiling the impact and mechanism of digital technology on agricultural economic resilience. Chin. J. Popul. Resour. Environ.22, 136–145. doi: 10.1016/j.cjpre.2024.06.004

  • CrossRef
  • Google Scholar
  • View reference in article

• 45

  RahmanM. M.IslamJ.HappyK.DebnathA. (2026). Unveiling the nexus of AI driven business strategies for organizational resilience: testing resource based view theory using SEM-NCA approach. Strateg. Bus. Res.2:100109. doi: 10.1016/j.sbr.2026.100109

  • CrossRef
  • Google Scholar
  • View reference in article

• 46

  RehmanO. U.LiW.WuK. (2025). Supplier concentration and corporate acquisitions: a risk-mitigation perspective. Account. Res. J.39, 243–259. doi: 10.1108/ARJ-07-2025-0257

  • CrossRef
  • Google Scholar
  • View reference in article

• 47

  RuanJ.SunY. (2026). Enhancing agricultural industry chain and supply chain resilience through artificial intelligence empowerment: logic, challenges and paths. J. Nanjing Agric. Univ.26, 24–37. doi: 10.3969/j.issn.1671-7465.2026.01.003

  • CrossRef
  • Google Scholar
  • View reference in article

• 48

  SahooS.KumarA.ManglaS. K.EmrouznejadA. (2026). Enhancing supply chain resilience through digital twin-enabled risk management: the role of logistic service coordination and distribution network efficiency. Int. J. Prod. Econ.297:110016. doi: 10.1016/j.ijpe.2026.110016

  • CrossRef
  • Google Scholar
  • View reference in article

• 49

  ShenY.ZhangX. (2024). Blue sky protection campaign: assessing the role of digital technology in reducing air pollution. Systems12:55. doi: 10.3390/systems12020055

  • CrossRef
  • Google Scholar
  • View reference in article

• 50

  ShenY.ZhangX.WangX. (2026). Innovation-driven sustainable development: analysing the impact of digital technology on synergistic governance of pollution and carbon reduction. J. Innov. Knowl.14:100978. doi: 10.1016/j.jik.2026.100978

  • CrossRef
  • Google Scholar
  • View reference in article

• 51

  ShiL.ZhaoJ.DuX.TanY.LeiT.XuM.et al. (2025). Achieving sustainable green agriculture: analyze the enabling role of data elements in agricultural carbon reduction. Front. Earth Sci.13:1618999. doi: 10.3389/feart.2025.1618999

  • CrossRef
  • Google Scholar
  • View reference in article

• 52

  ShoomalA.JahanbakhtM.ComponationP. J. (2026). An analytical framework for evaluating blockchain and IoT use cases in sustainable supply chains. Supply Chain Anal.13:100198. doi: 10.1016/j.sca.2026.100198

  • CrossRef
  • Google Scholar
  • View reference in article

• 53

  SinghG.RajeshR.DaultaniY.MisraS. C. (2023). Resilience and sustainability enhancements in food supply chains using Digital Twin technology: a grey causal modelling (GCM) approach. Comput. Ind. Eng.179:109172. doi: 10.1016/j.cie.2023.109172

  • CrossRef
  • Google Scholar
  • View reference in article

• 54

  SmithC.RondiE.De MassisA.NordqvistM. (2024). Rising every time we fall: organizational fortitude and response to adversities. J. Manage.50, 1865–1910. doi: 10.1177/01492063231164969

  • CrossRef
  • Google Scholar
  • View reference in article

• 55

  SongH.HanM.ShenL. (2024). The role of artificial intelligence in shaping supply chain resilience: a case study based on Maitron's global aftermarket supply chain management practice. China Ind. Econ.434, 174–192. doi: 10.19581/j.cnki.ciejournal.2024.05.010

  • CrossRef
  • Google Scholar
  • View reference in article

• 56

  SubeeshA.ChauhanN. (2026). Agricultural digital twin for smart farming: a review. Green Technol. Sustain.4:100299. doi: 10.1016/j.grets.2025.100299

  • CrossRef
  • Google Scholar
  • View reference in article

• 57

  SunH.SongY.ZhangR. (2026). Enhancing supply chain resilience: a fit mechanism between key core technology innovations and digital technology applications. Int. J. Inf. Manage.88:103040. doi: 10.1016/j.ijinfomgt.2026.103040

  • CrossRef
  • Google Scholar
  • View reference in article

• 58

  TanY.LiC.SongQ. (2023). Application of blockchain technology in the agricultural product supply chain: theoretical mechanism, development practice and policy implications. Issues Agric. Econ.517, 76–87. doi: 10.13246/j.cnki.iae.20220720.001

  • CrossRef
  • Google Scholar
  • View reference in article

• 59

  TanY.LiuY.XuM.ShiL.HeL.WangX. (2025). Environmental regulation, Spatial Spillovers and high quality Economic Development – the moderating effect of green technology innovation. Pol. J. Environ. Stud.34, 6763–6779. doi: 10.15244/pjoes/192933

  • CrossRef
  • Google Scholar
  • View reference in article

• 60

  TaoF.ZhuP.QiuC.WangX. (2023). The impact of digital technology innovation on enterprise market value. J. Quant. Technol. Econ.40, 68–91. doi: 10.13653/j.cnki.jqte.20230310.010

  • CrossRef
  • Google Scholar
  • View reference in article

• 61

  TukamuhabwaB. R.StevensonM.BusbyJ.ZorziniM. (2015). Supply chain resilience: definition, review and theoretical foundations for further study. Int. J. Prod. Res.53, 5592–5623. doi: 10.1080/00207543.2015.1037934

  • CrossRef
  • Google Scholar
  • View reference in article

• 62

  van der VegtG. S.EssensP.WahlströmM.GeorgeG. (2015). Managing risk and resilience. Acad. Manage. J.58, 971–980. doi: 10.5465/amj.2015.4004

  • CrossRef
  • Google Scholar
  • View reference in article

• 63

  VimalK. E. K.NadeemS. P.SunilS. M.SureshG.SanjeevN.KandasamyJ. (2022). Modelling the strategies for improving maturity and resilience in medical oxygen supply chain through digital technologies. J. Glob. Oper. Strateg. Sourcing15, 566–595. doi: 10.1108/JGOSS-10-2021-0088

  • CrossRef
  • Google Scholar
  • View reference in article

• 64

  WangD.LiuZ.ZhaoZ. (2016). Supply chain relationship and bank loans - an analysis based on supplier/customer concentration. Account. Res.348, 42–49. doi: 10.3969/j.issn.1003-2886.2016.10.006

  • CrossRef
  • Google Scholar

• 65

  WangH.ZhengJ.RenH. (2025). Can smart rural construction improve agricultural products supply chain resilience? Empirical evidence from 31 provinces in China. Front. Sustain. Food. Syst.9:1659012. doi: 10.3389/fsufs.2025.1659012

  • CrossRef
  • Google Scholar
  • View reference in article

• 66

  WangJ.LiuM.ChenY.YuM. (2023). Influencing factors on green supply chain resilience of agricultural products: an improved gray-DEMATEL-ISM approach. Front. Sustain. Food Syst.7:1166395. doi: 10.3389/fsufs.2023.1166395

  • CrossRef
  • Google Scholar
  • View reference in article

• 67

  WangM.LiuY.GengD.DoraM.HillA. (2026). Decarbonisation and firm financial performance: the roles of supply chain transparency and resilience. Int. J. Prod. Econ.296:109988. doi: 10.1016/j.ijpe.2026.109988

  • CrossRef
  • Google Scholar
  • View reference in article

• 68

  WangW.JiaJ. (2025a). Blockchain-driven tripartite game analysis of agricultural supply chain finance strategy. Sustain. Futures10:101253. doi: 10.1016/j.sftr.2025.101253

  • CrossRef
  • Google Scholar
  • View reference in article

• 69

  WangW.JiaJ. (2025b). Study on the driving mechanism and risk hedging effect of economic policy uncertainty on digital transformation of agricultural enterprises. Front. Sustain. Food Syst.9:1565343. doi: 10.3389/fsufs.2025.1565343

  • CrossRef
  • Google Scholar
  • View reference in article

• 70

  WangY.LongY.WangJ. (2025). Does the innovation-driven digital economy improve the resilience of industrial and supply chains?J. Innov. Knowl.10:100733. doi: 10.1016/j.jik.2025.100733

  • CrossRef
  • Google Scholar
  • View reference in article

• 71

  XingY.ZhangJ. Z.WangX. (2026). Digital innovation and transformation process in business growth: a systematic literature review and research agenda. Technovation151:103396. doi: 10.1016/j.technovation.2025.103396

  • CrossRef
  • Google Scholar
  • View reference in article

• 72

  XuB.XuH.WangY. (2025). Digital finance, technology business incubation, and industrial chain resilience. Financ. Res. Lett.86:108834. doi: 10.1016/j.frl.2025.108834

  • CrossRef
  • Google Scholar
  • View reference in article

• 73

  XuJ. (2024). Spatial spillover and threshold effects of digital rural development on agricultural circular economy growth. Front. Sustain. Food Syst.8:1337637. doi: 10.3389/fsufs.2024.1337637

  • CrossRef
  • Google Scholar
  • View reference in article

• 74

  XuM. (2025). Can the ecological protection red line policy promote food security? – based on the empirical analysis of land protection in China. Front. Environ. Sci.13:1654217. doi: 10.3389/fenvs.2025.1654217

  • CrossRef
  • Google Scholar
  • View reference in article

• 75

  XueX.GaoJ.XuS. (2025). The impact and mechanism of digital economy on agricultural development resilience: evidence from China. Front. Sustain. Food Syst.9:1574461. doi: 10.3389/fsufs.2025.1574461

  • CrossRef
  • Google Scholar
  • View reference in article

• 76

  YangY.LiY.LiangX. (2025). The role of data trading platforms (DTPs) in digital technology innovation: mechanisms and evidence from China. J. Policy Model.47, 1372–1396. doi: 10.1016/j.jpolmod.2025.09.003

  • CrossRef
  • Google Scholar
  • View reference in article

• 77

  YuZ.LiuJ. (2025). The digital revolution in banking: unpacking risk management in the age of transformation. Int. Rev. Econ. Financ.103:104444. doi: 10.1016/j.iref.2025.104444

  • CrossRef
  • Google Scholar
  • View reference in article

• 78

  YuanC.XiaoT.CengC.ShengY. (2021). Digital transformation and division of labor between enterprises: vertical specialization or vertical integration. China Ind. Econ.402, 137–155. doi: 10.3969/j.issn.1006-480X.2021.09.008

  • CrossRef
  • Google Scholar
  • View reference in article

• 79

  YuanY.GuoX.ShenY. (2024). Digitalization drives the green transformation of agriculture-related enterprises: a case study of A-share agriculture-related listed companies. Agriculture14:1308. doi: 10.3390/agriculture14081308

  • CrossRef
  • Google Scholar
  • View reference in article

• 80

  ZhangB.ChengL.ChenS.ZengH. (2025). How does digital economy affect food system resilience? A quasi-natural experiment on China's smart city pilot policy using double machine learning. Front. Sustain. Food Syst. 9:1672125. doi: 10.3389/fsufs.2025.1672125

  • CrossRef
  • Google Scholar
  • View reference in article

• 81

  ZhangJ. (2026). Supply chain digitalization and disruption risk: the roles of resilience and financing constraints. Financ. Res. Lett.93:109623. doi: 10.1016/j.frl.2026.109623

  • CrossRef
  • Google Scholar
  • View reference in article

• 82

  ZhaoX.XuY.XuY. (2023). How does digital collaboration affects supply chain resilience: an analysis based on resource and relationship integration. Ind. Econ. Rev.14, 24–42. doi: 10.14007/j.cnki.cjpl.2023.05.002

  • CrossRef
  • Google Scholar
  • View reference in article

• 83

  ZhouJ.LiuW.LiJ. (2024). Performance implications of a corporate political strategy in Chinese firms: state versus market logics in different institutional settings. Long Range Plann.57:102439. doi: 10.1016/j.lrp.2024.102439

  • CrossRef
  • Google Scholar
  • View reference in article

• 84

  ZhouY.HeW.TangH.LiuY. (2026). From complexity to strength: how digital transformation and innovation enhances agricultural supply chain resilience?J. Innov. Knowl.15:100985. doi: 10.1016/j.jik.2026.100985

  • CrossRef
  • Google Scholar
  • View reference in article