Do Industrial Robots Mitigate Supply Chain Risks? Evidence from Firm-Level Text Analysis

Abstract

Building a resilient and efficient supply chain system is critical for sustaining firm operations in an increasingly uncertain global environment. This study examines whether the firm-level exposure to industry-wide robot penetration mitigates firm-level supply chain risks. By adopting Bartik’s instrumental variable approach to decompose industry-level robot data to the firm level (from the International Federation of Robotics, IFR), and using a novel text-mining-based supply chain risk index, constructed via a tailored “supply chain risk” dictionary, to quantify sentences containing both keywords from firms’ annual report MD&A sections, we apply a fixed effects model, and find that robot adoption significantly reduces supply chain risk by enhancing firms’ discourse power and improving supply chain coordination. The effect is more pronounced in firms with higher capital intensity, greater international exposure, stronger regulatory oversight, and better ESG (Environmental, Social, and Governance) performance. By integrating automation adoption with supply chain risk management, this study extends the literature on production economics and supply chain resilience. Our findings reveal that industrial robots, beyond enhancing productivity, function as a risk-mitigating technology that strengthens supply chain stability and operational continuity in volatile global production networks.

1. Introduction

Supply chain risk is a phenomenon that has garnered widespread attention, especially in emerging markets. Stakeholders’ demands for responsible supply chain management have intensified [1]. A secure supply chain can ensure that firms will not be affected by the adverse actions of suppliers and customers, and can ensure the value maximization of the stakeholders.

However, global supply chains have become increasingly vulnerable to disruptions triggered by geopolitical tensions, trade conflicts, pandemics, and natural disasters [2,3]. These risks manifest across multiple dimensions—including supply, demand, operational, and informational flows—and often propagate in cascading patterns throughout entire networks. For instance, some Chinese high-tech companies, such as Huawei, have been included in the list of entities by the U.S. Department of Commerce (https://www.bis.doc.gov/index.php/allarticles/17-regulations/1555-addition-ofcertain-entities-to-the-entity-list-final-ruleeffective, accessed on 16 May 2019). As a result, these companies are unable to obtain key technologies and products, resulting in a huge impact on their global supply chain system. In addition, disruptions in a single supplier’s operations can trigger cascading effects across the entire supply chain; for example, the well-known mobile power brand Romas was shut down due to the suspension of product certification, and its upstream supplier Zhuoyi technology was seriously affected.

Against this backdrop, traditional supply chain risk management (SCRM) frameworks, while comprehensive, have struggled to keep pace with the velocity and interconnectedness of modern risks, underscoring the urgency of innovative technological interventions [4].

One emerging response has been the adoption of automation and digital technologies—including artificial intelligence (AI), robots, and advanced information systems—as enablers of supply chain resilience. Automation not only substitutes human labor in repetitive or high-risk activities but also strengthens firms’ analytical and decision-making capabilities. For instance, in the South African FMCG (Fast-Moving Consumer Goods) industry, AI-enabled information systems were found to enhance risk sensing, information integration, and adaptive responses, thereby reducing vulnerability to disruptions [5]. Similarly, in healthcare supply chains, digital transformation through the Internet of Medical Things (IoMT) has expanded the monitoring, traceability, and conformity assessment frameworks that safeguard security and resilience [6]. These findings collectively demonstrate that automation extends beyond cost efficiency—it plays a pivotal role in strengthening supply chain robustness and adaptability.

Another relevant dimension is the role of automation in supporting multi-level resilience strategies. During the COVID-19 crisis, the dairy sector demonstrated how resilience capacities could be developed at macro (industry), meso (supply chain), and micro (firm) levels [7]. Other industries are also intervening with transformative supply chains at the three levels mentioned above. For example, at the macro (industry) level, research has found that intelligent technology plays an important role in the digitization of industry supply chains [8]. From the mesoperspective, companies are pushed to innovate their supply chains to implement a circular supply chain [9]. At the micro (firm) level, the interaction between blockchain-based smart contracts and big data analytics creates value for the supply chains of various types of enterprises [10].

Automation and digitization facilitated the coordination of information, the stabilization of production, and the continuity of distribution, allowing firms to mitigate severe disruption effects [11]. These insights highlight that technological adoption supports resilience not only within individual firms but also across wider supply networks.

Theoretical perspectives further illuminate the mechanisms through which automation creates resilience. From a resource-based and resource-dependence view, technologies such as blockchain and circular economy practices provide firms with the localization, agility, and digitization (L-A-D) capabilities required to buffer against external shocks [12]. Automation can thus be conceptualized as a capability-enhancing investment that reduces dependency risks, improves transparency, and facilitates adaptive resource reallocation. In practical terms, this means that firms adopting automation are better positioned to withstand systemic disruptions, meet regulatory compliance requirements, and pursue sustainability goals simultaneously.

Prior research has explored various aspects of supply chain risk, including the definition of supply chain risk [13], factors causing supply chain risk [14], the consequences of supply chain risk on firms’ performance [15,16], and measures to reduce supply chain risk [17,18].

A notable limitation in the existing literature is the absence of reasonable and effective quantitative methodologies for assessing supply chain risk. Generally, theoretical analysis, case discussion, or questionnaire methods are used for research [17]. Therefore, there is an urgent need to further explore the development of measurement indicators and systems to more accurately capture firms’ supply chain risks. Furthermore, existing research has yet to fully investigate how to mitigate supply chain risks; most studies primarily focus on the supply chain’s inherent characteristics—such as supply chain concentration and digitalization level—while analyses from the perspective of productivity change remain scarce. Drawing from the resource dependence theory and collaborative governance theory [19,20], we argue that the firm-level exposure to industry-wide robot penetration reshapes firms’ dependence on supply chain partners and collaborative governance capabilities, thereby implying the potential mitigation of supply chain risks. Accordingly, this study addresses the following research question: Does, and how does, the firm-level exposure to industry-wide robot penetration affect the supply chain risk of firms?

Our research constructed a measurement index for the penetration of robots at the enterprise level in China, and used text analysis methods to evaluate the level of supply chain risk. We empirically tested the impact of industrial robot use on enterprise supply chain risk using the fixed effects model. Using data from Chinese listed companies from 2007 to 2019, our research found the following: (1) the firm-level exposure to industry-wide robot penetration can significantly reduce supply chain risks for enterprises, and this conclusion still holds true after endogeneity testing and a series of robustness tests; (2) based on the resource dependence theory and organizational synergy theory, we found that the negative effects of industrial robot use on supply chain risk have two mechanisms: one is increasing the discourse power of enterprises in the supply chain, and the other is promoting internal and external supply chain coordination of enterprises; and (3) the heterogeneity analysis results indicate that the firm-level exposure to industry-wide robot penetration has a more significant impact on supply chain risk in enterprises with strong regulation, low capital intensity, low internationalization, and high ESG scores. Academically, this study fills gaps by linking industrial robots to supply chain risk mitigation, innovates text-mining-based risk measurement, and clarifies mechanisms (discourse power, and coordination), thus bridging production economics and supply chain management. Practically, it encourages firms and policymakers to regard robots as powerful tools to enhance supply chain resilience.

Our study aims to make several theoretical contributions. First, by taking the supply chain—a networked organizational collaboration model—as the research starting point and focusing on the impact of industrial robots on supply chain risk, this study enriches the literature on the micro-level industrial effects of industrial robot adoption. Previous research has predominantly centered on the employment and labor market effects of industrial robots [21,22,23,24], as well as their influences on firm performance [25,26,27]. In contrast, our study extends the research boundary into two key aspects. On the one hand, it addresses how technological upgrading drives the optimization of firms’ organizational collaboration modes, including supply chain networks, thereby providing insights for the improvement of corporate governance systems. On the other hand, the findings of this study essentially reveal the pathway to constructing a robust defense for industrial and supply chain security and enhancing firms’ core competitiveness through the application of automation and intelligent technologies, with industrial robots as a representative.

Second, through a novel analytical perspective, this study elucidates the mechanism by which industrial robot adoption contributes to mitigating firm supply chain risks. It identifies two primary pathways: the enhancement of firms’ supply chain discourse power, and the strengthening of internal and external supply chain coordination. By establishing a connection between new technology implementation and the optimization of supply chain governance systems, our research further explicates how the technological revolution and industrial organizational transformation—driven by the emerging productivity of industrial robots—can streamline supply chain cycles, safeguard supply chain security and stability, and ultimately advance industrial transformation and upgrading as well as high-quality firm development.

Finally, this study employs text mining and analytical techniques to measure supply chain risk, utilizing a comprehensive supply chain risk dictionary constructed for this purpose and leveraging information disclosed in firms’ annual reports. This measurement approach represents an advancement and extension of existing methodologies for supply chain risk measurement in the literature, thereby facilitating more accurate capture and the identification of firms’ supply chain risk information.

2. Literature Review

2.1. Industrial Robots

Industrial robots are widely recognized for improving efficiency, standardization, and flexibility in production and logistics [28,29]. Applications extend from automated warehouses [30] to collaborative hybrid systems [29]. Robots also feature in debates on reshoring and labor substitution [31]. However, most studies frame robotics within productivity and sustainability discussions [32,33], leaving their role in supply chain risk (SCR) mitigation largely untested. The implicit assumption is that efficiency gains translate into resilience, but the empirical validation of this link remains scarce.

2.2. Supply Chain Risk

Supply chain disruptions are the primary source of supply chain risk, and the reasons for supply chain disruptions include supplier unreliability, transportation bottlenecks, demand volatility, and geopolitical shocks [3,34]. These risks are not isolated; they propagate through supply, demand, operational, and informational flows, often magnifying network vulnerabilities [2]. Their economic consequences extend beyond short-term financial losses to include reputational erosion and weakened competitive advantage [5]. While the literature has established the multidimensionality of SCR, it tends to catalog risks rather than develop robust metrics to quantify their firm-level impact. This limits the comparability of findings and constrains managerial insights.

Prior research also focuses on how supply chain risks are mitigated. Scholars emphasize diversification, redundancy, and digitalization as key mitigation strategies. Frameworks stress flexibility, agility, and collaboration as resilience enablers [4], while blockchain and circular economy practices are argued to improve transparency and reduce dependence on fragile nodes [12]. Artificial intelligence and advanced information systems are highlighted for their role in sensing, predicting, and responding to risks [5,6]. Moreover, some Industry 4.0 technologies are being utilized to manage supply chain disruptions Yet, these approaches are costly, context-dependent, and often lack scalability, raising questions about their broader applicability. The literature identifies “what” firms should do but provides limited evidence on “how” different technologies vary in effectiveness across industries and contexts.

2.3. The Effect of Industrial Robots on Supply Chain

Most researchers concentrate their interest in the role of industrial robots in supply chain management. In particular, new technology, including robots, plays a vital role in enhancing supply chain cooperation and supply chain flexibility [35,36]. In the process of deploying new technologies, a method that is conducive to fully achieving the benefits of supply chain integration is needed [37]. Meanwhile, a growing number of studies concentrate on the role of industrial robots in improving firms’ operational efficiency, flexible production, and risk response capabilities, considering the fact that industrial robots are deployed on a large scale in the manufacturing industry.

Some of the literature further explores other factors that could affect the effectiveness of industrial robots’ adoption, for instance, the strategic positioning and development path of a firm itself [38]. Furthermore, the spillover effect of supply chain when new technology is adopted in production has also been studied. In a highly integrated modern supply chain, information, products, and technologies between firms are intertwined, and the technological changes in firms themselves can drive the upgrading of related firms’ capabilities and value reshaping [39]. For example, the digital manufacturing model of core firms may drive system upgrades for suppliers or incentivize customers to improve operational efficiency [40].

Taken together, three gaps emerge. First, the literature rarely isolates industrial robots’ contribution to risk reduction, focusing instead on broader automation or digitalization. Second, evidence is fragmented across industry-specific studies, limiting its generalizability. Third, measurement challenges constrain rigorous testing of automation’s effects on SCR. This study addresses these gaps by constructing a firm-level SCR measure from textual disclosures and examining the effects of industrial robot adoption. In doing so, it repositions robots not only as productivity tools but also as potential enablers of supply chain resilience and governance.

The review highlights that while supply chain risk is increasingly complex and consequential, existing mitigation strategies remain fragmented and often lack scalability. Industrial robots stand out as a promising, yet underexplored, technological intervention. Prior research acknowledges their potential to enhance efficiency, coordination, and flexibility, but has not systematically examined whether these advantages translate into reduced risk exposure. At the same time, the persistent measurement gap in supply chain risk constrains empirical assessments of robotics as a resilience enabler.

Against this backdrop, the study investigates the role of industrial robots in mitigating supply chain risks. Specifically, we ask the following: Do industrial robots reduce firms’ supply chain risk exposure, and if so, through what mechanisms? Based on the research approach of Yu et al. [41], we develop a firm-level risk measure based on the text-mining of corporate disclosures and combine it with data on robot adoption. Building on the prior literature, we propose that robots affect risk through enhanced supply chain power and improved coordination. These insights form the basis for the study’s research questions and hypotheses.

3. Hypotheses Development

3.1. Positive Effect of Industrial Robots

Over-reliance on a single supplier or customer—signifying excessive concentration within supply chains—may exacerbate supply chain risks. The risk-mitigating effect of industrial robot adoption manifests primarily in the integration of automated and intelligent technologies with firms’ production operations, enabling accurate market information acquisition, operational efficiency improvements, and enhanced supply chain collaboration to diversify supply chain risks.

On the supply side, firms can rapidly evaluate market feedback on products and enhance product quality through interactions with intelligent environments. With the support of emerging technologies, firms can achieve real-time communication and information sharing with suppliers while establishing partnerships with a broader range of potential suppliers [42].

On the demand side, firms’ adoption of automated and intelligent technologies enhances demand response flexibility, alleviating risks of supply–demand mismatches, including the rapid conversion of production specifications and the dynamic adjustment of capacity elasticity [43]. In the alliance environment established by upstream and downstream firms in the supply chain, the use of robots helps firms to develop customer resources through sharing business networks, expanding customer positioning, and diversifying business risks, forming a diffusion effect of flow collaboration, thereby reducing customer risks.

Therefore, we hypothesize the following:

H1. 

The firm-level exposure to industry-wide robot penetration can reduce supply chain risk of firms.

3.2. Mechanism of Supply Chain Discourse Power

The supply chain constitutes a dynamic system linking upstream and downstream firms, necessitating an analysis of how industrial robot adoption influences supply chain risks from the perspective of inter-firm relationships across the supply chain. We argue that the firm-level exposure to industry-wide robot penetration may affect supply chain risks by increasing the power of supply chain discourse and strengthening internal and external coordination within the supply chain. Firstly, the firm-level exposure to industry-wide robot penetration can improve the production efficiency of firms and drive new firms to enter. By endowing firms with diversified supplier and customer networks to enhance their supply chain discourse power, supply chain risks can be mitigated. Secondly, the firm-level exposure to industry-wide robot penetration can promote cooperation and information sharing among firms on the supply chain, strengthen the external coordination of the supply chain, and thus affect supply chain risks. Thirdly, the firm-level exposure to industry-wide robot penetration can promote communication between various departments within the firm and can connect production with market demand in a timely manner, thereby strengthening internal coordination in the supply chain and playing a positive role in reducing supply chain risks.

The discourse power in the supply chain is a comprehensive reflection of a company’s position and bargaining power in the supply chain. Due to the fact that the more dispersed the supply chain, the stronger the bargaining power of firms, we use the supply chain concentration as a measure of the supply chain discourse power.

The resource dependency theory suggests that organizations need to alleviate external constraints by reducing dependence on a single resource [19]. Therefore, the more dispersed the supply chain, the more choices and opportunities for cooperation firms can have, and they also have more bargaining power when facing suppliers and customers, thus occupying a more advantageous position in the supply chain and reducing supply chain risks. The firm-level exposure to industry-wide robot penetration can reduce the concentration of a company’s supply chain and increase its discourse power in the supply chain through the following four methods:

(1)

The large-scale, firm-level exposure to industry-wide robot penetration by companies indicates their enhanced production capacity, which can convey signals of stable operation to the outside world and attract more supply chain companies to cooperate with them [44].

(2)

The research on the application of industrial robots in intelligent manufacturing systems shows that the intelligent production scheduling system built on the industrial Internet platform can realize the dynamic optimal configuration of manufacturing resources and significantly improve the production adaptability of standardized parts and modular semi-finished products by deeply integrating into the midstream link of the production process. This technology empowerment model can not only promote the evolution of the supply chain network structure towards a collaborative model with multiple nodes and wide coverage, but also effectively reduce the concentration index of the supply chain network and enhance the system’s ability to resist risks through decentralized production layout [45,46].

(3)

According to the theory of economies of scale, the firm-level exposure to industry-wide robot penetration can expand production scale, reduce labor costs, improve production efficiency, and increase operating profits, thereby attracting new firms to enter supply chain-related industries [47], creating more opportunities for cooperation, and thus enhancing the discourse power of firms in the supply chain.

(4)

The widespread application of intelligent manufacturing technologies such as robots can facilitate the rapid iteration and upgrading of firm products, promote the exchange and sharing of market information among suppliers or customers, help firms discover new partners in a timely manner according to market supply and demand changes, greatly expand market boundaries, and improve the supply chain matching of firm main business [26,48], thereby reducing the dependence of firms on a single supply chain partner.

Hence, we hypothesize the following:

H2. 

The firm-level exposure to industry-wide robot penetration can enhance a company’s discourse power in the supply chain, thereby reducing supply chain risks.

3.3. Mechanism of Supply Chain Collaboration

The theory of collaborative governance suggests that information sharing and cost sharing among organizational members can achieve maximum benefits [16]. Based on the theory of collaborative governance, our research believes that the firm-level exposure to industry-wide robot penetration can enhance internal and external coordination in the supply chain to jointly resist supply chain risks.

Firstly, industrial robots can enhance external coordination in the supply chain for the following reasons. (1) The large-scale deployment of industrial robots improves the accuracy and precision of firm production [26,49], stimulates firms to break through existing innovation boundaries [50], ensures the standardization and controllability of the production process, improves the production efficiency and product quality, enhances risk resistance by promoting the resilience of its upstream and downstream partners [41], and motivates chain firms to establish strategic alliances to achieve deep cooperation and better resist potential risks. (2) The intelligent transformation of industrial robots in the production process effectively eliminates information asymmetry between upstream and downstream firms, facilitates timely response to the personalized needs of upstream and downstream firms, deeply integrates the supply chain network, effectively improves collaborative efficiency, and reduces supply and demand coordination costs. (3) Industrial robots have strengthened information sharing and communication cooperation among firms on the supply chain, enhancing mutual trust between firms, suppliers, and customers. This has provided sufficient impetus for promoting cost sharing in the supply chain. At the same time, the firm-level exposure to industry-wide robot penetration has positive externalities, driving the transformation of production management models for on-chain firms and even the entire industry. As beneficiaries of the positive externalities of industrial robot use, on-chain firms tend to internalize this social benefit [51].

Secondly, the firm-level exposure to industry-wide robot penetration can enhance internal coordination within the supply chain. The reasons are as follows. (1) Industrial robots have achieved intelligence and automation in the production process, making it easy to quickly, and in a timely manner, capture and connect with market demand, accelerate product circulation, and improve inventory turnover speed [27]. (2) The firm-level exposure to industry-wide robot penetration not only enhances information sharing among firms on the chain, but also strengthens information sharing within firms, promoting smooth information flow between various business processes such as production, warehousing, logistics, and sales [52], in order to make timely adjustments to changes in external demand and reduce ineffective resource consumption.

Based on this, we propose the following hypothesis:

H3. 

The firm-level exposure to industry-wide robot penetration can improve internal and external collaboration in the supply chain, thereby reducing supply chain risk.

4. Data and Empirical Methodology

4.1. Data Collection

The sample studied in our study is mainly collected through the following methods. Firstly, supply chain risk data refers to sentences information that includes both “supply chain” and “risk” keywords publicly disclosed in the Management Discussion and Analysis (MD&A) text of the firm (specific measurement methods are described in the following text). Secondly, the industrial robot data is calculated based on the industry level industrial robot data published by the International Federation of Robotics (IFR) in China. Finally, the basic information of the listed companies comes from the CSMAR database. From the initial dataset, sample observations were excluded if they corresponded to firms with abnormal listing statuses (e.g., ST and PT) or contained missing values for key variables. Additionally, all continuous variables were winsorized at the 1% level to mitigate the influence of outliers. We finally obtain 18,235 observations with a time span of 2007–2019.

4.2. Empirical Methodology

Benchmark Model

$$Chainris{k}_{ijt}={\alpha }_0+\alpha Robo{t}_{ijt}+\beta Contro{l}_{ijt}+{\mu }_i+{\eta }_p+{\nu }_{jt}$$

(1)

Among them, Chainrisk_ijt is the dependent variable, Robot_ijt is the explanatory variable, and i, j, and t represent the firm, industry, and year, respectively. We select a series of control variables at the firm level, including company size, return on assets, property rights, capital structure, etc. (see Table 1 for details). μ_i represents the firm fixed effects, η_p represents the province fixed effects, and ν_jt represents industry–year fixed effects. In addition, the benchmark model controls for firm fixed effects, industry x year fixed effects, and province fixed effects, respectively.

Table 1. Variable definitions.

	Variable Name	Variables Symbol	Definition
Dependent variable	Supply chain risk	Chainrisk	See definition below
Independent variable	Firm-level exposure to industry-wide robot penetration rate	Robot	See definition below
Control variables	Firm size	Size	Ln (total asset)
	Return on assets	Roa	Total profit/total asset
	Nature of property rights	Soe	Soe = 1 if it is a state-owned firm, otherwise Soe = 0
	Debt-to-asset ratio	Lev	Total debt/total asset
	Proportion of fixed assets	FixedAsset	Fixed asset/total asset
	Inventory ratio	Inventory	Net inventory/total asset
	Sales growth	growth	The difference between the current year’s operating income and the previous year’s operating income divided by the previous year’s operating income
	Date of establishment	age	Add 1 to the logarithm of the difference between the observation year and the establishment year
	Cash ratio	cash	Net increase in cash and cash equivalents/total asset
	Ownership concentration	Stock	Shareholding ratio of the top ten shareholders
	Annual report text tone	Attitude	(Positive vocabulary − negative vocabulary)/annual report vocabulary
	Audit opinion	Audit	Audit = 1 if it is standard unqualified opinion, otherwise Audit = 0

Table 1 and Table 2, respectively, present the definitions of the main variables and descriptive statistical results. It can be seen that the average supply chain risk is 0.061, the maximum value is 0.144, the minimum value is 0.008, and the standard deviation is 0.030, indicating that the supply chain risk levels of different firms vary greatly. The maximum penetration rate of robots is 3.098, the minimum is 0, and the standard deviation is 0.437. Therefore, there is a significant difference in the number and installation density of robots used by different firms.

Table 2. Descriptive statistics of main variables.

	Observations	Mean	Sd.	Min	Max
Chainrisk	18,235	0.061	0.030	0.008	0.144
Robot	18,235	0.162	0.437	0.000	3.098
Size	18,235	22.157	1.271	19.723	26.007
Roa	18,235	0.039	0.058	−0.195	0.211
Soe	18,235	0.467	0.499	0.000	1.000
Lev	18,235	0.445	0.202	0.056	0.889
FixedAsset	18,235	0.245	0.165	0.004	0.722
Inventory	18,235	0.151	0.128	0.000	0.681
growth	18,235	0.182	0.420	−0.518	2.732
age	18,235	2.793	0.357	1.609	3.466
cash	18,235	0.001	0.085	−1.108	0.970
Stock	18,235	0.570	0.154	0.223	0.903
Attitude	18,235	0.002	0.010	−0.021	0.026
Audit	18,235	0.974	0.160	0.000	1.000

4.3. Variables Construction

4.3.1. Dependent Variables: Supply Chain Risk (Chainrisk)

The difficulty in measuring supply chain risks lies in the fact that companies often fail to fully disclose detailed information about their suppliers and customers in the same chain due to confidentiality motives, which makes it difficult to accurately identify their supply chain risks using listed company annual reports. In addition, the supply chain risk of firms is simultaneously affected by both primary and multi-level supply chains [49]. Based on the above factors, we use text analysis methods to capture supply chain risk information through the following path.

The first step was to construct relevant topic dictionaries separately. The screening of keywords such as “supply chain” and “risk” is the foundation of text analysis. In terms of the selection of vocabulary related to the theme of “supply chain”, we first compiled and collected the most frequently used vocabulary [18] from the latest version of the textbook “Supply Chain Management” (translated by Chen et al. (2022) published by the RUC Press [53]). Then, combined with policy-related expressions related to “supply chain” and the keywords selected by others, the previous dictionary was supplemented and improved. After obtaining a preliminary dictionary on the topic of “supply chain”, we randomly selected the MD&A text from 150 annual reports of listed companies to check whether the dictionary covers all categories of “supply chain” business, and further supplemented the missing parts. The “risk”-themed dictionary was constructed based on existing research [18]. In addition, our study also introduces the positive emotion vocabulary dictionary compiled by Dalian University of Technology to prevent expressions such as “optimistic about supply chain risks” from being mistakenly identified as supply chain risks.

The second step was to download the MD&A text in bulk and clean it. Hassan et al. (2023) and Ersahin et al. (2024) believe that the management analysis and discussion in the annual reports of listed companies are a summary of the company’s past operating conditions and an assessment of future development trends, which can accurately reflect the company’s perception and evaluate the risks it faces [18,33]. We refer to the ideas of Hassan et al. (2023) [33] and use the Management Discussion and Analysis (MD&A) chapter as the source of the text. Firstly, we downloaded the MD&A text of listed companies from 2007 to 2019 in bulk, deleted irrelevant information such as letters, numbers, and punctuation marks, and then used the jieba segmentation module in Python 3.13 to segment all sentences.

The third step was to measure supply chain risks. Firstly, we counted the number of sentences in each text that contained both “supply chain” and “risk” keywords, but did not contain positive words, and record them as SenChainrisk_it. Then, firm supply chain risk could be defined as:

$$Chainris{k}_{it}={\displaystyle \frac{SenChainris{k}_{it}}{N_{it}}}$$

(2)

Among them, N_it is the total number of sentences contained in each MD&A text.

4.3.2. Independent Variables: Firm-Level Exposure to Industry-Wide Robot Penetration Rate (Robot)

IFR only released robot data for different industries in China’s manufacturing industry. To further calculate the penetration rate of robots at the firm level, we used a comprehensive reference method to calculate the penetration rate of robots at the firm level through the following steps:

Firstly, it was necessary to calculate the penetration rate of robots at the industry level:

$$IndRobo{t}_{jt}={\displaystyle \frac{Stoc{k}_{jt}}{Employe{e}_{j,t=2010}}}$$

(3)

Among them, Stock_jt is the robot stock of industry j in year t, and Employee_j,t=2010 is the employment number of industry j in China’s base period. Due to some discrepancies between the industry classification in China by IFR and the actual situation, we matched industries with IFR based on the industry classification method in the National Economic Industry Classification. Next, we needed to calculate the penetration rate of robots at the firm level:

$$Robo{t}_{ijt}=IndRobo{t}_{jt}\times {\displaystyle \frac{PW{P}_{j,t=2012}}{Al{l}_{t=2012}}}$$

(4)

${\displaystyle \frac{PW{P}_{j,t=2012}}{Al{l}_{t=2012}}}$ denotes the ratio of the proportion of employees in the production department of firm i in industry j during the base period to the median proportion of employees in the production department of all firms in the same year. This weight utilizes changes in industry technology characteristics to decompose the penetration of robots at the industry level into the firm level [53].

5. Empirical Results

5.1. Main Results

Table 3 reports the benchmark regression results of the impact of firm-level exposure to industry-wide robot penetration rate on the firm supply chain risk. The first column of Table 3 did not include any control variables or fixed effects. The second column added control variables for regression based on the first column. The third column controlled for province, industry * year, and firm fixed effects based on the first column. The fourth column included all control variables and fixed effects in the model. The results showed that after gradually strengthening the control, the coefficients were significantly negative at the 1% level. The coefficient of “Robot” in column (4) was −0.009, indicating that for every 1 unit/100 employees increase in the penetration rate of industrial robots per 100 people in the firm, its own supply chain risk would decrease by 0.9%. From this result, it can be seen that the firm-level exposure to industry-wide robot penetration by firms can significantly reduce supply chain risks, which may be due to the combined effects of increasing supply chain discourse power (reducing supply chain concentration) and strengthening internal and external coordination of the supply chain.

Table 3. Benchmark regression results.

Variables	Chainrisk	Chainrisk	Chainrisk	Chainrisk
	(1)	(2)	(3)	(4)
Robot	−0.006 *** (0.001)	−0.006 *** (0.001)	−0.012 *** (0.003)	−0.009 *** (0.003)
Controls	No	Yes	No	Yes
Province fixed effects	No	No	Yes	Yes
Industry-year fixed effects	No	No	Yes	Yes
Firm fixed effects	No	No	Yes	Yes
Adjusted R2	0.007	0.100	0.370	0.401
Observations	18,235	18,235	18,063	18,063

Note. *** p < 0.01. The robust standard errors in parentheses are clustered at the firm level in this article, as shown in the table below.

5.2. Endogeneity

Our research adopts the instrumental variable method to address potential endogeneity issues arising from omitted variables. Based on the Bartik instrumental variable idea, the firm-level exposure to the industry-wide robot penetration rate of Chinese firms, calculated at the German industry level, is selected as the instrumental variable [54]. The rationality of this selection method lies in the following: firstly, the stock characteristics and development trends of industrial robots in Germany are similar to those in China; secondly, the bilateral trade volume between China and Germany accounts for a relatively low proportion of China’s total import and export trade, and China’s manufacturing industry is not significantly impacted only by Germany. The use of German industrial robots is difficult to directly relate to China’s supply chain risks and meet exogenous requirements. Specifically, drawing on the ideas of Goldsmith–Pinkham et al. (2020) [55]:

$${IndRobo{t}_{jt}}^{Ger}={\displaystyle \frac{{Stoc{k}_{jt}}^{Ger}}{Employe{e}_{j,t=2012}}}$$

(5)

Among them, IndRobot_jt^Ger is the stock of industrial robots in the j industry in Germany in year t, and Employee_j,t=2012 is the base employment number in the j industry in Germany. The ratio of the two represents the penetration rate of robots at the industry level in Germany. The instrumental variables in our study are as follows:

$$IVRobo{t}_{ijt}={IndRobo{t}_{jt}}^{Ger}\times {\displaystyle \frac{PW{P}_{j,t=2012}}{Al{l}_{t=2012}}}$$

(6)

Table 4 reports the regression results of the instrumental variable method. The first column of Table 4 replaces the explanatory variables in the benchmark regression with instrumental variables for regression. The sign and significance of the results are consistent with the benchmark regression, proving that the firm-level exposure to industry-wide robot penetration does significantly reduce supply chain risks for firms. Subsequently, we use the penetration rate of robots at the German industry level as an instrumental variable and conduct regression using the two-stage least squares (2SLS) method. The second column of Table 4 shows that the regression results of the first stage of the 2SLS method are significantly positive, indicating a significant positive correlation between the core explanatory variable and the instrumental variable. The second stage regression results in column (3) of Table 4 show that the firm-level exposure to industry-wide robot penetration can significantly reduce the supply chain risk of firms. The penetration rate of robots per 100 people in firms increases by one unit, and the average supply chain risk is reduced by 4.4%. At the same time, the instrumental variable passed the unidentifiable test and weak instrumental variable test, indicating that the research conclusion did not change while using instrumental variables to alleviate endogeneity issues.

Table 4. Endogeneity test results.

Variables	Minimalist Regression	First-Stage Regression	Second-Stage Regression
	Chainrisk	Robot	Chainrisk
	(1)	(2)	(3)
IV	−0.008 *** (0.003)	0.002 *** (0.000)
Robot			−0.044 *** (0.016)
Controls	Yes	Yes	Yes
Province fixed effects	Yes	Yes	Yes
Industry–year fixed effects	Yes	Yes	Yes
Firm fixed effects	Yes	Yes	Yes
Adjusted R2	0.4007
Observations	18,063	18,063	18,063
Under identification test			39.792 ***
Weak identification test			38.195
Stock–Yogo threshold			16.38

Note. *** p < 0.01. The robust standard errors in parentheses are clustered at the firm level in this article, as shown in the table below.

5.3. Robustness Tests

5.3.1. Changing Dependent Variable and Independent Variable

We first use different supply chain risk calculation methods to replace the previously explained variable for regression analysis. Firstly, we change the denominator in the formula from the total number of sentences contained in the previous MD&A text to the number of sentences containing “risk” topic vocabulary and regress. Secondly, we replace the dependent variable with the logarithm of the total number of sentences containing both “supply chain” and “risk” keywords. Next, we replace the explanatory variable of “Robot” with the logarithm of the number of imported robots by firms. The data will be obtained by matching the robot import data from the China Customs Trade Database with listed companies [21]. The regression results shown in columns (1)–(3) of Table 5 indicate that the sign and significance of the core explanatory variable did not change with the replacement of the variable.

Table 5. Replacing variables.

Variables	Changing the Denominator to the Number of Risk Sentences	Changing the Denominator to the Logarithm of the Total Number of Sentences	Robot Input
	Chainrisk1	Chainrisk2	Chainrisk
	(1)	(2)	(3)
Robot Input			−0.007 ** (0.003)
Robot	−0.017 ** (0.009)	−0.115 * (0.068)
Controls	Yes	Yes	Yes
Province fixed effects	Yes	Yes	Yes
Industry-year fixed effects	Yes	Yes	Yes
Firm fixed effects	Yes	Yes	Yes
Adjusted R2	0.326	0.436	0.401
Observations	18,063	18,063	17,896

Note. * p < 0.1, ** p < 0.05. The robust standard errors in parentheses are clustered at the firm level in this article, as shown in the table below.

5.3.2. Excluding the Impact of Strategic Disclosure

Firms may intentionally reduce the disclosure of supply chain risk information for various motives, resulting in underestimation of firm supply chain risk calculated from M&D texts [52,56]. To avoid the impact of selective disclosure, firstly, samples that have been punished by the China Securities Regulatory Commission or stock exchanges for illegal disclosure are excluded. Next, based on the information disclosure index of listed companies, only samples with high ratings (excellent and good ratings) are retained for regression analysis. It can be seen that the conclusion in Table 6 is still robust.

Table 6. Robustness test result, excluding the impact of strategic disclosure.

Variables	Removing Illegal Disclosures Samples	Keeping Highly Rating Samples
	Chainrisk	Chainrisk
	(1)	(2)
Robot	−0.011 *** (0.003)	−0.008 ** (0.003)
Controls	Yes	Yes
Province fixed effects	Yes	Yes
Industry–year fixed effects	Yes	Yes
Firm fixed effects	Yes	Yes
Adjusted R2	0.404	0.432
Observations	15,065	9083

Note. ** p < 0.05, *** p < 0.01. The robust standard errors in parentheses are clustered at the firm level in this article, as shown in the table below.

5.3.3. Controlling Contemporaneous Policies

To support the digital transformation of firms, the country has launched pilot projects for supply chain innovation and application, and has successively announced the list of pilot cities and pilot firms. In addition, a pilot zone system for innovative applications of artificial intelligence and a pilot demonstration project for intelligent manufacturing have been launched, both of which may interfere with the results of our research. Therefore, we successively add virtual variables at the firm and regional levels as control variables in the benchmark regression model: whether the firm is in the supply chain innovation and application pilot city, whether it is on the list of supply chain innovation and application demonstration firms, whether it is in the artificial intelligence pilot zone, and whether it belongs to the intelligent manufacturing pilot demonstration project. The results in columns (1)–(4) of Table 7 show that the research conclusions remain largely unchanged after controlling for policies that may have an impact on firm supply chain risks.

Table 7. Robustness test result, controlling the same period policy.

Variables	Controlling Supply Chain Innovation and Application Pilot Cities Policy	Controlling Supply Chain Innovation and Application Pilot Firms Policy	Controlling Artificial Intelligence Pilot Zone Policy	Controlling Smart Manufacturing Policy
	Chainrisk	Chainrisk	Chainrisk	Chainrisk
	(1)	(2)	(3)	(4)
Robot	−0.009 *** (0.003)	−0.009 *** (0.003)	−0.009 *** (0.003)	−0.009 *** (0.003)
Controls	Yes	Yes	Yes	Yes
Province fixed effects	Yes	Yes	Yes	Yes
Industry–year fixed effects	Yes	Yes	Yes	Yes
Firm fixed effects	Yes	Yes	Yes	Yes
Adjusted R2	0.401	0.401	0.399	0.400
Observations	18,062	18,062	18,063	18,063

Note. *** p < 0.01. The robust standard errors in parentheses are clustered at the firm level in this article, as shown in the table below.

5.3.4. Other Robustness Test Results

We further change the clustering method from clustering at the firm level to clustering at the industry level. The results in column (1) of Table 8 indicate that the basic conclusions have not changed after clustering at the industry level. To control for unobservable factors in provinces that change over time, this paper strengthens the control by incorporating province–year interaction fixed effects into the model. As shown in column (1) of Table 8, the coefficient of the core explanatory variable remains significantly negative. Therefore, after controlling for regional factors, the firm-level exposure to industry-wide robot penetration can still significantly reduce firm supply chain risks, which validates the rationality of the indicator measurement and the robustness of the research conclusions in this paper.

Table 8. Other robustness test results.

Variables	Cluster by Industry	Strengthen Control	Take Lagged Terms for Core Explanatory Variable	3% Tail Reduction
	Chainrisk	Chainrisk	Chainrisk	Chainrisk
	(1)	(2)	(3)	(4)
Robot	−0.009 *** (0.002)	−0.010 *** (0.003)		−0.015 ** (0.006)
L.Robot			−0.013 *** (0.003)
Controls	Yes	Yes	Yes	Yes
Province fixed effects	Yes	Yes	Yes	Yes
Industry–year fixed effects	No	Yes	No	No
Firm fixed effects	Yes	Yes	Yes	Yes
Controls	Yes	Yes	Yes	Yes
Adjusted R2	0.401	0.402	0.373	0.406
Observations	18,063	18,063	14,873	18,063

Note. ** p < 0.05, *** p < 0.01. The robust standard errors in parentheses are clustered at the firm level in this article, as shown in the table below.

Taking the lagged term of the core explanatory variable for regression can alleviate the endogeneity problem caused by reverse causality to a certain extent, and can also study whether there is a lag effect in the reduction in supply chain risk in firms by the firm-level exposure to industry-wide robot penetration. Therefore, in our study, the first-order lag term of the core explanatory variable “Robot” is added to the model. As shown in column (3) of Table 8, the significance of the coefficient remains unchanged, indicating that the firm-level exposure to industry-wide robot penetration can continue to have a positive effect on reducing firm supply chain risks—in other words, there is a lag effect. Subsequently, we conducted a 3% truncation process on all continuous variables, and the results remained robust.

5.4. Mechanism Test

We construct a regression model as shown in formula (7) to explore the main mechanisms by which the firm-level exposure to industry-wide robot penetration affects supply chain risks:

$$M_{ijt}={\beta }_0+{\beta }_{1}Robo{t}_{it}+{\beta }_{2}Contro{l}_{it}+{\mu }_i+{\eta }_p+{\nu }_{jt}$$

(7)

Among them, M_ijt are the mechanism variables, and the definitions of the remaining variables are consistent with the previous text.

5.4.1. Supply Chain Discourse Power

The theoretical analysis in the previous text points out that increasing the discourse power in the supply chain is an important mechanism for reducing supply chain risks in firms through the firm-level exposure to industry-wide robot penetration. Therefore, the next research object is the discourse power of the supply chain, examining the impact of industrial robot usage on the discourse power of firm supply chain. Firstly, the discourse power of firms in the middle and downstream, or, in other words, the discourse power relative to their customers, is studied. We measure the discourse power of firms relative to their customers using the proportion of sales to their top five customers against their total sales [56]. In column (1) of Table 9, the discourse power in the middle and downstream of the firm is taken as the dependent variable. The coefficient of the explanatory variable “Robot” in column (1) is significantly negative, indicating that as the penetration of industrial robots increases, the firm can sell products to more customers, thereby reducing the sales share of the top five customers and lowering the customer concentration of the firm. The decrease in customer concentration can improve the firm’s discourse power relative to customers.

Table 9. Supply chain discourse power mechanism test results.

Variables	Discourse Power	Discourse Power	Discourse Power
	Midstream–Downstream Discourse Power	Upstream–Midstream Discourse Power	Supply Chain Discourse Power
	(1)	(2)	(3)
	Customer	Purchase	Supply Chain
Robot	−6.157 *** (1.793)	−1.061 (2.185)	−3.493 ** (1.463)
Controls	Yes	Yes	Yes
Province fixed effects	Yes	Yes	Yes
Industry–year fixed effects	Yes	Yes	Yes
Firm fixed effects	Yes	Yes	Yes
Adjusted R2	0.712	0.511	0.626
Observations	17,186	17,186	17,186

Note. ** p < 0.05, *** p < 0.01. The robust standard errors in parentheses are clustered at the firm level in this article, as shown in the table below.

Secondly, we use the proportion of the procurement amount of the top five suppliers of the firm to the total procurement amount to measure the discourse power of the firm relative to the suppliers (i.e., upper middle reaches discourse power). The regression results are shown in column (2) of Table 9, where the coefficient of the explanatory variable is positive, indicating that the increase in the penetration of industrial robots in the firm does not reduce its concentration at the supplier level, and therefore cannot improve its discourse power relative to the suppliers.

Finally, the average of the proportion of procurement from the top five suppliers and the proportion of sales from the top five customers of the firm is taken as the dependent variables to study the impact of industrial robot usage on the overall discourse power of the supply chain. The results are shown in column (3) of Table 9. The coefficient of “Robot” in column (3) is significantly negative, indicating that the popularity of industrial robots has reduced the concentration of supply chain for firms, thereby increasing their discourse power in the supply chain. This impact is mainly manifested in term of promoting diversified customer allocation for firms.

5.4.2. Supply Chain Coordination

To confirm whether the firm-level exposure to industry-wide robot penetration can enhance the coordination mechanism of the firm supply chain and thus affect supply chain risks, we examine the impact of industrial robot use on the external and internal coordination of the firm supply chain.

Firstly, we take the establishment of strategic alliances between firms and those in the supply chain as a measure of strengthened external coordination in their supply chain. We collect announcement information about the establishment of strategic alliances by firms through text analysis methods, and manually select samples with supply chain firms as the contracting parties to the strategic alliance, assigning them a value of one. The remaining samples without strategic alliances with supply chain firms are assigned a value of zero [57]. As shown in column (1) of Table 10, the coefficient of the core explanatory variable “Robot” is significantly positive, indicating that the firm-level exposure to industry-wide robot penetration effectively promotes the establishment of strategic alliances between firms and their suppliers or customers, thereby strengthening external coordination of the supply chain and jointly resisting supply chain risks.

Table 10. Supply chain external coordination mechanism test results.

Variables	External Supply Chain Coordination of Supply	External Supply Chain Coordination of Supply	External Supply Chain Coordination of Supply
	Strategic Alliance	Coordination Degree	Cost Sharing
	(1)	(2)	(3)
Robot	0.099 ** (0.040)	−0.024 ** (0.011)	0.019 ** (0.010)
Controls	Yes	Yes	Yes
Province fixed effects	Yes	Yes	Yes
Industry–year fixed effects	Yes	Yes	Yes
Firm fixed effects	Yes	Yes	Yes
Adjusted R2	0.222	0.720	0.811
Observations	16,282	16,911	17,111

Note. ** p < 0.05. The robust standard errors in parentheses are clustered at the firm level in this article, as shown in the table below.

Secondly, we investigate whether the firm-level exposure to industry-wide robot penetration reduces supply chain coordination costs. Following the approach of Cachon et al. (2007), the degree of supply-and-demand matching in a company’s supply chain is measured by the deviation between production changes and demand changes [58,59]. The greater the deviation between supply and demand in the supply chain, the higher the coordination cost of the company’s supply chain. The second column of Table 10 reports this result, indicating that the widespread deployment of robots in production can improve the supply–demand matching of firms and reduce supply chain coordination costs.

Finally, we examine the impact of industrial robot usage on cost sharing among supply chain members. Firm supply chain cost sharing refers to the sharing of costs between upstream and downstream firms (suppliers and customers) in the supply chain through commercial credit cost sharing, research and development cost sharing, and other forms. Firm cost sharing between supply chains is a concrete manifestation of the coordination and cooperation mechanism among supply chain members. The reasons are as follows. Firstly, commercial credit is the credit convenience obtained by firms from the supply chain. If a company transfers financial pressure to upstream and downstream firms through commercial credit, it needs to establish an evaluation mechanism for upstream and downstream firms, which is a representation of supply chain coordination. Secondly, cost sharing in joint R&D requires the establishment of a cross-firm R&D project management system, which enables real-time synchronization of technology roadmaps, process parameters, testing data, and subsequent R&D cost sharing, thereby forming a collaborative evolution of technology between both parties and promoting supply chain cooperation and coordination. Thirdly, the cost sharing of commercial credit and cooperative research and development has a bidirectional synergistic effect. For example, the collaborative experience of capital flow accumulated from the cost sharing of commercial credit can be transferred to the research and development cooperation scenario. We define the net cost sharing of a company’s supply chain as (notes payable + accounts payable + unearned revenue − notes receivable − accounts receivable − prepaid expenses)/total assets, and then use the net cost sharing of the supply chain as the dependent variable for regression analysis. As shown in column (3) of Table 10, the results are significantly positive, indicating that the increase in the penetration of firm robots can promote cost sharing among firms on the chain.

The theoretical analysis in the previous text suggests that strengthening the internal coordination mechanism of the supply chain is also an important mechanism for the firm-level exposure to industry-wide robot penetration to affect supply chain risks. To verify this mechanism, we draw on the paper by Luo and Nagarajan (2015) and first take supply chain information transparency to be the dependent variable [60]. The results in column (1) of Table 11 show that as the penetration of firm robots increases, the transparency of supply chain information is significantly enhanced. Next, we investigate the impact of industrial robot usage on inventory turnover rate. The coefficient of Robot in column (2) of Table 11 is significantly positive, indicating that the firm-level exposure to industry-wide robot penetration can significantly promote inventory turnover. In summary, the firm-level exposure to industry-wide robot penetration has been validated as a mechanism for promoting internal coordination in the supply chain from two dimensions: improving supply chain information transparency and inventory turnover.

Table 11. Supply chain internal coordination mechanism test results.

Variables	Internal Supply Chain Coordination	Internal Supply Chain Coordination
	Information Transparency	Inventory Turnover Ratio
	(1)	(2)
Robot	0.071 *** (0.020)	1.603 * (0.905)
Controls	Yes	Yes
Province fixed effects	Yes	Yes
Industry–year fixed effects	Yes	Yes
Firm fixed effects	Yes	Yes
Adjusted R2	0.654	0.731
Observations	17,501	16,910

Note. * p < 0.1, *** p < 0.01. The robust standard errors in parentheses are clustered at the firm level in this article, as shown in the table below.

5.5. Heterogeneity Tests

5.5.1. Heterogeneity of Regulatory Distance Among Firms

Regulatory distance reflects the strength of external constraints and government policy influences that firms are subject to. Firms may face higher compliance costs in situations where regulatory distance is closer (regulatory intensity is greater), and the intelligent transformation of production processes by industrial robots can help alleviate this compliance pressure. To explore the heterogeneous impact of industrial robot use on supply chain risk in firms with different regulatory distances, this paper draws on the method of Tang et al. (2024) to measure the regulatory distance of firms using the geographical distance between the registered address of listed companies and the dispatched agency of the China Securities Regulatory Commission [61]. The samples are divided into those with long and short regulatory distances based on the median. The grouping regression results based on the heterogeneity of firm regulatory distance are shown in columns (1) and (2) of Table 12. It can be seen that the absolute value of the regression coefficient is greater in the group with closer regulatory distance than in the group with farther regulatory distance, indicating that when firms are subject to stronger regulation, the firm-level exposure to industry-wide robot penetration has a more significant effect on reducing supply chain risks for firms. The policy inspiration for this result is that the government can reduce supply chain risks by strengthening external regulatory pressure, promoting firms to accelerate the automation process of production processes.

Table 12. Regulatory distance heterogeneity test results.

Variables	Long Regulatory Distance	Close Regulatory Distance
	Chainrisk	Chainrisk
	(1)	(2)
Robot	−0.008 * (0.005)	−0.011 *** (0.003)
Controls	Yes	Yes
Province fixed effects	Yes	Yes
Industry–year fixed effects	Yes	Yes
Firm fixed effects	Yes	Yes
Adjusted R2	0.408	0.410
Observations	8931	8877

Note. * p < 0.1, *** p < 0.01. The robust standard errors in parentheses are clustered at the firm level in this article, as shown in the table below.

5.5.2. Heterogeneity of Capital Intensity

In theory, capital intensity reflects the relationship between capital and labor substitution within a company, and the deployment of industrial robots represents the process of “fixed capital replacing variable labor”. Therefore, the capital intensity of a company may affect its deployment decisions for industrial robots. We use the firm capital intensity data from the CSMAR database to divide the sample into two groups for regression: high capital intensity and low capital intensity. Columns (1) and (2) of Table 13 show the results of grouped regression, indicating that when capital intensity is low, companies have a stronger motivation to reduce supply chain risks by using industrial robots on a large scale. The policy significance of this result for firms with low capital intensity is that their supply chain security and stability can be ensured through the application of automation technology.

Table 13. Capital intensity heterogeneity test results.

Variables	High Capital Intensity	Low Capital Intensity
	Chainrisk	Chainrisk
	(1)	(2)
Robot	−0.005 (0.004)	−0.011 *** (0.003)
Controls	Yes	Yes
Province fixed effects	Yes	Yes
Industry–year fixed effects	Yes	Yes
Firm fixed effects	Yes	Yes
Adjusted R2	0.434	0.403
Observations	8814	8751

Note. *** p < 0.01. The robust standard errors in parentheses are clustered at the firm level in this article, as shown in the table below.

5.5.3. Heterogeneity of Internationalization Level

Compared to domestic firms, multinational corporations are more susceptible to adverse impacts from international supply chain networks and face dual risks. To investigate whether the integration of firm products into the international trade network will have an impact on the economic consequences of industrial robot use, we divide the sample of firms with positive operating income from overseas into international companies, and divide the sample of companies without overseas operating income into non-international companies for regression analysis. The results in columns (1) and (2) of Table 14 indicate that only in the sample of non-international companies, the coefficient before “Robot” is significantly negative, and the inhibitory effect of industrial robot use on firm supply chain risk is more significant. The inspiration from this result is that companies with international operations should combine technological upgrades in their production processes with optimization of their cross-border supply chain governance systems to enhance their supply chain risk resistance capabilities.

Table 14. Internationalization degree heterogeneity test results.

Variables	International Firms	Non-International Firms
	Chainrisk	Chainrisk
	(1)	(2)
Robot	−0.006 (0.004)	−0.015 *** (0.005)
Controls	Yes	Yes
Province fixed effects	Yes	Yes
Industry–year fixed effects	Yes	Yes
Firm fixed effects	Yes	Yes
Adjusted R2	0.419	0.419
Observations	8758	8917

*** p < 0.01. The robust standard errors in parentheses are clustered at the firm level in this article, as shown in the table below.

5.5.4. Heterogeneity of ESG Ratings

Existing research has found that the ESG performance of firms plays an important role in reshaping the discourse power of the supply chain, promoting upstream and downstream cooperation in the supply chain, and thereby reducing supply chain risks [62]. Therefore, based on the ESG rating data of firms in the China Research Data Service Platform (CNRDS), we divide the sample into two groups of high and low ESG ratings for regression analysis. From the results in columns (1) and (2) of Table 15, it can be seen that the inhibitory effect of industrial robot use on firm supply chain risk is only significant when the firm’s ESG rating is high. One plausible explanation is that companies with high ESG ratings transform the technological advantages of industrial robots into comprehensive management and control of supply chain risks through systematic environmental management, social responsibility fulfillment, and technological governance capabilities. However, low ESG firms lack long-term strategic synergy and risk integration mechanisms, making it difficult for robot applications to break through the bottleneck of local efficiency improvement. The inspiration for corporate management from this result is that companies with good ESG performance can use robotics technology to further amplify their governance advantages and build a supply chain governance framework of “intelligent manufacturing + ESG”; on the other hand, companies with poor ESG performance should ensure that industrial robots play a role in supply chain risk prevention while improving their internal governance.

Table 15. ESG ratings heterogeneity test results.

Variables	High ESG Ratings	Low ESG Ratings
	Chainrisk	Chainrisk
	(1)	(2)
Robot	−0.011 *** (0.004)	−0.005 (0.004)
Controls	Yes	Yes
Province fixed effects	Yes	Yes
Industry–year fixed effects	Yes	Yes
Firm fixed effects	Yes	Yes
Adjusted R2	0.446	0.419
Observations	8732	8748

*** p < 0.01. The robust standard errors in parentheses are clustered at the firm level in this article, as shown in the table below.

5.6. Further Analysis

The improvement of firm-level exposure to industry-wide robot penetration may have a more direct impact on enterprises by providing technical support in the form of software applications, online platforms, and other forms. This study calculates the proportion of intangible asset details related to digital technology in the financial report notes of listed companies, including keywords such as software, network, intelligent platform, client, management system, etc., and uses this indicator as a measure of software application programs, online platforms, and other application technologies for regression analysis. As shown in Table 16, the firm-level exposure to industry-wide robot penetration can support enterprises through the application of technology.

Table 16. Further analysis.

Variables	Tech Support
	(1)
Robot	0.056 ** (0.025)
Controls	Yes
Province fixed effects	Yes
Industry–year fixed effects	Yes
Firm fixed effects	Yes
Adjusted R2	0.356
Observations	17,566

** p < 0.05. The robust standard errors in parentheses are clustered at the firm level in this article, as shown in the table below.

6. Discussion and Conclusions

This study examines whether and how the adoption of industrial robots reduces supply chain risks, using firm-level data and text-mined disclosures of risk information. The findings provide several key insights.

Firstly, empirical results demonstrate that a higher penetration of industrial robots significantly lowers firms’ exposure to supply chain risks. This effect remains robust across alternative measurements, endogeneity checks, and multiple robustness tests. It should be pointed out that the use of robots at the firm level measured in this study mainly reflects the impact of industry automation trends on firms, and may not necessarily be its autonomous adoption behavior.

Secondly, mechanism analysis on supply chain discourse power proves Hypothesis 2, that the firm-level exposure to industry-wide robot penetration can enhance a company’s discourse power in the supply chain, thereby reducing supply chain risks. By using the discourse power indicators of midstream–downstream, upstream–midstream, and supply chain as the dependent variables, this conclusion reveals the dynamic reshaping effect of digital technology (such as robots) on discourse power, and also indicates that enterprises can enhance their production capacity through robots, thereby expanding their customer network and reverse enhancing their discourse power.

Thirdly, mechanism analysis on supply chain collaboration proves Hypothesis 2: the firm-level exposure to industry-wide robot penetration can improve internal and external collaboration in the supply chain, thereby reducing supply chain risk. In this analysis, strategic alliance, coordination degree, and cost sharing are used as indicators of external supply chain coordination, and information transparency and inventory turnover ratio are used as indicators of internal supply chain coordination. These mechanisms underscore the governing role of automation, which goes beyond productivity gains to reshape supply chain relationships and improve systemic resilience.

Finally, heterogeneity analysis shows that the risk-mitigating effect of industrial robots is particularly pronounced in firms with lower capital intensity, weaker internationalization, closer regulatory oversight, and stronger ESG performance. These results suggest that organizational and institutional contexts condition the benefits of automation, highlighting the importance of aligning technology adoption with governance and sustainability practices.

6.1. Theoretical Implications

Overall, this study contributes to the literature on supply chain resilience by positioning industrial robots as a strategic enabler of risk governance. The findings expand our understanding of how automation interacts with supply chain power dynamics and coordination mechanisms. This dual role of automation is particularly relevant in today’s turbulent global environment, where supply chain stability and adaptability have become central to competitive advantage.

Previous research based on resource dependence theory (Kooiman, 1999) [20] has suggested that supplier diversification can reduce supply chain risks, but has not answered the question of how to achieve supplier diversification. The value of this study lies in revealing that firm-level exposure to industry-wide robot penetration is a key technological path for achieving supplier diversification. This discovery supplements the research on resource dependency mitigation paths in resource dependency theory, extending solutions at the organizational behavior level to the technical level.

At the same time, this study also answered the question of how technology in collaborative governance theory can overcome barriers to information sharing. The efficient promotion of collaborative governance is often limited by information asymmetry and delayed transmission among multiple parties. However, this study found that the automation production characteristics and digital data integration capabilities of robots can capture and synchronize key information such as production progress, inventory levels, and quality inspection in real time. This result clarifies that automation technology can serve as an important empowering carrier for collaborative governance, making the theoretical explanation of supply chain collaboration in the digital age stronger.

6.2. Managerial Implications

Our findings offer theoretical implications for resource-dependence and governance perspectives. From a managerial standpoint, the results suggest that firms can leverage industrial robots not only to enhance efficiency but also to strengthen resilience against systemic disruptions. Through our findings, several important implications could be adopted for managers and practitioners seeking to strengthen supply chain resilience in turbulent environments.

First, leverage industrial robots as enablers of resilience rather than solely efficiency. While industrial robots are often introduced to reduce labor costs or increase productivity, our results suggest that they also enhance firms’ bargaining power and coordination capabilities, thereby reducing risk exposure. Managers should therefore integrate robot deployment into broader risk governance strategies, rather than treating it as a purely operational investment.

Second, align robot adoption with supply chain collaboration and governance mechanisms. The evidence shows that industrial robots foster stronger internal integration and external partnerships. To maximize these benefits, firms should design automation strategies that explicitly link with supplier diversification, customer relationship management, and information-sharing platforms. This alignment can help firms better anticipate disruptions and coordinate effective responses.

Third, consider organizational and contextual conditions when deploying automation. The heterogeneity analysis reveals that the risk-reducing effects of robots are more significant in firms with lower capital intensity, weaker internationalization, and stronger ESG performance. Managers in highly internationalized firms should complement automation with cross-border risk management practices, while firms with strong ESG capabilities can leverage robots to amplify sustainability-oriented governance advantages.

Finally, adopt a long-term perspective on technology-driven resilience. Industrial robots should be viewed as part of a dynamic capability portfolio that allows firms to adapt to shifting environments. This means balancing investments in automation with complementary resources such as digital platforms, workforce reskilling, and sustainability practices. In doing so, firms can build supply chains that are not only efficient but also robust and adaptable in the face of systemic shocks.

By reframing automation as a strategic enabler of resilience, managers can move beyond short-term efficiency gains and position their supply chains to have sustained competitiveness in volatile global markets.