Government Regulation and Safe Production in Agricultural Enterprises: Panel Tracking of Regulatory Perceptions and Cross-Sectional Analysis from China

Abstract

With the rapid advancement of agricultural modernization, ensuring production safety has become a pressing concern, yet the mechanisms through which government regulation fosters safe production remain underexplored. This study addresses this gap using a two-stage survey design: first, panel-tracked survey data collected from 2021 to 2024 are used to document the evolution of regulatory perceptions among agricultural enterprises; second, a cross-sectional analytical design based on three survey waves conducted in 2024 is employed to examine the effect mechanisms using structural equation modeling method. Drawing on survey data from 485 Chinese agricultural enterprises in 2024, the findings show that four regulatory types—normative, punitive, incentive, and service—promote safe production both directly and indirectly through dual pathways: knowledge acquisition (cognitive–technical capacity building) and risk awareness (preventive attitudinal orientation). Mediation comparison analysis reveals that these two mechanisms exert equivalent effects across all regulatory pathways, indicating complementary rather than competing roles. Theoretically, the study advances regulatory pluralism and dual-mediation frameworks in organizational safety research; practically, it offers guidance for policymakers to design integrated regulatory portfolios and for managers to strengthen both knowledge systems and risk-aware cultures.

1. Introduction

A crucial component of the national economy is agriculture, which is essential for ensuring food security, boosting farmers’ incomes, and driving rural economic development [1,2]. Concerns about agricultural safety production have become more prominent as China experiences rapid economic growth and advances in agricultural modernization [3,4]. The prevalent agricultural production model in China has resulted in resource depletion and negative environmental impacts [5]. The outdated extensive agricultural production practices pose significant threats to both ecological environments and the safety of agricultural products [6,7]. The deteriorating alignment between high-quality agricultural development and environmental carrying capacity is increasingly evident, as the tension escalates between ecological degradation and the people’s yearning for an aesthetically pleasing environment. The widening disparity between the supply of secure agricultural goods and the public’s craving for superior products continues to grow, while the pursuit of “carbon neutrality and peak carbon emissions” in agricultural progress remains an ongoing imperative [8]. In light of these issues, ensuring the safety of agricultural production in China is of paramount importance. Safe production directly impacts the nation’s economic stability, public health, and the overall global agricultural supply chain. Agriculture forms the backbone of China’s national economy, and its safety is tightly connected to food security, income levels for farmers, and rural economic sustainability [9,10]. Agricultural safety concerns, such as pesticide residues, machinery accidents, and natural disasters, not only jeopardize the lives and property of agricultural workers but also have far-reaching implications for product quality, consumer health, and market reputation [11].

In this research, the notion of “safety” is perceived through a multidimensional lens that encompasses the safeguarding of human health, responsible environmental management, and the guarantee of product integrity throughout the agricultural production cycle. In agricultural production, safety goes beyond simply protecting consumers; it also includes maintaining environmental integrity through sustainable approaches, minimizing occupational risks for those working in agriculture, and complying with rigorous standards that regulate the safety and quality of agricultural products. This holistic perspective on safety is evident in the various strategies designed to reduce risks associated with agricultural activities, ranging from effective land management and crop growing to machinery use and product processing. By incorporating these elements into this research, this study offers a nuanced insight into how governmental regulations impact safe production practices, thereby improving the quality of agricultural goods while also fostering sustainable development within the agricultural sector in China.

From a worldwide standpoint, as one of the leading agricultural nations on the planet, the efficacy of safe production in agricultural enterprises in China has a substantial impact on the security of global food supply and the steadiness of agricultural trade [12,13]. In today’s interconnected world, Chinese agricultural goods are distributed internationally, making safety production concerns a central focus of global market interest [14]. Additionally, technological advancements and management practices in safe production within China’s agriculture offer insights and beneficial models for other developing nations, aiding in the sustainable growth of worldwide agricultural production. Moreover, with the escalation of global climate change and environmental concerns, it is crucial for Chinese agricultural businesses to guarantee secure production while promoting green, ecological, and circular growth. This has a significant impact on global agriculture’s ability to combat environmental obstacles [7,15,16]. Consequently, a thorough examination of the correlation between governmental regulations and secure production in agricultural establishments not only improves the safety standards of Chinese agricultural enterprises but also offers essential theoretical and practical guidance for the sustainable progression of worldwide agriculture.

Government regulation plays a central role in safeguarding agricultural enterprises by establishing standards that ensure both worker safety and product quality [17,18,19]. Through the implementation of regulatory frameworks, governments can guide agricultural practices, reduce risks, and ensure the viability of production processes. Despite numerous regulatory initiatives aimed at enhancing agricultural safety, incidents of regulatory failure have continued to plague the sector, undermining both public trust in agricultural products and the broader economic potential of agriculture [5,20,21]. These incidents underscore the need for more effective regulation and enforcement mechanisms to address the persistent safety risks in agricultural production.

Nonetheless, the existing literature on the correlation between government oversight and the safe production within agricultural businesses remains lacking. While previous research has delved into how government regulations impact safe practices in agriculture [22,23,24], these studies typically emphasize immediate outcomes and static analysis, neglecting thorough examinations of enduring effects and evolving circumstances. The deficiencies within this investigative strategy have led to a deficient and superficial comprehension of the connection between government regulations and safe production within agricultural enterprises.

On the other hand, existing studies generally fail to systematically reveal how the effects of various regulatory forms (such as normative, punitive, incentive, and service-based) are transmitted to agricultural enterprises’ safety production behaviors through specific mediating mechanisms, such as knowledge acquisition and risk awareness. As a result, the existing literature does not provide a comprehensive theoretical explanation of how government regulation influences agricultural enterprises’ safety production through multiple pathways and mechanisms.

Given the specific context, there is a notable absence of extensive longitudinal investigations that can unveil the evolving dynamics in the correlation between governmental oversight and ensuring safe production in agricultural enterprises. By conducting longitudinal analyses over an extended period, we can effectively monitor the enforcement of regulatory mandates by the government, witness the adaptive measures taken by agricultural enterprises in response to these regulations at various junctures, and evaluate the tangible effects of such regulations on the overall safety standards within these enterprises. Embracing this research strategy will yield a wealth of comprehensive and profound data points and perspectives, enabling us to gain deeper insights into the interplay between government regulatory practices and safe production in agricultural enterprises, as well as the progression of this interrelationship.

This research will concentrate on the following areas:

(1)

The impact factors and mechanisms of the government regulation on the safe production practices of agricultural enterprises.

(2)

Analyzing the long-term evolutionary trends in the relationship between government regulation and the safe production of agricultural enterprises.

(3)

Investigating agricultural enterprises’ perspectives on various forms of government regulation approaches over time.

(4)

Strategies for optimizing safe production in agricultural enterprises.

(5)

Evaluating the functional effectiveness of government regulation in ensuring agricultural safe production and devising optimization strategies.

By addressing these issues, this study makes three distinct contributions to the literature. First, in terms of longitudinal design, it addresses the predominance of cross-sectional approaches by tracking regulatory perceptions over four years (2021–2024) and testing causal mechanisms with time-lagged data, thereby revealing how regulatory effectiveness evolves and how enterprises adapt over time. Second, with respect to the mediation mechanism, it opens the “black box” between regulation and behavior by showing how four regulatory types operate through dual pathways of knowledge acquisition and risk awareness, extending understanding beyond traditional compliance or economic channels. Third, in relation to the Chinese context, it provides systematic evidence on the performance of regulatory instruments in an environment marked by rapid agricultural modernization, institutional fragmentation, and strong state intervention, offering insights relevant to other developing and transition economies. Collectively, these contributions yield valuable implications for policymakers in designing integrated regulatory portfolios, for enterprise managers in strengthening safety capabilities, and for scholars in advancing regulatory theory.

The remaining structure of the paper is organized as follows: The second section is the literature review and research background, the third section focuses on the development of research hypotheses, the fourth section presents the methodology, the fifth section reports the results of the empirical tests, the sixth section includes the conclusion and discussion, and the final section is research limitations and future research.

2. Literature Review and Research Background

2.1. Government Regulation

Government regulation can be defined as the measures initiated by pertinent government entities to safeguard and fulfill certain public interests and societal objectives [22,25]. This process encompasses the creation of regulations and the application of governmental authority to intervene in markets by regulating, restricting, or overseeing the activities of microeconomic participants [18,26]. Research highlights that government regulation serves as a principal institutional driver for corporate environmental protection initiatives [27,28]. Figure 1 presents the key regulatory strategies and relevant regulations promulgated by the Chinese government from 2021 to 2023.

While the four-type regulatory classification adopted in this study draws on [22,29], it is essential to position this framework within the broader regulatory theory landscape. Alternative frameworks, such as responsive regulation [30] emphasizing regulatory pyramids and escalation strategies, and risk-based regulation [31] focusing on risk assessment and proportionate responses, have been widely applied in environmental and food safety contexts. However, these frameworks primarily address regulatory intensity and targeting, whereas our four-type classification captures the functional diversity of regulatory instruments—from rule-setting (normative) to enforcement (punitive), from economic steering (incentive) to capacity-building (service). This functional taxonomy is particularly suited to the Chinese agricultural context, where government intervention operates through multiple channels simultaneously [25]. Moreover, unlike binary classifications (e.g., command-and-control vs. market-based), our framework recognizes the complementarity among different regulatory types, aligning with recent calls for “smart mix” regulatory approaches [32].

Empirical research examining government regulation and firm behavior across agricultural and related sectors reveals three consistent gaps that motivate our study. First, most studies adopt cross-sectional designs that capture static relationships rather than regulatory dynamics over time [24]. Second, few studies examine the psychological and cognitive mechanisms, particularly knowledge acquisition and risk awareness, through which regulation influences behavioral changes [18,26]. Third, prior research predominantly focuses on single regulatory types in isolation rather than comparing multiple instruments’ effects simultaneously [33]. Our longitudinal study (2021–2024) addresses these gaps by examining how all four regulatory types influence safe production behavior through knowledge acquisition and risk awareness pathways.

Figure 2 demonstrates the specific meanings of the four distinct government regulatory types.

2.2. Agricultural Production Issues

Agricultural safety production issues create cascading effects across multiple dimensions, impacting not only farmers’ operational safety and economic interests, but also consumer health, environmental sustainability, food security, and social stability (Figure 3). Three interconnected dimensions warrant particular attention in understanding why government regulation and enterprise safe production behavior have become critical policy and research priorities.

The link between agricultural safety production and food security warrants specific attention. As the primary nexus in food production, the safety condition of agricultural enterprises directly impacts the volume, quality, and accessibility of food-critical pillars of national security [34,35]. With increasing population pressures and declining arable land, producing safe food with limited resources has become increasingly urgent [36,37].

Agricultural safety incidents including pesticide residues, veterinary drug abuse, and contamination during processing pose direct threats to consumer health and erode market confidence [11,20]. These incidents not only cause immediate health hazards but also trigger long-term reputational damage and market instability [21], highlighting the critical importance of safe production practices throughout the agricultural value chain.

Agricultural production involves inherent occupational risks including machinery accidents, chemical exposure, and climate-related hazards. Ensuring safe production practices protects the livelihoods and wellbeing of agricultural workers, contributing to rural social stability [9,10]. This dimension is particularly salient in China where smallholder farmers and family-based operations constitute many producers, often lacking formal safety training or protective equipment.

2.3. Agricultural Safe Production Behavior

The concept of safe production behavior, while related to green and sustainable production behaviors, requires precise definitional clarity [22]. Green production behavior emphasizes environmental outcomes such as emissions reduction and resource conservation, while sustainable production behavior encompasses long-term ecological and socioeconomic viability [38]. In contrast, safe production behavior focuses specifically on risk prevention and hazard mitigation to protect immediate human health, product quality, and environmental safety throughout the production cycle.

This distinction is critical for construct validity. Safe production behavior in agricultural enterprises encompasses three core dimensions: (1) input safety—proper use of seeds, pesticides, fertilizers, and veterinary drugs in compliance with safety standards [39]; (2) process safety—adoption of safe operational procedures to minimize occupational risks and contamination [40]; (3) output safety—quality control and proper handling to ensure product integrity [41]. While safe behavior may contribute to green outcomes, its primary objective is hazard control rather than resource optimization.

Agricultural practices for safety involve a range of measures to ensure the well-being of farmers, protect the environment, ensure product quality, and support sustainable agriculture [38,39,40]. These practices cover various aspects such as land management, crop cultivation, animal husbandry, machinery operation, pesticide and fertilizer use, and product processing. Safe agricultural practices encompass the entire production cycle, including pre-production, production, and post-production stages. In crop production, this includes precision pesticide application adhering to pre-harvest intervals and proper fertilizer management; in livestock production, it involves veterinary drug withdrawal periods and biosecurity measures [3,42]. See Figure 4 for details on each stage.

Safe agricultural production practices play a crucial role in elevating the overall safety standards of agricultural activities, protecting the lives and property of agricultural workers, and contributing to the sustainable development of the agriculture sector [41,43,44]. Understanding these behavioral dimensions provides the foundation for examining how government regulation (Section 2.1) influences enterprises’ adoption of safe practices through cognitive and attitudinal mechanisms, which we develop in the following hypothesis section.

3. Research Hypothesis Development

3.1. Government Regulation and Agricultural Safe Production Behavior

Numerous studies suggest that the main institutional driver of corporate environmental initiatives is government regulation [25,45]. According to the Porter Hypothesis, although government regulation can place extra burdens on companies, it also boosts incentives for creativity, leading to better product quality and efficiency [46]. Scholars worldwide have examined the Porter Hypothesis from different angles and have offered varying levels of support, as seen in the works of [47,48,49].

Furthermore, research from related studies highlights that firms under regulation often adhere to government guidelines in order to cultivate perceptions of political legitimacy within the social institutional context [45]. As a result, companies can secure access to governmental resources like subsidies and permits for market entry, aiding in the mitigation of financial hardship [50].

Hence, within the realm of agricultural safety in production, governmental oversight is vital. More precisely, governmental normative oversight usually includes explanatory statutes, directives, or guidelines with well-defined content and precise particulars, highlighting the importance of “engaging in proper, eco-friendly, and secure behavior” [51]. At the same time, normative oversight furnishes agricultural manufacturers with a structure by implementing a range of explicit production criteria and operational protocols. These guidelines encompass every facet from selecting seeds, utilizing pesticides, managing irrigation, to harvesting and preserving the crop. By establishing these norms, the government ensures that agricultural producers adopt best practices, which helps to minimize risks during the production process and improve the quality and safety of agricultural products. Furthermore, when considering safe agricultural production, such prescriptive regulatory requirements may induce herd behavior among farm operators, encouraging them to conform to mainstream values related to target expectations and ecological preservation [52]. In this way, governmental normative regulations can greatly advance safe agricultural production practices.

Government punitive regulation often uses specific directives or compulsory rules to direct and control individual intentions for behavior [53,54]. Punitive regulation sets up rigorous punitive measures for not following safety production rules, creating societal pressure on agricultural businesses throughout the production process and influencing their behavior towards safe production [45]. These consequences might involve financial penalties, withdrawal of production permits, and potentially criminal charges. Through this deterrent impact, farmers are incentivized to adhere to rules as they recognize the serious outcomes of not doing so. This regulatory strategy effectively diminishes non-compliant actions and guarantees the safety of agricultural production.

Additionally, incentive regulation in agriculture promotes the adoption of safety production measures through economic incentives like subsidies, tax breaks, and technical assistance [55,56]. For instance, financial aid may be extended to farmers practicing organic agriculture, while tax advantages could be granted to those utilizing eco-friendly pesticides. Such incentives lower the financial burden on agricultural producers, encouraging them to prioritize safety standards in their operations. In contrast, safe production demands higher expenses in comparison to conventional agricultural practices [57]. The implementation of government reward schemes often leads to an increase in agricultural income and a decrease in the incremental costs associated with safe production, achieved through subsidies and incentives. Additional studies suggest that a rise in the magnitude of governmental incentives correlates with an increased propensity to adopt safe production practices [33].

Ultimately, regulatory services play a crucial role in empowering agricultural producers to enhance their production safety capabilities and elevate their cognitive skills and perceived control in regard to safe production. This is achieved through the provision of various forms of assistance including information services, technical guidance, and access to market connections. The government supports producers in making informed decisions by furnishing them with up-to-date market data, weather forecasts, and alerts on pest/disease outbreaks. Furthermore, technical training and consultation services are offered to aid producers in mastering the latest agricultural technologies and management practices. These services collectively serve to bolster the technical expertise and market competitiveness of agricultural producers, while also reinforcing their understanding and proficiency in safe production practices [42]. Consequently, they are able to promote safe production behaviors effectively.

To summarize, the research proposes the following hypotheses:

H1a: 

Safe production in agricultural enterprises is facilitated by normative regulation.

H1b: 

Safe production in agricultural enterprises is facilitated by punitive regulation.

H1c: 

Safe production in agricultural enterprises is facilitated by incentive regulation.

H1d: 

Safe production in agricultural enterprises is facilitated by service regulation.

3.2. The Mediating Role of Knowledge Acquisition

The acquisition of knowledge involves firms recognizing their external surroundings and extracting information from them [58]. It encompasses the acquisition and conversion of external knowledge or various internal knowledge by firms [59]. According to [60], firms must not only gather new internal knowledge but also acquire fresh market and technological knowledge from the external environment in order to innovate and create new products, services, and technologies.

Furthermore, knowledge is an asset that allows organizations to maintain a sustainable edge over rivals, and it is essential for those who consider technology a key strength [61,62]. In highly competitive markets, companies must vigilantly observe competitors’ activities and react swiftly. Hence, when under pressure, organizations will seek to obtain knowledge in order to secure a competitive advantage.

Recent studies have shown that firms are pressured by regulations to gain expertise [60], with the impact mechanism mainly evident in the following areas (refer to Figure 5). Initially, governmental entities establish and execute rules, standards, and policies to provide clear directives and compliance requirements for enterprises, compelling them to seek and comprehend such information in order to ensure their operations are in accordance with the law. In the agricultural context, regulations on pesticide use compel farmers to acquire knowledge of chemical toxicity, safe dosage levels, and pre-harvest timing to minimize residues in food products. Similarly, animal welfare regulations require livestock enterprises to acquire knowledge of proper housing conditions, veterinary care protocols, and disease prevention measures. Such mandates directly foster domain-specific knowledge acquisition in agricultural practices.

Governmental regulations often accompany incentives like tax breaks, subsidies, and rewards that encourage firms to invest in research, development, and technological advancements, indirectly propelling knowledge acquisition in emerging technologies and methodologies. To qualify for these incentives, companies must continuously acquire and apply the most up-to-date knowledge in relevant sectors. Moreover, government regulations mandate that companies disclose information and share operational data and environmental impact reports. In agricultural production, traceability requirements including agricultural product quality safety tracing systems compel enterprises to acquire knowledge in digital record-keeping, data management systems, and supply chain tracking—skills essential for documenting inputs including seeds, fertilizers, and pesticides, as well as production processes and distribution channels. Additionally, climate adaptation subsidies encourage farmers to learn about drought-resistant crop varieties, water-saving irrigation technologies such as drip irrigation systems, and weather forecasting tools to enhance resilience against climate variability. This helps to boost corporate social responsibility and requires organizations to handle vast amounts of data to comply with regulations, thereby improving their data analysis and knowledge management skills.

Additionally, government regulations support knowledge acquisition in companies through educational initiatives and training programs. By offering vocational training, workshops, and online courses, the government assists in enhancing the expertise and knowledge of company staff. These training programs are aligned with current industry standards and best practices, enabling companies to remain current with industry knowledge.

Acquisition of knowledge is crucial in promoting safe production practices within agricultural businesses (as shown in Figure 5, which depicts its impact mechanism). Initially, it plays a key role in raising awareness about safety within the organization, ensuring that both staff and management fully appreciate the significance of safe production and consequently implement precautionary measures in their daily tasks. Furthermore, knowledge acquisition facilitates adherence to pertinent safety production regulations and standards, mitigating the risk of legal consequences and penalties resulting from non-compliance. Moreover, by obtaining up-to-date information on safety technologies and equipment, agricultural enterprises can integrate more advanced safety measures, thereby minimizing potential risks during production. Additionally, knowledge acquisition empowers organizations to effectively manage risks by identifying and evaluating potential safety threats, implementing appropriate preventive and control measures, and decreasing the likelihood of accidents. At the same time, it supports organizations in developing and improving emergency response strategies, boosting their ability to swiftly address unexpected incidents, and reducing the impact of accidents. Additionally, the ongoing learning and adoption of safety protocols compel businesses to enhance their safety protocols and operational procedures, leading to a consistent enhancement in safety practices. This doesn’t just elevate the company’s reputation and brand worth but also safeguards the welfare of workers, minimizing workplace injuries and illnesses, and promoting employee satisfaction and commitment.

To summarize, the research hypotheses presented are as follows:

H2a: 

Knowledge acquisition in enterprises is fostered by normative regulation.

H2b: 

Knowledge acquisition in enterprises is bolstered by punitive regulation.

H2c: 

Knowledge acquisition in enterprises is encouraged by incentive regulation.

H2d: 

Knowledge acquisition in enterprises is advanced by service regulation.

H3: 

The acquisition of knowledge in enterprises leads to enhanced safe production in agricultural enterprises.

H4a: 

The knowledge acquisition mediates the relationship between normative regulation and the enhancement of safe production in agricultural enterprises.

H4b: 

The knowledge acquisition mediates the relationship between punitive regulation and the enhancement of safe production in agricultural enterprises.

H4c: 

The knowledge acquisition mediates the relationship between incentive regulation and the enhancement of safe production in agricultural enterprises.

H4d: 

The knowledge acquisition mediates the relationship between service regulation and the enhancement of safe production in agricultural enterprises.

3.3. The Mediating Role of Risk Awareness

This research posits that governmental regulations could improve corporations’ awareness of risks. Through implementing strict regulations and standards for safety production, government supervision forces companies to recognize and evaluate potential risks and implement necessary preventive measures. This regulatory pressure not only increases companies’ awareness of risks but also encourages them to develop and improve risk management systems, ultimately decreasing the likelihood of safety incidents (Figure 6 demonstrates the impact mechanism).

Specifically, government regulation mandates that businesses adhere to specific safety, health, and environmental protection standards. Moreover, these standards translate into domain-specific risk awareness: pesticide drift that may contaminate neighboring organic farms or water sources; machinery operation hazards such as tractor rollovers and grain auger entrapments that endanger worker safety; biological risks from livestock diseases including avian influenza and African swine fever that threaten herd health and public safety; and soil degradation from excessive fertilizer use that undermines long-term productivity. Regulatory requirements for risk assessment plans, such as integrated pest management protocols or machinery safety audits, compel agricultural enterprises to systematically identify, evaluate, and prioritize these sector-specific risks. To comply with these regulations, companies must analyze risks in these areas and take steps to mitigate them. These regulatory demands compel companies to improve their awareness of potential risks.

Government mandates often entail requests for companies to divulge corporate information, including financial statements and reports on environmental impact. These disclosure rules push companies to focus more on risks that may arise from their operations and openly showcase their risk mitigation strategies to the public. Additionally, government regulations impact corporate conduct through offering incentives (like tax breaks, subsidies) and implementing penalties (such as fines, legal actions). To receive rewards or evade penalties, businesses will give greater importance to risk management, ultimately boosting their awareness of risks. Government regulatory bodies carry out routine or random inspections and audits on companies to ensure adherence to relevant rules. This monitoring process compels companies to continually prioritize their risk management procedures in preparation for potential reviews.

Governmental regulations also promote industry norms and best practices, typically covering sophisticated risk management concepts and techniques. To sustain a competitive edge, companies will adopt and incorporate these norms and methods, thus amplifying their risk awareness. Finally, governmental regulations mandate that companies establish emergency response protocols, including contingency plans and recovery strategies. In addition, agricultural preparedness extends to climate-related extremes such as flood evacuation plans for livestock farms, hail damage insurance for orchards, and heat stress mitigation protocols for greenhouse operations. It also encompasses biosecurity emergencies including rapid disease outbreak containment measures and mass culling procedures, as well as food safety incidents such as contamination recall systems and corrective action protocols. For example, China’s Emergency Response Plan for Major Animal Epidemics requires enterprises to cultivate risk-aware cultures that anticipate biological threats and maintain rapid response capabilities. These mandates encourage companies to contemplate potential crises and hazards in their everyday activities, thereby reinforcing their risk awareness.

Awareness of risks within corporations enhances the promotion of safe behaviors in agricultural enterprises (as shown in Figure 6 detailing its impact mechanism). This awareness encompasses the company’s capacity to recognize, evaluate, and comprehend potential risks, laying the groundwork for effective management of safe production [63,64]. Strong risk awareness enables companies to proactively implement measures aimed at risk prevention and mitigation [65,66], ultimately elevating the standard of safe production.

Companies with risk awareness are more proactive when it comes to recognizing and evaluating various potential hazards that may arise throughout the manufacturing process [67,68]. These hazards can range from natural disasters, pests, and diseases to chemical spills, mechanical breakdowns, and other unforeseen events. The ability to identify and evaluate these risks is essential for the development of efficient preventative strategies. Following the assessment of risks, companies can then devise appropriate preventative measures depending on the severity and likelihood of the hazards. These measures may involve enhancements to production processes, the utilization of safety gear, employee training programs, and the establishment of contingency plans. Furthermore, organizations that possess a solid understanding of potential risks will consistently observe the most recent advancements and top strategies in secure manufacturing, continually acquiring knowledge and enhancing their risk mitigation tactics [69,70]. This progressive enhancement assists organizations in reacting swiftly to newly arising risks and upholding the cutting-edge status of secure production standards [71].

On the flip side, the awareness of risks within companies also plays a significant role in fostering a culture of safety. This is achieved through effective internal communication and educational initiatives that instill a sense of safety consciousness within employees, integrating it seamlessly into the fabric of the organization [72]. Such cultural cultivation encourages employees to conscientiously follow safety protocols in their day-to-day tasks, cultivating positive practices for safe production. Firms that exhibit a robust sense of risk awareness are able to allocate their resources in a prudent manner, ensuring that the essential financial, technological, and human resources required for safe production are adequately provided for [73]. This optimization of resource distribution serves to bolster the infrastructure needed for safe production practices.

In summary, the research hypotheses proposed are as follows:

H5a: 

Normative regulation can increase enterprises’ risk awareness.

H5b: 

Punitive regulation can increase enterprises’ risk awareness.

H5c: 

Incentive regulation can increase enterprises’ risk awareness.

H5d: 

Service regulation can increase enterprises’ risk awareness.

H6: 

Risk awareness can promote safe production in enterprises.

H7a: 

Risk awareness acts as a mediator in the process of normative regulation that fosters safe production in agricultural enterprises.

H7b: 

Risk awareness acts as a mediator in the process of punitive regulation that fosters safe production in agricultural enterprises.

H7c: 

Risk awareness acts as a mediator in the process of incentive regulation that fosters safe production in agricultural enterprises.

H7d: 

Risk awareness acts as a mediator in the process of service regulation that fosters safe production in agricultural enterprises.

4. Methodology

The aim of this investigation is to acquire an in-depth comprehension of the enduring dynamic correlation and its influence mechanism between government regulation and safe production behaviors in agricultural enterprises. The study is partitioned into two primary sections, each carefully crafted to meet the goals of the research.

The first part, the longitudinal tracking survey conducted from 2021 to 2024, focuses on the enduring relationship between government regulation and enterprises. In contrast, the subsequent empirical study in 2024 delves into the intricate impact mechanism linking government regulation to safe production behaviors in enterprises. This systematic investigation entails thorough examination and documentation of empirical findings. The decision to analyze empirical data from 2024 is deliberate as it captures the most recent regulatory dynamics and corporate reactions, ensuring the research’s relevance and timeliness. Furthermore, leveraging the latest data allows for a more precise evaluation of the efficacy of existing regulatory measures and offers prompt insights for policy modifications.

To summarize, the design of this research thoroughly takes into account how time variables affect the efficiency of regulations. By conducting a longitudinal study and analyzing the most recent data, it extensively uncovers the link between governmental regulations and the safe production practices of companies, along with the mechanisms through which they influence each other. This approach not only boosts the credibility of the study findings but also offers valuable guidance for policymakers and businesses.

4.1. Scale Development

The study’s model, shown in Figure 7 below, encompasses seven variables for examination. Prior research was consulted, and research scales were utilized in a scientific manner. The scale was adapted through revisions and adjustments to align with the study’s investigative goals.

As such, this investigation has implemented multiple strategies to guarantee the scientific precision and integrity of the research framework. Initially, a thorough review of the literature was carried out to acquire a deep comprehension of the connection between regulatory measures by the government and the safe production within agricultural organizations. Building upon this foundation, current research frameworks and scales were methodically scrutinized and assessed. Enhancements and edits were implemented to the scales derived from prior studies to guarantee their ability to fully and precisely depict the correlation between government regulations and safe production within agricultural entities, along with the factors that impact it. Secondly, this research involved consultations with experts and scholars in pertinent fields alongside industry practitioners to gather insights through expert interviews and focus group discussions. These discussions aimed to collect their perspectives on safe practices within agricultural enterprises. The experts’ insights and practitioners’ views will help to enrich the research content, thereby enhancing its accuracy and reliability. Finally, to ensure the translation’s robustness, this study adhered to the forward-backward translation procedure specified by [74]. Additionally, the study enlisted a professional translation team for both forward and backward translations to ensure the scale’s accuracy and validity.

The research scale’s pre-testing holds equal importance. Before the final distribution, 100 enterprise participants were randomly selected for evaluating the initial questionnaire items. Reliability and validity tests were carried out on the scale using the collected data, leading to the elimination of measurement questions that did not meet the criteria. The measurement items underwent revision and refinement to address research process challenges and feedback from participants.

To clarify the criteria for modifying or deleting items during the pre-test phase, the research followed a systematic approach. The reference standards are as follows: (1) Factor loadings ≥ 0.60 were considered acceptable; (2) Average Variance Extracted (AVE) < 0.50 were considered problematic and removed to improve convergent validity; (3) Cronbach’s alpha and Composite Reliability (CR) values ≥ 0.70 were required to confirm scale reliability and internal consistency. These decision rules were applied systematically to remove or revise items, ensuring that only valid, reliable, and conceptually relevant measurement items were retained in the final survey instrument.

In Appendix A, the specific measurement items for each dimension are presented, covering Normative Regulation (NR), Punitive Regulation (PR), Incentive Regulation (IR), Service Regulation (SR), Knowledge Acquisition (KA), Risk Awareness (RA) and Safe Production Behavior (SPB). Also, a seven-point Likert scale is utilized in this research for evaluation purposes (ranging from 1 for strongly disagree to 7 for strongly agree). It is important to highlight that the accuracy and scientific integrity of the scale will be elaborated upon in the subsequent empirical analysis section. Appendix B presents the deleted measurement items along with the reasons for their removal.

4.2. Sample and Data Collection

This research project gathered participants through a variety of methods, such as sending out written invitations, utilizing online survey platforms, and making phone calls to potential participants. The research focused on agricultural companies in China, specifically targeting companies in 11 different major areas throughout the country, which include Henan, Anhui, Hainan, Shandong, Guangxi, Hebei, Jiangsu, Shanxi, Zhejiang, Guangdong, and Beijing. These regions span the North, East, South, and Central parts of China. The diverse latitudes and geological characteristics of these cities contributed to the broad range of the sample pool [75]. To summarize, the administration of surveys considered the diversity of the sample in relation to regional and geographical disparities, as well as policy discrepancies. The varied survey locations produced a sample that was more indicative and lessened any bias in sample selection [76].

Drawing on methods from previous research to mitigate common method bias threats, this study collected data from respondents in three separate waves at intervals of six weeks [77,78]. In the first round of the formal survey, items related to Normative Regulation (NR), Punitive Regulation (PR), Incentive Regulation (IR), and Service Regulation (SR) were included. The second round of the survey encompassed measurement items for Knowledge Acquisition (KA) and Risk Awareness (RA), while the third round focused on Safe Production Behavior (SPB) items (see Figure 8b). Respondents self-reported all measurement items. Since the official language in China is Chinese, the survey took place in Chinese.

The method of distributing the survey questionnaire was done in the following ways: (1) by conducting face-to-face interviews, where respondents were instructed to complete the questionnaire and submit the survey form after finishing; (2) through on-site visits, where respondents were requested to complete the questionnaire and send it back by mail within a specific timeframe; (3) via email. Participants were CEOs or high-level company executives who possessed a comprehensive knowledge of their organizations. Each company was surveyed with only one CEO or senior executive.

In this study, CEOs or senior executive were selected as the core informants due to their strategic oversight and in-depth understanding of the organization’s safety production practices and regulatory compliance. These individuals are typically responsible for decision-making at the highest level, including the implementation and monitoring of government regulations within the company. Given their extensive knowledge of both internal operations and external regulatory requirements, they are best positioned to provide accurate and comprehensive insights into how government regulations influence safety production practices. Moreover, CEOs and senior executive often have direct involvement in shaping organizational policies related to knowledge acquisition, risk awareness, and safety measures, making them ideal respondents for understanding the mechanisms through which different regulatory types—normative, punitive, incentive, and service—affect safety production outcomes in agricultural enterprises. This approach ensures the reliability of the data, as senior leaders are intimately familiar with both the strategic and operational aspects of safety within their organizations.

Figure 8a illustrates the overall framework and timeline of the survey design, which consists of two parts: (1) Panel Tracking of Regulatory Perceptions (from January 2021 to January 2024), with a random sample of 550 agricultural enterprises selected from the Chinese agricultural enterprise database. (2) Cross-sectional analysis of the effect mechanisms based on the 550 enterprises sample, involving three waves new survey from March 2024 to May 2024.

In the 2024 cross-sectional data survey, a total of 550 participants took part in the survey. After excluding 19 questionnaires with numerous missing values and subpar completion quality, 531 valid samples were collected, resulting in a successful response rate of 96.545%. Moving on to the next stage, the same set of questionnaires was dispatched to the 531 individuals who had presented valid answers in the first phase, garnering 509 satisfactory responses and achieving an effective response rate of 95.857%. Subsequently, a second round of survey invitations was extended to the 509 respondents, leading to the retrieval of 485 valid responses, which translated to an effective response rate of 95.285%. In summary, 485 surveyed enterprises provided complete feedback on all three survey questionnaires (see Figure 8b). Detailed descriptive statistics of the surveyed enterprises can be found in Figure 9. Moreover, the final sample of 485 enterprises is the same group of data observed over a four-year period.

During the recruitment of respondents and collection of samples, this research closely followed ethical guidelines and pertinent regulations to ensure the rights of the respondents were protected. Emphasis was significantly placed on safeguarding the respondents’ personal information throughout the survey process. Additionally, each participant provided informed consent before their participation. The research guaranteed the confidentiality and anonymity of all responses, and it was explicitly stated that the survey data would be utilized solely for this research purpose.

4.3. Trends in the Evolution of Enterprise Perceptions of Government Regulation (From 2021 to 2024)

In order to examine the changing perspectives of Chinese agricultural companies towards government oversight in four distinct categories (normative, punitive, incentive, service), a longitudinal study was devised spanning a duration of four years. The project commenced in January 2021 and wrapped up in January 2024, with the objective of tracking shifts in the perceptions of companies regarding various forms of regulatory actions via regular data gathering.

First, after thoroughly reviewing the literature and consulting with experts, the unique characteristics of and differences in the four regulatory types were clarified. Subsequently, a questionnaire was developed with questions presented on a scale for assessment. The questionnaire focused on the perspectives of companies regarding each type of regulation, their compliance status, assessments of consequences, and expectations for future regulatory advancements.

In order to guarantee the scientific validity of data collection in this research, a stratified random sampling method was utilized. This involved the random selection of 550 companies from the database of Chinese agricultural enterprises to serve as the sample. This approach ensured that the sample was representative of agricultural enterprises from diverse sizes and regions. The distribution of questionnaires was carried out via email and an online survey platform, with additional telephone follow-ups conducted to improve response rates. To ensure data consistency, a time-series survey was implemented, with fixed survey time points established (occurring every January) and standardized questions and response options maintained for each survey. This framework facilitated the comparison of data variations across different time periods. Following each survey administration, the research team executed data cleaning and validation processes to verify the accuracy and dependability of the collected data.

Figure 10 depicts how enterprises perceive the various forms of government regulation. Initially, the importance of normative regulations in setting industry norms has gained more acknowledgment from enterprises between 2021 and 2022. The mean score has climbed from 3.929 in 2021 to 4.429 in 2024 (out of a max score of 7). The effectiveness of normative regulations in improving product quality and safety is also on the rise. Significantly, the influence of the costs and intricacy of complying with normative regulations on enterprises is increasing as well.

Moreover, the deterrent impact of punitive regulations on prompting firms to adhere to regulations has gained more attention. Nonetheless, businesses are noticing a rise in the potential risk of excessively steep compliance expenses caused by relying too much on penalties as a regulatory tool. In contrast, the efficacy of incentive regulation (like grants and tax benefits) in encouraging companies to embrace new technologies and eco-friendly practices has been gaining recognition. Nevertheless, businesses are observing a marked decrease in the fairness and openness of incentive strategies, with the rating decreasing from 4.059 in 2021 to 3.782 in 2024.

Lastly, the importance of regulating services (such as offering guidance and technological assistance) to help businesses comprehend and adjust to rules is gaining more and more acknowledgment. The impact of Service regulation on improving the general competitiveness of businesses is also on the rise.

The survey results reveal that the significance of the four distinct categories of government regulation is growing. Specifically:

(1)

Generally, companies believe that normative regulations play a vital role in setting industry norms, improving the quality and safety of products. Nevertheless, they also highlight concerns regarding the costs and complexity of implementation.

(2)

Punitive regulations are viewed as a strong deterrent, compelling firms to adhere to rules and regulations. Despite this, there are worries among businesses that excessive reliance on fines could inflate compliance expenses.

(3)

Incentive regulation, such as subsidies and tax breaks, is seen as an effective way to encourage businesses to embrace new technologies and sustainable practices. However, it is crucial to ensure that incentive mechanisms are fair and transparent.

(4)

Business stakeholders consider service regulations, which offer advice and technical assistance, as particularly beneficial. This form of regulation not only helps companies better comprehend and conform to rules but also boosts their overall competitiveness.

However, from another perspective, while businesses acknowledge the importance of normative regulation in setting industry benchmarks, there is a decline in growth rate trajectory (see Figure 11). The growth percentages over the respective years are as follows: 6.541% (2021 to 2022), 4.000% (2022 to 2023), and 1.770% (2023 to 2024).

Organizations recognize that the influence of costs associated with the implementation and the complexity of normative regulations on their operations has been steadily rising, although the rate of this increase is slowing. Each year, the risk associated with an over-dependence on fines as a regulatory measure—which results in prohibitively high compliance costs—has been escalating, albeit at a decreasing pace.

The efficiency of incentive regulation (e.g., grants and tax breaks) in encouraging the implementation of innovative technologies and eco-friendly methods by businesses has shown a steady increase. Nevertheless, the annual growth rate has been on a downward trend, dropping from 3.500% (between 2022 and 2023) to 1.120% (between 2023 and 2024).

Concerning the ratings that organizations assign to the fairness and transparency of their existing incentive measures, there has been a decline from 2021 to 2024. However, the annual rate of this negative growth has been diminishing over the years. Specifically, the growth rate between 2023 and 2024 stands at −0.917%.

5. Empirical Analysis

This section uses three waves of cross-sectional survey data from March 2024 to May 2024 and the structural equation modeling method to analyze the effect mechanisms. No repeated-measures, growth, or cross-lagged models were estimated.

5.1. Common Methodological Bias Test

This research utilized two approaches to determine the presence of a potential common method bias in the study. Initially, the investigation employed Harman’s single-factor test method in the statistical analysis and performed exploratory factor analysis on all measurement items utilized in the research using SPSS 27.0 software. The findings post dimension reduction indicated that the first unrotated factor accounted for 21.124% of the variance, falling short of the 50% benchmark [77,79].

Following [80], a more comprehensive test for common method bias was carried out using the CFA comparison approach. Initially, all measurement items were organized into a single-factor structure known as Model 1 (single-factor model). Subsequently, the CFA model based on theory, including its original correlated variables and measurement items, was designated as Model 2 (multi-factor model). A comparison of the two models was made by assessing variations in degrees of freedom and chi-square values. The results of the comparative analysis (see

Table 1. Common methodological bias test results (CFA comparison approach).

Model	χ2	Df	∆χ2	∆Df	p
Single-factor	2615.752	189	2441.192	21.000	0.000
Multi-factor	174.560	168

) revealed that ∆χ² = 2441.192, ∆Df = 21.000, p = 0.000, which is below the significance level of 0.05.

Next, this study also employs a method that incorporates a non-measurable methodological factor to test for common methodological bias [81]. Specifically, a latent variable is added as a common methodological bias factor to the original model (as Model 2), and the model fit before (as Model 1) and after (as Model 1) adding this latent variable is compared. If the model fit indices after controlling for the latent variable do not differ substantially from those of the original path model (with changes in RMSEA and SRMR not exceeding 0.05, and changes in CFI, TLI, GFI, and IFI not exceeding 0.1), it suggests that no common methodological bias exists. The results show that the model controlling for the common methodological bias factor did not significantly improve the model fit (∆RMSEA < 0.05, ∆SRMR < 0.05, ∆CFI, ∆TLI, ∆GFI, and ∆IFI all < 0.1) (see

Table 2. Common methodological bias test results (CFA with common methodological bias factor added).

Model	RMSEA	SRMR	CFI	TLI	GFI	IFI	∆RMSEA	∆SRMR	∆CFI	∆TLI	∆GFI	∆IFI
Model 1	0.009	0.027	0.998	0.998	0.968	0.998	0.013	0.028	0.055	0.055	0.048	0.055
Model 2	0.022	0.055	0.943	0.943	0.920	0.943

). This analysis collectively indicates that common methodological bias in this study has been effectively controlled to a certain extent.

5.2. Exploratory Factor Analysis

SPSS 27.0 software was utilized to perform the exploratory factor analysis of the scales employed in this research, using the principal component analysis technique. The Kaiser–Meyer–Olkin (KMO) value obtained for the factor analysis equaled 0.771, indicating a satisfactory level of sampling adequacy, as it exceeded 0.7. Furthermore, the Bartlett Test of Sphericity yielded a result of 3773.142, with a df value of 210 and a significance level of 0.000, which is below 0.05. Overall, these findings support the appropriateness of the data for exploratory factor analysis.

Utilizing the gravel diagram to determine various factors, a rotation of maximum variance was implemented, ensuring that factors had eigenvalues exceeding 1. Subsequently, seven distinct factors were identified, and by eliminating measurement items with cross-loading, measurement indicators were categorized according to their respective factors. The findings from the exploratory factor analysis, presented in

Table 3. Exploratory factor analysis results.

Variables	Items	Loading	Eigenvalues	Explain the Variation Amount/%	Explain the Cumulative Variation Amount/%	Cronbach’s α
Risk Awareness (RA)	RA1	0.869	2.367	11.273	11.273	0.866
	RA2	0.866
	RA3	0.836
Knowledge Acquisition (KA)	KA1	0.859	2.292	10.916	22.189	0.846
	KA2	0.842
	KA3	0.840
Safe Production Behavior (SPB)	SPB1	0.834	2.210	10.522	32.711	0.819
	SPB2	0.825
	SPB3	0.843
Normative Regulation (NR)	NR1	0.841	2.129	10.140	42.851	0.785
	NR2	0.814
	NR3	0.824
Incentive Regulation (IR)	IR1	0.843	2.128	10.133	52.984	0.773
	IR2	0.815
	IR3	0.822
Service Regulation (SR)	SR1	0.804	2.064	9.826	62.810	0.761
	SR2	0.832
	SR3	0.798
Punitive Regulation (PR)	PR1	0.779	2.030	9.668	72.478	0.743
	PR2	0.832
	PR3	0.816

, reveal that all factor loadings for the measurement items surpass 0.600, with Cronbach’s α values exceeding 0.700. These outcomes highlight the successful analytical efficiency of the exploratory factor analysis carried out in this study.

5.3. Confirmatory Factor Analysis

Based on the research sample, seven first-order proposed structural models were developed with the AMOS 23.0 software. These models underwent comparison to uncover the most fitting factor structure. As displayed in

Table 4. Confirmatory factor analysis results.

Fit Indicators	χ2	df	χ2/df	RMSEA	CFI	IFI	TLI	GFI	AGFI	SRMR
Seven-factor model	174.560	168	1.039	0.009	0.998	0.998	0.998	0.968	0.955	0.027
Six-factor model	634.005	174	3.644	0.074	0.873	0.874	0.847	0.882	0.843	0.082
Five-factor model	1095.968	179	6.123	0.103	0.747	0.749	0.703	0.806	0.750	0.113
Four-factor model	1495.870	183	8.174	0.122	0.638	0.641	0.584	0.752	0.687	0.135
Three-factor model	2178.574	186	11.713	0.149	0.450	0.454	0.379	0.674	0.595	0.154
Two-factor model	2202.268	188	11.714	0.149	0.444	0.448	0.379	0.667	0.591	0.137
One-factor model	2615.752	189	13.840	0.163	0.331	0.334	0.256	0.621	0.537	0.141

Note: Seven-factor model: NR, PR, IR, SR, KA, RA, SPB; Six-factor model: NR + PR, IR, SR, KA, RA, SPB; Five-factor model: NR + PR + IR, SR, KA, RA, SPB; Four-factor model: NR + PR + IR + SR, KA, RA, SPB; Three-factor model: NR + PR + IR + SR + KA, RA, SPB; Two-factor model: NR + PR + IR + SR + KA + RA, SPB; One-factor model: NR + PR + IR + SR + KA + RA + SPB.

, the 7-factor model demonstrated the highest degree of fit (χ²/df = 0.009, TLI = 0.998, IFI = 0.998, CFI = 0.998, SRMR = 0.027, RMSEA = 0.009). The validation of the 7-factor model suggests that the study’s included factors possess strong discriminant validity and represent seven separate constructs. Consequently, utilizing a 7-factor model within this research aligns with the established criteria.

5.4. Model Fit Test

The results attained from the model fit test for the research model reveal that χ²/DF equals 1.047 [82], GFI equals 0.966 [83], AGFI equals 0.955, CFI equals 0.998 [84,85], TLI equals 0.997, RMSEA equals 0.010 [85,86], and SRMR equals 0.032.

All fit indices exceed conventional thresholds, and several values (e.g., CFI and TLI) were extremely high. While such results generally indicate a well-fitting model, we acknowledge that extremely high fit statistics may reflect potential overfitting rather than model “perfection”. Therefore, model adequacy was not judged solely on fit indices; instead, we relied on additional evidence including the scale’s reliability, validity and discriminant validity, and model stability tests. These complementary tests provide more balanced support for the model’s conceptual soundness and empirical robustness, reducing the risk of overinterpretation of fit metrics alone.

5.5. Measurement Model Analysis

Table 5. Scale reliability and validity tests results.

Variables	Items	Unstd.	S.E.	Z	p	Std.	Cronbach’s α	CR	AVE
Normative Regulation (NR)	NR1	1.000				0.797	0.785	0.788	0.554
	NR2	0.789	0.061	12.921	***	0.694
	NR3	0.956	0.072	13.252	***	0.738
Punitive Regulation (PR)	PR1	1.000				0.670	0.743	0.752	0.503
	PR2	0.947	0.084	11.214	***	0.711
	PR3	0.863	0.077	11.230	***	0.744
Incentive Regulation (IR)	IR1	1.000				0.759	0.773	0.788	0.553
	IR2	1.083	0.083	13.120	***	0.726
	IR3	1.421	0.107	13.231	***	0.746
Service Regulation (SR)	SR1	1.000				0.718	0.761	0.762	0.516
	SR2	0.999	0.084	11.912	***	0.723
	SR3	0.909	0.076	11.885	***	0.715
Knowledge Acquisition (KA)	KA1	1.000				0.840	0.846	0.848	0.651
	KA2	0.927	0.053	17.653	***	0.786
	KA3	0.825	0.046	17.770	***	0.793
Risk Awareness (RA)	RA1	1.000				0.916	0.866	0.870	0.693
	RA2	0.840	0.040	21.169	***	0.824
	RA3	0.846	0.045	18.864	***	0.748
Safe Production Behavior (SPB)	SPB1	1.000				0.754	0.819	0.819	0.601
	SPB2	1.086	0.072	15.187	***	0.800
	SPB3	1.052	0.070	14.969	***	0.771

Note: *** p < 0.01.

illustrates the outcomes of the reliability and validity assessments for the research scale. Each dimension has Cronbach’s alpha higher than 0.7, which signifies the research scale’s robust reliability [87]. Further, the composite reliability (CR) for every dimension surpasses 0.7, while the average variance extracted (AVE) for each dimension exceeds 0.5 [88]. Consequently, the research scale is considered to possess strong validity.

Moreover, the average extracted variance (AVE) square root for every variable exceeds the correlation coefficients among the variables, as shown in

Table 6. Discriminant validity test results. (a) AVE square root; (b) HTMT.

(a)
Variables	SPB	RA	KA	SR	IR	PR	NR
SPB	0.775
RA	0.315	0.832
KA	0.346	0.314	0.807
SR	0.238	0.289	0.209	0.719
IR	0.216	0.190	0.333	0.036	0.744
PR	0.192	0.269	0.129	0.017	0.031	0.709
NR	0.193	0.266	0.226	−0.015	0.070	0.005	0.744
(b)
Variables	NR	PR	IR	SR	KA	RA	SPB
NR
PR	0.059
IR	0.065	0.048
SR	0.051	0.067	0.047
KA	0.225	0.139	0.331	0.212
RA	0.280	0.267	0.171	0.287	0.315
SPB	0.183	0.191	0.218	0.237	0.348	0.310

Note: Values in bold are AVE open root value.

a. This implies that the scale demonstrates strong discriminant validity according to [89].

Following this, the evaluation of discriminant validity was conducted using the Heterotrait–Monotrait Ratio (HTMT). According to prior research, an HTMT value of less than 0.85 or 0.9 across all constructs indicates robust discriminant validity [90]. This study employs the stricter threshold of 0.85. The findings (see Table 6b) show that all HTMT values are below 0.85, which further substantiates the evidence for discriminant validity.

5.6. Structural Model Analysis

The research hypotheses and the results from the path analysis can be found in

Table 7. Direct effect test results.

Hypothesis	Path	Unstd. (β)	S.E.	Z	p-Value	Std.	Test Results
H1a	NR → SPB	0.121	0.043	2.767	***	0.155	Validated
H1b	PR → SPB	0.105	0.048	2.212	**	0.130	Validated
H1c	IR → SPB	0.103	0.053	1.966	**	0.116	Validated
H1d	SR → SPB	0.122	0.046	2.624	***	0.159	Validated
H2a	NR → KA	0.191	0.046	4.181	***	0.220	Validated
H2b	PR → KA	0.125	0.051	2.434	**	0.129	Validated
H2c	IR → KA	0.338	0.058	5.869	***	0.319	Validated
H2d	SR → KA	0.194	0.049	3.958	***	0.213	Validated
H3	KA → SPB	0.162	0.051	3.186	***	0.193	Validated
H5a	NR → RA	0.219	0.042	5.266	***	0.267	Validated
H5b	PR → RA	0.242	0.048	5.037	***	0.266	Validated
H5c	IR → RA	0.172	0.050	3.456	***	0.172	Validated
H5d	SR → RA	0.252	0.045	5.574	***	0.293	Validated
H6	RA → SPB	0.109	0.054	2.020	**	0.123	Validated

Note: ** p < 0.05; *** p < 0.01.

. The results demonstrate that every direct influence hypothesis suggested in this research shows significant positive correlations, and all the paths are deemed valid. As a result, each of the research hypotheses (direct influence) has been validated.

5.7. Mediation Effect Test

This study utilized the Bootstrap confidence interval method, as outlined by [91], to examine the mediating effects within the study. Initially, a confidence interval level of 95% was established, and the Bootstrap was programmed to conduct 1000 samples. Using Amos 29 software for analysis, a mediating effect is deemed present if the Z-value exceeds 1.96, and both the Bias-Corrected and Percentile confidence intervals do not encompass zero. Conversely, a mediating effect is considered absent if the interval includes zero.

In accordance with [91], this investigation utilized the Bootstrap approach to examine the mediating impacts within the study. Initially, a confidence interval of 95% was established, and the Bootstrap sampling frequency was adjusted to 1000 iterations. Utilizing the Amos software, a Z-value surpassing 1.96, in conjunction with non-zero values in both the Bias-Corrected and Percentile confidence intervals, indicates the presence of a mediating effect. Conversely, if the interval encompasses zero, the mediating influence is deemed non-existent.

Based on the findings presented in

Table 8. Mediation effect test results.

Effect Types	Point Estimate	Product of Coefficient		Bootstrapping
				Bias-Corrected 95% CI		Percentile 95% CI
		SE	Z	Lower	Upper	Lower	Upper
Mediation effects of KA
NR → KA → SPB	0.031	0.013	2.385	0.011	0.064	0.010	0.060
PR → KA → SPB	0.028	0.011	2.545	0.004	0.052	0.003	0.046
IR → KA → SPB	0.055	0.021	2.619	0.021	0.101	0.020	0.098
SR → KA → SPB	0.031	0.013	2.385	0.012	0.062	0.010	0.059
Mediation effects of RA
NR → RA → SPB	0.036	0.014	2.571	0.014	0.068	0.012	0.064
PR → RA → SPB	0.039	0.014	2.786	0.017	0.074	0.014	0.069
IR → RA → SPB	0.028	0.014	2.000	0.008	0.062	0.007	0.060
SR → RA → SPB	0.041	0.015	2.733	0.016	0.078	0.015	0.077
Mediation effects contrast of KA
NR → KA → SPB VS. PR → KA → SPB	0.011	0.014	0.786	−0.012	0.045	−0.014	0.041
NR → KA → SPB VS. IR → KA → SPB	−0.024	0.017	−1.412	−0.067	0.002	−0.063	0.004
NR → KA → SPB VS. SR → KA → SPB	0.000	0.012	0.000	−0.026	0.022	−0.025	0.024
PR → KA → SPB VS. IR → KA → SPB	−0.035	0.019	−1.842	−0.080	−0.006	−0.076	−0.004
PR → KA → SPB VS. SR → KA → SPB	−0.011	0.013	−0.846	−0.044	0.011	−0.042	0.013
IR → KA → SPB VS. SR → KA → SPB	0.023	0.016	1.438	0.000	0.068	−0.002	0.060
Mediation effects contrast of RA
NR → RA → SPB VS. PR → RA → SPB	−0.004	0.011	−0.364	−0.029	0.015	−0.026	0.017
NR → RA → SPB VS. IR → RA → SPB	0.008	0.013	0.615	−0.016	0.037	−0.019	0.034
NR → RA → SPB VS. SR → RA → SPB	−0.005	0.011	−0.455	−0.032	0.013	−0.030	0.016
PR → RA → SPB VS. IR → RA → SPB	0.011	0.013	0.846	−0.012	0.041	−0.017	0.038
PR → RA → SPB VS. SR → RA → SPB	−0.002	0.011	−0.182	−0.025	0.018	−0.024	0.019
IR → RA → SPB VS. SR → RA → SPB	−0.013	0.013	−1.000	−0.046	0.008	−0.041	0.011

, the mediating roles of KA and RA are significant. This significance is evidenced by their respective mediating pathways, which exclude zero in both the Bias-Corrected and Percentile 95% Confidence Intervals. Furthermore, the Z values for these pathways are all above 1.96.

Notably, by conducting a comparative analysis, it is evident that both variables have zero included in their Bias-Corrected 95% CI and Percentile 95% CI during the mediating effect contrast test. Additionally, their Z values are both below 1.96. Therefore, it is reasonable to conclude that there is no substantial variation in the mediating effects of these two variables through varying mediating pathways, and no statistically significant difference was detected. Further elaboration on this discovery will be provided in the subsequent sections.

5.8. Multi-Group Constancy Test

In order to assess whether the research model remains consistent across different groups, this study carried out a comparison of the model among multiple groups [92]. Sequential structural model invariance analysis was conducted to examine consistency in terms of structure, factor loadings, intercepts, factor variances and covariances, paths, structural residuals, and measurement residuals [93]. The outcomes of the examination of the gender variable are detailed in

Table 9. Test results of model invariance across different enterprise size groups.

Model	CMIN	DF	CMIN/DF	∆CMIN	∆DF	p	TLI	∆TLI	CFI	∆CFI
M1	565.436	525	1.077				0.987		0.989
M2	600.828	553	1.086	35.392	28	0.159	0.985	0.002	0.987	0.002
M3	630.146	581	1.085	29.318	28	0.397	0.985	0.000	0.987	0.000
M4	642.382	589	1.091	12.236	8	0.141	0.984	0.001	0.985	0.002
M5	648.790	595	1.090	6.408	6	0.379	0.984	0.000	0.985	0.000
M6	695.928	637	1.093	47.138	42	0.270	0.984	0.000	0.984	0.001

Note: M1: Morphological constancy; M2: Factor loading constancy; M3: Path constancy; M4: Factor variance/covariance constancy; M5: Structural residual constancy; M6: Measurement residual constancy.

. Equivalence limitations were applied in all assessments, and adherence to academic standards dictates that non-significant p-values (p > 0.05) indicate model invariance [94].

This study specifically categorized the companies surveyed into various research groups according to their size in order to investigate if there were notable distinctions among these groups within the model. Analysis of the test outcomes reveals that, in line with the specified criteria for uniformity, as all model p-values exceed 0.05, it can be deduced that there are no remarkable variations among the different-sized enterprise groups in the model. Consequently, the findings illustrate that the research model in this study showcases uniformity across the groups.

The discovery that the research model demonstrates cross-group measurement invariance implies that the mechanisms by which government regulation affects safe production behavior function similarly, mediated by knowledge acquisition and risk awareness, regardless of the agricultural enterprise’s size. This similarity can be attributed to the universal nature of regulatory compliance requirements, which are structured to apply to different size enterprises within the sector.

The validation of measurement uniformity among different size categories of farming businesses in this investigation emphasizes the strength and applicability of the research framework. This consistency implies that the concepts of government regulation, knowledge acquisition, risk awareness, and safe production behavior are assessed uniformly and can be effectively contrasted among companies of varying magnitudes. Therefore, the connections among these factors, as clarified by the framework, are expected to remain valid for farming businesses irrespective of their operational magnitude.

The results of this study carry significant implications for policymakers and practitioners serving the agricultural industry. Due to the proven invariance of the model, the insights gained from this research can be universally applied, guiding regulatory measures and safety protocols that are effective regardless of enterprise size. This universality increases the practical value of the research, indicating that the mechanisms identified for encouraging safe production behavior are applicable broadly across this sector, rather than being dependent on the size of the enterprise.

In the future, the research model’s consistency across businesses of varying sizes offers potential for further investigation and application. For instance, future studies could analyze the model’s effectiveness in different geographic locations or under varying regulatory frameworks to assess its versatility. Additionally, exploring the possibility of companies maintaining consistency despite differences in their founding dates and levels of innovation is worthwhile.

In summary, the research model shows consistent measurement across various sizes of agricultural businesses, confirming its reliability and potential for wider use. The uniform application of regulatory standards and knowledge and risk management practices within the industry may explain this consistency. This discovery suggests that the model’s insights are applicable beyond specific agricultural sectors, offering valuable information for policies and practices across a variety of enterprises.

Essentially, showcasing measurement invariance boosts the credibility of the research model, thereby enhancing its relevance and significance in practice and theory. However, it is important to emphasize that the invariance between different size groups in this study does not imply that the model can be automatically applied to different regulatory environments or countries.

5.9. Model Stability Test (Cross-Validation)

Following the suggestions made by [95], in order to confirm the reliability of the study results, this research utilized Amos’ group comparison function to perform cross-validation, which involved randomly splitting the sample into two separate groups.

Table 10. Model stability test result.

Model	∆DF	∆CMIN	p	∆NFI	∆IFI	∆RFI	∆TLI	∆CFI
Measurement weights	14.000	22.060	0.077	0.005	0.006	0.002	0.002	0.003
Structural weights	14.000	15.561	0.341	0.004	0.004	0.000	0.000	0.000
Structural covariances	4.000	2.065	0.724	0.001	0.001	−0.001	−0.001	−0.001
Structural residuals	3.000	3.039	0.386	0.001	0.001	0.000	0.000	0.000
Measurement residuals	21.000	19.107	0.578	0.005	0.005	−0.001	−0.001	0.000

displays the findings which reveal that every congruence configuration meets the criteria for acceptance (given that the p-values exceed 0.05), aligning with the standards set by [96]. As a result, this experimental research model satisfies the requirements for cross-validation and demonstrates reliability.

6. Conclusions and Discussion

6.1. Research Findings

The impact of government regulation on safe production practices of agricultural enterprises within the context of food safety and sustainable development was empirically explored in this study.

Over the course of research tracked from January 2021 to January 2024, this study uncovered several key findings: First, businesses largely perceive that normative regulation contributes to the creation of industry standards, thereby improving product quality and safety. However, they also highlight challenges related to implementation costs and the complexity involved. On the other hand, punitive regulation is viewed as a strong deterrent, effectively compelling businesses to adhere to established guidelines. Nonetheless, there is a worry that an over-reliance on fines might result in disproportionately high compliance expenses. Regulation incentives, like tax breaks and subsidies, are seen as successful in driving businesses to embrace new technologies and sustainable practices. However, maintaining fairness and transparency in these incentives is crucial. Enterprise’s view service regulations, such as offering consulting and technical assistance, as beneficial, though perceived fairness declined from 4.059 in 2021 to 3.782 in 2024. This type of regulation aids in improving their understanding and compliance with rules, ultimately boosting their competitiveness.

Using the three-wave survey from March 2024 to May 2024 and performing an empirical analysis using the structural equation modeling approach, the findings reveal that four distinct types of government oversight—normative regulation, punitive regulation, incentive regulation, and service regulation—not only directly foster safe production within agricultural businesses but also act indirectly through mediators such as knowledge acquisition and risk awareness. Additionally, each type of regulation bolsters knowledge acquisition and elevates risk awareness within enterprises. Simultaneously, knowledge acquisition and risk awareness play significant roles in promoting the safe production practices of agricultural enterprises. Furthermore, comparing the mediating impacts reveals that knowledge acquisition and risk awareness do not significantly differ in effect sizes across the four regulatory pathways (Table 8). This non-statistically-significant difference was detected suggests that both mechanisms—cognitive–technical capacity building (knowledge acquisition) and preventive attitudinal orientation (risk awareness)—play equally important and complementary roles in translating regulatory interventions into behavioral changes, instead of one mechanism dominating the other. This finding enriches prior research that focuses primarily on coercive compliance or economic incentives, instead highlighting the central role of information diffusion and normative influence in agricultural safety governance.

Additionally, verifying measurement invariance across different size categories of agricultural enterprises in this research emphasizes the robustness and generalizability of the study model. This invariance indicates that the constructs related to government regulation, knowledge acquisition, risk awareness, and safe production behavior are measured consistently and can be meaningfully compared among enterprises of varying sizes. Therefore, the relationships between these variables, as described by the model, are likely to be valid for agricultural enterprises irrespective of their scale. However, it is worth noting that the invariance between different size groups in this study does not imply that the model can be automatically applied to different regulatory environments or countries.

The panel-tracking design provides temporal context, while the 2024 SEM identifies stable structural relationships. The SEM analysis captures the interrelationships between government regulation, knowledge acquisition, risk awareness, and safety production at a single point in time, while the longitudinal data track how these relationships evolve over a three-year period. This temporal approach is particularly important in understanding how the effects of government regulation on safety production change over time, as enterprises become more attuned to regulatory changes and adjust their internal processes. The longitudinal survey also highlights the resilience and adaptability of agricultural enterprises in responding to different regulatory types, which might not be immediately apparent from the SEM analysis alone. Together, these two approaches offer a comprehensive view of the impact of government regulation on safety production, blending a snapshot of causal relationships with insights into their development over time.

6.2. Theoretical Contributions

To contextualize our contributions, it is essential to situate our findings within prior research. Existing studies reveal three main limitations. First, most rely on cross-sectional designs [22,24,97], which cannot capture regulatory dynamics over time. Second, prior work often examines single regulatory instruments in isolation [33,98], whereas our framework of four regulatory types enables systematic comparison. Third, few studies have empirically tested dual mediation mechanisms; our evidence that knowledge acquisition and risk awareness exert equivalent mediating effects challenges the prevailing emphasis on coercive or economic pathways as the primary regulatory mechanisms.

At its core, this study enriches the regulatory theory by delineating a nuanced landscape of government regulation in the agricultural sector, thereby advancing our understanding of how different regulatory strategies—normative, punitive, incentive, and service—affect enterprise behavior and outcomes. It enriches the monolithic view of regulation by elucidating the differential impacts and complexities involved.

Second, this research significantly bolsters the theoretical framework of institutional theory, particularly by examining how external regulatory pressures shape the internal mechanisms of organizations. The empirical evidence of the mediating role of knowledge acquisition and risk awareness in the regulatory context provides a robust foundation for understanding how institutions are internalized and operationalized within firms, contributing to a more dynamic and nuanced view of institutional isomorphism and adaptation.

Third, in the realm of the resource-based view (RBV), the study offers an innovative perspective by considering government regulation as a critical resource that can enhance a firm’s capabilities. By demonstrating that regulatory interventions can foster knowledge acquisition and heighten risk awareness, the research expands the RBV to include regulatory inputs as a source of competitive advantage, which is particularly relevant in the agricultural sector where safety standards are paramount.

Fourth, the study also makes an original contribution to the theory of organizational learning by highlighting the role of regulation in facilitating a learning environment that promotes safer production practices. This underscores the importance of regulatory frameworks in not only shaping immediate compliance but also fostering long-term learning and adaptation within organizations.

In addition, the finding of mediation effects’ non-statistically-significant difference was detected, wherein both knowledge acquisition and risk awareness exhibit similar mediating effects in the relationship between government regulation and safety production, which carries significant theoretical implications for understanding the mechanisms that drive regulatory impacts. This finding suggests that government regulations influence safety production in agricultural enterprises through two parallel yet equally important pathways: one that enhances the intellectual capacity of organizations (knowledge acquisition) and another that improves their understanding of safety risks (risk awareness). Theoretical frameworks often focus on either cognitive processes or psychological factors separately, but this study’s findings indicate that both are integral to the success of regulatory interventions. By demonstrating that both mediators play comparable roles, this study suggests that regulatory types—whether normative, punitive, incentive-based, or service-oriented—are equally effective in facilitating safety outcomes through these two dimensions. The implication is that regulatory effects are not dependent on a single mechanism but rather operate through a combination of knowledge transfer and heightened safety consciousness, offering a more comprehensive view of regulatory influence on organizational behavior.

Fifth, furthermore, utilizing a protracted longitudinal inquiry, this study showcases the ever-changing interplay between governmental regulation and corporate conduct. This temporal perspective is reinforced by recent evidence that leading enterprises can serve as knowledge intermediaries in diffusing safety standards [99]. This methodology not only assists scholars in gaining deeper insights into the progression of regulatory measures and their enduring influence on corporate practices but also establishes a factual basis for shaping the theory of regulatory dynamics. Ultimately, it serves as a valuable point of reference for forthcoming investigations.

Sixth, the study makes a significant contribution to the existing literature by presenting empirical proof that the elements of government regulation, acquisition of knowledge, awareness of risks, and engagement in safe behavior are consistently assessed in enterprises of different scales. This consistency in measurement is an important theoretical progression as it confirms the comparability of these elements and enables valuable comparisons across different groups. It indicates that the theoretical framework established in this study is robust and can be utilized to analyze safety dynamics in agricultural enterprises, regardless of their scale.

Seventh, the study’s findings have implications for future research by suggesting that the model’s constructs and relationships may be transferable to other contexts. The demonstrated invariance across enterprise sizes opens the door for the model to be tested in different geographical locations, regulatory environments, or even other industries where similar constructs may be relevant. This potential for generalization enhances the model’s theoretical value and suggests that the insights derived from this research could inform safety practices and regulatory strategies in a wide range of settings.

Eighth, this study holds substantial importance for global agricultural theoretical development from an international standpoint. By drawing on these conclusions, other countries and regions can create specific theoretical measures and actions that support safe agricultural production, improve product quality, and boost international competitiveness. Ultimately, this contributes to the sustainable development of agriculture on a global scale.

Lastly, the study offers a sophisticated theoretical contribution to the field of strategic management by integrating these various theoretical perspectives into a cohesive framework that explains the multifaceted influence of government regulation on safety production in agricultural enterprises. It provides a comprehensive model that can guide both policy-makers in designing effective regulatory strategies and managers in aligning their safety production behaviors with regulatory requirements, thus bridging the gap between theory and practice in a manner that is both enlightening and actionable.

6.3. Practice and Management Contributions

This study makes important contributions in practical and managerial aspects, both at the enterprise and government policy levels. These contributions align with recent policy analyses emphasizing differentiated regulatory strategies [28,100]. At the enterprise level, the research offers precise strategies and guidelines for enhancing safe production practices. Organizations can adjust their internal training and management practices in accordance with various forms of government regulation (normative, punitive, incentive, service), develop efficient risk management and compliance frameworks, invest in sustainable technologies and practices, and engage proactively in government-provided service-focused regulations, ultimately boosting their regulatory flexibility and competitiveness. Moreover, the research highlights the crucial intermediary function of acquiring knowledge and being aware of risks in connecting government regulations with safe production practices in corporations. It suggests that companies should boost employees’ grasp of regulations and risk awareness through ongoing training and educational initiatives, to better comply with government requirements and elevate safety standards. Additionally, businesses should proactively seek out and implement strategies like subsidies and tax breaks in incentive-driven regulations to reach both economic gains and environmental accountability.

According to the findings of this study, the following specific measures for enhancing safe production practices in agricultural enterprises are recommended:

Normative regulation: (1) Actively acquire and adhere to the agricultural standards for safe production and operational procedures set forth by the government to ensure that production activities are in accordance with legal and regulatory requirements. (2) Implement an internal management system for safe production, perform routine self-examinations and evaluations to guarantee the efficient execution of all safety measures. (3) Enhance employee training in safe production to improve their comprehension and adherence to safe production regulations.

Punitive regulation: (1) Implement a thorough internal system for reporting and addressing violations, urging employees to report safety hazards and non-compliance behaviors proactively. (2) Hold frequent training sessions on safe production laws and regulations to guarantee that employees grasp the repercussions of non-compliance and improve their self-control. (3) Establish consistent communication with government regulatory bodies to stay informed about current regulatory requirements and penalty guidelines, preventing unnecessary violations.

Incentive regulation: (1) Proactively seek government-offered incentives and financial aid aimed at enhancing safety production, and utilize these funds to upgrade safety facilities and technological processes. (2) Integrate safety production performance metrics into the employee assessment and rewards system, incentivizing staff to engage actively in safety practices. (3) Engage in safety production competitions and evaluations coordinated by governmental bodies, achieving recognition and rewards for excellent performance.

Service regulation: (1) Employ the secure production consultation services offered by the government to tackle safety concerns in the manufacturing process and boost the standard of secure production. (2) Engage in secure production workshops and technical conferences arranged by the government or industry groups to enhance employees’ understanding and abilities in secure production. (3) Create communication channels with government regulatory bodies to promptly receive policy and technical updates concerning secure production.

Knowledge acquisition: (1) Create an internal platform for sharing knowledge on safe production within the organization, motivating staff to exchange their insights and expertise in safe production. (2) Arrange frequent training sessions on safe production knowledge, featuring presentations by specialists to boost the proficiency of employees. (3) Partner with universities and research organizations to engage in research and development endeavors concerning safe production, gaining access to the most up-to-date knowledge in this domain.

Risk awareness: (1) Regularly conduct safety risk assessments in production to identify potential safety hazards and create appropriate preventive and response strategies. (2) Implement a safety production accident learning program, examining both internal and external industry incidents to improve staff’s understanding of risks. (3) Develop an early warning system for safety risks in production to quickly address government-issued safety alerts and implement preventative actions.

By implementing the aforementioned measures, agricultural businesses can more efficiently address government regulations, improve their internal safety production protocols, and advance their overall safety production with the intermediary benefits of knowledge acquisition and increased risk awareness.

At the government policy level, this study offers empirical proof for policymakers on enhancing regulatory strategies. Additionally, policymakers can use the direct evidence presented in the study regarding the efficiency of various government regulations to evaluate the effects of current regulatory measures and implement any required modifications to boost regulation efficiency and effectiveness.

Based on the tracking survey results of the long-term dynamic changes between government regulation and enterprises from 2021 to 2024, the following improvement measures are proposed to optimize the effectiveness of government regulation:

Normative regulation: (1) The role of normative regulation in setting industry standards is increasingly recognized as significant. However, the reduction in growth rates indicates potential regulatory fatigue or diminished regulatory efficiency. Therefore, the government should reevaluate normative regulation implementation to ensure its effectiveness and efficiency. It is advisable for the government to streamline regulatory requirements, offer clearer guidance, and use digital technology to improve the transparency and functionality of regulation. (2) The costs and complexity of implementation as perceived by enterprises continue to rise, though at a decelerated rate, which could mean that businesses are gradually adapting to regulatory measures. The government should keep optimizing regulatory processes, minimizing unnecessary administrative burdens, and providing training and support to help enterprises better understand and comply with regulations.

Punitive regulation: Relying too much on fines could result in increased compliance expenses, and despite the decrease in growth, it is necessary for the government to consider a variety of regulatory strategies. The suggestion is for the government to implement additional rewards for businesses that follow regulations and offer advisory support for compliance in order to lessen the reliance on fines.

Incentive regulation: (1) A decrease in the rate of growth in the effectiveness of incentive regulation suggests the need for potential updates or modifications to existing incentive measures. It is important for the government to periodically assess incentive policies to ensure they align with market advancements and technological innovations, while also improving policy transparency to mitigate uncertainties for businesses. (2) The diminishing fairness and transparency ratings for incentive measures, albeit at a slower rate, warrant careful consideration. The government should strive to enhance the transparency and predictability of regulatory actions to ensure all businesses have equal opportunity to access incentives, and clearly outline the criteria and benchmarks for eligibility.

Comprehensive regulatory approach: A holistic approach to regulation should be embraced by the government, combining normative, punitive, incentive, and service aspects to establish a balanced and efficient regulatory framework. It is advisable for the government to engage in ongoing communication with businesses, solicit input, and promptly modify regulatory actions to guarantee that regulation achieves desired outcomes without placing undue burdens on enterprises.

In addition, the non-statistically-significant difference in the mediating effects of knowledge acquisition and risk awareness has critical implications for regulatory policy design. Since both factors equally facilitate improvements in safety production, policymakers and regulatory bodies should prioritize interventions that address both knowledge and awareness in tandem. This suggests that regulatory policies should not merely focus on providing information or training (knowledge acquisition) or on enforcing safety compliance (risk awareness) in isolation. Instead, a more integrated approach is necessary, where both aspects are nurtured simultaneously. For example, safety regulations could include a balanced mix of educational initiatives that provide enterprises with the necessary knowledge, along with awareness campaigns or incentives that enhance recognition of safety risks. This dual approach ensures that safety production is enhanced on both cognitive and psychological fronts. Furthermore, it also suggests that the regulatory framework should be flexible and adaptable to various types of enterprises, acknowledging that different regulatory strategies—normative, punitive, incentive-based, and service-oriented—can all effectively foster safety outcomes, as long as they simultaneously encourage both knowledge enhancement and risk awareness.

Beyond the specific focus on the Chinese context in this study, we recognize that the findings may have broader implications, particularly for other emerging economies experiencing similar socio-economic transitions. While the specific institutional and cultural factors in China influence the results, many of the trends identified, such as the relationship between policy implementation and market response, may be observed in other developing nations facing comparable challenges in modernization and economic development. Therefore, while caution is warranted in generalizing across diverse contexts, the insights gained from this study could inform similar research and policy-making in other countries with evolving market structures and developmental trajectories.

To sum up, this study offers practical and managerial contributions by equipping enterprises with specific strategies and actionable guidelines to refine safety production behaviors, bolster risk management and knowledge acquisition, and foster sustainable development practices. Concurrently, it supplies the government with empirical evidence to refine regulatory approaches, enhance regulatory efficiency and effectiveness, and move closer to achieving food safety and sustainable development objectives.

7. Research Limitations and Future Research

The limitations of this study necessitate discussion, as several methodological and contextual constraints warrant acknowledgment.

Firstly, our reliance on self-reported data from senior managers may introduce social desirability bias, although we employed procedural remedies (anonymity assurances, three-wave collection) and statistical controls (Harman’s test). Future research could triangulate these findings with objective indicators (e.g., inspection records, accident reports) or adopt multi-informant designs. Also, while our longitudinal tracking spans 2021–2024, the SEM analysis relies on cross-sectional 2024 data, thereby limiting causal inference. In other words, although data were collected over multiple years, the study does not estimate longitudinal causal models. Future research should employ cross-lagged or growth SEM to test dynamic regulatory effects. Furthermore, our findings are situated within the Chinese agricultural context, characterized by strong governmental presence and Confucian norms emphasizing compliance; cultural differences may therefore affect regulatory effectiveness in other settings. Initially, the focus of the research was exclusively on agricultural businesses, restricting the extensiveness of the conclusions. Companies in diverse sectors may display differing reactions and adjustment tactics towards governmental policies, making the results non-transferrable to other fields. Subsequent research ought to contemplate broadening the sample size to incorporate businesses from a variety of sectors to corroborate and contrast the reactions and adjustment tactics towards governmental policies across various domains, consequently augmenting the inclusiveness and relevancy of the research findings.

Secondly, this study was conducted within China’s regulatory structure. The political, economic, and cultural background may differ significantly from that of other nations, potentially influencing the relevance of the findings elsewhere. To examine the transferability of these results, future research could conduct cross-national comparative analyses under distinct regulatory frameworks, assessing the influence of government oversight on corporate conduct across diverse cultural settings and exploring the similarities and differences in these effects. Specifically, while our core insights into dual mediation pathways—knowledge acquisition and risk awareness—may hold across diverse contexts, the relative effectiveness of the four regulatory types likely varies depending on institutional maturity and cultural orientations toward authority. These insights may be more directly transferable to developing or transition economies with similar governance characteristics, such as strong state involvement and smallholder production systems, than to developed economies with mature regulatory institutions and different risk cultures.

Thirdly, the research might not have covered all the factors affecting safety production behaviors in corporations, such as market rivalry, supply chain supervision, corporate ethos, and more, that could have a substantial influence on safety production practices. Subsequent studies ought to include a wider array of elements that could potentially influence safety production behaviors within organizations in order to forge a more all-encompassing research framework, allowing for more precise prognostications and interpretations of corporate conduct.

Fourthly, the research solely focused on a four-year observation period, potentially limiting its ability to fully comprehend the comprehensive range of enduring fluctuations in the interaction between government oversight and corporate perspectives. Furthermore, this study only used survey data from 2024 for its impact mechanism analysis; therefore, it acknowledges the limitations of using cross-sectional data for causal inference. To enhance comprehension of the sustained dynamic correlation between government regulations and corporate practices, future investigations should consider conducting longer-term tracking surveys to grasp the enduring impacts of regulatory adjustments and corporate response tactics.

Fifthly, data for this research was mainly gathered via questionnaires and surveys. Despite the absence of common method bias in this study, in order to diminish its effects, upcoming investigations could utilize various data collection techniques. For instance, incorporating interviews, surveys, and factual data could bolster the diversity and objectivity of the information gathered.

Finally, upcoming studies may also investigate the nonlinear connections between government regulations and corporate safe production practices, alongside the variations in regulatory efficacy during diverse economic periods and market situations. Moreover, the inquiry could delve deeper into how regulatory creativity can amplify the efficiency and efficacy of regulations, and how to maintain a harmonious blend of rigidity and adaptability in regulations to foster secure operations and lasting growth within businesses.