Enhancing Farmer Resilience Through Agricultural Insurance: Evidence from Jiangsu, China

Abstract

Against the backdrop of evolving global climate patterns, the frequency and intensity of extreme weather events have increased significantly, posing unprecedented threats to agricultural production. This change has particularly profound impacts on agricultural systems in developing countries, making the enhancement of farmers’ capacity to withstand extreme weather events a crucial component for achieving sustainable agricultural development. As an essential safeguard for agricultural production, agricultural insurance plays an indispensable role in risk management. However, a pronounced gap persists between policy aspirations and actual adoption rates among farmers in developing economies. This study employs the integrated theory of planned behavior (TPB) and protection motivation theory (PMT) to construct an analytical framework incorporating psychological, socio-cultural, and risk-perception factors. Using Jiangsu Province—a representative high-risk agricultural region in China—as a case study, we administered 608 structured questionnaires to farmers. Structural equation modeling was applied to identify determinants influencing insurance adoption decisions. The findings reveal that farmers’ agricultural insurance purchase decisions are influenced by multiple factors. At the individual level, risk perception promotes purchase intention by activating protection motivation, while cost–benefit assessment enables farmers to make rational evaluations. At the social level, subjective norms can significantly enhance farmers’ purchase intention. Further analysis indicates that perceived severity indirectly enhances purchase intention by positively influencing attitude, while response costs negatively affect purchase intention by weakening perceived behavior control. Although challenges such as cognitive gaps and product mismatch exist in the intention-behavior transition, institutional trust can effectively mitigate these issues. It not only strengthens the positive impact of psychological factors on purchase intention, but also significantly facilitates the transformation of purchase intention into actual behavior. To promote targeted policy interventions for agricultural insurance, we propose corresponding policy recommendations from the perspective of public intervention based on the research findings.

1. Introduction

Under the global context of climate change, the occurrence of natural disasters has become increasingly frequent and intensified in severity [1]. Agriculture, as a fundamental sector of the national economy, not only serves as a vital livelihood safeguard for rural populations in developing countries [2], but also constitutes a central pillar of their food security. However, its high reliance on natural conditions makes it the primary bearer of climate change impacts, a contradiction that is particularly pronounced in developing countries that heavily depend on agriculture yet lacking risk management resources [3]. It has been shown that the global warming phenomenon poses unprecedented challenges to rice production in Southeast Asian countries and China [4]. Meanwhile, according to statistics from the Emergency Events Database (EM-DAT), the number of global natural disaster events has surged from an average of 100 per year in the 1970s to over 400 per year in the past two decades, with the agricultural sector accounting for more than 70% of the damage. This trend not only poses a direct threat to agricultural productivity but also exacerbates the global food security crisis by undermining the sustainability of agricultural ecosystems.

The impact of climate change on agriculture is not only reflected in the frequency and intensity of disasters, but also significantly reduces the predictability of farmers’ harvests. China, for example, as the world’s largest agricultural producer, its food security is increasingly threatened by climate change. According to monitoring data from the Ministry of Agriculture and Rural Affairs, China loses up to 60 billion pounds of grain each year due to meteorological disasters, equivalent to approximately 5% of the country’s total grain output. Additionally, pests, diseases, and weeds lead to losses of 50–60 billion pounds of grain annually. These disasters are now forming a compound threat [5]. Lee et al. [6] have pointed out that climate change has significantly impacted China’s food security, introducing more uncertainty risks. This decline in harvest predictability, represented by yields, is fundamentally shaking the foundation of stability in agricultural production.

Against this backdrop, the importance of ex ante insurance mechanisms as a core tool for agricultural risk management has become increasingly prominent [7]. Agricultural insurance, through its risk transfer mechanism [8], can provide financial compensation to farmers for crop losses caused by natural disasters, thus effectively maintaining the stability of agricultural production [7].

In recent years, governments in many countries have attached great importance to the development of agricultural insurance and promoted the expansion of agricultural insurance market by introducing a series of public interventions such as premium subsidy policies. Nevertheless, despite these efforts, the voluntary participation rates among farmers in developing countries remain generally low [9]. Research shows that China’s agricultural insurance coverage level is only one-fifth that of the United States, one-third of Canada’s, and one-third of Japa’s, lagging far behind developed countries [10]. In addition, climate change also makes the risk assessment and product design on the supply side of agricultural insurance face great challenges. Due to the frequent occurrence of extreme events, the concentrated exposure of systemic risks will bring great instability to the existing risk prevention system. At the same time, the gap between farmers’ individualized demand for climate risks and insurance companies’ standardized products may further widen, escalating the conflict between supply and demand in the agricultural insurance market.

Against the above background, it has become crucial to study farmers’ willingness to purchase agricultural insurance and their purchasing behavior to help them improve their ability to withstand extreme weather events [11].

Given these considerations, we take Jiangsu Province, a high comprehensive risk area in China, as the research area, and use questionnaire surveys to obtain first-hand data. With an integration of the theory of planned behavior (TPB) and the protection motivation theory (PMT), we apply the structural equation modeling to explore the factors affecting farmers’ purchase intention. The aim is to answer the following core questions: what are the key factors driving farmers’ willingness to purchase agricultural insurance? How do these key factors play a role in purchase intentions? What is the relationship between farmers’ willingness to purchase agricultural insurance and their behavior? What is the role of institutional trust in farmers’ agricultural insurance purchase decisions? This is not only of great value for enhancing the farmers’ participation rate in agricultural insurance and perfecting China’s agricultural insurance system but also provides powerful support for stabilizing agricultural production and promote sustainable agricultural development.

Compared with prior studies, this study’s key contributions include: first, this study contributes to theoretical innovation by employing an integrated TPB-PMT model, addressing a gap in existing agricultural insurance research. Current studies on farmers’ insurance decisions predominantly adopt unidimensional analytical frameworks. For instance, Chen et al. [12] utilized Probit models to examine climate risks’ impact on pastoralists’ insurance purchasing behaviors, while Lan et al. [13] applied Logit models to investigate determinants of crop insurance participation in Vietnam’s Mekong Delta under climate change. Although these studies have identified several key factors, they failed to systematically reveal the underlying mechanisms and interactive effects. Regarding the integrated TPB-PMT model, while existing research has applied this approach in areas such as farmers’ technology adoption and environmental protection, its application in agricultural insurance remains underexplored. This study innovatively integrates both theories to examine factors influencing farmers’ agricultural insurance purchase intentions and behaviors. The integrated framework not only captures multidimensional decision-making factors but also enables comprehensive analysis of their interrelationships and underlying mechanisms.

Second, the innovation of the research background, Jiangsu Province, a key cereal-producing zone in China, but also a typical region with frequent natural disasters. The vulnerability of its agricultural production and the urgency of risk management coexist here. However, the existing literature has paid relatively little attention to the agricultural insurance decisions of farmers in the high comprehensive risk areas of China’s eastern coast. This study will fill this research gap in this specific region. Thirdly, this paper innovates in mechanism analysis by revealing the dual moderating role of institutional trust in farmers’ agricultural insurance decision-making. This finding breaks through the limitation of existing studies that mostly focus on the unidirectional influence of institutional trust and can provide theoretical support for policy makers to optimize the design of agricultural insurance system from the perspective of public intervention.

The rest of this paper is organized as follows: Section 2 reviews the previous literature. Section 3 introduces the relevant theories and proposes research hypotheses. Section 4 describes the data sources, research implementation, sample characteristics, and the constructed model. Section 5 presents the results of empirical analysis. The empirical results are further discussed in Section 6. Finally, Section 7 sets out the conclusions, policy recommendations in the light of the analysis results, limitations, and future implications of the study.

2. Literature Review

Scholars have conducted multifaceted studies on farmers’ intentions and behavior to purchase agricultural insurance, providing a robust theoretical and empirical foundation for understanding the complex phenomena in this field.

Early studies mainly focused on the product characteristics of agricultural insurance and market failure issues. Some scholars defined agricultural insurance as a quasi-public good, pointing out that insufficient government subsidies would lead to a shortage in supply [14]. Chinese scholars attributed market failure to its quasi-public good nature. For instance, Tu Guozhu and Wang Guojun [15] showed that insufficient government subsidies led to a scarcity in agricultural insurance supply, eventually resulting in a “double freeze” in both supply and demand. Additionally, adverse selection and moral hazard caused by information asymmetry were also regarded as key reasons for the supply shortages [16,17], which might cause insurance companies to suffer continuous losses and eventually withdraw from the insurance market.

Research on the demand side has concentrated on the internal and external factors influencing farmers’ decisions. Internal factors include farmers’ individual characteristics and family operation features, such as age, education level, gender, scale of cultivated land, and income structure [13,18,19]. Regarding external factors, government subsidy policies, the credit of insurance companies, and other factors can affect farmers’ decisions to participate in insurance [20,21]. In recent years, the research perspective on farmers’ purchasing decisions has shifted from traditional utility theory to behavioral economics [22]. Scholars have paid more attention to the influence of psychological and social factors such as risk perception, insurance cognition, and peer effects on farmers’ decisions. Empirical studies have shown that risk attitudes increase farmers’ likelihood of enrolling in agricultural insurance [23,24,25,26]. Meanwhile, insurance cognition also influences farmers’ purchasing behavior [27,28,29]. For example, Dragos et al. [27] pointed out that farmers’ knowledge of insurance terms and claims processes significantly influences their willingness to adopt agricultural coverage. Furthermore, research by Wu et al. [22] demonstrates that social influence among farming communities enhances participation rates by heightening risk awareness and improving understanding of insurance benefits.

Research on farmers’ intention and behavior to purchase agricultural insurance in the Chinese context has also been fruitful. Regarding the driving factors for participation, Liu et al. [30] conducted a study based on panel data from 31 provinces in China and the GMM dynamic panel data model, and found that agricultural insurance affordability, and management are the main influencing factors. Some studies have focused on specific region samples. For example, Guo et al. [10] conducted a survey of farmers in Inner Mongolia and used the Heckman two-stage model to comprehensively analyze the differentiated choices of different types of farmers regarding agricultural insurance. Peng et al. [25], based on data from farmers in Shandong Province, employed cluster analysis, variance analysis, hierarchical linear models, and structural equation modeling to investigate how catastrophic events and risk assessments influence farmers’ decisions to invest in insurance.

3. Theoretical Analysis

3.1. Theoretical Foundation

TPB, proposed by Icek Ajzen in 1991, is an extension of the theory of reasoned action. According to TPB, an individual’s intention to perform a specific behavior in the near future determines their actual behavior, and this intention is influenced by three key factors: attitude, subjective norms, and perceived behavior control [31]. PMT is a psychological risk assessment framework that explains individuals’ tendency to protect themselves when faced with threats. The core premise of PMT posits that individuals engage in two key cognitive processes when confronting risks: threat assessment and coping assessment. Threat assessment zeroes in on how dangerous someone “thinks” the risk is [32], boiling down to two main factors: how vulnerable they feel and how bad the consequences seem. On the flip side, coping assessment is all about whether they believe they can actually “do” anything about it. This involves weighing four different things: how effective they think their response will be, how easy it is to carry out that response, what the downsides of responding might be, and whether the rewards are worth the effort [5].

Compared with TPB, PMT employs a broader range of predictive factors, not only focusing on the individual costs of adaptive behaviors, but also considering various aspects of collective actions, such as reaction efficiency [33]. Given the strong openness of the TPB, researchers can adapt it to specific contexts to enhance explanatory power [34]. Therefore, we integrate the two theories to conduct an empirical analysis of farmers’ intentions to purchase agricultural insurance. This combination was also employed by Badsar et al. [35] to examine the primary determinants of sustainable environmental behaviors among farmers and their behavioral intentions in different agricultural sectors. Wang et al. [36] demonstrated that integrating these theories yields stronger predictions compared to independent applications of TPB and PMT.

As an important theoretical basis for the study of farmers’ behaviors, there is an intersection between the perceived behavior control in TPB and the reaction efficiency in PMT. Farmers’ agricultural insurance options is an assessment of their ability to purchase agricultural insurance. Therefore, in the subsequent research, perceived behavior control and reaction efficiency will be integrated.

3.2. Research Hypotheses Based on TPB-PMT

3.2.1. TPB and Agricultural Insurance Purchase Intention

Based on the TPB, attitudes, subjective norms and perceived behavioral control all have a significant impact on farmers’ willingness to purchase agricultural insurance.

Attitude contains instrumental attitude and emotional attitude, the former mainly reflects farmers’ rational cognition of the practicality and functionality of agricultural insurance. Agricultural production is highly dependent on the natural environment, and extreme weather, such as typhoons and floods, may lead to crop extinction, while market prices may also directly affect income stability. When farmers believe that agricultural insurance can effectively cover these potential losses through the risk transfer mechanism, they will regard it as a necessary production input [8,11]. Affective attitudes reflect farmers’ psychological preferences and emotional identification with agricultural insurance. If farmers have witnessed others reduce their losses through insurance claims, or if they themselves have experienced successful claims, they will develop an affective tendency to believe that insurance is reliable. When farmers’ attitudes toward insurance programs change from negative to positive, their willingness to pay increases significantly [37]. Combining the effects of farmers’ instrumental and affective attitudes on willingness to buy, we are able to find that behavioral attitudes can positively influence farmers’ willingness to buy agricultural insurance.

Subjective norms, as a perceived manifestation of social pressure, are categorized into descriptive norms and injunctive norms [38]. Descriptive norms essentially realize the social learning process through peer effects, and in rural areas, farmers have frequent daily exchanges with each other and influence each other’s production and business activities. When some farmers take the lead in purchasing agricultural insurance and reduce their losses and protect their incomes through insurance claims after natural disasters or market fluctuations, such actual cases will spread rapidly among neighbors. By observing and imitating the behavior of these “first movers”, other farmers can effectively reduce the potential risks arising from information asymmetry [39]. In this process, public intervention can also incentivize farmers to share their experiences and expand the demonstration effect through policy guidance and financial support.

Injunctive norms, on the other hand, create normative beliefs that “one should be insured” through authoritative guidance such as government policies and propaganda by village cadres [40], and Takahashi et al. [41] pointed out that public sector promotional effectiveness stems in part from trust in the authority of the system. Government public interventions in the insurance sector take various forms, including the introduction of premium subsidy policies that directly reduce the cost of insurance participation for farmers and enhance their purchasing power. When farmers perceive group practices or institutional support, their willingness to buy increases significantly.

Perceived behavior control contains both internal and external dimensions. Internal control is mainly reflected in farmers’ own cognitive ability and economic conditions. Farm households’ level of understanding of insurance products constitutes the cognitive ability underlying their decision to participate in insurance [42], and well-informed farmers are able to proactively integrate information to enhance their insurance perceptions, which in turn enhances their confidence in their ability to purchase insurance [43]. At the economic level, the ability to pay premiums creates a direct constraint. Relatively wealthy households can pay higher premiums for risk-hedging instruments [44], whereas low-income farmers in developing countries are often caught in the predicament of wanting to enroll in insurance but not being able to afford it, as premium expenses exceed their budgets [45].

External control mainly involves factors that exist objectively in the external environment and are not fully determined by farmers themselves, which are dominated by external institutions but still affect farmers’ convenience of participation and trust in insurance policies. From the perspective of transaction cost theory, the coverage of service outlets and the matching degree of insurance coverage directly determine the ease or difficulty for farmers to purchase insurance. When the cost of information search and transportation time increases, farmers may give up purchasing due to the inconvenience they feel. On the contrary, when the external environment is optimized, farmers’ willingness to purchase will then increase. In summary, this paper proposes integrative research hypotheses:

H1. 

Attitude, subjective norms and perceived behavioral control all positively affect farmers’ agricultural insurance purchase intention.

3.2.2. PMT and Agricultural Insurance Purchase Intention

Based on the PMT and related research, we believe that farmers’ threat assessment and coping assessment of agricultural risks jointly play a role in insurance purchase intention. Specifically, in the threat assessment, when farmers perceive more strongly the severity of risks such as yield fluctuations or price declines [25,46], or recognize more clearly the sensitivity of their own farming systems to climatic shocks [47], their willingness to participate in insurance will increase significantly. This threat assessment mechanism is often reinforced by personal risk experience. Farmers who have experienced climate-related natural disasters and extreme weather events not only recognize the severity of the threat more clearly, but also the limitations of their own adaptive capacity [21], which can lead them to proactively seek out new knowledge and technologies to help them overcome these limitations.

At the level of coping assessment, reaction efficiency refers to the degree of farmers’ trust in the risk mitigation effectiveness of the insurance program, mainly in terms of the adequacy of the amount of claims and the timeliness of payment. The positive effect of reaction efficiency is accentuated when insurance claims cover losses in full [48] and the claims payment process is efficient [47]. The reward appraisal essentially reflects farmers’ cognitive evaluation of the comprehensive value of insurance. The psychological security provided by insurance [49] and the long-term value of risk-sharing [50] can enhance farmers’ likelihood of participation through internal and external reward mechanisms.

However, response costs, especially premium burden, can have a significant inhibitory effect on farmers’ willingness to purchase agricultural insurance, and its influence mechanism is consistent with the theory of price elasticity. Premium expenditure, as a direct economic cost, will have a budget constraint effect on farmers’ decision to participate in insurance, especially for low-income farmers with high income elasticity, the premium burden may exceed their willingness to pay risk threshold. In addition, analyzing from the perspective of transaction cost theory, response costs not only include explicit premium expenses, but also implicit costs such as time and energy paid by farmers in the process of purchase and claims, and these implicit costs will also weaken the willingness of farmers to purchase. In developing countries, the inhibitory effect of response costs is reinforced by the generally lower financial literacy of farm households and their insufficient cognition of the risk-hedging value of insurance products, which makes them more likely to regard premium expenses as a purely economic burden rather than a necessary cost of risky asset allocation. Based on the above analysis, this study proposes the hypothesis:

H2. 

Perceived severity, perceived vulnerability, reaction efficiency, and reward appraisal all positively affect farmers’ agricultural insurance purchase intention, while response cost negatively affects purchase intention.

3.2.3. Farmers’ Agricultural Insurance Purchase Intention and Purchase Behavior

According to the TPB [40], when the actual control conditions are sufficiently satisfied, all the factors affecting behavior act indirectly on the behavior itself by influencing behavioral intention, from which this paper proposes the following hypothesis:

H3. 

Farmers’ agricultural insurance purchase intention significantly and positively influences agricultural insurance purchase behavior.

3.2.4. Mediating Role of Attitude and Perceived Behavioral Control

Attitude plays a crucial role in the link between perceived severity and willingness to purchase agricultural insurance. In high-risk environments, individuals usually act aggressively to transfer risks in order to alleviate inner stress [51]. This high level of risk perception prompts farmers to more rationally assess the instrumental value of agricultural insurance, and based on the cost–benefit trade-off, farmers will realize that insurance claims can effectively cover part of the economic losses in case of disasters [11], thus maintaining the continuity of agricultural operations. This positive attitude towards insurance as a necessary risk-hedging tool translates into a stronger willingness to purchase. Empirical studies support this logic, showing that farmers who have experienced disasters such as floods, having personally witnessed the severity of risk consequences, tend to proactively seek information about agricultural insurance and explore ways to enhance their risk resilience through insurance [21].

The association between response costs and farmers’ intention to purchase agricultural insurance depends on the moderating effect of the core mediating variable, perceived behavior control. Farmers gauge their behavior control by making a holistic judgment on the likelihood of engaging in agricultural insurance, taking into account their resources and how well they gather information. Among the many elements of response costs, premium payment is the most direct and crucial influencing factor. According to the resource base theory, farmers’ wealth and asset status directly determine their ability to cope with risks and their strategy choices [52]. When the premium of agricultural insurance exceeds farmers’ economic capacity, their assessment of the feasibility of participating in agricultural insurance will decrease, leading to a decline in the level of perceived behavior control. Since perceived behavior control has a significant impact on farmers’ purchase intention, the reduction in perceived behavior control caused by excessively high premiums will inevitably weaken farmers’ purchase intention. Therefore, we propose the following hypotheses:

H4. 

Perceived severity can positively affect farmers’ agricultural insurance purchase intention through attitude; while response cost can negatively affect farmers’ agricultural insurance purchase intention through perceived behavioral control.

3.2.5. The Moderating Role of Institutional Trust in Public Intervention Situations

Institutional trust, as a key link between public intervention policies and farmers’ decision-making, not only includes government credibility based on legal and political factors, but also covers collective trust related to collective behavior. Among these, the premium subsidy policy implemented by the government is one of the core elements in constructing institutional trust. Sherrick et al. [53] point out that subsidy policy, as a core tool of public intervention, is an incentive provided by the government in the form of cash, grants, or tax relief to farmers, aimed at improving the supply of certain goods and services. In some areas of China, farmers only need to pay 10% of the premium after receiving government subsidies. Good insurance payouts can significantly enhance farmers’ trust in insurance programs and stimulate their interest in continuous purchase [54]. Meanwhile, government subsidy policies not only lower the threshold for farmers to participate in insurance through economic incentives but also serve as a signal of institutional commitment to alleviate farmers’ concerns about the performance ability of insurance companies. Research in Nigeria showed that 32% of farmers refused to participate in insurance due to concerns about contract fulfillment, while China’s agricultural insurance practice indicates that when government subsidies are combined with strict regulatory mechanisms, farmers’ psychological resistance is significantly reduced [21].

In the process of transforming farmers’ intention to purchase agricultural insurance into actual purchasing behavior, institutional trust also plays an indispensable regulatory role. Existing research has confirmed that the motivation for farmers to continuously purchase insurance not only depends on their initial willingness to purchase, but also relies on their actual experience with insurance payouts and their level of trust in the institutional system [36]. Lessons from India showed that even with a premium subsidy rate as high as 90%, if farmers lack trust in the policy implementation, the actual participation rate is still difficult to increase effectively [55]. China’s agricultural insurance sector has mirrored this trend. Based on the above analysis, this paper proposes the following hypotheses:

H5. 

Under the background of public intervention, institutional trust can not only positively regulate the relationship between psychological factors and agricultural insurance purchase intention, but also play an important role in the positive relationship between farmers’ agricultural insurance purchase intention and purchase behavior.

Based on the above analysis, we constructed a hypothesized model to examine farmers’ agricultural insurance purchase intention and actual purchase behavior. The specific model diagram is shown in Figure 1.

4. Research Design

4.1. Research Area

Jiangsu Province, located in the eastern coastal area of China (longitude 116°18′ E–121°57′ E, latitude 30°45′ N–35°20′ N), covers an area of 107,200 square kilometers, with 16.9% of its territory being water. The terrain is mainly alluvial plains, featuring fertile soil and a dense network of rivers, providing a unique natural endowment for agricultural production. As one of the 13 major grain-producing provinces in China, Jiangsu holds strategic importance in the nation’s agricultural landscape. In 2024, the province’s total grain output reached 37.46 million tons, accounting for 5.2% of the national total, which has been stable at more than 37 million tons for eight consecutive years, making it the vital “rice bowl” of the Yangtze River Delta region.

However, the compounding effects of climate sensitivity and ecological vulnerability have rendered Jiangsu persistently exposed to high natural disaster risks. According to the National Comprehensive Risk Census Bulletin, the entire territory of Jiangsu Province has been classified as a high comprehensive risk area, with its risk level being particularly prominent among the disaster-prone zones such as the North China Plain and the southeast coast. Statistics over the past 20 years show that the average annual affected area in Jiangsu amounts to 11.404 million hectares, with a disaster occurrence rate exceeding 50%. The types of disasters are diversified. Influenced by the East Asian monsoon, the annual average precipitation in Jiangsu Province is 1000–1200 mm, and 60% of it is concentrated in the flood season. Coupled with its low-lying terrain and dense water system, it is highly vulnerable to disasters such as floods, typhoons, and prolonged continuous rainfall.

As one of China’s first pilot provinces for agricultural insurance innovation, Jiangsu has actively promoted agricultural insurance in recent years. Nevertheless, it continues to face challenges including low farmer participation rates and insufficient insurance density and depth. These circumstances make Jiangsu an eminently justified and urgent case study region (Figure 2).

4.2. Questionnaire Design

To ensure the reasonableness and applicability of the questionnaire, we first conducted a pre-survey before the formal research. According to the problems found in the pre-survey stage, we made targeted adjustments as follows: (1) There is a certain degree of repetition between the questions “I think agricultural insurance is an essential tool for stabilizing farm income” and “I think buying agricultural insurance can effectively reduce the income loss caused by disasters” in the section of attitude. The former was deleted from the formal questionnaire to avoid redundancy and ensure the validity of the scale. (2) Some variables are measured by only two questions, which does not meet the requirement of structural equation modeling that latent variables need at least three observables. The observational indicators for each variable were expanded to at least three in the formal questionnaire. (3) Farmers’ purchasing behavior of agricultural insurance has the situation of dropping out of the policy after taking out the policy, which may be related to “experienced regret”. We planned to include this variable in the pre-survey, but in the actual implementation process, we found that it was difficult to accurately cover the group of farmers who dropped out of insurance through random sampling, and the response rate of the sensitive question of the reasons for dropping out of insurance was low. Therefore, we decided not to include this variable in the formal questionnaire.

The modified formal questionnaire consists of three parts. The first part covered the basic information of the respondents and their agricultural operation conditions, including age, gender, education level, etc. The second section measured agricultural insurance purchase intention, behavior, and their determinants through scales developed based on two theoretical dimensions: TPB and PMT, and the third part is open-ended questions.

Among them, attitude mainly measures farmers’ perceptions of the value of purchasing agricultural insurance. Subjective norms mainly assess the influence of social networks (e.g., relatives, friends, and village cadres) on decision-making. Perceived behavior control examines both farmers’ understanding of insurance knowledge (including policy terms comprehension) and their perception of insurance product coverage adequacy and claims process simplicity. The threat assessment section focuses on risk perception, with perceived vulnerability and perceived severity focusing on quantifying farmers’ perceptions of the probability and severity of natural and market risks. Coping assessment measured reaction efficiency and response costs through two aspects: insurance benefit cognition and economic burden perception. In addition, regarding the measurement of the reward appraisal, we associate it with the factors such as willingness to expand production. Regarding institutional trust, this paper does not differentiate between specific trust objects but treats them as the government and insurance companies as a whole. This design is based on the policy characteristics of agricultural insurance in China, where the government and insurance companies have deep synergies in premium subsidies, loss adjustments and claims, which makes farmers’ trust evaluation more inclined to the overall system.

The questionnaire design(

Table 1. Design of the questionnaire.

Dimension		Variable Settings	Subjects	Literature Sources
TPB		Attitude	AT01-AT03	Ajzen [40]; Mishra & Goodwin [8]
		Subjective norms	SN01-SN04	Wu et al. [22]
		Perceived behavior control	PBC01-PBC06	Enjolras & Sentis [47]
P M T	Threat assessment	Perceived vulnerability	PV01-PV03	Peng et al. [25]
		Perceived severity	PS01-PS03	Rogers [56]; Wang et al. [46]
	Coping assessment	Reaction efficiency	RE01-RE03	Karlan et al. [57]
		Response cots	RC01-RC03	Binswanger-Mkhize [45]; Dragos et al. [27]
		Reward Appraisal	RA01-RA03	Carter et al. [49]
Institutional context		Institutional trust	IT01-IT03	Wang et al. [36]

) is based on Ajzen’s [40] TPB and Rogers’ [56] PMT. All items are measured on a 5-point Likert scale, where 1 indicates “strongly disagree”, 2 indicates “somewhat disagree”, 3 indicates “neutral”, 4 indicates “somewhat agree”, and 5 indicates “strongly agree”.

4.3. Research Implementation and Sample Characterization

4.3.1. Sampling Method

Survey respondents are selected by a stratified random sampling method. Firstly, the 13 cities in Jiangsu Province were divided into Central Jiangsu, Northern Jiangsu and Southern Jiangsu according to the characteristics of economic gradient to eliminate the interference of regional economic development level. Secondly, two prefectures and municipalities were sampled in each stratum to control the sample size within an operable range, and the specific cities were sampled, and then four districts or counties were sampled according to the value of agricultural output in the selected prefectures and municipalities. Finally, a certain number of farm households were sampled in the administrative village of each district or county according to the corresponding proportion. A total of 630 questionnaires were distributed and 608 valid questionnaires were collected, with an effective rate of 96.51%.

4.3.2. Ethics Statement

This study was approved by the Medical Ethics Committee of Jiangsu University (Approval No. JSDX20250427001). All participants provided written informed consent prior to enrollment. The research adopted an anonymous survey approach: neither the questionnaire nor the data collection process records personal identification information such as participants’ names or contact details, ensuring full protection of respondents’ privacy. Additionally, the study did not involve minors (individuals under 18 years old) or other vulnerable populations. All samples were adult farmers with full capacity for civil conduct, and participation followed the principle of voluntariness, allowing them to withdraw from the study at any time without affecting their rights and interests.

4.3.3. Sample Analysis

Considering that if the crop combinations operated by the farmers in the sample are highly correlated in terms of climate or price risk, it may have an impact on the robustness of the study’s conclusions. For this reason, this study ensured the validity of the analysis in the following ways: first, as shown by the statistics on the types of business activities of farmers in

Table 2. Sample agricultural operations.

Options	Response		Case Percentage
	Frequency	Percent
Planting industry (e.g., grain, vegetables, fruits, etc.)	247	20.3%	40.6%
Animal husbandry (e.g., cattle, sheep, pigs, chickens, etc.)	292	24.1%	48.0%
Aquaculture (e.g., fish farming, shrimp farming, etc.)	219	18.0%	36.0%
Forestry (e.g., tree planting, bamboo cultivation, etc.)	251	20.7%	41.3%
Leisure agriculture and rural tourism (e.g., farmhouse, picking gardens, etc.)	205	16.9%	33.7%
Total	1214	100.00%	199.7%

, the interviewed farmers generally adopt a composite business model, and this diversified business structure can reduce the concentration of a single risk factor to a certain extent. Second, there are significant differences in the main types of disasters faced by the dominant crops in the three major agricultural regions of Jiangsu Province (Southern Jiangsu, Central Jiangsu, and Northern Jiangsu), e.g., rice in Southern Jiangsu is mainly threatened by floods, while wheat in Northern Jiangsu is susceptible to droughts, and this spatial heterogeneity further reduces the correlation of crop risk. Therefore, the crop risk correlations of the samples in this study are within manageable limits and do not systematically affect the core findings.

This study used stratified random sampling to obtain a sample of farm households in Jiangsu Province and analyzed their demographic characteristics and agricultural operations through descriptive statistics. The specific results are shown in

Table 3. Statistical characteristics of the sample.

Characteristics	Options	Frequency	Percent (%)
Gender	Male	319	52.5
	Female	289	47.5
Age	20–35	42	6.9
	36–45	101	16.6
	46–55	224	36.8
	56–65	178	29.3
	66 or above	63	10.4
Education	Uneducated	49	8.1
	Secondary schools	159	26.2
	Junior high schools	266	43.8
	High school or junior college	108	17.8
	College or higher	24	3.9
	Undergraduate or above	2	0.3
Health status	very poor	15	2.5
	rather poor	43	7.1
	general	261	42.9
	better	172	28.3
	rare	117	19.2
Share of annual income from family agriculture	20% and below (very low)	13	2.1
	20–40% (less)	38	6.3
	40–60% (average)	134	22
	60–80% (more)	230	37.8
	80% and above (very high)	193	31.7
Number of persons working in agriculture in households	2 or less	15	2.5
	3–5	425	69.9
	5–10	128	21.1
	10 or above	40	6.6
Length of time in agriculture	Less than 5 years	22	3.6
	5–10 years	66	10.9
	10–15 years	55	9
	15–20 years	176	28.9
	20 years and above	289	47.5
Agriculture business area	Less than 10 acres	324	53.3
	10–50 acres	241	39.6
	50–200 acres	35	5.8
	200–500 acres	5	0.8
	500 acres and above	3	0.5

. In terms of gender, the distribution of the sample is relatively balanced. Among all respondents, the group aged 46–55 accounts for the highest proportion, reaching 36.8%, followed by those aged 56–65, reflecting to some extent the demographic reality of rural aging. Regarding educational level, junior high school education is dominant, with 34.3% having only primary school education or below, and only 0.3% having a bachelor’s degree or above. In terms of health status, 42.9% of the farmers assessed their health as average, 28.3% considered they were in good health, and 19.2% said they were in very good health. In terms of economic characteristics, and 69.5% of the households have agricultural income accounting for more than 60% of their total income, indicating that for a considerable number of interviews, agriculture still plays an important role in their family economy. Agricultural operation characteristics show that 69.9% of farming families comprised 3–5 members. Experience levels were polarized, with 47.5% possessing over 20 years’ farming experience versus merely 3.6% with less than 5 years. For the majority of farmers, the scale of their land operation is less than 50 mu (≈3.33 hectares).

4.4. Model Construction

Structural equation modeling is a methodology combining structural equation modeling and path analysis. It offers an observable and manageable manifest variable for latent variables that are difficult to observe directly.

SEM comprises two components: a measurement model and a structural one, where the former is expressed as

$$X=\Lambda x\xi +\delta$$

(1)

$$Y=\Lambda y\eta +\epsilon$$

(2)

where $\mathsf{\xi }$ is the exogenous variable, $\mathsf{\eta }$ is the endogenous latent variable; X and Y are the indicator observed variables of the exogenous latent variable and the endogenous latent variables, respectively. $\mathsf{\Lambda }\mathrm{x}$ is the factor loading of the indicator variable X, and $\mathsf{\Lambda }\mathrm{y}$ is the factor loading of the indicator variable Y. $\mathsf{\delta }$ and $\mathsf{\epsilon }$ are the measurement errors of the observed variables, also called residuals.

$$\eta =\beta \eta +\gamma \xi +\zeta$$

(3)

Equation (3) represents the expression of the structural model. Here, $\mathsf{\beta }$ is the matrix of coefficients that shows the effects among endogenous latent variables, $\mathsf{\gamma }$ is the matrix of coefficients that indicates the impact of exogenous latent variables on endogenous latent variables, and ε represents the residual term.

5. Empirical Results Analysis

5.1. Reliability and Validity Tests

Reliability and validity are research techniques used to assess the accuracy of measurement scales. We collected data based on a 5-point Likert scale and conducted reliability and validity analyses using SPSS 26.0 and AMOS 24.0 software respectively, to ensure the accuracy of latent variable measurement, so as to provide a reliable foundation for the subsequent construction of the structural equation modeling.

Reliability test (Reliability) reflects the internal consistency and stability of measurement tools, commonly evaluated using Cronbach’s α coefficient and Composite Reliability (CR). According to Nunnally [58], Cronbach’s α ≥ 0.7 indicates good reliability of the scale, and a CR of ≥0.7 suggests a high degree of consistency among the observed variables of the latent variable. As shown in

Table 4. Results of reliability and validity tests.

Constrcut	Indicator	Number of Items	Cronbach’s Alpha	Standardized Loading	AVE	CR
Attitude	AT01	3	0.797	0.727	0.569	0.798
	AT02			0.765
	AT03			0.770
Subject norms	SN01	4	0.844	0.780	0.575	0.844
	SN02			0.722
	SN03			0.773
	SN04			0.756
Perceived behavior control	PBC01	6	0.889	0.730	0.572	0.889
	PBC02			0.757
	PBC03			0.789
	PBC04			0.762
	PBC05			0.778
	PBC06			0.720
Perceived vulnerability	PV01	3	0.826	0.787	0.614	0.826
	PV02			0.739
	PV03			0.822
Perceived severity	PS01	3	0.812	0.757	0.574	0.802
	PS02			0.744
	PS03			0.772
Reaction efficiency	RE01	3	0.816	0.785	0.597	0.816
	RE02			0.774
	RE03			0.759
Response costs	RC01	3	0.825	0.791	0.600	0.818
	RC02			0.763
	RC03			0.770
Reward appraisal	RA01	3	0.802	0.756	0.575	0.802
	RA02			0.742
	RA03			0.776
Purchase intention	PI01	3	0.838	0.785	0.615	0.827
	PI02			0.780
	PI03			0.787
Purchase behavior	PB01	3	0.802	0.787	0.572	0.800
	PB02			0.757
	PB03			0.723

, the Cronbach’s α values of all latent variables in this study are higher than 0.7, and the CR values all exceed 0.7, indicating that the internal consistency of the measurement model is good and the reliability of the data meets the requirements for analysis.

Validity refers to the degree to which a scale accurately measures its intended target. It mainly includes content validity, criterion validity and construct validity. Among them, convergent validity is one of the important dimensions of construct validity. This study examines construct validity through confirmatory factor analysis (CFA), focusing particularly on convergent validity assessment using two key indicators: average variance extracted (AVE) and composite reliability (CR). Generally speaking, if the AVE value of each factor is greater than 0.5 and the CR value is greater than 0.7, it indicates that the model has good convergent validity. As shown in Table 4, all the latent variables meet the above requirements, and the model has good convergent validity.

Additionally, the AVE value can also be used for discriminant validity tests. In this study, the Fornell-Larcker criterion proposed by Fornell and Larcker [59] was adopted for assessment. When the square root of a variable’s AVE exceeds the absolute values of its correlation coefficients with all other variables, that variable demonstrates adequate discriminant validity. The results of discriminant validity are shown in

Table 5. Results of discriminant validity test.

	RA	RC	RE	PS	PV	SN	PBC	AT	PI	PB
RA	0.734
RC	−0.495	0.748
RE	0.453	−0.503	0.732
PS	0.533	−0.587	0.515	0.748
PV	0.460	−0.457	0.422	0.488	0.733
SN	0.480	−0.523	0.505	0.483	0.476	0.687
PBC	0.227	−0.458	0.231	0.269	0.209	0.240	0.660
AT	0.291	−0.320	0.281	0.545	0.266	0.263	0.147	0.718
PI	0.506	−0.563	0.511	0.580	0.478	0.515	0.344	0.406	0.760
PB	0.224	−0.29	0.227	0.259	0.211	0.230	0.318	0.173	0.423	0.722

. All the correlation coefficients of the latent variables are less than the square root of the AVE, indicating that the variables have good discriminant validity.

5.2. Overall Model Fit Test

We used AMOS 24.0 software to fit the structural equation modeling. As presented in

Table 6. Results of fit test for structural equation model.

	CMIN/DF	GFI	AGFI	NFI	IFI	TLI	CFI	RMSEA
Model results	1.867	0.916	0.900	0.908	0.955	0.949	0.955	0.038
Standard	1 < CMIN < 5	>0.9	>0.9	>0.9	>0.9	>0.9	>0.9	<0.08

, the model fit indices and their corresponding evaluation criteria are reported. The CMIN/DF index of the model is 1.867, GFI is 0.916, AGFI is 0.900, NFI is 0.908, IFI is 0.955, TLI is 0.949, CFI is 0.955, and RMSEA is 0.038. All indices meet established evaluation standards, indicating an ideal level of model fit.

5.3. Model Results Analysis

The structural equation model constructed in the paper reflects the path relationships among the variables. The specific model results are shown in

Table 7. Results of model path analysis.

Hypothesis	Path	Std. Coefficient	UnStd. Coefficient	S.E.	C.R.	p
H1	AT→PI	0.128	0.137	0.054	2.541	0.011
	SN→PI	0.133	0.144	0.058	2.48	0.013
	PBC→PI	0.109	0.127	0.051	2.51	0.012
H2	PV→PI	0.147	0.151	0.072	2.103	0.035
	PS→PI	0.11	0.116	0.053	2.186	0.029
	RE→PI	0.137	0.137	0.053	2.571	0.01
	RC→PI	−0.138	−0.133	0.061	−2.177	0.03
	RA→PI	0.122	0.134	0.059	2.25	0.024
H3	PI→PB	0.355	0.302	0.045	6.737	<0.01
H4	PS→AT	0.545	0.524	0.051	10.343	<0.01
	RC→PBC	−0.458	−0.38	0.041	−9.318	<0.01

. As shown, the hypothesized paths are significant at either the 5% or 0.1% significance level, and all the path directions are consistent with the theoretical expectations. Attitude, subjective norms, and perceived behavior control are the core of TPB all of which have significant positive effects on purchase intention, verifying hypotheses H1.

It is worth noting that among the three core variables subjective norms have the greatest effect on farmers’ willingness to purchase agricultural insurance. This result is similar to the findings of studies conducted in agriculture by Wu et al. [22] and Bhandari et al. [60]. The externality embedded in the peer effect has the ability to motivate individuals to make decisions. Positive actions by individuals who take the lead in decision-making can provide benefits to other members of the group that are difficult to measure in terms of market value, which in turn influences subsequent decisions. In the field of agricultural insurance, subjective norms can enhance farmers’ own willingness to purchase through information exchange and experience demonstration.

Among the variables related to PMT, perceived vulnerability and perceived severity have significantly positive impacts on the purchase intention, and hypotheses H2 are thus confirmed. Their standardized coefficients are 0.147 and 0.110, respectively. As part of the coping assessment, response efficacy and return factors similarly positively influence purchase intention, while response costs show a significant negative impact, i.e., the higher the response costs, the weaker the purchase intention of agricultural insurance among farmers. Furthermore, the analysis reveals that perceived severity strongly positively affects attitude, while response costs diminish perceived behavior control. Among all the variables, perceived vulnerability can have the most prominent effect on farmers’ agricultural insurance purchase intention.

After further comparing the strengths of the impacts of the two key variables, subjective norms and perceived vulnerability, we found that subjective norms drive willingness to buy slightly less than risk vulnerability, but the difference between the two is small. This result implies that in China’s high-risk agricultural environment, farmers’ direct perception of their own vulnerability may trigger conservation motives more than social pressure. From the perspective of risk preference theory, Pennings et al. [61] categorized farmers as risk averse, risk neutral, and risk loving based on their perceptions of the likelihood of adverse human or environmental impacts. According to utility function theory, farmers tend to be risk averse and are more willing to pay for agricultural insurance when aversion is high [53]. Specifically, Menapace et al. [62] showed that for every 1 percentage point increase in the constant relative risk aversion rate, the probability of a farmer purchasing crop insurance increases by 3%. Together, these studies confirm that when farmers clearly recognize the production vulnerabilities they face, their risk aversion traits motivate them to seek risk transfer instruments more actively.

The specific model result graph is shown in Figure 3 below.

Regarding the agricultural insurance purchase willingness and behavior of farmers, the results in Table 7 show the agricultural insurance purchase willingness of farmers can positively influence the purchase behavior, which verifies the hypothesis H3.

However, in the actual research process, we noticed that there is a special group of people who show the behavioral characteristics of “high willingness to participate in insurance but not actually purchasing”. This phenomenon is in line with the view that there is a significant “gap” between willingness and behavior, i.e., high willingness does not necessarily lead to behavior, and the process of willingness-to-behavior conversion is affected by a number of factors in reality.

In order to explore the factors that hinder the conversion of intention to behavior, we analyzed this group of people through an open-ended questionnaire. The specific steps were to average the scores collected in the “willingness to buy” and “buying behavior” sections and define those with a willingness to buy score of more than 3 and a buying behavior score of less than 3 as the “high willingness to participate in the insurance system, but did not participate in the purchase” group. Subsequently, their responses to the question “What advice do you have for agricultural insurance?” were analyzed using SPSS multiple-response method.

The results are shown in

Table 8. Results of open-ended question.

Options	Response		Case Percentage
	Frequency	Percent
Increase claim payment ratio	30	9.20%	27.80%
Expand insurance coverage scope, e.g., disaster types, crops/livestock varieties	54	16.50%	50.00%
Reduce premiums or increase government subsidies	44	13.50%	40.70%
Simplify claims procedures and accelerate settlement speed	34	10.40%	31.50%
Enhance transparency and readability of policy clauses	55	16.80%	50.90%
Expand publicity channels to improve farmers’ insurance literacy	47	14.40%	43.50%
Offer more customized insurance products, e.g., for different operation scales/types	22	6.70%	20.40%
Improve insurers’ service quality, e.g., response speed, communication efficiency	41	12.50%	38.00%
Total	327	100.00%	302.80%

. It can be seen from this table that for this group of people, for whom there is a certain “will-behavior” gap, they are more eager to optimize agricultural insurance in terms of enhancing the transparency and comprehensibility of the insurance terms, expanding the insurance coverage, and strengthening the publicity channels. From this, we can simply speculate that the factors hindering the transformation of farmers’ willingness to buy into behavior are: first, the existence of professional expression barriers in the insurance terms and conditions, and the fact that farmers are limited by their education level and have access to a single channel of information, which leads to an inadequate understanding of the core content of the terms and conditions; and second, the mismatch between the coverage of the existing types of insurance and the actual needs of the farmers. Against the background of the booming development of specialty agriculture in Jiangsu Province, traditional food crop insurance is difficult to meet the risk protection needs of aquaculture, facility vegetables and other specialty industries, and the lag in the supply of insurance products further exacerbates the “will-behavior” gap.

In addition, by analyzing the results of the structural equation model, we also found that perceived severity has a strong positive effect on behavioral attitudes, while response cost weakens perceived behavioral control. However, only through the solution of structural equation modeling, we cannot test whether the mediating effect exists, so we need to do the mediating effect test separately. In this paper, we use the SPSS macro program PROCESS tool developed by Hayes to conduct 5000 repetitions of the samples under the 95% confidence interval condition. If the confidence interval excludes 0, it indicates a meaningful mediation effect. See

Table 9. Results of mediation effect tests.

Path		Effect	BootSE	Bias-Corrected 95% CI
				Lower	Upper
PS→AT→PI	Aggregate effect	0.463	0.037	0.39	0.536
	Direct effect	0.352	0.038	0.277	0.428
	Intermediary effect	0.11	0.018	0.077	0.148
RC→PBC→PI	Aggregate effect	−0.447	0.037	−0.519	−0.375
	Direct effect	−0.349	0.037	−0.421	−0.276
	Intermediary effect	−0.098	0.016	−0.131	−0.068

for the results. The confidence intervals of the two paths, “PS→AT→PI” and “RC→PBC→PI”, do not include 0, respectively, indicating that attitude and perceived behavior control play significant roles in their respective mediating paths, supporting hypotheses H4. This finding is consistent with the study of tea plantation farmers in Guizhou Province, which also confirmed that risk perception positively significantly affects farmers’ perceived value of insurance [63].

5.4. Analysis of the Moderating Effect of Institutional Trust

To examine the mechanism of institutional trust’s influence on the relationship between psychological factors and farmers’ agricultural insurance purchase intention, we interacted institutional trust with attitude, perceived behavior control, perceived vulnerability, perceived severity, and reaction efficiency, respectively. The moderating effect of institutional trust was examined using the PROCESS plugin in SPSS software. The results are shown in

Table 10. Results of moderating effect tests.

Model1		Model2		Model3
Path	Coefficient	Path	Coefficient	Path	Coefficient
PI→PB	0.236 ***	AT→PI	0.374 ***	PBC→PI	0.369 ***
IT→PB	0.205 ***	IT→PI	0.241 ***	IT→PI	0.242 ***
PI × IT→PB	0.108 ***	AT×IT→PI	0.090 *	PBC × IT→PI	0.125 ***
Model4		Model5		Model6
Path	Coefficient	Path	Coefficient	Path	Coefficient
PS→PI	0.385 ***	PV→PI	0.244 ***	RE→PI	0.350 ***
IT→PI	0.202 ***	IT→PI	0.239 ***	IT→PI	0.246 ***
PS × IT→PI	0.058	PV × IT→PI	0.059	RE × IT→PI	0.106 ***

Note: * p < 0.10, *** p < 0.01, i.e., significant at the 1%, 5%, and 10% levels, respectively.

. Analyses reveal that institutional trust, as a moderating variable, has significant impact on specific pathways. In Model 2, the interaction between institutional trust and attitude hits significance at the 5% level. With a standardized coefficient of 0.090, it looks like institutional trust is giving a positive nudge to the way attitude influences purchase intention. Meanwhile, the standardized coefficients of the interaction terms between perceived behavior control, reaction efficiency and institutional trust are 0.125 and 0.106, respectively, and they are significant at the 0.1% level, suggesting that institutional trust positively enhances the effects of perceived behavior control and reaction efficiency on farmers’ purchase intention of agricultural insurance. Thus, hypothesis H5 is supported. However, the interaction terms involving perceived vulnerability and perceived severity do not pass the significance test, indicating that institutional trust does not have a significant moderating effect on the influence of perceived vulnerability and perceived severity on purchase intention.

Notably, the research revealed the substantial influence of institutional trust as a key moderator in the transition from purchase intention to actual purchase ac behavior under the framework of public intervention. According to model 1, the interaction term between purchase intention and institutional trust has a coefficient of 0.108 on purchase behavior, which is significant at the 0.1% level. This indicates that when farmers perceive that public interventions such as insurance subsidies and government regulation are continuous and reliable, institutional trust not only affects the psychological process of farmers’ willingness to purchase, but also promotes the transformation of willingness to actual behavior, and this double moderating effect highlights the key role of institutional trust in farmers’ agricultural insurance decision-making in the context of public intervention.

Based on the path analysis of the moderating effect, we drew the moderating effect diagram of institutional trust. Institutional trust is divided into two levels: high institutional trust and low institutional trust. As shown in Figure 4, under a higher level of institutional trust, the slope of the regression line between purchase intention and purchase behavior increases, indicating that the positive effect of purchase intention on purchase behavior is enhanced. Similarly, in the paths of attitude, perceived behavior control, and reaction efficiency, the incline of the regression line grows, meaning that the positive influence of attitude, perceived behavior control, and reaction efficiency on purchase intention is also strengthened under the moderating effect of institutional trust. In contrast, for perceived vulnerability and perceived severity, the slopes of the regression lines with purchase intention are not significantly different whether in a high or low institutional trust context, further confirming that institutional trust has no significant moderating effect on the influence of these two variables on purchase intention.

6. Discussion

6.1. Multiple-Dimensional Factors Drive Farmers’ Intention to Purchase Agricultural Insurance

Utilizing an interwoven model that incorporates TPB and PMT, we employ a structural equation modeling to reveal the multiple influencing factors of farmers’ willingness to invest in agricultural insurance. The research findings indicate that farmers’ decision regarding insurance participation is a multidimensional cognitive evaluation process, comprehensively influenced by psychological cognition, social environment, and institutional factors.

From the TPB perspective, attitude, as farmers’ value judgment on agricultural insurance, significantly and positively affects their purchase intention, which is consistent with the research findings of Santana et al. [64] on the role of attitude in risk decision-making. They pointed out that individuals’ positive evaluation of protective measures significantly increases the likelihood of adoption. Perceived behavior control reflects farmers’ assessment of their own ability to participate in insurance and can be analyzed in depth from both internal and external control aspects [65]. In terms of internal control, farmers’ understanding of insurance products and terms is crucial. Many farmers may have insufficient understanding of core contents such as insurance terms and claim conditions due to limited insurance literacy [66], and such cognitive limitations can significantly reduce their confidence in participating. Lan et al. [13] analyzed farm households in the Mekong Delta of Vietnam using logit regression and similarly found that information such as insurance terms and conditions is critical to farmers’ decision-making.

Economic affordability is also a key factor; a high proportion of premium expenditure to household income may deter farmers from purchasing [45]. Regarding external control, related services of insurance products are equally important, including whether the insurance coverage is comprehensive, whether the purchase channels are convenient, and whether the claim process is simple. When the claim process is overly complex or uncertain, even farmers with purchase intentions may eventually give up.

Analyzing from the perspective of PMT, farmers’ purchase intention is positively impacted by risk perception, which is consistent with the research conclusion of Dragos et al. [27] that risk perception drives insurance demand. In the context of agricultural insurance, reaction efficiency and reward appraisal reflect farmers’ judgment on the effectiveness of insurance measures, and such judgment often requires the support of actual claims cases. Our findings indicate that farmers are significantly more likely to buy agricultural insurance when they see it as both efficient and rewarding. This lines up with Karlan et al.’s [57] argument that, in developing countries, farmers’ insurance decisions hinge heavily on their real-world experiences with claims. Successful claim experiences not only directly enhance the renewal intention but also strengthen the trust of surrounding farmers in the insurance system through social learning effects. In contrast, response costs exhibit strong deterrent effects. When the premium rises to a relatively high level, the participation rate of farmers in insurance may drop sharply [57].

In fact, the decision-making process of farmers is essentially an adaptive choice under the constraint of bounded rationality. Due to information asymmetry, cognitive limitations, farmers’ risk coping behaviors are influenced by both individual cognitive factors and embedded in a specific cultural and institutional context. Given this complexity in decision-making, multiple factors need to be considered when designing policies.

6.2. The Driving Role of Subjective Norms from the Perspective of the TPB

The structural equation modeling outcomes reveal that subjective norms wield a stronger influence on farmers’ likelihood of buying agricultural insurance, outpacing both attitude and perceived behavior control. This observation aligns with Wu et al.’s [22] study, which notes that a 10% rise in neighbors enrolling in agricultural insurance boosts the farmers’ enrollment probability by 3.25%. This multiplier effect profoundly reflects the unique decision-making mechanism of Chinese rural society.

From a sociological perspective, this demonstration effect can be traced back to the theory of the differential patterns proposed by Fei Xiaotong. In the social network structure of rural China, individual economic decisions are not made independently but are embedded in a complex web of relationships bound by blood ties and geographical proximity. When making decisions, farmers tend to rely on peers’ practical experiences rather than independently assessing the risks and benefits of insurance products [67]. There are two reasons behind this decision-making pattern. On the one hand, it stems from the herd effect [39], in the face of uncertainty, individuals tend to imitate the behavior of the group as a way to avoid potential risks. On the other hand, the rural insurance market is inherently plagued by information asymmetry [68], and rural areas have relatively limited channels for information acquisition. Moreover, agricultural insurance is inherently complex and specialized, making it difficult for farmers to independently conduct a comprehensive assessment of agricultural insurance products. It is worth noting that this phenomenon is not unique to China, as a study of farmers in Northeast Ethiopia also found that social ties play a key role in shaping farmers’ attitudes toward agricultural risk management [50].

Meanwhile, apart from the horizontal demonstration effect among neighbors, the vertical guidance from authorities also plays a crucial role. A rural collective is a community of shared interests formed by a group of people, with village cadres serving as the “bellwether” within this community [69]. When deciding whether to purchase agricultural insurance, farmers subconsciously refer to the decisions made by village cadres. This also indirectly leads to village cadres having to consider the implicit pressure from ordinary farmers when deciding whether to purchase agricultural insurance for themselves. According to Boyd et al. [70], village cadres are relatively more likely to purchase agricultural insurance. In actual situations, when village cadres actively purchase agricultural insurance and publicize its benefits, farmers are more likely to try to purchase it.

6.3. The Driving Role of Risk Perception from the Perspective of PMT

For farmers, deciding whether to obtain agricultural insurance often hinges on how they personally size up the risks involved. Research generally suggests that the more farmers worry about potential problems, the more inclined they are to buy insurance. Our findings definitely back this up. Looking at it through the lens of PMT, it is clear that perceived vulnerability and the perceived severity of those vulnerabilities really push farmers toward purchasing insurance. When farmers foresee risks like natural calamities and market turbulence, they see agricultural insurance as a failsafe to safeguard against these uncertainties, thereby boosting their interest in obtaining coverage [33]. This push from risk awareness underscores the human instinct to shun losses.

Prospect theory points out that people’s aversion to potential losses is much greater than their preference for equivalent gains [71]. Therefore, strengthening farmers’ perception of agricultural risks can effectively activate their protection motivation. In terms of specific manifestations, the perception of natural risks is the most direct and intense. With the intensification of climate change, the frequent occurrence of extreme weather events has significantly enhanced farmers’ risk awareness [72]. Next is the perception of market risks. Although farmers are highly sensitive to price fluctuations of cash crops, due to the limitation of financial literacy, they often find it difficult to accurately evaluate the severity of market risks. The third is risk experience. Farmers who have been burned before by crop damage or losses tend to have better grasps of the inherent risks in farming. As a result, they are more likely to hedge their bets and protect their livelihood by buying insurance [73]. Osberghaus [74] found through analysis that households that have experienced disasters in the past are more inclined to take protective measures, and this memory of disasters can have a lasting impact.

This phenomenon varies across countries. A study of Romanian farmers found that experience of disaster losses significantly affects the decision to participate in insurance, and the higher the level of risk aversion, the stronger the willingness to participate in insurance [75]; and a study of Bangladeshi farmers’ willingness to pay for flood insurance came to a similar conclusion [76]. However, in contrast, the risk-averse attitude of Cambodian farmers has a weaker impact on the decision to enroll in insurance, and their decision is more dependent on the level of trust in the weather station [77].

6.4. The Mediating Role of Variables Under the TPB-PMT Integrated Framework

We found that perceived severity does indeed have a positive influence on farmers’ intention to purchase agricultural insurance through attitude, which verified hypothesis H4. According to PMT [56], when farmers perceive that agricultural risks may cause severe economic losses, it triggers an internal risk avoidance mechanism, thereby forming a positive perception of the instrumental value of agricultural insurance. This perception is transformed into an attitude that “purchasing insurance is a necessary choice”, ultimately enhancing their willingness to purchase. This finding is consistent with the conclusion of Liu et al. [30].

Regarding how response costs affect farmer interest in buying crop coverage, findings are consistent with the expectation, which states that response costs have an impact through perceived behavior control. Specifically, response cost elements such as premium payments have a significant negative impact on perceived behavior control. This conclusion is in line with the expectations of the resource-based theory [52]. When premium payments exceed a certain proportion of a household disposable income [57], farmers’ assessment of the feasibility of participating in insurance significantly decreases, thereby weakening their purchase intention. This phenomenon is particularly evident among low-income groups [45]. This finding further supports the social cognitive theory [78], indicating that individuals’ expectations about the outcomes of a new behavior and their self-assurance in executing it contribute to shaping behavioral intentions.

6.5. The Dual Moderating Effects of Institutional Trust in the Context of Public Intervention

It was found that institutional trust not only significantly moderates the influence path of psychological variables on purchase intention but also effectively promotes the translation of purchase intention into actual purchase behavior. Specifically, institutional trust really throws a positive wrench into the works when it comes to how attitude, perceived behavior control, and reaction efficiency affect things. However, it does not seem to budge the needle on threat assessment one way or the other. The underlying reason might be that high institutional trust can enhance farmers’ confidence in the effectiveness of policy implementation, amplifying their perception of behavior control while raising their expectations of insurance benefits [79]. Threat assessment, however, is more dependent on individuals’ personal experiences or direct observations of natural disasters. Institutional trust is difficult to alter the intensity of their risk perception through policy commitments. In other words, policy trust mainly affects the coping assessment rather than threat assessment, as the latter is rooted in objective risk exposure rather than subjective institutional evaluation.

Although this study confirms that farmers’ willingness to buy significantly and positively affects their purchasing behavior. However, a large number of previous studies have shown that there is a general barrier to the conversion of willingness to behavior in the field of protective behavior, i.e., there is a “willingness-behavior” gap [80]. This paper analyzes the open-ended question and finds that the blocking factors are the gap in farmers’ awareness and the lack of product suitability, which is consistent with the findings of the study by Kramer et al. [81] which pointed out that the base risk (i.e., the mismatch between the insurance payout rate and the actual loss rate) is one of the key factors leading to the low rate of participation in agricultural insurance in developing countries.

However, at the same time, this paper also found that institutional trust can play an important moderating role in the process of transforming purchase intention to purchase behavior. This result can be explained by the following reasons. At the cognitive level, high institutional trust can enhance farmers’ confidence in the effectiveness of policy implementation, which echoes the preference for timely claims settlement found in the study of Indian farmers [9]. At the behavioral level, insurance institutions with high trust usually simplify the enrollment process and provide relevant facilitation services, measures that can effectively reduce objective barriers to behavioral implementation [82]. At the economic level, government subsidy policies in public interventions not only directly reduce farmers’ participation costs but also send policy credible signals.

Nevertheless, we must recognize the complexity of public intervention. On the one hand, subsidy policies do help expand the demand for insurance and correct market failures [83]; but on the other hand, subsidies may have policy exit difficulties, especially in the face of global climate change, and simple insurance subsidies may have limited effects. This dilemma highlights the importance of establishing a sustainable institutional trust system, which requires the government to play the key role of in market coordination and policy formulation [84], while guarding against the possible negative effects of excessive intervention.

7. Conclusions and Policy Recommendations

7.1. Conclusions

Based on the TPB-PMT integrated model, we took a good hard look at what drives farmers’ intentions and actual decisions to buy agricultural insurance in Jiangsu Province, a high comprehensive risk area in China. The research finds that farmers’ decision to participate in insurance is a complex process that integrates psychological cognition, social norms, and institutional environment.

From the perspective of TPB, attitude, subjective norms and perceived behavior control all positively affect purchase intention. Among them, the driving role of subjective norms is particularly prominent, reflecting the key role of social networks in the rural society’s differential pattern in China. When making decisions, farmers highly rely on the insurance participation experience of their neighbors and the authoritative influence of village cadres, showing typical characteristics of “conformity” and “authority compliance”.

When it comes to PMT, perceiving the threat posed by risk serves as the catalyst to ignite motivation. Generally speaking, the higher the level of perceived severity and perceived vulnerability, the more likely farmers are to participate in agricultural insurance. The assessment of coping enhances farmers’ trust in the effectiveness of insurance, thereby increasing their willingness to purchase. Response cost markedly deters purchasing intention, indicating that farmers rationally weigh the cost inputs and benefits during the decision-making process.

Further comparing the impacts of two key variables, i.e., subjective norms and perceived vulnerability on strong riskiness, it can be seen that perceived vulnerability has a slightly higher direct driver effect on willingness to buy than subjective norms, a result that reveals that in China’s high-risk agricultural environments, farmers’ direct perception of their own vulnerability triggers protection motives more than social pressures.

In addition to the above, the study also found that perceived severity can shape positive attitudes by reinforcing the cognition of risk consequences, while response cost weakens perceived behavior control through economic rationality considerations. Although, in the process of will-to-behavior transformation, there are problems such as insufficient cognition and poor product suitability, institutional trust can effectively alleviate the problem and promote the transformation of farmers’ purchase willingness to purchase behavior. At the same time, it can also enhance the promotion effect of behavioral attitude, perceived behavioral control and response efficacy on purchase intention.

7.2. Policy Recommendations

Based on the empirical findings of this study, in order to effectively increase the participation rate of agricultural insurance among Chinese farmers, we will put forward policy recommendations from the following aspects.

First, at the level of risk perception, a systematic and regularized risk education mechanism should be established. Specifically, the agricultural department can take the lead and collaborate with meteorological and emergency management departments to compile risk prevention manuals that are suitable for the cognitive characteristics of local farmers. The content can cover the identification and prevention of common natural disasters, and specific cases of disaster impacts, in order to activate the “threat assessment” cognition of farmers. At the same time, regular risk knowledge lectures and case sharing sessions can be organized to enhance farmers’ awareness of risk, thereby activating their motivation to protect.

Secondly, in terms of the influence of subjective norms, it is necessary to fully leverage the demonstration effect and authoritative guidance role of the rural acquaintance society. On the one hand, a group of “insurance demonstration households” with exemplary effects can be cultivated, and their personal experiences can be used to influence the surrounding farmers. On the other hand, we can give full play to the normative influence of the authoritative subject by incorporating the participation rate into the performance appraisal of the village cadres and requiring them to take the lead to participate in the insurance and publicize the information of the insurance policy. The authoritative subject’s normative influence is fully utilized.

Thirdly, when optimizing insurance products and services, insurance companies need to focus on strengthening farmers’ awareness of the consequences of risk in order to shape positive attitudes and mitigate the weakening of perceived behavioral control by response costs through economically rational design. Specifically, a risk visualization publicity system can be constructed to show the difference in losses between uninsured and insured farmers through case presentations, thus promoting the formation of a positive attitude that insurance can effectively reduce losses. At the same time, to address the impact of reaction costs, insurance companies can introduce stepped premiums and provide discounts for farmers with small planting areas to reduce their payment pressure.

At the same time, considering the cognitive limitations and the lack of product suitability on the transformation of purchasing intention to behavior, we believe that we should establish a regular publicity mechanism in insurance publicity, and help farmers accurately understand the insurance terms through multiple channels, so as to effectively improve their cognitive level of insurance products. Regarding product design, we should consider the diversified needs of farmers, and enhance the match between agricultural products and insurance, so as to effectively solve the blockage of the transformation of willingness to behavior.

Lastly, in terms of institutional trust building, a sound insurance service supervision mechanism should be established, a feedback channel for farmers should be set up, and data on insurance participation and claim settlement should be publicized on a regular basis. The government’s trust should be built up through the fulfillment of claims promises, which will further enhance farmers’ trust and motivation to participate in the program, thus truly benefiting farmers.

7.3. Limitations and Future Research

7.3.1. Limitations

This study has certain limitations that can be further discussed and improved in future research. First, although stratified random sampling was used to ensure the representativeness of the sample, most of the research was conducted with the assistance of village cadres, and farmers may have tended to provide expected “standard answers” rather than true thoughts out of compliance with authority. Second, farmers with different insurance participation experiences may have differences in risk perception, attitudes, and subjective norms. However, we do not classify the research subjects in detail and our study lacks an in-depth discussion of factors such as farmers’ historical experiences and experienced regret in participation in agricultural insurance. I

7.3.2. Future Research

In the future, we will focus on three directions of expansion. First, we will use anonymous interviews or surveys assisted by third-party organizations to improve the authenticity of the data. Second, in terms of the content of the study, we will strengthen the tracking of the history of farmers’ participation in insurance, focusing on the impact of different participation experiences on the subsequent decision-making of farmers. Finally, in terms of theoretical framework, we try to introduce the “two-system” decision-making theory, the instinctive fear of systematic insurance fraud and the role of herd mentality in the decision-making of insurance participation, so as to further explain the phenomenon of “willingness-behavior” gap.