Research on Farmers’ Agricultural Disaster Insurance Purchase Decisions and Policy Implications Under Land Trusteeship

Abstract

Land trusteeship is an innovative agricultural management model that connects smallholder farmers with modern agriculture. It promotes large-scale agricultural operations, but still faces the impacts of conventional natural disasters. Although agricultural disaster insurance serves as a critical mechanism for farmers to mitigate these natural risks, its risk-mitigation potential remains underutilized due to the persistent challenge of low insurance participation rates. This study develops a decision-making model for farmers’ purchase of agricultural disaster insurance under land trusteeship, drawing on protection motivation theory, market failure theory, and quasi-public goods theory. Using structural equation modeling, we empirically analyze survey data from 319 land-trusteed farmers to uncover the mechanisms and pathways influencing their insurance purchase decisions. The results indicate that: (1) Vulnerability and severity are positively associated with protection motivation through perceived response efficacy and self-efficacy, and protection motivation is directly associated with purchase decisions; (2) Government support has both direct and indirect effects on purchase behavior; and (3) Individual and household characteristics are significantly associated with purchase decisions, with pure farmers, Type I part-time farmers, and farmers with larger landholdings tending to purchase agricultural disaster insurance more often.

1. Introduction

Agriculture is a fundamental pillar of human survival and development, and its modernization is essential for global food security and economic stability. Worldwide, smallholder farmers constitute the backbone of agricultural production. Although small farms account for 84% of all farms globally, they cultivate only about 12% of agricultural land but produce 35% of the world’s food, underscoring their indispensable role in the global food system [1]. However, fragmented operations, limited capacity for resource integration, and weak resilience to risk present common challenges for the modernization of smallholder-based agricultural systems [2,3]. In China, the world’s largest developing country, the smallholder economy remains particularly prominent. According to the Third National Agricultural Census, smallholders account for more than 98% of agricultural operators, highlighting their continued dominance in agricultural production [4]. In response to the dual challenges of smallholder transformation and agricultural modernization, China has promoted the land trusteeship model. By providing agricultural services, this model enables scaled operation and offers an institutional solution to the constraints faced by smallholders. However, while land trusteeship can generate economies of scale by consolidating fragmented plots, the expansion of operational scale and rising input costs may also increase exposure to natural-disaster risks. Indeed, economic losses caused by natural disasters have become a major obstacle to China’s agricultural modernization [5]. In response to the growing frequency of natural disasters, countries worldwide commonly rely on insurance as a mechanism for loss prevention and compensation [6,7]. Despite differences in policy priorities across countries, low insurance participation remains a global challenge [8,9]. Prior research suggests that improving participation requires not only external incentives but also the enhancement of farmers’ genuine demand for insurance, while respecting their preferences [9,10]. Against this backdrop, this study takes China’s land trusteeship model as its entry point to systematically examine the mechanisms influencing farmers’ decisions to purchase agricultural disaster insurance. The findings are intended to inform risk-protection strategies in the transformation of smallholder agriculture in China and provide insights for agricultural modernization in Asia, Africa, and Latin America.

Existing studies generally classify the determinants of farmers’ insurance purchase decisions into four dimensions: individual and household characteristics, risk perception, insurance awareness, and government subsidies. Some studies also consider the role of insurance product quality [11]. With respect to individual and household characteristics, education level, exposure to insurance promotion, and farm size are positively associated with purchase decisions, whereas age is negatively associated with them [12,13]. Farm type and labor force size also matter [14]. Regarding risk perception, farmers’ recognition of disaster risks, assessment of severity, and production objectives significantly shape their willingness to purchase insurance [8,15]. In addition, individual risk preferences, such as risk aversion, further influence these decisions [16]. With regard to insurance awareness, research shows that farmers’ understanding of insurance principles, policy terms, and benefits not only directly promotes purchasing behavior [17,18] but also improves the rationality of insurance decisions [19]. Beyond these internal factors, the external policy environment and product design are also critical. Government premium subsidies significantly encourage participation, although the marginal effect tends to diminish [20,21]. Moreover, as subsidy levels rise and agriculture becomes more specialized, the incentive effect weakens [22,23]. In addition, the quality and attractiveness of insurance products significantly affect farmers’ willingness to purchase. Insufficient incentives, mismatched products, and operational deficiencies on the part of insurers may negatively affect decisions [9].

Existing research has mostly investigated farmers’ agricultural disaster insurance purchase decisions from isolated dimensions, mainly including individual and household attributes, risk perception, insurance awareness, and government subsidies. Few studies have adopted social cognitive theory to unpack the mediating mechanisms and causal pathways among these factors. Addressing this research gap, this paper first systematically clarifies the intrinsic linkage between land entrustment and farmers’ agricultural insurance decisions. On this basis, it employs the Protection Motivation Theory to empirically explore the determinants and formation mechanisms of farmers’ insurance purchasing behavior under land entrustment arrangements. The main contributions of this study are twofold. First, it overcomes the limitations of single-dimensional research by integrating the above dimensions into a unified analytical framework, enabling a holistic synthesis of the factors shaping farmers’ insurance decisions. Second, combining Protection Motivation Theory with structural equation modeling, this study quantitatively analyzes multidimensional influencing factors. It not only reveals the internal mechanisms of farmers’ insurance purchase decisions but also provides an intuitive quantitative evaluation of the relative impact of each dimension.

The rest of the article is structured as follows. Section 2 presents the theoretical model and research hypotheses. Section 3 describes the research methods, including model construction, sample characteristics, and scale design. Section 4 reports the empirical findings, followed by Section 5, which concludes with a discussion of the results and policy implications.

2. Theoretical Model and Research Hypotheses

2.1. Theoretical Model

2.1.1. Protection Motivation Theory

Protection Motivation Theory (PMT), originally proposed by Rogers (1975) and grounded in expectancy theory, posits that individuals’ protective motivation is shaped by their perceptions of threat severity, vulnerability, and the efficacy of the recommended response. Together, these factors constitute the basis of fear appeals, which encourage adaptive responses. Rogers subsequently incorporated self-efficacy as a fourth core component [24]. Supporting this, studies by Bandura [25] and Condiotte [26] demonstrated that behavioral change is positively correlated with shifts in self-efficacy expectations. Thus, the revised PMT framework includes four key elements: susceptibility, severity, self-efficacy, and response efficacy. While some scholars have proposed integrating “maladaptive rewards” and “response costs”, this study excludes these constructs. The act of purchasing agricultural disaster insurance represents a protective behavior, and given the conceptual overlap between “response costs” and “self-efficacy”, their simultaneous inclusion is deemed redundant [27]. As illustrated in Figure 1, the PMT conceptualizes the adoption of protective behaviors as a cognitive process involving two appraisal stages: threat appraisal and coping appraisal. The combination of these appraisals generates protection motivation, which in turn guides decision-making [28].

2.1.2. The Theoretical Model of Agricultural Disaster Insurance Purchase Decision-Making Among Farm Households Under Land Trusteeship

Research on agricultural disaster insurance, both domestic and international, consistently highlights the significant influence of farmers’ cognitive levels on their purchase decisions. Protection Motivation Theory (PMT), a scientific framework that links individual cognition to behavior, provides a theoretical foundation for this phenomenon. Its four core elements of susceptibility, severity, response efficacy, and self-efficacy elucidate the internal logic through which farmers form protective motivation and make insurance decisions. Furthermore, the strategic importance of agriculture and the quasi-public good nature of agricultural insurance necessitate the inclusion of government support in analyzing farmers’ insurance uptake under land trusteeship models. From a market failure perspective, commercial agricultural insurance markets are often characterized by high costs and risks, which can lead insurers to raise premiums or withdraw from the market, thereby limiting farmers’ access to suitable products [29]. Government intervention, through policy support and subsidies, helps correct this imbalance and ensures the availability of agricultural insurance. Quasi-public goods theory further reinforces the need for government involvement. Agricultural insurance is neither a pure public good nor a fully market-driven commodity. It exhibits strong positive externalities—stabilizing farmer incomes and enhancing national food security [30]. The spillover effect, wherein individual insurance purchases generate broader social benefits, means that the social value of agricultural insurance exceeds its private value, underscoring the limitations of market mechanisms in providing adequate coverage. In this study, susceptibility and severity are treated as variables under risk perception, while response efficacy and self-efficacy belong to the dimension of insurance cognition. Farmers’ decisions are thus jointly shaped by their risk perception and insurance-related awareness. Building on this framework, we construct a theoretical model of agricultural disaster insurance purchase decisions among land-entrusted farmers (see Figure 2), using PMT as the foundation to reveal the mechanisms and pathways underlying their decision-making.

2.2. Research Hypotheses

2.2.1. The Impact of Farmers’ Risk Perception and Insurance Awareness on Agricultural Disaster Insurance Purchase Decisions Among Land Entrusted Farmers

The PMT provides a useful framework for understanding how cognitive processes influence behavior, making it highly applicable to the study of farmers’ decisions to purchase agricultural disaster insurance. The theory comprises two key dimensions—risk perception and insurance cognition—each containing two core elements: vulnerability and severity under risk perception, and response efficacy and self-efficacy under insurance cognition. From the perspective of risk perception, both vulnerability and severity contribute to the threat appraisal mechanism in farmers’ decision-making under the land trusteeship model. Vulnerability refers to a farmer’s perceived likelihood of agricultural disasters. When farmers recognize this vulnerability, they are more inclined to adopt protective behaviors. This perception is shaped by factors such as regional disaster frequency and personal experience with past disasters. In high-frequency disaster areas, farmers generally exhibit stronger crisis awareness and higher perceived vulnerability. Severity, on the other hand, reflects a farmer’s subjective assessment of the adverse consequences posed by agricultural disasters. Those who perceive greater threat severity are more motivated to take protective actions, thereby facilitating insurance purchase decisions. From the perspective of insurance cognition, farmers’ response efficacy and self-efficacy constitute the core components of the coping evaluation mechanism. In the context of agricultural disaster insurance purchases among land-entrusted farmers, response efficacy refers to farmers’ perception of how effective such insurance is in mitigating risks. The stronger this perceived effectiveness, the greater their response efficacy—ultimately fostering protection motivation and encouraging purchase decisions. Self-efficacy, on the other hand, reflects farmers’ confidence in their own ability to purchase agricultural disaster insurance. Rooted in Bandura’s self-efficacy theory, this concept underscores that psychological and behavioral changes are largely mediated by shifts in individuals’ beliefs in their capabilities [31]. Those with high self-efficacy are more likely to overcome obstacles, whereas those with low self-efficacy may become discouraged by similar challenges.

Based on the above analysis, the following hypotheses are proposed:

H1a. 

Perceived susceptibility is positively associated with protective motivation, thereby relating positively to farmers’ insurance purchase decisions.

H2a. 

Perceived Severity is positively associated with protective motivation, thereby relating positively to farmers’ insurance purchase decisions.

H3. 

Response efficacy is positively associated with protective motivation, thereby relating positively to farmers’ insurance purchase decisions.

H4. 

Self-efficacy is positively associated with protective motivation, thereby relating positively to farmers’ insurance purchase decisions.

In applications of PMT, some scholars conceptualize threat and coping assessments as independent processes. However, a growing body of research suggests that these two dimensions are, in fact, interrelated. Studies have shown that threat assessment exerts a significant positive influence on coping assessment [32,33]. This interdependence should be carefully considered in theoretical and empirical work. Further extending this line of inquiry to the agricultural context, farmers’ risk perceptions have been found to shape their intention to purchase agricultural insurance by influencing their understanding of such financial instruments [34]. As farmers enhance their knowledge of insurance products and evaluate their own risk-management capacities, they may opt to purchase insurance when they determine that self-reliance alone is insufficient to mitigate potential threats.

Based on the above analysis, this paper argues that the relationship between threat assessment and response assessment should be considered. Integrating the research context of agricultural insurance purchase decisions among land-entrusted farmers, the following hypotheses are proposed:

H1b. 

Farmers’ perceived susceptibility is positively associated with response efficacy, thereby linking to protective motivation and further to farmers’ insurance purchase decisions.

H1c. 

Farmers’ perceived susceptibility is positively associated with self-efficacy, thereby linking to protective motivation and further to farmers’ insurance purchase decisions.

H2b. 

Farmers’ perceived severity is positively associated with response efficacy, thereby linking to protective motivation and further to farmers’ insurance purchase decisions.

H2c. 

Perceived severity is positively associated with self-efficacy, thereby linking to protective motivation and further to farmers’ insurance purchase decisions.

The PMT originally aimed to explain intentions driven by protective motives. As the theory’s applications broadened, research attention shifted from examining intentions to understanding decision-making processes. A meta-analysis by [35] demonstrated that key components of PMT—threat severity, threat susceptibility, response efficacy, and self-efficacy—all contribute to strengthening protective motivation. Ref. [36] evaluated how these components influence decision-making, concluding that PMT effectively predicts behavioral decisions. Their findings also highlighted that the coping assessment component has a stronger predictive power than threat assessment. Based on this evidence, we propose the following hypothesis:

H5. 

Farmers’ protective motivation for agricultural disaster insurance is positively associated with their purchase decisions.

2.2.2. The Impact of Government Support on Land-Trusteeship Farmers’ Agricultural Disaster Insurance Purchase Decisions

In studies on farmers’ purchasing behavior for agricultural disaster insurance, the government is widely recognized as a key influencing factor [37]. This influence manifests through policy support, subsidy provisions, and public trust in government institutions. Policy support establishes the foundational framework for the development of agricultural insurance. Effective government-led promotion and guidance enhance farmers’ awareness of risk management and help broaden insurance coverage. The rise of land trusteeship services further facilitates the dissemination of agricultural insurance policies, improving farmers’ access to and comprehension of relevant information. Such outreach efforts encourage participation by increasing farmers’ understanding of insurance benefits and enabling them to better evaluate their own insurance needs and capacities—ultimately shaping their purchasing decisions. Government subsidies also play a crucial role. Premium subsidies positively influence farmers’ insurance awareness, allowing them to obtain basic production protection at lower costs. This affordability makes agricultural disaster insurance more accessible and appealing. Moreover, the level of public trust in government significantly affects farmers’ acceptance of agricultural insurance policies. Higher governmental credibility fosters greater understanding and approval of such policies, thereby shaping farmers’ perceptions and decisions regarding insurance adoption [38]. Based on the above analysis, the following hypotheses are proposed:

H6a. 

Government support is directly and positively associated with farmers’ agricultural disaster insurance purchase decisions.

H6b. 

Government support is positively associated with farmers’ response efficacy, thereby linking to protective motivation and further to farmers’ agricultural disaster insurance purchase decisions.

H6c. 

Government support is positively associated with farmers’ self-efficacy, thereby linking to protective motivation and further to farmers’ insurance purchase decisions.

2.2.3. The Impact of Farmers’ Differences on Agricultural Disaster Insurance Purchase Decisions

Under the land trusteeship model, older farmers often demonstrate lower willingness to purchase agricultural disaster insurance compared to their younger counterparts. This reluctance can be attributed to their limited exposure to new agricultural practices, lower insurance awareness, and greater reliance on personal farming experience. Educational attainment also plays a significant role. Farmers with higher levels of education are generally more inclined to purchase disaster insurance. Moreover, under land trusteeship arrangements, many farmers engage in off-farm employment. Such employment not only raises household income but also improves farmers’ understanding of agricultural insurance, thereby positively influencing their purchase decisions. However, the effect of non-agricultural employment is nuanced. As its degree increases, the share of agricultural income in total household income tends to decline, which may in turn weaken farmers’ incentive to insure agricultural production [39]. Finally, with the ongoing scaling of agricultural management, farm size has become another critical factor. Households operating larger farms tend to place greater emphasis on land use efficiency and production returns. Given their exposure to higher agricultural risks, these farmers are more motivated to adopt insurance as a risk mitigation tool and thus exhibit a stronger willingness to purchase disaster coverage.

Based on the above analysis, this study does not propose hypotheses for individual and household characteristics. Instead, we empirically examine how age, education, farm type, and land size are related to farmers’ insurance purchase decisions in Section 4.4.

Based on the above analysis, a decision-making model for farmers’ purchase of agricultural disaster insurance under land trusteeship arrangements has been constructed, as shown in Figure 3.

3. Research Design

3.1. Research Area Selection

To ensure the reliability and representativeness of the research sample, a multistage sampling approach was adopted. The sampling process encompassed three hierarchical levels: prefecture-level cities, townships, and farming households, serving as the primary, secondary, and tertiary sampling units, respectively. First, three prefecture-level cities in Hebei Province—Cangzhou, Langfang, and Tangshan—were selected as pilot study sites due to their well-established land trusteeship programs. Next, based on data provided by local agricultural and rural affairs bureaus, we identified 1 to 3 townships with well-established land trusteeship models in each selected city. These townships were purposively selected based on the recommendations of local agricultural bureaus, prioritizing those with well-established land trusteeship programs. Finally, this study designed standardized questionnaires via the Questionnaire Star platform, which were directionally distributed by village officials and farmer cooperatives in the selected townships through online communities exclusive to land trusteeship farmers. During this process, elderly respondents who were unfamiliar with smart devices were often assisted by their younger family members in completing the questionnaire. It is worth noting that “prefecture-level city” in this context refers to a second-tier administrative unit in China’s hierarchy, encompassing not only urban districts but also subordinate county-level cities, rural counties, and extensive agricultural areas. Unlike the conventional notion of a compact urban area, a prefecture-level city spans a broad administrative region. This study specifically targeted agricultural townships within these regions where land trusteeship programs have been implemented. This sampling framework ensures the study’s representativeness and robustness, capturing the dynamics of land trusteeship models in diverse agricultural settings.

Hebei experiences a temperate, semi-humid, semi-arid continental monsoon climate. Given its geographical and topographical conditions, agricultural production in the region is consistently vulnerable to natural disasters, resulting in considerable economic losses for local farmers. Cangzhou serves as a major grain-producing area in Hebei. According to data from its Municipal Bureau of Agriculture and Rural Affairs, the city hosted 3108 agricultural production service organizations in 2023, providing land stewardship services covering a total of 22.9 million mu (approximately 1.53 million hectares). Langfang functions as a key agricultural supply base for the Beijing–Tianjin–Hebei region. Since 2018, the city has actively promoted the adoption of managed farming services. Tangshan exhibits distinct advantages in crop cultivation. Official statistics indicate that the city’s entrusted agricultural service area reached 30.2 million mu in 2023.

3.2. Model Construction

This study employs a questionnaire survey to collect data, utilizing structural equation modeling supplemented by Bootstrap mediation effect testing and LSD multiple comparison methods to investigate the factors influencing farmers’ decisions to purchase agricultural disaster insurance under land trusteeship arrangements. Structural equation modeling quantifies and analyzes latent variables—those difficult to measure directly—through predefined observable and operationalized exogenous variables, thereby testing influence relationships or associative pathways between variables [40]. The structural equation model comprises measurement equations and structural equations, with the relationships between variables expressed as follows:

Measurement Equations:

$$\mathrm{X}={\Lambda }_x\mathsf{\xi }+\mathsf{\delta }$$

(1)

$$\mathrm{Y}={\Lambda }_y\mathsf{\eta }+\mathsf{\epsilon }$$

(2)

Structural Equations:

$$\mathsf{\eta }=\mathrm{B}\mathsf{\eta }+\mathrm{Г}\mathsf{\xi }+\mathsf{\zeta }$$

(3)

In this equation, $\mathsf{\xi }$ denotes the exogenous latent variables, indicated by pv, ps, re, se, and gov. Meanwhile, $\mathsf{\eta }$ represents the endogenous latent variables, indicated by mot and be. The observed variables—such as pv1, ps1, and re1, with a total of 21 items—are denoted by $\mathrm{X}$ and $\mathrm{Y}$. Matrix ${\Lambda }_x$ and ${\Lambda }_y$ capture the relationships between the latent variables and their respective observed indicators, allowing linear combinations to be derived for each latent variable from its indicators. The path coefficient matrix Γ is used to analyze the effects of pv, ps, re, and se on mot, as well as the effects of mot and gov on be. The terms $\mathsf{\delta }$, $\mathsf{\epsilon }$, and $\mathsf{\zeta }$ represent measurement errors.

Although structural equation modeling (SEM) is a powerful analytical tool for investigating complex relationships among latent variables, it is not without limitations. In the context of this study, the model incorporates 7 latent variables and 21 indicators, analyzed using a sample size of 319. While the parameter estimates meet standard criteria, caution is necessary when interpreting their stability, given the model’s complexity and sample size. Additionally, as the data are cross-sectional, the estimated paths (e.g., RE→MOT→BE) reflect statistical associations rather than causal relationships. It is important to note that the term “effect”, as used in this paper, is intended strictly in a statistical sense: within the specified model framework, a one-unit change in the exogenous variable corresponds to a β-unit change in the endogenous variable.

3.3. Scale Design

Through organizing and reviewing relevant literature and questionnaire examples, a five-point Likert scale method was adopted to construct the specific measurement scale. The scale items are organized around seven dimensions: susceptibility, severity, response efficacy, self-efficacy, government support, protective motivation, and purchase decision. Each dimension includes three items to accurately and comprehensively reflect farmers’ attitudes (see

Table 1. Collation of Scale Information.

Research Variables	Observed Variables	Measurement Items	Reference
Perceived susceptibility	pv1	The probability that disasters will damage agricultural production	[33,41,42]
	pv2	The probability that agricultural disasters will cause property losses
	pv3	The probability that your region will be threatened by agricultural disasters
Perceived severity	ps1	The degree of threat that agricultural disasters pose to agricultural production	[41,42]
	ps2	Agricultural disasters will cause property losses.
	ps3	The degree of harm that agricultural disasters cause to the environment and ecology
Perceived response efficacy	re1	Agricultural disaster insurance can diversify risks	[42,43]
	re2	Agricultural disaster insurance can compensate for losses.
	re3	Agricultural disaster insurance can stabilize agricultural production.
Perceived self-Efficacy	se1	Whether you purchase agricultural disaster insurance depends mainly on you.	[41,44,45]
	se2	It is relatively easy for you to purchase agricultural disaster insurance.
	se3	You have the ability to purchase, manage, and file claims for agricultural disaster insurance.
Government support	gov1	The impact of policy publicity on agricultural disaster insurance purchase decisions	[10,46,47]
	gov2	The impact of government subsidies on agricultural disaster insurance purchase decisions
	gov3	The impact of trust in the government on agricultural disaster insurance purchase decisions
Protection motivation	mot1	You are willing to purchase agricultural disaster insurance	[42,48,49]
	mot2	You plan to purchase agricultural disaster insurance.
	mot3	You are willing to recommend purchasing agricultural disaster insurance to relatives and friends.
Purchase decision	be1	Have you already purchased agricultural disaster insurance	[49]
	be2	Have you already recommended purchasing agricultural disaster insurance to relatives and friends?
	be3	Will you continue to purchase agricultural disaster insurance hereafter?

).

3.4. Sample Characteristics

A total of 341 questionnaires were distributed in the formal online survey, and 319 valid responses were obtained, resulting in a valid response rate of 93.55% (as shown in

Table 2. Basic Information of Surveyed Farmers.

Category	Classification Item	Frequency	Percentage (%)
Gender	Male	164	51.41%
	Female	155	48.59%
Age	Under 30 years old	21	6.58%
	31–40 years old	56	17.55%
	41–50 years old	80	25.08%
	51–60 years old	96	30.09%
	Over 60 years old	66	20.69%
Educational level	Primary school	67	21.00%
	Junior high school	174	54.55%
	Senior high school	50	15.67%
	Above senior high school	28	8.78%
Farm household type	Pure farm households	31	9.72%
	Part-time farm households (Type I)	34	10.66%
	Part-time farm households (Type II)	121	37.93%
	Non-farm households	133	41.69%
Land scale	1–10 mu	96	30.09%
	11–20 mu	117	36.68%
	21–30 mu	65	20.38%
	Over 30 mu	41	12.85%

Note: Bold is used for readability in the diagonal table.

). Sample characteristics are summarized as follows: The gender distribution was relatively balanced, with males accounting for 51.41% and females 48.59%. In terms of age, respondents aged 51–60 constituted the largest group (30.09%), followed by those over 60 (20.69%), which may reflect the outmigration of younger laborers for off-farm employment. Education levels were predominantly low and unevenly distributed. The majority of respondents (54.55%) had attained junior high school education, followed by elementary school. Based on established classifications [50], land-entrusting farmers were categorized into four household types according to the share of agricultural income in total household income: pure farmers (>80%), Type I part-time farmers (50–80%), Type II part-time farmers (30–50%), and non-agricultural households (<30%). In this sample, Type II part-time farmers and non-agricultural households together accounted for over 70%, indicating a high degree of part-time engagement among respondents. Regarding land holdings, the largest proportion of farmers (36.68%) operated 11–20 mu, followed by 1–10 mu (30.09%), suggesting that most farmers managed relatively small plots.

Over 70% of the study sample comprises Type II part-time and non-agricultural households, reflecting the pervasive off-farm employment in rural China and the growing relevance of the land trusteeship model. For these households, land trusteeship ensures standardized land use and stable farm income amid their primary non-agricultural work, making agricultural disaster insurance for entrusted land critical to mitigating production risks and stabilizing farm earnings. Despite most income coming from non-agricultural sectors, these households remain the primary decision-makers for such insurance purchases under the land trusteeship framework.

4. Results

4.1. Validity and Reliability Testing

Model reliability and validity were assessed using SPSS 26.0 and Amos 24.0. For reliability, the overall Cronbach’s α of the scale was 0.84, with α values for all dimensions ≥ 0.6, indicating favorable internal consistency. Before exploratory factor analysis, KMO and Bartlett’s sphericity test results (see

Table 3. Table of Reliability and Validity Test.

Research Variables	Observed Variables	Cronbach’s Alpha	CR	AVE
Perceived susceptibility	pv1	0.798	0.7974	0.5675
	pv2
	pv3
Perceived severity	ps1	0.787	0.7973	0.5707
	ps2
	ps3
Perceived response efficacy	re1	0.863	0.8698	0.6901
	re2
	re3
Perceived self-Efficacy	se1	0.828	0.8264	0.6143
	se2
	se3
Government support	gov1	0.904	0.9061	0.7628
	gov2
	gov3
Protection motivation	mot1	0.862	0.8638	0.6790
	mot2
	mot3
Purchase decision	be1	0.864	0.8587	0.6711
	be2
	be3

) showed KMO values of 0.661 for independent variables and 0.719 for dependent variables, both with p-values of 0.000 (p < 0.01). This passed the significance test at the 1% level, confirming the data’s suitability for factor analysis. Subsequently, principal component analysis was employed to extract factors with eigenvalues > 1. Results showed that 6 factors were extracted for the independent variables and 1 for the dependent variable. The dimensions of each factor aligned with the predefined framework, with factor loadings all >0.5. The cumulative explained variances were 69.566% and 67.268% (both >60%), indicating good construct validity. The results of the confirmatory factor analysis indicate: Regarding convergent validity, the seven variables exhibited standard factor loadings ranging from 0.6 to 0.95, with CR and Cronbach’s alpha coefficients both exceeding 0.7 and AVE > 0.5, all falling within acceptable ranges. This confirms good convergent validity for each variable. For discriminant validity (

Table 4. Results of Discriminant Validity Test.

	Perceived Susceptibility	Perceived Severity	Perceived Response Efficacy	Perceived Self-Efficacy	Government Support	Protection Motivation	Purchase Decision
Perceived Susceptibility	0.75
Perceived Severity	0.47	0.76
Perceived Response Efficacy	0.63	0.50	0.83
Perceived Self-Efficacy	0.61	0.45	0.49	0.78
Government Support	0.23	0.19	0.38	0.41	0.87
Protection Motivation	0.58	0.46	0.61	0.62	0.31	0.82
Purchase Decision	0.47	0.38	0.52	0.53	0.38	0.81	0.82

Note: Bold and italics are applied in the diagonal table only for better readability of data.

), the square roots of variance extracted (0.753–0.819) exceeded the inter-variable correlation coefficients. For structural validity, model fit assessment should include absolute fit indices, incremental fit indices, and parsimony fit indices.

Table 5. Model Fit Test Results.

Index Name		Model Index Value	Fit Criterion	Fit Result
Absolute Fit Index	GMIN/DF	1.895	<3	Good
	RMR	0.032	<0.08	Good
	RMSEA	0.053	<0.08	Good
	GFI	0.913	>0.9	Good
Incremental Fit Index	IFI	0.960	>0.9	Good
	CFI	0.960	>0.9	Good
	TLI	0.951	>0.9	Good
Parsimonious Fit Index	PCFI	0.795	>0.5	Good
	PNFI	0.761	>0.5	Good

Note: CFI > 0.9, RMSEA < 0.08, RMR < 0.05 indicate good fit.

confirms good model fit, with all indices within acceptable ranges.

4.2. Hypotheses Testing and Model Modification

To further examine the relationships among variables, path analysis and hypothesis testing were performed on the structural equation model after satisfactory model fit was established (with *** denoting p < 0.001 and ** denoting p < 0.01). The results are summarized in

Table 6. Standardized Results of Structural Equation Model Path Analysis.

Path	Standardized Coefficient	p-Value	Significance	Hypotheses	Test Result
PV→MOT	—	>0.05	—	H1a	Not Supported
PV→RE	0.467	0.000	***	H1b	Supported
PV→SE	0.473	0.000	***	H1c	Supported
PS→MOT	—	>0.05	—	H2a	Not Supported
PS→RE	0.246	0.000	***	H2b	Supported
PS→SE	0.194	0.003	**	H2c	Supported
RE→MOT	0.401	0.000	***	H3	Supported
SE→MOT	0.431	0.000	***	H4	Supported
MOT→BE	0.767	0.000	***	H5	Supported
GOV→BE	0.134	0.004	**	H6a	Supported
GOV→RE	0.232	0.000	***	H6b	Supported
GOV→SE	0.269	0.000	***	H6c	Supported

Note: **, *** represent statistical significance levels; — indicates missing data; → denotes the direction of influence.

:

(1)

Farmers’ perceived susceptibility is positively and indirectly associated with their insurance purchase decisions, primarily through response efficacy and self-efficacy. The corresponding path coefficients are 0.467 and 0.473, both indicating substantial statistical effects. These results support Hypotheses H1b and H1c.

(2)

Perceived severity is positively and indirectly associated with insurance purchase decisions via response efficacy and self-efficacy, with path coefficients of 0.246 and 0.194, respectively. Hypotheses H2b and H2c are thus supported.

(3)

Both response efficacy and self-efficacy show strong direct effects on protection motivation, with path coefficients of 0.401 and 0.431, respectively, supporting Hypotheses H3 and H4.

(4)

Protection motivation is positively and indirectly associated with insurance purchase decisions, with a path coefficient of 0.767. This result supports Hypothesis H5 and confirms that protection motivation is the strongest predictor of farmers’ insurance purchase behavior, consistent with protection motivation theory.

(5)

Government support is positively related to insurance purchase decisions through two pathways: a direct effect (path coefficient = 0.134), and indirect effects mediated by response efficacy and self-efficacy (path coefficients = 0.232 and 0.269, respectively). These findings validate Hypotheses H6a, H6b, and H6c.

Additionally, the paths from perceived susceptibility to protection motivation (PV→MOT) and from perceived severity to protection motivation (PS→MOT) were not statistically significant. This indicates that, within the constructed model of agricultural disaster insurance purchase decisions among land-entrusted farmers, susceptibility and severity influence protection motivation mainly through efficacy perceptions rather than directly. Therefore, Hypotheses H1a and H2a are rejected. These non-significant paths were removed to form the final model, as illustrated in Figure 4.

4.3. Mediation Analysis

To further investigate the role of mediating variables in explaining the relationships among variables in the model, the Bootstrap method was used to conduct a mediation analysis. As shown in

Table 7. Bootstrap Mediation Effect Test.

Bootstrap Mediation Effect Test
Path	SE	Effect Size	Bias-Corrected 95%CI		Percentile 95%CI
			Lower	Upper	Lower	Upper
PV→RE→MOT→BE	0.039	0.144	0.076	0.232	0.071	0.222
PV→SE→MOT→BE	0.041	0.157	0.088	0.25	0.085	0.244
PS→RE→MOT→BE	0.027	0.076	0.032	0.137	0.027	0.131
PS→SE→MOT→BE	0.032	0.064	0.012	0.141	0.011	0.137
GOV→RE→MOT→BE	0.025	0.071	0.031	0.132	0.027	0.125
GOV→SE→MOT→BE	0.030	0.089	0.039	0.162	0.037	0.154

Note: → in the Path column indicates the hypothesized direct influence path between consecutive variables.

, the confidence intervals for all paths in the model do not include zero, confirming the significant mediating associations across all paths.

Table 8. Path Coefficients for Protective Motivation and Purchase Decision of Agricultural Catastrophe Insurance.

Variable Relationship	Path	Direct Path Coefficient	Indirect Path Coefficient	Total Association
PV→MOT	PV→RE→MOT	—	0.187	0.391
	PV→SE→MOT	—	0.204
PS→MOT	PS→RE→MOT	—	0.099	0.182
	PS→SE→MOT	—	0.083
RE→MOT	RE→MOT	0.401	—	0.401
SE→MOT	SE→MOT	0.431	—	0.431
GOV→MOT	GOV→RE→MOT	—	0.093	0.209
	GOV→SE→MOT	—	0.116
PV→BE	PV→RE→MOT→BE	—	0.144	0.300
	PV→SE→MOT→BE	—	0.156
PS→BE	PS→RE→MOT→BE	—	0.076	0.140
	PS→SE→MOT→BE	—	0.064
RE→BE	RE→MOT→BE	—	0.308	0.308
SE→BE	SE→MOT→BE	—	0.331	0.331
GOV→BE	GOV→BE	0.134	—	0.295
	GOV→RE→MOT→BE	—	0.072
	GOV→SE→MOT→BE	—	0.089

Note: — means no direct path coefficient; → in the Path column indicates the hypothesized direct influence path between consecutive variables.

details the effect values for each specific path, with the following conclusions:

(1)

Farmers’ perceived susceptibility to agricultural disasters is positively associated with their purchase decisions for agricultural disaster insurance, mediated by both perceived efficacy and self-efficacy. Specifically, susceptibility is positively associated with purchase decisions through perceived efficacy (path coefficient = 0.467) and self-efficacy (path coefficient = 0.473). The 95% confidence intervals for these paths are (0.076, 0.232) and (0.071, 0.222), respectively, with additional intervals ranging from (0.088, 0.25) to (0.085, 0.244). None of these intervals includes zero, indicating that both perceived efficacy and self-efficacy fully mediate the relationship between susceptibility and purchase decisions within the framework of protective motivation.

(2)

The perceived severity of agricultural disasters is positively associated with insurance purchase decisions, mediated by response efficacy and self-efficacy. The path coefficients for severity via response efficacy and self-efficacy are 0.076 and 0.064, respectively. Confidence intervals for these indirect effects are (0.032, 0.137) and (0.027, 0.131), with supplementary intervals ranging from (0.012, 0.141) to (0.011, 0.137). All intervals exclude zero, confirming that response efficacy and self-efficacy fully mediate the pathway from severity to purchase decisions through protective motivation.

(3)

Government support shows both a direct and an indirect association with farmers’ decisions to purchase agricultural disaster insurance. The direct effect is 0.134, while the total indirect effect is 0.161. Specifically, the indirect effect mediated by response efficacy is 0.071, with confidence intervals of (0.031, 0.132) and (0.027, 0.125). The indirect effect mediated by self-efficacy is 0.089, with confidence intervals of (0.039, 0.162) and (0.037, 0.154). Since none of the intervals include zero, both mediators are statistically significant, indicating that perceived efficacy and self-efficacy partially mediate the effect of government support on insurance purchase decisions.

4.4. Analysis of Farmer Variations

This study employed SPSS 26.0 to conduct an analysis of variance (ANOVA) to determine the significance of differences in farmers’ purchasing decisions for agricultural disaster insurance based on individual characteristics. Since ANOVA cannot specify the exact differences between groups, LSD post hoc tests were used for comparative analysis. In ANOVA, the F-value reflects the relative magnitude of variance between groups versus within groups, while the p-value measures the significance of differences. If the F-value is large and p < 0.05, the difference between groups is statistically significant.

(1)

Analysis of Age Differences. Analysis of variance (ANOVA) was conducted to examine age differences. Results indicate that the F-value for “farmers’ decision to purchase agricultural disaster insurance” was 0.436 (p = 0.783 > 0.05). This suggests no significant age-related differences in the decision to purchase agricultural disaster insurance among land-entrusted farmers.

(2)

Analysis of Differences in Educational Attainment. Analysis of variance (ANOVA) was conducted to examine differences in educational attainment. Results indicate that the F-value for “farmers’ decision to purchase agricultural disaster insurance” was 1.778 (p = 0.151 > 0.05). This suggests no significant difference in the decision to purchase agricultural disaster insurance among land-entrusted farmers based on educational attainment.

(3)

Analysis of Differences Among Farm Household Types. An analysis of variance (ANOVA) was performed to assess variations in agricultural disaster insurance purchase decisions across different types of farm households. The result shows a significant F-value of 10.838 (p < 0.001), indicating statistically significant differences among groups. To pinpoint where these differences lie, post hoc LSD tests were conducted. The results reveal that pure farmers (those relying predominantly on agricultural income) and Part-time Farmers I have a significantly higher demand for agricultural disaster insurance. In contrast, Part-time Farmers II show a moderate level of demand, while Non-farmers—those with negligible agricultural income—exhibit the lowest demand and purchase willingness among all groups.

(4)

Analysis of Differences in Land Scale. Analysis of variance (ANOVA) was conducted to examine differences in land scale. Results indicate that the F-value for “farmers’ agricultural disaster insurance purchase decisions” was 17.860 (p = 0.000 < 0.05), confirming significant differences in agricultural disaster insurance purchasing behavior among farmers with varying land scales. To further identify which farmer groups exhibited significant differences in insurance purchase decisions, post hoc LSD tests were performed. Results indicate that farmers with larger land holdings are more likely to purchase agricultural disaster insurance than those with smaller holdings.

5. Discussion and Conclusions

5.1. Driving Factors Associated with Farmers’ Agricultural Disaster Insurance Purchase Decisions from the Perspective of the Protection Motivation Theory

This study develops a decision-making model for farmers’ purchase of agricultural disaster insurance within land trusteeship arrangements, drawing on protection motivation theory, market failure theory, and quasi-public goods theory. Using structural equation modeling, we empirically examine the mechanisms and pathways influencing farmers’ insurance purchase decisions. The main findings are as follows:

First, farmers’ response efficacy and self-efficacy are positively associated with their protection motivation, which is subsequently linked to their insurance purchase decisions [12]. This finding aligns with Chen [9], who reported that risk perception promotes purchase intention through protection motivation in Jiangsu, China. Some farmers have recognized the role of insurance in mitigating agricultural risks and believe in their own ability to purchase coverage. This awareness of both “insurance usefulness” and “personal capacity to buy insurance” is increasingly translating into actual purchase decisions. Given that agriculture is highly vulnerable to uncontrollable factors such as extreme weather and natural disasters—and agricultural income remains a major source of livelihood for many rural households—effective risk management tools are essential. Perceived efficacy helps farmers view insurance as a viable means of offsetting disaster losses, while self-efficacy reflects their confidence in completing the purchase process. As farmers’ trust in insurance and their perceived ability to buy it grow, so does their motivation to seek protection, thereby increasing the likelihood of insurance adoption.

Second, farmers’ insurance purchase decisions are associated with their susceptibility and severity perceptions, both of which are linked to protective motivation through perceived efficacy and self-efficacy, and thereby are indirectly associated with their purchasing decisions. Susceptibility and severity are key factors associated with farmers’ agricultural disaster insurance purchase decisions. However, unlike Fahad [8], who found a direct effect of risk perception on purchase decisions in Pakistan, we found that susceptibility and severity are fully mediated by efficacy beliefs. This discrepancy likely reflects contextual differences: more frequent climate shocks in Pakistan may create a direct risk–action pathway, while China’s land trusteeship model—where service organizations assume part of daily risk management—weakens that direct link.

Third, farm household type and land scale exert significantly differentiated effects on agricultural disaster insurance purchase decisions [39]. Pure farmers and part-time farmers of Category I have a higher proportion of agricultural income in their total economic income, making them naturally more inclined to purchase agricultural disaster insurance when faced with disaster impacts. Non-farmers exhibit the lowest willingness to purchase. Large-scale farmers also demonstrate stronger purchasing intent than small-scale farmers. These factors collectively result in significant differences in insurance decision-making among farmers of varying types and scales.

5.2. Farmers’ Protection Motivation, Purchase Decisions, and Policy Support

Farmers’ protection motivation is defined as their psychological tendency to proactively adopt risk prevention measures when facing uncontrollable agricultural risks—such as extreme weather and natural disasters—in order to safeguard their core interests. The stronger this motivation, the more likely farmers are to purchase agricultural disaster insurance. Government support serves as a catalyst for such motivation. Policy backing, direct premium subsidies, and governmental credibility together form the core pillars that stimulate farmers’ engagement in protective behaviors.

From the universal patterns observed in the global promotion of agricultural insurance, despite variations in implementation approaches across countries, the core consensus lies in the following: (1) Government policy support and public trust are pivotal in overcoming farmers’ insurance-related misconceptions. By clearly communicating the risk-diversification value of insurance, helping farmers understand the underlying logic of insurance operations and its adequate compensation function, and strengthening farmers’ response efficacy to agricultural disaster risks, these factors are positively associated with their purchase decisions [10,47]. (2) Direct premium subsidies serve as the primary means to alleviate farmers’ economic burdens. By reducing financial costs through premium subsidies, farmers’ self-efficacy in participating in insurance is enhanced, thereby influencing their purchasing decisions.

In addressing natural disaster risks, China has long relied predominantly on government relief, with such assistance and disaster support accounting for 90% of disaster compensation [51]. Since 2003, China has explored establishing a policy-based agricultural insurance system. While fiscal support for agricultural disaster insurance has steadily increased, and coverage has expanded in both scope and scale, issues such as insufficient farmer awareness of insurance and weak willingness to bear premiums persist, constraining participation rates in agricultural disaster insurance.

5.3. Impact on Public Policy Makers

Agricultural disaster insurance is a critical instrument for mitigating natural risks and stabilizing agricultural production. Within large-scale land trusteeship systems, it plays an essential role in dispersing concentrated disaster losses and protecting the incomes of entrusted farmers. However, several challenges—such as low farmer awareness of insurance, limited capacity to pay premiums, and inconsistent quality of insurance products—have resulted in low participation rates. This undermines the core functions of insurance in disaster relief and risk transfer, particularly under the land trusteeship model. As this model expands the scale of land management, risks become more concentrated, making the consequences of low insurance uptake even more severe.

To enhance farmers’ participation in agricultural disaster insurance under land trusteeship and fully realize its risk-sharing and disaster compensation functions, the government must effectively leverage its macro-regulatory role while stimulating market vitality. The government should strengthen policy guidance and support by increasing publicity to break down farmers’ cognitive barriers, improving premium subsidy mechanisms to alleviate their financial burden, and reinforcing the review and supervision of insurance products to standardize market operations [52]. Concurrently, insurance institutions should gain an in-depth understanding of farmers’ actual needs while adhering to policy directives. By continuously improving insurance products and optimizing claims settlement processes, they can enhance service quality [53], effectively overcome obstacles in policy uptake and claims handling, and better fulfill the risk protection needs of farmers under trusteeship models.

Rooted in its national context as a “large country with small-scale farmers”, China has pursued a distinct path toward agricultural modernization. Through land trusteeship, it addresses the constraints of smallholder fragmentation, while agricultural disaster insurance helps mitigate risks associated with scaled production. This experience offers valuable insights for other developing countries. Many African nations, for instance, are in the early stages of agricultural modernization and face similar challenges—a high proportion of smallholders and weak risk resilience. Agricultural insurance has thus become a common strategy for protecting farmers’ livelihoods [54,55]. This study identifies key factors influencing farmers’ insurance uptake and underscores a fundamental principle: the promotion of agricultural insurance must align with national and agricultural realities. It cannot overlook the dominance of smallholders and resource constraints. Insurance systems must be tailored to specific developmental stages, leveraging governmental leadership while activating market potential.

5.4. Limitations and Future Prospects

This study extends the traditional Protection Motivation Theory—originally applied in health behavior decision-making—to the domain of agricultural risk management. This provides a new theoretical framework for analyzing farmers’ agricultural disaster insurance purchase decisions.

However, several limitations should be noted. First, the questionnaire was distributed via online communities, resulting in a non-probability convenience sample of land trusteeship farmers with WeChat group access; thus, results should be interpreted as exploratory rather than strictly population-representative. Second, as survey data were collected exclusively from Hebei Province, China, the findings—while relevant to regions with similar agricultural conditions in China—require further location-specific validation for broader generalizability. Additionally, the cross-sectional design of this study precludes definitive causal inference. Although structural equation modeling can test directional hypotheses, the associative pathways are correlational in nature. To address these limitations, future research should adopt longitudinal designs, quasi-experimental methods, or instrumental variable approaches to establish causal links with greater rigor. Another promising direction is to integrate theories such as institutional economics and contract governance to explore how different land trusteeship models shape farmers’ agricultural disaster insurance demand.