Efficiency of Agricultural Insurance in Facilitating Modern Agriculture Development: From the Perspective of Production Factor Allocation

Abstract

Agricultural insurance is instrumental in consolidating the gains of poverty alleviation and advancing rural revitalization. It significantly aids in the efficient allocation of agricultural production factors, which in turn enhances agricultural output and bolsters the evolution of modern agriculture. Therefore, utilizing data from 583 household surveys and employing endogenous transformation and intermediary effect models, this paper analyzes the production factor allocation effect and specific mechanism of agricultural insurance. It focuses on small-scale farmers and new agricultural operators, exploring how insurance contributes to the advancement of modern agricultural practices. The results show the following: (1) Agriculture insurance can significantly affect the agricultural scale input behavior of farmers such as land input scale and input scale, agricultural machinery application behavior such as the degree of mechanization and water conservancy application, agricultural technology adoption behavior, and planting structure selection behavior, thereby helping to modernize agriculture. (2) There is heterogeneity in the impact of agriculture insurance on the allocation of production factors for small farmers and new agricultural operators. For small farmers, agriculture insurance has a significant promoting effect on their agricultural machinery application behavior, agricultural technology adoption behavior, and planting structure selection behavior. For new agricultural operators, agriculture insurance significantly promotes their agricultural scale input behavior, agricultural machinery application behavior, and agricultural technology adoption behavior. (3) In terms of the mechanism of action, agriculture insurance mainly promotes agricultural scale input behavior through land transfer, facilitates agricultural machinery application behavior by purchasing agricultural machinery equipment and services, encourages agricultural technology adoption behavior by strengthening agricultural technology training, and enhances professional production levels by increasing the scale of insured planting, thereby contributing to the development of modern agriculture. Based on this, several policy suggestions have been proposed. These include enhancing the directionality of agriculture insurance policies, improving the collaborative interaction mechanism between agriculture insurance and agricultural credit financing, and adopting certain reward and punishment measures to curb moral hazard.

1. Introduction

Agriculture, being a vulnerable sector, is often exposed to the uncertainties of natural calamities and market fluctuations. These factors can significantly impede the progress of agricultural modernization and the broader initiative of rural revitalization. Currently, China has entered a new stage of consolidating poverty alleviation achievements and promoting rural revitalization. However, most farmers still have weak risk resistance capabilities [1], and are highly likely to fall into relative poverty when facing risk shocks. Among them, agricultural risk shocks account for more than 20%, and their incidence and disaster damage rates tend to be high [2,3]. These risks can severely damage agricultural production infrastructure, impede advancements in agricultural technology, and dampen the intrinsic motivation for development among farmers. Farmers are particularly susceptible to relative poverty, and addressing this issue has emerged as a critical new focus for enhancing the efficacy of poverty alleviation and management strategies. Agriculture insurance can give full play to its unique advantages in poverty prevention work. Especially for groups with low income levels and weak risk resistance capabilities, agriculture insurance can effectively provide their agricultural risk avoidance needs [4]. Since the pilot work of central government premium subsidies was launched in 2007, China’s agriculture insurance has developed rapidly and played an increasingly important role in promoting poverty alleviation and agricultural development. However, as China enters a new stage of rural revitalization, the role of agricultural insurance expands beyond merely addressing fundamental risk management and ensuring production security. It is also crucial for supporting farmers in their efforts to achieve long-term poverty alleviation and for stimulating growth in agricultural production [5]. However, under the current “wide coverage, low guarantee” and “lightweight” system in China, the overall level of agriculture insurance coverage is far lower than that in countries such as the United States and Japan. The role of agricultural insurance in risk dispersion and production assurance has not been optimally utilized, which undermines its effectiveness and limits its capacity to foster agricultural modernization. Therefore, the question of whether China’s agricultural insurance serves to enhance the optimization of production factor allocation and, in turn, supports agricultural modernization is a significant one. In addition, the mechanism through which this might occur should also be delved into.

Existing studies have explored the impact of agriculture insurance on agricultural production factors in various aspects [6,7,8,9,10,11,12,13], including agricultural production scale, agricultural technology adoption, and agricultural green production. Overall, most scholars hold a positive attitude towards the impact of agriculture insurance on agricultural production factors. They hold that agriculture insurance can effectively promote agricultural business scale, promote the adoption of new agricultural technologies, and promote green agricultural production [14]. Some scholars have explored the impact of agriculture insurance on the scale of agricultural operations, demonstrating agriculture insurance as an important factor in promoting the expansion of agricultural production and operation [15]. Some scholars have analyzed the impact of agriculture insurance on the use and adoption of agricultural technologies, pointing out that agriculture insurance can effectively promote the adoption of protective tillage techniques by farmers, especially combinations of protective tillage techniques [16,17]. Some scholars argue that agricultural insurance can impact the adoption of water-saving irrigation technology by farmers in both direct and indirect ways. Directly, it encourages farmers to adopt water-saving irrigation technologies. Indirectly, it promotes the adoption through the evaluation of agricultural risk transfer effects [18]. Some scholars have examined the effectiveness of agriculture insurance in promoting green emissions reduction in agriculture, pointing out that agriculture insurance can effectively promote carbon reduction in agriculture. However, there is a negative effect of operating scale on the carbon reduction effect of agriculture insurance, that is, the carbon reduction effect of agriculture insurance decreases with an increase in production scale [19]. Some scholars have analyzed the impact of agriculture insurance on agricultural green development. They argue that agricultural insurance can effectively promote the growth of agricultural green total factor productivity by optimizing the allocation of agricultural production factors, thereby enhancing the level of green agricultural development [20]. However, at the same time, some scholars hold a negative attitude towards the impact of agriculture insurance on agricultural production factors. They hold that agriculture insurance has a negative impact on the level of agricultural green development in major grain producing areas, and the impact of agriculture insurance compensation expenditures on the level of agricultural green development shows an “inverted U-shape” [21]. Some scholars believe that the implementation of full cost insurance and income insurance can effectively reduce agricultural carbon emissions in high natural risk areas, while increasing agricultural carbon emissions in low natural risk areas [22].

Numerous studies have explored the impact of agricultural insurance on agricultural production, but most focus on single dimensions such as land scale, fertilizer and pesticide application, or agricultural machinery and equipment, lacking a systematic approach. There is a lack of systematic research on the impact of agriculture insurance on the allocation of agricultural production factors, and the internal logic has not been fully revealed. Meanwhile, existing studies mostly focus on the poverty alleviation period, lacking analysis of the impact of agriculture insurance factor allocation from the perspective of assisting agricultural modernization in the current context of consolidating poverty alleviation achievements and promoting rural revitalization goals. The question of whether and how agriculture insurance affects production factor adjustment to assist agricultural modernization remains unresolved. In light of this, this study, set against the backdrop of reinforcing poverty alleviation and fostering rural revitalization, takes into account the evolving landscape of China’s agriculture and agricultural insurance sectors. It encompasses a range of agricultural management entities of varying sizes and conducts a comprehensive theoretical analysis and empirical examination. The focus is on assessing the impact of agricultural insurance on the optimization of production factor allocation, which is instrumental in driving the process of agricultural modernization from the perspective of resource distribution. Meanwhile, from the perspective of assisting agricultural modernization, this study focuses on four aspects, including agricultural scale input behavior, agricultural machinery application behavior, agricultural technology adoption behavior, and planting structure selection behavior. It examines the specific mechanisms by which agriculture insurance affects effective agricultural production decisions to facilitate agricultural modernization. Overall, the efficacy and operational mechanisms of agricultural insurance are also assessed, offering guidance for enhancing its quality and efficiency, and for stimulating agricultural productivity.

Compared to existing studies, the possible marginal contributions of this study are as follows: Firstly, in terms of theoretical mechanisms, based on the perspective of production behavior, this study comprehensively considers four aspects, including agricultural scale input behavior, agricultural machinery application behavior, agricultural technology adoption behavior, and planting structure selection behavior. It systematically analyzes the specific mechanisms of how agriculture insurance affects the allocation of production factors and aids in agricultural modernization, revealing the internal logic and framework system of how agriculture insurance contributes to development. Secondly, from a research perspective, the consolidation of poverty alleviation achievements and the promotion of rural revitalization are taken into full consideration, and the factor allocation effect of agriculture insurance on agricultural production behavior is examined from the perspective of promoting agricultural modernization. Thirdly, recognizing the diversity among research subjects, such as small farmers and new agricultural entities, the study accounts for their distinct demands as key participants in agricultural insurance. It investigates how different types of agricultural insurance cater to this heterogeneity and contribute to the development of modern agriculture. The findings aim to inform the refinement of agricultural insurance policies and practices.

2. Theoretical Analysis and Research Hypotheses

The stronger the market-oriented dispersion and transfer function of agriculture insurance on agricultural risks, the more helpful it is in promoting farmers to adjust the allocation of agricultural production factors and enhancing their agricultural production and management capabilities. A virtuous cycle is thus formed to boost agricultural production and assist in agricultural modernization (Figure 1).

2.1. Agriculture Insurance and Agricultural Scale Input Behavior

In terms of land input scale, agricultural risks have a strong spatiotemporal correlation. When natural risks occur, agricultural production in a certain area will suffer disaster damages, and market risks such as price fluctuations can also have an impact on regional and even global agricultural production. Therefore, farmers tend to make relatively conservative agricultural production decisions [23]. If they are unwilling to expand planting areas, they may even choose to “abandon land” or even reduce agricultural production investment [24]. Agriculture insurance effectively achieves market-oriented diversification and transfer of agricultural risks, which is conducive to reducing the risk of large-scale production for farmers. This encourages farmers to engage in moderate-scale management by expanding land input or transferring land, and even reusing abandoned land and inefficient land [15,25,26]. Scaling up the land input scale is beneficial for achieving economies of scale, improving agricultural production efficiency, and thus supporting the development of modern agriculture.

In terms of capital input scale, financial institutions have a low enthusiasm for carrying out agricultural loan business due to the insufficient stability of agricultural returns and lack of collateral for farmers. The long-standing problem of “financing difficulties” for farmers is even more prominent [27], so they cannot provide sufficient capital input for agricultural production. Agriculture insurance mitigates the risks inherent in farming operations and addresses the practical challenges that farmers encounter, particularly those related to the absence of adequate collateral [28]. Meanwhile, it can transfer credit risks caused by agricultural system risks from the financial market to the insurance market, which is conducive to ensuring the sustainability of banks and other credit institutions, ensuring the sustainable reproduction capacity of farmers [29]. In this way, the production capital difficulties of farmers are alleviated through the synergistic effect of agriculture insurance credit financing, and the enthusiasm of farmers to increase agricultural capital input will be increased [30]. Hence, increasing capital input is conducive to promoting the accumulation of agricultural material capital, thereby facilitating the development of modern agriculture.

2.2. Agriculture Insurance and Agricultural Machinery Application Behavior

Farmers are cautious in their investment decisions to upgrade agricultural machinery and purchase agricultural machinery services due to the weak nature of agriculture. Even with the support of relevant government policies and funds, farmers do not have a high willingness to invest. Meanwhile, the lack of stability in agricultural returns and the lack of collateral among farmers result in agricultural credit constraints, making it difficult for farmers to obtain sufficient financial support to purchase and upgrade agricultural machinery and equipment. In this context, the risk protection function of agriculture insurance enhances the risk resistance ability of farmers. Agriculture insurance can cover the potential risks of purchasing agricultural machinery and services [31], and its credit financing synergy can alleviate the financial difficulties in purchasing agricultural machinery and services to some extent, thereby encouraging farmers to invest in purchasing agricultural machinery equipment and services, including machine seeding, machine tillage, machine harvesting, machine drying, etc. [32,33]. Strengthening the application of agricultural machinery is beneficial for improving production efficiency [34], thus facilitating the development of modern agriculture.

2.3. Agriculture Insurance and Agricultural Technology Adoption Behavior

Due to the weak nature of agriculture, coupled with the high cost and risk of introducing and adopting emerging agricultural technologies, farmers are also cautious in their decision-making regarding the application of emerging agricultural technologies. Meanwhile, farmers themselves cannot well resist natural disasters and market risks, and are thus more inclined to continuously choose traditional agricultural technologies [35]. In addition, new agricultural technologies have the triple characteristics of high cost, high risk, and high efficiency, failing to attract farmers with lower economic levels. In this context, agriculture insurance ensures stable production returns and shields farmers from risks associated with new technology adoption. It also enhances credit access, easing financial strains in implementing modern agricultural practices, thus encouraging farmers to embrace innovative technologies and production methods [16,17,36]. Strengthening the adoption of agricultural technologies is beneficial for promoting the output of high value-added agricultural products, improving production efficiency and market competitiveness, and thus facilitating the development of modern agriculture.

2.4. Agriculture Insurance and Planting Structure Selection Behavior

Due to the fragmented cultivated land characteristics of traditional agriculture in China, coupled with self-insurance measures such as crop rotation and diversified planting by farmers, the professional and intensive development of agriculture is quite limited. Agriculture insurance, with its risk diversification and protection capabilities, fosters a higher inclination among farmers to specialize in cultivating insured varieties [37,38]. Meanwhile, agriculture insurance can cover the potential risk brought about by professional and intensive planting, thereby alleviating the risk-related psychological expectations of farmers [39], and further promoting the development of modern agriculture through the advantages of professional, intensive, and industrialized planting.

Therefore, Hypothesis 1 is proposed: agriculture insurance can effectively promote the optimization of agricultural production factor allocation, including agricultural scale input behavior, agricultural machinery application behavior, agricultural technology adoption behavior, and planting structure selection behavior, thereby facilitating the development of modern agriculture.

In addition, the distinct profiles and risk management needs of small farmers and new agricultural operators result in varying levels of protection from current agriculture insurance policies. This variation influences their agricultural production and income differently, thereby affecting the allocation of production factors in distinct ways for each group [40,41]. For small farmers, their production scale is relatively limited, and they usually have a high degree of part-time farming, with a relatively low proportion of agricultural income, and correspondingly lower requirements for agricultural risk management and protection. Therefore, the existing agriculture insurance, with its enhanced guarantee level, is more aligned with the risk management preferences of small farmers. It plays a more significant role in protecting their agricultural production practices and income. This alignment can more effectively encourage farmers to make optimized decisions regarding the allocation of production factors. For new agricultural operators, agricultural production and operation are their main businesses, and they have a higher-level demand for agricultural security [42]. However, the current agriculture insurance system, characterized by a “low guarantee level,” fails to consistently address the actual protection needs of new agricultural operators. Its influence on their agricultural production behavior and the level of agricultural assets is limited, resulting in a reduced incentive to adjust the allocation of production factors.

Therefore, Hypothesis 2 is proposed: agriculture insurance exhibits subject heterogeneity in its effect on promoting agricultural modernization through production factor allocation. In other words, there is a notable disparity in how agricultural insurance impacts the allocation of production factors between small farmers and new agricultural operators.

3. Materials and Methods

3.1. Endogenous Transformation Model: Examining the Factor Allocation Role of Agriculture Insurance in Supporting Modern Agricultural Production

Previous theoretical analysis has revealed that agriculture insurance can affect the allocation of agricultural production factors. Therefore, in order to examine the factor allocation role of agriculture insurance in facilitating modern agricultural production, the following model is constructed [6,43]:

$${\mathrm{F}\mathrm{a}\mathrm{c}\mathrm{t}}_{\mathrm{i}}={\mathsf{\alpha }}_0+{\mathsf{\beta }}_0{\mathrm{I}\mathrm{n}\mathrm{s}\mathrm{u}}_{\mathrm{i}}+\mathsf{\delta }\sum _{\mathrm{j}=1}^{\mathrm{n}}{\mathrm{X}}_{\mathrm{i}}+{\mathsf{\tau }}_{\mathrm{i}}$$

(1)

${\mathrm{F}\mathrm{a}\mathrm{c}\mathrm{t}}_{\mathrm{i}}$ represents the allocation of agricultural production factors, including agricultural scale input behavior, agricultural machinery application behavior, agricultural technology adoption behavior, and planting structure selection behavior. ${\mathrm{I}\mathrm{n}\mathrm{s}\mathrm{u}}_{\mathrm{i}}$ is a dummy variable for agriculture insurance: ${\mathrm{I}\mathrm{n}\mathrm{s}\mathrm{u}}_{\mathrm{i}}=1$ represents farmers who are insured, while ${\mathrm{I}\mathrm{n}\mathrm{s}\mathrm{u}}_{\mathrm{i}}=0$ represents those who are not insured. ${\mathrm{X}}_{\mathrm{i}}$ is a series of control variables, including three categories: basic characteristics of farmers, agricultural production characteristics, and agricultural disaster damage characteristics. ${\mathsf{\tau }}_{\mathrm{i}}$ denotes the error term.

However, the participation of farmers in agriculture insurance decision-making is based on their own basic characteristics and comparative advantages. Meanwhile, there are unobservable variables impacting the participation decision-making of farmers and the allocation of agricultural production factors. Therefore, an endogenous transformation model is hereby employed to correct invalid biased estimates [6,44]. This model boasts the advantages of solving the problem of “self-selection” and “simultaneous decision-making”, incorporating unobservable biases to correct sample selection bias, and separately identifying and analyzing the influencing factors of factor allocation for insured and uninsured farmers. Additionally, the role of agriculture insurance in promoting modern agricultural development is analyzed based on counterfactual analysis.

The specific estimation approach for endogenous transformation models is as follows: The first stage is to estimate the participation of agriculture insurance in decision-making, as shown in Equation (2). The second stage involves estimating the impact effect equations of production factor allocation for insured and uninsured farmers, as shown in Equations (2a) and (2b):

$${\mathrm{I}\mathrm{n}\mathrm{s}\mathrm{u}}_{\mathrm{i}}={\mathsf{\alpha }}_1+\mathsf{\delta }{\mathrm{Z}}_{\mathrm{i}}+{\mathrm{k}}_{\mathrm{i}}{{\mathrm{I}}_{\mathrm{i}}+\mathsf{\vartheta }}_{\mathrm{i}}$$

(2)

$${\mathrm{F}\mathrm{a}\mathrm{c}\mathrm{t}}_{\mathrm{i}\mathrm{T}}=\mathsf{\delta }{\mathrm{X}}_{\mathrm{i}}^{\prime }+{\mathsf{\vartheta }}_{\mathrm{T}}, \mathrm{if}{\mathrm{I}\mathrm{n}\mathrm{s}\mathrm{u}}_{\mathrm{i}}=1$$

(2a)

$${\mathrm{F}\mathrm{a}\mathrm{c}\mathrm{t}}_{\mathrm{i}\mathrm{U}}=\mathsf{\delta }{\mathrm{X}}_{\mathrm{i}}^{\prime }+{\mathsf{\vartheta }}_{\mathrm{U}}, \mathrm{if}{\mathrm{I}\mathrm{n}\mathrm{s}\mathrm{u}}_{\mathrm{i}}=0$$

(2b)

${\mathrm{Z}}_{\mathrm{i}}$ is the factor that affects whether farmers participate in insurance, and ${\mathrm{I}}_{\mathrm{i}}$ is the instrumental variable, namely medical insurance participation behavior. ${\mathrm{F}\mathrm{a}\mathrm{c}\mathrm{t}}_{\mathrm{i}\mathrm{T}}$ and ${\mathrm{F}\mathrm{a}\mathrm{c}\mathrm{t}}_{\mathrm{i}\mathrm{U}}$ represent the allocation of agricultural production factors for insured and uninsured farmers, respectively. In addition, due to the existence of unobservable factors that affect the decision-making behavior of farmers in participating in insurance and the allocation behavior of agricultural production factors, there is a correlation between the residuals of the insurance decision-making equation and the allocation equation of production factors. Therefore, the endogenous transformation model introduces the inverse Mills ratio (${\mathsf{\lambda }}_{\mathrm{i}}$) calculated from the insurance decision equation into the agricultural production factor allocation equation to solve this problem. Consequently, the production factor allocation equations for insured and uninsured farmers are transformed into Equations (3) and (4), respectively:

$${\mathrm{F}\mathrm{a}\mathrm{c}\mathrm{t}}_{\mathrm{i}\mathrm{T}}=\mathsf{\delta }{\mathrm{X}}_{\mathrm{i}}^{\prime }+{\mathsf{\sigma }}_{\mathrm{T}}{\mathsf{\lambda }}_{\mathrm{T}}+{\mathsf{\vartheta }}_{\mathrm{T}}, \mathrm{if}{\mathrm{I}\mathrm{n}\mathrm{s}\mathrm{u}}_{\mathrm{i}}=1$$

(3)

$${\mathrm{F}\mathrm{a}\mathrm{c}\mathrm{t}}_{\mathrm{i}\mathrm{U}}=\mathsf{\delta }{{\mathrm{X}}_{\mathrm{i}}^{\prime }}^{\prime }+{\mathsf{\sigma }}_{\mathrm{U}}{\mathsf{\lambda }}_{\mathrm{U}}+{\mathsf{\vartheta }}_{\mathrm{U}}, \mathrm{if}{\mathrm{I}\mathrm{n}\mathrm{s}\mathrm{u}}_{\mathrm{i}}=0$$

(4)

${\mathsf{\lambda }}_{\mathrm{T}}$ and ${\mathsf{\lambda }}_{\mathrm{U}}$ control for the selectivity bias caused by unobservable variables. ${\mathsf{\sigma }}_{\mathrm{T}}=\mathrm{c}\mathrm{o}\mathrm{v}({\mathsf{\vartheta }}_{\mathrm{T}},\mathsf{\tau })$ and ${\mathsf{\sigma }}_{\mathrm{U}}=\mathrm{c}\mathrm{o}\mathrm{v}({\mathsf{\vartheta }}_{\mathrm{U}},\mathsf{\tau })$ represent the covariance of the insurance decision equation and the production factor allocation equation, respectively. Significant values necessitate addressing “simultaneous decision-making” and “self-selection” issues to mitigate biases from unobservable factors, ensuring the derivation of consistent and unbiased estimates. Furthermore, based on counterfactual analysis, the role of agriculture insurance in supporting modern agricultural production factor allocation is further analyzed, and the differences in production factor allocation between insured and uninsured farmers under both realistic and counterfactual conditions are compared.

The expected conditions for the allocation of agricultural production factors for insured farmers (processing groups) are

$$\mathrm{E}\left[{\mathrm{F}\mathrm{a}\mathrm{c}\mathrm{t}}_{\mathrm{i}\mathrm{T}}\right|{\mathrm{I}\mathrm{n}\mathrm{s}\mathrm{u}}_{\mathrm{i}}=1]=\mathsf{\delta }{\mathrm{X}}_{\mathrm{i}}^{\prime }+{\mathsf{\sigma }}_{\mathrm{T}}{\mathsf{\lambda }}_{\mathrm{T}}$$

(5)

The expected conditions for the allocation of agricultural production factors for uninsured farmers (control group) are

$$\mathrm{E}\left[{\mathrm{F}\mathrm{a}\mathrm{c}\mathrm{t}}_{\mathrm{i}\mathrm{U}}\right|{\mathrm{I}\mathrm{n}\mathrm{s}\mathrm{u}}_{\mathrm{i}}=0]=\mathsf{\delta }{\mathrm{X}}_{\mathrm{i}}^{\prime }+{\mathsf{\sigma }}_{\mathrm{U}}{\mathsf{\lambda }}_{\mathrm{U}}$$

(6)

The expected conditions for the allocation of agricultural production factors when the processing group is not insured are

$$\mathrm{E}\left[{\mathrm{F}\mathrm{a}\mathrm{c}\mathrm{t}}_{\mathrm{i}\mathrm{U}}\right|{\mathrm{I}\mathrm{n}\mathrm{s}\mathrm{u}}_{\mathrm{i}}=1]=\mathsf{\delta }{\mathrm{X}}_{\mathrm{i}}^{\prime }+{\mathsf{\sigma }}_{\mathrm{U}}{\mathsf{\lambda }}_{\mathrm{T}}$$

(7)

The expected allocation conditions of agricultural production factors for the control group involved in insurance are

$$\mathrm{E}\left[{\mathrm{F}\mathrm{a}\mathrm{c}\mathrm{t}}_{\mathrm{i}\mathrm{T}}\right|{\mathrm{I}\mathrm{n}\mathrm{s}\mathrm{u}}_{\mathrm{i}}=0]=\mathsf{\delta }{\mathrm{X}}_{\mathrm{i}}^{\prime }+{\mathsf{\sigma }}_{\mathrm{T}}{\mathsf{\lambda }}_{\mathrm{U}}$$

(8)

Therefore, the average treatment effect (ATT) of the treatment group is the difference between Equations (5) and (7):

$$\mathrm{A}\mathrm{T}\mathrm{T}=\mathrm{E}\left[{\mathrm{F}\mathrm{a}\mathrm{c}\mathrm{t}}_{\mathrm{i}\mathrm{T}}\right|{\mathrm{I}\mathrm{n}\mathrm{s}\mathrm{u}}_{\mathrm{i}}=1]-\mathrm{E}[{\mathrm{F}\mathrm{a}\mathrm{c}\mathrm{t}}_{\mathrm{i}\mathrm{U}}|{\mathrm{I}\mathrm{n}\mathrm{s}\mathrm{u}}_{\mathrm{i}}=1]={\mathrm{X}}_{\mathrm{T}\mathrm{i}}^{\prime }({\mathsf{\delta }}_{\mathrm{T}}-{\mathsf{\delta }}_{\mathrm{U}})+{\mathsf{\lambda }}_{\mathrm{T}}$$

(9)

where, ${\mathsf{\delta }}_{\mathrm{T}}$ and ${\mathsf{\delta }}_{\mathrm{U}}$ represent the parameter estimation results of Equations (5) and (6), respectively, and are a series of explanatory variables in Equation (2a).

The average treatment effect (ATT) of the control group is the difference between Equations (6) and (8):

$$\mathrm{A}\mathrm{T}\mathrm{T}=\mathrm{E}\left[{\mathrm{F}\mathrm{a}\mathrm{c}\mathrm{t}}_{\mathrm{i}\mathrm{T}}\right|{\mathrm{I}\mathrm{n}\mathrm{s}\mathrm{u}}_{\mathrm{i}}=0]-\mathrm{E}[{\mathrm{F}\mathrm{a}\mathrm{c}\mathrm{t}}_{\mathrm{i}\mathrm{U}}|{\mathrm{I}\mathrm{n}\mathrm{s}\mathrm{u}}_{\mathrm{i}}=0]={\mathrm{X}}_{\mathrm{U}\mathrm{i}}^{\prime }({\mathsf{\delta }}_{\mathrm{T}}-{\mathsf{\delta }}_{\mathrm{U}})+{\mathsf{\lambda }}_{\mathrm{U}}({\mathsf{\sigma }}_{\mathrm{T}}-{\mathsf{\sigma }}_{\mathrm{U}})$$

(10)

where, ${\mathrm{X}}_{\mathrm{U}\mathrm{i}}^{\prime }$ is a series of explanatory variables in Equation (2b).

3.2. Mediation Effect Model: Examining the Mechanism of Agriculture Insurance Influencing Factor Allocation

To further investigate the specific mechanism by which agriculture insurance affects the allocation of production factors, a mediation effect model is constructed and validated using the stepwise regression method. The specific model settings are as follows:

$${\mathrm{F}\mathrm{a}\mathrm{c}\mathrm{t}}_{\mathrm{i}}={\mathsf{\gamma }}_1+{\mathsf{\beta }}_2{\mathrm{I}\mathrm{n}\mathrm{s}\mathrm{u}}_{\mathrm{i}}+\mathrm{b}\sum _{\mathrm{j}=1}^{\mathrm{n}}{\mathrm{X}}_{\mathrm{i}}+{\mathsf{\epsilon }}_{\mathrm{i}}$$

(11)

$${\mathrm{I}\mathrm{n}\mathrm{t}\mathrm{e}\mathrm{r}}_{\mathrm{i}}={\mathsf{\gamma }}_2+{\mathsf{\beta }}_2{\mathrm{I}\mathrm{n}\mathrm{s}\mathrm{u}}_{\mathrm{i}}+\mathrm{r}\sum _{\mathrm{j}=1}^{\mathrm{n}}{\mathrm{X}}_{\mathrm{i}}+{\mathrm{u}}_{\mathrm{i}}$$

(12)

$${\mathrm{F}\mathrm{a}\mathrm{c}\mathrm{t}}_{\mathrm{i}}={\mathsf{\gamma }}_3+{\mathsf{\beta }}_3{\mathrm{I}\mathrm{n}\mathrm{s}\mathrm{u}}_{\mathrm{i}}+{\mathsf{\beta }}_4{\mathrm{I}\mathrm{n}\mathrm{t}\mathrm{e}\mathrm{r}}_{\mathrm{i}}+\mathrm{t}\sum _{\mathrm{j}=1}^{\mathrm{n}}{\mathrm{X}}_{\mathrm{i}}+{\mathsf{\theta }}_{\mathrm{i}}$$

(13)

${\mathrm{I}\mathrm{n}\mathrm{t}\mathrm{e}\mathrm{r}}_{\mathrm{i}}$ refers to the mediating variable, involving the area of land reclamation, land transfer area, whether to apply for agricultural credit, the number of purchases of agricultural machinery equipment or services, the amount of agricultural technology training, and the area of insured planting.

3.3. Data Source

The research data are sourced from a survey conducted by the research group from September to November 2021 with farmers in Gansu and Shandong provinces. The survey adopted a stratified random sampling method to determine the study area and sample farmers. Firstly, considering the operation of agriculture insurance and the level of agricultural development, Shouguang City in Shandong Province was selected, which has a relatively good level of development in agriculture and the agriculture insurance market. In addition, Tianshui City in Gansu Province, featuring a relatively average development level in agriculture and the agriculture insurance market, was adopted. This survey mainly focused on crop insurance with a wide implementation scope. Secondly, 4 townships in each city, 4 villages in each township, and 15–20 farmers in each village were randomly selected for research. A total of 614 questionnaires were distributed, and 583 valid questionnaires were collected, presenting a validity rate of 94.95%. A total of 348 insured farmers, 225 small farmers, and 123 new agricultural operators were involved.

3.4. Variable Selection

In this study, the dependent variable is the allocation of agricultural production factors, including agricultural scale input behavior, agricultural machinery application behavior, agricultural technology adoption behavior, and planting structure selection behavior. The four variables are treated as independent variables and are also included in the unified model as dependent variables for overall analysis. Agricultural scale input behavior includes land input scale and capital input scale [40], the scale of land input is measured by the actual cultivated area, and the scale of capital input is determined by the annual capital input per acre of agricultural land (agricultural investment is divided into 5 levels equally based on 0–100%). The agricultural machinery application behavior is a measure of the overall use of agricultural machinery. On the one hand, it is reflected in the degree of agricultural mechanization application of machine seeding, machine tillage, machine harvesting, etc. [32], while on the other, it is reflected in the degree of agricultural water conservancy application of water-saving irrigation. The agricultural mechanization application degree is measured by summing up the number of items used in activities such as machine seeding, machine tillage, and machine harvesting, or the number of purchases of agricultural machinery. The agricultural water conservancy application degree is measured based on the proportion of effective irrigation area equipped with irrigation projects or equipment [2,45]. The agricultural technology adoption behavior can reflect the level of agricultural technology, that is, the citation and adoption of emerging agricultural technologies, measured by the sum of the number of agricultural technology adoption items including plastic film, greenhouses, and straw returning. The planting structure selection behavior can reflect the level of professional production, measured by the Herfindahl index (HI). The core explanatory variable is agriculture insurance, measured based on whether the farmers are insured. Meanwhile, three types of control variables are involved, including basic characteristics of farmers, agricultural production characteristics, and agricultural disaster damage characteristics. Among them, gender, age, and education level are selected to measure the basic characteristics of farmers. Furthermore, the degree of fragmentation of agricultural land (according to the actual total number of plots of land for farmers, 5 levels are divided based on every 3 plots as one level) and part-time employment status are selected to measure the characteristics of agricultural production and operation. Moreover, the frequency and degree of disasters are selected to measure the characteristics of agricultural risk and damage [46]. The specific variables are shown in

Table 1. Descriptive statistics of variables.

Variable Name			Variable Description/Unit	Average	Standard Deviation
dependent variable	agricultural scale input behavior	land input scale	actual cultivated area (ha): [0, 0.13] = 1; (0.13, 0.33] = 2; (0.33, 0.67] = 3; (0.67, 2] = 4; (2, 3.33] = 5; (3.33, 6.67] = 6; (6.67, 13.33] = 7; >13.33 = 8	4.220	2.050
		capital input scale	annual average input in agriculture as a percentage (%): ≤20 = 1; (20, 40] = 2; (40, 60] = 3; (60, 80] = 4; >80 = 5	2.361	0.881
	agricultural machinery application behavior	mechanization application degree	accumulated number of agricultural technology items or number of purchases of agricultural machinery	0.700	0.772
		water conservancy application degree	the proportion of effective irrigation area on cultivated land (%): ≤10 = 1; (10, 30] = 2; (30, 50] = 3; (50, 80] = 4; >80 = 5	1.489	1.552
	agricultural technology adoption behavior	agricultural technology level	accumulated number of agricultural technology items (items)	0.670	0.753
	planting structure selection behavior	professional production level	HI: sum of squared proportions of the planting area of the top three crops	4.220	2.050
explanatory variable	agriculture insurance		insured = 1, not insured = 0	0.597	0.241
control variable	basic characteristics of farmers	age	actual age (years)	45.636	12.481
		gender	female = 0; Male = 1	0.548	0.319
		education level	primary school = 1; junior high school = 2; high school = 3; college degree or above = 4	1.896	0.828
	agricultural production and operation characteristics	degree of fragmentation of agricultural land	total number of plots actually planted (blocks): [1, 3] = 1; [4, 6] = 2; [7, 9] = 3; [10, 12] = 4; >12 = 5	2.943	1.209
		part-time employment status	farming as a supplement = 1; mainly engaged in agriculture = 2; pure farming = 3	1.979	0.776
	characteristics of agricultural risk and disaster losses	disaster frequency	actual number of disasters in the past 5 years (times)	1.166	1.249
		disaster damage level	the proportion of disaster affected area to total planting area/(%): ≤10 = 1; (10, 30] = 2; (30, 50] = 3; >50 = 4	1.166	1.172

.

4. Results and Discussion

4.1. Benchmark Regression Results

4.1.1. Joint Estimation of Agriculture Insurance and Agricultural Production Factor Allocation Model

Table 2. Estimation results of endogenous transformation models for agriculture insurance and agricultural production factor allocation.

	Agriculture Insurance	The Impact of Agricultural Production Factor Allocation
		Agricultural Scale Input Behavior				Agricultural Machinery Application Behavior
		Land Input Scale		Capital Input Scale		Mechanization Application Degree
	Equation (1)	Equation (2)		Equation (3)		Equation (4)
		Insured	Uninsured	Insured	Uninsured	Insured	Uninsured
Age	−0.113 (0.101)	0.094 * (0.053)	0.047 (0.041)	−0.054 (0.049)	0.057 (0.039)	0.079 *** (0.027)	−0.024 (0.012)
Gender	−0.060 (0.185)	0.285 ** (0.100)	−0.140 (0.078)	0.048 * (0.043)	−0.159 (0.073)	0.034 (0.022)	0.062 (0.060)
Education level	−0.036 (0.011)	0.063 (0.060)	−0.022 * (0.044)	0.033 (0.026)	0.005 (0.002)	0.138 *** (0.034)	−0.010 (0.005)
Degree of fragmentation of agricultural land	0.102 (0.099)	−0.475 *** (0.041)	−0.610 *** (0.062)	−0.213 *** (0.039)	−0.305 * (0.055)	−0.017 *** (0.013)	−0.043 (0.032)
Part-time employment status	0.706 *** (0.156)	0.140 (0.099)	0.013 (0.064)	0.094 (0.090)	0.005 (0.004)	0.144 *** (0.052)	0.073 (0.049)
Disaster frequency	0.181 * (0.096)	−0.042 (0.051)	−0.165 *** (0.063)	−0.016 (0.008)	−0.056 * (0.049)	−0.046 (0.026)	−0.083 * (0.048)
Disaster damage level	0.382 *** (0.107)	−0.022 (0.052)	−0.129 * (0.078)	0.047 (0.039)	−0.159 ** (0.073)	−0.018 (0.017)	−0.090 (0.059)
Participation in medical insurance	0.688 *** (0.219)	-	-	-	-	-	-
Regional variables	controlled	controlled	controlled	controlled	controlled	controlled	controlled
Crop type	controlled	controlled	controlled	controlled	controlled	controlled	controlled
Constant	−5.718 *** (0.547)	−2.231 *** (0.481)	0.067 (0.261)	0.934 * (0.497)	0.904 *** (0.245)	1.492 *** (0.217)	0.379 * (0.196)
ρ1u	-	1.230 *** (0.360)	-	0.214 *** (0.208)	-	0.191 ** (0.127)	-
ρ0u	-	-	−0.245 (0.168)	-	−0.219 *** (0.163)	-	−0.167 *** (0.133)
Equation independence test	-	9.20 ***		2.31 *		3.75 *
Model goodness of fit test	-	1096.32 ***		129.09 ***		238.26 ***
Log likelihood	-	−696.944		−668.793		−465.240
N	583	583		583		583
		Agricultural Machinery Application Behavior		Agricultural Technology Adoption Behavior		Planting Structure Selection Behavior
		Water Conservancy Application Degree		Agricultural Technology Level		Professional Production Level
		Equation (5)		Equation (6)		Equation (7)
Age		−0.020 (0.018)	0.020 (0.017)	0.084 (0.030)	0.028 (0.018)	0.011 (0.010)	0.002 (0.001)
Gender		0.080 (0.049)	0.004 (0.001)	0.073 (0.057)	0.026 (0.023)	0.045 ** (0.019)	−0.007 (0.001)
Education level		0.105 (0.089)	−0.028 (0.019)	0.092 *** (0.034)	0.033 (0.030)	0.021 * (0.017)	0.014 * (0.012)
Degree of fragmentation of agricultural land		−0.078 (0.024)	−0.141 (0.097)	−0.052 (0.026)	−0.063 (0.039)	0.007 (0.002)	0.001 (0.003)
Part-time employment status		0.445 *** (0.148)	0.044 (0.041)	0.051 (0.047)	0.016 (0.014)	0.031 (0.019)	0.028 (0.017)
Disaster frequency		−0.028 (0.025)	−0.032 * (0.041)	0.059 ** (0.029)	−0.016 (0.013)	−0.004 (0.010)	−0.009 (0.007)
Disaster damage level		−0.045 (0.036)	−0.083 * (0.050)	−0.035 (0.029)	−0.007 (0.002)	0.002 (0.001)	−0.058 *** (0.020)
Participation in medical insurance		-	-	-	-	-	-
Regional variables		controlled	controlled	controlled	controlled	controlled	controlled
Crop type		controlled	controlled	controlled	controlled	controlled	controlled
Constant		2.885 *** (0.586)	0.346 ** (0.163)	0.946 *** (0.234)	0.254 (0.170)	0.741 *** (0.080)	0.635 *** (0.069)
ρ1u		0.586 *** (0.206)	-	0.241 * (0.124)	-	0.092 * (0.072)	-
ρ0u		-	−0.301 * (0.169)	-	−0.430 *** (0.155)	-	−0.055 * (0.152)
Equation independence test		12.60 ***		13.31 ***		7.65 *
Model goodness of fit test		26.59 ***		183.03 ***		71.50 ***
Log likelihood		−609.239		−438.335		−140.735
N		583		583		583

*, **, and *** indicate significance at the 1%, 5%, and 10% levels, respectively, with values in parentheses indicating standard error.

shows the simultaneous estimation results of the endogenous transformation models. The estimated result of the influencing factors of agriculture insurance participation is presented in Equation (1), and Equations (2)–(7) demonstrate the estimated influencing factors of agricultural production factor allocation, including land input scale, capital input scale, mechanization application degree, water conservancy application degree, agricultural technology adoption behavior, and planting structure selection behavior. As indicated by the estimation results, each equation rejects the null hypothesis that the agriculture insurance participation equation and the impact of agricultural production factor allocation are independent of each other at different statistical levels, and the ${\rho }_{1u}$ or ${\rho }_{0u}$ values of each equation are significant at different statistical levels, indicating the existence of estimation bias caused by self-selection and unobservable factors. Hence, endogenous transformation models should be established for correction. The Wald test values for the model fit of each equation are significant at the 1% statistical level, indicating that the overall regression coefficients of the models are significant.

At the same time, the instrumental variables “participation in medical insurance” and agriculture insurance in each equation are also significant at the 1% statistical level. Furthermore, the estimation results using the OProbit model for the impact of “participation in medical insurance” on land input scale, capital input scale, mechanization application degree, water conservancy application degree, agricultural technology adoption behavior, and planting structure selection behavior are all insignificant. In addition, the instrumental variable model (IV-2SLS) is employed to estimate the impact of agriculture insurance on the allocation of agricultural production factors. The first-stage results, with an F-value of 260.02, are significant at the 1% level, indicating strong instrumental variables. Additionally, the non-significant p-values from the overidentification tests suggest that the null hypothesis of exogeneity for the instrumental variable cannot be rejected, confirming the validity of the model’s instrumental variables. Therefore, “participation in medical insurance” can be appropriately chosen as a tool variable for agriculture insurance.

As indicated by the estimation results of production factor allocation, there are differences in factors affecting the allocation of various production factors. In terms of agricultural scale input behavior, the degree of fragmentation of agricultural land has a significant negative impact on the land input scale. Meanwhile, the number of disasters and the degree of damage have a significant negative impact on the land input scale and capital input scale of uninsured farmers at different statistical levels. However, the impact on insured farmers is not significant. This indicates that participating in insurance has reduced the negative inhibitory effect of risk and disaster losses on land input scale and capital input scale to some extent. In terms of agricultural machinery application behavior, part-time employment status has a significant positive impact on agricultural mechanization application degree and water conservancy application degree of insured farmers at the 1% statistical level, but the impact on uninsured farmers is not significant. At the same time, the degree of fragmentation of agricultural land has a significant negative impact on agricultural mechanization application degree among insured farmers at the 1% statistical level. In addition, the disaster frequency has a significant negative impact on the degree of agricultural mechanization application degree and water conservancy application degree of uninsured farmers at the 10% statistical level, but the impact on insured farmers is not significant. This indicates that participating in insurance can to some extent reduce the inhibitory effect of risk disasters on improving machinery application behavior. In terms of agricultural technology adoption behavior, the number of disasters and the education level significantly affect the agricultural technology adoption behavior of insured farmers at different statistical levels. This indicates that insured farmers reduce the negative inhibitory effect of risk and disaster losses on their agricultural technology adoption behavior to a certain extent, which is conducive to enhancing their enthusiasm for adopting new agricultural technologies and production models. In terms of planting structure selection behavior, the degree of disaster damage has a significant negative impact on the specialization production level of uninsured farmers at the 1% statistical level, but the impact on insured farmers is not significant. This indicates that participating in insurance has reduced the negative inhibitory effect of disaster damage on professional production to some extent.

4.1.2. Estimation of the Processing Effect of Agriculture Insurance and Agricultural Production Factor Allocation

Table 3. Estimated results of the treatment effect of agriculture insurance on the allocation of agricultural production factors.

	Agricultural Scale Input Behavior		Agricultural Machinery Application Behavior		Agricultural Technology Adoption Behavior	Planting Structure Selection Behavior
	Land Input Scale	Capital Input Scale	Mechanization Application Degree	Water Conservancy Application Degree	Agricultural Technology Level	Professional Production Level
ATT	0.085 *** (0.056)	0.160 *** (0.064)	0.240 *** (0.013)	0.491 *** (0.049)	0.194 *** (0.034)	0.056 *** (0.015)
ATU	0.039 *** (0.020)	0.031 *** (0.019)	0.381 *** (0.024)	0.434 *** (0.061)	0.062 *** (0.014)	0.004 *** (0.003)
Sample quantity	583	583	583	583	583	583

*** indicated significance at the 1% levels, with values in parentheses indicating standard error.

presents the estimated effects of agriculture insurance on the allocation of agricultural production factors. The results show that insurance participation has a significant positive impact on agricultural scale input behavior including land input scale and capital input scale, agricultural machinery application behavior including the mechanization application degree and water conservancy application degree, agricultural technology adoption behavior, and planting structure selection behavior. The impact of agricultural insurance on the application of agricultural machinery is the most pronounced, whereas its influence on the selection of planting structures is comparatively minimal. Therefore, agriculture insurance plays a promoting role in influencing the allocation of production factors, thereby facilitating the development of modern agriculture. Hypothesis 1 is thus verified.

From the perspective of agricultural scale input behavior, the average treatment effect of participating in insurance on land input scale and capital input scale is significantly positive at the 1% statistical level. This indicates that participating in insurance significantly promotes the scale input behavior of farmers, and that agricultural land input scale and capital input scale are significantly improved. In addition, agriculture insurance can simultaneously diversify and manage natural risks and market risks in agricultural products, thus ensuring agricultural production and stable expected returns. Based on this, agriculture insurance can, on one hand, effectively motivate farmers to expand their agricultural land production scale. On the other hand, it can amplify agricultural capital investment by leveraging the synergistic effects of agricultural insurance with credit financing. This approach facilitates both the scaling up of land input and the infusion of capital necessary for agricultural advancement.

From the perspective of agricultural machinery application behavior, the average treatment effect of participating in insurance on mechanization application degree and water conservancy application degree is significantly positive at the 1% statistical level, indicating that participating in insurance significantly enhances the agricultural machinery application behavior of farmers. Meanwhile, the agricultural mechanization application degree and water conservancy application degree are significantly improved. Agriculture insurance helps to improve the risk resistance ability of agricultural production and operation, thereby enhancing the enthusiasm of farmers to apply and upgrade agricultural production machinery and equipment. Meanwhile, the synergistic effect of agriculture insurance credit financing also provides financial support for improving agricultural mechanization application degree and water conservancy application degree, thereby promoting the agricultural machinery application behavior of farmers and enhancing the level of agricultural mechanization.

From the perspective of agricultural technology adoption behavior, the average treatment effect of participating in insurance on agricultural technology adoption is significantly positive at the 1% statistical level. This indicates that participating in insurance effectively promotes the agricultural technology adoption behavior of farmers. Agriculture insurance bolsters farmers’ willingness to introduce and adopt innovative agricultural technologies and production models by positively influencing their risk-related expectations. Meanwhile, the synergistic effect of agriculture insurance credit has also solved the funding problem of agricultural technology introduction, thereby inspiring farmers to adopt agricultural technology and improving the level of agricultural technology application.

From the perspective of planting structure selection behavior, the average treatment effect of participating in insurance on the level of professional production is significantly positive at the 1% statistical level, indicating that participating in insurance effectively improves the level of agricultural specialization. Agriculture insurance can reduce the tendency of farmers to diversify their risk by improving their risk-related psychological expectations. This, in turn, effectively strengthens their professional planting behavior and promotes the professional and intensive development of agriculture.

4.2. Robust Testing

4.2.1. Robustness Test One: Excluding Extreme Samples

To test the robustness of the benchmark estimation results, the largest 10% and smallest 10% of land use scale farmers are removed, and the same method is used to re-estimate the impact of agriculture insurance participation on the allocation of agricultural production factors. The specific results are shown in

Table 4. Estimated results of processing effects for excluding extreme samples.

	Scale Level		Mechanization Level		Technology Level	Professional Level		Green Level
	Land Scale	Investment Scale	Mechanization Degree	Water Conservancy Degree		HI	MI	Fertilizer Application Amount
ATT	0.076 *** (0.040)	0.153 *** (0.072)	0.227 *** (0.029)	0.475 *** 0.065	0.169 *** (0.024)	0.047 *** (0.012)	0.082 *** (0.026)	−0.186 *** (0.037)
ATU	0.034 *** (0.027)	0.027 ** (0.014)	0.362 *** (0.026)	0.416 *** 0.069	0.058 *** (0.016)	0.003 * (0.002)	0.044 *** (0.014)	−0.385 *** (0.089)
Sample quantity	467	467	467	467	467	467	467	467

*, **, and *** indicate significance at the 1%, 5%, and 10% levels, respectively, with values in parentheses indicating standard error. (2) Due to space limitations, only the estimated results of processing effects are reported. The estimated results of agriculture insurance and factors affecting the allocation of agricultural production factors for the two types of agricultural operators are basically consistent with the full sample estimation results.

.

The results show that the average treatment effects of agriculture insurance on land input scale, capital input scale, mechanization application degree, water conservancy application degree, agricultural technology level, and professional production level are significantly positive at different statistical levels. This suggests that agriculture insurance can effectively promote agricultural scale input behavior, agricultural machinery application behavior, agricultural technology adoption behavior, and agricultural professional planting behavior, thereby facilitating the development of modern agriculture. Therefore, after excluding extreme samples, the estimated impact of agriculture insurance participation on the allocation of agricultural production factors is consistent with the benchmark regression results, confirming the robustness.

4.2.2. Robustness Test Two: Changing Estimation Methods

To further test the robustness of the benchmark estimation results, the Heckman two-stage model is employed to estimate the impact of agriculture insurance participation on the allocation of agricultural production factors [47]. The specific results are shown in

Table 5. Joint estimation results of farmers’ insurance decision-making model and agricultural production factor allocation model.

	Insurance Decision Model	Impact of Agricultural Production Factor Allocation
	Agriculture Insurance Participation	Agricultural Scale Input Behavior		Agricultural Machinery Application Behavior		Agricultural Technology Adoption Behavior	Planting Structure Selection Behavior
		Land Input Scale	Capital Input Scale	Mechanization Application Degree	Water Conservancy Application Degree	Agricultural Technology Level	Professional Production Level
Participation behavior in agriculture insurance	-	0.057 ** (0.044)	0.006 * (0.037)	0.167 *** (0.066)	0.641 *** (0.029)	0.012 *** (0.015)	0.059 *** (0.051)
Whether to participate in medical insurance	0.827 *** (0.257)	-	-	-	-	-	-
λ	-	−0.024 ***	−0.088 **	−0.315 ***	−0.157 ***	−0.605 ***	−0.033 ***
Wald test value	-	4443.65 ***	582.99 ***	1088.04 ***	1602.94 ***	1074.76 ***	385.43 ***
Regional variables	controlled	controlled	controlled	controlled	controlled	controlled	controlled
Crop type	controlled	controlled	controlled	controlled	controlled	controlled	controlled
N	583	583	583	583	583	583	583

*, **, and *** indicate significance at the 1%, 5%, and 10% levels, respectively, with values in parentheses indicating standard error. (2) Due to space limitations, only the estimated results of processing effects are reported. The estimated results of agriculture insurance and factors affecting the allocation of agricultural production factors for the two types of agricultural operators are basically consistent with the full sample estimation results.

.

The results present a significant inverse Mills ratio (λ) in the Heckman two-stage model at different statistical levels, indicating the existence of sample selection bias. This suggests that there is a correlation between farmer participation decisions and agricultural production factor allocation decisions. Meanwhile, the Wald test values are all significant at the 1% statistical level, indicating a significant overall regression coefficient for the model. In addition, the impact of agriculture insurance on land input scale, capital input scale, mechanization application degree, water conservancy application degree, agricultural technology level, and professional production level is significantly positive at different statistical levels. This demonstrates that agriculture insurance can effectively promote agricultural scale input behavior, agricultural machinery application behavior, agricultural technology adoption behavior, and agricultural professional planting behavior, thereby facilitating the development of modern agriculture. The above results indicate that, after changing the estimation method, the estimation results of the impact of agriculture insurance on the allocation of agricultural production factors are consistent with the benchmark regression results, confirming the robustness.

4.3. Further Analysis

4.3.1. Mechanism Analysis

Based on the previous analysis, agriculture insurance can effectively promote the allocation of agricultural production factors, but the specific mechanism still needs further investigation. Therefore, a mediation effect model is hereby used to verify the mediating transmission mechanism of agriculture insurance in promoting the allocation of agricultural production factors. Additionally, robustness tests are conducted using the Sobel method [18]. The estimated results are shown in

Table 6. Intermediary transmission mechanism of agriculture insurance promoting modern agricultural development.

Mechanism of Action		Path 1	Coefficient	Path 2	Coefficient	Mediating Effect	Sobel
Agricultural scale input behavior	land input scale	insured→reclamation	0.141 (0.059)	reclamation→land input scale	0.068 (0.027)	0.010	0.604
		insured→land transfer	0.107 *** (0.021)	land transfer→land input scale	0.046 *** (0.013)	0.005 ***	0.005 ***
	capital input scale	insured→agricultural financing credit	0.046 (0.021)	agricultural financing credit→capital input scale	0.020 (0.011)	0.001	0.216
Agricultural machinery application behavior		insured→purchase agricultural machinery equipment or services	0.087 *** (0.026)	purchase agricultural machinery equipment or services→mechanization application degree	0.142 *** (0.047)	0.012 ***	0.004 ***
Agricultural technology adoption behavior		insured→agricultural technology training	0.055 *** (0.010)	agricultural technology training→agricultural technology level	0.086 *** (0.019)	0.005 ***	0.007 ***
Planting structure selection behavior		insured→scale of insured product cultivation	0.168 *** (0.064)	scale of insured product cultivation→professional production level	0.213 *** (0.044)	0.036 ***	0.009 ***

*** indicates significance at the 1% levels, with values in parentheses indicating standard error.

.

The estimation results indicate that agriculture insurance mainly promotes agricultural scale input behavior through land transfer, promotes agricultural machinery application behavior by purchasing agricultural machinery equipment or services, promotes agricultural technology adoption behavior by strengthening agricultural technology training, and enhances professional production level by increasing the scale of insured planting. Meanwhile, the intermediary transmission paths of agriculture insurance promoting land input scale through land reclamation and increasing agricultural financing credit to increase capital input scale are not significant. In addition, the robustness test results acquired using the Sobel method indicate robustness of the mediating effect.

Firstly, from the perspective of land input scale, agriculture insurance mainly increases land input scale by promoting land transfer. In addition, the impact of land reclamation on land input scale is not significant. This indicates that agriculture insurance helps to enhance the vitality of the rural land transfer market, activate idle and abandoned land, and thereby expand the scale of agriculture production and operation. Secondly, from the perspective of capital input scale, the transmission mechanism of agriculture insurance increasing capital input scale through agricultural financing credit is not significant. A possible reason is that the synergistic effect of agriculture insurance financing credit has increased the enthusiasm of farmers to apply for agricultural credit. However, the proportion of successfully approved credit is relatively small. Research indicates that in the past three years, 34.82% of farmers have sought agricultural credit, yet only 14.75% have been granted approval. This discrepancy highlights that the barrier to credit access for agricultural operators remains substantial, particularly for small farmers. It underscores the need for financial institutions to enhance the availability of agricultural credit to better serve this sector. Thirdly, from the perspective of agricultural machinery application behavior, agriculture insurance helps to enhance the enthusiasm of farmers to purchase agricultural machinery equipment or services, promote their investment in agricultural machinery application, and thereby enhance the degree of mechanization application of agricultural machinery in sowing, tillage, and harvesting. Fourthly, from the perspective of agricultural technology adoption behavior, participating in insurance can enhance the agricultural technology literacy of farmers by promoting their participation in agricultural technology training. This, in turn, promotes their enthusiasm for introducing and adopting emerging agricultural technologies and new production models, and strengthens their practical application level of emerging agricultural technologies and new production models. Fifthly, from the perspective of planting structure selection behavior, agriculture insurance can smooth out the self- insurance tendency of farmers towards diversified planting, while also increasing the expected returns of insured varieties. This helps to increase the planting and operation area of insured products, encourages farmers to engage in professional planting and production, and improves the level of agricultural specialization production.

4.3.2. Heterogeneity Analysis

(1) Farmer heterogeneity analysis

According to the scale of land management, the farmers are divided into small farmers (land area < 3.33 hectare) and new agricultural operators (land area ≥ 3.33 hectare) to examine the impact of agriculture insurance on the allocation of agricultural production factors among heterogeneous operating entities. The estimated results are shown in

Table 7. Estimated results of the treatment effect of agriculture insurance on the allocation of agricultural production factors among heterogeneous operating entities.

		Agricultural Scale Input Behavior		Agricultural Machinery Application Behavior		Agricultural Technology Adoption Behavior	Planting Structure Selection Behavior
		Land Input Scale	Capital Input Scale	Mechanization Application Degree	Water Conservancy Application Degree	Agricultural Technology Level	Professional Production Level
Small farmers	ATT	0.034 (0.061)	0.124 (0.073)	0.123 *** (0.036)	0.249 *** (0.024)	0.076 *** (0.019)	0.038 *** (0.014)
	Sample quantity	333	333	333	333	333	333
New agricultural operators	ATU	0.122 *** (0.072)	0.193 *** (0.098)	0.407 *** (0.085)	0.630 *** (0.062)	0.245 *** (0.065)	0.081 (0.024)
	Sample quantity	250	250	250	250	250	250

*** indicates significance at the 1% levels with values in parentheses indicating standard error. (2) Due to space limitations, only the estimated results of processing effects are reported. The estimated results of agriculture insurance and factors affecting the allocation of agricultural production factors for the two types of agricultural operators are basically consistent with the full sample estimation results.

.

The estimation results reveal significant differences in the impact of agriculture insurance on the allocation of production factors between small farmers and new agricultural operators. For small farmers, participating in insurance has a significant promoting effect on their agricultural machinery application behavior, agricultural technology adoption behavior, and planting structure selection behavior, but its impact on agricultural scale input behavior is not significant. This suggests that agriculture insurance mainly facilitates small farmers to achieve modernization of agricultural production by improving agricultural mechanization application degree, agricultural water conservancy application degree, and professional production level. For new agricultural operators, participating in insurance significantly promotes their agricultural scale input behavior, agricultural machinery application behavior, and agricultural technology adoption behavior, while having no significant impact on their planting structure selection behavior. Agriculture insurance mainly facilitates new agricultural operators to achieve modernization of agricultural production by increasing land and capital input scale, agricultural mechanization, water conservancy application degree, and agricultural technology level. Hypothesis 2 is thus verified.

From the perspective of agricultural scale input behavior, compared to small farmers, the income level of new agricultural operators is relatively high, and they have strong enthusiasm for scale expansion and financial support. Therefore, under agriculture insurance incentives, the impact on the land and capital scale of new agricultural operators is more pronounced. In terms of planting structure selection behavior, small farmers are more vulnerable to risks and more inclined to self-insurance through diversified planting and crop rotation. Hence, participating in agriculture insurance can enhance their risk resistance ability, thereby incentivizing small farmers to engage in professional production. However, since new agricultural operators typically employ specialized and intensive methods for their production and operations, there is limited room for them to further enhance their level of specialized production through incentives provided by agriculture insurance.

(2) Regional heterogeneity analysis

To further analyze the differences in the allocation of agricultural production factors under agriculture insurance across different regions, the impact of farmer participation in insurance on the allocation of agricultural production factors is classified according to households in Gansu and Shandong. The estimated results are shown in

Table 8. Estimated results of the treatment effect of agriculture insurance on the allocation of agricultural production factors among heterogeneous operating entities.

		Agricultural Scale Input Behavior		Agricultural Machinery Application Behavior		Agricultural Technology Adoption Behavior	Planting Structure Selection Behavior
		Land Input Scale	Capital Input Scale	Mechanization Application Degree	Water Conservancy Application Degree	Agricultural Technology Level	Professional Production Level
Gansu province	ATT	0.051 *** (0.345)	0.009 * (0.124)	0.044 ** 0.174)	0.032 (0.218)	0.017 *** (0.163)	0.082 *** (0.117)
	sample quantity	356	356	356	356	356	356
Shandong province	ATU	0.034 * (0.128)	0.042 *** (0.119)	0.026 (0.043)	0.013 (0.036)	0.019 (0.051)	0.056 *** (0.047)
	sample quantity	227	227	227	227	227	227

*, **, and *** indicate significance at the 1%, 5%, and 10% levels, respectively, with values in parentheses indicating standard error. (2) Due to space limitations, only the estimated results of processing effects are reported. The estimated results of agriculture insurance and factors affecting the allocation of agricultural production factors in the two regions are basically consistent with the full sample estimation results.

.

The estimation results reveal a significant difference in the impact of agricultural insurance on the allocation of production factors among farmers in Gansu Province and Shandong Province. For farmers in Gansu Province, participating in insurance significantly promotes their agricultural scale input behavior, agricultural machinery application behavior, agricultural technology adoption behavior, and planting structure selection behavior. However, for those in Shandong Province, participating in insurance only has a significant promoting effect on their agricultural scale input behavior and planting structure selection behavior, while exerting no significant impact on their agricultural machinery application behavior and agricultural technology adoption behavior. This may be attributed to a significant gap in the level of agricultural development between the two regions. Shandong, being a key agricultural province with advanced agricultural development, high levels of mechanization, and technological adoption, has relatively less room for further optimizing the allocation of agricultural production factors via agricultural insurance. This is particularly true given the current low guarantee level of the agricultural insurance operation mode. However, Gansu has a comparatively lower level of agricultural development, and farmers there have less robust risk management capabilities. Consequently, after enrolling in insurance programs, the risk mitigation effects of agriculture insurance and the credit synergy have a more pronounced impact on their agricultural production and the allocation of production factors.

5. Conclusions

Systematically grasping the impact of agriculture insurance on agricultural production behavior and factor allocation from the perspective of facilitating modern agricultural production not only contributes to clarifying the function of agriculture insurance in poverty prevention, but also provides decision-making reference for agriculture insurance to boost efficiency, consolidate poverty alleviation achievements, and promote rural revitalization. Using farmer survey data and employing endogenous transformation models along with mediation effect models, this study provides a theoretical framework and empirical investigation into the role and mechanisms by which agricultural insurance influences the allocation of factors, thereby promoting the development of modern agriculture. The results show that, firstly, agriculture insurance has a significant positive impact on scale input behavior including land input scale and capital input scale, agricultural machinery application behavior including the mechanization and water conservancy application degree, agricultural technology adoption behavior, and planting structure selection behavior. The impact of agriculture insurance on the application behavior of agricultural machinery is the most significant, while its influence on the selection behavior of planting structures is the least pronounced. Secondly, agriculture insurance mainly promotes agricultural scale input behavior through land transfer, promotes agricultural machinery application behavior through the purchasing of agricultural machinery and equipment or services, promotes agricultural technology adoption behavior by strengthening agricultural technology training, and enhances professional production levels by increasing the scale of insured planting. However, the intermediary transmission path of promoting land input scale through land reclamation and increasing agricultural financing credit to increase capital input scale is not significant, which may be related to the serious moral hazard issues of farmers under the current high-subsidy insurance model. Thirdly, there are significant differences in the impact of agriculture insurance on the allocation of production factors between small farmers and new agricultural operators. For small farmers, participating in agriculture insurance significantly promotes their agricultural machinery application behavior, agricultural technology adoption behavior, and planting structure selection behavior. However, for new agricultural operators, participating in agriculture insurance has a significant promoting effect on their agricultural scale input behavior, agricultural machinery application behavior, and agricultural technology adoption behavior, while exerting no significant impact on their planting structure selection behavior.

Based on the above research conclusions, the following policy implications can be obtained. Firstly, the directionality of agriculture insurance and subsidy policies should be enhanced, and the autonomous and endogenous development of agricultural operators should be encouraged. Attention should also be paid to the directional role of agriculture insurance and fiscal subsidy policies in guiding the autonomous and endogenous development of agricultural operators, and farmers should be inspired to shift towards intensive and modern agricultural production and operation models. Meanwhile, it is also advisable to place emphasis on improving government policy support and financial subsidies for agriculture insurance, guiding farmers to further optimize the allocation of production factors, enhancing their driving force for upgrading and optimizing agricultural production and operation, promoting agricultural modernization, and promoting long-term poverty prevention. Secondly, efforts should be made to further enhance the overall level of security, expand the implementation scope of full cost insurance and income insurance, and set multiple protection levels after considering the differentiation factors of different regions and farmers. At the same time, refining the insurance contract is essential, such as enhancing the existing tiered compensation regulations, fine-tuning the deductible and compensation ratios, etc. These adjustments aim to further improve the efficiency and protection level of agriculture insurance disaster compensation. Thirdly, the collaborative and interactive mechanism between agriculture insurance and agricultural credit financing should be strengthened. Furthermore, to address the difficulties faced by farmers in financing due to insufficient stability of agricultural returns and a lack of collateral, it is necessary to highlight the effective combination of agriculture insurance and agricultural credit financing. The financial difficulties faced by farmers should be solved by further underscoring the synergistic effect of agricultural insurance credit financing, enhancing the enthusiasm of farmers to increase agricultural capital investment, and providing financial support for the adoption and upgrading of agricultural machinery, technology, etc. Meanwhile, it also holds considerable significance for enriching the financial products and service models related to agriculture insurance, strengthening the financial support of agriculture insurance for agricultural technological progress, optimizing agricultural production materials, and improving human capital, thereby achieving a win–win situation for all parties.