A Study on the Nonlinear Impact of Agricultural Insurance on the Resilience of Agricultural Economy

Abstract

With the deepening implementation of agricultural full-cost insurance and crop income insurance, agricultural insurance has gradually become a significant force in promoting agricultural and rural modernization and achieving the strategic goal of building a strong agricultural nation. Based on data from 30 provinces in China from 2011 to 2023, a comprehensive evaluation index system for agricultural economic resilience was constructed, and the impact of agricultural insurance on agricultural economic resilience, along with its underlying mechanisms, was systematically analyzed. The findings reveal that: (1) There exists a nonlinear “U-shaped” relationship between agricultural insurance and agricultural economic resilience, a conclusion that remains robust after a series of tests; (2) Agricultural insurance can positively influence agricultural economic resilience by promoting agricultural technological progress; (3) When the level of industrial structure exceeds 7.108, agricultural insurance has a significant effect on agricultural economic resilience, and as the industrial structure level improves, the promoting effect of agricultural insurance becomes more pronounced; (4) The “U-shaped” impact of agricultural insurance on agricultural economic resilience is more prominent in the eastern, central, and northeastern regions, while it is not significant in the western region.

1. Introduction

A strong nation must first have a strong agricultural sector; only when agriculture is strong can the country be strong. The report to the 20th National Congress of the Communist Party of China established, for the first time, the strategic goal of “expediting the building of an agricultural powerhouse” and incorporated it into the overall plan for modernizing the country [1]. However, as the once-in-a-century global transformation accelerates, China’s development has entered a period where strategic opportunities and risks coexist, and uncertain and unpredictable factors are increasing. In this context, the development of the agricultural economy also faces numerous uncertainties and shocks. The Central Document No. 1 of 2023 highlights “building a diversified food supply system,” “strengthening emergency grain supply capabilities,” and “enhancing agricultural disaster prevention and mitigation capacities” [2]. This indicates that the state has put forward new and higher requirements for the resilience of the agricultural economy—that is, to strengthen the ability of the agricultural system to quickly adjust itself in response to external pressures, thereby absorbing and withstanding shocks. Therefore, enhancing the resilience of the agricultural economy and improving its “resistance–adaptability–reconstruction capacity” are of great significance for ensuring national food security and achieving the strategic goal of building a strong agricultural nation.

As an important tool for ensuring the high-quality development of the agricultural economy, agricultural insurance helps prevent economic losses caused during agricultural production. Since 2007, the central government has provided premium subsidies to farmers purchasing agricultural insurance, marking the beginning of the development of policy-oriented agricultural insurance. In 2019, the Ministry of Finance, the Ministry of Agriculture and Rural Affairs, the China Banking and Insurance Regulatory Commission, and the National Forestry and Grassland Administration jointly issued the Guiding Opinions on Accelerating the High-Quality Development of Agricultural Insurance, which clarified the goals and direction for the high-quality development of agricultural insurance in China. According to data from the Ministry of Agriculture and Rural Affairs, in 2024, the premium income of agricultural insurance in China reached approximately 150 billion yuan, providing risk protection exceeding 5 trillion yuan for farmers [3]. This demonstrates that agricultural insurance plays an important role in the reform of the national “agriculture, rural areas, and farmers” sector and within the agricultural support policy system [4].

At present, the development of China’s agricultural economy faces numerous uncertainties and challenges from external risks. These include not only increasingly complex economic conditions and issues related to the structure of the agricultural industry but also vulnerabilities arising from environmental and social changes [5,6]. Research on how to leverage agricultural insurance to enhance the resilience of the agricultural economy, the effectiveness of its impact, and the mechanisms at play is conducive to better harnessing the role of agricultural insurance, promoting the enhancement of agricultural economic resilience, and is of vital importance for achieving the high-quality development of agricultural insurance and the strategic goal of building a strong agricultural nation.

2. Literature Review

Research on resilience can be traced back to Holling (1973), who applied the concept in ecology, defining resilience as the ability of a system to quickly recover to an equilibrium state after being subjected to uncertain shocks—that is, the capacity to respond to and adjust in the face of uncertainty [7], also referred to as adaptive resilience [8,9]. Subsequently, many scholars incorporated resilience into the analytical framework of economics, giving rise to the concept of “economic resilience.” This refers to the characteristic of an economic system to exhibit strong resistance and recovery when faced with shocks during its development and to reach a new equilibrium state after enduring pressure [10,11]. Applied to the agricultural economy, scholars define agricultural economic resilience as the regulatory capacity of the agricultural economic system to mitigate the negative impacts brought about by external pressures [12,13,14].

In specific studies, researchers have primarily focused on the evaluation of agricultural economic resilience and the analysis of its influencing factors. Regarding the evaluation of agricultural economic resilience, some scholars have constructed indicator systems from the two dimensions of resistance and reconstruction to assess it [15]; others have used the Pressure-State-Response (PSR) model to develop evaluation indicators for agricultural economic resilience [16]; and still others have selected indicators from three dimensions—risk resistance and recovery capacity, adaptive regulation capacity, and transformation and innovation capacity—to evaluate agricultural economic resilience [17]. In terms of the influencing factors of agricultural economic resilience, scholars argue that crop diversification, the diversification and commercialization of agricultural production, and improvements in total factor productivity in agriculture are important factors that enhance agricultural economic resilience [18,19,20]. Some also suggest that the integration and optimization of industrial structures can promote the enhancement of agricultural economic resilience [14,21]. Additionally, the overall level of agricultural infrastructure is regarded as a crucial factor affecting agricultural economic resilience, as it can improve the system’s adaptive regulation, innovation and transformation, and risk resistance capacities [22].

On the basis of clarifying the connotation of agricultural economic resilience and its influencing factors, how to enhance agricultural economic resilience through effective policy instruments has become a critical issue. In this context, as an important risk management and policy instrument, agricultural insurance is regarded as a potential key pathway for enhancing agricultural economic resilience. Existing literature suggests that agricultural insurance has a welfare effect on farmers. It not only increases farmers’ income by promoting the scale, mechanization, specialization, and greening of agricultural production but also enhances farmers’ risk awareness and enables them to benefit from disaster prevention services [23,24,25]. Meanwhile, some scholars argue that the increase in agricultural insurance coverage helps improve farmers’ total factor productivity and guides agricultural green development. It can also curb agricultural carbon emissions through production specialization and agricultural technological progress, thereby promoting efficient and green agricultural development [26,27,28]. Additionally, in related empirical studies, scholars often use agricultural insurance premiums as a proxy variable to measure its development level [29] or construct policy variables based on the timing of the central government’s premium subsidy policies in various provinces to examine the actual impact of agricultural insurance [30]. It is worth noting that most existing studies focus on the direct impact of agricultural insurance on agricultural economic resilience, explore the mechanisms through which agricultural insurance affects agricultural economic resilience from a macro perspective, or emphasize the positive role of insurance’s loss compensation function in short-term recovery [31]. However, they often overlook the fact that in the early stages of its development, the positive economic effects of agricultural insurance may not be significant due to institutional frictions. Based on this, this paper integrates insurance function theory and behavioral economics perspectives to systematically explain the impact of agricultural insurance on agricultural economic resilience and its underlying mechanisms.

The potential contributions of this paper are as follows: First, in terms of research perspective, most studies tend to explore the linear relationship between agricultural insurance and agricultural economic resilience, often overlooking the possibility of nonlinear effects. This paper adopts a nonlinear perspective, revealing a U-shaped impact of agricultural insurance on agricultural economic resilience—initially inhibitory and later promotional—thereby providing a more refined depiction of the complex relationship between the two. Second, in terms of mechanistic explanation, it elaborates on the internal mechanism through which agricultural insurance influences agricultural economic resilience by promoting agricultural technological progress, offering new insights into understanding the long-term enabling effects of agricultural insurance. Third, in terms of research methodology, the introduction of a panel threshold regression model identifies the level of industrial structure as a key threshold variable in the influence of agricultural insurance on agricultural economic resilience, facilitating the recognition of significant heterogeneous characteristics in the enabling effects of agricultural insurance across different stages of industrial structure.

3. Theoretical Analysis and Research Hypotheses

Agricultural production is inherently vulnerable, as the production process is highly dependent on natural conditions and relatively weak in resisting disasters and uncertainties. Therefore, risk management in agricultural production is particularly important. Agricultural insurance primarily reduces the burden of uncontrollable risks through risk transfer, that is, it functions by dispersing risks to enhance agricultural economic resilience [31]. From the perspective of institutional complementarity and adaptive learning, agricultural insurance in its initial promotion stage can be regarded as a newly embedded institutional arrangement. Due to the path dependence of agricultural production and management systems, the introduction of agricultural insurance contracts creates institutional friction with existing production practices, risk response patterns, and incomplete factor markets. This manifests specifically as high contract comprehension costs, complex claims procedures, insufficient trust among farmers, and conflicts with traditional risk management methods [32]. At this stage, the risk protection function of agricultural insurance is partially offset by high transaction costs and behavioral adaptation issues. It may even lead to short-term efficiency losses by altering farmers’ original production incentives, thereby weakening its role in enhancing economic resilience and forming the left half of the “U-shaped” curve. Moreover, existing theoretical analyses have also found that the moral hazard effect of agricultural insurance may reduce farmers’ investment in maintaining disaster prevention facilities or decrease preventive labor inputs such as pest control and drought resistance [33,34]. In the long run, this may prevent fundamental improvements in the vulnerability of the agricultural system and could even exacerbate future disaster losses.

With policy advancement and the passage of time, the development of agricultural insurance enters a dynamic process of bidirectional learning and adaptation. On the supply side, insurance institutions achieve supply-side learning through product iteration, claims process optimization, and actuarial data accumulation, thereby improving coverage suitability and service efficiency. On the demand side, farmers and new types of agricultural operators enhance their understanding, trust, and ability to utilize agricultural insurance tools through “learning by doing” and information diffusion, gradually transitioning from passive acceptance to active application. This bidirectional adaptation significantly reduces institutional friction costs, laying the foundation for the effective functioning of agricultural insurance. More crucially, the maturation of agricultural insurance is not an isolated process but one that gradually forms a synergistic network with complementary institutions. When the insurance system interacts and strengthens with institutional arrangements such as agricultural technology extension systems, rural credit markets, digital infrastructure, and farmers’ cooperatives, its function expands from mere risk compensation to a comprehensive mechanism that incentivizes technology adoption, alleviates credit rationing, and improves resource allocation efficiency. At this point, insurance not only stabilizes income but also enhances the adaptive and transformative capacity of the agricultural economy in responding to shocks by promoting systemic capital deepening and technological penetration.

As agricultural insurance matures, its inhibitory effect on resilience gradually diminishes, and its positive role enters a strengthening phase. On one hand, as farmers’ participation rates increase, insurance institutions’ “law of large numbers” advantage becomes evident. The systematic expansion of the risk pool not only reduces unit premium costs but also enhances overall underwriting capacity through risk dispersion effects [35]. In this process, insurance institutions can further strengthen ex-ante moral hazard prevention mechanisms and ex-post claims risk control capabilities [36]. Particularly important is the long-term accumulation of underwriting and claims data, which provides a dynamic optimization basis for actuarial pricing, effectively curbing adverse selection and forming a positive feedback loop [37]. On the other hand, when insurance coverage is sufficiently broad and claims are settled promptly, the uncertainty in farmers’ production and operations is significantly reduced [30]. This leads to a shift in risk preferences, making farmers more willing to invest in drought-resistant crops, ecological agriculture, or technical training, which directly enhances the recovery capacity and long-term adaptability of the agricultural industry. Furthermore, in the mature stage, agricultural insurance interacts more closely with policies such as credit, futures, and subsidies, forming a composite risk management tool that helps stabilize production expectations and withstand market fluctuations [38]. More importantly, a long-term sustainable insurance system not only meets individual risk transfer needs but also promotes the transformation of agricultural risk governance from dispersed individual decision-making to organized collective and coordinated responses through institutionalized risk-sharing arrangements [39]. The evolution of this governance model represents a deep institutional safeguard for enhancing the resilience of the agricultural economic system.

Based on this, the paper proposes

Hypothesis 1.

There is a nonlinear relationship between agricultural insurance and agricultural economic resilience.

Agricultural economic resilience, as the core capacity of the agricultural system to cope with external shocks, maintain production functions, and achieve sustainable development, essentially reflects a dynamic balance among resistance, adaptability, and transformative capacity. Enhancing resilience relies not only on the stability of production scale but also on the strengthening of intrinsic industrial efficiency and self-adaptive capabilities. Agricultural technological progress serves as a fundamental driver for improving agricultural development efficiency and acts as a key transmission pathway through which agricultural insurance influences agricultural economic resilience. On one hand, agricultural insurance reduces the potential risks for farmers in adopting new technologies through its risk-hedging mechanism. According to the theory of rational smallholders, farmers pursue expected utility maximization in production decisions. However, due to the higher costs and uncertainties associated with new technologies, farmers tend to stick to traditional methods to avoid risks in the absence of external risk-sharing mechanisms [40,41]. Agricultural insurance transforms potential individual losses from disasters into relatively fixed premium expenditures, thereby stabilizing farmers’ income expectations. This encourages them to more actively adopt advanced technologies such as resilient crop varieties and smart agricultural machinery, promoting the diffusion and application of agricultural technologies in practice. On the other hand, the loss compensation function of agricultural insurance provides essential financial support for the continuity of technology application. When disasters occur, insurance compensation can promptly cover production losses, preventing farmers from facing capital shortages for reinvestment. This ensures their ability to repair and upgrade technological elements such as irrigation facilities and new machinery, thereby maintaining the stability and continuity of technological investments. Moreover, agricultural technological progress exhibits significant spatial spillover effects, as the successful application of new technologies in one region can encourage neighboring areas to learn and adopt similar technologies or management models through demonstration effects, forming a virtuous cycle of regional technology diffusion [42]. Research indicates that agricultural technological progress can significantly enhance the stability of agricultural productivity, resource utilization efficiency, and post-disaster recovery capabilities, all of which are key dimensions of agricultural economic resilience.

Based on the aforementioned mechanisms, this paper proposes the following research hypothesis:

Hypothesis 2.

Agricultural insurance exerts a positive influence on agricultural economic resilience by promoting agricultural technological progress.

The impact of agricultural insurance on agricultural economic resilience may vary depending on the level of industrial structure. When the level of industrial structure is low, there are fewer high-value-added industries such as manufacturing and services. Underdeveloped sectors create numerous challenges for economic development, especially in rural areas. Specifically, China’s agriculture is dominated by small-scale, fragmented operations. Farmers have limited awareness of agricultural insurance, coupled with insufficient capital and weak risk management consciousness, making it difficult to form stable, long-term demand for agricultural insurance [43]. Furthermore, most farmers struggle to provide relevant agricultural data needed by insurers for risk assessment, increasing operational costs for insurance companies [44]. All these factors hinder the development of the agricultural insurance market, naturally limiting its role in enhancing agricultural economic resilience. As the level of industrial structure improves, positive interactions emerge among the industrial, service, and agricultural sectors. Through integration in technology, information, productivity, and services, agriculture transitions toward greater diversification, efficiency, technological advancement, and modernization. This development also broadens farmers’ perspectives, increasing their recognition of the benefits of agricultural insurance and thereby boosting demand. In addition, insurance companies can leverage modern technologies to disseminate knowledge about agricultural insurance, innovate insurance products, and reduce risk assessment costs, thereby better assisting farmers in managing agricultural risks. In other words, when the level of industrial structure surpasses a certain threshold, the role of agricultural insurance in enhancing agricultural economic resilience becomes more pronounced.

Based on this, the paper proposes

Hypothesis 3.

The impact of agricultural insurance on agricultural economic resilience exhibits a threshold effect related to the level of industrial structure.

As the industrial structure improves, the positive effect of agricultural insurance on agricultural economic resilience strengthens.

China’s different regions vary significantly in terms of economic development, factor endowments, infrastructure, and policy support—all of which influence the role of agricultural insurance in enhancing agricultural economic resilience. For example, in terms of economic development, the eastern coastal regions are the most economically advanced in China, featuring strong industrial advantages and well-developed market mechanisms. Compared to the underdeveloped, resource-scarce, and economically challenged central and western regions, the agricultural economy in the east is relatively more resilient. In terms of factor endowments and infrastructure, agricultural production and natural resources are more abundant in southern regions than in the north, making the south more suitable for farming and inherently more resilient. The eastern region also enjoys more complete infrastructure in transportation, energy, water conservancy, and communications, whereas infrastructure in central and western regions lags due to historical and geographical factors, potentially weakening agricultural resilience. Additionally, the focus and intensity of policy support vary across different regions. For example, in the developed eastern regions, policies are more inclined toward promoting high-level agricultural openness and industrial integration. Initiatives such as establishing the Zhejiang Inclusive Financial Services Rural Revitalization Reform Pilot Zone and advancing the development of high-tech agriculture and cross-border supply chains in the Guangdong-Hong Kong-Macao Greater Bay Area have endowed the agricultural sector in the eastern regions with stronger market adaptability and competitiveness. In the major grain-producing areas of central China, policies prioritize ensuring national food security and enhancing comprehensive agricultural production capacity. Measures like implementing specialized support policies for grain production functional zones in provinces such as Henan and Heilongjiang strengthen the stability and supply guarantee functions of agricultural production. In the ecologically fragile and less developed western regions, policies emphasize ecological conservation and the development of characteristic industries. In the northeastern region, efforts are focused on consolidating the advantages of agricultural scale and mechanization while extending the industrial chain. Examples include supporting Jilin and Liaoning in implementing black soil protection projects and developing agricultural product deep-processing industrial parks. These region-specific policies, aligned with local resource endowments and developmental stages, collectively shape the varying capacities of agricultural economies to respond to risks.

Based on this, the paper proposes

Hypothesis 4.

The impact of agricultural insurance on agricultural economic resilience exhibits nonlinear regional heterogeneity.

4. Methods and Data Description

4.1. Variable Selection

This study utilizes panel data from 30 provinces in China from 2011 to 2023. Tibet was excluded from the analysis due to a high number of missing values. This time period fully captures the policy evolution of agricultural insurance from wide coverage to high-quality development, making it well-suited to examine its long-term effects under changing policy environments. The data are primarily sourced from the China Rural Statistical Yearbook, China Grain Yearbook, China State Farms Statistical Yearbook, China Statistical Yearbook for Regional Economy, China Statistical Yearbook, and China Labor Statistical Yearbook. A small number of missing values—specifically, 7 for agricultural disaster-affected area, 12 for rural households’ fixed asset investment in agriculture, and 4 for urbanization rate, totaling 23 missing data points—were imputed using linear interpolation.

4.1.1. Dependent Variable

The dependent variable in this study is agricultural economic resilience (Resi). In constructing the evaluation index system for agricultural economic resilience, based on the connotation of agricultural economic resilience and drawing on previous studies [14,15,16,17,21], this paper establishes the criterion layer from three dimensions: resistance, adaptability, and reconstruction capacity. Resistance primarily refers to the ability of the agricultural system to withstand shocks after being affected by uncertain factors. It includes three element layers: economic foundation, production conditions, and Informa ionization level, and encompasses nine indicators, such as the gross output value of agriculture, forestry, animal husbandry, and fishery and per capita disposable income of rural residents. Adaptability mainly reflects the agricultural system’s ability to adapt to damage after experiencing shocks. It consists of the adaptability element layer, and includes three indicators, such as the number of rural doctors and health workers, per capita living consumption expenditure of rural households, and grain output. Reconstruction capacity refers to the ability of the agricultural system to recover to its original state—or even surpass it—through adjustments after a shock. It includes three element layers: financial support, technological progress, and ecological governance, and contains seven indicators, such as agriculture-related loans and government expenditure on agriculture, forestry, and water affairs. The specific indicators are shown in

Table 1. Comprehensive Evaluation Index System for Agricultural Economic Resilience.

First-Level Indicators	Second-Level Indicators	Third-Level Indicators	Direction	Weight
resistance (0.447)	economic foundation (0.100)	gross output value of agriculture, forestry, animal husbandry, and fishery	+	0.051
		rural residents’ disposable income	+	0.049
	production conditions (0.263)	total power of agricultural machinery	+	0.060
		rural electricity consumption	+	0.121
		sown area of crops	+	0.046
		application of chemical fertilizers in agriculture	−	0.010
		use of pesticides	−	0.018
		use of agricultural plastic film	−	0.008
	level of informatization (0.084)	rural broadband internet subscribers	−	0.084
adaptability (0.175)	capacity for adaptation (0.175)	rural doctors and health personnel	−	0.061
		per capita living consumption expenditure of rural households	−	0.055
		grain output	−	0.059
reconstruction capacity (0.373)	financial support (0.171)	agriculture-related loans	−	0.080
		government expenditure on agriculture, forestry, and water affairs	−	0.035
	technological progress (0.123)	rural residents’ personal fixed asset investment in agriculture	−	0.056
		effective irrigated area	−	0.056
		number of people in the grain industry with national vocational qualification certificates	−	0.067
	ecological environment (0.079)	total afforestation area	−	0.053
		agricultural disaster-affected area	−	0.026

.

Furthermore, Table 1 also presents the weights of the indicators for agricultural economic resilience calculated using the entropy method. Among the first-level indicators, the order of weights is resistance (0.447) > reconstruction capacity (0.373) > adaptability (0.175), indicating that in the development of agricultural economic resilience, the capacity to prevent and withstand shocks should take precedence over post-disaster recovery and long-term adaptation. This aligns with the seasonal and vulnerable nature of agricultural production. Among the second-level indicators, the top three in terms of weight are production conditions, adaptability, financial support, and technological progress, with weights of 0.263, 0.175, 0.171, and 0.123, respectively. This suggests that production conditions serve as the core support for enhancing agricultural economic resilience, with their importance significantly higher than that of other factors. Meanwhile, the adaptive capacity of the agricultural system and the level of financial security are also crucial for improving resilience. While technological progress is important, its relative role at the current stage is slightly less prominent than the top three factors.

4.1.2. Core Explanatory Variable

The core explanatory variable in this study is agricultural insurance (Insu). The premium income of agricultural insurance can directly measure the development of agricultural insurance, reflecting both the scope and amount of coverage it provides [29], as well as indicating, to a certain extent, the level of agricultural insurance development in each province. Therefore, this study uses the share of agricultural insurance premium income in the value-added of the primary industry as a proxy variable for agricultural insurance.

4.1.3. Mechanism Variable

The mechanism variable in this study is agricultural technological progress (Tech), which encompasses both advancements in hardware and software technologies. Gong B.L. et al. argue that agricultural total factor productivity measures the portion of agricultural output growth that cannot be explained by all input factors, representing broad agricultural technological progress in essence [45]. Therefore, this paper follows the approach of Tian Y. and Yin M.H. and uses agricultural total factor productivity to measure agricultural technological progress [46]. Specifically, the DEA-Malmquist index method is employed for calculation. In measuring agricultural technological progress, agricultural output is represented by the gross output value of agriculture, forestry, animal husbandry, and fishery. Labor input is measured by the number of employees in the primary industry, land input by the total sown area of crops, and energy input by rural electricity consumption. Capital input includes the total power of agricultural machinery, chemical fertilizers, pesticides, agricultural films, and diesel. Data related to the measurement of agricultural technological progress are sourced from the National Bureau of Statistics of China and the statistical yearbooks of various provinces and cities. As the calculation method of the DEA-Malmquist index is well-established, it is not detailed here.

4.1.4. Control Variables

Agricultural economic resilience is not only influenced by agricultural insurance but also by other factors. Therefore, this paper selects the regional economic level, fixed asset investment, crop sown area, urban-rural income ratio, agricultural disaster ratio, and urbanization as control variables. Specifically, the regional economic level is measured by per capita GDP (RGDP). Fixed asset investment refers to the fixed asset investment by households in agriculture (Invest). Crop sown area is measured by the grain sown area (BoZ), and a logarithmic transformation is applied to eliminate the effects of different scales. The urban-rural income ratio (Income) is measured by the per capita disposable income of urban residents divided by that of rural residents. The agricultural disaster ratio (Disaster) is measured by the proportion of crop disaster-affected area to crop disaster-stricken area. Urbanization (Urban) is measured by the urbanization rate.

4.1.5. Threshold Variable

The level of industrial structure reflects the adaptability and reconstruction capacity of a region’s agriculture in the face of risks and challenges. The development of the secondary and tertiary industries determines the region’s level of economic development and also constrains agricultural development to a certain extent. Drawing on the research of Yang Z. S. et al. [47], this paper uses the ratio of the combined output value of the secondary and tertiary industries to that of the primary industry to represent the level of industrial structure (Struc).

4.1.6. Data Analysis

(1)

Descriptive statistics

To analyze the characteristics of each variable more intuitively, the descriptive statistics results are presented in

Table 2. Descriptive Statistics of Variables.

VarName	Obs	Mean	SD	Min	Max
Resi	390	0.260	0.126	0.064	0.581
Insu	390	0.575	0.580	0.021	3.781
Insu2	390	0.667	1.605636	0.000	14.296
GPE	390	0.651	0.142	0.360	0.970
Struc	390	26.674	58.297	2.827	372.666
RGDP	390	10.841	0.436	9.706	12.013
Invest	390	3.667	1.487	2.302	5.754
BoZ	390	8.191	0.142	0.360	9.610
Income	390	2.635	0.432	1.850	3.980
Disaster	390	0.033	0.0355004	0.000	0.264
Urban	390	59.006	12.218	35.030	89.600

. The results show that the mean value of agricultural economic resilience (Resi) is 0.260, indicating that China’s agricultural economic resilience still has considerable room for improvement in recent years. The difference between the maximum and minimum values is 0.517, suggesting significant regional imbalance in agricultural economic resilience across different areas. Meanwhile, the mean value of agricultural insurance premium income (Insu) is 6.857, with a difference of 5.971 between the maximum and minimum values, reflecting substantial disparities in the development level of agricultural insurance across regions.

(2)

Test of Multicollinearity

A multicollinearity test was conducted on the variables using Stata 18 MP software, and the results are presented in

Table 3. Test Results of Multicollinearity.

VarName	VIF	1/VIF
Insu	1.840	0.544
RGDP	6.630	0.151
Invest	4.630	0.216
BoZ	5.530	0.181
Income	2.230	0.449
Disaster	1.600	0.627
Urban	9.990	0.100
Mean VIF	4.640	/

. As shown, the VIF values for all variables are below the threshold of 10, indicating that there is no severe multicollinearity issue.

4.2. Methods

4.2.1. Entropy Value Method

The entropy weight method can avoid the influence of subjective factors on the determination of weights, thereby enhancing the credibility of the indicator weights. Therefore, the entropy weight method is employed to conduct a comprehensive evaluation and analysis of agricultural resilience. The specific steps are as follows:

Select M provinces and N indicators, where X_ij represents the value of the j-th indicator for the i-th province (i = 1, 2, …, M; j = 1, 2, …, N).

Step 1: Data Standardization

Standardize the original data to eliminate the influence of different units of measurement (dimensionality) and the differing natures of positive and negative indicators on the comprehensive evaluation results.

Formula for Positive Indicators:

$$X\prime ={\displaystyle \frac{X_{ij}-\mathit{min}\left(X_{ij}\right)}{\mathit{max}\left(X_{ij}\right)-\mathit{min}\left(X_{ij}\right)}}$$

(1)

Formula for Negative Indicators:

$$X\prime ={\displaystyle \frac{\mathit{max}\left(X_{ij}\right)-X_{ij}}{\mathit{max}\left(X_{ij}\right)-\mathit{min}\left(X_{ij}\right)}}$$

(2)

Step 2: Calculate the proportion of the value of the j-th indicator of the i-th province relative to the total value of that indicator:

$$P_{ij}={\displaystyle \frac{X_{ij}}{{\sum }_{i=1}^M{X}_{ij}}}$$

(3)

Step 3: Compute the entropy value for the j-th indicator:

$$e_j={\displaystyle \frac{1}{\mathit{ln}M}}\times {\sum }_{i=1}^M\left(P_{ij}\times \mathit{ln}{P}_{ij}\right)$$

(4)

Step 4: Compute the difference coefficient for the j-th indicator:

$$d_j=1-e_j$$

(5)

Step 5: Compute the weight of each evaluation indicator:

$$W_j={\displaystyle \frac{d_j}{{\sum }_{j=1}^N{d}_j}}$$

(6)

Step 6: Compute the composite score for each evaluation unit based on the indicator weights:

$$Z_i={\sum }_{j=1}^N{W}_j\times X_{ij}$$

(7)

4.2.2. Panel Data Model

To test Hypothesis 1 that agricultural insurance has a nonlinear impact on agricultural economic resilience, the following model is constructed:

$${Resi}_{it}={\beta }_0+{\beta }_1{Insur}_{it}+{\beta }_2{Insur}_{it}^2+{\beta }_3{Controls}_{it}+{\lambda }_i+{\mu }_t+{\epsilon }_{it}$$

(8)

where $Resi$ represents agricultural economic resilience, $Insur$ denotes agricultural insurance, $Controls$ refer to a set of control variables, i and t indicate province and year, respectively, ${\lambda }_i$ is the province fixed effect, ${\mu }_t$ is the year fixed effect, and ${\epsilon }_{it}$ is the random disturbance term. To test whether there is a nonlinear relationship between agricultural insurance and agricultural economic resilience, the quadratic term of agricultural insurance is incorporated into the regression model, denoted as ${Insur}^2$. ${\beta }_1$, ${\beta }_2$ and ${\beta }_3$ represent the coefficients of agricultural insurance, the quadratic term of agricultural insurance, and the control variables, respectively.

4.2.3. Mediation Effect Model

Agricultural insurance mainly influences the resilience of the agricultural economy by promoting agricultural technological progress. This scenario reflects the mechanism whereby a mediating variable plays a role between the independent variable and the dependent variable. Therefore, a mediation effect model is constructed.

$${Resi}_{it}={\beta }_0+{\beta }_1{Insur}_{it}+{\beta }_2{Insur}_{it}^2+{\beta }_3{Controls}_{it}+{\lambda }_i+{\mu }_t+{\epsilon }_{it}$$

(9)

$${Tech}_{it}={\beta }_0^{\prime }+{\beta }_1^{\prime }{Insur}_{it}+{\beta }_2^{\prime }{Insur}_{it}^2+{\beta }_3^{\prime }{Controls}_{it}+{\lambda }_i^{\prime }+{\mu }_t^{\prime }+{\epsilon }_{it}^{\prime }$$

(10)

$${Resi}_{it}={\beta }_0^{″}+{\beta }_1^{″}{Insur}_{it}+{\beta }_2^{″}{Insur}_{it}^2+{\beta }_3^{″}{Tech}_{it}+{\beta }_4^{″}{Controls}_{it}+{\lambda }_i^{″}+{\mu }_t^{″}+{\epsilon }_{it}^{″}$$

(11)

where $Tech$ represents Agricultural technological progress (mediating variable), ${\beta }_1^{\prime }$ and ${\beta }_2^{\prime }$ represents the effect of agricultural insurance on the mediating variable, and ${\beta }_1^{″}, {\beta }_2^{″}$ and ${\beta }_3^{″}$ represent the effects of agricultural insurance and the mediating variable on agricultural economic resilience, respectively.

4.2.4. Threshold Effect Model

To verify Research Hypothesis 3, a threshold model was constructed to analyze the role of industrial structure level in the impact of agricultural insurance on agricultural economic resilience. Assuming there are k threshold values, the impact of agricultural insurance on agricultural economic resilience is divided into k + 1 segments.

$$Res{i}_{it}=\alpha +{\delta }_{1}Insu{r}_{it}\times I\left(Th<{\theta }_1\right)+{\delta }_{2}Insu{r}_{it}\times I\left({\theta }_1<Th<{\theta }_2\right)+\dots +{\delta }_{k+1}Insu{r}_{it}\times I\left(Th>{\theta }_k\right)+\gamma X_{it}+{\mu }_i+{\lambda }_i+{\epsilon }_{it}$$

(12)

where θ₁, θ₂, …, θₖ are the threshold values to be estimated, δₖ represents the effect of agricultural insurance on agricultural economic resilience after the level of industrial structure and the scale of agricultural development cross the (k − 1)-th threshold, X denotes the control variables, uᵢ represents the province fixed effect, λₜ represents the year fixed effect, and εᵢₜ is the random disturbance term.

5. Analysis of Research Results

5.1. Model Selection

When analyzing panel data, it is necessary to determine whether to adopt a fixed-effects model, a random-effects model, or a pooled estimation model. To identify the optimal model, an F-test, LM test, and Hausman test were conducted. The results of the F-test showed that F(8, 352) = 135.450, Prob > F = 0.000, indicating that the fixed-effects model is superior to the pooled estimation model. The results of the LM test showed that chi2(1) = 23.24, Prob > chi2 = 0.000, indicating that the random-effects model is superior to the pooled estimation model. The results of the Hausman test showed that chi2(8) = 23.050, Prob > chi2 = 0.003, indicating that the fixed-effects model is superior to the random-effects model. Therefore, the fixed-effects model should be selected.

Furthermore, additional tests for heteroskedasticity and serial correlation were conducted. The results of the heteroskedasticity test show: chi2(30) = 980.320, Prob > chi2 = 0.000, rejecting the null hypothesis at the 1% significance level, indicating the presence of heteroskedasticity. The serial correlation test results show: F(1, 29) = 9.848, Prob > F = 0.004, rejecting the null hypothesis at the 1% significance level, indicating the presence of first-order serial correlation. Based on the above test results, this paper employs a fixed-effects model with clustered robust standard errors for estimation. The main reasons for selecting this model are as follows: First, clustered robust standard errors can simultaneously address both heteroskedasticity and serial correlation issues, without requiring prior assumptions about the specific forms of heteroskedasticity and serial correlation, thus providing good robustness. Second, the fixed-effects model can effectively control for individual heterogeneity and avoid omitted variable bias. Third, clustered robust standard errors exhibit favorable statistical properties in large samples, delivering consistent and efficient standard error estimates to ensure the reliability of statistical inference.

5.2. Regression Result Analysis

To examine the relationship between agricultural insurance and agricultural economic resilience, this paper presents regression results both without and with control variables, as shown in

Table 4. Regression results.

Variable	(1)	(2)
Insu	−0.0804 ***	−0.0600 ***
	(0.0168)	(0.0141)
Insu2	0.0106 ***	0.0081 **
	(0.0032)	(0.0032)
RGDP		0.0697 ***
		(0.0139)
Invest		0.0095 ***
		(0.0033)
BoZ		0.0518 ***
		(0.0141)
Income		0.0358 **
		(0.0169)
Disaster		−0.0760
		(0.0655)
Urban		0.0001
		(0.0012)
Year	YES	YES
Pro	YES	YES
_cons	0.2988 ***	−1.0227 ***
	(0.0076)	(0.2475)
Obs	390	390
R-squared	0.9865	0.9895

** and *** indicate statistical significance at the 5%, and 1% levels, respectively.

. In column (1), it can be observed that the coefficient of the linear term of agricultural insurance is significantly negative at the 1% significance level, while the coefficient of the quadratic term is significantly positive. These preliminary estimation results suggest a significant U-shaped relationship between agricultural insurance and agricultural economic resilience. In column (2), after incorporating control variables, the linear term of agricultural insurance remains significantly negative at the 1% level, and the quadratic term is significantly positive at the 5% level. The results further indicate a significant U-shaped relationship between agricultural insurance and agricultural economic resilience. Additionally, it is worth noting that the R-squared values in columns (1) and (2) are 0.9865 and 0.9895. This is because the core models include both province and year fixed effects. The R-squared values primarily reflect the model’s explanatory power for within-province variation. Since the fixed effects capture all time-invariant province characteristics and common annual shocks across provinces, their explanatory power is strong, leading to typically high R-squared values. Therefore, the high R-squared values in this study should be interpreted as evidence that the model effectively controls for these layered confounding structures, rather than raising concerns about overfitting. Agricultural insurance is a compensatory form of insurance whose main function is to prevent major losses to agricultural production caused by disasters, mitigating farmers’ exposure to risks from weather and other force majeure events through risk transfer. The “U-shaped” effect between agricultural insurance and agricultural resilience implies that, under certain conditions, agricultural insurance can alleviate disaster impacts and enhance agricultural resilience. However, when the amount of agricultural insurance coverage is excessively high or policies contain discriminatory or unfair elements, it may produce counterproductive effects. Thus, only when agricultural insurance develops to a certain extent can it improve agricultural economic resilience.

Furthermore, a U-shape test was conducted to examine the nonlinear relationship between agricultural insurance and the dependent variable. According to the U-shaped relationship test developed by Lind and Mehlum [48], the null hypothesis is the presence of a monotonic or an inverted U-shaped relationship, while the alternative hypothesis is the presence of a U-shaped relationship. The research results show that the overall test yields t = 2.37, p = 0.012. At the lower limit of the data range, the slope is negative and significant (p = 0.006); at the upper limit of the range, the slope is positive and significant (p = 0.012), and the inflection point is located around 2.05, respectively, with an inflection point around 2.05. Therefore, there is a significant U-shaped relationship between agricultural insurance and agricultural economic resilience. Thus, Hypothesis 1 is validated.

5.3. Endogeneity and Robustness Test

Firstly, Conduct an endogeneity test. A dynamic panel system GMM model was employed by introducing the one-period lag of the dependent variable as an instrumental variable. This effectively controls for endogeneity bias caused by the bidirectional causality and temporal inertia between agricultural insurance and agricultural economic resilience. The results, as shown in

Table 5. Results of endogeneity and robustness tests.

Variable	SYS-GMM	Replace Variables	Change the Model	2011–2018	2019–2023
L.Resi	0.6720 ***
	(0.0200)
Rinsur	−0.0248 **	−0.0214 **	−0.0212 **	−0.0456 ***	−0.0730 *
	(0.0098)	(0.0099)	(0.0107)	(0.0142)	(0.0384)
Rinsur2	0.0047 **	0.0028 *	0.0056 *	0.0043 ***	0.0054 **
	(0.0024)	(0.0015)	(0.0029)	(0.0014)	(0.0027)
RGDP	0.0287 ***	0.0533 ***	0.1242 ***	0.0472 *	0.0787 ***
	(0.0043)	(0.0135)	(0.0117)	(0.0264)	(0.0152)
Invest	0.0057 ***	0.0122 ***	0.0097 ***	0.0116 ***	0.0123 ***
	(0.0009)	(0.0030)	(0.0025)	(0.0025)	(0.0037)
BoZhong	0.0373 ***	0.0713 ***	0.0808 ***	0.0177	0.0447
	(0.0052)	(0.0186)	(0.0087)	(0.0198)	(0.0341)
Income	0.0074 **	0.0393 **	0.0313 ***	0.0062	0.0842 *
	(0.0029)	(0.0169)	(0.0087)	(0.0079)	(0.0463)
Disaster	0.0102	−0.0066	−0.0159	−0.1439 ***	−0.0711
	(0.0193)	(0.0643)	(0.0458)	(0.0433)	(0.0710)
Urban	0.0019 ***	0.0027 *	0.0030 ***	−0.0012	0.0026
	(0.0005)	(0.0013)	(0.0007)	(0.0021)	(0.0024)
Constant	−0.6664 ***	−1.2954 ***	−2.0364 ***	−0.2909	−1.1221 ***
	(0.0580)	(0.2820)	(0.1474)	(0.2864)	(0.3693)
Obs	360	390	390	240	150
R-squared		0.9900	0.7899	0.9955	0.9928
Chowtest				3.1700 ***
AR(1)	0.0820
AR(2)	0.1202
Sargan	0.9934

*, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively.

, indicate that the “U-shaped” relationship between agricultural insurance and agricultural economic resilience remains statistically significant. Further model validity tests reveal that the AR(1) test is significant (p = 0.0820), confirming the presence of first-order serial correlation in the disturbance term, while the AR(2) test is insignificant (p = 0.1202), ruling out second-order autocorrelation and supporting the rationality of the lag order selection for the instrumental variables. The Sargan test yields a p-value of 0.9934, indicating no over-identification of the instrumental variables. In summary, the “U-shaped” relationship between agricultural insurance and agricultural economic resilience has passed rigorous endogeneity diagnostic criteria under the dynamic GMM method, demonstrating strong robustness for hypothesis H1.

Secondly, Conduct a robustness test. On the one hand, the method of replacing the explanatory variable is employed for robustness testing. Specifically, agricultural insurance is measured by the proportion of agricultural insurance claim payments to the agricultural labor force. The results, as shown in column (2) of Table 5, indicate that both the linear and quadratic terms of agricultural insurance are significant at the 5% and 10% levels, respectively. This suggests that the nonlinear effect of agricultural insurance on agricultural economic resilience remains reliable. On the other hand, while the baseline regression applies a fixed-effects model, a random-effects panel model is further used for robustness testing. The results, presented in column (3) of Table 5, also show that the linear and quadratic terms of agricultural insurance are significant at the 5% and 10% levels, respectively, consistent with the previous findings. Furthermore, recognizing the potential impact of the 2019 Guidelines on Accelerating the High-Quality Development of Agricultural Insurance, we conducted split-sample regressions with 2019 as the cutoff, specifically for the periods 2011–2018 and 2019–2023. The results show that the signs and significance of the core explanatory variables remain largely consistent across the two subsamples. However, the absolute values of the estimated coefficients for both the linear and quadratic terms of agricultural insurance are larger in the 2019–2023 period compared to 2011–2018. This suggests that the 2019 policy altered the intensity of the impact of agricultural insurance on agricultural economic resilience, without changing the fundamental conclusion. Additionally, we performed a Chow test to examine coefficient differences between the groups, confirming the reliability of the results. All the above regression results confirm the robustness of the baseline regression.

5.4. Mechanism Analysis

Table 6. Results of mechanism analysis.

Variable	Resi	Tech	Resi
Insu	0.0189 **	0.1503 ***	0.0153 *
	(0.0091)	(0.0568)	(0.0092)
Tech			0.0244 ***
			(0.0094)
RGDP	0.1953 ***	0.0195	0.1949 ***
	(0.0211)	(0.1309)	(0.0209)
Invest	−0.0112 **	−0.0200	−0.0107 **
	(0.00549)	(0.0336)	(0.0053)
BoZ	0.1207 ***	−0.0509	0.1219 ***
	(0.007309)	(0.0454)	(0.0073)
Income	−0.0464 ***	0.0819	−0.0484 ***
	(0.0121)	(0.0751)	(0.0119)
Disaster	−0.6638 ***	1.5064 *	−0.7005 ***
	(0.1422)	(0.8837)	(0.1415)
Urban	−0.0038 ***	−0.0325 ***	−0.0030 ***
	(0.0009)	(0.0058)	(0.0010)
Year	YES	YES	YES
Pro	YES	YES	YES
_cons	−2.7781	40.5571 **	−3.7661
	(2.8735)	(17.8626)	(2.8704)
Obs	390	390	390
R-squared	0.8088	0.4355	0.8132

*, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively.

reports the mediating role of agricultural technological progress in the process of agricultural insurance promoting agricultural economic resilience. The results show that agricultural insurance significantly enhances agricultural technological progress. Furthermore, after introducing the variable of agricultural technological progress, both agricultural insurance and agricultural technological progress significantly and positively promote agricultural economic resilience. However, compared with the regression results in the first column, the coefficient of agricultural insurance’s impact on agricultural economic resilience decreases after including the agricultural technological progress variable, indicating the presence of a mediating effect of agricultural technological progress between the agricultural insurance and agricultural economic resilience.

Further, a Sobel test was conducted for the mediating effect, and the results are shown in

Table 7. Results of Sobel Test.

	Est	Std_Err	z	p > |z|
Sobel	0.004	0.002	1.856	0.063
Aroian	0.004	0.002	1.792	0.073
Goodman	0.004	0.002	1.927	0.054
a_coefficien	0.150	0.057	2.646	0.008
b_coefficient	0.024	0.009	2.604	0.009
Indirect_effect_aXb	0.004	0.002	1.856	0.063
Direct_effect_c’	0.015	0.009	1.667	0.096
Total_effect_c	0.019	0.009	2.071	0.038
Proportion of total effect that is mediated:		0.194
Ratio of indirect to direct effect:		0.240
Ratio of total to direct effect:		1.240

. It can be seen that the z-value for the Sobel test is 1.856, with a p-value of 0.063, which is not statistically significant at the 0.05 level but is marginally significant at the 10% level. The p-values for the Aroian and Goodman tests are 0.073 and 0.054, respectively, also falling at the marginally significant 10% level. The results of the path coefficients indicate that the impact of agricultural insurance on agricultural technological progress is significant (p = 0.008). After controlling for the independent variable, the impact of agricultural technological progress on agricultural economic resilience is also significant (p = 0.009). However, after introducing agricultural technological progress, the direct effect of agricultural insurance on agricultural economic resilience is only marginally significant at the 10% level (p = 0.096), while the total effect of agricultural insurance on agricultural economic resilience is significant at the 5% level (p = 0.038). The effect size results show that the proportion of the mediating effect to the total effect is 19.4%, and the ratio of the indirect effect to the direct effect is 0.240. Overall, although the Sobel test did not reach significance at the 5% level, the pattern of results—where the direct effect is marginally significant—still suggests the potential existence of a mediating pathway. Therefore, a Bootstrap method was further employed for verification.

Table 8. Results of Bootstrap Test.

	Observed	Bootstrap
	Coefficient	Bias	Std. Err.	[95% Conf. Interval]
ind_eff	0.0037	0.0003	0.0021	[−0.0005, 0.0078]	N
				[0.0008, 0.0086]	P
				[0.0008, 0.0088]	BC
				[0.0007, 0.0085]	BCa
dir_eff	0.0153	−0.0002	0.0099	[−0.0042, 0.0347]	N
				[−0.0036, 0.0350]	P
				[−0.0029, 0.0356]	BC
				[−0.0029,0.0355]	BCa

reports the supplementary test using 1000 Bootstrap samples to determine whether the mediating effect holds. It can be seen that the coefficient for the indirect effect (ind_eff) generated by agricultural technological progress is 0.0037. The Bootstrap 95% confidence interval (BCa method) is [0.0007, 0.0085]. Since this interval does not include 0, it indicates that the indirect effect is statistically significant, meaning the mediating effect is established. After controlling for the mediating variable, the direct effect (dir_eff) of agricultural insurance on agricultural economic resilience is 0.0153, and its Bootstrap 95% confidence interval (BCa method) is [−0.0029, 0.0355]. This interval includes 0, indicating that the direct effect is not significant. Combining this with the previously analyzed result that the total effect of agricultural insurance on agricultural economic resilience is significant, it can be concluded that the mediating variable plays a complete and significant mediating role between the independent and dependent variables. That is, the impact of agricultural insurance on agricultural economic resilience is realized through the pathway of agricultural technological progress. Therefore, Hypothesis 2 of this study is confirmed.

5.5. Threshold Effect Analysis

To test the threshold effect, the level of industrial structure (Struc) is used as the threshold variable, and single-threshold, double-threshold, and triple-threshold tests are conducted. Using the bootstrap method with 500 repeated resamples, the F-statistics and corresponding p-values are obtained, and the final threshold effect test results are presented in

Table 9. Results of threshold effect analysis.

Threshold Variable	Model	Threshold Value	F-Statistic	p-Value	Number of Bootstrap Replications
Struc	Single threshold	7.108	24.910	0.080 *	500
	Double threshold	20.592	27.920	0.052 *	500
	Triple threshold	25.767	12.590	0.588	500

* indicates statistical significance at the 10% level.

. The results show that the single threshold of the level of industrial structure passes the significance test at the 1% level, the double threshold passes at the 10% level, and the triple threshold is not significant.

According to the panel threshold regression model test results, this paper conducts a double threshold regression for the level of industrial structure (Struc), and the regression results are shown in Table 7. The two threshold values for the industrial structure level (Struc) are 7.108 and 20.592, respectively. When Struc < 7.108, the regression coefficient of agricultural insurance on agricultural economic resilience is insignificant. At this stage, the industrial structure is relatively primary, characterized by a singular agricultural industry, short industrial chains, and low added value. This results in insufficient demand for agricultural insurance, making it difficult for the financial security function of agricultural insurance to translate into productivity and risk resilience, thereby limiting its role in enhancing economic resilience. When 7.108 < Struc < 20.592, the regression coefficient of agricultural insurance on agricultural economic resilience is 0.005, which is significant at the 10% level. At this stage, the industrial structure enters an optimization phase, with preliminary integration between agriculture, industry, and services. However, it also faces more complex market and natural risks. Insurance begins to serve as a necessary risk management tool, gradually demonstrating its stabilizing effect on agricultural economic resilience. When Struc > 20.592, the regression coefficient of agricultural insurance on agricultural economic resilience is 0.009, which is significant at the 1% level. At this stage, the industrial structure gradually enters an advanced phase, characterized by deep integration of primary, secondary, and tertiary industries in rural areas, along with continuous enhancement of the value chain. However, as the agricultural industrial system becomes more complex, localized disasters or price fluctuations are more easily transmitted through the industrial chain, potentially triggering systemic risks. At this point, the risk dispersion and loss compensation functions of agricultural insurance become particularly critical. Its role is not only reflected in the direct protection of production activities but also manifests through multiple transmission mechanisms, forming a significant multiplier effect, thereby further enhancing its impact on agricultural economic resilience. Therefore, as the level of industrial structure improves, the positive effect of agricultural insurance on agricultural economic resilience becomes more pronounced. Based on the above analysis, Hypothesis 3 is validated (

Table 10. Results of threshold regression.

Variable	Resi
Struc < 7.108	0.030 ***
	(0.002)
7.108 < Struc < 20.592	0.037 ***
	(0.002)
Struc > 20.592	0.042 ***
	(0.003)
RGDP	0.091 ***
	(0.009)
Invest	0.008 ***
	(0.002)
BoZ	0.044 ***
	(0.010)
Income	0.029 ***
	(0.008)
Disaster	0.019
	(0.041)
Urban	0.003 ***
	(0.000)
F-statistic	128.620 ***
Obs	390
R-squared	0.830

*** indicates statistical significance at the 1% level.

).

5.6. Heterogeneity Analysis

The promoting effect of agricultural insurance on agricultural economic resilience may be influenced by economic–geographic location. From the perspective of economic–geographic location, there are substantial differences among China’s regions in terms of economic development, factor endowments, and infrastructure. Following the official classification, the sample is divided into eastern region, central region, western region, and northeastern region for grouped regressions, and the results are reported in

Table 11. Results of heterogeneity analysis.

Variable	Eastern Region	Central Region	Western Region	Northeastern Region
Insur	−0.0831 ***	−0.2052 ***	−0.0404 **	−0.3636 ***
	(0.0266)	(0.0496)	(0.0173)	(0.1097)
Insur2	0.0054 **	0.0148 ***	0.0028	0.0251 **
	(0.0022)	(0.0036)	(0.0017)	(0.0083)
RGDP	0.0843 ***	0.1332 ***	0.0865 ***	−0.0637
	(0.0193)	(0.0347)	(0.0198)	(0.0464)
Invest	0.0076 **	0.0149 ***	0.0154 ***	0.0253 ***
	(0.0032)	(0.0035)	(0.0056)	(0.0068)
BoZhong	0.0734 ***	−0.0078	0.1231 ***	0.3402 **
	(0.0155)	(0.1063)	(0.0276)	(0.1183)
Income	0.0187	−0.0289	−0.0069	0.0667 **
	(0.0171)	(0.0218)	(0.0156)	(0.0251)
Disaster	−0.0913	−0.0940	−0.1894 **	0.0009
	(0.1182)	(0.0999)	(0.0751)	(0.0663)
Urban	0.0012	−0.0004	−0.0016	0.0029
	(0.0012)	(0.0023)	(0.0024)	(0.0083)
Constant	−1.0532 ***	−0.2992	−1.5340 ***	−1.1849
	(0.3767)	(0.8147)	(0.3831)	(0.9938)
Year	YES	YES	YES	YES
Pro	YES	YES	YES	YES
Obs	130	78	143	39
R-squared	0.9963	0.9926	0.9781	0.9952

** and *** indicate statistical significance at the 5% and 1% levels, respectively.

. The results show that, at least at the 5% significance level, the linear term coefficients of agricultural insurance for the Eastern, Central, Western, and Northeastern regions are negative, while the quadratic term coefficients for the Eastern, Central, and Northeastern regions are positive. Only the quadratic term coefficient for the Western region is not significant. This indicates that in the Eastern, Central, and Northeastern regions, the impact of agricultural insurance on agricultural economic resilience exhibits a significant “U-shaped” nonlinear relationship—first inhibiting and then promoting resilience—while this relationship is relatively insignificant in the western region. In regions where the “U-shaped” relationship is significant, most provinces are traditional major agricultural hubs in China, with relatively well-developed agricultural insurance systems, robust technology extension systems, and sound agricultural infrastructure. This enables insurance to act as a lever and catalyst, synergizing with factors such as credit and advanced technologies to collectively enhance the modernization and resilience of the agricultural system. In the Western region, the “U-shaped” relationship is not evident, which may be attributed to constraints in infrastructure and market development. The relatively weak transportation, communication, and financial infrastructure, along with underdeveloped market systems, hinder the effective transmission of insurance signals to production and investment decisions. Moreover, even with insurance coverage, farmers face challenges in expanding production or adopting new technologies.

Furthermore, as the Chow test can provide an objective statistical criterion to examine whether these inter-group differences truly exist in the population, the Chow test method is adopted to calculate the F-statistic and its corresponding p-value, thereby conducting a rigorous test of the null hypothesis that there are no differences between groups. This approach helps avoid subjective judgment. Pairwise grouping is performed for the Eastern, Central, Western, and Northeastern regions, resulting in a total of six tests. The test results are as follows: the Chow test statistic between the Eastern and Central regions is 14.400 with p = 0.000; between the Eastern and Western regions, it is 39.620 with p = 0.000; between the Eastern and Northeastern regions, it is 19.530 with p = 0.000; between the Central and Western regions, it is 12.220 with p = 0.000; between the Central and Northeastern regions, it is 22.020 with p = 0.000; and between the Western and Northeastern regions, it is 8.920 with p = 0.000. The p-values for all six tests are highly significant at the 1% level, which strongly rejects the null hypothesis of no structural differences in the models across regions. Therefore, the test results statistically confirm that there are significant structural differences in the impact mechanisms of agricultural insurance on agricultural economic resilience among the Eastern, Central, Western, and Northeastern regions. Conducting analysis using subsample regression models is both necessary and appropriate, which methodologically supports the research hypothesis of regional heterogeneity in this study. Therefore, Hypothesis 3 is validated.

6. Discussion

Existing literature has extensively explored the relationship between agricultural insurance and the agricultural economy, reaching a consensus that agricultural insurance positively influences agricultural economic development. For instance, Zeng S et al., taking prefecture-level cities in Zhejiang Province, China, as a case study, empirically demonstrated that agricultural insurance has a significant positive impact on agricultural economic growth [49]. Further, some scholars have shifted their focus to the relationship between agricultural insurance and agricultural economic resilience. For example, research by Chen T et al. not only supported the positive role of agricultural insurance in enhancing agricultural economic resilience but also revealed the positive moderating effect of the digital economy in this relationship [50]. Similarly, a study by Li H et al., based on county-level data and a DID model, also indicated that full-cost insurance policies have a positive effect on agricultural economic resilience [51].

However, it is noteworthy that existing studies tend to interpret a linear positive relationship between agricultural insurance and agricultural economic resilience. In contrast, the empirical results of this study reveal that the impact of agricultural insurance on agricultural economic resilience is not a simple linear promotion but rather a nonlinear, U-shaped relationship characterized by initial inhibition followed by promotion. This finding theoretically supplements and deepens existing research. We argue that this is primarily due to the two-stage effect in the development process of agricultural insurance. During the inhibitory phase, the coverage level of agricultural insurance is limited, claims settlement efficiency is relatively low, and it is insufficient to effectively hedge against production risks. Moreover, during this period, farmers’ risk management awareness and ability to utilize insurance are relatively weak, resulting in an insignificant promotional effect of agricultural insurance in the short term. Once it enters the promotional phase, agricultural insurance development matures gradually, and the acceptance capacity of new agricultural business entities improves. The core functions of insurance are fully realized, not only stabilizing production expectations through risk loss compensation but also fundamentally enhancing the resistance and adaptability of the agricultural economic system to external shocks by promoting the adoption of new technologies and optimizing resource allocation efficiency. This study further identifies, through mechanism testing, that agricultural technological progress serves as a significant mediating pathway through which agricultural insurance influences agricultural economic resilience. This discovery, from a dynamic capability perspective, reveals the internal process by which agricultural insurance operates on economic resilience. It demonstrates that agricultural insurance functions not only through risk loss compensation but also by incentivizing technological advancement, thereby enhancing the adaptability and transformative capacity of the agricultural economic system. Theoretically, this enriches the explanatory framework for the relationship between agricultural insurance and agricultural economic resilience and provides new mechanistic evidence for understanding the long-term developmental benefits of agricultural insurance.

7. Conclusions and Implications

7.1. Conclusions

From a theoretical perspective, this paper systematically reviews the mechanisms through which agricultural insurance affects agricultural economic resilience, puts forward three theoretical hypotheses, introduces a panel threshold model, and employs panel data from 30 provinces in China over the period 2011–2023 to empirically examine the mechanism by which agricultural insurance influences agricultural economic resilience. The main findings are as follows. First, there exists a nonlinear U-shaped relationship between agricultural insurance and agricultural economic resilience, and this conclusion remains robust after a series of robustness tests. Second, agricultural technological progress serves as a critical channel through which agricultural insurance affects agricultural economic resilience. By promoting agricultural technological advancement, agricultural insurance exerts a positive facilitating effect on agricultural economic resilience. Third, the impact of agricultural insurance on agricultural economic resilience exhibits a threshold effect, whereby the influence becomes gradually stronger as the level of industrial structure optimization improves. Fourth, the impact of agricultural insurance on agricultural economic resilience displays regional heterogeneity characterized by a nonlinear U-shaped pattern, with the effect of agricultural insurance being more pronounced in the eastern, central, and northeastern regions.

7.2. Implications

Based on the above conclusions, this paper proposes the following policy recommendations: First, it is essential to accelerate the improvement of rural financial infrastructure by expanding the coverage of financial institutions in rural areas and enhancing the breadth and quality of their services. This will enable more farmers to access convenient agricultural insurance through inclusive finance. Second, agricultural insurance products should be innovated by developing differentiated offerings linked to green agricultural technologies and smart agricultural machinery. Such products can incentivize farmers to adopt advanced technologies that increase yield, improve efficiency, conserve resources, and protect the environment. At the same time, the “insurance + credit + services” model should be strengthened to alleviate financing constraints for farmers adopting new technologies. Insurance institutions are encouraged to collaborate with agricultural technology companies to provide integrated technological solutions and risk management services to policyholders. Third, differentiated agricultural insurance support strategies should be implemented according to the threshold effect of industrial structure. In regions with more advanced industrial structures, agricultural insurance should be deeply integrated with modern agricultural value chains, and comprehensive insurance products should be developed for high-value-added agricultural products, large-scale production, and green technologies—with enhanced coverage levels and claims settlement efficiency. In regions with less developed industrial structures, policy efforts should prioritize industrial optimization and upgrading while improving basic insurance provision to support agricultural transition with adequate risk protection. Fourth, regionally tailored support strategies should be adopted in response to spatial variations in how agricultural insurance affects agricultural economic resilience. In eastern, central, and northeastern regions, where a significant U-shaped relationship is observed, agricultural insurance should be further integrated with credit and technology extension systems. In western regions, efforts should focus on improving transportation, communications, financial infrastructure, and market systems, while promoting the linkage of agricultural insurance with industrial development and technology dissemination. This will establish the foundational conditions for insurance to play an effective role and gradually guide these regions onto the upward segment of the U-shaped curve.

7.3. The Limitations of the Research

This study provides important evidence for understanding the complex relationship between agricultural insurance and agricultural economic resilience. However, it also has certain limitations. First, regarding data and sample coverage, while provincial panel data from 2011 to 2023 were used, the period does not fully encompass all critical stages of agricultural insurance policy evolution, and the analysis lacks examination of long-term shocks, such as climate disasters and policy shifts. Future research could incorporate household-level data to explore the impact of insurance on the resilience of different agricultural operators, as well as introduce data from longer time spans. Second, in terms of methodology, a two-way fixed effects model was employed to analyze the relationship between agricultural insurance and agricultural economic resilience. However, the estimates derived from this approach are more indicative of a robust correlation, and challenges remain in establishing strict causal inference. Therefore, to more accurately identify the causal effect of agricultural insurance on economic resilience, future studies could explore natural experiment settings, such as leveraging the rollout of agricultural insurance pilot policies or exogenous events like natural disaster shocks, to strengthen causal identification. Third, in the theoretical analysis section, although we have attempted to provide theoretical support for the “U-shaped” relationship between agricultural insurance and agricultural economic resilience from the perspective of institutional evolution, the model constructed remains a highly simplified theoretical framework. Future research could develop more complex general equilibrium models or incorporate additional realistic dimensions, such as farmer heterogeneity and the diversity of insurance products, into the theoretical model to achieve a clearer analysis of the relationship between agricultural insurance and agricultural economic resilience.