Examining Whether Participation in Industrial Integration Can Enhance Farmers’ Income Based on Empirical Evidence from the “Hundred Villages and Thousand Households” Survey in Jiangxi Province

Abstract

Against the backdrop of China’s Rural Revitalization Strategy, rural industrial integration is regarded as a critical pathway to boosting farmers’ income, yet its specific impact and heterogeneous characteristics remain to be explored. Using biennial panel data from the 2021 and 2023 “Hundred Villages and Thousand Households” survey in Jiangxi Province, this study employs two-way fixed effects models, the instrumental variable method, and quantile regression to investigate the effect of farmers’ participation in rural industrial integration on their income. The findings show that participation in industrial integration significantly increases household income by an average of 28.6%, with causal relationships confirmed by instrumental variable analysis. Among different integration modes, industrial chain extension has the most significant effect, followed by functional expansion and internal multi-format integration, while technology penetration shows no significant effect; overlapping multiple modes exhibits a negative interactive effect. Additionally, high-standard farmland construction amplifies the income-increasing effect, and the effect is more pronounced for low-income farmers, those in mountainous areas, and farmers in the Central Jiangxi region. This study provides micro-level empirical evidence for optimizing industrial integration policies and advancing rural revitalization in central and western agricultural provinces.

1. Introduction

Against the backdrop of accelerated global urbanization, rural decline has become a core challenge threatening sustainable development [1]. Developing countries, in particular, face the predicament of a widening urban–rural income gap; the per capita disposable income of rural residents in China is only 38.2% of that of urban residents [2], and addressing this predicament has become a key issue in the implementation of the rural revitalization strategy. The report of the 20th National Congress of the Communist Party of China explicitly listed “promoting the integrated development of primary, secondary, and tertiary industries in rural areas and expanding channels for farmers to increase their income and achieve prosperity” as an important path for rural revitalization, pointing out the direction for agricultural and rural development. Under this policy framework, Jiangxi Province innovatively proposed the “Four Integrations and One Common” strategy for building beautiful and harmonious villages in 2023, promoting common prosperity through four paths: the integration of scenic areas and villages, the integration of industries and villages, the integration of three types of governance (self-governance, rule of law, and rule of virtue), and urban–rural integration. In 2025, the Opinions on Further Deepening Rural Reform and Solidly Promoting Comprehensive Rural Revitalization issued by Jiangxi Province further proposed the “Double Hundred Action” for agricultural industrialization, aiming to promote the transformation of agriculture to high-value-added fields through industrial chain extension, internal integration, functional expansion, and technological penetration, providing a typical policy context for the research on the income-increasing effect of industrial integration.

As a key path for rural revitalization [3], rural industrial integration has been proven to significantly increase farmers’ income by reconstructing the value chains of agriculture, agricultural product processing, services, and other industries [4], and providing more non-agricultural employment opportunities [5,6]. International cases also confirm the income-increasing effect of integration: India’s “agriculture + aromatic crops” model has increased smallholders’ profits by 250% [7], and the promotion of climate-smart agricultural technologies in Ghana has increased maize income by 44% [8].

From the perspective of the influence mechanism, the path through which rural industrial integration affects farmers’ income exhibits multi-dimensional characteristics. On the one hand, industrial integration directly enhances the economic benefits of agricultural production by promoting large-scale agricultural operations [9], advancing the application of green technologies [10], and optimizing the efficiency of resource allocation [11]; On the other hand, non-agricultural employment opportunities generated through attracting investment and encouraging return to hometowns for entrepreneurship [12], in the process of rural industrial integration [13], further increase farmers’ income through the growth of wage income and property income.

However, existing studies have three key limitations that this study aims to address: first, they overemphasize macro-policy effects [2,5] while lacking dynamic tracking of farmers’ micro-level behavioral decisions and income changes; second, they insufficiently address regional heterogeneity—focusing mostly on the national level or developed eastern regions [14] and leaving a critical gap in empirical research on central China’s major agricultural provinces (especially Jiangxi, which has unique ecological advantages but a weak industrial foundation, with industrial integration paths distinct from coastal areas); third, they fail to conduct in-depth, systematic analysis on the moderating mechanisms, integration model differences, and context dependence (e.g., terrain, region, farmers’ income stratification) of farmers’ participation in industrial integration. In short, micro-level empirical evidence targeting Jiangxi Province is scarce, and the heterogeneous effects and moderating mechanisms of industrial integration on farmers’ income remain understudied.

The topography of Jiangxi Province consists mainly of mountains and hills (Figure 1), with mountainous areas accounting for 36% of the province’s total area, and hilly areas being more extensive, reaching 42%. By contrast, plain areas are relatively scarce, accounting for only 12% of the province’s total area. These topographic distribution characteristics have led Jiangxi Province to form a unique geographical pattern of “six parts mountains, one part water, and two parts farmland”. Under this pattern, arable land resources exhibit obvious fragmentation characteristics, being distributed scattered and discontinuous. Meanwhile, the proportion of agricultural population in Jiangxi Province is relatively high, exceeding 40% of the total population. Due to the constraints of arable land fragmentation, it is difficult for farmers’ income growth to achieve significant improvement through large-scale agricultural operations. Therefore, agricultural development in Jiangxi Province has a high dependence on industrial integration and urgently needs to promote the steady growth of farmers’ income through diversified industrial integration models.

The marginal contributions of this study are as follows:

(1)

Based on micro-panel data encompassing 462 rural households in Jiangxi Province (2021–2023), and by using software tools such as Stata 18 and ArcGIS 10.8, this study pioneers a systematic examination of the income effects associated with farmers’ participation in industrial integration, thereby addressing the research gap concerning micro-empirical evidence within Jiangxi Province.

(2)

Employing a combined approach of a two-way fixed-effects (FE) model and an instrumental variable (IV) methodology—utilizing ‘policy awareness among village collective economic organizations’ as the IV—this research effectively mitigates endogeneity stemming from self-selection bias, enabling the precise identification of a causal relationship between industrial integration and farmer income augmentation.

(3)

The analysis reveals three principal dimensions of heterogeneity and moderating mechanisms: (a) differential income-enhancing effects across four distinct integration models; (b) the positive moderating role of high-standard farmland construction; and (c) heterogeneous performance contingent upon income levels, regional contexts, and topographic conditions. These findings also provide targeted empirical support for policy-making.

2. Theoretical Analysis and Research Hypotheses

Rural industrial integration provides diversified paths for increasing farmers’ income by breaking the single production function of traditional agriculture and reconstructing the allocation of production factors. As shown in Figure 2, its effects exhibit heterogeneity due to differences in integration models. At present, mainstream scholars at home and abroad classify farmers’ participation behaviors into four types: industrial chain extension type, internal integration type, functional expansion type, and technology penetration type [14,15].

First, this study examines the impact of whether farmers participate in industrial integration on their income. Existing studies have shown that through optimizing resource allocation and improving production efficiency, industrial integration enables participating farmers to have a significantly higher income level than non-participants [16]. Therefore, this study hypothesizes as follows:

H1. 

Farmers’ participation in rural industrial integration can significantly increase their income level.

Different industrial integration models have different impact mechanisms on farmers’ income, thus resulting in differences in the magnitude of their impact on farmers’ income.

(1)

Specifically, industrial chain extension integration extends the value chain through agricultural product processing, e-commerce sales, etc., and directly obtains added value, and is expected to have the greatest impact on farmers’ income. For example, the development of rural e-commerce can significantly increase farmers’ household income [17], and after farmers participate in grain processing cooperatives, the income per unit product increases [18]. In addition, farmers can also promote the use of the Internet such as live-streaming sales by participating in cooperatives [19]. Based on this, the following is proposed:

H2a. 

Among the four integration modes, farmers’ participation in industrial chain extension integration has the most significant income-increasing effect.

(2)

The income-increasing effect of internal integration (such as crop–livestock integration and diversification within agriculture) is reflected in improved resource utilization efficiency. Crop–livestock integration can reduce the income volatility of Indian farmers by 30% and increase their average income by 15%. Diversified farming systems (such as intercropping and agroforestry) increase smallholder farmers’ income by 25–50% by reducing risks and improving resource efficiency [20]. Crop diversification can directly bring about income diversification for farmers [21], reduce the market risks of a single crop, and lower farmers’ income fluctuations [22]. Internal integration models such as intercropping legumes and vegetables increase farmers’ income by 53–198% [23]. However, the essence of multi—format integration within agriculture is the horizontal integration of internal agricultural formats. Its value creation is confined to the “primary agricultural product production” stage, failing to reach the high-value-added links downstream of the industrial chain or high-value-added areas outside agriculture. As a result, its effect on increasing farmers’ income is relatively weaker compared with industrial chain extension-type integration and function expansion-type integration. Therefore, this study hypothesizes the following:

H2b. 

Farmers’ participation in internal multi-format integration can significantly increase their income, but the effect is weaker than that of industrial chain extension integration.

(3)

Functional expansion integration (such as ecotourism and agricultural study tours) achieves income growth by exploring the non-production functions of agriculture. The income level of farmers participating in rural tourism is significantly higher than that of non-participants [24], with the core mechanism being the value conversion of ecological products and the distribution of tourism benefits. For example, West African agroforestry systems, by integrating agricultural and ecological protection functions, enable farmers to obtain income from forest by-products, with their total income increasing by more than 30% [25]. In addition, farmers can also achieve asset income through land transfer, collective economic equity, etc. For instance, farmers develop homestays through homestead transfer, with the proportion of non-agricultural income increasing to 40% [26]; The essence of functional expansion-type integration is to explore the non-production functions of agriculture. Its creation of added value breaks through the boundary of “primary agricultural product production”, yet it does not traverse the entire “production–processing–sales” value chain like industrial chain extension. It is expected that the increase in farmers’ income from this type of integration lies between that of the other two types. Therefore, this study hypothesizes the following:

H2c. 

Farmers’ participation in functional expansion integration can significantly increase their income, with an effect between industrial chain extension integration and internal multi-format integration.

(4)

Technology penetration integration introduces modern technological means; for example, green production technologies increase income by raising output and sales prices, thereby offsetting the incurred costs [27]. However, the income-increasing effect of technology penetration integration is controversial. Although agricultural digitalization can improve production efficiency, in the initial stage of technology promotion, smallholders find it difficult to benefit due to insufficient skills; smart agricultural equipment has a long investment return cycle, bringing no short-term improvement to smallholders’ income [28], and may even increase costs due to equipment investment. In addition, farmers find it difficult to adopt these agricultural technologies due to their insufficient awareness and infrastructure barriers [29]. Many technological applications in agriculture often fail to increase smallholders’ income, due to low digital literacy and insufficient value chain integration [30]. As the study utilizes data exclusively from 2021 and 2023, the time interval is limited. Moreover, farmers in Jiangxi Province generally exhibit low educational attainment. Consequently, the following is proposed:

H2d. 

Farmers’ participation in technology penetration-type industrial integration has no significant effect on income in the short term.

When farmers participate in multiple models simultaneously, resource constraints may lead to “synergistic inhibition”; participating in multiple integration models will cause the diversion of management efforts, resulting in a decline in technical efficiency [31]. Excessive integration of agriculture with industry and service industries may trigger competition for land resources, which in turn inhibits income growth [32]. Participation in multiple value chains will increase transaction costs, resulting in net income lower than that of a single integration model [33]. Farmers participating in both enterprise-led and cooperative-led integration have a lower income growth rate than those participating in a single model due to conflicts in the profit distribution mechanism [34]. Based on this, the following is proposed:

H3. 

The superposition of multiple integration models has a negative interaction effect, that is, the income-increasing effect of participating in two or more models simultaneously is weaker than that of a single model.

Land consolidation represented by high-standard farmland construction projects can improve arable land quality and grain production capacity, and enhance resource utilization efficiency [35]. Improved land infrastructure enables farmers participating in industrial integration to reduce production costs, promote crop specialization, and increase investment in agricultural innovation [36]. It is anticipated that villages implementing high-standard farmland construction projects will experience more pronounced income growth among farmers engaged in industrial integration. Based on this, the following is proposed:

H4. 

High-standard farmland construction plays a positive moderating role in the relationship between farmers’ participation in industrial integration and income growth.

Farmers’ participation in industrial integration has different impacts on farmers with different income levels, in different regions, and in different terrain. Due to the lack of initial resources, low-income farmers enjoy higher marginal returns from industrial integration, while high-income farmers, who already have a foundation for large-scale operations, experience relatively limited incremental effects from industrial integration. The poverty reduction effect of rural industries on the poorest 20% of farmers is 1.5 times that on high-income groups [37], and the income increase margin for low-income farmers is significantly higher than that for high-income groups [38,39]. Studies based on apple growers show that such income gaps among farmers mainly stem from differences in non-agricultural income, and industrial integration can narrow this gap, with low-income groups benefiting more significantly [40,41].

Regional resource differences within Jiangxi Province lead to differentiation in effects. Central Jiangxi, relying on the market radiation of the Nanchang metropolitan circle, has a significant superimposed effect of industrial chain extension and functional expansion [14]. Mountainous areas, relying on resources such as characteristic forest and fruit as well as ecotourism, have a significant income-increasing effect [5], hilly areas, due to land fragmentation restricting the exertion of effects, have a relatively low income increase margin [26]. Based on this, the following is proposed:

H5. 

The income-increasing effect of industrial integration on low-income farmers is stronger than that on high-income farmers.

H6. 

The income-increasing effect of industrial integration in the Central Jiangxi region is the highest.

H7. 

The income-increasing effect of industrial integration in mountainous areas is the strongest.

3. Data Sources, Variable Selection and Model Construction

3.1. Data Sources

This study relies on the China Rural Revitalization Jiangxi “Hundred Villages and Thousand Households” Database (hereinafter referred to as CRRJDB), jointly constructed by the Institute of Rural Revitalization Strategy of Jiangxi Agricultural University and the China Center for Agricultural Policy at Peking University. The research team conducted four rounds of household surveys in 2019, 2020, 2022, and 2024. Using the method of stratified random sampling, they have cumulatively collected micro—household survey data from 12 county—level administrative regions in 11 prefecture—level cities, 108 villages, and over 1000 rural households in Jiangxi Province in 2018, 2019, 2021, and 2023. These data provide a solid foundation for academic research, the compilation of policy analysis reports, and graduate education. With its high data reliability and validity, CRRJDB has established its authoritative status in the field of micro-level farmer research in Jiangxi Province and is regarded as a representative sample dataset. Given the significant differences in items between the questionnaires of 2018 and 2019 and those of 2021 and 2023 in the Hundred Villages and Thousand Households Database, this study selected data from 2021 and 2023 as research materials based on the principles of timeliness and consistency of questionnaire items. After excluding the missing values of the explained variable, farmers’ income, this study integrated the original questionnaire data from 2021 and 2023 according to the farmers’ numbers. That is, only the farmers with the same number were retained. Finally, a panel dataset of 924 valid samples over two years from 462 farmers was obtained. As shown in Figure 3, the samples cover areas such as Wanan County and Yongxin County of Ji’an City, Dayu County and Ruijin County of Ganzhou City, Fengxin County of Yichun City, Pengze County of Jiujiang City, Xinjian District and Jinxian County of Nanchang City, Fuliang County of Jingdezhen City, Yushan County of Shangrao City, Luxi County of Pingxiang City, and Zixi County of Fuzhou City.

3.2. Variable Selection

As shown in

Table 1. Variable selection and description.

Variable Type and Name	Variable Symbol	Variable Meaning	Variable Description
Explained Variable
Farmers’ Income	income	Farmers’ Annual Income	Yuan/Year
Explanatory Variable
Farmers’ Participation in Industrial Integration Behavior	participate	Whether farmers participate in industrial integration	Yes = 1, No = 0
Industrial chain extension-type integration	chain	Whether your family engage in online sales of agricultural products	Yes = 1, No = 0
		Whether your family is a member of an agricultural cooperative	Yes = 1, No = 0
Internal multi-format integration	multi	Whether your family have allocated or managed forest land	Yes = 1, No = 0
		Whether your family have animal husbandry	Yes = 1, No = 0
		Whether your family have aquaculture	Yes = 1, No = 0
		Whether your family plant other crops	Yes = 1, No = 0
Functional expansion-oriented integration	func	Whether your family have the transfer of homestead and housing use rights?	Yes = 1, No = 0
		Whether your family have equity shares in the collective economic organization of this village	Yes = 1, No = 0
		Whether your family have arable land that is entrusted or centrally transferred by the village collective?	Yes = 1, No = 0
		Have you had self-operated industry and commerce?	Yes = 1, No = 0
Technological penetration-oriented integration	tech	Have you had soil testing and fertilizer recommendation?	Yes = 1, No = 0
		Does your family use drones for agricultural production?	Yes = 1, No = 0
Control variable
Individual characteristics	gender	Gender	Male = 1, Female = 0
	age	Age	Year
	edu	Years of education	Year
Family characteristics	family	Number of family members	person
	area	Area of own cultivated land managed	mu
	plots	Number of plots	piece
	homestead	Number of homestead plots	piece
Social capital	cpc	Whether you are a member of the Communist Party of China	Yes = 1, No = 0
	leader	Have you ever served as a village cadre
Instrumental variable
Policy cognition	heard	Have you heard of the village’s collective economic organization	Yes = 1, No = 0
Moderating variable
Village type	farm	Whether the village has implemented the high-standard farmland construction and upgrading project	Yes = 1, No = 0

, combined with the items of the “Hundred Villages and Thousand Households” survey questionnaire and the theoretical analysis in the previous text, the explained variable of this study is the total income of farmers, and the core explanatory variable is farmers’ participation in rural industrial integration behavior. Farmers’ participation in industrial integration is divided into four categories: industrial chain extension type, internal integration type, functional expansion type, and technology penetration type. Industrial chain extension integration is indicated by whether farmers engage in online sales of agricultural products; internal integration is measured by whether farmers have been allocated or manage forestland, whether they engage in animal husbandry, whether they are involved in aquatic products, and whether they grow other crops; technology penetration integration is indicated by items such as whether farmers conduct soil testing and formula fertilization, and whether they use drones for pesticide application. If a farmer participates in one or more of these, their participation in industrial integration is recorded as 1; otherwise, it is recorded as 0. Control variables include individual characteristics, household characteristics, and social capital; the instrumental variable is farmers’ policy cognition; and village type is selected as the moderating variable.

3.2.1. Explained Variable

Farm households’ annual income, as the core explained variable, serves as a key indicator for measuring farm households’ economic well-being and can comprehensively reflect the comprehensive benefits that farm households obtain through participating in rural industrial integration [42]. In some studies, farm households’ income is often decomposed into agricultural operating income and non-agricultural income. However, in this study, farm households’ participation in rural industrial integration directly affects their overall income level through approaches such as expanding agricultural industrial chains and creating employment opportunities.

In this study, farmer income is logarithmically transformed to mitigate fluctuation in the variance of the error term and alleviate heteroscedasticity. Concurrently, the coefficients derived from logarithmic transformation represent income elasticity, which enhances economic interpretability. Given the presence of extreme values in annual farmer income, a 1% two-sided winsorization is applied during robustness testing to mitigate potential bias from data entry errors or anomalous events. This treatment preserves the core sample distribution.

3.2.2. Core Explanatory Variable

This study defines farmers’ participation in industrial integration behavior as a unary discrete variable, that is, whether farmers have the behavior of participating in industrial integration. Farmers’ participation in industrial integration can be divided into four major types: industrial chain extension integration, internal integration, functional expansion integration, and technology penetration integration [43,44].

(1)

Industrial Chain Extension Integration. The popularization of Internet technology has promoted online sales of agricultural products, becoming an important way for industrial chain extension. The proportion and income of online sales of agricultural products directly reflect the degree of farmers’ participation in the e-commerce field and their income status, and serve as a key indicator to measure the extension of the industrial chain to the circulation and sales links. Online sales of agricultural products significantly increase farmers’ profits [45], which is a manifestation of farmers’ participation in industrial chain extension integration. Joining agricultural cooperatives is a way for small-scale farmers to overcome their limitations in the market and increase their income [46], and it is also an important form for farmers to participate in industrial chain extension integration. Cooperatives can integrate resources, enhance market negotiation capabilities, reduce production costs, provide technical support and information services, etc., which help increase farmers’ income [17]. Agricultural cooperatives are important new-type agricultural business entities. They increase the sales price of agricultural products by providing agricultural materials procurement, technical services, etc. [47]. Participation in cooperatives and the types of services obtained reflect the degree of organization and resource integration capabilities of farmers in the industrial chain.

(2)

Multi-business Integration within Agricultural Internal Integration. Multi-business integration within agriculture forms an ecological closed loop through the combination of planting and breeding as well as the complementation of agriculture and forestry, reducing external input costs and increasing per unit land income. Internal integration builds a diversified income structure through multi-category operations (such as the combination of planting and breeding, aquaculture, and under-forest economy, etc.) [48], which reduces the risk of fluctuations in a single industry. Growing multiple crops, especially high-value crops, is one of the main sources for increasing farmers’ income [49,50].

(3)

Functional Expansion Integration. Farmers can obtain additional income such as rent by transferring the use rights of their idle homesteads [51], and this is also conducive to revitalizing idle rural resources and improving resource utilization efficiency [52]. Land trusteeship and transfer can promote large-scale land operation, and improve technology adoption and agricultural production efficiency [53]. Self-operated industry and commerce and other non-agricultural operations are key to income diversification [54], and diversified income sources can enhance risk resistance capacity.

(4)

Technology Penetration Integration. Green production technologies, such as precision fertilization like soil testing and formula fertilization, can reduce costs and increase yields, and they are an important manifestation of agricultural technological innovation and sustainable development [55]. In the field of agricultural production, traditional field operation modes often rely on manual labor, which is not only inefficient. Moreover, in the pesticide spraying link, operators are exposed to pesticide environments for a long time and are highly vulnerable to harm from chemical agents. Drones are one of the latest newly added tools in the precision agriculture technology toolkit [56], With their flexible flight performance and accurate positioning systems, they can accomplish large-area field operation tasks quickly and efficiently [57], thereby reducing the workload of farmers. The application of this technology minimizes the use of pesticides and water, which not only improves the economic benefits of agricultural production but also reflects the important value of modern science and technology in the agricultural field, providing strong support for the sustainable development of agriculture.

3.2.3. Control Variables

(1)

Individual Characteristics: Individual characteristics such as gender, age, and education level affect farmers’ production decisions and participation capabilities [58,59,60]. Farmers with a higher education level are more likely to accept new technologies and ideas [59], and have a significant impact on the willingness and effectiveness of participating in industrial integration.

(2)

Household Characteristics: The number of family laborers and the area of contracted land reflect the family’s production resource endowment [61], and these factors are directly related to farmers’ ability and scale of participating in industrial integration. The number of plots reflects the degree of cultivated land fragmentation, which increases production costs, reduces the advantages of economies of scale [62], directly affects mechanization efficiency and supervision costs, and serves as an important constraint condition for farmers’ production decisions. Multiple homesteads correspond to scattered residence, which exacerbates redundancy of infrastructure, reduces the agglomeration effect [52], and also inhibits industrial integration. The number of plots and homesteads reflects the degree of dispersion of family production and living materials, affecting farmers’ integrated utilization of resources and operation and management efficiency.

(3)

Social Capital: Party membership is usually positively correlated with access to policy information and opportunities for organizational participation. For example, farmer households with Party members are more likely to participate in agricultural operation organizations, which is attributed to the information advantages of Party members in policy interpretation and project application [63]. As core subjects of rural governance, village cadres are endowed with advantages in resource integration due to their status. Generally, the probability of village cadres’ families participating in industrial integration is higher than that of ordinary farmers, and the proportion of their operating income coming from secondary and tertiary industries is also higher. Therefore, taking “whether being a Communist Party of China (CPC) member” and “whether being a village cadre” as control variables of social capital can effectively capture the heterogeneous participation behaviors of farmers in industrial integration and provide a more rigorous basis for causal inference for the model.

3.2.4. Instrumental Variable

This study selects policy cognition as the instrumental variable. Policy cognition affects farmers’ decision-making in participating in industrial integration, but does not directly affect farmers’ income [64]. The policy of developing and strengthening the village collective economy is an important policy in rural areas. Farmers’ policy cognition of collective economic organizations reflects their understanding of the rural land system and affects their decision-making in participating in village collective economic equity and industrial integration [65], which can serve as an instrumental variable at the policy level.

3.2.5. Moderating Variable

Large-scale construction of high-standard farmland in China has improved the quality of cultivated land and crop yields, and enhanced ecological environment security [66]. The construction of high-standard farmland creates more favorable conditions for rural industrial integration through such paths as improving agricultural production efficiency, adjusting agricultural planting structure, and expanding the scale of agricultural operations. In areas participating in the construction, land is more concentrated and contiguous, suitable for mechanized operations, which can attract more enterprises and capital to enter, expand channels for farmers to participate in industrial integration, and thereby affect the income that farmers obtain from industrial integration.

3.3. Empirical Model Design

To systematically identify the causal effect of farmers’ participation in rural industrial integration on their income, this study constructs a multi—level econometric model system by integrating the characteristics of panel data and the complexity of the research problem. The standard errors of all regression models are clustered at the farmer level to mitigate the underestimation of standard errors caused by unobserved heterogeneity. The specific settings are as follows:

3.3.1. Multiple Linear Regression Model

To preliminarily and quickly test the correlation between participation in industrial integration and farmers’ income, the basic linear regression model is set as follows:

$${Ln\_income}_{it}={\alpha }_0+{\alpha }_1 {participate}_{it}+{\displaystyle \sum {\alpha }_k}{X}_{kit}+{\mu }_{it}$$

(1)

In Equation (1), $i$ denotes individual farmers, and $t$ denotes time, and $Ln\_incom{e}_{it}$ is the explained variable, representing the logarithm of the income of farmers $i$ in year $t$. $participat{e}_{it}$ is the core explanatory variable, indicating whether farmer $i$ participates in industrial integration in year $t$. $X_{kit}$ is a set of control variables, encompassing individual characteristics, household characteristics, and social capital. ${\alpha }_0$ is the constant term, ${\alpha }_1$ is the coefficient of the core explanatory variable, ${\alpha }_k$ is the coefficient of the control variables, and ${\mu }_{it}$ is the random error term.

3.3.2. Two-Way Fixed Effects Model

Given that farmer income may be influenced by time-invariant individual heterogeneity factors and time-varying factors such as macro policies and market fluctuations, thus, employing a two-way fixed effects model to simultaneously control for individual-specific and time-specific effects enables a more accurate identification of the causal impact of participating in industrial integration. The model specification is as follows:

$$Ln\_incom{e}_{it}={\beta }_0+{\beta }_{1}participat{e}_{it}+{\displaystyle \sum {\beta }_k}{X}_{kit}+{\gamma }_i+{\lambda }_t+{\epsilon }_{it}$$

(2)

In Equation (2), ${\gamma }_i$ represents the individual fixed effect, ${\lambda }_t$ denotes the time fixed effect, ${\epsilon }_{it}$ is the random error term. Through this model, the net effect of participating in industrial integration on income can be isolated, and the core coefficient ${\beta }_1$ measures the marginal change in income resulting from participation. ${\beta }_0$ is the constant term, ${\beta }_k$ is the coefficient for the control variables.

To further analyze the heterogeneous effects of different integration models, the core explanatory variables are replaced with four types of industrial integration models:

$$\begin{array}{l}Ln\_incom{e}_{it}={\beta }_0+{\beta }_{1}chai{n}_{it}+{\beta }_{2}mult{i}_{it}+{\beta }_{3}fun{c}_{it}\\ +{\beta }_{4}tec{h}_{it}+{\displaystyle \sum {\beta }_k}{X}_{kit}+{\gamma }_i+{\lambda }_t+{\epsilon }_{it}\end{array}$$

(3)

Among them, $chain$, $multi$, $func$ and $tech$ respectively represent four integration models: industrial chain extension type, internal multi-format integration type, functional expansion type, and technology penetration type integration. Coefficient ${\beta }_1, {\beta }_2, {\beta }_3, {\beta }_4$ reflect the differences in the income-increasing effects of different models. The meanings represented by other symbols are consistent with those in the above formula, and will not be repeated here.

3.3.3. Quantile Regression Model

To further examine the differential impacts of industrial integration on farmers with different income levels, the quantile regression model is adopted:

$$\begin{array}{l}{Q}_{\tau }(Ln\_incom{e}_{it}\mid participat{e}_{it},X_{kit})={\delta }_0(\tau )+{\delta }_1(\tau )participat{e}_{it}\\ +{\displaystyle \sum {\delta }_k}(\tau )X_{kit}+{\gamma }_i+{\lambda }_t+{\nu }_{it}\end{array}$$

(4)

Among them, $Q_{\tau }(\cdot )$ represents the conditional quantile of the explained variable at the $\tau$ quantile $(\tau =0.1, 0.25, 0.5, 0.75, 0.9)$; ${\delta }_1(\tau )$ is the coefficient of the core explanatory variable at the $\tau$ quantile, which is used to identify the marginal effects of industrial integration at different positions of the income distribution; ${\delta }_0$ is the constant term; ${\delta }_k$ is the coefficient of the control variable, $v_{it}$ is the random error term.

3.3.4. Instrumental Variable Model

Given that high-income farmers may be more inclined to participate in industrial integration, which leads to endogeneity issues, this study selects “whether they understand village collective economic organizations” as the instrumental variable. This variable may affect farmers’ willingness to participate in industrial integration, but does not directly affect their income level. Therefore, to address potential endogeneity issues, this study employs the instrumental variable two-stage least squares (IV—2SLS) for parameter estimation, with the specific implementation steps as follows:

First stage: Regress the core explanatory variable $R_{it}$ on the instrumental variable and all control variables to capture exogenous variation.

$$participat{e}_{it}={\theta }_0+{\theta }_{1}hear{d}_{it}+{\displaystyle \sum {\theta }_k}{X}_{kit}+{\gamma }_i+{\lambda }_t+{\xi }_{it}$$

(5)

In the above formula, $heard$ is the instrumental variable “whether they have heard of village collective economic organizations”; ${\theta }_0$ is the constant term; ${\theta }_1$ is the coefficient of the instrumental variable; ${\theta }_k$ is the coefficient of the control variable, ${\xi }_{it}$ is the random error term.

Second stage: Regress the explained variable $\ln ({income}_{it})$ on the predicted participation probability value of ${participate}_{it}$ obtained from the first stage.

$$Ln\_incom{e}_{it}={\theta }_0^{\prime }+{\theta }_1^{\prime }{\widehat{participate}}_{it}+{\displaystyle \sum {\theta }_k^{\prime }}{X}_{kit}+{\gamma }_i+{\lambda }_t+{\xi }_{it}^{\prime }$$

(6)

In the above formula, ${\widehat{participate}}_{it}$ is the first-stage fitted value; Coefficient ${\theta }_1^{\prime }$ is the net effect after addressing endogeneity. ${\theta }_0^{\prime }$ is the constant term; ${\theta }_k^{\prime }$ is the coefficient of the control variable, ${\xi }_{it}^{\prime }$ is the random error term.

3.3.5. Moderating Effect Model

This study introduces “the village has implemented high-standard farmland construction and improvement projects” as a moderating variable for farmers’ participation in industrial integration to examine the strengthening effect of agricultural production infrastructure on the income-increasing effect of industrial integration:

$$\begin{array}{l}Ln\_incom{e}_{it}={\varphi }_0+{\varphi }_{1}participat{e}_{it}+{\varphi }_{2}far{m}_{it}\\ +{\varphi }_3\left(participat{e}_{it}\times far{m}_{it}\right)+{\displaystyle \sum {\varphi }_k}{X}_{kit}+{\gamma }_i+{\lambda }_t+{\omega }_{it}\end{array}$$

(7)

In the above formula, $far{m}_{it}$ is the moderating variable, ${\varphi }_2$ is its coefficient, and the coefficient ${\varphi }_3$ of the interaction term $participat{e}_{it}\times far{m}_{it}$ is the moderating effect, which is used to measure the moderating role of whether the village has carried out high-standard farmland construction and improvement projects on the relationship between farmers’ participation in industrial integration and income. ${\varphi }_0$ is the constant term, ${\varphi }_k$ is the coefficient of the control variable, ${\omega }_{it}$ is the random error term.

4. Empirical Results and Analysis

4.1. Descriptive Statistics of Variables

4.1.1. Full Sample Characteristics

To intuitively observe the situation of sample data and the basic characteristics of each variable, we first conduct descriptive statistical analysis on the full sample, as shown in

Table 2. Descriptive Statistical Analysis.

Variable	Count	Mean	Std	Min	Max
income	924	53,083.961	29,038.350	1200	300,000
Ln_income	924	10.728	0.603	7.090	12.612
participate	924	0.462	0.499	0	1
chain	924	0.029	0.093	0	1
multi	924	0.302	0.459	0	1
func	924	0.240	0.427	0	1
tech	924	0.018	0.134	0	1
gender	924	0.595	0.491	0	1
age	924	43.574	11.592	19	78
edu	924	7.206	4.876	0	19
family	924	4.318	2.017	1	12
area	924	1.757	4.453	0	60
plots	924	1.530	3.767	0	70
homestead	924	1.118	0.417	0	3
cpc	924	0.081	0.273	0	1
leader	924	0.047	0.211	0	1
heard	924	0.392	0.500	0	1
farm	924	0.472	0.499	0	1

.

According to the results of the descriptive statistical analysis presented in Table 2, the mean value of annual household income of farmers in the sample is CNY 53,083.961, with a standard deviation of CNY 29,038.35. A relatively large standard deviation indicates that income differences among farmers are relatively significant. Logarithmic transformation is often used to make the data more consistent with the normal distribution to facilitate analysis.

In terms of core explanatory variables, approximately 46.2% of farm households have participated in at least one form of rural industrial integration. Further analysis shows that farm households’ participation exhibits significant heterogeneity: among them, internal multi-business integration and function expansion-oriented integration are the main forms of participation, while the participation rates of industrial chain extension-oriented and technology penetration-oriented integration are relatively low.

In terms of control variables, farm household groups are predominantly male, accounting for nearly 60%, with an average age of 43.57 years, an average of 7.20 years of education, and an average household size of 4.32 people. In terms of land management, the mean value of own cultivated land area is 1.76 mu, but its standard deviation is as high as 4.45 mu, which indicates that the distribution of land management area is extremely dispersed; the mean number of plots is 1.53, with a standard deviation of 3.77, which also reflects the discrete characteristics of plot distribution. In terms of social capital, the proportion of Communist Party of China (CPC) members is 8.1%, and the proportion of those who have served as village cadres is 4.7%.

For the instrumental variable, 39.2% of farm households have heard of village collective economic organizations; the moderating variable “high-standard farmland construction” covers 47.2% of villages. These variable characteristics lay a solid foundation for addressing endogeneity and analyzing moderating effects.

4.1.2. Comparison of Income Gaps Between Participating and Non-Participating Groups

Before conducting regression analysis, to intuitively determine whether there are differences in income between farmers who participate in industrial integration and those who do not, as well as the income status of farmers under different participation modes, this study uses the original values of farmers’ per capita income before taking logarithms for analysis.

According to the differences in annual income and results of the t-test regarding whether farmers participate in industrial integration and under different participation modes in

Table 3. Differences in farmers’ income in participation in industrial integration and under different modes.

Participation Mode	Participating Group	Non-Participating Group	Difference	t-Value	p-Value
participate	65,885.902	42,085.111	23,800.791	13.604	0.000
chain	100,844.444	51,646.355	49,198.090	9.046	0.000
multi	60,172.043	50,017.953	10,154.090	4.941	0.000
func	69,548.108	47,877.350	21,670.758	10.221	0.000
tech	64,729.412	52,865.689	11,863.723	1.671	0.095

, overall, the annual income of farmers participating in industrial integration is significantly higher than that of those not participating. Specifically, in the modes of industrial chain extension-oriented integration, internal multi-business integration, and function expansion-oriented integration, there are significant differences in annual income between participating and non-participating farmers. The income level of the participating group is higher, and it is statistically significant at the 1% significance level; In the technology penetration-oriented integration mode, although its significance level is lower than that of the previous three integration modes, it is significant at the 10% significance level and thus still statistically significant. This result shows that most industrial integration modes can effectively increase farmers’ income, while the effect of technology penetration-oriented integration in improving farmers’ income still requires further in-depth exploration.

4.2. Baseline Regression Results

This study aims to conduct an in-depth investigation into the impact of farmers’ participation in rural industrial integration on improving their income levels. It analyzes based on micro balanced panel data obtained from household surveys conducted on farmers in Jiangxi Province in 2021 and 2023. First, a multiple linear regression model (Model 1) is used for analysis. This model can intuitively show the correlation between variables, quickly verify whether there is an association between farmers’ participation in industrial integration and income, and thereby provide a preliminary basis for the in-depth analysis of the subsequent fixed effects model.

Although multiple linear regression analysis can reveal the correlation between variables, it has certain limitations in addressing endogeneity issues caused by individual heterogeneity. By controlling for individual fixed effects, the fixed effects model eliminates confounding factors that do not change over time, focusing only on the impact of changes in the same farmer’s participation in industrial integration at different time points on income, thereby upgrading correlation analysis to more reliable causal inference. Therefore, after conducting multiple linear regression, this paper further adopts the time and individual two-way fixed effects model to analyze the income effect of farmers’ participation in industrial integration (Model 2), and based on the fixed effects model, further analyzes the impact of farmers’ participation in different industrial integration modes on income. Models 3–6 are sequentially the time and individual two-way fixed effects regression results for four modes: farmers’ participation in industrial chain extension-oriented integration mode, internal multi-business integration mode, function expansion-oriented integration mode, and technology penetration-oriented integration mode.

As shown in

Table 4. Regression analysis results of farmers’ different participation modes.

Variable	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
participation	0.474 *** (0.035)	0.286 *** (0.039)
chain			0.652 *** (0.115)
multi				0.186 *** (0.042)
func					0.199 *** (0.044)
tech						0.020 (0.137)
gender	0.098 *** (0.034)	0.245 * (0.131)	0.271 ** (0.134)	0.296 ** (0.136)	0.224 (0.136)	0.278 ** (0.139)
age	−0.015 *** (0.001)	−0.032 *** (0.004)	−0.031 *** (0.004)	−0.032 *** (0.004)	−0.033 *** (0.004)	−0.033 *** (0.005)
edu	0.016 *** (0.004)	0.013 *** (0.004)	0.008 ** (0.004)	0.014 *** (0.004)	0.009 ** (0.004)	0.011 *** (0.004)
family	−0.011 (0.008)	−0.026 ** (0.010)	−0.007 (0.010)	−0.018 * (0.011)	−0.009 (0.010)	−0.004 (0.010)
area	0.011 ** (0.005)	0.018 ** (0.007)	0.018 ** (0.008)	0.020 ** (0.008)	0.017 ** (0.008)	0.019 ** (0.008)
plots	−0.014 ** (0.006)	−0.007 (0.005)	−0.002 (0.006)	−0.006 (0.006)	−0.002 (0.006)	−0.003 (0.006)
homestead	−0.175 *** (0.039)	−0.049 (0.045)	0.003 (0.046)	−0.030 (0.047)	−0.004 (0.046)	−0.000 (0.047)
cpc	0.229 *** (0.081)	0.142 * (0.085)	0.280 *** (0.085)	0.220 ** (0.087)	0.227 *** (0.087)	0.274 *** (0.088)
leader	0.209 ** (0.103)	0.600 * (0.313)	0.599 * (0.320)	0.523 (0.324)	0.564 * (0.323)	0.589 * (0.331)
Constant	11.230 *** (0.096)	11.862 *** (0.209)	11.803 *** (0.214)	11.819 *** (0.216)	11.874 *** (0.216)	11.839 *** (0.221)
R2	0.362	0.265	0.234	0.214	0.216	0.179
N	924	924	924	924	924	924

Note: *, ** and *** denote significance at the 10%, 5% and 1% levels, respectively, with standard errors in parentheses.

, in Model 1, the coefficient of the variable “whether to participate in industrial integration” is 0.474, and it is significant at the 1% significance level. This indicates that farmers’ participation in industrial integration can significantly increase their income levels, and promoting the development of rural industrial integration is an effective measure to boost farmers’ income. The coefficient of Model 2 drops to 0.286 (p < 0.01), remaining highly significant. The reason for this decline is that the fixed effects model controls for individual time-invariant characteristics (such as personal ability, family background) and time trends, eliminating potential omitted variable bias and endogeneity bias in Ordinary Least Squares (OLS), thereby enhancing the credibility of the core coefficient. This means that the actual impact of participating in industrial integration on income is approximately 28.6%, i.e., farmers participating in industrial integration have an income about 28.6% higher than non-participants, which is close to a causal relationship, and Hypothesis 1 is verified.

At the level of segmented modes, the coefficient of industrial chain extension-oriented integration in Model 3 is 0.652 (p < 0.01), ranking the highest among the four modes. This indicates that participating in industrial chain extension has the most significant effect on increasing farmers’ income. The reason is that industrial chain extension can directly enhance the added value of agricultural products and broaden sales channels. The coefficient of internal multi-business integration in Model 4 is 0.186 (p < 0.01), which is significant but with a relatively weak effect. This means that internal multi-business integration helps farmers achieve synergetic development of multiple businesses, thereby broadening income channels. However, this mode is significantly affected by scale or market factors, and it is necessary to further explore the potential functions of agriculture in the future. The coefficient of function expansion-oriented integration in Model 5 is 0.199 (p < 0.01), which has a similar effect to multi-business integration. It reflects the marginal contribution of expanding agriculture to ecological and cultural functions to farmers’ income, but this mode is restricted by regional resources or market demand. The coefficient of the technology penetration-oriented integration mode in Model 6 is 0.020 (p > 0.1), which does not reach a significant level. This indicates that the effect of technology penetration-oriented integration in promoting farmers’ income increase has not been fully reflected. The reasons include the high threshold for farmers’ technology adoption, high short-term investment costs, or limited popularization of technology, which have not yet transformed applications into actual outputs. To sum up, Hypotheses 2a–2d are verified.

In the dimension of the impact of control variables on farmers’ income: from the perspective of individual characteristics, the gender coefficient is positive and highly significant in multiple models, indicating that male farmers have certain advantages in income acquisition compared with female farmers. Based on this, training and support programs for female farmers can be carried out. In all models, the age coefficient is negative and highly significant, meaning that farmers’ income decreases with age. The years of education coefficient is positive and significant in all models, which verifies that human capital promotes agricultural production efficiency. Therefore, investment in rural education should continue to be increased.

From the perspective of family characteristics, the coefficients of the number of family members are mostly insignificant, indicating that the number of family members has no significant impact on farmers’ annual income. This may be due to excessive family population leading to the dilution of per capita resources or low efficiency of labor allocation. In all models, the coefficients of the area of self-owned cultivated land operated are mostly positive and significant, indicating that for each 1-unit increase in cultivated land scale, income increases by 1.1–2.0%, reflecting the economies of scale effect in agricultural production. The coefficients of the number of plots and the number of homestead plots are both negative, but they are insignificant in the fixed effects models. Scattered plots are not conducive to large-scale operation, and excessively scattered homesteads also occupy production resources, which will have a negative impact on farmers’ income. However, the specific degree of impact needs further exploration.

At the level of social capital, the coefficient of the variable of Communist Party of China (CPC) membership is positive and significant. This is because party members have certain advantages in information acquisition, policy support, etc., and can give full play to the leading and exemplary role of party member farmers. The variable of having served as a village cadre has a significantly positive coefficient in some models, and the effect is more prominent especially in fixed effects models, which is related to the higher participation of village cadres in industrial integration projects. The reason is that the social networks and resource allocation capabilities brought by the experience as a village cadre can increase income. Therefore, experienced village cadres should be encouraged to engage in rural industrial development.

Overall, farmers’ participation in industrial integration can significantly increase their income levels at the overall level. Different industrial integration modes show differences in their impact on farmers’ income. Among them, the integration effects of industrial chain extension-oriented, internal multi-business integration, and function expansion-oriented modes are relatively significant, while the integration effect of the technology penetration-oriented mode is not yet significant. At the same time, control variables such as individual characteristics, family characteristics, and social capital have also exerted varying degrees of influence on farmers’ income.

4.3. Interaction Effects of Participating in Multiple Industrial Integration Modes

From the above analysis, it can be seen that except for farmers’ participation in the technology penetration-oriented integration mode, which has no significant impact on income, participating alone in any one of the industrial chain extension-oriented, internal multi-business integration, or function expansion-oriented industrial integration modes can achieve income growth. This raises the question: if farmers participate in two or three of these three integration modes simultaneously, will it lead to more significant income growth? Next, this paper will further explore the income effect of farmers’ simultaneous participation in multiple integration modes by introducing interaction terms and using the time and individual fixed effects model. It is mainly divided into four scenarios: simultaneous participation in industrial chain extension-oriented and internal multi-business integration modes (Model 7), simultaneous participation in industrial chain extension-oriented and function expansion-oriented modes (Model 8), simultaneous participation in internal multi-business integration and function expansion-oriented integration modes (Model 9), and simultaneous participation in all three integration modes (Model 10).

From the results of Models 7–10 in

Table 5. Income effects of farmers’ simultaneous participation in multiple industrial integration modes.

Variable	Model 7	Model 8	Model 9	Model 10
chain	0.814 ***	0.534 ***		0.571 ***
	(0.152)	(0.155)		(0.122)
multi	0.172 ***		0.262 ***	0.171 ***
	(0.041)		(0.048)	(0.040)
func		0.164 ***	0.291 ***	0.185 ***
		(0.044)	(0.053)	(0.042)
Chain × multi	−0.430 **
	(0.198)
Chain × func		0.105
		(0.190)
Multi × func			−0.233 ***
			(0.090)
Chain × multi × func				−0.244
				(0.248)
Control for individual characteristics	Yes	Yes	Yes	Yes
Control for family characteristics	Yes	Yes	Yes	Yes
Control for social capital	Yes	Yes	Yes	Yes
Individual fixed effects	Yes	Yes	Yes	Yes
Time fixed effects	Yes	Yes	Yes	Yes
Constant	11.866 ***	11.857 ***	11.864 ***	11.815 ***
	(0.212)	(0.214)	(0.210)	(0.207)
R2	0.266	0.260	0.264	0.289
N	924	924	924	924

Note: ** and *** denote significance at the 5% and 1% levels, respectively, with standard errors in parentheses.

, it can be seen that after incorporating interaction terms of multiple mode combinations, farmers’ participation in any single industrial integration mode can significantly increase their income levels. However, simultaneous participation in multiple industrial integration modes does not show better income-increasing effects compared with participating in a single mode alone. For example, the interaction term coefficient of industrial chain extension-oriented and function expansion-oriented integration in Model 8 is 0.105, which is not significant, indicating that the two have not formed a significant synergistic effect. In addition, simultaneous participation in multiple industrial integration modes may even have negative impacts. For instance, the interaction term coefficient of industrial chain extension and internal multi-business integration is −0.430, significant at the 5% level; the interaction term coefficient of internal multi-business and function expansion-oriented integration is −0.233, significant at the 1% level. This is because when farmers participate in industrial integration, the resources they can invest, such as land, labor, funds, time, and energy, are limited. Simultaneous participation in multiple modes may disperse the core resources of farmers’ production, leading to loss of management efficiency and thus reduced efficiency in their main business. Therefore, in the process of participating in rural industrial integration, farmers should focus on one industrial integration mode and avoid blindly superimposing multiple modes. Hypothesis 3 is verified.

4.4. Treatment of Endogeneity with Instrumental Variable

When exploring the impact of farmers’ participation in industrial integration on their income, the existence of endogeneity issues will significantly interfere with the accuracy of causal inference. The endogeneity in this study mainly stems from reverse causality: it may not be industrial integration that drives income growth, but rather farmers with higher income levels, relying on more abundant funds, resources, and stronger risk-bearing capacity, are more qualified to participate in industrial integration projects. Instrumental variable analysis is an effective way to address endogeneity issues. Its core logic is to find a variable that is highly correlated with the endogenous variable (whether to participate in industrial integration) but uncorrelated with the error term (including unobserved confounding factors). In this study, “whether one has heard of the collective economic organization in the village” is used as an instrumental variable, which theoretically meets these two conditions: on the one hand, farmers who have heard of the collective economic organization are more likely to obtain information related to industrial integration, thereby increasing their participation probability, satisfying relevance; on the other hand, this variable usually does not directly affect farmers’ income, but exerts an indirect effect by influencing participation decisions, satisfying exclusivity. With the instrumental variable method, the part of the endogenous variable that is correlated with the error term can be isolated, thereby obtaining more accurate estimates of causal effects, ensuring the reliability of research conclusions and the effectiveness of policy recommendations.

As shown in

Table 6. Results of two-stage regression by instrumental variable method (IV-2SLS).

Variable	Model 11	Model 12
Explained variable	participation	Ln_income
Heard	0.650 *** (0.045)	-
Participation (predicted)	-	0.263 *** (0.075)
Control for individual characteristics	Yes	Yes
Control for family characteristics	Yes	Yes
Control for social capital	Yes	Yes
Constant	−0.206 (0.202)	11.862 *** (0.217)
R2	0.525	0.212
First-stage f-statistic	45.36	-
N	924	924

Note: *** denotes significance at the 1% level, with standard errors in parentheses.

, Model 11 is the first-stage regression of the instrumental variable method (IV-2SLS), i.e., the results verifying the correlation between the instrumental variable and the endogenous variable. The coefficient of the variable “whether one has heard of the collective economic organization in the village” is 0.65, indicating that the probability of farmers who have heard of the village’s collective economic organization participating in industrial integration is 65% higher than that of farmers who have not, and this correlation is significant at the 1% significance level. This indicates that collective economic organizations, as hubs for policy transmission, their information penetration capacity directly affects farmers’ perception of industrial integration and their participation decisions. The first-stage f-statistic is 45.36, which is far greater than the empirical threshold of 10, strongly rejecting the “weak instrumental variable” hypothesis and verifying that there is a strongly significant positive correlation between the instrumental variable “whether one has heard of the village’s collective economic organization” and the endogenous variable “farmers’ participation in industrial integration”.

Model 12 is the second-stage regression of the instrumental variable method, i.e., the regression results of industrial integration on the logarithm of annual income. The results show that after controlling for characteristics such as farmers’ individual, household, and social capital, farmers’ participation in industrial integration has a significant positive impact on the logarithm of their income, with a coefficient of 0.263, significant at the 1% level. This implies that participating in industrial integration can increase farmers’ income by an average of about 26.3%, indicating that farmers’ participation in rural industrial integration can significantly increase their income.

4.5. Robustness Tests

To further test the income effect of farmers’ participation in industrial integration, this study adopts three methods—winsorization (Model 13), excluding some samples (Model 14), and replacing the core explanatory variable (Model 15)—to conduct robustness tests.

Winsorization involves performing 1% winsorization on both the left and right tails of the original values of farmers’ annual income, followed by logarithmic transformation. In the basic regression process, to most truly reflect the income effect of farmers’ participation in industrial integration, this paper did not perform winsorization on the explained variable. However, there are indeed extreme values in the data. For example, some farmers have an annual income of CNY 3,000,000,000. These extreme values may result from data entry errors, special occasional events, or individual abnormal factors. In addition, some farmers may obtain extremely high incomes due to one-time huge subsidies or rare business opportunities. Such extremely high income values will have a significant impact on the overall statistical analysis results, leading to potential biases in conclusions drawn from ordinary statistical methods, which fail to accurately reflect the actual situation of most farmers. Similarly, extremely low income values may also interfere with the analysis. Through 1% winsorization, the impact of these extreme values can be controlled within a certain range, making the data more in line with the actual distribution.

In the process of excluding some samples, village samples from Xinjian District and Jinxian County under Nanchang City, the capital of Jiangxi Province, were excluded. As a provincial capital, Nanchang has a more developed transportation network and a broader market, which makes the opportunities and challenges faced by local farmers in participating in industrial integration different from those in other regions, and thus leads to differences in their income change mechanisms compared with other regions. If the village samples under Nanchang are included in the overall analysis, the research results may be interfered by these special circumstances, making it impossible to accurately reflect the impact of farmers’ participation in industrial integration on income under general circumstances. By excluding these samples to conduct robustness tests, we can verify whether the conclusions previously drawn based on the full sample still hold. That is, after excluding the impact of Nanchang’s special circumstances, we can judge whether the relationship between farmers’ participation in industrial integration and income has stability and universality, thereby enhancing the reliability and persuasiveness of the research conclusions.

Replacing the core explanatory variable: the core explanatory variable “whether to participate in industrial integration” is adjusted from originally participating in any of the four integration modes to participating in any of the two modes: internal multi-business integration or function expansion-oriented integration. This is to test whether the original research conclusions are accidental results caused by including the two integration modes with fewer participants: industrial chain extension-oriented integration and technology penetration-oriented integration, and further confirm whether the previous research results based on the definition of mainstream industrial integration participation modes have robustness. The results of the robustness test are shown in

Table 7. Results of robustness test.

Variable	Model 13	Model 14	Model 15
participate	0.263 *** (0.036)	0.270 *** (0.042)	0.201 *** (0.028)
Control for individual characteristics	Yes	Yes	Yes
Control for family characteristics	Yes	Yes	Yes
Control for social capital	Yes	Yes	Yes
Individual fixed effects	Yes	Yes	Yes
Time fixed effects	Yes	Yes	Yes
Constant	11.664 *** (0.188)	11.808 *** (0.404)	11.853 *** (0.400)
R2	0.278	0.265	0.253
N	924	864	924

Note: *** indicates significance at the 1% significance level, with standard errors in parentheses.

.

As shown in Table 7, Models 13–15 all controlled for individual characteristics, household characteristics, and social capital, and incorporated individual fixed effects and time fixed effects. The coefficients of the core explanatory variable “participation in industrial integration” are all positive at the 1% significance level, indicating that regardless of the robustness treatment method adopted, farmers’ participation in rural industrial integration significantly increases their income levels. The conclusion that “farmers’ participation in rural industrial integration can increase income” holds consistently, and there have been no substantial changes in the significance level. Combined with strictly specified control variables and fixed effects, the research results support the core hypothesis that “participation in rural industrial integration has a positive promoting effect on farmers’ income”. Based on the results of endogeneity treatment and robustness tests using the aforementioned instrumental variable, Hypothesis 1 regarding the core research question of this study is further verified.

4.6. Moderation Effect Analysis

Moderation effect analysis, by introducing the interaction term between the core explanatory variable and the moderator variable “a high-standard farmland construction and improvement project has been implemented in the village”, aims to examine whether the causal effect of the core explanatory variable varies with different levels of the moderator variable. For example, whether the impact of industrial integration on income is more significant in areas with high-standard farmland. At the theoretical level, this analysis helps deepen the understanding of the boundary conditions of causal mechanisms and reveal the context-dependent characteristics of the impact of independent variables on dependent variables; At the practical level, it can identify heterogeneous groups or scenarios of policy effects and provide support for targeted policy design.

As shown in

Table 8. Regression results of the moderation effect of high-standard farmland.

Variable	Model 16	Model 17	Model 18
participate	0.200 ***	0.355 ***	0.166 ***
	(0.041)	(0.070)	(0.049)
farm	0.000
	(omitted)
Participate × farm	0.135 *
	(0.071)
Control for individual characteristics	Yes	Yes	Yes
Control for family characteristics	Yes	Yes	Yes
Control for social capital	Yes	Yes	Yes
Individual fixed effect	Yes	Yes	Yes
Time fixed effects	Yes	Yes	Yes
Constant	10.631 ***	10.600 ***	10.665 ***
	(0.020)	(0.033)	(0.025)
R2	0.781	0.746	0.811
N	924	436	488

Note: * and *** denote statistical significance at the 10% and 1% levels, respectively, with standard errors in parentheses.

, Model 16 is a moderation effect model containing an interaction term, Model 17 is the high-standard farmland group, and Model 18 is the group without high-standard farmland construction projects in the village. In these three models, the coefficient of “participation in industrial integration” is significantly positive at the 1% significance level, which once again confirms that farmers’ participation in rural industrial integration can significantly improve their income level.

In Model 16, the coefficient of the interaction term “participation in industrial integration × high-standard farmland construction” is 0.135, which is significant at the 10% significance level, indicating that high-standard farmland construction has a positive moderating effect on the relationship between “participation in industrial integration and improvement of farmers’ income”. That is to say, when villages carry out high-standard farmland construction, the promoting effect of participating in industrial integration on farmers’ income will be strengthened. The grouped regression results of Model 17 and Model 18 further verify this conclusion. In the village group represented by Model 17, which has implemented the high-standard farmland construction and improvement project, the coefficient of participation in industrial integration is 0.355, which is significantly larger than the coefficient of 0.166 in Model 18. This indicates that in villages with high-standard farmland, the promoting effect of participating in industrial integration on farmers’ income is more significant. The mechanism of this moderating effect lies in that high-standard farmland reduces the initial investment costs of industrial integration and enhances the potential of large-scale land management through measures such as land consolidation and improvement of water conservancy facilities. Farmers participating in industrial integration are more likely to achieve resource integration through land transfer, cooperatives and other forms, thereby forming a virtuous cycle effect of “improvement of agricultural production infrastructure—participation in industrial integration—income growth”, and Hypothesis 4 is verified.

4.7. Heterogeneity Analysis

4.7.1. Differences in the Income-Increasing Effect of Participating in Industrial Integration on Farmers with Different Income Levels

Table 9. Results of quantile regression.

Variable	Model1 9	Model 20	Model 21	Model 22	Model 23	Model 24	Model 25	Model 26	Model 27
participate	0.522 *** (0.087)	0.399 *** (0.049)	0.381 *** (0.046)	0.411 *** (0.044)	0.407 *** (0.038)	0.410 *** (0.035)	0.385 *** (0.034)	0.360 *** (0.035)	0.382 *** (0.045)
Control for individual characteristics	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Control for family characteristics	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Control for social capital	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Individual fixed effects	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Time fixed effects	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Constant	10.853 *** (0.239)	10.930 *** (0.135)	11.088 *** (0.127)	11.090 *** (0.120)	11.141 *** (0.104)	11.043 *** (0.096)	11.126 *** (0.093)	10.950 *** (0.097)	11.046 *** (0.124)
Pseudo R2	0.229	0.221	0.203	0.209	0.205	0.220	0.215	0.231	0.259
N	924	924	924	924	924	924	924	924	924

Note: *** indicates significance at the 1% significance level, with standard errors in parentheses.

and Figure 4 present the estimation results of the full-quantile regression of the income effects of farmers with different incomes participating in industrial integration. Models 19–27, respectively, show the impacts of participating in industrial integration on the income of farmers at the 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, and 90% quantile levels.

All models consistently indicate that the variable “whether to participate in industrial integration” has a consistently positive impact on the logarithm of farmers’ income, and is significant at the 1% significance level. This implies that the income-increasing effect of industrial integration is highly robust in a statistical sense, and is not affected by income quantiles or sample variation. It can be seen that participating in industrial integration has an income-increasing effect on farmers with different income levels. However, the income-increasing effect shows a decreasing trend as the income quantile increases, decreasing from 0.522 at the 10% quantile to 0.382 at the 90% quantile. This result indicates that the income-increasing margin of industrial integration is more significant for low-income farmers, while its marginal contribution to high-income farmers is relatively limited.

As shown in Figure 4, the results of the full-quantile regression of the impact of participating in industrial integration on farmers’ income clearly reveal the heterogeneous characteristics of the income-increasing effect of industrial integration, that is, industrial integration can not only generally benefit farmers at different income levels, but also have a more “breakthrough” impact on low-income groups. This provides empirical support for the targeted formulation of industrial integration policies and the promotion of rural revitalization.

4.7.2. Analysis of Income Differences Among Farmers in Different Regions in Participating in Industrial Integration

There are certain differences in resource endowments, industrial structures, and market environments among different regions in Jiangxi Province, and these differences may profoundly affect the effect of industrial integration on farmers’ income. From the perspectives of geography and natural sciences, southern Jiangxi, central Jiangxi, and northern Jiangxi have significant differences in natural endowments such as geographical location and climatic conditions. Therefore, this study conducts a detailed analysis of different regions from the perspectives of southern Jiangxi, central Jiangxi, and northern Jiangxi, which helps policymakers formulate policies in accordance with local conditions. In the sample, Dayu County and Ruijin City belong to southern Jiangxi, with a total of 234 samples; counties such as Wanan, Fengxin, Yongxin, Luxi, and Zixi belong to central Jiangxi, with a total of 368 samples; Pengze, Xinjian, Fuliang, Yushan, and Jinxian belong to northern Jiangxi, containing a total of 322 samples. In

Table 10. Regional differences in the income effect of farmers’ participation in industrial integration.

Variable	Model 28	Model 29	Model 30
participate	0.260 *** (0.090)	0.344 *** (0.060)	0.239 *** (0.063)
Control for individual characteristics	Yes	Yes	Yes
Control for family characteristics	Yes	Yes	Yes
Control for social capital	Yes	Yes	Yes
Individual fixed effects	Yes	Yes	Yes
Time fixed effects	Yes	Yes	Yes
Constant	10.413 *** (0.480)	12.513 *** (0.404)	10.797 *** (0.855)
R2	0.257	0.451	0.202
N	234	368	322

Note: *** indicates significance at the 1% significance level, with standard errors in parentheses.

, Models 28–30, respectively, reflect the income effects of farmers’ participation in industrial integration in the three regions of southern Jiangxi, central Jiangxi, and northern Jiangxi.

The research results show that industrial integration has a significantly positive promoting effect on farmers’ income in all regions of Jiangxi Province (p < 0.01), but the intensity of its effect presents obvious regional heterogeneity characteristics: the elasticity coefficient in central Jiangxi is the highest (0.344), which is significantly higher than that in southern Jiangxi (0.269) and northern Jiangxi (0.239). This gradient difference reflects the interaction between the initial development level, institutional environment, and factor endowments. Central Jiangxi, with the Nanchang Metropolitan Circle as its core, has a more solid economic foundation and more sophisticated market-oriented systems. Industrial integration can more effectively stimulate factor reorganization and innovation spillover effects, thereby generating a stronger income-increasing effect. Overall, in the three regions of southern Jiangxi, central Jiangxi, and northern Jiangxi, the variable of whether farmers participate in industrial integration has a significantly positive impact on the logarithm of farmers’ annual income, indicating that participating in industrial integration can indeed increase farmers’ income.

4.7.3. Analysis of Income Differences Among Farmers Participating in Industrial Integration in Different Terrain Types

Under different topographic conditions, there are significant differences in the distribution of land resources, climatic conditions, and water resources. Counties and districts such as Dayu, Yongxin, Fuliang, Yushan, and Zixi in the sample belong to mountainous terrain, where the terrain is relatively undulating and arable land shows fragmented characteristics, which to some extent restricts the development of large-scale mechanized agricultural production. However, this region has unique mountain forest resources, which are suitable for developing industries such as characteristic fruit and forestry, and under-forest economy. Counties such as Wanan, Fengxin, Luxi, and Ruijin are located in hilly areas, where the terrain is relatively gentle and land resources have a certain degree of diversity. Such terrain is suitable not only for developing economic fruit and forestry but also for carrying out small-scale agricultural breeding activities. However, compared with plain areas, large-scale production in these areas is still subject to certain constraints, and diversified operation models can be implemented. Counties and districts such as Xinjian, Jinxian, and Pengze are of plain terrain with flat and open land, which is conducive to the development of industries such as large-scale grain cultivation and aquaculture. These different agricultural production methods will have an impact on farmers’ participation in industrial integration, and thus affect farmers’ income. Models 31–33 are sequentially the analysis of the income effects of farmers’ participation in industrial integration in plain, hilly, and mountainous terrains.

The results of Models 31–33 in

Table 11. Income effects of farmers’ participation in industrial integration in different terrain types.

Variable	Model 31	Model 32	Model 33
participate	0.268 *** (0.082)	0.226 *** (0.064)	0.325 *** (0.062)
Control for individual characteristics	Yes	Yes	Yes
Control for family characteristics	Yes	Yes	Yes
Control for social capital	Yes	Yes	Yes
Individual fixed effects	Yes	Yes	Yes
Time fixed effects	Yes	Yes	Yes
Constant	10.982 *** (0.178)	10.887 *** (0.323)	12.353 *** (0.288)
R2	0.242	0.249	0.340
N	162	312	450

Note: *** indicates significance at the 1% significance level, with standard errors in parentheses.

indicate that participation in industrial integration has a significantly positive impact on farmers’ income in plain, hilly, and mountainous areas, with coefficients of 0.268, 0.226, and 0.325, respectively, all significant at the 1% significance level. Converting the logarithmic coefficients to actual income growth rates using the exponential function, the corresponding actual income growth rates are approximately 30.7%, 25.4%, and 38.4%, respectively. The income effect in mountainous terrain is the most significant, which is due to the abundant mountain forest resources, prompting farmers to develop composite integration models such as characteristic fruits and forests, and under-forest economy. Plain areas, relying on large-scale production, have the second-highest income effect; hilly areas, constrained by land fragmentation, have a relatively weaker income effect. After the models control for individual characteristics, household characteristics, social capital, and dual fixed effects, the goodness of fit of the mountain model is higher than that of the plain and hilly area models, and due to the largest sample size in mountainous areas, its results are more stable. Through the above heterogeneity analysis, Hypotheses 5–7 are verified.

5. Discussion and Conclusions

5.1. Discussion

(1)

Marginal Contribution: This study supplements the micro-level empirical evidence on industrial integration in major agricultural provinces in central and western China, and reveals the transmission mechanism of “factor reorganization–model selection–income effect”. The study finds that participating in industrial integration increases farmers’ per capita net income by 23.6%, which is consistent with the research conclusions on industrial integration by Xie et al. (2025) and Tschora & Cherubini (2020) [25,34], that is, industrial integration can significantly improve farmers’ income by optimizing factor allocation. The negative interaction effect of multiple model superposition indicates that smallholder farmers face a “risk of resource dilution” under capital and labor constraints. In this case, participating in multiple modes will lead to the dispersion of resources among incompatible activities. For example, if farmers carry out both aquaculture (in—agricultural integration) and drone—based plant protection (technology—penetrating integration) simultaneously, they may neglect aquaculture management due to the energy invested in learning drone operation, resulting in a decline in aquatic product yields. This logic also explains why the interaction term in Table 5 is significantly negative, which is consistent with the conclusion of Mzyece & Ng’ombe (2020) that multi-activity operations of African smallholders lead to a decline in management efficiency [31]. This breaks the traditional perception that “the more integration models, the better” and provides a new perspective for understanding farmers’ optimal decision-making under resource constraints. From the analysis of the regulatory mechanism, high-standard farmland construction enhances agricultural production efficiency by improving the level of agricultural infrastructure, thereby promoting farmers’ participation in industrial integration and thus boosting income growth. This is basically consistent with the logic put forward by Zhang & Fan (2023) that the digital transformation of agriculture increases income through improving production efficiency [2]. The study provides micro-level evidence for the income growth through industrial integration in central agricultural provinces. It not only verifies the consensus in the international academic community that “industrial integration promotes rural sustainable development” but also reveals the particularities of Jiangxi region, providing a basis for the formulation of differentiated policies.

(2)

Practical Implications: Analysis of regional heterogeneity shows that, unlike the eastern large-scale agricultural areas studied by Li et al. (2024) [15], the characteristics of resource fragmentation in the hilly and mountainous areas of Jiangxi have prompted farmers to tend to choose integration models with low input thresholds and high dependence on local resources (such as under-forest economy and characteristic breeding and cultivation), which verifies the regional adaptability of industrial integration models proposed by Ye et al. (2023) [4]. The income effect of industrial integration in mountainous areas is higher than that in plain areas, which is consistent with the findings of Su et al. (2023) on urban–rural industrial integration in Xinjiang [13], who hold that mountain resources can offset geographical disadvantages through premiums from characteristic industries. From the perspective of participating in industrial integration models, farmers participating in industrial chain extension-oriented integration have the most significant income growth. This is consistent with the conclusions of Luo et al. (2023) and Chen et al. (2022) that agricultural-tourism integration drives the growth of non-agricultural income through industrial chain extension [67,68], and also highly aligns with the case of income increase through the “agriculture + aromatic crops” value chain extension model by Khan & Verma (2018) in India [7], This advantage stems from two core mechanisms: directly increasing added value and expanding market channels (such as selling agricultural products via e-commerce). This indicates that the industrial chain extension-oriented integration model can be a priority choice for farmers to participate in rural industrial integration. The ineffectiveness of technology-penetrating integration is attributable not only to the prevalent low educational attainment among the agricultural population (predominantly below junior high school level), which engenders digital divides and impedes technology diffusion, but also exhibits significant association with income heterogeneity. Specifically, farmers in the lowest income decile (10th percentile, Table 9) face dual constraints: they lack the capital for upfront investments in smart agricultural equipment (exemplified by drones exceeding ¥10,000) and subsequently lack the operational competencies for effective utilization post-adoption. Consequently, the anticipated short-term transformation pathway of “technology adoption → income growth” remains unrealized. This mechanistic explanation accounts for the statistically non-significant effects of technology-penetrating integration observed in the short-term panel data spanning 2021–2023. This finding suggests that when promoting technology-penetrating integration, we should fully consider farmers’ technology acceptance capability and local technology promotion conditions, and adopt targeted training and support measures to improve the actual effectiveness of technology-penetrating integration. Participation in industrial integration has a higher income growth elasticity for low-income groups, confirming the “pro-poor nature” of industrial integration. Therefore, smallholder farmers should be encouraged to actively participate in industrial integration, and relevant capacity building should be strengthened to ensure that they can fairly share the value-added benefits brought by industrial integration.

(3)

Research Limitations: The sample only includes longitudinal survey data from 462 rural households in 12 counties of Jiangxi Province over 2 years, which may not fully reflect the common characteristics of the central and western regions; Due to the limitations in the time span of micro panel data, the long-term effects of technology-penetrating integration require further tracking, and short-term data may underestimate its potential. In future studies, the sample scope will be expanded to cover a wider range of regions and data with a longer time span, so as to more comprehensively reveal the impact of industrial integration on farmers’ income and its dynamic change process. Furthermore, further in-depth exploration can be conducted on the impact of industrial integration on farmers’ agricultural operating income and non-agricultural income, as well as how to optimize the selection of industrial integration models through policy guidance, so as to provide more comprehensive and forward-looking theoretical support and policy recommendations for promoting the development of rural industrial integration.

5.2. Main Conclusions

This study is based on the “Baicun Qianhu” survey data in Jiangxi Province, and employs various econometric analysis methods such as fixed effect model, instrumental variable method, and quantile regression to systematically explore the impact of farmers’ participation in rural industrial integration on income and its heterogeneous characteristics. The main research conclusions are as follows:

(1)

Participation in industrial integration significantly promotes the growth of farmers’ income. Empirical research results show that, based on the two-way fixed effects model, farmers’ participation in rural industrial integration can increase their income by an average of 28.6%. Based on the fact that the average annual income of sample farmers is CNY 53,083.96, participation in industrial integration can increase farmers’ income by approximately CNY 15,182 on average. After addressing the endogeneity issue using the instrumental variable method, the causal effect is further confirmed, indicating that the participation behavior itself is a direct driving factor for the growth of farmers’ income. This means that industrial integration has had a substantive promoting effect on farmers’ income by restructuring agricultural production relations and expanding income-increasing channels, verifying the causal relationship of “industrial integration → income growth”.

(2)

There are significant differences in the income-increasing effects of different integration models. Among the four industrial integration models, the income-increasing effect of industrial chain extension integration is the most significant, with a coefficient of 0.652; followed by functional expansion integration and internal multi-format integration, with coefficients of 0.199 and 0.186, respectively, while the effect of technology penetration integration is insignificant. Such differences stem from variations in factor input thresholds, market docking capabilities, and risk-bearing requirements across different models. For instance, industrial chain extension directly connects to high-value-added links, whereas technology penetration requires long-term investment with insignificant short-term returns. Further analysis of interaction terms reveals that the superposition of multiple models has a synergistic inhibitory effect, meaning that farmers participating in multiple integration models simultaneously may lead to resource dispersion and reduced management efficiency. This indicates that farmers should focus on a single advantageous model and avoid blind diversified development.

(3)

The income-increasing effect of participating in industrial integration is characterized by context dependence and heterogeneity. High-standard farmland construction significantly enhances the income-increasing effect of industrial integration (interaction term coefficient = 0.135, p < 0.1). In villages implementing high-standard farmland projects, the income-increasing effect of industrial integration reaches 35.5%, which verifies the “enabling effect” of the improvement of agricultural production infrastructure on industrial integration, i.e., high-standard farmland construction amplifies the effect of industrial integration. In addition, the income effect of farmers’ participation in industrial integration shows obvious heterogeneous characteristics. From the income perspective, the income-increasing effect of industrial integration on low-income farmers is significantly higher than that on high-income farmers, indicating that it has positive significance in alleviating relative poverty and consolidating the achievements of poverty alleviation. From the regional perspective, the income-increasing effect in central Jiangxi is stronger than that in southern and northern Jiangxi, which is consistent with the region’s characteristics of a better economic foundation and a higher degree of marketization. From the aspect of terrain differences, mountainous areas have achieved the strongest income-increasing effect by relying on characteristic resources, followed by plain areas, while hilly areas have a relatively weak income-increasing effect due to the constraint of land fragmentation.

6. Policy Recommendations

6.1. Actively Guide Farmers to Participate in Rural Industrial Integration

6.1.1. Give Priority to Promoting Industrial Chain Extension—Type Integration

It is necessary to vigorously support the construction of an agricultural product e-commerce platform system covering the entire province, provide all-round support including logistics subsidies, packaging subsidies, etc., for farmers actively participating in online sales, and systematically carry out training on new sales skills such as short-video marketing and live-streaming commerce. Encourage qualified agricultural cooperatives to take the lead in integrating resources and build a complete closed-loop industrial chain system of “production—intensive processing—brand marketing—terminal sales”. For the Northern Jiangxi plain areas, give full play to their advantages in large-scale agriculture, with a focus on developing the construction of cold-chain logistics infrastructure and contract farming models; for the Southern and Central Jiangxi mountainous and hilly areas, it is necessary to highlight regional characteristics and focus on building a brand system of characteristic agricultural products, including cultural and creative packaging of Gannan navel oranges and selenium-rich agricultural products.

6.1.2. Steadily Advance Internal Multi-Business Format and Function-Expanding Integration

In hilly areas, promote integration models with low input costs and technical thresholds such as “rice–fish farming” and “under-forest planting” by adjusting measures to local conditions. Agricultural and rural affairs departments shall organize experts to compile easy-to-understand technical manuals and establish regular market docking service mechanisms. In suburban villages, innovatively advance the reform of the transfer of homestead use rights, improve the collective equity dividend system, and establish a diversified interest linkage mechanism of “farmers + village collectives + leading enterprises” to ensure that farmers can fairly and reasonably share the value-added benefits of the industry.

6.1.3. Optimize the Promotion Strategies for Technology-Penetrating Integration

Given the actual situation of the generally low educational level of farmers, adopt “hands-on” and “face-to-face” immersive technical training methods, and at the same time significantly reduce the threshold for using modern agricultural equipment such as drone pesticide application through government rental subsidies and other means. Give priority to laying out smart agriculture demonstration sites in concentrated areas of high-standard farmland, forming a progressive technology diffusion path of “technology display–farmer observation–practical application”.

6.1.4. Focus on Integration Models and Enhance Management Effectiveness

Give full play to the leading role of village collective economic organizations and agricultural cooperatives, guide farmers to focus on one industrial integration model based on actual conditions, and avoid excessive dispersion of resources. It is suggested that farmers comprehensively consider resource endowment conditions such as household labor structure and land management scale, and select one core business with the greatest development potential. Government departments can provide public welfare consulting services such as “adaptability assessment of integration models” to assist farmers in making scientific decisions.

6.2. Strengthen Agricultural Infrastructure Construction and Institutional Guarantees

6.2.1. Improve the Synergistic Effect of High-Standard Farmland

Integrate the special plan for industrial integration development into high-standard farmland construction projects in an organic manner. While carrying out traditional projects such as land leveling and the construction of irrigation and drainage facilities, simultaneously build supporting new facilities like primary processing workshops for agricultural products and e-commerce service stations, so as to create a trinity development pattern of modern agriculture featuring “high-standard farmland–processing base–e-commerce platform”. The Central Jiangxi region should fully utilize the location advantages of the Nanchang Metropolitan Circle, and focus on building an integrated development demonstration belt that combines high-standard farmland, leisure agriculture, and agritourism.

6.2.2. Perfect Support Policies for Smallholder Farmers

Provide special policy support for low-income farmers participating in industrial integration (such as reducing or exempting e-commerce platform entry fees, providing interest subsidies for entrepreneurship loans, etc.), establish and improve a long-term mechanism of paired assistance featuring “Party member demonstration households + ordinary smallholder farmers”, and give full play to the exemplary and leading role of Party member farmers in information acquisition, market connection, etc.

6.3. Implement Rural Industrial Integration Policies in Line with Local Conditions

The Central Jiangxi region should fully rely on the advantages of its sound economic foundation, focus on cultivating a number of industrial integration-oriented leading agricultural enterprises, and vigorously develop high-value-added emerging business forms such as “agriculture + cultural tourism” and “agriculture + health and wellness”. Southern Jiangxi region should closely integrate with policies for the revitalization and development of old revolutionary base areas, focusing on developing integrated projects that combine red tourism with characteristic breeding and cultivation; Northern Jiangxi region, on the other hand, should base itself on the advantages of Poyang Lake’s ecological resources and actively explore new paths for green and low-carbon integrated development such as “ecological agriculture + carbon sink trading”. For mountainous areas, efforts to support characteristic industries should be strengthened; for hilly areas, focus should be placed on promoting the regulation of land fragmentation and moderate-scale operation. For low-income farmers, such as special groups like female farmers and elderly farmers, we must focus on developing integrated projects with low labor intensity such as handicraft cultural and creative industries and the courtyard economy to ensure that all groups can fairly participate in the development of industrial integration.