Production Process Outsourcing, Farmers’ Operation Capability, and Income-Enhancing Effects

Abstract

Production process outsourcing not only enhances farmers’ operation capability but also contributes to income growth. Utilizing field survey data from five provinces—Inner Mongolia, Gansu, Ningxia, Henan, and Shaanxi—this study employs an endogenous switching regression model to analyze the impact of production process outsourcing on the enhancement of farmers’ operation capability and the income-enhancing effect. The results reveal the following: (1) Production process outsourcing significantly improves farmers’ operation capability and increases income. (2) A higher degree of adoption of production process outsourcing correlates with greater improvements in farmers’ operation capability. (3) The impact of production process outsourcing on farmers’ operation capability varies with individual endowments; farmers with higher education levels, a larger number of laborers, and smaller planting areas experience more pronounced improvements in management capabilities when participating in outsourcing. (4) Production process outsourcing partially mediates the income-enhancing effect through its influence on farmers’ operation capability. To further promote income growth, it is essential to enhance the agricultural outsourcing market supply system, expand farmers’ access to production service information, and prioritize the development of farmers’ operation capability.

1. Introduction

China’s agricultural landscape, characterized by numerous smallholder farmers, necessitates the seamless integration of these small-scale operations with modern agricultural practices to unlock their full developmental potential [1]. However, the significant outflow of young and able-bodied workers from rural areas has impeded the ability of these small farmers to adapt to the requirements of modern agricultural development [2,3]. Recent rapid advancements in production process outsourcing within China’s agricultural production chain have enabled farmers to access new machinery and technological innovations, thereby overcoming constraints related to farmland size and capital [4]. Theoretically, if smallholders can adeptly leverage the outsourcing market to address challenges related to labor efficiency, work quality, and cost-effectiveness, it could significantly enhance farmers’ operation capability and household incomes. Can farmers’ participation in production process outsourcing enhance their operation capability and increase their income? What are the specific mechanisms of optimization and the magnitude of these effects?

Research on the optimization effects of production process outsourcing in agriculture has predominantly explored its impact on crop yields, technological advancements, and cost benefits. Studies have largely found that outsourcing services enhance crop yields and quality, expand planting areas, and promote large-scale operations [5,6,7,8]. In the realm of technological progress, outsourcing services boost total factor productivity in grain and improve technical efficiency in rice production. By incorporating plot characteristic variables, they provided a more nuanced measure of technological progress [9,10,11]. Regarding cost–benefit analysis, production process outsourcing has been shown to reduce production costs, increase farmers’ incomes, and enhance overall welfare [3,12,13,14]. Additionally, production process outsourcing can optimize household labor allocation and create novel income opportunities for farmers [10,11]. However, the relationship between firm capabilities and service outsourcing remains underexplored. Sean’s study [15] indicated that the erosion of capabilities not only negatively impacts outsourcing performance but also impairs firms’ ability to build and sustain effective commitments and cooperative relationships with outsourcing providers. In developing countries, international service outsourcing has been identified as a key driver of economic development, offering opportunities for growth and integration into global value chains [16,17]. Host country firms benefit from knowledge and technology spillovers, starting from low value-added segments and gradually enhancing their technological innovation capabilities [18]. Yet, the specific mechanisms through which outsourcing services affect agricultural operations, particularly in terms of farmers’ operation capability and income, remain inadequately understood. Therefore, a comprehensive analysis of the effects of production process outsourcing on farmers’ operation capability and income-enhancing effects at the farm household level is essential.

The limitations of the existing studies primarily fall into four key areas: First, existing research has scarcely examined the impact of production process outsourcing on farmers’ operation capability. Enhancing the operation capability of smallholder farmers through outsourcing services is a crucial strategy for their integration into modern agricultural frameworks. Second, the literature on the effect of farmers’ operation capability on their income-enhancing effect is limited, with no comprehensive studies exploring the mechanisms and outcomes of these capabilities in the context of production process outsourcing. Third, much of the existing research enumerates the sub-indicators of farmers’ operation capability but lacks a scientific approach to measure these abilities comprehensively. For instance, Zhu et al. [6] demonstrated that improving farmers’ operation capability and effectively matching factors significantly promote the formation of large-scale farming operations. Similarly, Luo et al. [19] empirically analyzed the impact of land status and farmers’ operation capability on agricultural income. Lastly, the predominant use of traditional linear regression methods in the current literature to study the impact of outsourcing on farmers and agriculture often overlooks the endogeneity of farmers’ outsourcing decisions, leading to potentially biased estimations [20,21].

In response, this paper seeks to address these limitations and contribute to the existing body of research: (1) Utilizing a sample of 2447 farm households, we employ the entropy value method to quantify farmers’ operation capability across two dimensions: agricultural production operation capability and transaction operating capability. (2) We assess the influence of production process outsourcing on both farmers’ operation capability and the income-enhancing effect of farm households. (3) Employing intermediary effect analysis, we examine the mechanism by which production process outsourcing impacts the income-enhancing effect by altering farmers’ operation capability. (4) To address potential endogeneity, we adopt an endogenous switching regression model to calculate the net effect of outsourcing, thereby providing a precise identification of causal relationships. The findings of this paper offer a theoretical foundation for the government to enhance the market supply system for agricultural outsourcing and provide empirical evidence to support the improvement of smallholder farmers’ operation capability and income in developing countries.

The remainder of this paper is structured as follows: Section 2 presents the conceptual framework, Section 3 details the data and model, Section 4 discusses the empirical results, Section 5 provides a discussion, and the final section offers this paper’s conclusions and policy implications.

2. Conceptual Framework

In the context of rural revitalization, enhancing farmers’ agricultural production operation capability has become a critical issue. This paper builds on the research of Zhu et al. [6], conceptualizing farm household operations as an “enterprise”. According to Conmans, a prominent figure in institutional economics, the essence of an enterprise can be categorized into two attributes: production decision-making attributes and transaction operating attributes. The production decision-making attributes, derived from the “human–object” relationship, involve the collection of resources to deliver products or services to the market. Conversely, transaction operating attributes, stemming from the “human–human” relationship, involve utilizing the enterprise’s influence to replace market mechanisms and align the interests of internal stakeholders [22]. Consequently, farmers’ operation capability is classified into agricultural production operation capability and transaction operating capability.

Regarding the impact of production process outsourcing on the agricultural production operation capability of farm households, the increasing trends of rural labor non-agriculturalization, aging, and feminization have led to a decline in farmers’ operation capability. Production process outsourcing helps mitigate these challenges by compensating for the reduced quantity and quality of family labor, thereby enhancing agricultural yields [10]. Furthermore, the migration of rural young adults, combined with the cyclical and seasonal nature of agriculture and shifts in family labor availability, creates a mismatch between the supply and demand for agricultural labor [6]. Production process outsourcing effectively substitutes for family labor and plays a significant role in enhancing farmers’ agricultural production operation capability [23,24]. Regarding the impact of production process outsourcing on farmers’ transaction operating capability, first, the substantial non-farm transfer of labor has led to increased demand for hired agricultural workers, reduced supply, and rising wages. The seasonality of agricultural production and labor imbalances further exacerbate the uncertainty and risk associated with hiring workers [5]. Additionally, the absence of low-cost, quantifiable monitoring systems [25,26], coupled with the cyclical nature of agriculture and the specialization required for farm machinery, results in investment lock-in and high sunk costs for small farmers who purchase their own equipment [6]. Consequently, when farmers engage in production process outsourcing, they benefit from the advanced technology and specialized labor provided by outsourcing institutions, which reduces costs while enhancing operational efficiency. Moreover, farmers’ participation in production process outsourcing often necessitates access to credit to purchase services, and greater access to credit correlates with higher participation rates, increased income, and improved operational efficiency. Finally, outsourcing services represent a new driving force for integrating small farmers into modern agriculture [27], serving as a critical pathway for them to connect with modern agricultural practices. This integration can increase farmers’ land demand, reinforce their commitment to securing land rights, and ultimately enhance their transaction operating capability. Based on this, Hypothesis 1 is proposed: production process outsourcing can enhance farmers’ operation capability.

Moreover, production process outsourcing can influence the income-enhancing effect for farmers by improving their operation capability. As 2020 marked the final year of China’s efforts to build a moderately prosperous society, the rural population living in poverty under the current standards has been lifted out of poverty. However, to consolidate these achievements and address the ongoing challenges of poverty alleviation, those on the brink of poverty and with a higher vulnerability to poverty still require effective means to increase their income. Sustainable income growth can only be ensured if farmers continue to enhance their own capabilities and explore long-term income-generating strategies from the source [28]. This approach will reduce the risk of farmers falling back into poverty and support the sustained and stable increase in their income. Based on this, Hypothesis 2 is proposed: production process outsourcing influences the income-enhancing effect by improving farmers’ operation capability; that is, by enhancing farmers’ operation capacity, production process outsourcing subsequently contributes to an increase in their income.

For rational farmers, the voluntary adoption of production process outsourcing and the reconfiguration of agricultural production factors can lead to higher profit opportunities [3]. This is reflected in two key aspects: first, production process outsourcing enhances agricultural efficiency, reduces production costs, and increases output, thereby contributing to the income-enhancing effect for farmers (Figure 1). A study by Zhang et al. [29] found that specialized outsourcing services employing advanced techniques can optimize agricultural input, such as reducing pesticide overuse, leading to cost savings. Outsourcing also lowers the learning curve for farmers to master critical agricultural skills, effectively improving labor productivity and boosting crop yields [12]. Second, in the context of aging and the transfer of rural labor, many female and older laborers have limited agricultural participation due to physical constraints and skill gaps. By participating in production process outsourcing and delegating labor-intensive activities to specialized service providers, these groups can increase their involvement in agriculture. Simultaneously, other household members can engage in non-agricultural employment, thereby raising overall household income and reducing the risk of poverty [30]. Based on this, Hypothesis 3 is proposed: production process outsourcing aids in enhancing the income of farmers.

3. Data and Methods

3.1. Model Specification

3.1.1. Entropy Value Method

The entropy value method objectively assigns weights to indicators based on the amount of information they provide, effectively mitigating the biases associated with subjective weighting [31]. Given the complexity and nonlinearity of factors influencing farmers’ agricultural production operation capability—considered a dissipative structural system—the entropy value method is well suited for analysis. Accordingly, this paper employs the entropy value method to quantify and evaluate the index system, with the specific procedures outlined as follows:

• The n evaluation indicators across m samples are organized into the original data matrix X = (x_ij)_m×n (where 1 ≤ i ≤ m, and 1 ≤ j ≤ n), with x_ij representing the j-th indicator for the i-th farmer.
• To make the data dimensionless and thus comparable, the maximum value of the j-th indicator is denoted as M_j, and the minimum value is denoted as m_j, Dimensionless processing is applied to x_ij, resulting in new data denoted as ${\mathrm{x}}_{\mathrm{i}\mathrm{j}}^{’}$: ${\mathrm{x}}_{\mathrm{i}\mathrm{j}}^{’}={\displaystyle \frac{{\mathrm{x}}_{\mathrm{i}\mathrm{j}}-{\mathrm{m}}_{\mathrm{j}}}{{\mathrm{M}}_{\mathrm{j}}-{\mathrm{m}}_{\mathrm{j}}}}$. To eliminate the effect of a value of 0, the dimensionless processed data are adjusted by adding a small constant value close to 0, set at 0.001 in this study. The adjusted data are denoted as ${\mathrm{x}}_{\mathrm{i}\mathrm{j}}^{\mathrm{*}}$: ${\mathrm{x}}_{\mathrm{i}\mathrm{j}}^{\mathrm{*}}={\mathrm{x}}_{\mathrm{i}\mathrm{j}}^{’}+\mathsf{\alpha }$.
• For standardized treatment, p_ij represents the weight of the j-th indicator for the i-th farmer within the overall dataset: ${\mathrm{p}}_{\mathrm{i}\mathrm{j}}={\displaystyle \frac{{\mathrm{x}}_{\mathrm{i}\mathrm{j}}^{\mathrm{*}}}{{\sum }_{\mathrm{i}=1}^{\mathrm{m}}{\mathrm{x}}_{\mathrm{i}\mathrm{j}}^{\mathrm{*}}}}$.
• The entropy value e_j of the j-th indicator is calculated: ${\mathrm{e}}_{\mathrm{j}}=-{\displaystyle \frac{1}{\mathrm{l}\mathrm{n}\mathrm{m}}}\sum _{\mathrm{i}=1}^{\mathrm{m}}{\mathrm{p}}_{\mathrm{i}\mathrm{j}}\mathrm{l}\mathrm{n}{\mathrm{p}}_{\mathrm{i}\mathrm{j}}$.
• The coefficient of differentiation g_j for the j-th indicator is calculated: g_j = 1 − e_j.
• The weight W_j of the j-th indicator is determined: ${\mathrm{W}}_{\mathrm{j}}={\displaystyle \frac{{\mathrm{g}}_{\mathrm{j}}}{{\sum }_{\mathrm{j}=1}^{\mathrm{n}}{\mathrm{g}}_{\mathrm{j}}}}$.
• The composite score S_i for each sample’s operational capability is calculated: S_i = $\sum _{\mathrm{j}=1}^{\mathrm{n}}{\mathrm{W}}_{\mathrm{j}}\times {\mathrm{x}}_{\mathrm{i}\mathrm{j}}^{\mathrm{*}}$.

3.1.2. Endogenous Switching Regression (ESR) Model

1. Model Setting: Addressing the endogeneity issue in assessing the impact of farmers’ participation in production process outsourcing on their operation capability is crucial. This endogeneity arises from two primary sources: first, the observable factor is a result of farmers’ self-selection; second, unobservable factors influence the effect of participation in production process outsourcing on operation capability. Ignoring these issues and relying solely on ordinary least squares (OLS) for estimation would likely yield biased results. While the propensity score matching (PSM) method is commonly employed to address selectivity bias, it fails to account for the endogeneity associated with unobservable variables. Similarly, the instrumental variable (IV) approach does not adequately address the heterogeneity in treatment effects. To overcome these challenges, this study employs an endogenous switching regression (ESR) model to assess the welfare implications of production process outsourcing. This method offers several advantages: (i) it addresses both observable and unobservable factors when resolving the self-selection and endogeneity issues associated with outsourcing decisions; (ii) it allows for the separate estimation of factors influencing the agricultural production operation capability of farmers who outsource and those who do not, thereby revealing differential impacts; (iii) it utilizes full information maximum likelihood estimation to mitigate the problem of omitted variable bias; and (iv) it facilitates counterfactual analysis.

The endogenous switching regression (ESR) model typically involves a two-stage estimation process. In the first stage, a Probit or Logit model is employed to estimate the selection equation, determining the factors that influence whether farmers opt for production process outsourcing. In the second stage, the outcome equation evaluates the impact of this decision on the farmers’ operation capability and income-enhancing effects, comparing those who participate in production process outsourcing with those who do not. The second stage specifically focuses on the result equation. The modeling process is outlined as follows:

First, the behavioral choice equation is formulated:

A_i = δ′Z_i + K′I_i + μ_i

(1)

Second, the resulting equation is constructed:

S_iα = β_αX_iα + σ_μαλ_iα + ε_iα

(2)

S_in = β_nX_in + σ_μnλ_in + ε_in

(3)

In Equation (1), A_i denotes the latent variable of farmers’ production process outsourcing; Z_i is an exogenous explanatory variable that influences whether farmers engage in production process outsourcing; μ_i is the error term; I_i is the identification vector. Equation (2) represents the estimation of farmers’ operation capability and the income-enhancing effect for those engaged in production process outsourcing, while Equation (3) represents the estimation for farmers not participating in outsourcing. In these equations, S_iα and S_in represent the levels of the operation capability of farmers engaged in production process outsourcing and those not engaged, respectively; X_iα and X_in capture the factors influencing both operation capability and the income-enhancing effect; ε_iα and ε_in represent the error terms. To address potential sample selection bias due to unobservable factors, the inverse Mills ratios, λ_iα and λ_in, along with their covariances, σ_μα = cov(μ_i, ε_iα) and σ_μn = cov(μ_i, ε_in), are introduced. The full information maximum likelihood method is then applied to estimate Equations (1)–(3), with β_α and β_n as the parameters to be estimated.

2. Average treatment effect

The average treatment effect of production process outsourcing on farmers’ operation capability and the income-enhancing effect is determined by comparing the expected values of these outcomes for farmers who engage in production process outsourcing with those who do not, under both real and counterfactual scenarios. The estimation process is outlined as follows:

For the expected outcomes for the operation capability and income-enhancing effects of farmers engaged in production process outsourcing, the following equation is used:

$$\mathrm{E}\left[\mathrm{S}{\mathrm{i}}_{\mathsf{\alpha }}\right|{\mathrm{A}}_{\mathrm{i}}=1]={\mathsf{\beta }}_{\mathsf{\alpha }}^{’}{\mathrm{X}}_{\mathrm{i\alpha }}+{\mathrm{\sigma }}_{\mathrm{\mu \alpha }}{\mathrm{\lambda }}_{\mathrm{i\alpha }}$$

(4)

For the expected outcomes for the operation capability and income-enhancing effects of farmers not engaged in production process outsourcing, the following equation is used:

$$\mathrm{E}\left[{\mathrm{S}}_{\mathrm{in}}\right|{\mathrm{A}}_{\mathrm{i}}=0]={\mathsf{\beta }}_{\mathrm{n}}^{’}{\mathrm{X}}_{\mathrm{in}}+{\mathrm{\sigma }}_{\mathrm{\mu n}}{\mathrm{\lambda }}_{\mathrm{in}}$$

(5)

In the counterfactual scenario, for the expected outcomes for the operation capability and income-enhancing effects of farmers engaged in production process outsourcing, had they not participated in outsourcing, the following equation is used:

$$\mathrm{E}\left[{\mathrm{S}}_{\mathrm{in}}\right|{\mathrm{A}}_{\mathrm{i}}=1]={\mathsf{\beta }}_{\mathrm{n}}^{’}{\mathrm{X}}_{\mathrm{i\alpha }}+{\mathrm{\sigma }}_{\mathrm{\mu n}}{\mathrm{\lambda }}_{\mathrm{i\alpha }}$$

(6)

For the expected outcomes for the operation capability and income-enhancing effects if farmers not currently engaged in production process outsourcing were to participate in outsourcing, the following equation is used:

$$\mathrm{E}\left[\mathrm{S}{\mathrm{i}}_{\mathsf{\alpha }}\right|{\mathrm{A}}_{\mathrm{i}}=0]={\mathsf{\beta }}_{\mathsf{\alpha }}^{’}{\mathrm{X}}_{\mathrm{in}}+{\mathrm{\sigma }}_{\mathrm{\mu \alpha }}{\mathrm{\lambda }}_{\mathrm{in}}$$

(7)

By comparing Equation (4) with Equation (6), the average treatment effect (ATT) on the operation capability and income-enhancing effect of farmers who actually participate in production process outsourcing can be determined:

$${\mathrm{ATT}}_{\mathrm{i}}=\mathrm{E}\left[\mathrm{S}{\mathrm{i}}_{\mathsf{\alpha }}\right|{\mathrm{A}}_{\mathrm{i}}=1]-\mathrm{E}[{\mathrm{S}}_{\mathrm{in}}|{\mathrm{A}}_{\mathrm{i}}=1]=({\mathsf{\beta }}_{\mathsf{\alpha }}^{’}-{\mathsf{\beta }}_{\mathrm{n}}^{’}){\mathrm{X}}_{\mathrm{i\alpha }}+({\mathrm{\sigma }}_{\mathrm{\mu \alpha }} - {\mathrm{\sigma }}_{\mathrm{\mu n}}){\mathrm{\lambda }}_{\mathrm{i\alpha }}$$

(8)

By comparing Equation (5) with Equation (7), the average treatment effect (ATU) for the operation capability and income-enhancing effect of farmers not engaged in production process outsourcing can be derived:

$${\mathrm{ATU}}_{\mathrm{i}}=\mathrm{E}\left[{\mathrm{S}}_{\mathrm{in}}\right|{\mathrm{A}}_{\mathrm{i}}=0]-\mathrm{E}[\mathrm{S}{\mathrm{i}}_{\mathsf{\alpha }}|{\mathrm{A}}_{\mathrm{i}}=0]=({\mathsf{\beta }}_{\mathrm{n}}^{’}-{\mathsf{\beta }}_{\mathsf{\alpha }}^{’}){\mathrm{X}}_{\mathrm{in}}+({\mathrm{\sigma }}_{\mathrm{\mu n}} - {\mathrm{\sigma }}_{\mathrm{\mu \alpha }}){\mathrm{\lambda }}_{\mathrm{in}}$$

(9)

3.1.3. Intermediary Effect

Employing the methodologies proposed by Wen et al. [32] and Baron et al. [33], this study utilizes stepwise regression to conduct an intermediary effect analysis, investigating whether production process outsourcing enhances farmers’ income-enhancing effects by improving their operation capability.

Y_i = β₀ + β₁X_i + γ_ijZ_ij + ε_i1

(10)

S_i = δ₀ + αX_i + δ₂Z_i + ε_i2

(11)

Y_i = φ₀ + φ₁X_i + γS_i + φ₃Z_i + ε_i3

(12)

In the above model, Y_i denotes the income-enhancing effect of farmer i; X_i denotes the production process outsourcing of farmer i; Z_ij denotes the j-th control variable for farmer i; S_i is the intermediary variable, indicating the operation capability of farmer i; β₀, δ₀, and φ₀ are constant terms; β₁, γ_ij, α, δ₂, φ₁, γ, and φ₃ are the coefficients to be estimated; and ε_i1, ε_i2, and ε_i3 are the random perturbation terms.

This study also builds on the work of Iacobucc et al. [34] to examine whether production process outsourcing can enhance farmers’ income by improving their operation capability. This approach is applicable for testing the intermediary effect when the intermediary variable has a continuous dependent variable [35]. However, the estimated coefficients from these regression models are on different scales and must be standardized for comparison. In this study, after transforming the regression coefficients α and γ into Z_α and Z_γ, respectively, both are standardized to the same scale. The size of the mediation effect within the model is represented by the product Z_α × Z_γ. The significance of this mediation effect is evaluated through the Sobel test. The relevant test statistics are calculated as follows:

Z_α = α/S_α, Z_γ = γ/S_γ

(13)

$$\mathrm{Z}{\mathsf{\alpha }}_{\mathsf{\gamma }}={\mathrm{Z}}_{\mathsf{\alpha }}{\mathrm{Z}}_{\mathsf{\gamma }},\mathrm{SE}(\mathrm{Z}{\mathsf{\alpha }}_{\mathsf{\gamma }})=\sqrt{Z_{\alpha }^2+Z_{\gamma }^2+1}$$

(14)

$$\mathrm{Z}=\mathrm{Z}\mathsf{\alpha }\mathsf{\gamma }/\mathrm{SE}(\mathrm{Z}{\mathsf{\alpha }}_{\mathsf{\gamma }})={\mathrm{Z}}_{\mathsf{\alpha }}{\mathrm{Z}}_{\mathsf{\gamma }}/\sqrt{Z_{\alpha }^2+Z_{\gamma }^2+1}$$

(15)

Here, the statistic Z follows a standard normal distribution. If the absolute value of Z falls between 1.96 and 2.58, the intermediary effect is significant at the 5% level. If the absolute value of Z exceeds 2.58, the intermediary effect is significant at the 1% level.

3.2. Data Sources

The data for this analysis were collected by our research group between July and August 2022, from farmers across five provinces in the Yellow River Basin: Inner Mongolia, Gansu, Ningxia, Henan, and Shaanxi. These provinces were chosen primarily for two reasons: (1) the region’s fragile natural environment, characterized by heavy rainfall, flooding, low temperatures, and frost damage, posing significant challenges to agricultural production, necessitating improvements in farmers’ operation capability and risk management abilities and (2) the level of outsourcing services varies among these provinces. Using a combination of stratified and random sampling methods, 2–3 counties (districts) were randomly selected from each city, 1–3 townships were randomly selected from each county (district), and 4–8 villages were selected from each township using a stratified approach. Approximately 20 farmers from each village were then randomly selected for one-on-one surveys. The questionnaire covered personal characteristics, family characteristics, business characteristics, and other relevant information. A total of 2452 questionnaires were distributed, with 2447 valid samples obtained after excluding those with missing key information or unreasonable data, yielding an effective response rate of 99.80%.

3.3. Definition of Variables

3.3.1. Dependent Variable

Farmers’ operation capability: As previously described, farmers’ operation capability is categorized into agricultural production operation capability and transaction operating capability. Based on this classification, indices for measuring both aspects of operation capability are designed according to the hypothesis (see

Table 1. Indicator system for evaluation of farmers’ operation capability.

Level 1 Indicators	Level 2 Indicators	Calculation Method	Variable Assignment
agricultural production operation capability	Production capability	The yield levels of individual farmers were assessed in comparison to those of their peers in the surrounding area	1 = Low; 2 = moderate; 3 = high
	Employed or not	Whether the farmer utilized hired labor in agricultural production	1 = Yes; 0 = No
transaction operating capability	Investment in agricultural machinery *	The purchase of agricultural machinery or not	1 = Yes; 0 = No
		The value of agricultural machinery	Net present value (CNY ten thousand)
	Agricultural credit	The amount of agricultural loans	Actual amount (CNY ten thousand)
	Stability of land rights	Whether the land is adjusted	1 = No; 2 = some; 3 = all
		Whether the land is titled	1 = Yes; 0 = No

* Note: agricultural machinery encompasses equipment such as seeders, mulching machines, tractors, harvesters, threshing machines, irrigation systems, and rotary tillers.

). Agricultural production operation capability is assessed through two metrics: first, the relative yield level of a household’s agricultural production, measured by the question ‘How does your household’s agricultural production yield compare to others?’; second, the employment of labor, measured by the question ‘Do you employ laborers in agricultural production?’. Transaction operating capability is evaluated through three dimensions: the substitution of labor by machinery, assessed by the ownership of farm machinery, indicated by responses to ‘Do you purchase farm machinery?’ and ‘What is the net asset value of your farm machinery?’; agricultural credit inputs, measured by the amount of bank loans received for agricultural investment; and the stability of land rights, gauged by questions such as ‘Has the land been adjusted?’ and ‘Has the land been titled?’. The entropy value method is employed to calculate the level of farmers’ operation capability [6].

Income-enhancing effect: Drawing upon the study by Baiyegunhi et al. [12], which examines the welfare effects of production process outsourcing, this research utilizes per capita net household income from agriculture indicators to measure the income-enhancing effect. This methodology serves to evaluate and test the income-generating potential of farm households comprehensively. Specific formula: per capita net income from family agriculture is calculated as the net income from family agriculture divided by the family size, expressed in million yuan.

3.3.2. Independent Variable

Production process outsourcing: Production process outsourcing involves delegating production tasks to external specialized resources to reduce costs, mitigate risks, enhance efficiency, and improve competitiveness. In this study, the variable ‘production process outsourcing’ is operationalized by classifying farmers who ‘hire specialized service teams’ as engaging in production process outsourcing, while those utilizing other labor sources are classified as not participating. Consistent with the previous literature [21,36,37], which employs binary dummy variables to analyze farmers’ outsourcing decisions, and based on the research group’s findings, the scope of production process outsourcing in the study area encompasses land preparation, mulching, irrigation, pest and disease control, and harvesting. Consequently, the outsourcing decision-making variable is defined by whether farmers engage in any of these processes. Farmers who utilize at least one of these services are assigned a value of 1, while those who do not are assigned a value of 0.

Degree of production process outsourcing: Given the variation in capital, technology, and labor among farmers, the degree of outsourcing is described in terms of the number of production link services purchased. Therefore, farmers are scored as 1 for outsourcing one process, 2 for two processes, 3 for three processes, and so forth, reflecting the degree of their production process outsourcing [38].

3.3.3. Instrumental Variable

To ensure model identifiability, the cohort effect is employed as an instrumental variable. It is defined as follows: village outsourcing rate excluding the household in question = (total number of outsourced households in the village sample − whether this household is outsourced)/(total number of households in the village sample − 1). The justification for using this variable is based on the premise that farmers’ behavioral decisions are often influenced by the actions of their peers [39], leading them to adopt similar production process outsourcing behaviors. However, this variable does not directly influence the farmers’ operation capability or the income-enhancing effect.

3.3.4. Intermediary Variable

Farmers’ operation capability: To evaluate whether production process outsourcing influences farmers’ income-enhancing effect through the enhancement of their operation capability, this study uses farmers’ operation capability as the intermediary variable to assess this effect. The calculation method employed is consistent with that used for the explanatory variables previously discussed.

3.3.5. Control Variable

Drawing upon the existing literature [3,10,11,13,40], this study selects the personal attributes of household heads, family characteristics, and business traits as control variables. Regional dummy variables are included to account for estimation bias due to regional disparities. As shown in

Table 2. Descriptive statistics of variables.

Variable Name	Variable Definition and Assignment	Average Value			Discrepancy
		Total Sample	Outsourcing	Not Outsourced
Dependent variable
Farmers’ operation capability	Measured by the entropy method	0.125	0.169	0.086	−0.082 ***
Per capita net household income from agriculture	Net family income from agriculture/family size (CNY ten thousand)	1.506	1.710	1.327	−0.383 ***
Explanatory variables
Outsourcing decision	1 = Yes; 0 = No	0.467	1	0
Degree of outsourcing	0 = Not used, 1 = 1 item used, 2 = 2 items used, and so on	0.891	1.898	0.008	−1.889 ***
Control variable
The personal attributes of household heads
Age	Actual age (years)	57.349	56.915	57.729	0.814 *
Gender	1 = male; 0 = female	0.933	0.934	0.933	−0.001
Educational attainment	Years	7.021	7.178	6.883	−0.296 **
Length of experience in agriculture	Years	33.154	32.608	33.633	1.025 *
Family characteristics
Number of agricultural laborers	persons	2.756	2.807	2.712	−0.094 **
Non-farm payroll ratio	Non-farm payrolls/total	0.319	0.306	0.330	0.023 *
Social network	Number of relatives and friends who visit regularly (persons)	28.238	26.597	29.677	3.080 *
Business traits
Planting area	Actual area cultivated by households (mu #)	9.301	9.331	9.300	−0.031
Access to cooperatives	1 = Accession; 0 = non-accession	0.034	0.038	0.031	−0.007
Technical training	1 = Accepted; 0 = not accepted	0.310	0.368	0.259	−0.109 ***
Distance to town hall	Actual distance from home to town hall (km)	6.346	6.346	6.346	0.000
Instrumental variable
Cohort effect	Village outsourcing rate excluding the household in question	0.467	0.516	0.425	−0.091 ***
Regional dummy variables
Inner Mongolia	1 = Inner Mongolia; 0 = Inner Mongolia	0.135	0.128	0.142	0.014
Gansu	1 = Gansu; 0 = Gansu	0.220	0.243	0.199	−0.044 ***
Ningxia	1 = Ningxia; 0 = Ningxia	0.223	0.193	0.248	0.055 ***
Henan	1 = Henan; 0 = Henan	0.222	0.185	0.255	0.070 ***
Shaanxi	1 = Shaanxi; 0 = Shaanxi	0.099	0.143	0.061	−0.082 ***

Note: ^# mu = 1/15 ha; ***, **, and * indicate significance at the 1%, 5%, and 10% levels, respectively, and parametric t-tests were used to compare differences.

, farmers’ operation capability is generally low, with a mean value of 0.125. The average age of household heads is 57 years, indicating significant aging in rural areas. Educational attainment is low, typically at the junior high school level, and the average farming experience spans 33 years, reflecting high levels of agricultural expertise. On average, households have three laborers, and the proportion of non-farm employment is 31.9%, suggesting that non-farm employment influences agricultural production and validates the need for production process outsourcing.

4. Empirical Results

4.1. Model Linkage Estimation of Production Process Outsourcing and Farmers’ Operation Capability

The estimation results of the model linking production process outsourcing and farmers’ operation capability are presented in

Table 3. Model linkage estimation of production process outsourcing and farmers’ operation capability.

Variable	Production Process Outsourcing Model	Farmers’ Operation Capability Model
		Outsourcing	Not Outsourced
Personal attributes of household heads
Age	0.005 (0.003)	0.000 (0.000)	−0.001 ** (0.000)
Gender	−0.172 * (0.099)	0.002 (0.007)	0.008 (0.009)
Educational attainment	0.059 (0.039)	0.004 (0.003)	−0.006 * (0.003)
Length of experience in agriculture	−0.004 (0.003)	0.000 (0.000)	0.000 * (0.000)
Family characteristics
Number of agricultural laborers	0.014 (0.022)	0.004 ** (0.002)	0.006 ** (0.002)
Non-farm payroll ratio	−0.017 (0.081)	−0.016 ** (0.006)	−0.025 ** (0.007)
Social network	−0.001 (0.001)	0.000 * (0.000)	0.000 (0.000)
Business traits
Planting area	−0.000 (0.001)	−0.004 (0.041)	−0.001 (0.050)
Access to cooperatives	0.004 (0.134)	0.038 *** (0.009)	−0.003 (0.012)
Technical training	0.099 * (0.055)	0.005 (0.004)	−0.005 (0.005)
Distance to town hall	0.002 * (0.001)	0.000 (0.000)	−0.000 (0.000)
Instrumental variable
Cohort effect	0.546 *** (0.102)
Regional dummy variables
Inner Mongolia	−0.322 ** (0.102)	−0.052 *** (0.007)	0.051 *** (0.009)
Gansu	−0.059 (0.092)	−0.039 *** (0.006)	0.022 ** (0.008)
Ningxia	−0.108 (0.091)	−0.051 *** (0.007)	0.009 (0.008)
Henan	−0.012 (0.095)	−0.079 *** (0.007)	0.002 (0.008)
Shaanxi	0.222 ** (0.110)	0.019 ** (0.007)	−0.028 ** (0.010)
Constant term (math.)	0.409 ** (0.217)	0.181 *** (0.017)	0.041 ** (0.019)
/lns1		−2.835 *** (0.028)
Rho1		−0.199 (0.143)
/lns0			−2.394 *** (0.026)
Rho0			−0.985 *** (0.004)
LR	175.190 ***
Loglikelihood	1981.810

Note: *, **, and *** indicate 10%, 5%, and 1% significance levels, respectively.

. Column 2 provides the estimation of factors influencing farmers’ outsourcing decisions, while columns 3 and 4 present the estimations of factors affecting the operation capability of outsourced and non-outsourced farmers, respectively. The coefficients rho1 and rho2 represent the correlation between the decision-making model’s error terms and the operation capability models of outsourced and non-outsourced farmers, respectively, with one coefficient being significant at the 1% level. This significance suggests the presence of unobservable factors in the sample that impact both outsourcing and operation capability, necessitating correction to eliminate bias.

4.1.1. An Analysis of the Model Estimation Results for Farmers’ Production Process Outsourcing Decisions

The estimation results in column 2 of Table 3 reveal several significant factors influencing farmers’ decisions regarding production process outsourcing. Among personal characteristics, gender exhibits a significant negative effect, indicating that women are more inclined to engage in production process outsourcing due to their relative technical and physical disadvantages in agricultural production. Regarding business characteristics, technical training has a significant positive effect on farmers’ production process outsourcing decisions. This is understandable, as farmers who have undergone technical training are more likely to recognize the benefits of adopting advanced technology, with outsourcing being a convenient method to incorporate such technology. Additionally, the estimation results indicate that the proximity of a farmer’s home to the township government has a significant positive effect on the decision to outsource the production process. Farmers situated closer to the government benefit more from its services and influence, making them more likely to participate in production process outsourcing.

4.1.2. Analysis of Estimation Results for Farmers’ Operation Capability

The model’s estimation results for farmers’ operation capability reveal several significant factors. Among personal characteristics, both age and education have a significant negative impact on the operation capability of farmers who do not engage in production process outsourcing. This may be attributed to older farmers experiencing physical decline, which hampers their operational capability. Additionally, farmers with higher education levels often have opportunities to earn higher wages outside of agriculture, detracting from their focus on enhancing agricultural operation capability. Conversely, the number of years of farming experience has a significant positive effect on the operation capability of non-outsourced farmers, likely because extensive farming experience translates to richer skills and stronger proficiency in agricultural production. In terms of household characteristics, both the number of laborers and the proportion of non-farm employment significantly affect the operation capability of both outsourced and non-outsourced farmers. The number of laborers has a positive effect, while the proportion of non-farm employment has a negative effect. This suggests that the status of labor resources greatly influences the operation capability of farm households. A larger number of laborers and a lower proportion of non-farm employment increase the available workforce for agricultural production, thereby enhancing operational capability. Social networks have a significant positive impact on the operation capability of farmers who engage in production process outsourcing. This indicates that social networks not only expand access to information but also facilitate material and capital support, thereby improving operational capability. Among business characteristics, the variable of joining cooperatives has a positive effect on the operation capability of outsourced farmers at the 1% significance level, while it is not significant for non-outsourced farmers. This suggests that cooperatives play a crucial role in promoting and developing production process outsourcing, thereby enhancing farmers’ operation capability.

4.1.3. Instrumental Variable Validity Testing

The estimation results in Table 3 indicate that the cohort effect, used as an instrumental variable, significantly influences farmers’ decisions to engage in production process outsourcing. To further assess its validity, this study conducts additional regressions incorporating control variables: first, with the instrumental variable as the independent variable and farmers’ operation capability as the dependent variable; second, with both production process outsourcing and the instrumental variable as independent variables and farmers’ operation capability as the dependent variable. The findings reveal that the instrumental variable has no significant impact on farmers’ operation capability in either regression. Moreover, the instrumental variable estimation for the effect of production process outsourcing on farmers’ operation capability yields an F-value of 95.53 in the first stage, surpassing the empirical threshold of 10, thereby rejecting the hypothesis of weak instruments [41]. Consequently, the selected instrumental variable is deemed valid.

4.2. Estimation Results of Model Linking Production Process Outsourcing to Income-Enhancing Effect

The estimation results of the model linking production process outsourcing to the income-enhancing effect are presented in

Table 4. Estimation results of model linking production process outsourcing to income-enhancing effect.

Variable	Production Process Outsourcing Model	Income-Enhancing Effect Model
		Outsourcing	Not Outsourced
Personal attributes of household heads
Age	−0.001 (0.003)	0.002 (0.002)	−0.004 * (0.002)
Gender	−0.169 * (0.096)	−0.012 (0.065)	0.062 (0.068)
Educational attainment	0.014 (0.038)	0.015 (0.023)	0.021 (0.027)
Length of experience in agriculture	0.000 (0.003)	0.001 (0.002)	−0.003 (0.002)
Family characteristics
Number of agricultural laborers	0.011 (0.022)	0.027 * (0.015)	0.009 (0.016)
Non-farm payroll ratio	0.191 ** (0.079)	0.207 *** (0.060)	0.201 *** (0.057)
Social network	−0.000 (0.001)	−0.000 (0.001)	0.000 (0.000)
Business traits
Planting area	0.000 (0.001)	−0.000 (0.000)	−0.000 (0.000)
Access to cooperatives	−0.005 (0.136)	0.123 (0.079)	0.066 (0.099)
Technical training	0.208 *** (0.055)	0.091 ** (0.040)	0.132 ** (0.039)
Distance to town hall	0.002 (0.001)	−0.002 ** (0.001)	−0.000 (0.001)
Instrumental variable
Cohort effect	0.309 *** (0.079)
Regional dummy variables
Inner Mongolia	−0.295 ** (0.099)	0.304 *** (0.071)	0.214 ** (0.071)
Gansu	−0.006 (0.092)	−0.177 ** (0.053)	−0.103 (0.066)
Ningxia	−0.289 *** (0.090)	−0.124 * (0.068)	0.138 ** (0.065)
Henan	−0.245 ** (0.095)	−0.031 (0.069)	0.088 (0.068)
Shaanxi	0.329 ** (0.112)	0.124 * (0.072)	−0.143 * (0.081)
Constant term (math.)	−0.029 (0.210)	0.908 *** (0.188)	0.011 (0.149)
/lns1		−0.307 ** (0.024)
Rho1		−0.996 ** (0.001)
/lns0			−0.730 *** (0.022)
Rho0			−0.022 (0.395)
LR	284.940 ***
Loglikelihood	3084.009

Note: *, **, and *** indicate 10%, 5%, and 1% significance levels, respectively.

. Column 2 displays the estimation results for factors influencing outsourcing decisions, while columns 3 and 4 present the estimation results for factors affecting the income-enhancing effect of outsourced and non-outsourced farmers, respectively. The coefficients rho1 and rho0 represent the correlation between the error terms of the decision-making model and the income-enhancing effect models for outsourced and non-outsourced farmers, with one coefficient being significant at the 1% level. This significance indicates the presence of unobservable factors that simultaneously affect outsourcing decisions and the income-enhancing effect, necessitating correction to eliminate bias.

4.2.1. Analysis of Estimation Results for Farm Households’ Production Process Outsourcing Decisions

The estimation results in column 2 of Table 4 reveal that the proportion of non-farm employment has a significant positive effect on the decision of farm households to engage in production process outsourcing. This effect can be attributed to the reduction in the agricultural labor force caused by an increase in non-farm employment, prompting households to outsource to sustain their agricultural production. The effects of the gender variable and the proportion of technical training are consistent with the explanations provided in Table 3.

4.2.2. Analysis of Estimation Results for Income-Enhancing Effects

The estimation results in columns 3 and 4 of Table 4 reveal several significant factors affecting the income-enhancing effect of farm households. Among individual characteristics, age has a significant negative impact on the income-enhancing effect of non-outsourced farmers. This is likely due to the decline in physical capacity among older farmers, which adversely affects agricultural output and income. Regarding household characteristics, the quantity of labor has a significant positive effect on the income-enhancing effect of outsourced farmers, whereas it fails the significance test for non-outsourced farmers. This indicates that the status of labor resources has a more substantial impact on the income-enhancing effect in households that engage in production process outsourcing. Additionally, the proportion of non-farm employment has a significant positive impact on the income-enhancing effect for both outsourced and non-outsourced households, with a more pronounced effect in outsourced households. This suggests that non-farm employment significantly contributes to household income, and the combination of non-farm employment and production process outsourcing amplifies this effect. Among business characteristics, technical training has a positive effect on the income-enhancing effect for both outsourced and non-outsourced households at the 5% significance level. This implies that receiving training promotes the development of production process outsourcing and contributes to increasing household incomes. The results also show that the distance to the township government has a significant negative effect on the income-enhancing effect of outsourced farmers. This may be due to the fact that greater distances from the township government result in less support, hindering the increase in farmers’ income.

4.2.3. Instrumental Variable Validity Testing

The estimation results presented in Table 4 indicate that the influence of the instrumental variable on farmers’ outsourcing decisions is significantly positive at the 1% statistical level. To further assess the validity of this finding, this study conducts additional regressions with the inclusion of control variables: first, with the instrumental variable as the independent variable and the income-enhancing effect as the dependent variable; second, with both production process outsourcing and the instrumental variable as independent variables and the income-enhancing effect as the dependent variable. The analysis reveals that the instrumental variable has no significant effect on the income-enhancing effect in either case. Furthermore, the instrumental variable estimation for the effect of production process outsourcing on the income-enhancing effect yields an F-value of 11.46 in the first stage, exceeding the empirical threshold of 10, thereby rejecting the hypothesis of weak instruments [41]. Hence, the selected instrumental variable is considered valid.

4.3. Analysis of Treatment Effects

Table 5. Analysis of treatment effects.

Variable	Type of Farm Household	ATT	ATU
farmers’ operation capability	Outsourcing	0.222 *** (0.001)
	Not outsourced		0.088 *** (0.001)
income-enhancing effect	Outsourcing	0.653 *** (0.015)
	Not outsourced		0.231 *** (0.015)

Note: *** denotes 1% significance level; ATT and ATU denote the average treatment effect for farmers in the participating and non-participating groups, respectively.

presents the treatment effects of production process outsourcing on farmers’ operation capability and the income-enhancing effect. The results for operation capability indicate that ATT = 0.222, significant at the 1% level, suggesting that if a farmer shifts from participating in production process outsourcing to not outsourcing, their operation capability decreases by an average of 0.222. Conversely, ATU = 0.088, also significant at the 1% level, reflecting that non-outsourced farmers experience an average increase in operation capability of 0.088. This confirms Hypothesis 1, indicating that production process outsourcing is a primary driver for improving farmers’ operation capability and enhancing income. The income-enhancing effect results show that ATT = 0.653, significant at the 1% level, implying that if a farmer changes from outsourcing to not outsourcing, the income-enhancing effect decreases by an average of 0.653. Similarly, ATU = 0.231, significant at the 1% level, indicating that non-outsourced farmers see an average increase in the income-enhancing effect of 0.231. This supports Hypothesis 3, demonstrating that production process outsourcing significantly contributes to increasing farmers’ incomes.

4.4. Variation in Treatment Effects of Degree of Production Process Outsourcing on Farmers’ Operation Capability

Farmers in the sample area primarily outsource five production processes: land preparation, mulching, irrigation, pest control, and harvesting. Farmers’ needs for these processes vary due to financial, technological, and labor constraints, and there are significant disparities in the development of social services across different production processes. Considering the actual adoption of outsourcing in the sample area, the number of farmers adopting four or five outsourcing processes is very small. Therefore, this study focuses on farmers adopting one, two, or three outsourcing processes. The estimation results are presented in

Table 6. Variation in treatment effects of production process outsourcing on farmers’ operation capability.

Variable	Type of Farm Household	ATT	ATU
1 item used	Outsourcing	0.196 *** (0.001)
	Not outsourced		0.079 *** (0.001)
2 items used	Outsourcing	0.196 *** (0.002)
	Not outsourced		0.094 *** (0.001)
3 items used	Outsourcing	0.226 *** (0.005)
	Not outsourced		0.241 ***(0.001)

Note: *** denotes 1% significance level; ATT and ATU denote the average treatment effect for farmers in the participating and non-participating groups, respectively.

.

The estimation results indicate that the average treatment effect (ATT) of outsourcing on the operation capability of farmers adopting one, two, or three processes is positively significant at the 1% level. Specifically, the ATT results show that the operation capability decreases by 0.196, 0.196, and 0.226, respectively, if farmers who adopt one, two, or three production process outsourcing processes cease to participate in these activities. Conversely, the ATU results reveal that farmers’ operation capability increases by 0.079, 0.094, and 0.241, respectively, if non-outsourced farmers begin to participate in one, two, or three outsourcing processes. Overall, the findings suggest that the more outsourcing processes a farmer adopts, the greater the improvement in their operation capability.

4.5. Difference Analysis

Differences in farmers’ education levels, labor force size, and planting area exert distinct influences on their participation in production process outsourcing and their agricultural production operation capability. To further elucidate these differences, we adopt the difference analysis method of Zeng et al. [42]. We calculate the mean values of the variables for education, labor force, and planting area and subsequently divide the samples into ‘greater than the mean’ and ‘less than the mean’ groups for analysis. The results are presented in

Table 7. Difference analysis.

Variable		Type of Farm Household	ATT	ATU
Educational attainment	Less than the mean	Outsourcing	0.203 *** (0.002)
		Not outsourced		0.106 *** (0.001)
	Greater than the mean	Outsourcing	0.246 *** (0.002)
		Not outsourced		0.075 *** (0.001)
Number of agricultural laborers	Less than the mean	Outsourcing	0.214 *** (0.002)
		Not outsourced		0.093 *** (0.001)
	Greater than the mean	Outsourcing	0.221 *** (0.002)
		Not outsourced		0.105 *** (0.002)
Planting area	Less than the mean	Outsourcing	0.204 *** (0.001)
		Not outsourced		0.107 *** (0.001)
	Greater than the mean	Outsourcing	0.076 *** (0.003)
		Not outsourced		0.089 *** (0.003)

Note: *** denotes 1% significance level; ATT and ATU denote the average treatment effect for farmers in the participating and non-participating groups, respectively.

.

The estimation results in Table 7 indicate that farmers’ involvement in production process outsourcing has a significant positive effect on their operation capability. The enhancement effect of outsourcing on operation capability is more pronounced for farmers with higher education levels and larger labor forces compared to those with lower education levels and smaller labor forces. However, the effect is less pronounced for farmers with larger planting areas than for those with smaller planting areas. These findings highlight the significant heterogeneity in the impact of production process outsourcing on operation capability, which varies considerably based on the endowments of the farm households.

4.6. Robustness Check

To test the robustness of the matching results, we employed the nearest neighbor matching method (with K set to 4), the kernel matching method (with bandwidth set to 0.06), and the marginal distance matching method for comparison. As shown in

Table 8. PSM robustness check.

Matching Method	Treatment Group Mean	Control Group Mean	ATT	Standard Error	T-Value
The kernel matching method	0.168	0.091	0.078 ***	0.003	25.67
The nearest neighbor matching method	0.168	0.088	0.080 ***	0.003	22.85
The marginal distance matching method	0.168	0.090	0.078 ***	0.003	27.96

Note: *** denotes 1% significance level; ATT denotes the average treatment effect for farmers in the participating groups.

, the average treatment effect (ATT) for the treatment group involved in production process outsourcing indicates that the mean ATT value, calculated using the three matching methods, is 0.079. This demonstrates that production process outsourcing significantly enhances farmers’ operation capability, further confirming the robustness of the ESR results.

4.7. Analysis of Impact Mechanisms

This study examines whether the adoption of production process outsourcing by farmers enhances their operation capability, which subsequently impacts their income-enhancing effect, as discussed in the previous section. The test results are presented in

Table 9. Analysis of impact mechanisms.

Variable	Income-Enhancing Effect	Farmers’ Operation Capability	Income-Enhancing Effect
Production process outsourcing	0.226 *** (0.046)	0.074 *** (0.003)	0.150 *** (0.054)
Farmers’ operation capability			1.023 *** (0.370)
Control variable	Controlled	Controlled	Controlled
Constant term (math.)	8.786 *** (0.192)	0.122 *** (0.011)	8.660 *** (0.197)
R2	0.098	0.397	0.101
Observations	2447	2447	2447
Z-value	2.748 ***

Note: *** denotes 1% significance level.

. Column 2 shows that production process outsourcing has a significant direct effect on income generation, consistent with the benchmark regression results. Column 3 demonstrates that production process outsourcing significantly improves the farmers’ operation capability. Both variables in column 4 pass the significance test, indicating that after controlling for the effect of production process outsourcing, the mediating variable—farmers’ operation capability—still has a significant impact on income generation. Further analysis, using the estimated coefficients and standard errors from both production process outsourcing and farmers’ operation capability, confirms the mediation effect with a Z-value of 2.748 (Table 9), significant at the 1% level. This indicates that farmers’ operation capability mediates the relationship between production process outsourcing and income generation, thereby supporting Hypothesis 2: production process outsourcing increases farm household income by enhancing operation capability.

5. Discussion

Utilizing field survey data from five provinces in the Yellow River Basin—Inner Mongolia, Gansu, Ningxia, Henan, and Shaanxi—this study investigates the impact of production process outsourcing on the enhancement of farmers’ operation capability and the income-enhancing effect within a counterfactual framework, employing an endogenous switching regression model (ESR). The findings reveal that production process outsourcing is crucial in improving the agricultural production operation capability of farm households and in driving their income growth. Unlike previous research, this study incorporates farmers’ operation capability into the analytical framework, offering a novel complement to earlier mechanistic studies on production process outsourcing and farm household income growth [12,13]. Furthermore, by representing the degree of outsourcing through the number of production links outsourced, this study provides an additional dimension to the existing literature, which typically examines each production link in isolation [43]. The ESR model is selected for its ability to account for both observable and unobservable factors, addressing the issues of self-selection and endogeneity in outsourcing decisions. This approach allows for the separate estimation of factors influencing the operation capability of farmers engaged in and not engaged in production process outsourcing, enabling an examination of differential impacts. This study employs the all-informative maximum likelihood estimation method to mitigate the omission of valuable information [21].

This study concludes that production process outsourcing significantly enhances both the operation capability and income-enhancing effect of farm households. Participation in production process outsourcing mitigates quantitative and qualitative labor deficiencies, boosts farmland productivity, reduces agricultural production costs, and drives income growth. Analyzing the impact of the outsourcing extent on farmers’ operation capability reveals that limited outsourcing restricts farmers’ exposure to standardized service provider operations. Consequently, much of the agricultural production and decision-making remains in the farmers’ hands, constrained by limitations in labor, technology, capital, and knowledge. In contrast, farmers who engage in more extensive outsourcing experience greater improvements in operation capability. Variance analysis indicates that higher education and greater labor availability correlate with higher levels of human capital, making these farmers more likely to participate in outsourcing and enhance their operation capability. However, farmers with smaller planting areas are less likely to invest in advanced technology and facilities, whereas those with larger areas tend to do so and participate in outsourcing. Interestingly, the latter group experiences relatively smaller gains in operation capability from outsourcing. The roles of education level, labor force size, and planting area in promoting operation capability through production chain outsourcing are crucial. Mediation effect tests suggest that production process outsourcing encourages farmers to improve their operation capability, save labor while increasing agricultural output, and enhance income. Therefore, it is imperative for the government to continue developing the outsourcing service market, improving farmland infrastructure, and increasing outreach efforts. Social service organizations should offer high-quality outsourcing services at competitive prices to reduce the supervision costs for farmers.

There are certain limitations in our study that warrant further discussion. Firstly, due to the underdeveloped outsourcing market in the sample area and the low adoption rates among farmers in certain segments, data on the degree of outsourcing involving four or five items are lacking, an area that will require further attention in future research. Secondly, since our study focuses exclusively on regions within the Yellow River Basin, it does not account for the developmental disparities across different regions in China. Lastly, given that we utilized cross-sectional data, which may not capture the dynamic nature of farmers’ behaviors over time, future studies could employ panel data to more accurately identify causal effects and further explore the findings.

6. Conclusions and Policy Recommendations

6.1. Conclusions

This study investigates the impact of production process outsourcing on farmers’ operation capability and the income-enhancing effect using the endogenous switching regression (ESR) model to construct a counterfactual framework. The analysis is based on data from 2447 farmers across five provinces within the Yellow River Basin. The findings are as follows: First, production process outsourcing significantly enhances farmers’ agricultural production operation capability and income. In the counterfactual scenario, if farmers currently participating in outsourcing were to cease, their operation capability and income-enhancing effects would decrease by 22.2% and 65.3%, respectively. Conversely, if non-participating farmers were to engage in outsourcing, their operation capability and income-enhancing effects would increase by 8.8% and 23.1%, respectively. Second, a higher degree of production process outsourcing correlates with greater improvements in agricultural production operation capability. Third, the impact of production process outsourcing on operation capability varies based on farmers’ endowments. Higher education levels, larger labor forces, and smaller planting areas are associated with greater improvements in operation capability through outsourcing. Fourth, production process outsourcing directly affects farmers’ operation capability and income-enhancing effects. Additionally, it partially mediates income generation through improved operation capability, with the mediation effect accounting for 33.5% of the total effect.

6.2. Policy Recommendations

Based on the aforementioned conclusions, the following policy recommendations are proposed: First, the government should actively support outsourcing service organizations for small farmers. By implementing preferential policies, these organizations can be guided to continuously improve the quantity and quality of their technology, labor, and other services, thereby enhancing farmers’ operation capability and increasing their incomes. Second, higher degrees of production process outsourcing are more beneficial for farmers’ development. Therefore, the role of media such as radio, television, and the Internet should be utilized to increase the awareness and dissemination of outsourcing services, enabling farmers to access service information promptly, adopt scientific production methods, and engage in more outsourcing processes. Third, there should be a focus on cultivating small farmers’ operation capability and sense of cooperation through education and technical training, facilitating their integration into modern agriculture. Fourth, infrastructure development in farmland, such as roads and water conservancy, should be strengthened to create favorable conditions for the growth of socialized production services. Fifth, considering the group differences in the impact of production process outsourcing on operation capability, the government should implement measures such as education and training, rational land transfer, and the development of the hired labor market to address farmers’ shortcomings and promote the improvement of their operation capability.