Economic Effects of Green Pest Control Technology Adoption on Apple Farmers’ Income: Evidence from China

Abstract

With the increasing importance of green transformation in agricultural production, green pest control technologies (GPCTs), defined as a set of eco-friendly methods aimed at managing agricultural pests with reduced reliance on synthetic chemical pesticides, play a key role in improving agricultural production efficiency, ensuring product quality, and protecting the ecological environment. Based on field survey data from apple farmers in Yantai and Linyi cities, Shandong Province, collected in 2022, this paper employs endogenous treatment effects regression (ETR) and instrumental variable quantile regression (IVQR) models to analyze the impact of adopting green pest control technologies on household income and explores the heterogeneity of this effect across different income levels. The results show that the adoption of green pest control technologies significantly increases apple farmers’ net apple income and household income, confirming their income-boosting effect. Moreover, the income-boosting effect is more significant for lower-income farmers, suggesting that these farmers benefit more from the adoption of green pest control technologies by improving pest management and thus enhancing apple production efficiency. This study provides empirical evidence for the promotion of green pest control technologies and offers valuable references for policymakers, especially in supporting technology adoption among lower-income farmers.

1. Introduction

Over the past few decades, the rapid expansion of urban economies has significantly influenced rural labor dynamics. The creation of diverse employment opportunities in urban areas has encouraged many rural residents to diversify their income sources, engaging in both agricultural production and off-farm activities [1]. This trend, compounded by an increasing proportion of elderly populations in rural communities, has made the efficient allocation of agricultural resources through advanced technologies a critical factor for improving farmers’ livelihoods [2,3]. Furthermore, due to the time-sensitive nature of cash crop production, farmers are increasingly adopting more intensive and precise farming techniques to optimize yields and enhance income. Improving agricultural technology has therefore become a key mechanism for increasing productivity and raising household income [4].

The concept of “green transformation” in agriculture, as alluded to earlier, refers to a fundamental shift from traditional resource-intensive and polluting agricultural practices toward more sustainable, environmentally friendly, and ecologically sound alternatives. This transformation aims to mitigate negative environmental impacts, such as soil degradation, water pollution, and biodiversity loss, while simultaneously enhancing agricultural productivity and ensuring food security. It aligns with broader national efforts in China to foster sustainable development and rural revitalization.

Central to this green transformation are “green pest control technologies (GPCTs)”. These technologies encompass a range of eco-friendly methods designed to manage agricultural pests with reduced reliance on synthetic chemical pesticides. Examples of effective GPCTs widely promoted and adopted in China include (1) ecological control, such as promoting natural enemy populations by enhancing farm biodiversity and intercropping; (2) biological control, which involves the use of beneficial organisms (e.g., predatory mites, parasitic wasps, microbial pesticides like Bacillus thuringiensis); (3) physical control methods, such as insect traps (e.g., yellow sticky traps, light traps, pheromone traps) and physical barriers; and (4) scientific pesticide use, which emphasizes the targeted, efficient, and reduced application of low-toxicity and low-residue biopesticides or highly specific chemical pesticides when necessary. These practices collectively aim to reduce pesticide use intensity, minimize environmental pollution, and improve the quality and safety of agricultural products. Specifically in apple cultivation, the application of these GPCTs is crucial for maintaining fruit quality, reducing chemical residues, and ensuring the long-term sustainability and economic viability of orchards.

Green pest control technologies, as a core component of sustainable farming practices, hold significant potential for transforming agricultural productivity while simultaneously ensuring environmental protection. The adoption of such technologies is particularly relevant for understanding their impact on farm household income [5]. Although existing studies have explored the income-enhancing effects of various green agricultural technologies, a gap remains in understanding the specific economic outcomes of adopting green pest control technologies [6,7]. Most research has concentrated on the yield-enhancing and income-boosting effects of green production technologies in the context of crop yields, with less focus on how these technologies affect broader household income. Additionally, many studies that examine the economic effects of new technologies fail to properly account for the issue of endogeneity in their methodological approaches. While some research has attempted to address selectivity bias or estimated average treatment effects (ATT), few studies have directly evaluated the causal income effects of adopting green pest control technologies [7,8]. Furthermore, there has been limited attention paid to the heterogeneous impacts of these technologies across different income strata within farming communities. Understanding these complex and potentially varied impacts is crucial, as the economic benefits of such technologies can be influenced by diverse factors, including local agricultural conditions, market access, and farmer-specific characteristics.

To fill these gaps, this study leverages data from a 2022 survey of apple farmers in Yantai and Linyi cities, Shandong Province, China. Using advanced empirical methods, including endogenous treatment effects regression (ETR) and instrumental variable quantile regression (IVQR), this study aims to examine the heterogeneous income effects of adopting green pest control technologies. Specifically, this paper seeks to answer two core questions: (1) What is the causal impact of adopting green pest control technologies on apple farmers’ household income and (2) Does this income effect vary across different income levels among apple farmers? The findings are intended to provide empirical evidence that can guide policymakers in designing more effective strategies for promoting green agricultural practices and supporting the income growth of farmers, particularly those in lower-income brackets.

The decision to focus specifically on apple farmers in this study is based on several key factors. Firstly, apples serve as a significant economic crop and a primary source of income for local farmers in the study regions. This economic importance often leads farmers to engage in more intensive cultivation practices, which in turn fosters a higher propensity to adopt innovative technologies like green pest control. Furthermore, the established regional brand effect of local apples encourages farmers to maintain high product quality and sustainability, reinforcing their inclination toward adopting such technologies. This context is crucial for obtaining a rich and diverse sample regarding green pest control technology adoption, which is essential for robust mathematical model establishment.

Secondly, from a methodological perspective, the adoption of GPCTs in apple cultivation is not strongly tied to large-scale mechanized operations, unlike some staple field crops. This means that farmers’ decisions to adopt GPCTs are often explicit and deliberate, rather than being incidental to purchasing mechanized services or engaging in large-scale farming. This distinction is vital, as it helps avoid confounding effects (e.g., spillover effects from large-scale operations or agricultural services) that could obscure the clear causal relationship between farmers’ autonomous adoption behavior and its economic outcomes. Therefore, studying apple farmers allows for a clearer assessment of the direct impact of GPCT adoption.

2. Mechanism of Income Increase Through Green Pest Control Technologies

2.1. The Impact of Green Pest Control Technology Adoption on Farmers’ Household Income

According to the assumption of rational economic behavior, farmers’ actions are the result of maximizing utility under given constraints. Therefore, this study adopts the theoretical framework of farmer behavior, under the basic assumption of dividing household income into agricultural income and non-farm income, and constructs a theoretical analysis framework for examining the impact of green pest control technology adoption on the household income of apple farmers [9].

Based on this assumption, the study posits that the labor force available to apple farmers for agricultural production is denoted by $l_1$, corresponding to agricultural income $Y_1$, while the labor force available for non-farm employment is denoted by $l_2$, corresponding to non-farm income $Y_2$ [10]. It should be noted that, theoretically, the adoption of green pest control technologies does not directly lead to the liberation of labor for shifting to non-agricultural employment [11]. Therefore, building upon the previous chapter’s analysis of the impact of green technology adoption on the agricultural technical efficiency of apple farmers, this study primarily focuses on analyzing the effects of green pest control technology adoption on the household income and agricultural income of apple farmers [12]. As a result, the income composition of apple farmers can be expressed as

$$Y=Y_1+Y_2$$

(1)

$$Y_1=P(A,l_1,k,\varsigma )·Q(A,l_1,k,\varsigma )-C(A,l_1,k,\varsigma )$$

(2)

$$Y_2=w{l}_2$$

(3)

In Equations (1)–(3), $P(\cdot )$ represents the function for the sales price of agricultural products; $Q(\cdot )$ and $C(\cdot )$ are the functions for the yield per unit area and production cost, respectively; $A$ denotes the technology level; $k$ refers to material capital investment; $\varsigma$ represents household characteristic information; and $w$ stands for the wage level for non-farm income. Substituting Equations (2) and (3) into Equation (1) and incorporating the adoption of green pest control technologies into this income function, the income function of apple farmers with respect to the adoption of green pest control technologies can be expressed as

$$Y=P(A,l_1,gpct,k_u,\varsigma )·Q(A,l_1,gpct,k_u,\varsigma )-C(A,l_1,k,\varsigma )+w_{2}l$$

(4)

In Equation (4), $gpct$ represents the adoption of green pest control technology, and $k_u$ refers to capital investment other than the costs associated with adopting green pest control technology. Taking the partial derivative of Equation (4) with respect to $gpct$, we obtain the expression for the impact of green pest control technology adoption on farmers’ household income:

$$\partial Y/\partial gpct=\partial P/\partial gpct·\partial Q/\partial gpct-\partial C/\partial gpct+w_2\partial l/\partial gpct$$

(5)

In Equation (5), the first half of the right-hand side represents the impact of green pest control technology adoption on the net agricultural income of apple farmers, while the second half represents the impact on their non-farm income. Since biological pest control methods, such as biopesticides, can be applied alongside other pesticides, and physical control methods, such as physical trapping and the use of insect-proof nets, can be carried out in parallel with the routine management of orchards, green pest control technologies do not require substantial additional labor. Therefore, in theory, the adoption of green pest control technologies does not affect non-farm income. As a result, this study does not delve deeply into the impact of green pest control technology adoption on non-farm income.

Unlike field crops, apples, as an economic crop, have special characteristics in apple orchards that bind and constrain farmers’ choices of employment types and crop management [8]. On the one hand, most agricultural land in the study area is located in mountainous or hilly regions, where ecological conditions are suitable for the growth of shrub crops but not for large-scale field crop operations, making it difficult to promote socialized services. Apples, therefore, are the best crop option for farmers in the region [13]. On the other hand, the initial investment in building an apple orchard is substantial, with a long payback period [14]. Once a newly planted orchard begins to bear fruit, the income increases gradually from low to high over the years. The replacement of old or diseased trees and the substitution of quality rootstocks also require continuous fixed investments. If farmers abandon their orchards to seek employment elsewhere or switch to growing other crops, they face significant sunk costs. Therefore, apple farmers are more concerned with refining their cultivation techniques and increasing apple-related income to improve household income. This study therefore primarily focuses on examining the impact of adopting green pest control technologies on both the household income and net income from apples for apple farmers. The following hypotheses are proposed:

H1: 

The adoption of green pest control technologies positively affects the net income from apples for apple farmers.

H2: 

The adoption of green pest control technologies positively affects the household income of apple farmers.

2.2. Heterogeneous Analysis of the Impact of Green Pest Control Technology Adoption on Apple Farmers’ Household Income

From the perspective of heterogeneous farm households, under the framework of total factor productivity, income levels may vary when adopting and applying green pest control technologies, leading to different income effects [15,16]. From a theoretical standpoint, farmers with higher income levels generally have stronger economic capabilities and higher technical expertise. Therefore, the income effects of adopting green pest control technologies may be weaker for high-income farmers compared to low-income farmers [17]. High-income farmers are better able to efficiently allocate various production factors, and their non-apple income may constitute a higher proportion of total household income, or they may have larger apple farming operations with higher management standards [18]. As a result, the correlation between the income effects created by green pest control technology adoption and their total income tends to be weaker. On the other hand, low-income farmers often face more challenges in factor allocation, with shortcomings in both their operational scale and management level. The adoption of green pest control technologies is likely to improve pest and disease control, enhance apple production quantity and quality, and increase apple income. Consequently, low-income farmers are more likely to benefit from adopting these technologies. Based on this, the following hypothesis is proposed:

H3: 

The income-increasing effect of green pest control technology adoption is stronger for low-income apple farmers.

3. Model Construction and Variable Selection

3.1. Model Construction

3.1.1. Analysis Model for the Income Effect of Green Pest Control Technology Adoption

There are various methods to quantify the “effect”. Broadly speaking, these methods can be divided into two categories: parametric and non-parametric methods. Among non-parametric methods, the most commonly used is the propensity score matching (PSM) method, which is limited to addressing selection bias caused by observable factors [16]. In the parametric methods, the most widely applied model is the endogenous switching regression (ESR) model, which is effective in addressing selection bias in most cases, including both observable and unobservable factors [18]. However, both methods are indirect evaluation methods, which can only assess the effects through the average treatment effect (ATT) value [19] and do not provide a direct evaluation of the effects [20]. To overcome this limitation, this study adopts the more advanced endogenous treatment effects regression (ETR) model within the parametric methods to estimate the relationship between green pest control technology adoption and farmers’ income [7]. This model builds on the advantages of the ESR model while offering unique advantages, the most important of which is its ability to directly assess the effects [21].

The ETR model regression process consists of two steps. The first step is the selection equation regression, which aims to reflect the relationship between various variables, such as the personal characteristics of respondents, family production and management, and the decision to adopt green pest control technologies. The second step is the outcome equation regression, which represents the result equation determining apple farmers’ income. This step estimates the direct impact of adopting green pest control technologies on the income of apple farmers, under the premise of controlling for endogeneity.

Specifically, the first and second stages are as follows:

$$O_i^{*}={\partial }_i{I}_i+{\delta }_i{X}_i+{\epsilon }_i,O_i=\{\begin{array}{c}1,if.O_i^{*}>0\\ 0,if.O_i^{*}\le 0\end{array}$$

(6)

$$Y_i={\alpha }_i{O}_i+{\beta }_i{X}_i+{\mu }_i$$

(7)

Equation (6) represents the selection equation, where $O_i$ is the binary choice variable representing apple farmers’ decision to adopt green pest control technologies. This decision is determined by a random utility model, with the utility of adopting green pest control technology $(U_{adopted})$ compared to the utility of not adopting it $(U_{unadopted})$. If $O_i^{*}=U_{adopted}-U_{unadopted}>0$, then $O_i=1$, indicating that the apple farmer adopts green pest control technology. Conversely, if $O_i^{*}=U_{adopted}-U_{unadopted}\le 0$, then $O_i=0$, indicating that the farmer does not adopt green pest control technology.

Equation (7) represents the outcome equation, where $Y_i$ denotes the household income of the farmer, and $X_i$ includes the factors influencing the adoption decision of green pest control technologies and household income. ${\partial }_i,{\delta }_i,{\alpha }_i,{\beta }_i$ are the parameters to be estimated. Here, ${\alpha }_i$ represents the direct impact of green pest control technology adoption on farmers’ income, while ${\epsilon }_i$ and ${\mu }_i$ are the random error terms. $I_i$ represents the instrumental variable.

It should be noted that the ETR model uses the full information maximum likelihood estimation method to jointly estimate the correlation coefficients $({\rho }_{\epsilon \mu })$ of the error terms ${\epsilon }_i$ and ${\mu }_i$. By introducing the selection bias term from the first stage into the second stage, the model addresses the selection bias caused by both observable and unobservable factors, thereby mitigating the endogeneity problem caused by these factors. Additionally, the ETR model reports the Wald independence test results to measure the correlation between the selection equation and the outcome equation.

3.1.2. Model Construction for the Heterogeneous Analysis of the Income Effect of Green Pest Control Technology Adoption

The ETR model estimates the income effect of adopting green pest control technologies but overlooks the differences in the income effect across apple farmers at different income levels. Quantile regression (QR) can be used for regression analysis, but it does not effectively address endogeneity issues. Therefore, in this study, the author adopts the instrumental variable quantile regression (IVQR) model to further estimate the heterogeneity of the income effect of adopting green pest control technologies.

The construction principle of the target function for the IVQR model is divided into two steps. First, the structural equation is constructed:

$$\{\begin{array}{c}{Y}_i=E_i^{\prime }{\eta }_i(U_i)+X_i^{\prime }+v_i(U_i)\\ E_i=\rho (X_i,V_i,I_i)\\ \ddot{\tau }\to E_i^{\prime }{\eta }_i(\tau )+X_i^{\prime }{v}_i(\tau )\end{array}$$

(8)

In Equation (8), $Y_i$ represents the household income variable of apple farmers, and $X_i$ denotes the factors influencing household income. $U_i$ represents unobservable factors related to household income; $V_i$ represents other ignored and unrelated factors; and $I_i$ is the instrumental variable, while $E_i$ is the endogenous variable determined jointly by $X_i$, $V_i$, and $I_i$. $\rho (\cdot )$ is a function form, and $\ddot{\tau }$ is a strictly increasing function concerning the quantile point $\tau$.

Second, the equation $S_{Y_i}(\tau |E_i{X}_i)=e_i^{\prime }{\eta }_i({\eta }_i)+x_i^{\prime }{v}_i(\tau )$ is established. Based on the structural Equation (8), it can be determined that it is equivalent to

$$P[Y_i\le S_{Y_i}(\tau |E_i{X}_i)|I_i,X_i]=\tau$$

(9)

The above equation is equivalent to

$$Q_{Y_i-S_{Y_i}(\tau |E_i{X}_i)}(\tau |I_i,X_i)=0$$

(10)

Following the method of Koenker et al. [22], the objective function of quantile regression is

$$Q_{Y_i}(\tau |W_i)=\arg {\min }_{\theta (\tau )}E\left\{{\phi }_{\tau }[Y_i-f(W_i)]\right\}$$

(11)

In the equation, $f(\cdot )$ is the parameter function; $W_i$ represents all the factors related to the household income of apple farmers; and $\theta (\tau )$ is the parameter combination to be estimated corresponding to the quantile point $\tau$. By solving Equations (11) and (10) simultaneously, the objective function of the IVQR can be determined as

$$\arg {\min }_{\theta (\tau )}E\left\{{\phi }_{\tau }[Y_i-S_{Y_i}(\tau |E_i{X}_i)-f(I_i,X_i)]\right\}=0$$

(12)

3.2. Data Source

The micro-level data for this study were collected in 2022 through a questionnaire survey targeting apple farmers in the primary apple-producing regions around the Bohai Sea. The Bohai Bay apple production area is one of China’s most significant apple-producing regions, boasting the longest history of apple cultivation. Its continental monsoon climate, characterized by four distinct seasons, is highly suitable for ripening apple varieties. Furthermore, convenient water transport facilitates distribution. In 2021, this region alone accounted for approximately 15 million tons of apples, contributing significantly—up to 33%—to the national apple output. Within this crucial area, Shandong Province stands out for its renowned apple production, contributing 65% of the region’s total apple output. Specifically, Yantai and Linyi are major apple-producing areas within Shandong Province. These regions are characteristic of China’s agricultural landscape, where farming is predominantly carried out by smallholder households. While our survey data indicate an average apple farming area of approximately 6.29 Mu (1.03 acres) per household, reflecting this typical small-scale operation, the overall agricultural sector in Shandong Province, including Yantai and Linyi, is diversified. Beyond apple cultivation, the key agricultural products in these areas also include major food crops, such as wheat, corn, and economically important cash crops like peanuts, which are vital to the local economy and farmers’ livelihoods. The selection of Yantai and Linyi for this study is strategically justified due to their prominent role as major apple-producing centers within the Bohai Bay region and Shandong Province. These cities represent typical agricultural areas where apple cultivation is a significant economic activity, making them ideal for examining the adoption and economic impacts of green pest control technologies among apple farmers. Their diverse agricultural practices also provide a robust context for understanding broader farming decisions.

The surveyed areas included Muping District, Penglai District, and Qixia City in Yantai, Shandong Province, as well as Mengyin County and Yishui County in Linyi. The survey employed one-on-one questionnaire interviews and utilized stratified sampling to account for regional economic development levels, agricultural production scales, and geographical characteristics.

Our sampling frame was constructed using a multi-stage stratified random sampling method to ensure representativeness. First, Yantai and Linyi cities were selected as study sites due to their prominence in apple production within Shandong Province and the Bohai Bay region. Within these cities, specific counties/districts (Muping District, Penglai District, Qixia City in Yantai; Mengyin County, Yishui County in Linyi) were chosen based on their significant apple cultivation areas and typical agricultural characteristics.

In the second stage, within each selected county/district, two to three townships were purposefully selected based on their importance in apple production and their accessibility. For the third stage, two to three administrative villages were then randomly selected from a list of all apple-producing villages within each chosen township. This list was obtained from the local agricultural bureaus.

Finally, in the fourth stage, within each selected village, 10–20 apple farmers were randomly chosen from a list of all apple farming households provided by the village committee. These farmers were then approached for one-on-one interviews and questionnaire surveys. A total of 475 questionnaires were distributed, of which 409 were valid, yielding an effective response rate of 86.11%.

It should be noted that while the survey was conducted in 2022, the income and production data collected primarily pertain to the 2020 apple production and sales cycle. This temporal aspect is crucial because apples, as an economic crop, are frequently placed in cold storage and sold progressively across the year. Therefore, to obtain a complete and accurate representation of income derived from the 2020 harvest, data collection in 2022 was necessary to account for the full sales period that extended into the subsequent years.

3.3. Variable Selection

3.3.1. Outcome Variable

The outcome variable is the household income of the farmers. Total household income is the sum of wages, apple income, non-apple crop income, livestock and fishery income, self-employed business income, and other income, which is then logged. The natural logarithm of household income is used to mitigate the impact of heteroscedasticity and to normalize the distribution of the income variable, which often exhibits a skewed distribution in empirical studies. This transformation helps ensure that the assumptions of the regression models are better met, thereby improving the robustness and interpretability of our findings.

It should be noted that in 2020, some of the surveyed farmers had negative net apple income. This study considers such cases to be consistent with normal logic and general patterns, so they are retained. However, the logarithmic value of their apple net income is treated as zero. This approach is adopted for two primary reasons: First, the logarithm of a negative number is undefined, and treating negative income as zero allows for the inclusion of these observations in the logarithmic transformation of total household income without losing valuable data. Second, from an economic perspective, a negative net apple income, while indicating a loss for the apple farming activity in that specific year, still represents the farmer’s engagement in apple production. For the purpose of analyzing the impact of GPCT adoption on income, setting a minimum threshold (zero) for net apple income during the logarithmic transformation ensures that the income-generating potential, even when negative, is still accounted for in a manner that allows for consistent econometric analysis. This is a common practice in empirical studies dealing with income variables that can occasionally take negative values, aiming to retain as much information as possible while allowing for the application of logarithmic models.

3.3.2. Treatment Variable

The treatment variable is the decision to adopt green pest control technology. This is a binary choice variable. Based on existing studies [23,24,25], this study defines it as follows: In 2020, if a surveyed farmer adopted green pest control technology in their apple farming, they are assigned a value of 1, categorized into the treatment group; if they did not adopt the technology, they are assigned a value of 0, categorized into the control group.

3.3.3. Control Variables

Based on the literature reviewed by Chen et al. [26], Lou et al. [27], Ren et al. [28], and Singh et al. [29], the author identified the following control variables:

• Personal characteristics of the surveyed farmers, such as gender, age, education level, health status, village leader identity, party membership, etc.
• Household production and management characteristics, including the proportion of farming population, family size, operational area, relationship with neighbors, road conditions around the orchard, etc.
• Village characteristics, including whether there are agricultural subsidies and the distance between the village and the county town.

It should be noted that to exclude the interference of regional characteristics, economic levels, and other factors, the model also uses a set of regional dummy variables.

3.3.4. Instrumental Variables

The study selects two instrumental variables:

• The apple farmer’s concern about reduced yields from green technology. This variable refers to the question “Do you think green production technology will lead to a reduction in apple yields?”. It significantly affects the green pest control technology adoption decision of the surveyed apple farmers, but it is considered an exogenous variable with respect to their household income, making it suitable as an instrumental variable.
• The distance to the nearest green technology promotion location. This variable refers to the question “How far is your home from the nearest green technology promotion location?”. This variable is used as an alternative instrumental variable in the robustness check section. It significantly influences the adoption decision of green pest control technology by the surveyed farmers, but from the perspective of household income, it is exogenous, making it an ideal instrumental variable.

3.4. Descriptive Statistical Analysis

The variables involved in this study and their meanings are shown in

Table 1. Variable definition and descriptive statistics.

Variable Name	Variable Definition	Mean	Standard Deviation	Min	Max
Outcome Variables
Household Income	Total household income in 2020 (in CNY 10,000)	9.862	0.695	1.207	86.891
Net Apple Income	Net apple income in 2020 (in CNY 10,000)	6.255	1.566	1	64.238
Treatment Variables
Green Pest Control Adoption	Whether at least one green pest control technology was adopted in apple farming: Yes = 1, No = 0	0.545
Control Variables
Gender	Gender of the household head: 1 = Male, 0 = Female	0.968	0.176	0	1
Age	Actual age (in years)	55.232	9.854	18	90
Education Level	Years of education	8.112	2.794	0	19
Health Status	Compared to other farmers in the same village, how is your health status: 1 = Very poor; 2 = Poor; 3 = Average; 4 = Good; 5 = Very good	3.936	0.793	0	5
Village Leader Status	Whether the household head is a village leader: Yes = 1, No = 0	0.064	0.244	0	1
Party Member Status	Whether the household head is a party member: Yes = 1, No = 0	0.227	0.42	0	1
Farming Population Proportion	Proportion of household population involved in farming	0.858	0.202	0.143	1
Household Size	Number of household members living together	3.005	1.103	1	7
Farming Area	Area of apple farm (in Mu, logarithmic transformation applied)	1.57	0.704	0	4.382
Relationship with Relatives/Neighbors	Level of communication with relatives, friends, and neighbors: 1 = Very rarely; 2 = Rarely; 3 = Average; 4 = Often; 5 = Very often	3.941	0.932	1	5
Orchard Road Conditions	Whether there are motor vehicle accessible roads near the orchard: Yes = 1, No = 0	0.946	0.226	0	1
Agricultural Subsidies	Whether agricultural subsidies are provided in the village: Yes = 1, No = 0	0.665	0.473	0	1
Distance to Town	Distance from the residential location to the nearest town or township (in km)	4.241	3.133	0.01	14
Regional Dummy Variable	Yantai = 1, Linyi = 0	0.389	0.488	0	1
Instrumental Variables
Concern about Yield Reduction	Opinion on whether green production technology will reduce yields: 1 = Strongly disagree; 2 = Disagree; 3 = Neutral; 4 = Agree; 5 = Strongly agree	3.33	1.282	1	5
Distance to Green Technology Promotion Site	Distance from the residential location to the nearest green technology promotion site (in km)	6.067	4.84	0.1	25

Data Source: Compiled from the survey data.

. From the analysis of the table, we can see that 223 respondents adopted green pest control technology, accounting for 54.52% of the surveyed individuals, while the remaining 186 respondents did not adopt it, accounting for approximately 45.48%. The average household income of the respondents is CNY 98,665.76 (approximately USD 13,813.21), with a minimum value of CNY 2070 (approximately USD 289.80) and a maximum value of CNY 869,000 (approximately USD 121,660.00). The average net apple income of the respondents is CNY 62,275.92 (approximately USD 8718.63), with a minimum value of CNY −30,558 (approximately USD −4278.12) and a maximum value of CNY 851,500 (approximately USD 119,210.00). The average actual age of the respondents is 55.23 years, with a minimum of 18 years and a maximum of 90 years. The average number of years of education is 8.11 years, with a maximum of 19 years. Among the household heads, 26 households (6.36%) hold village leader positions, and 93 households (22.74%) are party members. The average family size is 3.005 people, with a minimum of 1 person and a maximum of 7 people. The average proportion of the farming population is 0.86. The average apple farming area is 6.29 Mu (approximately 1.03 acres), with a minimum of 1 Mu (approximately 0.16 acres) and a maximum of 80 Mu (approximately 13.12 acres). In total, 387 households (94.62%) have motor vehicle roads around their orchards. A total of 272 households (66.50%) are in villages that offer various agricultural subsidies. The average distance from the residential location to the town is 4.24 km, with a maximum of 14 km. The average opinion of the surveyed farmers regarding the potential yield reduction from green production technology is 3.33. Overall, 31 respondents (7.58%) strongly disagree; 101 respondents (24.69%) somewhat disagree; 75 respondents (18.34%) are neutral; 106 respondents (25.92%) somewhat agree; and 96 respondents (23.47%) strongly agree. The average distance from the residential location to the nearest green technology promotion site is 6.07 km, with a maximum of 25 km.

4. Empirical Results on the Income Effect of Green Pest Control Technology Adoption and Its Heterogeneity

4.1. Impact of Green Pest Control Technology Adoption on Apple Farmers’ Net Apple Income

The estimated results of the correlation coefficient $({\rho }_{\epsilon \mu })$ between the random error terms $({\epsilon }_i,{\mu }_i)$ of the selection equation and the outcome equation, as well as the Wald independence test, are shown in

Table 2. Joint estimation results of the ETR model on the impact of green pest control technology on farmers’ net apple income.

Variable	ETR Method		Robust Standard Error OLS Method
	Green Pest Control Technology Adoption	Net Apple Income	Net Apple Income
Green Pest Control Technology Adoption		0.242 *	1.942 ***
		(0.124)	(0.271)
Gender	0.722 *	0.933 ***	0.619 *
	(0.369)	(0.361)	(0.531)
Age	0.019 **	0.012	0.008
	(0.009)	(0.008)	(0.007)
Education Level	0.060 **	−0.000	−0.056 **
	(0.028)	(0.024)	(0.024)
Health Status	−0.272 ***	−0.110	−0.042
	(0.100)	(0.082)	(0.065)
Village Leader Status	−0.312	−0.202	−0.064
	(0.333)	(0.283)	(0.170)
Party Member Status	0.009	0.277 *	0.356 **
	(0.178)	(0.165)	(0.131)
Proportion of Farming Population	0.657	−0.953 **	−1.059 ***
	(0.579)	(0.433)	(0.372)
Household Size	0.000	−0.147 *	−0.137 *
	(0.090)	(0.082)	(0.061)
Farming Area	0.254 **	1.046 ***	1.051 ***
	(0.111)	(0.094)	(0.163)
Relationship with Relatives/Neighbors	0.004	0.039	0.011
	(0.085)	(0.069)	(0.055)
Road Conditions	0.945 ***	1.616 ***	1.556 ***
	(0.320)	(0.283)	(0.660)
Agricultural Subsidies	0.455 ***	0.204	0.082
	(0.164)	(0.140)	(0.113)
Distance to Town	−0.117	−0.060	−0.013
	(0.120)	(0.069)	(0.075)
Regional Dummy Variables	Controlled	Controlled	Controlled
Concerns about Yield Reduction from Green Technology	−0.163 ***
	(0.047)
Constant	−2.548 **	8.022 ***	8.200 ***
	(1.148)	(0.975)	(0.861)
Tests and Other Information
${\rho }_{\epsilon \mu }$	1.755 ** (0.128)
Lnsigma	−0.216 *** (0.038)
Goodness-of-Fit Test	217.99 ***
Log-Likelihood Value	−710.96788
Wald Independence Test	${\chi }^2(1)=61.42,prob>{\chi }^2=0.0000$
F-Test			26.75
R-Squared			0.5026
Sample Size	409	409	409

Note: Data were collected through the survey in 2022, and all calculations were performed using Stata 17; *, **, and *** represent statistical significance at the 10%, 5%, and 1% levels, respectively; values in parentheses are the standard errors.

. Analysis of the data in the table reveals that the correlation coefficient of 1.755 is statistically significant at the 1% level, indicating the presence of unobservable factors that simultaneously promote both the decision to adopt green pest control technology and net apple income. This means that the outcome equation is associated with selection bias. Without effective control of these factors, the income effect would be overestimated. The Wald independence test value of 217.99, statistically significant at the 1% level, rejects the hypothesis that the selection equation and the outcome equation are independent of each other. Therefore, it is necessary to jointly estimate both equations, which clearly applies to the ETR model.

Regarding the instrumental variable test, first, the Pearson correlation coefficient between the adoption decision of green pest control technology and farmers’ concerns about reduced yield due to green production technology is −0.413, statistically significant at the 5% level, demonstrating a relationship between these two variables.

To assess the validity of the instrumental variable, a weak instrument test is performed, and the F-statistic is 83.54 (well above 10), indicating no weak instrument problem. Second, in the second column of Table 2, the estimated value of farmers’ concern about reduced yield from green production technology is −0.163, statistically significant at the 1% level. This sufficiently demonstrates that concerns about yield reduction are a promoting factor in farmers’ decisions to adopt green pest control technology, thus confirming the validity of the instrumental variable.

4.1.1. ETR Model Estimation Results for the Selection Equation

The second column of Table 2 reports the estimation results of the ETR model’s selection equation. It can be seen that education level has a significant positive impact on the decision to adopt green pest control technology. Apple farmers with higher education levels have better learning and operational abilities to use green pest control technologies and are better able to understand the economic and ecological benefits of these technologies for apple farming. As a result, they are more likely to adopt them.

Farm size also has a significant positive effect on the decision to adopt green pest control technology. The larger the farming scale, the more necessary it is to adopt green pest control technologies to reduce the likelihood of pest and disease occurrence and spread. Therefore, farmers with larger operations are more likely to use green pest control technologies to ensure the quality and safety of apple products and mitigate pest and disease risks.

Road conditions and agricultural subsidies have significant positive effects on the adoption decision as well. Compared to orchards with poor road conditions, apple farmers are more likely to engage in intensive farming in orchards with better roads. Additionally, it is easier for them to use, install, and maintain the products and facilities associated with green pest control technologies. Agricultural subsidies, as external incentives, can help offset the high capital investment required for adopting green pest control technologies. Thus, both road conditions and agricultural subsidies promote the adoption of these technologies.

4.1.2. ETR Model Estimation Results for the Outcome Equation

The third and fourth columns of Table 2 report the estimation results of the ETR model’s outcome equation and the OLS model estimation results for the factors affecting net apple income, using robust standard errors. From the third column, it can be seen that the estimated coefficient for the adoption decision of green pest control technology is 0.242, statistically significant at the 1% level, which is lower than the 1.942 obtained in the fourth column. This suggests that if the ETR model is not used, and instead, the OLS estimation is applied, the result will be overestimated. The regression results indicate that the adoption of green pest control technology helps increase farmers’ net apple income, which supports the validity of hypothesis H1. Furthermore, overall, both in terms of significance and direction of effect, the estimation results from the ETR model and the OLS model are very similar. Therefore, the analysis will focus on the ETR model’s estimation results moving forward.

From the perspective of control variables, regarding household characteristics, gender and party membership of the household head have a significant positive impact on net apple income. The estimated coefficient for gender is 0.933, statistically significant at the 1% level, indicating that male household heads have a significant positive effect on improving net apple income. This could be because most female household heads lack the strong male labor force in the family, which limits the production capacity of apples due to a shortage of labor. Party membership, in the context of rural China, often implies greater engagement with local governance structures and community organizations. This engagement can facilitate access to valuable social capital, information networks, and agricultural support services. For instance, party members might have preferential access to training programs, agricultural extension services, or information regarding subsidized inputs and market prices, which can lead to more efficient farming practices and improved market opportunities. This enhanced access to resources and information, rather than mere compliance, contributes to higher net apple income. Similar findings on the role of social networks and institutional embeddedness in rural income generation have been observed in other studies [30,31,32]. As a result, the net apple income for party-member household heads is higher compared to non-party-member heads.

Regarding household production and management characteristics, the number of household members and the proportion of the farming population in the household have a significant negative impact on net apple income. This may be because, on the one hand, daily apple farming does not require a large labor force, and significant labor shortages only occur during bagging and harvesting. Even with a larger household population or a higher proportion of farming labor, the labor shortage is often not fully compensated, usually relying on mutual help from neighbors or hired labor. On the other hand, a larger household population typically means more elderly and children. Even if the elderly and children assist in apple farming, their help is minimal and may even have a counterproductive effect due to physical limitations or lack of experience.

Additionally, families with a lower proportion of farming labor might have family members working elsewhere, which can enhance the farming level of apple growers by broadening their horizons, increasing capital stock, and strengthening their social network. This, in turn, can increase net apple income. Therefore, the number of household members and the proportion of the farming population have a negative impact on net apple income.

Regarding farm size and road conditions, both have a significant positive impact on net apple income. This is easy to understand: larger orchards benefit from economies of scale, and orchards with better road conditions make it easier to transport agricultural inputs, equipment, and apple products. Therefore, farm size and road conditions positively influence the net apple income of farmers.

4.2. Impact of Green Pest Control Technology Adoption on Apple Farmers’ Household Income

To further analyze the income effect of adopting green pest control technology, this study continues to use the ETR model to estimate the impact of green pest control technology adoption on apple farmers’ household income.

Table 3. Joint estimation results of the ETR model on the impact of technology on farmers’ household income.

Variable	ETR Method		Robust Standard Error OLS Method
	Green Pest Control Technology Adoption	Household Income	Household Income
Green Pest Control Technology Adoption		0.209 *	0.411 ***
		(0.117)	(0.074)
Gender	0.786	0.321 **	0.284 *
	(0.508)	(0.137)	(0.167)
Age	0.012	−0.001	−0.001
	(0.011)	(0.003)	(0.003)
Education Level	0.128 ***	0.002	−0.004
	(0.036)	(0.010)	(0.010)
Health Status	−0.353 ***	−0.070 **	−0.062 *
	(0.136)	(0.031)	(0.032)
Village Leader Status	−0.410	−0.108	−0.091
	(0.412)	(0.106)	(0.105)
Party Member Status	−0.284	0.094	0.103
	(0.230)	(0.062)	(0.064)
Proportion of Farming Population	0.893	−0.336 **	−0.349 **
	(0.804)	(0.162)	(0.156)
Household Size	0.088	0.000	0.002
	(0.122)	(0.031)	(0.029)
Farming Area	0.126	0.671 ***	0.671 ***
	(0.149)	(0.035)	(0.047)
Relationship with Relatives/Neighbors	0.052	0.006	0.003
	(0.120)	(0.026)	(0.025)
Road Conditions	0.667	0.331 ***	0.324 **
	(0.429)	(0.106)	(0.149)
Agricultural Subsidies	−0.551 **	0.147 ***	0.181 ***
	(0.228)	(0.055)	(0.062)
Distance to Town	−0.294 *	−0.036	−0.030
	(0.158)	(0.026)	(0.026)
Regional Dummy Variables	Controlled	Controlled	Controlled
Concerns about Yield Reduction from Green Technology	−0.399 ***
	(0.075)
Constant	−1.661	9.935 ***	9.956 ***
	(1.577)	(0.366)	(0.366)
Tests and Other Information
${\rho }_{\epsilon \mu }$	0.306 ** (0.150)
Lnsigma	−0.765 *** (0.037)
Goodness-of-Fit Test	479.28 ***
Log-Likelihood Value	−381.90328
Wald Independence Test	${\chi }^2(1)=4.08,prob>{\chi }^2=0.0434$
F-Test			26.32
R-Squared			0.5611
Sample Size	409	409	409

Note: Data were collected through the survey in 2022, and all calculations were performed using Stata 15; *, **, and *** represent statistical significance at the 10%, 5%, and 1% levels, respectively; values in parentheses are the standard errors.

presents the estimation results of the correlation coefficient $({\rho }_{\epsilon \mu })$ between the random error terms $({\epsilon }_i,{\mu }_i)$ of the selection equation and the outcome equation, as well as the Wald independence test. The data in the table show that the correlation coefficient $({\rho }_{\epsilon \mu })$ of 0.307 is statistically significant at the 5% level, indicating the presence of unobservable factors that simultaneously promote both the adoption decision of green pest control technology and household income. This means that the outcome equation suffers from selection bias, and without effective control measures, the income effect would be overestimated. The Wald independence test value of 479.28, statistically significant at the 1% level, rejects the hypothesis that the selection equation and the outcome equation are independent of each other. Therefore, it is necessary to jointly estimate both equations, and the ETR model is fully appropriate.

Regarding the instrumental variable test, first, the Pearson correlation coefficient between the adoption decision of green pest control technology and apple farmers’ concerns about yield reduction due to green production technology is −0.413, statistically significant at the 5% level. This demonstrates a relationship between the adoption decision and farmers’ concerns about yield reduction. To verify the validity of the instrumental variable, the author performed a weak instrument test, and the F-statistic of 83.54 (well above 10) rules out the weak instrument problem. Additionally, in the second column of Table 3, the estimated value of farmers’ concerns about yield reduction from green production technology is −0.399, statistically significant at the 1% level, which is sufficient to show that yield reduction concerns significantly influence the decision to adopt green pest control technology. Therefore, the choice of the instrumental variable is appropriate.

From the third column of Table 3, it can be seen that the estimated coefficient for the adoption decision of green pest control technology is 0.209, statistically significant at the 10% level, indicating that the adoption of green pest control technology helps increase the apple farmers’ household income. In the fourth column, the estimated coefficient for the adoption decision of green pest control technology is 0.411, statistically significant at the 1% level. This shows that if the ETR method is not used and the robust standard errors from the OLS method are applied, the impact of green pest control technology adoption on apple farmers’ household income will be overestimated. Therefore, the model results from the OLS method should not be used for analysis. Based on the results from the ETR model, hypothesis H2, as mentioned earlier, is supported.

4.3. Heterogeneous Analysis of the Income Effect of Green Pest Control Technology Adoption

The previous analysis confirmed the income effect of adopting green pest control technology. This section primarily focuses on heterogeneous analysis, specifically examining the differences and dynamic changes in the income effect of green pest control technology adoption across apple farmers with different income levels. Therefore, the author continues to use farmers’ concerns about yield reduction from green production technology as the instrumental variable and performs a regression analysis using the IVQR model. To provide a complete report, quantile points of 10%, 30%, 50%, 70%, and 90% are selected. It is worth mentioning that a particular quantile point represents the proportion of the sample below that quantile point relative to the total sample.

The p-value of the model’s J-test is 0.072, which is statistically significant at the 10% level. According to

Table 4. IVQR model estimation results for the impact of green pest control technology on household income at different quantiles.

Variable	IVQR Model
	10% Quantile	30% Quantile	50% Quantile	70% Quantile	90% Quantile
Green Pest Control Technology Adoption	0.646 ***	0.451 ***	0.357 ***	0.285 ***	0.161
	(0.120)	(0.075)	(0.073)	(0.083)	(0.113)
Gender	−0.187	−0.085	−0.036	0.002	0.066
	(0.154)	(0.189)	(0.223)	(0.254)	(0.311)
Age	−0.025 ***	−0.020 ***	−0.017 ***	−0.015 ***	−0.011 **
	(0.006)	(0.004)	(0.004)	(0.004)	(0.005)
Education Level	−0.027	−0.015	−0.010	−0.005	0.002
	(0.026)	(0.016)	(0.012)	(0.011)	(0.013)
Health Status	−0.208 ***	−0.153 ***	−0.126 ***	−0.106 ***	−0.071
	(0.050)	(0.040)	(0.041)	(0.043)	(0.052)
Village Leader Status	−0.043	−0.085	−0.105	−0.121	−0.147
	(0.262)	(0.160)	(0.125)	(0.113)	(0.130)
Party Member Status	0.039	0.042	0.043	0.044	0.046
	(0.138)	(0.084)	(0.070)	(0.068)	(0.085)
Proportion of Farming Population	−2.381 ***	−1.560 ***	−1.167 ***	−0.860 ***	−0.341
	(0.602)	(0.329)	(0.237)	(0.214)	(0.289)
Household Size	−0.319 ***	−0.187 ***	−0.123 ***	−0.074 *	0.010
	(0.086)	(0.050)	(0.041)	(0.040)	(0.052)
Farming Area	0.725 ***	0.656 ***	0.623 ***	0.597 ***	0.554 ***
	(0.075)	(0.054)	(0.050)	(0.052)	(0.061)
Relationship with Relatives/Neighbors	−0.085 **	−0.048 *	−0.031	−0.017	0.006
	(0.041)	(0.029)	(0.028)	(0.030)	(0.037)
Road Conditions	0.022	0.030	0.034	0.037	0.042
	(0.116)	(0.127)	(0.148)	(0.168)	(0.207)
Agricultural Subsidies	0.256 **	0.187 **	0.153 **	0.127 *	0.083
	(0.123)	(0.080)	(0.072)	(0.074)	(0.093)
Distance to Town	−0.013	−0.059	−0.081 **	−0.098 ***	−0.127 ***
	(0.055)	(0.038)	(0.035)	(0.036)	(0.044)
Regional Dummy Variables	Controlled	Controlled	Controlled	Controlled	Controlled
Constant	15.367 ***	13.984 ***	13.322 ***	12.805 ***	11.932 ***
	(0.856)	(0.619)	(0.613)	(0.662)	(0.809)
Sample Size	409	409	409	409	409

Note: Data were collected through the survey in 2022, and all calculations were performed using Stata 15; *, **, and *** represent statistical significance at the 10%, 5%, and 1% levels, respectively; values in parentheses are the standard errors.

, at the 10%, 30%, 50%, and 70% quantile points, the estimated coefficients for the decision to adopt green pest control technology are 0.646, 0.451, 0.357, and 0.285, all statistically significant at the 1% level. At the 90% quantile point, the coefficient is 0.161, which is not statistically significant. This strongly demonstrates that the income effect of green pest control technology adoption is more pronounced at lower quantiles. In other words, the lower the income level of the farmers, the more significant the income effect of adopting green pest control technology. This is likely because high-income apple farmers have advantages in terms of economic foundation, cultural quality, and technical expertise; therefore, the income effect of adopting green pest control technology represents a smaller proportion of their overall income. On the other hand, for low-income farmers, the income effect from adopting green pest control technology accounts for a higher proportion of their household income. Therefore, the income effect of adopting green pest control technology is more significant for low-income farmers, which supports the validity of hypothesis H3.

4.4. Robustness Test

This study uses the distance between the respondents’ residences and the nearest green pest control technology promotion location as an instrumental variable. The ETR model is used again to regress the net income from apples and household income of the respondents. Additionally, the IVQR model is used to conduct a heterogeneous analysis of the income effect of green pest control technology adoption, testing the robustness of the regression results.

Regarding the instrumental variable test, first, the Pearson correlation coefficient between the decision to adopt green pest control technology and the distance between apple farmers’ residences and the promotion location of the green pest control technology is −0.374, which is statistically significant at the 10% level. This confirms the relationship between the decision to adopt green pest control technology and the distance to the promotion location. To check the validity of the instrumental variable, a weak instrument test is conducted. The F-statistic is 83.54 (much greater than 10), ruling out the problem of weak instruments.

Second, in

Table 5. Robustness test results of baseline regression.

Variable	ETR Method		ETR Method
	Green Pest Control Technology Adoption	Net Apple Income	Green Pest Control Technology Adoption	Household Income
Green Pest Control Technology Adoption		0.143 *		0.193 *
		(0.118)		(0.132)
Gender	0.777 **	0.952 ***	0.982 *	0.324 **
	(0.366)	(0.366)	(0.505)	(0.137)
Age	0.022 ***	0.013	0.015	−0.001
	(0.008)	(0.008)	(0.011)	(0.003)
Education Level	0.054 **	0.003	0.129 ***	0.003
	(0.026)	(0.024)	(0.035)	(0.010)
Health Status	−0.225 **	−0.114	−0.250 *	−0.071 **
	(0.096)	(0.083)	(0.131)	(0.031)
Village Leader Status	−0.497	−0.210	−0.749 *	−0.109
	(0.322)	(0.286)	(0.400)	(0.106)
Party Member Status	0.132	0.273	−0.052	0.093
	(0.173)	(0.167)	(0.225)	(0.062)
Proportion of Farming Population	0.433	−0.946 **	0.746	−0.335 **
	(0.574)	(0.438)	(0.810)	(0.163)
Household Size	−0.015	−0.148 *	0.030	0.000
	(0.088)	(0.083)	(0.117)	(0.031)
Farming Area	0.258 **	1.046 ***	0.121	0.671 ***
	(0.108)	(0.095)	(0.147)	(0.035)
Relationship with Relatives/Neighbors	0.016	0.041	0.062	0.007
	(0.090)	(0.069)	(0.121)	(0.026)
Road Conditions	0.879 **	1.619 ***	0.503	0.332 ***
	(0.352)	(0.286)	(0.445)	(0.106)
Agricultural Subsidies	−0.404 **	−0.220	−0.525 **	0.144 **
	(0.158)	(0.142)	(0.227)	(0.056)
Distance to Town	−0.013	−0.063	−0.113	−0.036
	(0.114)	(0.069)	(0.161)	(0.026)
Regional Dummy Variables	Controlled	Controlled	Controlled	Controlled
Concerns about Yield Reduction from Green Technology	−0.037 ***		−0.078 ***
	(0.012)		(0.022)
Constant	−3.165 ***	−3.165 ***	−3.162 **	9.933 ***
	(1.139)	(1.139)	(1.539)	(0.366)
Tests and Other Information
${\rho }_{\epsilon \mu }$	1.859 *** (0.128)		0.309 * (0.166)
Lnsigma	0.229 *** (0.038)		−0.764 *** (0.038)
Goodness-of-Fit Test	210.40 ***		476.53 ***
Log-Likelihood Value	−713.221		−390.508
Wald Independence Test	${\chi }^2(1)=73.26,prob>{\chi }^2=0.0000$		${\chi }^2(1)=3.22,prob>{\chi }^2=0.0729$
Sample Size	409	409	409	409

Note: Data were collected through the survey in 2022, and all calculations were performed using Stata 15; *, **, and *** represent statistical significance at the 10%, 5%, and 1% levels, respectively; values in parentheses are the standard errors.

, the estimated values for the distance between the apple farmers’ residences and the green pest control technology promotion location in both adoption models are statistically significant at the 1% level. This sufficiently indicates that the distance to the promotion location significantly influences the decision to adopt green pest control technology, validating the choice of the instrumental variable.

From the third and fifth columns of Table 5, it can be seen that the estimated coefficients for the decision to adopt green pest control technology are positive and statistically significant at the 10% level. The adoption of green pest control technology has a significant positive effect on both the net income from apples and household income for the farmers. Additionally, from Table 5, it is determined that at the 10%, 30%, and 50% quantile points, the estimated coefficients for the decision to adopt green pest control technology are positive and statistically significant at the 1% level. However, the regression results at the 70% and 90% quantile points are not statistically significant, which essentially verifies that the income effect of green pest control technology adoption is stronger for low-income apple farmers.

This fully proves that the robustness test results from Table 5 and

Table 6. Robustness test results of heterogeneous regression.

Variable	IVQR Model
	10% Quantile	30% Quantile	50% Quantile	70% Quantile	90% Quantile
Green Pest Control Technology Adoption	0.658 **	0.410 **	0.248 **	0.127	−0.023
	(0.339)	(0.196)	(0.128)	(0.115)	(0.159)
Gender	0.245	1.943	3.055	3.879	4.910
	(1.362)	(2.056)	(2.663)	(3.134)	(3.744)
Age	−0.044	−0.041 *	−0.040 **	−0.038 *	−0.037 ***
	(0.035)	(0.026)	(0.020)	(0.016)	(0.011)
Education Level	0.019	0.012	0.008	0.005	0.001
	(0.071)	(0.042)	(0.026)	(0.020)	(0.025)
Health Status	−0.372	−0.295 *	−0.245 **	−0.207 *	−0.160
	(0.251)	(0.164)	(0.120)	(0.104)	(0.114)
Village Leader Status	−0.542	−0.238	−0.038	0.109	0.294
	(2.051)	(1.121)	(0.546)	(0.283)	(0.630)
Party Member Status	0.058	−0.059	−0.135	−0.192	−0.263
	(0.265)	(0.178)	(0.146)	(0.145)	(0.172)
Proportion of Farming Population	−4.811	−4.051	−3.553 **	−3.185 ***	−2.723 ***
	(2.945)	(1.996)	(1.449)	(1.133)	(0.980)
Household Size	−0.653	−0.525	−0.440	−0.378 **	−0.300 **
	(0.494)	(0.344)	(0.251)	(0.187)	(0.126)
Farming Area	0.838 *	0.779 ***	0.740 ***	0.711 ***	0.675 ***
	(0.295)	(0.192)	(0.131)	(0.097)	(0.087)
Relationship with Relatives/Neighbors	−0.128	−0.106	−0.091	−0.080	−0.066
	(0.166)	(0.104)	(0.073)	(0.064)	(0.076)
Road Conditions	−0.200	−0.518	−0.727	−0.882	−1.075
	(0.491)	(0.504)	(0.657)	(0.805)	(1.011)
Agricultural Subsidies	0.214	0.133	0.080	0.041	−0.009
	(0.342)	(0.207)	(0.137)	(0.115)	(0.147)
Distance to Town	−0.016	−0.110	−0.172 **	−0.217 **	−0.275 **
	(0.116)	(0.078)	(0.082)	(0.099)	(0.131)
Regional Dummy Variables	Controlled	Controlled	Controlled	Controlled	Controlled
Constant	19.093 *	17.136 ***	15.854 ***	14.905 ***	13.716 ***
	(7.537)	(6.197)	(5.456)	(4.994)	(4.567)
Sample Size	409	409	409	409	409

Note: Data were collected through the survey in 2022, and all calculations were performed using Stata 15; *, **, and *** represent statistical significance at the 10%, 5%, and 1% levels, respectively; values in parentheses are the standard errors.

are consistent with the estimation results from the baseline regression using the ETR model and IVQR model. Although the regression coefficients are not identical, the results confirm that the baseline regression meets the robustness requirements of this study.

5. Discussion

Our study provides robust empirical evidence on the economic effects of green pest control technology (GPCT) adoption on apple farmers’ income in China, particularly highlighting the heterogeneous impacts across different income strata. The significant positive effect of GPCT adoption on both net apple income and overall household income, as revealed by both endogenous treatment effects regression (ETR) and instrumental variable quantile regression (IVQR) models, consistently supports our main hypothesis. This finding aligns with the principles of rational farmer behavior, where farmers, as economic agents, strategically adopt innovations that are perceived to enhance their utility and profitability. GPCT adoption, in this context, represents a strategic decision by farmers to optimize their factor allocation efficiency by reducing reliance on costly and environmentally damaging chemical inputs while simultaneously improving the quality and quantity of their output. This approach is consistent with a growing body of literature that underscores the income-enhancing potential of sustainable agricultural practices, emphasizing that environmentally friendly technologies can also be economically beneficial for farmers, especially in understanding their impact on farm household income [5].

The income improvement from GPCT adoption can be attributed to several synergistic mechanisms. Firstly, by reducing the reliance on synthetic chemical pesticides, farmers incur lower input costs, directly improving their net agricultural income. Secondly, effective green pest management often leads to healthier apple trees and higher quality produce, reducing pesticide residues and physical damage to fruits, which can command premium prices in the market and increase overall marketable yield. Thirdly, diminished exposure to harmful chemicals can lead to improved farmer health and labor productivity, thereby indirectly contributing to household income by reducing health-related expenditures and increasing available labor days. This multi-faceted impact demonstrates how GPCT adoption contributes to a more efficient and productive agricultural system for smallholder farmers.

A particularly salient finding is the more pronounced income-boosting effect for lower-income farmers. The IVQR model results explicitly demonstrate a larger positive impact at the 10%, 30%, and 50% quantiles of income distribution, with the effect becoming insignificant at higher quantiles (70% and 90%). This heterogeneity can be attributed to several factors. Lower-income farmers often face greater resource constraints, have less access to capital-intensive traditional pest control methods, and may experience higher baseline pest-induced losses due to suboptimal management. Consequently, the adoption of effective GPCTs represents a more substantial improvement in their pest management strategies, directly translating into enhanced apple production efficiency and, subsequently, a more significant marginal increase in income. This aligns with economic theories suggesting diminishing marginal returns due to innovation as initial resource endowments or capabilities increase and echoes previous research indicating that technological innovations tend to yield higher relative benefits for resource-constrained households due to a larger initial gap in capabilities and resource endowments [33,34]. High-income farmers, conversely, may already operate near their production frontier through existing advanced inputs or management practices, thus experiencing a comparatively smaller incremental gain from GPCT adoption. Their more robust economic standing might also mean that the immediate income benefit from a single technology adoption is less impactful on their overall larger income base.

Our findings align with a growing body of empirical research highlighting the economic benefits of sustainable agricultural technologies, both in China and globally. For example, studies on the adoption of integrated pest management (IPM) and organic farming practices have consistently shown positive impacts on farmer income through mechanisms such as reduced input costs and improved product quality. Zhang et al. [35] examined smallholder rice farming in Poyang Lake, China, and found that while rice farmers with a heavy reliance on rice cultivation had higher yields, their income was lower compared to those with diversified livelihoods, primarily due to the low returns from rice farming. The study suggests that increasing the economic returns from rice farming could boost smallholder incomes. Similarly, Cai et al. [36] found that agricultural socialized services (ASS) significantly improved the technical efficiency of smallholder rice farmers in southern China by providing access to modern techniques and encouraging sustainable practices, ultimately enhancing productivity and income. Mussa et al. [37] reported that the use of inorganic fertilizers increased farm incomes for smallholder rice farmers in Tanzania, indicating that better access to fertilizers and extension services could further enhance productivity and livelihoods.

While many studies affirm a positive income effect from green technology adoption, our research provides novel insights, particularly regarding the heterogeneous effects across income groups. While some studies have explored general average treatment effects, the detailed quantile regression analysis presented here offers a more nuanced understanding of who benefits most from GPCTs. This finding, namely that lower-income farmers experience a disproportionately larger income boost, is a key contribution to the literature. It suggests that GPCTs may serve as a potent tool for poverty alleviation and income redistribution within agricultural communities, contrasting with some technologies that might disproportionately benefit larger, more resourced farms due to high initial investment costs or economies of scale. The observed consistency of our results with the general trend of income benefits from green technologies, without significant unexpected deviations, further reinforces the reliability of GPCTs as a viable pathway for sustainable agricultural development and income enhancement for smallholder farmers.

Furthermore, our findings on the influence of demographic and farm-specific factors corroborate existing understanding of agricultural household economics. The positive correlation between household size and farming area with income reinforces the notion that economies of scale continue to play a role in agricultural productivity and income generation, even for specialized crops like apples. The negative impact of health status on income underscores the critical importance of human capital in farming; farmers facing health challenges may experience reduced labor capacity, hindering their ability to engage in productive agricultural practices and adopt new technologies efficiently. The significant positive effect of agricultural subsidies on income further highlights the potential of policy support in fostering technology adoption and improving livelihoods, suggesting that financial incentives can effectively mitigate initial adoption costs or risks, particularly for vulnerable farmer groups. These interpretations collectively not only confirm the direct economic benefits of GPCTs but also shed light on the mechanisms through which these benefits are realized, especially within heterogeneous farming populations. Our study extends the current literature by providing nuanced evidence on the differential impacts across income groups, which is often overlooked in average treatment effect analyses.

The findings of this study, while empirically grounded in the context of apple farming in China, offer insights into the mechanisms by which green pest control technologies contribute to farmer income. The theoretical framework, based on rational economic behavior and utility maximization, is broadly applicable to agricultural decision making globally. Similarly, the pathways through which these technologies influence income—primarily via enhanced yields and improved product quality—are common to agricultural systems worldwide. The observed heterogeneous impacts across different farmer income levels also resonate with general agricultural economic principles. However, it is crucial to note that the specific magnitudes and direct quantitative implications of these findings are tied to the unique characteristics of Chinese apple cultivation and its associated socio-economic and policy environment. Therefore, while the mechanisms discussed may be transferable, the results themselves should be interpreted within this specific context. Future research should aim to test these hypotheses and mechanisms in diverse agricultural settings and geographical regions to further explore their generalizability.

6. Conclusions

This study rigorously explored the impact of adopting green pest control technology on the income levels of apple farmers in China, focusing on its heterogeneity across different income strata and the influence of socio-economic factors. Using endogenous treatment effects regression (ETR) and instrumental variable quantile regression (IVQR) models, our analysis provides compelling evidence that GPCT adoption significantly and positively affects apple farmers’ net apple income and overall household income, confirming its income-boosting effect. Importantly, this income-boosting effect is found to be more pronounced for lower-income farmers, aligning with our hypothesis that these farmers benefit more from the adoption of green pest control technologies through improved pest management and enhanced apple production efficiency. The robustness of our findings is further corroborated by various tests, including the use of an alternative instrumental variable, which strengthens the validity of the causal relationship identified.

The implications of these findings are substantial and extend beyond the immediate context of apple farming. Firstly, they underscore the critical importance of targeted interventions aimed at encouraging green pest control technology adoption, particularly among lower-income farmers. Policies and programs designed to provide financial and technical support to these farmers could significantly amplify the benefits of GPCTs, thereby contributing to poverty alleviation and sustainable agricultural development. The positive effect of agricultural subsidies on income further highlights the potential of integrating financial support with technological innovations to foster greater adoption of green technologies in agriculture. These insights offer valuable references for policymakers seeking to promote sustainable agricultural practices and enhance rural livelihoods.

While this study offers significant contributions, it also has limitations that suggest avenues for future research. Our focus on apple farmers means that the results may not necessarily generalize directly to other types of crops or agricultural settings. Future studies could explore the impact of green pest control technology across diverse crops and regions, particularly in different economic and ecological contexts, to test the broader applicability of our findings. Additionally, while this study primarily concentrates on income effects, future research could consider a wider range of social and environmental outcomes, such as the impact on biodiversity, ecosystem health, and community well-being. A comprehensive understanding of the full spectrum of benefits and costs associated with green pest control technologies will be essential for designing more effective and holistic agricultural policies.

Finally, future research could further investigate the role of knowledge dissemination and farmer education in enhancing the adoption of green pest control technologies. While our study focused on the socio-economic factors influencing adoption, it would be valuable to explore how information accessibility, extension services, and peer learning networks influence farmers’ willingness and ability to adopt these technologies, thus contributing to more effective promotion strategies.