EN中文

Journal of Resources and Ecology >

2024 , Vol. 15 >Issue 2: 243 - 257

DOI: https://doi.org/10.5814/j.issn.1674-764x.2024.02.001

Agroecology and Agricultural Development

The Impact of Agricultural Technology Services on the Efficiency of Green Grain Production: An Analysis based on the Generalized Stochastic Forest Model

  • ZHANG Yuedong ,
  • ZHENG Yifang ,
  • XU Jiaxian , *
Expand
  • School of Public Administration and Law, Fujian Agriculture and Forestry University, Fuzhou 350002, China
*XU Jiaxian, E-mail:

ZHANG Yuedong, E-mail:

Received date: 2023-09-05

  Accepted date: 2023-11-15

  Online published: 2024-03-14

Supported by

The Major Project Funding for Social Science Research Base in Fujian Province Social Science Planning(FJ2022MJDZ022)

Fold

Abstract

The green production of food currently faces the challenge of technological shortage. As agricultural technology services are an important source of technical knowledge for farmers, it is of great significance to explore how agricultural technology services influence the efficiency of green food production in order to improve its productivity. This study utilized data from the China Land and Economy Survey (CLES). Firstly, the three-stage DEA model was used to calculate the efficiency of green food production. Secondly, the generalized random forest model was used to empirically test the impact, heterogeneity, and pathways of agricultural technology services on the efficiency of green food production. This study found that: (1) Agricultural technology services have a significant positive impact on the efficiency of green food production. Among the stages, pre-production and mid-production agricultural technology services significantly improve the efficiency of green food production. However, post-production agricultural technology services do not show a significant impact. Additionally, the combination of agricultural technology services has a significant improvement effect on the efficiency of green food production. (2) The marginal effects of resource allocation by farmers have heterogeneity in terms of the impact on the efficiency of agricultural technology services in improving green food production. An increase in the proportion of the family agricultural labor force and the expansion of family-operated arable land scale significantly reduce the returns of agricultural technology services. (3) An analysis of mechanism variables showed that agricultural technology services mainly affect the efficiency of green food production through pesticide and fertilizer usage behavior, and significantly contribute to its improvement. The government should expand the scope of agricultural technology services and fully utilize their potential to improve the efficiency of green food production. Tailored strategies and policies should be implemented to guide the concept of green production during the provision of agricultural technology services, thereby promoting sustainable agricultural practices among farmers.

Key words： agricultural technical services; green food production efficiency; generalized random forests

Cite this article

ZHANG Yuedong , ZHENG Yifang , XU Jiaxian . The Impact of Agricultural Technology Services on the Efficiency of Green Grain Production: An Analysis based on the Generalized Stochastic Forest Model[J]. Journal of Resources and Ecology, 2024 , 15(2) : 243 -257 . DOI: 10.5814/j.issn.1674-764x.2024.02.001

1 Introduction

The discussion on carbon emissions is currently focused on the manufacturing industry, as it is the primary contributor to carbon emissions (Wang et al., 2016). However, agricultural carbon emissions play a significant role in China’s overall carbon emissions due to its status as an agricultural country. In recent years, grain production has faced various challenges, including high costs, high energy consumption, and low efficiency (Cui et al., 2022; Yu et al., 2022), which are primarily attributed to the current shortage of rural labor force. Farmers, in their pursuit of increased production, often resort to the indiscriminate use of fertilizers, pesticides, and other production materials, leading to significant carbon emissions and exacerbating unsustainable land use, thereby posing a threat to China’s food security. Therefore, enhancing the efficiency of green grain production is essential not only for implementing energy-saving and emission reduction measures in agriculture, but also for ensuring China’s food security.
Agricultural technology plays a pivotal role in enhancing the efficiency of green grain production, while facilitating the saving of energy and emission reduction in the process, and ensuring food security. As key contributors to the agricultural sector, farmers directly influence the efficiency of green grain production through their command of advanced production techniques. By enhancing farmers’ proficiency in production technology, we can optimize the allocation of agricultural resources, increase the farmers’ income, and achieve environmentally-friendly development by saving energy and reducing emissions. Agricultural technology services are essential for enhancing farmers’ production technology and advancing the modernization of agricultural production. Effective agricultural technology should encourage farmers to adopt advanced agricultural production techniques and enhance the efficiency of green grain production (Tong et al., 2018).
Agricultural technology services are highly effective in improving the efficiency of green food production. The purpose of agricultural technology services is to enhance the efficiency of green food production and facilitate the modernization of the industry. Effective agricultural technology should encourage farmers to adopt advanced agricultural production techniques (Ruttan, 1986; Hu et al., 2022). In 2021, the State Council issued the “14th Five-Year Plan (2021-2025)” to promote the modernization of agriculture and rural areas, which clearly points out that it is necessary to strengthen the construction of the agricultural science and technology socialized service system. Promoting the development of agricultural technology services is not only a practical necessity for achieving the “Dual-Carbon”①(①China is committed to increasing its nationally owned contribution, adopting stronger policies and measures, striving to peak CO2 emissions by 2030, and striving to achieve carbon neutrality by 2060. Abbreviation: “Dual-Carbon”.) goal and guaranteeing food security, but it also provides important support for the development of agricultural modernization.
This study contributes to that strategy in several ways. Firstly, it differs from the current studies that primarily focus on macro data and measure the green production efficiency of food in different regions. Instead, this study utilizes micro subjective data to examine the green production efficiency of different plots. It delves deeper into the impact of agricultural technology services on the green production efficiency of food, thereby providing more targeted recommendations. Secondly, while some existing studies have investigated the production efficiency of plots, they often overlook the differences between different plots, leading to a certain amount of bias in measuring green production efficiency. In contrast, this study accounts for the varying disaster conditions and land quality of different plots. It employs a three-phase DEA model that excludes the effects of external environmental factors and random errors on the input variables. This approach allows for a more accurate measurement of the green production efficiency of individual grains. Thirdly, this study applies machine learning methods to analyze the impact of agricultural technology services on food green production efficiency. This extends the quantitative research methods used in this field and provides a practical case study for the application of machine learning in policy recommendation. Overall, these contributions enhance our understanding of the relationship between agricultural technology services and green production efficiency, and provide valuable insights for policymakers and practitioners in the field.

2 Literature review

The current academic research on agricultural technology services mainly focuses on several aspects. The first is the impact of agricultural technology services on the adoption of green agricultural technologies. Some scholars believe that, overall, agricultural technology services have a positive effect on the adoption of green technologies by farmers. However, the effectiveness of agricultural technology services is influenced by various individual factors. A disparity exists between the advancement of green agricultural technologies and the technological proficiency of farmers. Consequently, farmers make adoption decisions based on the benefits associated with implementing these technologies (Feder and Slade, 1986). Xia et al. (2013) and other researchers have emphasized that agricultural progress can only be achieved through effective agricultural extension. They also argue that the effectiveness of agricultural extension is determined by the provision of green agricultural technologies that meet the needs of the farmers. Ying and Zhu (2015) conducted an empirical analysis of agricultural rice farmers and found that the extension of green agricultural technologies facilitates the adoption of technology by farmers. However, Tong et al. (2018) also noted that the impact of extending green agricultural technologies on farmers’ technology adoption is influenced by the economic status of the farmers. Additionally, agricultural technology services positively contribute to the environmentally-friendly production of grain. The promotion of agricultural technology can expedite the transfer of green agricultural technologies, thereby significantly advancing agricultural development. Current research indicates that agricultural technology services can improve the quality of production factor inputs, which in turn improves agricultural productivity (Wen and Wu, 2016). The present study concluded that agricultural technology services can enhance the quality of environmentally friendly grain production inputs, leading to an increase in agricultural productivity. Previous studies have investigated the impact of agricultural technology services on enhancing rice productivity (Xiang et al., 2023). According to a study by Rosegrant and Pingali (1994) on India’s agricultural productivity measurement, that country’s high agricultural productivity is primarily attributed to public agricultural research and extension efforts. Agricultural technology training is an important method for assisting farmers in learning and adopting new technologies. It can directly or indirectly facilitate the adoption of advanced technologies, thereby enhancing the farmers’ green grain productivity (Yu and Will, 2019). This, in turn, improves the overall efficiency of agricultural production.
In the existing literature, current research focuses primarily on the influences of agricultural technology services on technology adoption and the efficiency of green grain production. However, an overall analysis can identify several deficiencies. Firstly, there is insufficient research on the impact of agricultural technology services on the efficiency of green grain production. Given the current “Dual-Carbon” background, since agriculture is a significant contributor to carbon emissions, it is crucial to focus not only on improving the efficiency of green grain production but also on exploring strategies for sustainable agricultural development. Secondly, most studies use the traditional linear regression model, and there are different degrees of selection problems in agricultural technical services. The use of the traditional regression model also introduces an unsolvable issue of endogeneity. Additionally, fixed effects and propensity score matching methods are often employed, but their effectiveness diminishes as the number of variables increases.
The aim of this study is to address the impact of agricultural technology services on the efficiency of green food production. This will be accomplished by considering the perspective of food-growing farmers. The research questions that will be answered are as follows: What is the impact of agricultural technology services on the efficiency of green food production? How do the diverse characteristics of farmers impact the benefits of agricultural technology services? What are the mechanisms by which agricultural technology services enhance the efficiency of green food production? This study was conducted in two stages. Firstly, it accounted for the varying disaster conditions and land quality of different plots. Secondly, a three-stage DEA model was employed to assess the efficiency of green food production. In the second step, we improved upon previous studies by using the Tobit model to measure grain green production efficiency. However, we acknowledge that there may be selective bias in the selection of agricultural technology service samples, which can lead to endogeneity problems. To address this issue, we employed machine learning techniques for causal identification. Specifically, we constructed a generalized random forest model to estimate the treatment effect of agricultural technology services. This approach accounts for the heterogeneity of individuals and ensures reliable estimation of the treatment effect (Gao et al., 2017; Liu et al., 2021). By using this model, we were able to fully consider individual heterogeneity and enhance the reliability of the treatment effect estimation. This can help to address the limitations of previous studies in terms of causal identification.

3 Theoretical analysis

Hayami (1985) proposed the theory of induced technological change, which suggests that when there is a change in factor scarcity within agricultural production, leading to a change in relative prices, new innovations will be triggered to optimize the use of the relatively scarce factors. Currently, due to the significant loss of rural labor, labor has become a scarce resource, prompting the search for alternative factors to replace labor (Zheng and Xu, 2017). In the current agricultural production system, large amounts of chemical fertilizers and pesticides are used to replace the labor force, but this has resulted in low efficiency and environmental pollution. Agricultural technology services have emerged as a new alternative, providing substitution and technological effects, and effectively replacing labor (Zhang and Yan, 2021).
The definition of agricultural technology service can vary among scholars. Some believe that it refers to the main body providing technology, information, and support to farmers engaged in agricultural production. Others view it as the agricultural extension department providing agricultural machinery, technical knowledge, and scientific management knowledge to address the issues faced by farmers during the production process. In this study, agricultural technology service is considered to be the main body providing farmers with knowledge on agricultural production, advanced production technology, equipment, and management concepts to address their production-related challenges. These services may include soil testing and fertilization, seed services, water-saving irrigation, clean energy utilization, information on agricultural policies, and agricultural product processing.

3.1 Agricultural technology services and the efficiency of green food production

The role of agricultural technology services is to disseminate and provide farmers with current and advanced agricultural production technologies, thereby expanding their learning opportunities and enhancing their production capacity. The theoretical basis for the improvement of green food production efficiency through agricultural technology services is derived from Terblanché’s theory (Terblanché, 2008), which suggests that these services can bridge the gap between farmers’ technological and managerial knowledge. In reality, a disparity exists between farmers’ optimal production processes and their actual production processes. Agricultural technology services play a crucial role in transmitting information to farmers, promoting the adoption of new technologies, and educating farmers to enhance their agricultural production efficiency (Terblanché, 2008).
Agricultural technology services contribute to the improvement of green production efficiency by “educating” farmers in two ways. Firstly, they assist farmers in acquiring knowledge and skills related to the current advanced production technologies. The adoption of these new technologies leads to progress in agricultural production and stimulates the growth of the agricultural sector (Cunguara and Darnhofer, 2011). Secondly, agricultural technology services help conserve resources and mitigate the need for excessive labor inputs. By optimizing the allocation of production factors, farmers can avoid the inefficient resource allocation caused by labor shortages and other challenges (Zhan, 2013).
In summary, agricultural technology services play a vital role in promoting the dissemination of agricultural technologies, educating farmers, and improving agricultural productivity. These services contribute to the growth of green food production efficiency by facilitating the adoption of advanced technologies, conserving resources, and optimizing production inputs. This situation leads to research hypothesis H1.
H1: Agricultural technology services have a significant effect on the efficiency of green food production.

3.2 Farmers’ resource allocation, agricultural technology services and green food production efficiency

The benefits generated by agricultural technology services are not equally distributed among all farmers. Research by Tong et al. (2018) supports this notion, highlighting that some farmers may benefit more from these services while others may benefit less. The variation in benefits can be attributed to differences in farmers’ resource allocation and the specific characteristics of the agricultural technologies they receive.
Even if two farmers receive the same agricultural technology services, the benefits derived from these services can differ due to variations in resource allocation and individual circumstances. Therefore, evaluating the impact of agricultural technology services on food green production efficiency from the perspective of resource allocation in individual farm households provides a more comprehensive understanding of the role of these services in improving production efficiency.
Differences in resource allocation among farm households primarily stem from two factors: the size of farmland and the number of agricultural laborers. Agricultural land is a crucial input factor for food production (Yuan et al., 2023). According to the theory of rational economic behavior, larger farmers, who possess more land, are more motivated to pursue agricultural technology updates and optimize their production inputs. On the other hand, labor availability also influences the demand for agricultural technology. Farmers with insufficient labor resources can benefit from agricultural technology services, as these services can help address labor shortages and enhance productivity.
In light of these considerations, hypothesis H2 can be proposed to examine the impact of agricultural technology services on food green production efficiency, taking into account the different resource allocations of individual farm households. This approach provides a more comprehensive assessment of the role of agricultural technology services in improving production efficiency. Currently, hypothesis H2 is proposed as follows.
H2: There is heterogeneity in the effect of agricultural technology services on the efficiency improvement of green food production, since the effect is influenced by differences in resource allocation among farmers.

3.3 Agricultural technology services, green production behavior and food green production efficiency

What is the approach of agricultural technology services in improving the efficiency of green food production? The production of green grain requires a high level of cultural and environmental literacy. However, grain growers currently have a low cultural level and face a labor shortage, resulting in the misuse of fertilizers, pesticides, and other practices (Zhang et al., 2023a). Agricultural technology service organizations possess specialized equipment, skilled personnel, and green production materials, enabling them to provide farmers with cost-effective green production materials and technologies, thereby alleviating the labor and technology shortages in the production process (Zhang et al., 2023b). Agricultural technology services assist farmers in avoiding the excessive investments in fertilizers and pesticides caused by labor and technology shortages, thereby achieving reduced and efficient utilization of these inputs, lowering production costs, and promoting green production practices among the farmers (Du et al., 2021). Currently, hypothesis H3 is proposed as follows.
H3: Agricultural technology services affect the efficiency of green food production by influencing the green production behavior of farmers, and thus the efficiency of green food production.

4 Research data and methodology

4.1 Modeling

The design of the research process in this study considers two important points. Firstly, it focuses on determining the green production efficiency of food. Currently, the two commonly used methods for determining green production efficiency are the stochastic frontier production function method (SFA) and data envelopment analysis (DEA). The SFA method involves setting a specific function form to estimate production efficiency, while DEA does not require a specific function form and can decompose efficiency into pure technical efficiency and scale efficiency. Both methods have their advantages, but traditional DEA does not consider the impact of external environmental differences on production efficiency, which may introduce bias. In contrast, the three-stage DEA model can account for external environmental factors and provide a more accurate measurement of the actual green production efficiency. Therefore, this study selected the three-stage DEA model to measure the green production efficiency of grain.
Secondly, this study addresses the issue of self-selection bias in the sample of agricultural technology services, which can lead to endogeneity problems. To overcome this bias and obtain more reliable causal inference results, this study employed a more advanced machine learning algorithm for causal identification and compared it with traditional measurement methods. This approach ensures that the causal relationships between agricultural technology services and green production efficiency are accurately assessed.
By considering these points in the research process, this study provides a robust analysis of the impact of agricultural technology services on green production efficiency.

4.1.1 Three-stage DEA model

The three-stage DEA model is able to remove environmental and stochastic perturbations and more accurately measure productivity efficiency (Dong and Mu, 2019). The process is as follows:
Stage 1: Traditional DEA modeling.
Stage 2: SFA modeling. SFA regression was used to exclude the effects of external environmental factors and random disturbances.
Stage 3: Adjusted DEA model. The inputs and outputs adjusted by the second stage were used to measure the green production efficiency of food again using the DEA model.

4.1.2 Generalized random forest model

The random forest (RF) model, proposed by Breiman, is an integrated machine learning algorithm that consists of multiple decision tree models. Compared to individual decision trees, Random Forest combines the predictions of many decision trees to improve the robustness of the fitting effect and prediction results (Breiman, 2001; Mullainathan and Spiess, 2017). It has been widely used due to its nonparametric advantage, which allows it to capture the nonlinear relationships in data and effectively handle multiple covariates (Athey et al., 2019).
However, one limitation of Random Forest is its lack of explanatory power in causality identification. This is because Random Forest operates as a black-box model, so it is unable to directly present the average treatment effect. To address this limitation, the Generalized random forest (GRF) model was developed as an extension of Random Forest. GRF calculates the conditional mean treatment effect of each individual and then regresses it using the traditional Random Forest model, which enables the estimation of the conditional mean treatment effect of heterogeneity (Friedman, 2001; Athey et al., 2019; Chen, 2021).
By incorporating the GRF model, researchers can obtain more interpretable results and better understand the causal relationships between variables. These advantages make GRF a valuable tool for conducting causal inference analysis in machine learning research.

4.2 Data and descriptive statistics

The data for this study were collected from the China Land Economy Survey (CLES) conducted between 2020 and 2022. The survey utilized the Probability Proportional to Size (PPS) sampling method to select a representative sample from Jiangsu Province in China.
The sampling process involved selecting 26 sample districts and counties from 13 prefecture-level cities in Jiangsu Province. From each selected district or county, two sample townships were chosen. Then, one administrative village was selected from each township. Finally, 50 farm households were randomly selected from each sample village.
The questionnaire used in the survey collected data at two levels: farming households and villages. This study specifically focused on examining the impact of agricultural technology services on the efficiency of green food production. To ensure data quality, samples where the largest plot was planted with non-food crops or had missing values were excluded.
In total, the survey yielded 2264 samples for analysis. Among these samples, 1320 did not have access to agricultural technology services, while 944 samples had access to such services. This dataset allowed us to investigate the relationship between agricultural technology services and the efficiency of green food production in the context of Jiangsu Province.

4.2.1 Three-stage DEA inputs and outputs

The evaluation of food production efficiency in this study was conducted by considering both input and output factors. The construction of the food production efficiency evaluation index system primarily consists of two aspects: input indicators and output indicators. The selection of these indicators aimed to accurately reflect the green production efficiency of food while excluding environmental impact factors.
The input indicators focus on land inputs, labor inputs, and capital inputs. Land inputs were measured by the area of land used for food production. Labor inputs were represented by the number of self-employed farmers involved in the production process. Capital inputs included expenses such as pesticide fees, seed fees, and machinery operating costs.
On the other hand, the output indicators can be divided into desired outputs and non-desired outputs. Desired outputs were primarily measured by the total rice production of the plot, which represents the main goal of food production. Non-desired outputs were measured by carbon emissions, which have a negative impact on the environment. The coefficient values for the carbon emission indicators were derived from the studies conducted by Gai et al. (2022) and Li et al. (2011).
It is important to note that grain green production efficiency is not solely influenced by the input of production factors. Other environmental factors, which cannot be easily changed by human intervention, also affect the efficiency of green food production. These factors include the disaster situation, soil condition, and fertility status of the plot. Therefore, this study included these environmental variables in the evaluation. The descriptions of the indicators for each variable can be found in Table 1.
Table 1 Inputs and outputs of green production efficiency in food production
Variable Variable name Description of variable Average value Standard
deviation		 Minimum value Maximum value
Input indicators Land input Area of the plot (ha) 0.62 3.47 0.00 93.33
Labor input Number of labor invested in the plot (working hours) 40.82 73.45 0.00 720.00
Capital investment Number of land investment funds (yuan) 31137.50 338131.20 0.00 10403319.50
Environmental variables Situation of disasters Number of disasters (times) 0.60 1.09 0 4
Soil type 1=sandy; 2=loam; 3=clay 2.44 0.89 1 3
Fertility (of soil) 1=poor; 2=moderate; 3=good 2.43 0.62 0 3
Output indicators Rice output Rice production (kg) 5285.61 30539.86 0.50 840000.00
Carbon footprint Carbon emissions (t) 562.58 5630.13 0.16 200968.95
By considering both input and output indicators, as well as environmental variables, this evaluation framework provides a comprehensive understanding of the efficiency of green food production, taking into account the factors that influence it.

4.2.2 Descriptive statistical analysis of variables

In this study, the explained variable is agricultural green production efficiency, and it was calculated using the three-stage DEA model based on the input-output theory. The results presented in Table 2 indicate that there is still the potential for a 22% improvement in grain green production efficiency in Jiangsu Province. Furthermore, the grain green production efficiency of the treatment group, which consisted of farmers receiving agricultural technology services, was significantly higher than that of the control group. This suggests a positive correlation between receiving agricultural technology services and achieving higher green production efficiency.
Table 2 Descriptive statistics of the variables
Variable Description of the variable All Control
group		 Treatment
group		 Difference
(Control-treatment)
Sample size (statistics) 2264 1320 944
Explanatory
variable		 Efficiency of green
food production		 Measured by three-stage DEA 0.78 0.77 0.81 -0.04***
Control
variable		 Stability of
land rights		 Unspecified duration of land operation = 1
Land operation for a definite period of less than 5 years = 2
Land operation for a definite period of 5 to 10 years = 3
Land operation with a definite duration > 10 years = 4
Contracted land = 5		 4.47 4.56 4.33 0.23***
Planting subsidies Number of subsidies received for planting (yuan) 5474.06 1981.95 10000 -8400***
Distinguishing
between the sexes		 0 = female; 1 = male 0.86 0.84 0.89 -0.05***
(a person’s) age Actual age (years) 60.92 62.12 59.26 2.86***
Cultures Educational attainment 2.72 2.57 2.92 -0.35***
Health status Health status from low to high 1 to 5 4.04 3.92 4.20 -0.28***
Geographical position Distance to county, hospital, bank; entropy value gained, value range: 0-1 0.77 0.77 0.76 0.01
Economics Whether it is an economically weak village, 0=No; 1=Yes 0.14 0.16 0.11 0.04***
Transportation Distance to high speed rail, highways, paved roads; Entropy
value obtained		 0.32 0.33 0.32 0.01*
Proportion of
agricultural labor force		 Household participation in agricultural labor/Family labor force 0.55 0.54 0.58 -0.04***
Scale of household
farmland operations		 1 = Operating area of 5 acres or less
2 = Operating area of 5 acres to 50 acres
3 = Operating area of 50 acres or more		 1.54 1.40 1.73 -0.34***
Mechanism
variable		 Agricultural film
recycling		 Whether they recycle agricultural films, 0=No; 1=Yes 0.04 0.03 0.04 -0.01
Pesticide use Use of low-toxicity pesticides, 0=No; 1=Yes 0.67 0.65 0.75 -0.10***
Fertilizer use Whether they use organic fertilizers, formulated fertilizers,
0=No; 1=Yes		 0.05 0.03 0.07 -0.04***
Type of
technical
service		 Prenatal technical
services		 Access to network information services, agricultural policy
information services, capital credit and other services are
assigned a value of 1, while the rest are assigned a value of 0		 0.44 - - -
Mid-production
technical services		 Access to good seeds, soil testing and fertilizer application
services, crop cultivation management services, pest control
technology, water-saving irrigation technology, etc., are
assigned a value of 1, while the rest are assigned a value of 0		 0.30 - - -
Pre-production
technical services		 Access to crop straw utilization technology, agricultural market
information services, agricultural product processing
insurance technology and other services are assigned a value
of 1, while the rest are assigned a value of 0		 0.03 - - -

Note: *, **, *** denote coefficients significant at the 10%, 5%, and 1% levels, respectively. The same below.

The core explanatory variable in this study is whether farmers receive agricultural technology services. Agricultural technology services play a crucial role in promoting agricultural technology adoption and can impact the efficiency of green food production. Agricultural technology services encompass various services provided by organizations to farmers, including soil testing and fertilization, seed selection, policy information dissemination, and agricultural product processing techniques.
The treatment group in this study comprised farmers who had received agricultural technology services, while the control group consisted of farmers who had not received such services. According to Table 2, the proportion of samples in the treatment group was 41.82% of the total samples, indicating that more than half of the farmers in Jiangsu Province do not have access to agricultural technology services. This high proportion highlights the need for further improvement in the promotion and accessibility of agricultural technology services. Efforts should be made to ensure that more farmers can benefit from these services, thereby expanding the reach and impact of agricultural technology services at the grassroots level.
This study also considered several control variables based on relevant literature, including land rights stability, gender, age, culture, health status, village location, economy, and transportation (Qian et al., 2022). The grouping t-test conducted on these control variables, as shown in Table 2, revealed significant differences in the mean values between the treatment group and the control group. This suggests that there may have been some self-selection issues in the sample selection process for farmers receiving agricultural technology services.
To examine the mechanism through which agricultural technology services influence green production efficiency, this study selected green production behavior as the mechanism variable. This study focused on three aspects of the farmers’ green production behavior: agricultural film recycling, use of low-toxicity pesticides, and selection of organic fertilizers. These three variables were chosen as the mechanism variables to be tested.
Individual differences in various characteristics were found to also play a role in determining the impact of agricultural technology services on green production efficiency. This study found that household resource allocation, particularly the scale of land management and the proportion of agricultural laborers in the household, are important characteristics that influence the effectiveness of agricultural technology services.

5 Impact of agricultural technology services on the efficiency of green food production

5.1 Basic regression model

This study employed the truncated regression model (TRM) as the baseline regression model, considering that food green production efficiency values range between 0 and 1. It initially used traditional linear regression models, including OLS and Tobit regression models, as well as a PSM (Propensity Score Matching) model for the causal identification of the impact of agricultural technology services on food green production efficiency. The regression results are presented in Table 3.
Table 3 Basic regression results
Variable name OLS TRM OLS+FE TRM+FE FE+PSM
(1) (2) (3) (4) (5)
Agricultural technical services 0.041*** 0.041*** 0.043*** 0.043*** 0.042***
(0.013) (0.013) (0.012) (0.012) (0.012)
Stability of land rights 0.052*** 0.052*** 0.052*** 0.052*** 0.052***
(0.005) (0.006) (0.004) (0.005) (0.004)
Planting subsidies 0.000** 0.000** 0.000*** 0.000*** 0.000***
(0.000) (0.000) (0.000) (0.000) (0.000)
Distinguishing between the sexes 0.011 0.011 0.015 0.015 0.012
(0.018) (0.018) (0.018) (0.018) (0.018)
Age -0.004*** -0.004*** -0.003*** -0.003*** -0.003***
(0.001) (0.001) (0.001) (0.001) (0.001)
Cultures -0.018** -0.018** -0.017** -0.017** -0.018**
(0.007) (0.007) (0.008) (0.007) (0.008)
Health status -0.014** -0.014** -0.011* -0.011* -0.012**
(0.006) (0.006) (0.006) (0.006) (0.006)
Geographical position -0.091*** -0.091*** -0.111*** -0.111*** -0.113***
(0.031) (0.031) (0.029) (0.031) (0.029)
Economics -0.159*** -0.159*** -0.078*** -0.078*** -0.077***
(0.017) (0.017) (0.026) (0.020) (0.026)
Transportation -0.082** -0.082** 0.062 0.062 0.059
(0.040) (0.040) (0.072) (0.051) (0.072)
Year fixed effects No No Yes Yes Yes
Individual fixed effect No No Yes Yes Yes
Constant 0.962*** 0.962*** 0.916*** 0.881*** 0.919***
(0.067) (0.067) (0.071) (0.067) (0.071)
N 2264 2264 2264 2264 2255

Note: The standard errors given in parentheses.

In Table 3, columns (1)‒(2) show the findings of the OLS and TRM regressions, indicating a significant positive effect of agricultural technology services on food green production efficiency. Columns (3)‒(4) in Table 3 demonstrate that agricultural technology services continued to have a significant effect on food green production efficiency even after incorporating fixed effects.
However, the descriptive statistics in Table 2 reveal significant differences in the mean values of each variable between the treatment group and the control group. Additionally, agricultural technology services face self-selection issues. To address these concerns and reduce the imbalance between the treatment and control groups, this study calculated propensity scores and employed kernel matching to select the samples and identify the causal factors. The results after matching passed the balance test. The regression results after propensity score matching are presented in column (5) of Table 3, and they indicate a significant positive effect of agricultural technology services on the efficiency of green food production.

5.2 Generalized random forest

The OLS and TRM models, as well as propensity score matching, are known to have limitations in addressing the issue of two-way causality. Additionally, the instrumental variable method, while capable of causal identification, relies on the assumption of exogeneity of instrumental variables, which may involve subjective judgment. To overcome these challenges, this study employed the generalized random forest for empirical estimation and causal identification.
In the generalized random forest identification, control variables are used initially, and variables with an importance degree less than 0.05 are subsequently excluded to improve the model's accuracy by eliminating the weaker explanatory variables. The estimation using generalized random forest is then conducted. Columns (1) to (4) in Table 4 indicate that as the number of decision trees increases, the average treatment effect of agricultural technology services on green food production efficiency remains around 0.031, with minimal changes in the standard deviation. This result aligns closely with the numerical estimation of traditional econometric models, demonstrating consistent findings with no significant differences.
Table 4 Generalized random forest estimation results
Variable name Efficiency of green food production
(1) (2) (3) (4)
Agricultural technical services 0.030*** (0.013) 0.030** (0.013) 0.030** (0.013) 0.030** (0.013)
Number of trees 500 1000 1500 2000
Mould Causal forest Causal forest Causal forest Causal forest
Clustering Yes Yes Yes Yes
Observed value 2264 2264 2264 2264

Note: The standard errors given in parentheses.

The data in Table 4 further illustrate that agricultural technology services have a significant and positive impact on the efficiency of green food production. This can be attributed to several possible reasons. Firstly, agricultural technology services enhance farmers’ agricultural technology level, thereby promoting improved management capabilities and the adoption of advanced agricultural production techniques. This facilitates the rational allocation of agricultural resources, preventing wastage and pollution resulting from inadequate mastery of agricultural technology. Consequently, this enhances green food production efficiency. Secondly, agricultural technology services drive progress in agricultural technology, thereby reducing farmers' labor demand and alleviating labor shortages. This helps to avoid efficiency losses caused by labor shortages and ultimately improves the efficiency of green food production.
In summary, the benefits of agricultural technology services primarily stem from resource saving and resource substitution, thus supporting hypothesis H1.
This study examined the impacts of different types ofagricultural technology services on food green production efficiency, specifically focusing on three stages: pre-production, mid-production, and post-production. Table 5 presents the findings regarding the influences of these services on food green production efficiency.
Table 5 Split-sample estimation results
Variable Efficiency of green food production
Pre-production Mid-production Post-production Pre-production×
Mid-production		 Pre-production×
Post-production		 Mid-production× Post-production Pre-production×
Mid-production× Post-production
Agricultural
technical services		 0.027**
(0.016)		 0.027*
(0.014)		 0.043
(0.031)		 0.030*
(0.016)		 0.069*
(0.036)		 0.092*
(0.050)		 0.125**
(0.056)
Number of trees 2000 2000 2000 2000 2000 2000 2000
Clustering Yes Yes Yes Yes Yes Yes Yes
Observed value 2264 2264 2264 2264 2264 2264 2264

Note: The standard errors given in parentheses.

The results indicate that both pre-production and mid- production agricultural technology services have positive and significant effects on food green production efficiency. The coefficients for these two stages are consistent, suggesting a consistent impact. On the other hand, the coefficient for post-production technical services is positive but not statistically significant.
Post-production agricultural technology services primarily affect food green production efficiency through two aspects. Firstly, technologies related to agricultural product processing and marketing, such as preservation techniques, play a crucial role in maintaining the value of agricultural products. These technologies can impact farmers’ income, thereby influencing the efficiency of food production. Secondly, the treatment of agricultural waste, particularly the management of straw, is essential for reducing carbon emissions in agriculture. The handling of straw can have implications for the efficiency of green food production.
However, this study found that the impact of post- production agricultural technology services on food green production efficiency is not significant, which may be attributed to several factors. One possibility is that agricultural technical service organizations may neglect the provision of post-production services. The descriptive statistics mentioned earlier indicated that only 3% of the samples received post-production agricultural technology services. The limited availability of these services or their poor quality could hinder the potential benefits of post-production agricultural technology services.
In summary, these results highlight the positive and significant impacts of pre-production and mid-production agricultural technology services on food green production efficiency. However, the significance of post-production services was not supported by the data, potentially due to the limited provision or quality of these services.
This study also examined the potential benefits of combined agricultural technology services compared to individual services. The results suggest that combined agricultural technology services may generate more spillover effects and yield higher returns. By cross-multiplying the different stages of agricultural technology services, this study explored the impacts of these combinations on green food production efficiency.
The findings reveal that all the cross-multipliers significantly increased the efficiency of green food production. Moreover, the coefficients for these cross-multipliers are higher than those for the individual agricultural technology services. This indicates that the combination of agricultural technology services has a greater impact on improving green food production efficiency compared to the individual services.
In particular, the combination of all three agricultural technology services has the highest impact on green food production efficiency, as its coefficient surpasses those of the individual agricultural technology services and the other two two-way combinations. This finding suggests that the combination of agricultural technology services can generate spillover effects and enhance the overall effectiveness of these services.
In summary, these results highlight the potential advantages of combined agricultural technology services in improving green food production efficiency. The cross-multipliers of these combinations contribute significantly to the efficiency gains, with the combination of all three services showing the highest impact.

5.3 Resource allocation heterogeneity

5.3.1 Heterogeneity test

Figure 1 illustrates the distribution of treatment effect estimates for agricultural technology services on green food production efficiency. The majority of the samples show an average treatment effect, and the values are concentrated in the range of 0 to 0.05. However, the overall treatment effect estimates for the entire sample range from -0.05 to 0.25.
Fig. 1 Distribution of the treatment effects of agricultural technology services on green food production efficiency
This range indicates that there are significant variations in the returns to agricultural technology services among different individuals. To further investigate this heterogeneity, this study conducted a heterogeneity test and analyzed the heterogeneity of the treatment effects. These analyses aimed to uncover the specific impacts of agricultural technology services on green food production efficiency, taking into account the diverse characteristics and circumstances of different individuals.
By examining the heterogeneity of the treatment effects, this study aims to provide a more comprehensive understanding of how agricultural technology services influence green food production efficiency. This analysis considered the variations in treatment effects among different individuals to shed light on the factors that contribute to the differences in returns from agricultural technology services.
Table 6 presents the coefficients of the between-group differences, which are all significantly positive. This indicates that there is substantial heterogeneity in the impact of agricultural technology services on the efficiency of green food production. To further investigate this heterogeneity, this study conducted an analysis based on the different characteristics of the samples.
Table 6 Heterogeneity test
Variable Ratio Standard error
Average value 1.060*** 0.563
Difference between groups 0.700*** 0.425
Model GRF
Observed value 2264
Unlike traditional econometric models that use the group over interaction form to test for differences in average treatment effects, this study employed a generalized random forest approach. The generalized random forest calculates an estimate of the average treatment effect for each sample, thereby making the estimation of heterogeneity more accurate.
The study analyzed the input resources of farmers in food production, specifically focusing on the proportion of the agricultural labor force and the scale of agricultural operations. By exploring the differences in resource inputs, this study aimed to understand how these factors impact the efficiency of green food production upon receiving agricultural technology services.
By examining the relationship between resource inputs and the efficiency of green food production, this study provides insights into the extent of the impact of agricultural technology services on different groups. This analysis helps to identify the specific characteristics and circumstances that influence the effectiveness of agricultural technology services in improving green food production efficiency.

5.3.2 Heterogeneity in resource allocation

The differences in resource allocation among farm households have an impact on the returns from receiving agricultural technology services. Farm households that allocate more resources to agriculture are more likely to actively seek out and benefit from agricultural technology services, leading to improved efficiency in green food production.
To analyze this heterogeneity, this study focused on two factors: the number of agricultural laborers in the household and the size of the family arable land operation. Figure 2 shows the heterogeneity of agricultural technology service reporting between two types of resources.
Fig. 2 Treatment effects of the proportion of working population in agriculture and the size of household farmland operations
The results indicate that as the number of family agricultural laborers increases, the returns from agricultural technology services in terms of green food production efficiency decrease. This may be attributed to the labor-intensive nature of food cultivation technology, where labor inputs have a significant impact on efficiency. In other words, as the number of agricultural laborers in the household increases, the marginal effect of agricultural technology services gradually diminishes. In contrast, when the family agricultural labor force is small, farmers tend to increase other inputs due to labor shortages. Agricultural technology services provide access to advanced production technology, helping farmers to avoid excessive reliance on material inputs. Therefore, agricultural technology services have a greater marginal effect on improving green food production efficiency for households with a smaller agricultural labor force.
The results also indicate that the scale of the family farmland operation affects the returns from agricultural technology services. Small-scale farmers benefit more from agricultural technology services compared to large-scale farmers. This is because small-scale farmers face greater challenges in accessing the latest agricultural technology, and agricultural technology services serve as an important means for obtaining such technology. On the other hand, large-scale farmers can leverage their scale advantage to acquire agricultural technology from various sources, making agricultural technology services just one of many sources. Consequently, the returns from agricultural technology services are higher for small-scale farmers, and the marginal effect of these services decreases as the scale of the family farmland operation expands.
In conclusion, the results verify that the returns from agricultural technology services vary among farmers due to differences in resource allocation. This finding supports hypothesis H2, which suggests that the impact of agricultural technology services differs among farmers.

5.4 Further discussion: Analysis of mechanisms

Agricultural technology promotion is an important tool for realizing China's food security, but also an important aid for the modernization of food production. The above empirical study confirms that agricultural technology services play a positive role in promoting the efficiency of the green production of food, but it does not explain the way by which agricultural technology services promote it, or what the intrinsic role of agricultural technology services is in the efficiency of green production of food paths. According to the theoretical analysis, the agricultural technology service is an important channel for helping farmers to learn new technology, and farmers can acquire advanced green production technology through agricultural technology services and then apply the green production technology to their food production. Therefore, agricultural technology services need to be really applied to food production, so this study tested whether agricultural technology services affect food green production efficiency by influencing farmers’ green production behavior through the analysis of that behavior.
Panel A of Table 7 shows the results of an initial investigation of the impacts of agricultural film recycling, pesticide use, and chemical fertilizer use on the green production efficiency of food. The data indicate that all three practices have significant positive effects on the green production efficiency of food. Panel B in Table 7 further investigates the influence of agricultural technology services on the mechanism variables to explore how these services enhance the green production efficiency of food. The results reveal that agricultural technology services do not have a significant inhibitory effect on agricultural film recycling, but they do significantly promote the use of low-toxic pesticides and organic fertilizers.
Table 7 Mechanism analysis
Impact of Panel A mechanism variables on green food production efficiency
Variable name Efficiency of green food production
Agricultural film recycling 0.108*** (0.032)
Pesticide use 0.271*** (0.012)
Fertilizer use 0.079*** (0.028)
Control variable Controlled Controlled Controlled
Observed value 2264 2264 2264
Impact of Panel B mechanism variables on the efficiency of green food production
Variable name Agricultural film recycling Pesticide use Fertilizer use
Agricultural technical services -0.001 (0.007) 0.053** (0.020) 0.027** (0.011)
Tree volume 2000 2000 2000
Clustering Yes Yes Yes
Observed value 2264 2264 2264

Note: The standard errors given in parentheses.

This difference may be due to two main reasons. Firstly, most agricultural films are made of plastic. If farmers do not recycle the agricultural films, the films cannot degrade in the soil, which directly affects the cultivation in the next season. The presence of residual agricultural films can also reduce soil aeration and water permeability, thereby impacting crop growth. Farmers have firsthand experience with the harm caused by agricultural films, so even without agricultural technical services, they would still take the appropriate measures to properly dispose of the residual agricultural films.
Secondly, agricultural technology services encourage farmers to invest their resources rationally. In the face of labor shortages and natural disasters, farmers without sufficient agricultural technology are more likely to invest more resources to ensure production. Pesticides and fertilizers are the two most convenient means and resources for farmers. According to the theory of induced technological change, various factors can be substituted for each other. Agricultural technology services can help farmers obtain more advanced agricultural technology, so they would replace pesticides, fertilizers, and other factors to help reduce the losses caused by natural disasters.

6 Discussion

This study makes several contributions to the measurement of green production efficiency in grain. Firstly, it utilizes micro-subjective data, which provides a more realistic perspective compared to the national and inter-provincial macro-data that are commonly used in the current studies. By using micro-subjective data, this study can provide more practical recommendations for improving green production efficiency in grain.
Secondly, this study employs the third-stage DEA model to calculate green production efficiency, which takes into account different farmland qualities and excludes the influence of environmental factors. This approach is different from the existing plot-based calculation method and allows for a more accurate assessment of green production efficiency in grain. This study found an average green production efficiency of 0.78.
Furthermore, this study explores the relationship between agricultural technical services and green production efficiency in grain using the generalized random forest model. By applying machine learning methods to the study of green production efficiency, the research field is expanded, and the quality of agricultural services and green production efficiency in grain can be improved. This study also contributes to the field of grain green production efficiency research by introducing quantitative research methods and providing a practical case study for using machine learning methods in analyzing influencing factors and making policy recommendations for eco-efficiency.
Compared to the traditional econometric models and random forest models used in previous studies, the generalized random forest model utilized in this study has advantages in avoiding the bias caused by covariate selection and achieving consistent estimation of treatment effect heterogeneity.
Based on the existing research results, the following discussion will build upon these findings.
(1) Agricultural technology services enhance the efficiency of green food production.
While there are limited discussions on this topic, the existing studies support the notion that agricultural technology services can improve agricultural production efficiency. For example, Owens et al. (2003) conducted a study in Zambia and found that agricultural technology services promoted the growth of agricultural production efficiency. Tian and Wang (2016) also concluded that strengthening farmer training can improve the efficiency of green food production.
The empirical results of these studies align with the majority of research in this area, which suggests that agricultural technology services can enhance the efficiency of green food production. Additionally, this study examined the impact of agricultural technology services at different stages of production and found that the services were most effective during the mid-production stage, followed by the pre-production and post-production stages.
(2) Agricultural technology services enhance the agricultural production efficiency path.
Many current studies suggest that agricultural technology services can promote the adoption of agricultural technology by farmers. For example, Meili et al. (2022) argues that the acceptance of agricultural technology services by farmers can facilitate the adoption of new technologies. This finding is generally consistent with the results of the study mentioned above, which also found that agricultural technology services promote green production by farmers, thereby enhancing production efficiency.
Furthermore, Xiang et al. (2023) reported that agricul-tural technology services enhance agricultural production efficiency through technical training and the involvement of experts in rural areas. These findings present alternative pathways to the ones explored in the study mentioned above.
(3) Resource endowment differences affect the return on agricultural technology services.
According to current research, there are differences in the returns of agricultural technology services among different farmers. Yuan et al. (2023) suggested that due to the existence of differences in the scale of farmers’ operations, their demands for agricultural technology also vary. Tong et al. (2018) argue that the social and economic status of farmers has an impact on the promotion of agricultural technology. These differences are related to the resources possessed by the farmers. Overall, the findings of this study are consistent with the existing research conclusions, as both suggest that individual differences among farmers affect the returns from agricultural technology services.
However, this study has several limitations. Firstly, it relies on data from the China Land Economy Survey (CLES), which was conducted in Jiangsu province. Jiangsu is a prosperous province in China, known for its strong economy and significant rice production. Therefore, the generalizability of the results from this study to economically disadvantaged areas that are not major rice producers and are located in remote regions requires further investigation and confirmation.
Furthermore, the data in this study are primarily focused on rice production. However, due to variations in the production processes of rice compared to other food crops, it is necessary to conduct further investigation and verification to determine whether the conclusions of this study can be applied to other food crops.
Lastly, this study examines the influence of farmers’ heterogeneity in resource allocation on the marginal effect of agricultural technology services. However, this study did not further explore the diverse data limitations in the current main body of agricultural technology services and the content of the service.
Future studies should investigate the impacts of agricultural technology services on agricultural production efficiency, farm household income, and other aspects of agricultural technology service performance. They should also focus on the main body of agricultural technology services, considering the content of the service and the adoption of technology.

7 Conclusions

The study employed a combination of machine learning algorithms and traditional social science empirical analysis methods to identify the causal relationships and empirically examine the impact of agricultural technology services on the efficiency of food green production. The main findings are as follows.
Agricultural technology services have a significant and positive effect on food green production efficiency. Specifically, pre-production and mid-production agricultural technology services greatly enhance the efficiency of food green production, while the return on agricultural technology services in the post-production stage is not significant. The combination of agricultural technology services plays a crucial role in improving food green production efficiency.
There is heterogeneity in the marginal effect of resource allocation in agricultural households on the enhancement of food green production efficiency through agricultural technology services. An increase in the proportion of family agricultural laborers and the expansion of family-run arable land size significantly reduce the returns from agricultural technology services.
The analysis of mechanism variables indicated that agricultural technology services primarily influence food green production efficiency through pesticide use and chemical fertilizer use. These services have a significant enhancing effect on food green production efficiency.
Based on the above research several insights are derived.
The scope of agricultural technology services should be expanded since only 41.8% of the total sample had received such services. Expanding the scope of agricultural technology services will maximize their impact on enhancing the efficiency of green food production. The current supply of post-production agricultural technology services lacks sufficient coverage. Greater emphasis should be placed on the supply of post-production stage agricultural technology services to establish a comprehensive system covering the pre- production, in-production, and post-production stages. Emphasizing the supply of agricultural technology services in the post-production stage and establishing a comprehensive system covering all three stages will fully utilize the benefits of integrated agricultural technology services.
The impacts of agricultural technology services on enhancing the efficiency of green food production vary and should be customized to local conditions and the specific needs of individuals. This customization will assist in addressing the labor shortages faced by farmers, particularly small-scale farmers, and help to resolve their challenges. To address the challenge of integrating small farmers’ production with agricultural modernization, agricultural technology services should prioritize the needs of small farmers and refrain from favoring large households in their service delivery.
Providing agricultural technology services strengthens the guidance of green production concepts and promotes green production behavior of farmers. Agricultural technology services are an important agricultural technology access channel for farmers, so agricultural technology services should be strengthened in the process of green production concept output, thereby promoting the green production behavior of farmers, and enhancing the efficiency of green food production. Agricultural technology services serve as a crucial channel for farmers to access agricultural technology. Therefore, it is essential to reinforce these services in order to promote the adoption of green production concepts and behaviors among farmers, ultimately improving the efficiency of green food production.

References

Publishing order | Descend order by publishing year | Descend order by cited within
[1]
Athey S, Tibshirani J, Wager S. 2019. Generalized random forests. The Annals of Statistics, 47(2): 1148-1178.

[2]
Breiman L. 2001. Random forests. Machine Learning, 45(1): 5-32.

DOI

[3]
Chen Q. 2021. Machine learning and python applications. Beijing, China: Higher Education Press. (in Chinese)

[4]
Cui X F, Wang Y F, Zhang G H. 2022. Measurement and analysis of spatial and temporal evolution of agroecological efficiency for low-carbon development—Based on SBM-ESDA model. Issues in Agricultural Economy, (9): 47-61. (in Chinese)

[5]
Cunguara B, Darnhofer I. 2011. Assessing the impact of improved agricultural technologies on household income in rural Mozambique. Food Policy, 36(3): 378-390.

DOI

[6]
Dong Y, Mu Y Y. 2019. Empirical evidence of path selection and efficiency mechanism of environmentally friendly technology adoption by farmers. China Rural Survey, (2): 34-48. (in Chinese)

[7]
Du S X, Luo X F, Huang Y Z, et al. 2021. Risk perception, agricultural socialized services, and rice farmers’ adoption behavior of biopesticides technology. Resources and Environment in the Yangtze Basin, 30(7): 1768-1779. (in Chinese)

[8]
Feder G, Slade R. 1986. The impact of agricultural extension: The training and visit system in India. The World Bank Research Observer, 1(2): 139-161.

DOI

[9]
Friedman J H. 2001. Greedy function approximation: A gradient boosting machine. Annals of Statistics, 29(5): 1189-1232.

DOI

[10]
Gai Z X, Zhan W X, Wang H Y, et al. 2022. Characteristics of spatial and temporal variability of carbon emissions from cropland utilization transition and its formation mechanism. Transactions of the Chinese Society for Agricultural Machinery, 53(7): 187-196. (in Chinese)

[11]
Gao Y, Wang N, Li X F, et al. 2017. Decision analysis of farmers’ adoption decision of eco-friendly farmland soil management technology—A case study of Shandong Province. Agricultural Economic Issues, 38(1): 38-47, 110-111. (in Chinese)

[12]
Hayami Y, Ruttan V W. 1985. Agricultural development:An international perspective (revised and expanded). Baltimore, USA: Johns Hopkins University Press.

[13]
Hu R R, Yu C X, Jin Y H, et al. 2022. Impact of government policies on research and development (R&D) investment, innovation, and productivity: Evidence from pesticide firms in China. Agriculture, 23(4): 962-974.

[14]
Li B, Zhang J B, Li H P. 2011. Spatial and temporal characteristics of agricultural carbon emissions in China and decomposition of influencing factors. China Population, Resources and Environment, 21(8): 80-86. (in Chinese)

[15]
Liu S L, Zhang X M, Yang Z S. 2021. The impact of Internet use on rural residents’ income. Journal of Quantitative & Technological Economics, 38(4): 103-119. (in Chinese)

[16]
Meili H, Fengxia D, Liang C. 2022. Mechanization services, factor allocation, and farm efficiency: Evidence from China. Review of Development Economics, 26(3): 1618-1639.

DOI

[17]
Mullainathan S, Spiess J. 2017. Machine learning: An applied econometric approach. Journal of Economic Perspectives, 31(2): 87-106.

DOI

[18]
Owens T, Hoddinott J, Kinsey B. 2003. The impact of agricultural extension on farm production in resettlement areas of Zimbabwe. Economic Development and Cultural Change, 51(2): 337-357.

DOI

[19]
Qian L, Lu H, Gao Q, et al. 2022. Household-owned farm machinery vs. outsourced machinery services: The impact of agricultural mechanization on the land leasing behavior of relatively large-scale farmers in china. Land Use Policy, 115: 106008. DOI: 10.1016/j.landusepol.2022.106008.

[20]
Rosegrant M W, Pingali P L. 1994. Policy and technology for rice productivity growth in Asia. Journal of International Development, 6(6): 665-688.

DOI

[21]
Ruttan V W. 1986. Assistance to expand agricultural production. World Development, 14(1): 39-63.

DOI

[22]
Terblanché S E. 2008. Towards an improved agricultural extension service as a key role player in the settlement of new farmers in South Africa. South African Journal of Agricultural Extension, 37(1): 58-84.

[23]
Tian X, Wang S G. 2016. Analysis of environmental efficiency of food production and its influencing factors in China. Resource Science, 38(11): 2106-2116. (in Chinese)

[24]
Tong D J, Huang W, Ying R Y. 2018. Impacts of grassroots public agricultural extension on farmers’ technology adoption—Taking rice technology demonstration as an example. China Rural Survey, (4): 59-73. (in Chinese)

[25]
Wang Z, Deng X, Bai Y, et al. 2016. Land use structure and emission intensity at regional scale: A case study at the middle reach of the Heihe River Basin. Applied Energy, 183: 1581-1593.

DOI

[26]
Wen C C, Wu J X. 2016. Analysis of factors influencing the adoption behavior of farmers’ “two-type agriculture” technology—Based on the survey data of corn and rice farmers in Liaoning Province. Journal of China Agricultural University, 21(9): 179-187. (in Chinese)

[27]
Xia Z X, Zhang S Y, Hao C C. 2013. Evaluation of service effectiveness of agricultural extension organizations based on farmers’ perspective. Guangdong Agricultural Science, 40(4): 198-201. (in Chinese)

[28]
Xiang S, Li Y X, Jiang J Y. 2023. Hiding grain in technology: Grass-roots agricultural extension and technical efficiency of grain production—An empirical analysis based on 859 growers’ survey data in the main grain-producing areas of Anhui Province. Rural Economy, (4): 114-125. (in Chinese)

[29]
Ying R Y, Zhu Y. 2015. The impact of agricultural technology training methods on farmers’ agricultural chemical input use behavior—Evidence from experimental economics. China Rural Survey, (1): 50-58, 83, 95. (in Chinese)

[30]
Yu S, Will C. 2019. Exploring the relationship between farm size and productivity: Evidence from the Australian grains industry. Food Policy, 84: 196-204.

DOI

[31]
Yu Z J, Gong Y Z, Zheng S. 2022. Carbon neutrality in China’s agriculture and rural areas: Theoretical logic, practical path and policy orientation. Reform of Economic System, (6): 74-81. (in Chinese)

[32]
Yuan R L, Liao W M, Qiu H L. 2023. The effects of agricultural technology training and scale of operation on agricultural production efficiency—A case study of rice cultivation technology. Chinese Journal of Agricultural Resources and Regional Planning, 44(7): 216-226. (in Chinese)

[33]
Zhan J T. 2013. Differential study of technical extension services, factor inputs and farmers’ rice output effects—Analysis based on Quantile regression. Journal of Nanjing Agricultural University (Social Science Edition), 13(3): 40-46. (in Chinese)

[34]
Zhang M L, Chen Z J, Weng Z L, et al. 2023a. Study on the impact of agricultural socialization services on the reduction of chemical fertilizer application—Analysis of the regulatory effect based on factor allocation. Journal of Agrotechnical Economics, (3): 104-123. (in Chinese)

[35]
Zhang X, Yan T W. 2021. Analysis of the impact of agricultural technology services on farmers’ straw return behavior in the context of labor transfer—Taking Hubei Province as an example. Journal of China Agricultural University, 26(1): 196-207. (in Chinese)

[36]
Zhang Y N, Yin Y S, Zhang K J, et al. 2023b. Can agricultural socialized services promote the green production transformation of wheat growers—Based on evidence from household surveys in Henan, Shandong, and Shanxi. China Population, Resources and Environment, 33(6): 172-181. (in Chinese)

[37]
Zheng X Y, Xu Z G. 2017. Resource endowment constraints, factor substitution and induced technological change—The mechanization of grain production in China as an example. Economics (Quarterly), 16(1): 45-66. (in Chinese)

Options
Outlines

/

〈 〉