Do Agricultural Productive Services Impact the Carbon Emissions of the Planting Industry in China: Promotion or Inhibition?

Abstract

Reducing carbon emissions from planting (PCE) is crucial for achieving the “dual carbon” objective within the agricultural sector. This paper aims to investigate the impact of agricultural productive services (APSs) on carbon emissions in the planting industry, providing novel insights into reducing carbon emissions in this field. Based on the panel data of 30 provinces and regions in China from 2005 to 2021, this study measured the PCE and the level of APSs. The internal relationship between the APSs and PCE is subsequently examined empirically using various statistical models, including the fixed effect model, instrumental variable model, mediating effect model, and threshold effect model. The findings suggest that the PCE experienced an initial increase, followed by a subsequent decrease between 2005 and 2021, with the peak observed in 2015. APSs has a significant inhibitory effect on PCE, which remains significant even after conducting various robustness tests. However, the carbon emission reduction effect of APSs varies across different crop-producing areas. The study also revealed that APSs can inhibit PCE by facilitating land transfer. Additionally, the inhibitory effect of APSs on PCE is influenced by the scale of land management and exhibits a U-shaped nonlinear relationship. To effectively reduce PCE and achieve sustainable agricultural development, policymakers should actively promote the integration of APSs with modern agriculture and form moderate-scale operations by facilitating land transfer to maximize APSs’ carbon emission reduction effect.

1. Introduction

The issue of global warming is becoming increasingly severe. As the primary driver of climate change, greenhouse gas emissions resulting from human activities have attracted widespread attention across various sectors worldwide [1]. Being a traditional agricultural country, China ranks among the highest in the world in terms of grain output and planting area. The acceleration of China’s agricultural modernization has led to the emergence of new trends in agriculture; however, it has also resulted in significant environmental problems. Studies have indicated that agricultural activities contribute to 17% of the total carbon emissions in China [2]. Therefore, achieving agricultural carbon emission reduction holds greater practical significance for China compared to other countries [3]. Additionally, as a narrow section of agriculture, the planting industry holds an extremely special and important position within the entire agricultural sector. It is one of the most fundamental and essential components of agricultural production, and its carbon emissions should not be underestimated [4]. The inefficient utilization of resources and high environmental costs during the production process in China’s planting industry have led to a 180% increase in fertilizer consumption and a 103% increase in greenhouse gas emissions over the past 30 years [5,6]. Consequently, reducing agricultural carbon emissions still faces significant challenges in China.

Given the fundamental national conditions of “small farmers in a large country,” small farmers will continue to be the mainstay of China’s agricultural operations for a long time in the future [7], and whether their production methods can be integrated with modern agriculture will directly impact the progress of low-carbon agricultural development [8]. Moreover, given the significant population of approximately 230 million small-scale farmers in China, some scholars have begun to question the sustainability of Chinese agriculture [9]. To address this question, the Chinese government has actively advocated and implemented the “development of new agricultural socialized services” and “promotion of deep integration between the modern service industry and modern agriculture” to provide new opportunities for the organic connection between small farmers and the development of modern agriculture. Indeed, the realization of agricultural servitization is undoubtedly a crucial aspect of China’s agricultural modernization transformation [10]. On one hand, it can facilitate the transition of small farmers’ production mode from independent to entrusted service operation, thereby fostering a new approach that enhances the connection between small farmers and modern agricultural development [11]. On the other hand, it has the potential to significantly enhance agricultural resource utilization efficiency and reduce environmental costs associated with agricultural production, thus presenting new opportunities for promoting sustainable agriculture [12]. Can the development of APSs provide a new driving force for reducing PCE? The interpretation and exploration of this issue have significant practical implications in promoting the deep integration of the modern service industry and modern agriculture, while also offering new ideas for reducing PCE.

Currently, scholars primarily focus on the construction and calculation of carbon emission index systems for the planting industry [13,14,15], carbon emission efficiency [16,17,18], accounting and characteristics of carbon footprints [19], as well as other related aspects. Additionally, attention has been given to exploring the impact of factors such as operational scale [20,21], land transfer [22], planting structure [23], agricultural technology [24,25], and agricultural mechanization level [26,27] on PCE. On the other hand, scholars have fully acknowledged the economic and ecological effects of APSs and contend that it can not only improve the welfare of smallholder farmers [28,29] and promote farmers’ income growth [30,31], but also facilitate land transfer [32,33], adjust agricultural production modes [34], and enhance agricultural production efficiency [35,36]. Moreover, it holds significant implications for stimulating farmland protection behavior among farmers [37,38] and reducing chemical input [39,40,41]. However, despite extensive research conducted by numerous scholars on PCE and the value function of APSs from various perspectives, there remains a lack of studies directly elucidating the relationship between APSs and PCE. Further research is still required to investigate the impact and mechanism of APSs on PCE.

In light of this, the research question of this study is to investigate the impact of agricultural productive services (APSs) on carbon emission reduction within the planting industry, aiming to provide innovative insights for mitigating carbon emissions in this sector. Based on panel data from 2005 to 2021 encompassing 30 provinces and regions in China, this study employs a two-way fixed effect model to empirically examine the direct impact of APSs on PCE, while considering the carbon emissions of major crops and APS levels in China. The robustness of the benchmark regression results is assessed by altering core variables and employing an instrumental variable model. Additionally, a mediating effect model is utilized to investigate its mechanism, along with a threshold effect model to explore the nonlinear relationship between APSs and PCE across different operational scales. Furthermore, considering variations in planting area, input of production factors, farming methods, and other aspects among different crops as well as varying total greenhouse gas emissions [42], this study aims to explore further potential disparities in the impact of APSs on carbon emissions across various crops.

This paper contributes in three main ways. Firstly, the method of measuring PCE takes into account variations across regions, crop growth habits, and climatic conditions, thereby enhancing the accuracy of calculation results. Secondly, by integrating APSs and PCE within the same analytical framework, this study not only enriches research on the carbon emission reduction effect of APSs but also provides a novel perspective for sustainable agricultural development. Finally, in terms of research content, this paper incorporates the arable land operation area into the analytical framework to analyze the nonlinear impact of APSs on PCE and explore their nonlinear relationship using a threshold model. It refutes arguments suggesting that expanding farmland management scale reduces carbon emissions while advocating for moderate-scale operations instead. These findings offer policymakers new ideas to maximize the carbon emission reduction effect of APSs.

The remainder of this paper is organized as follows: Section 2 consists of the theoretical analysis and research hypotheses. Section 3 describes the study design, including the research methodology, variable selection, and data sources. Section 4 presents and analyzes the main findings. Further discussion is provided in Section 5. Finally, key conclusions and policy implications are summarized in Section 6.

2. Theoretical Analysis and Research Hypothesis

The theory of agricultural division of labor suggests that as the specialized division of labor in the field of agricultural production continues to develop, professional service organizations or individuals can be entrusted with part or all of the links in agricultural production, ultimately promoting the formation of a market for APSs [36]. Concurrently, APSs permeate the entire agricultural production process and exerts a substantial influence on agriculture [43]. On the one hand, the division of labor in agricultural production can subdivide the relatively complex production process into independent segments, effectively facilitating the promotion and application of APSs [44]. Additionally, it enables the integration of smallholder economies in an era characterized by agricultural division of labor, thereby enhancing the efficiency of agricultural operations [45]. On the other hand, APSs can provide corresponding services in all facets of agricultural production, including guidance on production and operations, support for green technologies, and agricultural machinery services [46]. From the theoretical analysis, while APSs can facilitate the rational allocation of agricultural input factors and reduce carbon emissions at the source by offering guidance services for agricultural production and operations [47], they may also contribute to increased carbon emissions due to elevated fuel consumption during mechanized services such as farmland irrigation and land tillage, especially when providing mechanical operation services on finely fragmented arable land [48]. Therefore, how do APSs impact the PCE? Is they inhibitory or promotive? The aforementioned analysis suggests that while it is possible to conclude that APSs can directly impact the PCE, making a clear judgment on the carbon emission reduction effect of APS at a theoretical level remains difficult. The following competitive hypothesis is proposed based on this.

Hypothesis 1_a. 

APSs significantly inhibit PCE.

Hypothesis 1_b. 

APSs significantly promote PCE.

According to the theory of farmers’ adaptive production behavior, small-scale farmers, as the primary actors in agricultural operations, possess the ability to adapt swiftly to changes in the external market environment [49]. They can effectively adapt by modifying their agricultural production habits and adjusting their methods accordingly. This will stimulate both organizational advantages and labor division potential among resilient small-scale farmers, injecting new vitality into the organic connection between small-scale farmers and modern agriculture while promoting green and low-carbon development in agriculture [34]. In addition, the formation of the APS market enables service organizations to effectively introduce modern production factors into agriculture in a market-oriented manner, which can impact the inherent mode of traditional agricultural production [7] and induce small-scale farmers to engage in agricultural land transfer [36], thereby further reducing the PCE through the scale effect. On the one hand, APSs guide farmers to participate in decision-making regarding agricultural land transfer through the advantage of division of labor [33], reintegrate fragmented land by “assembling the parts into a whole”, promote large-scale operation in the planting industry, realize optimal allocation of agricultural input factors [32], and, thus, improve the efficiency of utilizing agricultural resources. On the other hand, as rational small-scale farmers striving for profit maximization [50], they strategically transfer fallow cultivated land to more efficient large-scale management entities after careful consideration of their production conditions. These entities can implement centralized land management more effectively than ordinary farmers [51]. As a result, this approach efficiently reduces the requirement for excessive inputs of agricultural fertilizers, pesticides, and other chemicals due to limited land resources and inseparable factors, while simultaneously decreasing carbon emissions at their source [52]. Based on the above analysis, this paper proposes the second research hypothesis.

Hypothesis 2. 

APSs can indirectly reduce PCE by facilitating the transfer of land.

The current agricultural management system in our country is predominantly composed of small farmers, who constitute 98% of the total number of agricultural managers [53]. Decentralized smallholder production methods have long faced challenges in achieving scale and intensification, resulting in significant agricultural non-point source pollution and hindering the progress of agricultural modernization [54]. The improvement of China’s APS level has gradually resolved the shortcomings in technology accumulation, production management, and scale effect among traditional small farmers, leading to an upward trend in the marginal benefits of agricultural production thereof [45]. The continuous expansion of operational scale, however, may result in an inverse relationship between the area of cultivated land operations and agricultural production efficiency [55]. The marginal return of agricultural productive services also begins to decline simultaneously, primarily manifested in the absolute increase in agricultural factor resource input and total consumption of machinery fuel. This phenomenon partially aligns with the Theory of Metabolic Rifts and the substitution effect, ultimately resulting in an overall rise in total carbon emissions [56]. For this reason, certain scholars have raised doubts regarding the compatibility of large-scale operations with sustainable agricultural development [57]. They argue that while large-scale cultivation may reduce the intensity of non-point source pollution in agriculture, it does not necessarily guarantee a reduction in overall carbon emissions compared to small-scale operations [58]. This indicates that as the scale of land management increases, the impact of APSs on PCE may exhibit non-linear characteristics. Based on the above analysis, this paper proposes the third research hypothesis.

Hypothesis 3. 

The effect of APSs on PCE is non-linear and is based on the scale of land operation as a threshold.

Based on the above analysis, the purpose of this paper is to empirically examine the relationship between APSs and PCE, based on theoretical analysis and research hypotheses. The main framework is illustrated in Figure 1.

3. Methodology and Data

3.1. Model Specification

3.1.1. Panel Benchmark Regression Model

Referring to the existing literature [59], this paper utilizes a two-way fixed effect model with city and year variables to estimate the impact of APSs on PCE. The choice of this model is motivated by its ability to address endogenous issues through the inclusion of city-fixed effects and capture regional common trends and fluctuations through the inclusion of year-fixed effects. Additionally, cluster-robust standard errors are employed to mitigate any potential influence from heteroscedasticity on the model’s results. The following model is constructed in this paper based on this analysis.

$${PCE}_{i,t}={\alpha }_0+{\alpha }_1{APS}_{i,t}+{\alpha }_2{\displaystyle \sum }{X}_{i,t}+{\lambda }_{city}+{\lambda }_{year}+{\epsilon }_{i,t}$$

(1)

In Equation (1), ${\mathrm{P}\mathrm{C}\mathrm{E}}_{\mathrm{i},\mathrm{t}}$ represents the PCE in region i during year t, ${\mathrm{A}\mathrm{P}\mathrm{S}}_{\mathrm{i},\mathrm{t}}$ represents the level of APSs in region i during year t, and $\sum {\mathrm{X}}_{\mathrm{i},\mathrm{t}}$ represents a series of control variables related to PCE. ${\mathsf{\lambda }}_{\mathrm{c}\mathrm{i}\mathrm{t}\mathrm{y}}$ and ${\mathsf{\lambda }}_{\mathrm{y}\mathrm{e}\mathrm{a}\mathrm{r}}$ denote fixed effects for regions and times respectively. ${\mathsf{\epsilon }}_{\mathrm{i}\mathrm{t}}$ is the random disturbance term, ${\mathsf{\alpha }}_0$ is the model intercept term, and ${\mathsf{\alpha }}_1$ and ${\mathsf{\alpha }}_2$ are coefficients corresponding to APS and control variables, respectively.

3.1.2. Impact Mechanism Model

The previous theoretical analysis suggests that APSs can indirectly impact PCE by facilitating land transfer. Consequently, a three-stage mediating effect model is constructed, drawing on existing research findings [47]. The first stage of the mediating effect model is represented by Equation (1), while the second and third stages are illustrated in Equations (2) and (3).

$$M_{i,t}={\beta }_0+{\beta }_1{APS}_{i,t}+{\beta }_2{\displaystyle \sum }{X}_{i,t}+{\lambda }_{city}+{\lambda }_{year}+{\epsilon }_{i,t}$$

(2)

$${PCE}_{i,t}={\gamma }_0+{\gamma }_1{APS}_{i,t}+{\gamma }_2{M}_{i,t}+{\gamma }_3\sum X_{i,t}+{\lambda }_{city}+{\lambda }_{year}+{\epsilon }_{i,t}$$

(3)

In Equations (2) and (3), ${\mathrm{M}}_{\mathrm{i},\mathrm{t}}$ is the mediating variable; ${\mathsf{\beta }}_0$ and ${\mathsf{\gamma }}_0$ represent the intercept term; ${\mathsf{\beta }}_1$ and ${\mathsf{\gamma }}_1$ are the coefficients of APS; ${\mathsf{\gamma }}_2$ represents the coefficient of the mediating variable; while other variables have the same meanings as Equation (1).

3.1.3. Panel Threshold Regression Model

The Hansen panel regression model [60] is employed in this study to examine the presence of a threshold effect. The explanatory variable used is PCE, while the core explanatory variable is APSs, and the threshold variable is SCAL, which represents the land management scale. The threshold effect regression model is constructed accordingly, using the following equation.

$$\begin{gathered} {\mathrm{P}\mathrm{C}\mathrm{E}}_{\mathrm{i},\mathrm{t}}={\mathsf{\omega }}_0+{\mathsf{\omega }}_1{\displaystyle \sum }{X}_{i,t}+{\phi }_1{APS}_{i,t}\times \mathrm{I}\left({\mathrm{S}\mathrm{c}\mathrm{a}\mathrm{l}}_{\mathrm{i},\mathrm{t}}<{\mathsf{\theta }}_1\right)+{\phi }_2{APS}_{i,t} \\ \times \mathrm{I}\left({\mathsf{\theta }}_1<{\mathrm{S}\mathrm{c}\mathrm{a}\mathrm{l}}_{\mathrm{i},\mathrm{t}}<{\mathsf{\theta }}_2\right)+······+{\phi }_n{APS}_{i,t}\times \mathrm{I}\left({\mathsf{\theta }}_{\mathrm{n}-1}<{\mathrm{S}\mathrm{c}\mathrm{a}\mathrm{l}}_{\mathrm{i},\mathrm{t}}<{\mathsf{\theta }}_{\mathrm{n}}\right)+{\phi }_{n+1}{APS}_{i,t}\times \mathrm{I}({\mathrm{S}\mathrm{c}\mathrm{a}\mathrm{l}}_{\mathrm{i},\mathrm{t}}>{\mathsf{\theta }}_{\mathrm{n}})+{\epsilon }_{i,t} \end{gathered}$$

(4)

In Equation (4), ${\mathsf{\theta }}_1$,${\mathsf{\theta }}_2$,$······$,${\mathsf{\theta }}_{\mathrm{n}}$ denote the threshold values at n different levels. $\mathrm{I}(·)$ is the indicator function and ${\mathrm{S}\mathrm{c}\mathrm{a}\mathrm{l}}_{\mathrm{i},\mathrm{t}}$ is the threshold variable, while other variables have the same meanings as in Equation (1).

3.2. Variable Selection

3.2.1. Explained Variable

The explained variable in this study is PCE. Drawing on the research findings of numerous scholars [13,14], this paper calculates the PCE in China from the following two aspects. The first aspect is the carbon emissions generated by inputs in agriculture. These inputs encompass six categories of carbon sources: chemical fertilizers, pesticides, agricultural diesel fuel, agricultural plastic film, agricultural irrigation, and land tillage [15]. The carbon emission coefficients and their corresponding reference sources are presented in

Table 1. Carbon emission coefficients of different carbon sources in agricultural inputs.

Carbon Source	Emission Coefficient	Unit	Data Reference Source (Basis)
Fertilizer	0.8956	kg CE/kg	ORNL (Oak Ridge National Laboratory)
Pesticide	4.9341	kg CE/kg	ORNL (Oak Ridge National Laboratory)
Diesel oil used in agriculture	0.5927	kg CE/kg	IPCC (Intergovernmental Panel on Climate Change)
Agricultural film	5.18	kg CE/kg	IREEA (Institute of Resource, Ecosystem and Environment of Agriculture)
Irrigation	25	kg CE/hm2	Dubey et al. [61]
Tillage	312.6	kg CE/km2	College of Agronomy and Biotechnology, China Agricultural University

. The second aspect is the carbon emissions generated during crop cultivation. China’s main crops are rice, wheat, and corn, which are extensively grown and equally important in agricultural production [20]. Therefore, this study aims to investigate the greenhouse gas emissions generated during the cultivation of three major food crops in China, namely rice, corn, and wheat. The gas emission types and their corresponding emission factors for the relevant crops are presented in

Table 2. The emission types and their corresponding coefficients of major grain crops in China.

Carbon Source	Emission Coefficient		Unit	Data Reference Source (Basis)
Rice	North China	234	kg (CH4)/hm2	Guidelines for the Compilation of Provincial GHG Inventories in China
	Eastern China	215.5
	Central and South China	236.7
	Southwest China	156.2
	Northeast China	168
	Northwest China	231.2
Corn	2.53		kg (N2O)/hm2	Li et al. [62]
Spring wheat	0.4		kg (N2O)/hm2
Winter wheat	2.05		kg (N2O)/hm2

Note: Taking into account the specific circumstances of wheat cultivation in China, this study utilized the carbon emission coefficient of spring wheat for seven provinces and regions, namely Heilongjiang, Jilin, Liaoning, Xinjiang, Ningxia, Inner Mongolia, and Gansu; while, for the remaining provinces (autonomous regions and municipalities directly under the Central Government), the carbon emission coefficient of winter wheat was employed.

.

The formula for calculating PCE is as follows:

$${\mathrm{P}\mathrm{C}\mathrm{E}}_{\mathrm{i},\mathrm{t}}={\displaystyle \sum }_{\mathrm{k}=1}^{\mathrm{n}}{\mathrm{e}}_{\mathrm{k},\mathrm{i},\mathrm{t}}={\displaystyle \sum }_{\mathrm{k}=1}^{\mathrm{n}}{\mathsf{\delta }}_{\mathrm{k}}{\mathsf{\omega }}_{\mathrm{i},\mathrm{t}}$$

(5)

In Equation (5), ${\mathrm{P}\mathrm{C}\mathrm{E}}_{\mathrm{i},\mathrm{t}}$ represents the PCE in region i in year t, $\mathrm{k}$ represents the types of carbon emission sources ($\mathrm{k}=\mathrm{1,2},3······$), ${\mathrm{e}}_{\mathrm{k},\mathrm{i},\mathrm{t}}$ represents the carbon emissions from different sources in the planting industry, and ${\mathsf{\delta }}_{\mathrm{k}}$ and ${\mathsf{\omega }}_{\mathrm{i},\mathrm{t}}$ represent the carbon emission coefficient of each carbon emission source and the corresponding factor input, respectively. In order to facilitate analysis and dispel any reader’s doubts, the carbon emissions mentioned in this paper specifically pertain to the emissions resulting from greenhouse gases such as CO₂, CH₄, and N₂O generated by various sources within the planting industry, which have been uniformly converted into carbon equivalent (CE), with the following conversion rates: 1 t CH₄ = 6.82 t CE and 1 t N₂O = 81.27 t CE [20].

3.2.2. Core Explanatory Variable

The core explanatory variable in this study is the level of APS, which is quantified as the output value of APSs per unit of crop planting area. In other words, it represents the ratio between the output value of APSs and crop planting area [36]. However, the data obtained from the China Tertiary Industry Statistical Yearbook represents the total output value of agriculture, forestry, animal husbandry, and fishery services rather than agricultural productive services. Therefore, it is necessary to calculate the output value of agricultural productive services using the following formula:

$$\mathrm{A}\mathrm{P}\mathrm{S}={\displaystyle \frac{\mathrm{O}\mathrm{A}}{\mathrm{T}\mathrm{O}\mathrm{A}}}\times \mathrm{T}\mathrm{O}\mathrm{S}$$

(6)

In Equation (6), APS represents the output value of agricultural productive services; OA denotes the total value of agricultural output; TOA signifies the total output value of agriculture, forestry, animal husbandry, and fishery combined; TOS refers to the total output value of agriculture, forestry, animal husbandry, and fishery services.

3.2.3. Control Variables

To mitigate the adverse effects of other factors on the regression outcomes, control variables that may influence PCE are incorporated into the econometric model based on relevant findings from previous studies [21,47]. The level of agricultural mechanization (AML) is quantified by the ratio between the aggregate power of agricultural machinery and the cultivated area. Grain productivity (GP) is measured by the ratio between grain output and cultivated area. The Agricultural Disaster Rate (ADR) is calculated based on the ratio of the agricultural disaster area to the crop planting area. Agricultural Financial Support (AFS) is measured by the ratio of expenditure on agricultural and forestry affairs to general expenditure in the local budget. The Income Level of Rural Residents (ILRR) is directly determined by per capita disposable income among rural residents. The non-agricultural employment level (NAEL) is calculated by dividing the total employment in the secondary and tertiary industries by the overall employment figure. The multiple cropping index (MCI) is determined by dividing the area of land used for growing crops by the total cropland area.

3.2.4. Mediating Variable

According to the previous theoretical analysis and hypothesis, the agricultural land transfer rate (ALTR) is chosen as the mediating variable. The ALTR is calculated by dividing the total area of household contracted cultivated land transferred by the area of household contracted cultivated land for operation [32].

3.2.5. Threshold Variable

The selected threshold variable in this paper is the scale of land operation (SCAL), which is represented by the per capita planting area. Specifically, it is measured as the ratio of crop planting area to agricultural labor inputs [20].

3.3. Data

This study focuses on a research area of 30 provincial-level administrative regions in China, covering the period from 2005 to 2021. Due to the limited availability of sample data, Hong Kong, Macao, Taiwan, and Tibet are not included within the scope of this study. The gross output values of agriculture, forestry, animal husbandry, and fishery services are sourced from the China Tertiary Industry Statistical Yearbook for data analysis in this study. Other variables are obtained from the China Statistical Yearbook, the China Rural Statistical Yearbook, the China Agricultural Statistical Yearbook, as well as regional statistical yearbooks. Excel 2021 software was used to process outliers and missing values in the data, while an interpolation method was employed to supplement the missing values. Descriptive statistics for all variables used in this paper are presented in

Table 3. Descriptive statistics of variables.

Variables	Unit	Mean	Std. Dev.	Min	Max
Explained variable	Million tons	5.079	3.478	0.153	12.751
PCE
Core explanatory variable	100 million CNY/1000 hm2	3.951	3.481	0.199	21.603
APS
Control variables	10,000 KW/1000 hm2	0.603	0.250	0.211	1.416
AML
GP	Tons/hm2	5.274	1.040	3.046	8.479
ADR		0.194	0.145	0.000	0.936
AFS		0.108	0.033	0.027	0.204
ILRR	10,000 CNY	1.037	0.634	0.197	3.852
NAEL		0.653	0.155	0.248	0.982
MCI		1.313	0.397	0.488	2.427
Mediating variable		0.671	0.506	0.119	3.179
ALTR
Threshold variable	Hm2/person	0.698	0.366	0.272	2.92
SCAL

.

4. Results and Analysis

4.1. Results and Analysis of PCE

The utilization of the Chord graph has been extensively employed in previous research [62,63,64], and this study employs the Origin2022 version to generate a Chord graph that illustrates the temporal and source-based variations in total carbon emissions within the planting industry (Figure 2). There have been fluctuations and an overall increase in China’s PCE over recent years. The PCE increased from 135.47 million tons (Mt) of carbon equivalent (CE) in 2005 to the highest level of 163.57 Mt CE in 2015, and began to decline slightly after that year, possibly due to the government’s implementation of the Action Plan for Zero Growth of Pesticide Use by 2020 [36]. In addition, carbon emissions from various carbon sources associated with cropping activities were further analyzed. Taking 2021 as an example, agricultural inputs and crop growth accounted for 63.1% and 36.9% of the total carbon emissions from the planting industry, respectively. The CH₄ emissions from rice accounted for the largest source of carbon emissions generated by crop growth, totaling 42.66 Mt CE. Fertilizer inputs emerged as the primary source of carbon emissions resulting from agricultural inputs, amounting for 46.16 Mt CE. Overall, in 2021, carbon emissions from fertilizer inputs constituted the primary source of PCE and accounted for approximately 30.9% of its total emissions. The emission of CH₄ from rice cultivation ranked second and contributed to around 28.6% of the total emissions. Other carbon sources such as agricultural film, pesticides, diesel, irrigation, and tillage make up a small proportion of overall carbon emissions.

4.2. The Effect of APSs on PCE

4.2.1. Benchmark Regression Analysis

Since short panel data (N > T) is utilized in this study, the regression analysis avoids spurious regression issues that are encountered in simple time series models [65]. Consequently, there is no necessity to consider the unit root problem. However, to ensure the reliability of regression results, this paper employs the variance inflation factor (VIF) to test for multicollinearity among variables (

Table 4. VIF test results.

Variables	APS	AML	GP	ADR
VIF	2.14	1.44	1.81	1.50
Variables	AFS	ILRR	NAEL	MCI
VIF	1.07	4.10	3.02	1.32

). The results indicate that the maximum VIF value for each variable is 4.10, which is significantly lower than 10 and satisfies the condition of no severe multicollinearity. These findings demonstrate that there are no significant issues with multicollinearity among the selected variables in this study, allowing for subsequent benchmark regression analysis to be conducted.

The benchmark regression results are presented in

Table 5. Benchmark regression results.

Variables	Model (1)	Model (2)	Model (3)
	PCE	PCE	PCE
APS	−0.092 ***	−0.131 ***	−0.106 ***
	(0.015)	(0.015)	(0.015)
AML		2.009 ***	1.782 ***
		(0.237)	(0.251)
GP		0.520 ***	0.560 ***
		(0.110)	(0.116)
ADR		0.282	0.338
		(0.236)	(0.248)
AFS			1.357 **
			(0.633)
ILRR			−0.586 ***
			(0.135)
NAEL			−0.951
			(0.584)
MCI			0.060
			(0.144)
Regional and time-fixed effects	YES	YES	YES
Constant	5.444 ***	1.588 **	2.411 ***
	(0.057)	(0.654)	(0.873)
R2	0.982	0.985	0.985
N	510	510	510

Note: Significance at the 1%, 5% levels are expressed by ***, **, respectively. Standard errors are presented in parentheses.

. Model (1) represents the direct impact of APS on PCE, without controlling for any variables. Models (2) and (3) display the regression results after gradually incorporating control variables; this can also provide evidence for testing the robustness of regression results. The findings indicate that the influence coefficient of APS on PCE consistently exhibits a significantly negative effect at a 1% confidence level. This outcome confirms the competitive Hypothesis 1_a, demonstrating that APS exerts a substantial inhibitory impact on PCE. The reason is that APSs can effectively provide farmers with corresponding services and technical guidance, promoting rational use of fertilizers, pesticides, and other agricultural resource inputs, thereby facilitating the green and low-carbon transformation of the planting industry [47].

Regarding the control variables, the level of agricultural mechanization (AML), food productivity (GP), and agricultural financial support (AFS) will all contribute to an increase in PCE. The possible explanation is that the improvement of AML implies the need to increase diesel consumption, which aligns with the findings of Guan et al. [48]. The improvement of GP means that the yield of food crops per unit planted area increases, as well as a corresponding rise in the overall emission levels of greenhouse gases such as CH₄ and N₂O during crop growth, which is consistent with the findings of Shakoor et al. [42]. The increase in AFS means that, on the one hand, rational smallholders will receive more agricultural financial subsidies, leading to an increase in the purchase of agricultural machinery and tools, which directly results in higher total mechanical fuel consumption. On the other hand, they may also increase their input of agricultural resources such as pesticides and fertilizers to maximize output, which is generally consistent with the results of Wang et al. [56]. However, the increase in rural residents’ income levels will lead to a significant reduction in PCE. This indicates that higher income levels among rural residents result in an increased willingness among farmers to purchase APSs, thereby effectively enhancing agricultural production efficiency and reducing PCE.

4.2.2. Robustness Test

To test the robustness of the benchmark regression results in Equation (1), control variables have been gradually added to Table 5 above for a preliminary test. However, to provide more compelling evidence for the robustness of the benchmark regression results, this chapter will also conduct a robustness test using the replacement of key variables and instrumental variables (

Table 6. Robustness test results.

Variables	Replacing Key Variables		Instrumental Variables (2SLS)
	Model (1)	Model (2)	Model (3)	Model (4)
	TPCE	PCE	APS	PCE
APS	−0.031 ***			−0.149 ***
	(0.010)			(−0.050)
PAPS		−0.470 **
		(0.203)
L.APS			1.086 ***
			(0.020)
Control variables	YES	YES	YES	YES
Regional and time-fixed effects	YES	YES	YES	YES
Constant	0.257	3.047 ***	0.022	−0.280
	(0.451)	(0.878)	(0.203)	(1.079)
R2	0.883	0.984	0.970	0.478
F			28.219
LM			14.12 ***
N	510	510	510	510

Note: Significance at the 1%, 5%levels are expressed by ***, **, respectively. Standard errors are presented in parentheses.

).

Firstly, the robustness test is conducted by replacing key variables. In Model (1), the explained variable PCE is substituted with TPCE, which represents the carbon emission intensity of the planting industry and is calculated as the ratio of PCE to agricultural labor input [47]. In Model (2), the core explanatory variable APS is replaced with PAPS, which denotes per capita APS output and is computed as the ratio of APS to agricultural labor input [21]. The findings indicate that even after substituting the proxy variable, the coefficient of the core explanatory variable on the explained variable remains statistically significant and negative at both 1% and 5% levels.

Secondly, the benchmark regression results may be affected by endogeneity problems. On one hand, there could be a reverse causality between APS and PCE. On the other hand, there is a possibility of missing time-varying characteristics that might impact the regression results. To address this issue, this study incorporates the lagged variable (L.APS) of APS as the core explanatory variable based on existing research findings [47]. It is used as an instrumental variable to conduct a robustness test through the two-stage least square method (2SLS). Model (3) presents the results of the first-stage regression analysis, where the coefficient of the instrumental variable is significantly positive at a 1% level. This indicates that there is a correlation between the selected instrumental variable and APS. According to the judgment criteria of Stock et al. [66], the F-value in the first stage is 28.219, which exceeds the critical value of a 10% error (16.83). This indicates that the selected instrumental variable successfully passes the weak instrumental variable test. Additionally, with an LM value of 14.12 and a statistically significant P value at a level of 1%, it can be concluded that the instrumental variable also passes the unidentifiable test. Model (4) represents the regression result of the second stage. The coefficient of APS remains negative and statistically significant at the 1% level, even after incorporating instrumental variables to address endogeneity concerns. This reaffirms the robustness of the regression results in Equation (1).

4.3. The Effect Mechanism of APS on PCE

The mediating mechanism analysis is conducted using the stepwise test regression coefficient method, as presented in

Table 7. Results of mediating mechanism analysis.

Variables	Model (1)	Model (2)
	ALTR	PCE
APS	0.079 ***	−0.088 ***
	(0.007)	(0.021)
ALTR		−0.238 **
		(0.118)
Control	YES	YES
Regional and time-fixed effects	YES	YES
Constant	0.300	2.483 ***
	(0.272)	(0.869)
Sobel test		−0.270 * (0.044)
Bootstrap test (ind_eff)		−0.270 *** (0.043)
Bootstrap test (dir_eff)		0.148 *** (0.054)
R2	0.924	0.986
N	510	510

Note: Significance at the 1%, 5%, and 10% levels are expressed by ***, **, and *, respectively. Standard errors are presented in parentheses.

. The impact of APS on the agricultural land transfer rate (ALTR) is estimated firstly, according to Equation (2). The results from Model (1) demonstrate a significant impact of APS on ALTR at the 1% confidence level, with an estimated coefficient of 0.079. This suggests that APS plays a crucial role in guiding farmers’ involvement in decision-making regarding farmland transfer, facilitating the consolidation of fragmented and decentralized cultivated land, and promoting the scale effect [33]. Secondly, based on Equation (3), we estimate the impact of APS and ALTR on PCE. As indicated in Model (2), both APS and ALTR coefficients are found to be negative and significant at the levels of 1% and 5%, respectively. This demonstrates that APS can reduce PCE through ALTR, with ALTR playing a partial intermediary role. The above results are further validated in this study through Sobel analysis and Bootstrap sampling, confirming the consistency between the direct and indirect effects of mediating variables as observed in the aforementioned analysis. Consequently, Hypothesis 2 is verified.

4.4. Threshold Effect Based on the Scale of Land Operation

The model should be tested for threshold effect before estimating its threshold value [20]. The Hansen panel regression model in Equation (4) is used to conduct the multiple threshold effect test, and the Bootstrap method in Stata 18.0 software is utilized for performing 300 iterations of sampling (

Table 8. Results of the threshold effect test.

Threshold Variable	Threshold Test	F-Statistic	Bootstrap Times	10% Critical Value	5% Critical Value	1% Critical Value
SCAL	Single	117.25 ***	300	31.5654	38.8033	52.3216
	Double	6.84	300	118.6179	155.9860	226.2801
	Triple	6.69	300	22.9362	94.8019	185.3973

Note: *** indicates significance at the significance level of 1%.

). The results indicate that when considering APS as the core explanatory variable and scale of land operation (SCAL) as the threshold variable, the model only exhibits a significant single threshold effect at a 1% level. However, it does not demonstrate any significant double or triple-threshold effects. This demonstrates that the impact of APS on PCE exhibits nonlinear characteristics, and there exists a single threshold value for SCAL.

The regression analysis of the threshold effect is conducted after successfully passing the test for threshold effects and determining the precise number of thresholds (

Table 9. Regression results of the threshold model.

Statistical Magnitude	Results
Explanatory variable	APS
Threshold variable	SCAL
Threshold number	Single threshold
Threshold value (θ)	1.536
APS·D(Scal ≤ 1.536)	−0.093 *** (0.024)
APS·D(Scal > 1.536)	1.840 *** (0.537)
p value	0.0000
F value	117.25
N	510

Note: *** represents a significance level of 1%. Standard errors are presented in parentheses.

). The results indicate that the threshold variable has a single threshold value of 1.536, and APS initially has a negative impact on PCE before turning positive. Specifically, when the SCAL is below 1.536 hm², there exists a significant impact of APS on PCE at a confidence level of 1%, with an estimated coefficient of −0.093. Conversely, when the SCAL exceeds 1.536 hm², there remains a significant impact of APS on PCE at a confidence level of 1%, with an estimated coefficient of 1.840. The reason can be explained as follows: currently in China, APSs are primarily charged based on the operating area. When farmers have a planting scale smaller than the threshold value, they incur relatively lower costs to purchase APSs. Consequently, rational smallholders exhibit a higher willingness to adopt APSs [34] and then promote the improvement of agricultural production efficiency. At the same time, APSs can also provide green technical guidance for small-scale farmers, promote rational application of agrochemicals by farmers, and reduce PCE [52]. However, when the scale of planting exceeds the threshold value, there is a decline in the marginal return of APSs. This is evidenced through an increase in fuel consumption for mechanization services such as farmland irrigation and land tillage [56], as well as a rise in the absolute amount of agricultural chemicals required during the production process [58]. These factors objectively contribute to an overall increase in PCE. The findings demonstrate that as the scale of land management expands, there is a U-shaped nonlinear correlation between APS and PCE. Therefore, Hypothesis 3 is verified.

4.5. Heterogeneity Analysis

Considering factors such as natural geographical location [36], agricultural functional areas [67], and productivity disparities [68], numerous scholars have examined the heterogeneity of carbon emission reduction effects in APSs, providing ample and reliable grounds for policy optimization. Based on previous research, this paper attempts to categorize the carbon emissions of the planting industry into three groups: rice, wheat, and corn, for regression analysis. The formula used for calculating carbon emissions of different crops is as follows:

$${CE}_i={\displaystyle \frac{S_i}{S_{crop}}}\times C_{input}+C_i$$

(7)

In Equation (7), ${CE}_i$ represents the carbon emissions of class i crops, $S_i$ represents the planting area of class i crops, $S_{crop}$ represents the total planting area of crops, $C_{input}$ represents the total carbon emissions generated by agricultural input, and $C_i$ represents the total carbon emissions generated by class i crops during their growth process ($C_{input}$ and $C_i$ are calculated based on the carbon emission coefficients in Table 1 and Table 2 respectively).

The carbon emissions resulting from the cultivation of rice, wheat, and corn can be calculated based on Equation (7), and these emissions for different crops can then be incorporated into Equation (1) for grouped regression analysis (referring to

Table 10. Results of heterogeneity analysis.

Variables	Model (1)	Model (2)	Model (3)	Model (4)	Model (5)	Model (6)	Model (7)	Model (8)	Model (9)
	RCE	WCE	CCE	PCE	PCE	PCE	PCE	PCE	PCE
APS	−0.194 **	−0.236 ***	−0.031	−0.215 ***	−0.020	0.081 ***	−0.023 *	−0.092 ***	0.125 ***
	(0.088)	(0.068)	(0.034)	(0.043)	(0.014)	(0.031)	(0.014)	(0.027)	(0.032)
Control	YES	YES	YES	YES	YES	YES	YES	YES	YES
Regional and time-fixed effects	YES	YES	YES	YES	YES	YES	YES	YES	YES
Constant	0.815	−0.607	0.153	2.921 *	2.136 ***	1.502	4.179 ***	0.326	1.472
	(0.563)	(0.474)	(0.232)	(1.527)	(0.778)	(1.197)	(0.981)	(1.853)	(1.332)
R2	0.987	0.963	0.990	0.963	0.998	0.971	0.996	0.985	0.971
N	510	510	510	221	119	170	204	153	153

Note: Significance at the 1%, 5%, and 10% levels are expressed by ***, **, and *, respectively. Standard errors are presented in parentheses.

Models 1–3). The findings indicate that APSs have a significant adverse impact on both rice carbon emissions (RCE) and wheat carbon emissions (WCE), while no discernible effect is observed on corn carbon emissions (CCE). Possible explanations are as follows: On one hand, taking agricultural machinery services as an example, the characteristics of crop adaptation to mechanization mainly include plant morphological characteristics, growth duration, and resistance to weather conditions. In comparison to corn, with tall plants and well-developed roots, rice and wheat possess growth habits that better align with the requirements of large-scale mechanized operations [69]. On the other hand, in China, the levels of plant protection services for fertilizer and water management, as well as disease and insect control, are comparatively higher in rice- and wheat-producing areas than in corn-producing areas based on the current state of APSs. Moreover, compared to rice and wheat-producing regions, corn cultivating areas suffer from severe land fragmentation and poor soil quality, which significantly exacerbate challenges faced by APSs [20]. The aforementioned factors have exerted a certain influence on the inhibitory effect of APSs on carbon emissions from corn.

Additionally, this study examines the heterogeneity of carbon emission reduction effects associated with APSs across different grain planting functional regions (referring to Table 10 Models 4–6) and various geographical regions (referring to Table 10 Models 7–9). The difference analysis of different grain planting function areas, such as the main grain production area, the main grain marketing area, and the grain production and marketing balance area, is conducted in Models 4–6. The results indicate that APSs can still inhibit PCE in the main grain producing areas; however, their impact on PCE is not significant in the main grain marketing areas. Moreover, it is worth noting that APSs can promote PCE in regions with balanced grain production and marketing. A possible explanation is that the level of APSs is relatively high in major grain-producing regions, and the availability of service organizations is relatively optimized. However, in major grain marketing areas, the carbon emission reduction effect of APSs may not be fully realized, especially in balanced grain production and marketing regions where further improvements are needed for the APSs’ service system. On the other hand, Models 7–9 represent the analysis of differences among various geographical regions in China, including eastern China, central China, and western China, respectively. The results indicate that APSs can inhibit PCE in the eastern and central regions of China; however, it can promote PCE in the western region. This finding is consistent with the reasons mentioned above. The well-developed agricultural facilities and established APS service system in the eastern and central regions contribute to this effect, while the relatively underdeveloped service supply organizations in the western region hinder fully harnessing the carbon emission reduction potential of APSs.

5. Discussion

The analysis of the impact and mechanism of APSs on PCE is crucial for the sustainable development of agriculture, particularly considering the deep integration between the modern service industry and modern agriculture. The existing literature analysis reveals two types of literature that can be referenced and compared in this paper, both sharing a common foundation—APSs. The first topic concerns the relationship between APSs and agricultural green total factor productivity (AGTFP). Xu et al. utilized the GML productivity index for calculating AGTFP, revealing that APSs played a crucial role in fostering AGTFP growth, with technological progress being recognized as the primary driving force [36]. Li et al., using the Super-SBM model to assess AGTFP, found that both specialization and scale of APS supply organizations positively influenced local provinces’ AGTFP; however, specialization and scale of individual APS suppliers negatively impacted it [27]. The second one is about the relationship between APSs and agricultural carbon emission efficiency (ACEE). Zhu et al. specifically analyzed the impact of APSs on ACEE and its mechanism and believed that APSs could improve ACEE by optimizing input and increasing output. This improvement was mainly reflected in technological progress, adjustment of planting structure, optimization of factor allocation, and space overflow [70]. Although previous studies have started to acknowledge the carbon emission reduction impact of APSs, in contrast to these studies, the primary focus of this study is to examine the correlation between APSs and PCE, aiming to provide decision-making guidance for formulating carbon emission reduction policies related to the planting industry. The findings indicate that APSs can directly produce a significant inhibitory effect on PCE, which is mainly reflected in the provision of production and management guidance, green technology, and agricultural machinery services in agricultural production. The aforementioned factors can have a direct impact on agricultural input. Furthermore, the heterogeneity analysis can help us optimize carbon emission reductions in regards to planting, by making policy more targeted. Although previous studies have examined the heterogeneity of APS carbon emission reduction effects in terms of physical geographical location, agricultural functional area, and productivity differences, this paper distinguishes itself by employing a grouping regression approach to divide PCE into rice, wheat, and corn. This novel methodology constitutes one of the significant contributions made by this study. The inhibitory effect of APSs on both RCE and WCE was found to be significant, while no significant effect was observed on CCE. This paper believes that there are two reasons for this situation: firstly, it can be attributed to factors such as plant morphological characteristics, growth duration, and resistance to weather conditions [69]; secondly, it can be attributed to the unbalanced development level of APSs in different crop planting areas [20].

Existing studies mainly discuss the mechanism path of APSs influencing PCE from two aspects: technological progress [71] and scale management [41], but lack discussion on ALTR. We innovatively use ALTR as a mediator variable to analyze the mechanism of APS influence on PCE, providing a new research perspective for the study of the APS carbon emission reduction path. The technology and scale effects are not considered as direct paths of influence in this paper for the following reasons: Firstly, previous studies have already analyzed these factors, rendering further analysis redundant. Secondly, this paper argues that the technological effect and scale effect are largely influenced by the centralized management of land [70]. The study conducted by Adamopoulos and Restuccia revealed that land fragmentation and small planting scale are the primary reasons for the low ACCE in developing countries [72]. In China, it is necessary to consolidate fragmented land before achieving contiguous and centralized land management. This paper believes that APSs can effectively compensate for these shortcomings by promoting ALTR, while simultaneously maximizing the carbon emission reduction effect of APSs. As far as we know, the present study potentially represents the first attempt in China to reveal the impact of APSs on PCE at the provincial level and its mechanism.

In addition, the present study further investigates the nonlinear relationship between APSs and PCE across different SCAL, with SCAL serving as the threshold variable. Differing from previous studies, this paper considers that the relationship between APSs and PCE exhibits a U-shaped nonlinearity. The expansion of SCAL can inhibit (increase) PCE when the value of SCAL is below (above) 1.536. From the perspective of carbon emissions, agricultural chemical inputs and energy consumption are the two main sources of PCE. With the expansion of SCAL, APSs can reduce carbon emissions by optimizing agrochemical inputs. However, the increase in total agricultural machinery services and their fuels will directly or indirectly lead to an increase in carbon emissions [56]. Furthermore, from an effectiveness perspective, the carbon emission effect of energy consumption is greater than the carbon emission reduction effect of optimizing agrochemical input [70]. This shows that APSs, as an input factor, adhere to the principle of diminishing marginal returns in short-term production. There exists an optimal combination between APSs and SCAL. When this ratio is exceeded (the lowest point of the U-shaped curve), the marginal benefit of APSs decreases. The paper therefore suggests that advocating for moderate-scale operations and maintaining the operational scale at the lowest point of the U-shaped curve would maximize the effectiveness of carbon emission reduction in APSs.

6. Conclusions and Implications

PCE is a realistic constraint faced by China’s agricultural transformation towards modernization, and APSs play an important role in promoting the organic connection between small farmers and modern agriculture. A scientific analysis of the relationship between APSs and PCE holds great practical significance for the sustainable development of China’s agriculture. The impact mechanism of APSs on PCE was theoretically analyzed in this paper first. Subsequently, employing panel data from 30 provinces in China spanning the period from 2005 to 2021, this research investigates the impact, mechanism, and threshold effect of APSs on PCE using fixed effects, instrumental variable techniques, intermediary effect models, and threshold effect models. The main findings are as follows: firstly, the PCE in China exhibited an inverted U-shaped pattern from 2005 to 2021, initially increasing and then decreasing, with a peak in 2015. Secondly, APSs have a significant inhibitory effect on PCE. This result remains consistent even after conducting robustness tests and endogeneity tests using relevant instrumental variables. Thirdly, APSs have a significant positive impact on ALTR. By introducing ALTR as a mediating variable, both APSs and ALTR demonstrate a substantial inhibitory effect on PCE. Therefore, ALTR plays a partial mediating role in the process of APSs inhibiting PCE. Fourthly, when considering SCAL as a threshold variable with a single threshold effect, there is a U-shaped nonlinear relationship between APSs and PCE. Lastly, further research reveals that the inhibitory effect of APSs on carbon emissions varies among different crop types; particularly evident is their impact on reducing carbon emissions from rice and wheat cultivation.

The following suggestions are put forward based on the above conclusions:

(1)

Optimize the structure of APSs and fully leverage the carbon reduction effect of these services. On one hand, we should actively promote traditional agricultural services such as soil testing and fertilization, disease and pest control, and straw incorporation into fields to effectively mitigate non-point source pollution in agriculture. Simultaneously, we should also encourage the adoption of modern intelligent information technology services like UAV-based plant protection and digital agricultural assistance to drive agricultural modernization. On the other hand, we propose further implementation of APS policies to optimize the supply balance of APSs in various food functional areas and geographical regions. This will comprehensively enhance the level of agricultural servitization and effectively harness the carbon emission reduction potential of APSs.

(2)

Accelerate the improvement of China’s agricultural land transfer system, as the moderate scale effect formed by land transfer is conducive to reducing PCE. First of all, we should further clarify the property rights of agricultural land and actively introduce policies, legal documents, and measures related to agricultural land transfers so that such transfers can be conducted in accordance with laws and regulations. In the second place, under policy and legal compliance conditions, local government land management departments should relax control over agricultural land transfers, truly activate the operational and transfer rights of agricultural land, and fully leverage the carbon reduction effects brought about by such transfers.

(3)

Promote the moderate-scale management of agricultural land to maximize the carbon emission reduction effect of APSs. On one hand, agricultural land should be scientifically planned, and, after determining the amount of cultivated land strictly to meet the requirements, the mode of land transfer should be gradually and orderly transformed into a moderate-scale transfer, which is conducive to promoting the overall layout of moderate-scale agricultural management. On the other hand, it should be combined with the actual situation in each region, land production factors should be properly concentrated to exert maximum economic benefits from the land. Additionally, to achieve the optimal marginal efficiency of agricultural production, the scale of operation should be maintained at the lowest point of the U-shaped curve as far as possible, so as to maximize the carbon reduction effect of APSs.

The above research conclusions and suggestions can provide decision-making references for promoting the development of APSs and formulating carbon emission reduction policies related to agriculture. However, there are still certain limitations in this study: Firstly, the APSs studied in this paper are broad concepts, and different types of agricultural services have varying impacts on PCE. This paper only analyzed the carbon emission reduction effect of APSs on the planting industry from a macro perspective, without conducting a specific analysis of different types of agricultural services. Additionally, due to limitations in data samples, this study solely focused on PCE and did not consider the impact of APSs on carbon emissions in other sectors such as forestry, animal husbandry, and fishery. The above questions can provide potential ideas for further research, and we will wait to see how they are answered.