How Land Transfer Affects Agricultural Carbon Emissions: Evidence from China

Abstract

The effects of land transfer on agricultural carbon emissions and their underlying mechanisms must be investigated if we are to achieve sustainable development and environmentally friendly high-quality agricultural development. This research experimentally investigated the spatial impacts of land transfer on agricultural carbon emissions and their underlying causes using multiple econometric models based on provincial panel data covering the years 2010 to 2022. The results allow us to draw the following conclusions: (1) Land transfer significantly inhibits agricultural carbon emissions. This conclusion remained valid after various robustness checks, including a reduction in sample size, change model type, and adjustment of geographical regions. (2) Agricultural socialized services play a positive moderating role in the process of land transfer to curb agricultural carbon emissions. (3) Land transfer has a substantial spillover effect on agricultural carbon emissions, resulting in significantly reduced emissions in the immediate area and nearby regions.

1. Introduction

With the worsening of global climate change, reducing greenhouse gas emissions has become a major challenge faced by the international community [1,2]. As a significant source of greenhouse gas emissions, agricultural carbon emissions have attracted increasing attention [3,4,5]. Agricultural carbon emissions stem from the use of fertilizers, pesticides, and other inputs during agricultural production processes [6]; they are closely related to changes in land use [7,8]. As the largest developing country [9] and a major agricultural nation [10,11], the effect of China’s agricultural carbon emissions on national and global climate change must be addressed by researchers [12]. For a long time, decentralized and fragmented land management models have hindered growth in productivity, exacerbated agricultural carbon emissions, and constrained sustainable development. Breaking the pattern of small-scale farming and promoting large-scale farming is crucial if we are to achieve high-quality development and foster a low-carbon agricultural system. Within the current system of land management, the circulation of management rights under the ‘separation of three rights’ framework enables us to effectively reconcile the contradiction between land fragmentation and the need for large-scale farming; in doing so, we can reduce carbon emissions and initiate the green transformation of agriculture. In recent years, the No. 1 Central Document has repeatedly emphasized the importance of the orderly circulation of land management rights to promote moderate-scale farming. By the end of 2022, the area of rural land under circulation had exceeded 402 million hectares. This action has not only significantly improved the efficiency with which agricultural resources are allocated, but has also played a pivotal role in advancing low-carbon agriculture. Given China’s goal of ‘carbon neutrality’, it is vital that we conduct in-depth research into the impact of land circulation on carbon emissions and their underlying mechanisms; such research is of significant theoretical and practical value.

Theoretical research on agricultural carbon emissions provides an essential foundation for developing successful carbon reduction policies, sparking numerous academic debates and discussions. First, agricultural carbon emissions must be measured and analyzed. Existing studies have primarily adopted methods such as the emission factor approach [13,14], field measurement [15], life cycle assessment [16], carbon footprint [17,18], and the Tapio decoupling model [19] to measure agricultural carbon emissions in various metrics. In-depth analyses of the current status and evolutionary characteristics of agricultural carbon emissions have been conducted based on the scientific measurement of agricultural carbon emissions [20,21,22]. The second section of this study discusses factors that influence agricultural carbon emissions. Existing research has shown that factors such as level of urbanization [23], agricultural mechanization [24], land use efficiency [25], the rural digital economy [26], digital inclusive finance [27], green technological innovation within agriculture [28], agricultural socialized services [29], the structure of agricultural production [30], and the rural planting structure [31] impact agricultural carbon emissions.

The existing literature on land transfer primarily focuses on land transfer policies and their success, mainly exploring the impacts of land transfer on economic development [32], urbanization [33], the ecological environment [34], industrial structure [35], land use efficiency [36], green total factor productivity [37], and other aspects. As an essential component of China’s agricultural reform in recent years [38], land transfer is a crucial means of promoting moderately sized agricultural operations, profoundly influencing agricultural development [39,40]. Land transfer can also significantly improve the environment and promote green development while driving economic development [41]. However, there is currently a scarcity of research focusing on the relationship between land transfer and agricultural carbon emissions. According to existing studies, land rotation alters the organizational structure and size of agricultural output [42], optimizes the allocation of land resources [43], and thereby reduces carbon emissions [44]. Additionally, land transfer promotes land consolidation through the effects of scaled production and knowledge spillover, optimizes rural land resources, and drives the green transformation of agriculture, thereby curtailing agricultural carbon emissions.

In conclusion, researchers have conducted pertinent studies on land transfer and agricultural carbon emissions from several perspectives, providing a strong basis for more research in this area. Nevertheless, the existing literature does have some gaps. Previous research has not considered the spatial association between land transfer and agricultural carbon emissions; it has not examined the combined effects of land transfer and agricultural socialized services on agricultural carbon emissions. Thus, it is crucial that we investigate the intrinsic relationship between land transfer and agricultural carbon emissions and whether said relationship affects agricultural carbon emissions in nearby areas; the findings of such an investigation will help to promote green and sustainable agricultural development and reduce overall agricultural carbon emissions.

As such, in this study, we unified the concepts of land transfer, agricultural socialized services, and agricultural carbon emissions, and empirically tested the impact of land transfer on agricultural carbon emissions using benchmark regression, the moderating effect, and spatial Durbin models to provide theoretical guidance and support for the implementation of pertinent policies that will reduce agricultural carbon emissions. Empirical testing was based on a prior theoretical analysis. This study’s potential additions to the literature include (1) the scientific quantification of agricultural carbon emissions and an analysis of their spatio-temporal evolution; (2) a systematic study of the mechanisms regulating land transfer and their effect on agricultural carbon emissions from the perspective of socialized services; and (3) an exploration of the relationship between land transfer and agricultural carbon emissions from a spatial perspective, which will clarify the spatial effects of land transfer and open up further opportunities for related research.

2. Analysis of Mechanisms and Research Hypotheses

2.1. Mechanism of the “Direct Effect” of Land Transfer on Agricultural Carbon Emissions

The main ways in which land transfer has a “direct effect” on agricultural carbon emissions are as follows. Land transfer facilitates the consolidation of land from small-scale, fragmented farmers into larger, specialized farming enterprises and agricultural enterprises, resulting in the centralization of previously fragmented land and the formation of large-scale operations [45]. This procedure lowers the cost of land management, improves the efficiency with which land resources are used [46], and thus reduces agricultural carbon emissions. Moreover, land transfer boosts agricultural revenue, lowers the cost of agricultural output, and broadens the scope of farmers’ operations [47]. This encourages farmers to adopt low-carbon and eco-friendly farming methods, using organic instead of chemical fertilizers, using biological pest control techniques, and establishing green prevention and control measures for pests and diseases. These actions reduce agricultural carbon emissions by preventing the overuse of chemical pesticides and fertilizers. We therefore present Hypothesis 1, which will be examined in the coming sections.

Hypothesis 1. 

Land transfer and agricultural carbon emissions are negatively correlated.

2.2. The Mechanism Underlying the “Moderating Effect” of Land Transfer on Agricultural Carbon Emissions

The impact of land transfer on carbon emissions in agriculture may be mitigated by the actions of agricultural socialized services in rural areas. First, technological innovations and their use and promotion are closely linked with agricultural socialized services [48]. Alongside promoting agricultural modernization and green development [49] and lowering agricultural carbon emissions after land transfer, implementing cutting-edge green and environmentally friendly agricultural technologies can dramatically reduce the use of pesticides, fertilizers, and other harmful inputs [49,50]. Also necessary are policies encouraging and ensuring the growth of agricultural socialized services [51]. The government’s policy environment in land circulation policies, agricultural green transformation, agricultural technology, and other aspects has an important impact on land circulation and agricultural carbon emissions [52]. Consequently, we posit Hypothesis 2.

Hypothesis 2. 

Agricultural socialized services can regulate and enhance the inhibitory effect of land transfer on agricultural carbon emissions.

2.3. The Mechanism of the “Spatial Effect” of Land Transfer on Agricultural Carbon Emissions

“Tobler’s First Law of Geography” postulates that any two objects have a spatial dependency [53], and the degree of this dependency is determined by the distance between two points on Earth. Based on this, cross-regional land transfer can easily facilitate the transfer of land from low-efficiency farmers to large-scale and highly efficient agricultural operators [54]. This may lead to the integration of land resources, optimized allocation of regional resources, and an eventual reduction in agricultural carbon emissions. Additionally, land transfer may be accompanied by a spillover effect (through which geographical and spatial limits are overcome via knowledge or technology spillovers [55,56]), enabling the efficient use of resources and reduced regional agricultural carbon emissions. Drawing on this theoretical analysis, we propose Hypothesis 3.

Hypothesis 3. 

Land transfer has a significant spatial spillover effect on agricultural carbon emissions.

3. Materials and Methods

3.1. Model Design

3.1.1. Benchmark Regression Model

In this study, we created a fixed-effects panel model that can be used in regression analysis to examine the ways in which land mobility affects agricultural carbon emissions. The model may be described as follows:

$$AG{C}_{it}={\alpha }_0+{\alpha }_{1}L{T}_{it}+{\beta }_c\sum _{c=5}{X}_{it}+{\mu }_i+{\lambda }_t+{\epsilon }_{it}$$

(1)

where AGC represents agricultural carbon emissions, and LT represents rural land transfer and is a series of control variables, including industrial structure (ST), agricultural mechanization level (ML), transportation infrastructure (TR), planting structure (PL), and financial support for agriculture (FS). i represents the province/region, t represents the year, ${\alpha }_0$ is the constant term, and ${\alpha }_1$ and ${\beta }_c$ are the estimated parameters. The province’s fixed influence is denoted by ${\mu }_i$, the time effect is denoted by ${\lambda }_t$, and the random disturbance term is denoted by ${\epsilon }_{it}$.

3.1.2. Moderation Effect Model

We also constructed a method of modeling the moderating effect of agricultural socialized services on the relationship between rural land transfer and agricultural carbon emissions, as shown in Equations (2) and (3):

$$AG{C}_{it}={\phi }_0+{\phi }_{1}L{T}_{it}+{\phi }_2{SE}_{it}+{\nu }_c\sum _{c=5}{X}_{it}+{\mu }_i+{\lambda }_t+{\epsilon }_{it}$$

(2)

$$AG{C}_{it}={\phi }_0+{\phi }_{1}L{T}_{it}+{\phi }_2{SE}_{it}+{\phi }_3{L{T}_{it}\times SE}_{it}+{\nu }_c\sum _{c=5}{X}_{it}+{\mu }_i+{\lambda }_t+{\epsilon }_{it}$$

(3)

where SE represents agricultural socialized services, which is the moderating variable in this study.

3.1.3. Spatial Econometric Model

Spatial effects must be accounted for before a spatial correlation study on land transfer and agricultural carbon emissions can be performed. This is known as spatial autocorrelation. Equation (4) illustrates how one might test for spatial effects among variables using the global Moran’s I index.

$$I={\displaystyle \frac{{\sum }_{i=1}^n{\sum }_{j=1}^n{W}_{ij}(X_i-\overline{X})(X_j-\overline{X})}{{\sum }_{i=1}^n{\sum }_{j=l}^n{W}_{ij}{\sum }_{i=1}^n{(X_i-\overline{X})}^2}}$$

(4)

In Formula (4), n represents the 30 provincial regions; ${ X}_i$ and $X_j$ indicate the attribute values of the random variable X in distinct spatial units (i, j); the sample attribute values of 30 geographical units are represented by the average value; $\overline{X}$ is the spatial weight matrix; and the global Moran’s I index is indicated by $W_{ij}$.

The spatial Durbin, spatial lag, and spatial error models are the three main categories of geographic econometric models. The spatial Durbin model can be transformed into other models under certain circumstances. This research built upon the spatial Durbin model to better examine the spatial impact of land transfer on agricultural carbon emissions. The model is as follows:

$$AG{C}_{it}=\rho WAG{C}_{it}+{\varphi }_{1}L{T}_{it}+{\varphi }_2\sum _{c=5}{X}_{it}+{\theta }_{1}WL{T}_{it}+{\theta }_{2}WL{T}_{it}+{\mu }_i+{\lambda }_i+{\epsilon }_{it}$$

(5)

In Equation (5), W represents the spatial weight matrix, the spatial autoregressive coefficient is denoted by $\rho$ and the calculated coefficients are denoted by ${\varphi }_1$, ${ \varphi }_2$, ${\theta }_1$, and ${\theta }_2$.

3.2. Variable Selection

3.2.1. Explained Variable

In this study, we focused on narrow-sense agriculture, specifically in the form of crop farming, in order to calculate agricultural carbon emissions (AGCs). We computed the carbon emissions resulting from the input of agricultural materials within crop production using an Intergovernmental Panel on Climate Change (IPCC)-approved technique. The calculation proceeded as follows:

$$AG{C}_{it}=\sum {AGC}_K=H_K\times G_K$$

(6)

In Equation (6), ${\mathrm{C}\mathrm{a}\mathrm{r}\mathrm{b}\mathrm{o}\mathrm{n}}_{\mathrm{k}}$ represents total agricultural carbon emissions,${\mathrm{H}}_{\mathrm{k}}$ represents carbon emissions from various agricultural materials, ${\mathrm{G}}_{\mathrm{k}}$ represents the quantity of various agricultural materials, and $\sum {\mathrm{A}\mathrm{G}\mathrm{C}}_{\mathrm{k}}$ represents the carbon emission coefficients for various types of agricultural materials. Their values are presented in

Table 1. Carbon emission coefficients for agricultural materials.

Source of Agricultural Carbon Emissions	Emission Coefficient
Agricultural Fertilizer	0.8956 kg/kg
Agricultural Pesticides	4.9341 kg/kg
Agricultural Plastic Mulch	5.1800 kg/kg
Agricultural Diesel	0.5927 kg/kg
Agricultural Irrigation Area	25.0000 kg/m²
Agricultural Sown Area	3.1260 kg/m²
Agricultural Fertilizer	0.8956 kg/kg

.

3.2.2. Core Explanatory Variable

In China, rural land operates under the system of household contract responsibility, within which farmers, as contractors, legally enjoy the right to manage contracted land. Land transfer typically refers to the circulation of this right, i.e., farmers’ ability to transfer land use rights (i.e., management rights) to other farmers or economic organizations for a certain period of time while maintaining unchanged land ownership and contracted rights. Such transfers can be accomplished through various means, such as subcontracting, leasing, swapping, transferring, and shareholding cooperation. This practice aims to enhance land use efficiency, increase farmers’ income, and promote large-scale and intensive agricultural operations. To assess the extent and spatial coverage of land transfer, this paper uses the ratio of the transferred area (including transferred and swapped areas) of contracted management rights to the total area of household contracted land in each province (i.e., the autonomous region or municipality directly under the central government) as an indicator, on the grounds that it accurately reflects the actual land transfer context of that region.

3.2.3. Moderating Variable

We measured the extent of agricultural socialized services (SEs) by calculating the output value of the agricultural service industry per unit sown area, drawing on an existing study by Zhang (2024) [57].

3.2.4. Control Variables

In this study, we accounted for the impacts of other factors on agricultural carbon emissions to increase the accuracy of our research findings. In theory, agricultural carbon emissions are intimately tied to industrial structure, the degree of agricultural mechanization, transportation infrastructure, cropping structure, and financial assistance given to the agricultural sector. Therefore, the control variables in this paper included (1) industrial structure (ST), determined by dividing the overall value of the agricultural output by the total value of forestry, fisheries, animal husbandry, and sideline activity; (2) the level of agricultural mechanization (ML), measured via the total power of agricultural machinery (in ten thousand kilowatt-hours); (3) transportation infrastructure (TR), measured using the ratio of the rural road mileage to the rural population (km/10,000 persons); (4) the cropping structure (PL), determined by dividing the area devoted to grain sowing by the total area of crops; and (5) financial support for agriculture (FS), determined by the percentage of the general public budget allocated to forestry, water, and agriculture.

3.3. Data Sources and Descriptive Statistics

To maintain data continuity and validity, Hong Kong, Macao, Taiwan, and Tibet were omitted from the sample selection in this study due to severe data gaps. As a result, this article focused on panel data from 30 provinces in Mainland China between 2010 and 2022, providing an in-depth empirical analysis. The China Agriculture Yearbook, China Macro Database, and Agriculture and Forestry Database within the Electronic Port Supervision (EPS) Data Platform were the primary data sources, alongside authoritative materials such as the China Rural Statistical Yearbook, China Agricultural Machinery Industry Yearbook, the China Agriculture and Forestry Management and Operation Annual Report, and various local statistical yearbooks. We utilized linear interpolation as a viable supplement to ensure data integrity where individual missing data were present. Logarithmic transformations of some critical variables were used to improve data stationarity and effectively minimize concerns regarding heteroskedasticity.

Table 2. Variables and their descriptive statistics.

Variable	Number of Observations	Mean	Standard Deviation
lnAGC	390	5.4693	1.0352
LT	390	0.3603	0.2444
SE	390	390	0.8313
lnST	390	−0.6459	0.1610
lnML	390	7.6818	1.1204
lnTR	390	4.4370	0.4316
lnPL	390	−0.4548	0.2250
lnFS	390	−2.227	0.3338

summarizes descriptive statistics for all variables.

4. Results and Discussion

4.1. Analysis of Spatio-Temporal Evolution

Figure 1 depicts the evolution in the spatio-temporal dynamics of agricultural carbon emissions at the provincial level in China in 2010, 2014, 2018, and 2022, categorizing the emissions into four tiers: low, lower, higher, and high. In terms of temporal evolution, in 2010, eight provinces had low agricultural carbon emissions, eleven had lower levels, eight had higher levels, and three had high levels. By 2022, this distribution had slightly shifted, with the number of provinces in the low-level tier dropping to six, the lower-level tier expanding to nine, the higher-level tier rising to twelve, and the high-level tier remaining at three. This chart demonstrates that overall agricultural carbon emissions in China were largely steady from 2010 to 2022. In terms of spatial distribution patterns, agricultural provinces in central and eastern China—such as Henan, Shandong, and Hebei—have largely produced higher agricultural carbon emissions. This aspect of the spatial distribution has grown more apparent over time, indicating a degree of agglomeration in agricultural carbon emissions rather than isolated distribution. This conclusion has implications for our collective understanding of the regional characteristics of agricultural carbon emissions and their dynamic evolutionary patterns.

4.2. Benchmark Regression and Robustness Tests

Before conducting the empirical analysis, we employed the variance inflation factor (VIF) test to identify and exclude potential collinear variables and thus ensure the model’s robustness and reliability.

Table 3. Results of the variable multicollinearity test.

Variable	VIF	1/VIF
LT	1.12	0.891
ST	1.13	0.882
lnML	1.73	0.579
lnTR	1.42	0.702
lnPL	1.33	0.752
lnFS	1.88	0.531
Mean VIF	1.44

shows the comprehensive findings of the multicollinearity test, among which the variance inflation factor (VIF) values for all variables rigidly remained below three, meaning they were substantially lower than the commonly accepted threshold of multicollinearity difficulties. This finding indicated no significant multicollinearity among the variables studied, providing a sound foundation for a further in-depth empirical examination.

The benchmark regression findings in this work are shown in

Table 4. Results of robustness tests and benchmark regression.

Variable	(1)	(2)	(3)	(4)	(5)
	FE	FE	2011–2021	Change Model	Adjusted Region
LT	−0.205 ***	−0.154 ***	−0.096 *	−0.049 *	−0.054 *
	(0.072)	(0.057)	(0.054)	(0.029)	(0.030)
lnST		−0.099	−0.071	0.110	0.124
		(0.080)	(0.078)	(0.107)	(0.110)
lnML		0.287 ***	0.247 ***	0.505 ***	0.505 ***
		(0.028)	(0.027)	(0.032)	(0.033)
lnTR		0.379 ***	0.262 ***	−0.057 *	−0.063 *
		(0.054)	(0.055)	(0.032)	(0.032)
lnPL		−0.002	0.006	0.168	0.234 **
		(0.081)	(0.078)	(0.104)	(0.109)
lnFS		0.100 **	0.046	0.057	0.044
		(0.041)	(0.039)	(0.053)	(0.055)
Constant	5.480 ***	1.878 ***	2.560 ***	2.132 ***	2.145 ***
	(0.021)	(0.293)	(0.294)	(0.328)	(0.338)
Region	YES	YES	YES	NO	YES
Year	YES	YES	YES	NO	YES
Observations	390	390	330	390	377
Number of ids	30	30	30	30	29
R-squared	0.331	0.610	0.594	0.260	0.245

Note: values included in parentheses are t-values; significance levels of 1%, 5%, and 10% are indicated by ***, **, and *, respectively.

. Columns (1)–(2) demonstrate that the estimated coefficient of the primary explanatory variable—land transfer—stayed negative and significant at the 1% level after the control variables were introduced. This indicated a significant negative correlation between land transfer and agricultural carbon emissions, supporting Hypothesis 1. The possible reasons for this finding are various. Firstly, rural land circulation fosters large-scale farming, which, through the adoption of advanced agricultural technologies and equipment, including precision agriculture technology and water-saving irrigation systems, effectively minimizes the consumption of resources and the production of carbon emissions during agricultural practices. Secondly, as the scale of rural land circulation expands, the integration of modern management models and production methods like crop rotation, fallow farming, and organic farming practices will lead to a decrease in the usage of fertilizers and pesticides, thereby further mitigating the impact of carbon emissions resulting from agricultural processes. This analysis is consistent with results in the existing literature [58,59].

As indicated in Columns (3)–(5) of Table 4, we performed robustness tests by lowering the sample size, altering the model, and modifying the studied area to confirm the model’s resilience. Column (3) presents the regression results after reducing the sample size. Given that the scale of land transfer is continually growing, the time period was adjusted to 2011–2021 to avoid the influence of extreme values. Column (4) shows the regression results after changing the model. The fixed-effects model was replaced with a random-effects model to avoid estimation errors caused by model singularity. Column (5) displays the regression results after adjusting the regions. The samples from Xinjiang were excluded given the significant differences in geographical location, climatic conditions, local customs, and culture between Xinjiang and other provinces in China, all of which substantially impact land transfer. The robustness tests still revealed a significant negative relationship between land transfer and agricultural carbon emissions, validating the benchmark regression model’s strong robustness.

4.3. Heterogeneity Analysis Based on Functional Areas

Given the significant geographical inequalities in China’s grain production, land transfer techniques in major and non-major grain-producing areas differed, consequently influencing the dynamic variations in agricultural carbon emissions. This necessitated a thorough investigation of the specific effects of land transfer on agricultural carbon emissions, considering functional positioning changes among regions. China is divided into seventeen non-major grain-producing provinces and thirteen major grain-producing provinces. The major grain-producing regions can be further divided into six provinces in the Yangtze River Basin (Anhui, Hubei, Hunan, Jiangsu, Jiangxi, and Sichuan), four provinces in the Yellow River Basin (Shandong, Hebei, and Henan, and Inner Mongolia), and three provinces in the Songhua River Basin (Heilongjiang, Jilin, and Liaoning). We conducted a group study on several river basins and important grain production areas using regression analysis, aiming to reveal the heterogeneity of land transfer on agricultural carbon emissions.

Table 5. Heterogeneity analysis based on differences in functional positioning.

Variable	(1)	(2)	(3)	(4)	(5)
	Yangtze River Basin	Yellow River Basin	Songhua River Basin Areas	Major Grain-Producing Areas	Non-Major Grain-Producing Areas
LT	−0.043	−0.232 *	−0.100	−0.311 **	−0.141
	(0.067)	(0.132)	(0.103)	(0.147)	(0.093)
lnST	−0.202 *	0.312	0.123	0.156 *	−0.408 ***
	(0.119)	(0.297)	(0.142)	(0.079)	(0.127)
lnML	0.017	0.657 ***	0.353 ***	0.172 ***	0.325 ***
	(0.034)	(0.109)	(0.117)	(0.029)	(0.045)
lnTR	−0.020	0.119	−0.282	0.001	0.486 ***
	(0.079)	(0.193)	(0.215)	(0.068)	(0.080)
lnPL	−0.456 **	0.621	1.089	−0.064	0.065
	(0.175)	(0.417)	(0.680)	(0.058)	(0.124)
lnFS	0.076	0.319 **	0.297 *	0.248 ***	0.026
	(0.058)	(0.118)	(0.146)	(0.046)	(0.060)
Constant	5.848 ***	1.234	5.146 ***	5.239 ***	0.463
	(0.531)	(1.161)	(1.736)	(0.434)	(0.403)
Region	YES	YES	YES	YES	YES
Year	YES	YES	YES	YES	YES
Observations	78	52	39	169	221
Number of ids	6	4	3	13	17
R-squared	0.842	0.912	0.951	0.768	0.629

Note: T-values are shown in parentheses; significance levels of 1%, 5%, and 10% are represented by ***, **, and *, respectively.

shows the detailed results of this regression analysis.

The findings of the functional region-based heterogeneity regression analysis are presented in Table 5. Columns (1) and (2) show that both large and non-major grain-producing areas negatively affected the coefficient of the key explanatory variable, land transfer. Land transfer did not pass the significance test in non-major grain-producing areas but did pass in large grain-producing areas. This research highlights the potential of land transfer to lower agricultural carbon emissions while indicating the differences in its impact on various functional zones. The reason could be that the larger scale of agricultural production and greater agricultural intensification often characterizing major grain-producing regions allow them to use land transfer most effectively to reduce agricultural carbon emissions. Meanwhile, the more dispersed agricultural production patterns in minor grain-producing regions may constrain the considerable impact of land rotation in reducing carbon emissions. A further regression analysis within major grain-producing areas demonstrated that land transfer significantly negatively impacts agricultural carbon emissions in the Yellow River Basin, as shown in Columns (3)–(5). By contrast, the regression results for the Yangtze River Basin and Songhua River Basin did not pass the significance test. The Yellow River Basin’s effective reduction in agricultural carbon emissions is primarily due to its unique model of agricultural production, land use structure, support for lowering emissions via policy, and protection of its ecological environment. The region primarily practices dry farming, which, compared to the cultivation of rice in the Yangtze and Songhua River Basins, requires less irrigation, thereby reducing water use and methane emissions from wetlands and giving the area an advantage in lowering carbon emissions. Additionally, the relatively flat terrain of the Yellow River Basin is suitable for large-scale mechanized farming, the format of which enhances production efficiency and reduces carbon emissions per unit of output. Supported by national policies, the region has actively promoted low-carbon agricultural technologies (such as conservation tillage, straw incorporation, and water-saving irrigation), effectively reduced the use of fertilizers and pesticides, and reduced its consumption of energy, thereby furthering the mission to lower carbon emissions.

4.4. Analysis of the Moderating Effect

Our investigation into the moderating effect of land transfer on agricultural carbon emissions is presented in

Table 6. Regression results of the moderating effect.

Variable	All of China		Grain-Producing Areas		Non-Major Grain-Producing Areas
	(1)	(2)	(3)	(4)	(5)	(6)
LT	−0.178 ***	−0.134 **	−0.904 ***	−0.411 ***	−0.177 **	−0.132
	(0.054)	(0.061)	(0.122)	(0.062)	(0.088)	(0.093)
SE	−0.146 ***	−0.126 ***	−0.082 *	−0.393 ***	−0.134 ***	−0.112 ***
	(0.022)	(0.025)	(0.050)	(0.073)	(0.026)	(0.031)
LT&SE		−0.454 ***		−0.193 **		−0.051
		(0.161)		(0.097)		(0.037)
Control variables	YES	YES	YES	YES	YES	YES
Constant	1.767 ***	1.823 ***	5.148 ***	4.852 ***	0.298	0.372
	(0.277)	(0.279)	(0.369)	(0.377)	(0.380)	(0.383)
Region	YES	YES	YES	YES	YES	YES
Year	YES	YES	YES	YES	YES	YES
Observations	390	390	169	169	221	221
Number of ids	30	30	13	13	17	17
R-squared	0.654	0.657	0.834	0.832	0.674	0.678

Note: T-values are shown in parentheses; significance levels of 1%, 5%, and 10% are represented by ***, **, and *, respectively.

. Columns (1) and (2) of Table 6 display the nationwide regression results, indicating that land transfer and agricultural socialization services significantly inhibit agricultural carbon emissions, thereby confirming their effectiveness in reducing agricultural carbon emissions. Additionally, the coefficient of the interaction term between agricultural socialization services and rural land transfer was highly negative, at −0.454. This result shows that land transfer negatively impacts agricultural carbon emissions, which increase with the extent of agricultural socialization. This suggests that agricultural socialization services significantly moderate the effect of rural land transfer on agricultural carbon emissions, which supports Hypothesis 2. The regression results for major and non-major grain producing regions are presented in Columns (3) through (6) of Table 6, respectively. In the main grain-producing areas, the coefficient of the interaction term between rural land transfer and agricultural socialized services is significantly negative, indicating that agricultural socialized services have a pronounced moderating effect on suppressing agricultural carbon emissions through rural land transfer. Specifically, as the level of agricultural socialized services improves, the suppressive effect of rural land transfer on agricultural carbon emission intensity intensifies. Conversely, in non-grain-producing areas, the interaction coefficient between rural land transfer and agricultural socialized services is not statistically significant, suggesting that the moderating effect of agricultural socialized services on reducing carbon emissions through land transfer has not been fully realized. In addition to examining the direct influence of land transfer on agricultural carbon emissions relative to previous research [60,61], this study also demonstrates the inherent role that agricultural socialization plays in moderating land transfer’s influence on agricultural carbon emissions. This contribution broadens the field’s scope of investigation by offering fresh viewpoints and theoretical insights that will form the basis of future studies.

4.5. Spatial Effects

4.5.1. Spatial Autocorrelation Test

Table 7. Results of spatial autocorrelation tests for various variables from 2010 to 2022.

Year	Agricultural Carbon Emissions		Land Transfer
	Moran’s I	p-Value	Moran’s I	p-Value
2010	0.203	0.048	0.255	0.014
2011	0.193	0.059	0.250	0.017
2012	0.183	0.072	0.270	0.011
2013	0.174	0.085	0.303	0.005
2014	0.151	0.026	0.366	0.001
2015	0.141	0.047	0.355	0.001
2016	0.138	0.054	0.431	0.000
2017	0.135	0.063	0.388	0.001
2018	0.134	0.166	0.599	0.000
2019	0.128	0.080	0.411	0.000
2020	0.135	0.062	0.410	0.000
2021	0.131	0.074	0.341	0.002
2022	0.137	0.059	0.289	0.008

comprehensively presents the statistical test results of the global Moran’s I index of agricultural carbon emissions and rural land transfer. The results indicate that the Moran’s I value for agricultural carbon emissions was significantly greater than zero, suggesting a relatively strong spatial autocorrelation. Rural land transfer showed a clear positive spatial autocorrelation, as indicated by all Moran’s I index values for the primary independent variable—land transfer—being positive at least at the 5% significance level. As such, this research used spatial econometric models as analytical tools to investigate the complex geographical connections and consequences of rural land transfer and agricultural carbon emissions, aiming to offset potential estimation biases caused by standard models that fail to completely account for spatial variables and improve the precision and scientific rigor of our research findings.

4.5.2. Spatial Econometric Model Validation

In this work, the spatial Durbin model (SDM) was used as a research instrument to investigate the spatial impacts of land transfer on agricultural carbon emissions. The model was systematically validated to ensure the rationality and scientific rigor of the model’s selection. The findings are shown in

Table 8. Results of spatial econometric model validation.

Test Method	0-1 Matrix		Economic Geography Weight		Economic Matrix
	Statistic	p-Value	Statistic	p-Value	Statistic	p-Value
LM-error	10.818	0.001	0.890	0.019	0.970	0.001
Robust LM-error	18.594	0.000	12.114	0.001	3.014	0.083
LM-lag	0.495	0.482	48.093	0.000	8.743	0.003
Robust LM-lag	8.271	0.004	60.187	0.000	11.756	0.000
LR-SDM-SAR	69.91	0.000	106.72	0.000	112.23	0.000
LR-SDM-SEM	120.96	0000	125.81	0000	127.89	0000
Hausman	104.50	0000	104.50	0000	57.52	0000

. Both the LM-error and LM-lag statistics reached significance levels when 0–1 contiguity, economic geography weight, and economic distance matrices were used as spatial weight matrices, providing strong evidence for SDM’s applicability as the model of choice in this study. Furthermore, the LR statistic was significant at the 1% level, demonstrating the SDM’s superiority over the spatial error model (SEM) or the spatial lag model (SLM). This indicates that the SDM does not degenerate into any of these simplified models, thereby reinforcing the rationale and reliability of selecting the SDM. Additionally, the Hausman test results justify the choice of the dual-fixed effects model, further validating the model specification in this study. As a result, the spatial Durbin model with dual fixed effects was deemed adequate and reasonable for this study.

4.5.3. Analysis of Regression Test Results

Table 9. Analysis of the spatial Durbin regression results.

Variable	(1)	(2)	(3)
	0–1 Matrix	Economic Geography Weight	Economic Matrix
LT	−0.114 **	−0.090 *	−0.120 ***
	(0.048)	(0.049)	(0.047)
W × LT	−0.137 *	−0.168 *	−1.013 ***
	(0.083)	(0.089)	(0.200)
Control	YES	YES	YES
Region	YES	YES	YES
Year	YES	YES	YES
Rho	0.384 ***	0.123	0.179 **
	(0.062)	(0.077)	(0.091)
Sigma2_e	0.003 ***	0.003 ***	0.003 ***
	(0.000)	(0.000)	(0.000)
Observations	390	390	390
Number of id	0.225	0.225	0.225
R-squared	30	30	30

Note: T-values are shown in parentheses; significance levels of 1%, 5%, and 10% are represented by ***, **, and *, respectively.

presents the findings of the SDM regression analysis of the ways in which rural land transfer affects agricultural carbon emissions, assessed via contiguity, economic geography, and economic matrix conditions. The estimated coefficients of the primary explanatory variable, agricultural land transfer, were negative under all three matrix conditions, suggesting the model’s resilience.

4.5.4. Spatial Effect Decomposition

Table 10. Decomposition effect of the spatial Durbin model.

Variable	Direct Effect	Indirect Effect	Total Effect
LT	−0.023	−0.212 *	−0.235 *
	(0.048)	(0.123)	(0.136)
Control	YES	YES	YES
Region	YES	YES	YES
Year	YES	YES	YES
Observations	390	390	390
Number of id	0.225	0.225	0.225
R-squared	30	30	30

Note: T-values are shown in parentheses; significance levels of 10% are represented by *, respectively.

illustrates how the regional spillover effects of rural land transfer on agricultural carbon emissions can be separated into three categories: direct, indirect, and total effects.

Table 10 demonstrates that the total effect of rural land transfer on agricultural carbon emissions is negative. Land transfer has a direct negative influence on agricultural carbon emissions, that is, it reduces agricultural carbon emissions. Land transfer exhibits an indirect but large spatial spillover effect on agricultural carbon emissions; its indirect effect on agricultural carbon emissions is significantly negative, meaning it can act to curb agricultural carbon emissions in the immediate vicinity and in neighboring areas. There may be a number of reasons for this: Firstly, land transfer consolidates fragmented small plots into large-scale farmland, creating favorable conditions for the adoption of modern agricultural technologies and mechanized operations. This scale-up increases the efficiency of agricultural production while reducing reliance on fertilizers, pesticides, and energy, thereby lowering carbon emissions. This effect is not only evident on transferred land but also spreads to surrounding areas through the diffusion of technology and expertise, leading to an overall reduction in carbon emissions. Secondly, land transfer drives the adjustment and optimization of the agricultural industrial structure, gradually phasing out inefficient and high-emission traditional farming methods in favor of efficient, low-carbon agricultural models. The transferred land is increasingly used for cultivating economically valuable crops or forests with a strong carbon sequestration capacity, thereby enhancing the agricultural system’s absorption of carbon and reducing net carbon emissions. This form of optimization, similarly, is not limited to the transferred land but also extends to neighboring areas through demonstrative effects, further contributing to the overall lowering of agricultural carbon emissions in the wider region.

5. Conclusions and Policy Recommendations

5.1. Research Conclusions

Rural land transfer is a significant transformation in the development of China’s agriculture, while agricultural carbon emissions pose a practical constraint on its development path. Therefore, exploring the impact of rural land circulation on agricultural carbon emissions holds great practical significance.In this paper, we developed a comprehensive analytical framework within which to study the impact of land transfer on agricultural carbon emissions, thoroughly investigating the direct effect moderating effects, and spatial spillover effects. Using this theoretical framework, we conducted a comprehensive empirical analysis of panel data at the provincial level in China from 2010 to 2022. Our main conclusions are as follows: (1) Rural land transfer is negatively correlated with agricultural carbon emissions; this finding remained valid after a series of robustness tests. (2) Nationwide, the impact of rural land transfer in lowering agricultural carbon emissions increases with the quality of agricultural social services. Agricultural social services have a moderating influence, in key grain-producing areas, on the lowering of agricultural carbon emissions via land transfer; nonetheless, this moderating effect is negligible in non-major grain-producing areas. (3) Land transfer has a significant spatial spillover effect on agricultural carbon emissions, as reflected by its significant direct impact on local agricultural carbon emissions.

5.2. Policy Recommendations

5.2.1. Improve the Rural Land Transfer System

Firstly, the land transfer market mechanism should be optimized in accordance with China’s contemporary context. Local governments should be encouraged to create and explore novel land transfer models that can ensure the rational flow and optimal allocation of land resources. This will help to broaden opportunities for agricultural development and promote the concurrent progress of land transfer and green, low-carbon agricultural development. Secondly, guidance regarding rural land transfer prices should be established and refined and dynamic management practices should be implemented. This will require the statistical tracking, monitoring, and analysis of land transfer prices to be strengthened if the healthy and orderly development of the rural land transfer market is to be maintained. Lastly, clear implementation plans and steps should be developed, including the creation of a mechanism for adjusting land transfer prices, as well as regularly evaluating and revising land transfer policies to enhance their feasibility and practical impact.

5.2.2. Enhance Agricultural Socialized Services

Firstly, support in the form of policy and the allocation of resources should be used to encourage the development of social service organizations—such as agricultural cooperatives, family farms, and agricultural service companies—to establish a diversified and multi-tiered service supply framework that fully supports land transfer and agricultural production activities. Secondly, technologies such as water-saving irrigation, soil testing with formulated fertilization, and green pest management should be promoted to curb the excessive use of fertilizers and pesticides, thereby reducing agricultural carbon emissions. To ensure the practical implementation of these measures, detailed action plans must be developed, including the establishment of professional teams for the dissemination of technology and the introduction of policies incentivizing farmers’ participation in agricultural social services. Lastly, a performance evaluation mechanism should be instituted to regularly assess and enhance service quality, thereby facilitating the continuous improvement of the agricultural social service system and contributing to sustainable agricultural development.

5.2.3. Promote Coordinated Regional Development

Firstly, regular joint meetings, information sharing on emission reductions, and coordinated enforcement actions should be implemented to enhance communication and collaboration between regions, thereby jointly addressing agricultural carbon emission issues. It is essential that differentiated carbon reduction policies are tailored to the specific natural conditions, economic development levels, and agricultural industrial structures of various regions. In areas with a high carbon emission intensity, stricter reduction measures should be adopted, accompanied by technical support and financial investment, to effectively reduce emissions. In regions that have already achieved significant results in lowering emissions, the government should encourage ongoing exploration and innovation in emission reduction methods and promote successful practices in other areas. Secondly, to ensure the feasibility of these policies, detailed implementation plans should be developed, including the establishment of cross-regional cooperation mechanisms, the creation of specialized emission reduction task forces, and regular assessments of policies’ efficacy.

5.3. Limitations and Future Prospects

Since the provinces studied actually placed spatial and temporal restrictions on the data used in this analysis, it was not possible to completely capture the complex relationship between land transfer and agricultural carbon emissions on a larger geographic scale.

Future research should concentrate on several key areas. First, future research should explore how land transfer alters farmers’ cultivation practices, adoption of technology, and allocation of resources and consequently influences agricultural carbon emissions. By collecting micro-level survey data and constructing behavioral models, researchers can more accurately assess the actual impact of changes in farming practices on carbon emissions, providing more targeted guidance for policy making. Second, future research should focus on constructing time series data spanning longer periods and combining them with dynamic tracking analyses of land transfer and agricultural carbon emissions to uncover the time-lag effects of and long-term trends in these two factors. This approach will help to foster a more comprehensive understanding of the environmental effects of land transfer and provide empirical support for sustainable land use policies.