Impact of Farmers’ Livelihoods on Agricultural Carbon Emission Efficiency Under the Background of Population Urbanization: Evidence from China

Abstract

In the context of China’s population urbanization, the quality and pattern of farmers’ livelihoods are undergoing significant changes. Reducing emissions and sequestering carbon in agriculture is a crucial pathway for China to achieve its “dual carbon” goals. How to balance low-carbon agricultural development with the changing livelihood patterns of farmers has become an important issue in China’s agricultural and rural development. This study examines the impact of farmers’ livelihoods on agricultural carbon emission efficiency and explores regional disparities based on panel data from 31 provinces and municipalities in China from 2002 to 2020. The findings are as follows: (1) The quality of farmers’ livelihoods in China is conducive to an improvement in agricultural carbon emission efficiency (ACEE). (2) From a national perspective, the higher the livelihood of non-farm employment in the region, the higher the agricultural carbon emission efficiency. The livelihood of agricultural production has a significantly negative correlation with agricultural carbon emission efficiency. (3) Improvement in the quality of farmers’ livelihoods (QFL) in the eastern and western regions has a significant positive impact on the efficiency of agricultural carbon emissions, and the impact is larger in the western region, while there is no statistically significant relationship in the central region. The reason for this difference may be that the migration of agricultural labor from the western region to the eastern region and the local urban employment of eastern farmers have led to this, while the migration and local urban employment of agricultural labor in the central region is relatively limited. This paper provides policy insights into promoting both farmers’ income growth and low-carbon agricultural production in a coordinated manner.

1. Introduction

Global warming, along with the accelerated melting of glaciers and permafrost due to carbon emissions, has gained significant attention in recent years [1,2]. In response to climate change, nations worldwide have implemented various strategies, including “Total Quantity Control and Carbon Emissions Trading Systems” [3,4]. As one of the largest carbon emitters, China accounts for approximately 27% of global emissions [5]. The Chinese government has actively promoted “Carbon Peaking” and “Carbon Neutrality” initiatives, committing to rigorous emission reductions [6]. Agriculture remains a major global source of greenhouse gas emissions, responsible for over 30% of total anthropogenic greenhouse gases, with around 17% of China’s carbon emissions stemming from this sector [7]. China’s agricultural carbon emissions grow at an annual rate of 4.08%, underscoring agriculture as a key focus in the national “Double Carbon” initiative [8].

Chinese agriculture is transitioning from traditional to modernized practices. In this modernization—characterized by increased mechanization, chemical use, and adoption of information technology—carbon emissions are inevitably generated through greater consumption of fuel and materials. The technical reliance on current agricultural practices poses challenges for rapid transformation. Therefore, pursuing solely emission reductions to develop sustainable agriculture may not align with the imperative to improve production efficiency [9,10]. This study posits that the low-carbon development pathway in Chinese agriculture should seek the maximum feasible reduction in emissions without compromising production efficiency. Improving agricultural carbon emission efficiency (ACEE) is fundamental to achieving a sustainable, low-carbon agriculture system defined by “low energy consumption, low emissions, low pollution, and high efficiency” [11].

China’s strategies for urbanization and rural revitalization are reshaping farmers’ livelihoods and narrowing the urban–rural income gap. According to the China Statistical Yearbook, the per capital disposable income of rural residents has reached CNY 18,931 in 2021, more than double that in 2012. Notably, this growth rate outpaced that of urban residents and GDP growth. Concurrently, farmers’ income structures have shifted, with wage income increasingly surpassing business income as the main income source; in 2021, wage, business, and property incomes contributed 42.0%, 34.7%, and 2.5% of total income, respectively. Against the backdrop of increasing urbanization, changes in farmers’ livelihoods hold substantial implications for low-carbon agricultural development in China.

Most research on carbon emissions focuses on urban and industrial sectors, with limited studies on the agricultural sector. Existing research on farmers’ livelihoods and agricultural carbon emission efficiency mainly focuses on the following two key dimensions. Firstly, a consensus regarding the impact of the urban–rural income disparity on carbon emissions has yet to be reached. Some scholars argue that widening income gaps may prompt rural residents to overexploit natural resources, causing environmental degradation and elevated carbon emissions [12,13,14]. In contrast, others suggest that urban residents exhibit stronger environmental awareness, emphasizing environmental conservation and emission reduction [15,16,17]. Secondly, several studies have highlighted that alterations in farmers’ income compositions can profoundly impact the utilization of chemicals, including fertilizers and pesticides, thereby potentially enhancing agricultural production efficiency via bolstering human capital and adopting advanced agricultural technologies [18,19,20,21]. However, existing research lacks an in-depth exploration of the impact of farmers’ livelihoods on ACEE.

This study addresses this gap by examining the influence of farmers’ livelihoods on agricultural carbon emission efficiency and analyzing regional disparities using panel data from 31 Chinese provinces and municipalities from 2002 to 2020. It offers several novel contributions: Firstly, this study explores the potential impact of farmers’ livelihood quality on agricultural carbon emission efficiency, which complements existing research related to farmers’ livelihoods. Secondly, from the perspective of agricultural carbon emission efficiency, this study attempts to assess the differences in the intensity of impact among different types of farmers’ livelihoods, providing a new perspective for existing research. Lastly, this study examines the differences in impact across different regions, offering targeted recommendations on how to achieve high efficiency in agricultural carbon emissions within the context of farmers’ livelihood transformation in various areas.

2. Theoretical Analysis and Research Hypothesis

2.1. Mechanism of Improved Farmer Livelihoods on ACEE Under the Trend of Urbanization

Against the backdrop of increasing urbanization in China, improvements in farmers’ livelihoods are reflected in their shift towards urban living patterns [22]. Moreover, the income gap between urban and rural areas is gradually decreasing. First, as farmers’ livelihoods improve, rural inhabitants are increasingly motivated to invest in human capital to close the gap with their urban counterparts [23]. This investment elevates the quality of the rural labor force and encourages urban residents to similarly invest in enhancing their skills to maintain their competitive edge. Collectively, this results in an overall improvement in labor force quality, advancing environmentally sustainable and efficient agricultural practices. Second, improvements in farmers’ livelihoods reduce the disparity in interests between urban and rural areas [24]. This enhancement increases rural inhabitants’ capacity to invest in agriculture, thus heightening their willingness to adopt advanced production technologies. Furthermore, it slows the migration of agricultural labor to urban centers, partially mitigating the adverse effects of labor shortages and demographic aging on agricultural productivity. Additionally, it also encourages skilled labor to return to rural areas, raising the average quality of the rural workforce. This, in turn, facilitates the spread of eco-friendly production techniques, contributing to a reduction in agricultural carbon emissions [25].

The foregoing analysis demonstrates that enhancing farmers’ livelihoods can improve agricultural carbon emission efficiency through mechanisms involving human capital investment, the adoption of advanced agricultural technologies, and the revitalization of the rural workforce [26,27]. Thus, the first hypothesis of this study is that improvements in farmers’ livelihoods can significantly enhance agricultural carbon emission efficiency in the context of urbanization.

2.2. The Relationship Between Farmers’ Diverse Livelihoods and Agricultural Carbon Emission Efficiency

China’s rural population is increasingly migrating to urban areas for employment and residence, leading to a significant transformation in farmers’ livelihoods [28]. Previously reliant solely on agricultural production, farmers now engage in a combination of non-agricultural employment, agricultural operations, and asset leasing. Non-agricultural employment refers to labor in non-agricultural sectors for wages, agricultural operations involve income generated from agricultural production, and asset leasing pertains to rental income from land and homestead leases [29,30].

The proportion of wage income serves as a metric for the level of non-agricultural employment. A higher wage income among farmers indicates substantial time allocation to non-agricultural activities, potentially including urban employment. As a result, farmers’ participation in non-agricultural employment will reduce the time available for farming activities, leading to a decrease in the use of carbon-intensive inputs like fertilizers and pesticides and consequently reducing agricultural carbon emissions [31]. Furthermore, the decreasing availability of agricultural labor increases reliance on modern production factors such as capital and technology. This reliance is expected to spur technological advancements in agriculture, enhancing crop varieties, planting management practices, and agricultural equipment [32,33]. Ultimately, farmers who participate in non-agricultural work, particularly in urban settings, are more inclined to adopt innovative technologies due to their heightened knowledge and enhanced learning capabilities. These farmers are likely to invest in advanced agricultural technologies and equipment aimed at improving production efficiency, minimizing waste and resource use, and enhancing the effectiveness of eco-friendly agricultural practices [34,35].

Accordingly, this study proposes a second hypothesis: Non-agricultural employment-based livelihoods of farmers positively affect agricultural carbon emission efficiency.

The proportion of business income serves as an indicator of the extent of agricultural operations. Farmers with a significant share of business income are likely to dedicate more time and resources to agriculture production. Those with larger farming operations often benefit from economies of scale, leading to greater production efficiency, and they are more inclined to adopt environmentally sustainable technologies. However, in their quest to maximize yields, these farmers may excessively use fossil fuel-based inputs such as fertilizers and pesticides, potentially diminishing agricultural carbon emission efficiency [36].

Therefore, this study posits a third hypothesis: Agricultural operation-based livelihoods of farmers may exert both positive and negative effects on ACEE.

Farmers’ asset leasing activities primarily refer to the transfer of land or homestead rights. Farmers who transfer their cultivated land out of their possession tend to decrease the application of agricultural inputs like pesticides and fertilizers, thereby directly curbing agricultural carbon emissions. In contrast, farmers who acquire farmland tend to adopt large-scale and intensive farming practices. On one hand, large-scale operations can enhance eco-friendly agricultural production efficiency and reduce carbon emissions; on the other hand, they increase the use of agricultural machinery. While the use of machinery can improve production efficiency, it also significantly raises fossil fuel consumption, especially in hilly regions where machinery efficiency is lower, potentially reducing ACEE [37,38,39].

Based on this, this study presents a fourth hypothesis: Asset leasing-based livelihoods of farmers may have both positive and negative impacts on ACEE.

3. Materials and Methods

3.1. Data

Since China’s official accession to the World Trade Organization (WTO) in December 2001, the national economy has experienced rapid growth. Concurrently, the urbanization process in China has accelerated significantly, with a substantial rural population gradually moving to urban areas and diversifying their sources of income [40,41,42]. In line with the global call for environmental protection and sustainable development, the Chinese government introduced the “National Ecological Environment Protection ‘Fifteenth’ Plan” in 2002, setting clear policy objectives to enhance environmental protection and promote carbon emission reduction. Given these contexts, and considering the availability and integrity of data, this study selects data from 31 provinces in China (excluding Hong Kong, Macau, and Taiwan) from 2002 to 2020 for analysis. The data are sourced from the “China Statistical Yearbook”, the “China Rural Statistical Yearbook”, and Chinese provincial statistical yearbooks [43,44], which come from the Chinese government (National Bureau of Statistics of China: https://www.stats.gov.cn/sj/ndsj/, accessed on 18 December 2024). The missing part of the data is compensated by linear fitting interpolation. In addition, all economic variables are deflated in the model, with 2002 as the base period.

3.2. The Accounting Method for Agricultural Carbon Emission Efficiency

Agricultural production efficiency measures a region’s overall agricultural production potential, providing a crucial evaluation of resource allocation and input–output effectiveness in agriculture. However, traditional metrics for assessing agricultural production efficiency often focus on inputs and outputs, overlooking the environmental impacts of agricultural practices. In reality, agricultural activities involving the consumption of energy and materials can result in unexpected consequences, such as agricultural carbon emissions. Therefore, it is essential to evaluate agricultural carbon emission efficiency (ACEE) by considering both the input–output aspects associated with agricultural production and the emissions arising from these activities.

The DEA model is often used to study efficiency problems. This method does not need to set production function and weight in advance and uses a linear programming method to measure the efficiency of multiple decision units with the same input and output [45]. However, the output index of the traditional DEA model is mostly the expected output index, which is not suitable for a situation with a non-expected output index, and the possible influence of the relaxation variable on the efficiency value is not considered. As compared to the general radial DEA model (radial BCC/CCR), the Super-SBM model takes into account the “relaxation” of the factors. It can effectively solve the problem of the deviation of agricultural carbon emission efficiency evaluation results caused by ignoring the non-expected output, and make up for the defect that the traditional DEA model cannot sort and distinguish the effective decision units [46,47]. Therefore, the Super-SBM model is employed within the framework of data envelopment analysis (DEA) to calculate the ACEE across different Chinese provinces over various years. The Super-SBM model is presented below.

$$p=min{\displaystyle \frac{\frac{1}{m}\sum _{i=1}^m\frac{\overline{X}}{X_{ik}}}{\frac{1}{r_1+r_2}\left(\sum _{s=1}^{r_1}\frac{\overline{Y^d}}{Y_{sk}^d}+\sum _{q=1}^{r_2}\frac{\overline{Y^u}}{Y_{qk}^u}\right)}}$$

(1)

$$s.t.\left\{\begin{array}{c}\sum _{\begin{array}{c}j=1\\ j\ne k\end{array}}^n{X}_{ij}{\lambda }_j\le \overline{X } \\ \sum _{\begin{array}{c}j=1\\ j\ne k\end{array}}^n{Y}_{sj}^d{\lambda }_j\ge \overline{Y^d} \\ \sum _{\begin{array}{c}j=1\\ j\ne k\end{array}}^n{Y}_{qj}^d{\lambda }_j\le \overline{Y^d}\#\\ \overline{X}\ge X_k;\overline{Y^d}\le Y_k^d;\overline{Y^u}\ge Y_k^u\\ {\lambda }_j\ge 0\\ i=1, 2,\dots , m;j=1, 2,\dots , n\left(j\ne k\right)\\ s=\mathrm{1,2},\dots ,r_1;q=\mathrm{1,2},\dots ,r_2\end{array}\right.$$

(2)

In Formulas (1) and (2), P signifies the ACEE of the province, $\overline{X}$ signifies the input slack, $\overline{Y^d}$ signifies the expected output slack, and $\overline{Y^u}$ signifies the unexpected output slack. X_ij denotes the i-th input of the j-th province, Y_sj signifies the s-th expected output of the j-th province, Y_qj denotes the q-th non-expected output of the j-th province, and λ_j represents the weight coefficient.

According to the relevant literature, agricultural carbon emissions are mainly considered carbon emissions produced by agricultural production [48,49]. The input index of agricultural carbon emission efficiency was determined from four dimensions: labor, capital, land, and resources. The output includes expected output and unexpected output. The corresponding specific indicators are shown in

Table 1. Evaluation index system of ACEE.

Index	Aspect	Variable	Unit	Mean	Standard Deviation
Input	Labor	Number of agricultural employees	Per 10,000 people	926.598	701.970
	Capital	Fixed asset investment in agriculture	CNY (Chinese Yuan) in billions	69.200	68.966
	Land	Crop planting area	1000 hectares	5167.020	3766.219
	Resources	Pesticide input	10,000 tons	5.116	4.225
		Fertilizer input	10,000 tons	173.229	140.391
		Fuel input	10,000 tons	63.355	65.039
		Agricultural plastic film usage	10,000 tons	7.021	6.468
		Water resource input	100 million cubic meters	119.629	101.112
Output	Expected output	Agricultural output value	CNY (Chinese Yuan) in billions	882.673	789.784
	Unexpected output	Agricultural carbon emissions	Dimensionless (already normalized to negative direction)	69.947	22.988

.

Labor input is measured by the number of people employed in primary industry at the end of the year. Capital input is measured by the number of fixed asset investments in agriculture. Considering the difference between multiple cropping indexes and the influence of the fallow phenomenon in some areas, the crop planting area was used to measure land input. Based on the current situation of agricultural production, pesticide input, fertilizer input, fuel input, agricultural plastic film usage, and water resource input are used to measure resource input [50,51]. The total agricultural carbon emissions are taken as the unexpected output, and the total agricultural output value is taken as the expected output [52]. According to the Intergovernmental Panel on Climate Change (IPCC) coefficient method, there are six main contributors to agricultural carbon emissions: the manufacture and use of fertilizers, pesticides, agricultural film, and agricultural machinery, as well as the release of carbon from the use of electric energy for irrigation and the loss of organic carbon to the atmosphere as a result of agricultural plowing [53]. Therefore, the primary contributors to carbon emissions are fertilizer agricultural film, pesticides, diesel, irrigation, and tillage. The total amount of agricultural carbon emissions is calculated by multiplying the usage of the above carbon emission sources by the corresponding carbon emission coefficient. The relevant carbon emission coefficients and sources are shown in

Table 2. Agricultural carbon emission source factors.

Sources	Factor	Data Sources:
Fertilizer	0.8956 kg/kg	Oak Ridge National Laboratory, United States
Agricultural film	5.18 kg/kg	Institute of Agricultural Resources and Eco-Environment, Nanjing Agricultural University
Pesticides	4.9341 kg/kg	Oak Ridge National Laboratory, United States
Diesel	0.5927 kg/kg	Intergovernmental Panel on Climate Change
Irrigation	25 kg/hm2	Intergovernmental Panel on Climate Change
Tillage	312.6 kg/km2	College of Biology and Technology, China Agricultural University

.

The precise calculation formulas are presented below.

$$P=\sum P_i\times T_i$$

(3)

In Formula (3), P denotes the agricultural carbon emissions. The unit is 10⁴ t. Pi denotes the emission coefficient of the corresponding i-type carbon source, Ti denotes the consumption of the corresponding i-type carbon source, and i refers to each of these sources of carbon emissions.

3.3. Variable Selection and Descriptive Statistics

The explained variable in this study is ACEE, calculated using the Super-SBM model, as described in Formula (1). The core explanatory variables encompass the quality of farmers’ livelihoods (QFL) and the diverse livelihoods of farmers, which are categorized into three types: livelihood of non-farm employment (LNE), livelihood of agricultural production (LAP), and livelihood of asset leasing (LAL). QFL is measured by the inverse of the gap between the per capita disposable income of urban residents and the per capita disposable income of rural residents. LNE is measured by the proportion of wage income to total income, LAP is measured by the proportion of business income to total income, and LAL is measured by the proportion of property income to total income. Control variables include several metrics: the rural population, per capita consumption by rural residents, the proportion of primary industry (measured by the ratio of primary industry output value to total output value), the degree of disaster (assessed through the ratio of crop-affected area to total sown area), and the policy of main grain-producing areas (classified as 1 for main grain-producing provinces and 0 otherwise) [8,54,55]. Descriptive statistics for these variables are presented in

Table 3. Descriptive statistics of variables.

	Name of Variable	Unit	Mean	Standard Deviation
Explained Variable	ACEE		0.286	0.323
Core explanatory variables	Quality of farmers’ livelihoods		0.098	0.056
	Livelihood of non-farm employment	%	0.261	0.263
	Livelihood of agricultural production	%	0.244	0.138
	Livelihood of asset leasing	%	0.019	0.021
Control variables	Rural population	Per 10,000 people
	Per capita consumption by rural residents	CNY	7.328	0.989
	Proportion of primary industry	%	8.609	0.707
	Degree of disaster	%	0.195	0.102
	Policy of main grain-producing areas	Virtual variables	0.214	0.151

.

3.4. Two-Way Fixed-Effects Model

The data used in this paper are a long panel of 19 years of data covering 31 provinces in China. It is necessary to control the influence of province factor and time factor on the model. The two-way fixed-effects model for panel data not only effectively controls for factors that remain constant over time or across different sections, but also mitigates the impact of multicollinearity. Consequently, this study utilizes a two-way fixed-effects model as the baseline regression model to investigate the impact of farmers’ livelihoods on ACEE. The detailed formulation of the model is provided below.

$$Y_{n,t}={\alpha }_0+{\alpha }_1\times I_{i,n,t}+\sum _2^k{\alpha }_k\times X_{k,n,t}+{\phi }_t+{\mathrm{\mu }}_n+{\epsilon }_{n,t}$$

(4)

In Formula (4), $Y_{n,t}$ represents the explained variable, signifying the ACEE of the nth province in t-year. $I_{i,n,t}$ represents the core explanatory variables (i = 1, 2, 3, 4), among which $I_{1,n,t}$ represents farmers’ livelihoods, $I_{2,n,t}$ represents non-agricultural employment-based livelihoods, $I_{3,n,t}$ represents agricultural operation-based livelihoods, and $I_{4,n,t}$ represents asset leasing-based livelihoods. $X_{k,n,t}$ represents the control variables that may affect ACEE. ${\alpha }_0$ represents constant terms, ${\alpha }_1$ represents the estimated coefficients of the core independent variables on agricultural carbon emissions, ${\alpha }_k$ represents the estimated coefficients of each control variable, ${\phi }_t$ and ${\mu }_n$ represent the year fixed effects and area fixed effects, and ${\epsilon }_{n,t}$ represents the random error terms of the formula.

4. Results

4.1. Descriptive Analysis

Figure 1 illustrates the evolution of the mean values of ACEE and QFL in China from 2002 to 2020. The ACEE in China exhibited a fluctuating upward trend, particularly noticeable post-2012, as it increased from 0.259 to 0.669 by 2020, while the QFL steadily decreased from 0.213 in 2002 to 0.043 in 2020 (Figure 1a). This suggests that, in recent years, the Chinese government has effectively reduced high-carbon energy consumption and greenhouse gas emissions throughout the agricultural production and supply chain by various means, including structural adjustments in industries, technological and institutional innovations, and the utilization of renewable energy. These efforts have notably become more evident following the 18th National Congress of the Communist Party of China in 2012. Concurrently, the disparity in living standards between rural and urban residents in China has been diminishing, indicating an alleviation of the urban–rural divide.

Regionally, the trend in ACEE in eastern China aligns closely with the national pattern (Figure 1b), which implies that the low-carbon development of agriculture in the eastern region has set an exemplary standard for the country. In this region, coastal developed areas, suburban areas of major cities, state-owned farms, and national modern agricultural demonstration zones have largely achieved agricultural modernization, transitioning from extensive growth to low-carbon sustainable development. In contrast, the trends in ACEE in the central and western regions are more volatile (Figure 1c,d), and the degree of improvement is considerably lower than in the eastern region. This suggests that agriculture in central and western provinces remains relatively extensive, constrained by socioeconomic and natural conditions, with limited innovation capacity in agricultural technology, indicating significant potential for further reductions in emissions. The trend of GURL in the eastern, central, and western regions mirrors the national trend, with a consistent decline.

The observed inverse trajectories of ACEE and QFL call for further examination to determine whether a negative causal relationship exists between them, as discussed in the subsequent sections.

4.2. Empirical Analysis

Table 4. Effects of the quality of farmers’ livelihoods on ACEE.

Name of Variable	(1)	(2)	(3)
	ACEE	ACEE	ACEE (Two-Way Fixed-Effects Model)
QFL	5.686 ***	2.387 ***	3.897 ***
	(0.853)	(0.483)	(0.786)
Rural population		−0.138 ***	0.107
		(0.0132)	(0.0970)
Per capita consumption by rural residents		0.179 ***	0.364 ***
		(0.0412)	(0.0663)
Proportion of primary industry		−0.463 ***	−1.026 ***
		(0.140)	(0.373)
Degree of disaster		−0.176 **	0.0800
		(0.0857)	(0.0813)
Policy of main grain-producing areas		−0.0542 **	−0.0144
		(0.0274)	(0.0488)
Area control	Controlled		Controlled
Year control	Controlled		Controlled
Constant	−0.492 ***	−0.330	−2.205
	(0.165)	(0.422)	(2.046)
Observations	589	589	589
R-squared	0.518	0.314	0.527

Note: ***, **, * indicate significant at 1%, 5%, 10% levels of significance, respectively, and within () are standard errors.

presents the regression results from the baseline model, revealing a positive relationship between QFL and ACEE, which is significant at the 1% level. This positive relationship persists across the first, second, and third models, with an increase in the R² value after incorporating control variables and accounting for both area and year fixed effects, reinforcing the reliability of the findings. The results suggest that the improvements in QFL in China significantly enhance ACEE, supporting the first hypothesis. The reason for this is that the improvements in QFL have encouraged the mutual flow of people and capital between urban and rural areas and the elevation of the rural labor force’s overall quality. It has also promoted the adoption of green production technologies, leading to more sustainable, scientific, and efficient agricultural practices. This result not only confirms the relevant research results but also supplements the existing research. Households’ livelihood styles trigger their production and consumption activities, which drive the utilization of various energy and materials that emit carbon emissions [56]. Some research indicates that higher household incomes are associated with higher indirect energy consumption and carbon emissions [57]. Other studies propose that rural households also have specific livelihoods (agriculture and reforestation activities) that lead to carbon sequestration, which is closely related to household carbon emissions [58]. Meanwhile, recent findings also highlight that in state-owned farms, the increase in the level of economic development can reduce carbon emissions and promote the green transformation of agriculture [53].

In terms of controlling variables, the coefficient of the proportion of primary industry is significantly negative at the 1% level. This finding is consistent with the research of by other scholars [53], which also demonstrates the suppressive effect of the proportion of primary industry on ACEE. An increase in the proportion of primary industry within the economic structure may indicate a rise in agricultural output. However, traditional agricultural practices often rely on inputs such as fertilizers, pesticides, and irrigation, all of which are associated with high carbon emissions. Without effective technological upgrades and management measures, the resulting increase in carbon emissions may exceed the benefits of output growth, leading to a decline in ACEE. The coefficient of the policy of main grain-producing areas is −0.0144 but does not pass the significance test, suggesting that its negative impact on ACEE is not statistically significant. This result is inconsistent with previous studies, such as Some research found that the policy of main grain-producing areas exerts a significant “efficiency-enhancing” effect [59]. In contrast, the coefficient of per capital consumption by rural residents is 0.364 and significant at the 1% level. This indicates that an increase in per capital consumption by rural residents positively contributes to ACEE improvement. Higher consumption levels drive the transition of agriculture from extensive production to more efficient, low-carbon practices. Additionally, rising income and upgrading consumer demand stimulate the development and adoption of low-carbon agricultural technologies, thereby significantly enhancing ACEE.

To address potential endogeneity concerns, this study utilizes the conditional mixed process (CMP) approach, using the first-order lagged term of QFL as an instrumental variable. The CMP method evaluates the homogeneity of the core explanatory variable through the endogeneity test parameter (atanhrho_12). A significant deviation of the endogeneity test parameter from zero indicates the presence of endogeneity issues in the model [60,61]. Table 4 presents the CMP estimation results, illustrating the impact of QFL on ACEE. The first-stage estimation reveals that the instrumental variables are statistically significant at the 1% level with respect to ACEE, thereby meeting the requirement for instrument relevance. The endogeneity test parameter (atanhrho_12) is not significant, suggesting the absence of apparent endogeneity issues in the model, thus confirming the consistency and reliability of the benchmark regression results. The subsequent sections of this study employ the CMP method to identify and mitigate endogeneity (

Table 5. Results of the CMP model.

Name of Variable	(4) ACEE
	First stage	Second stage
QFL		2.493 ***
		(0.487)
Instrumental variable	0.865 ***
	(0.008)
Control variable	Controlled	Controlled
Atanhrho_12	−0.053
	(0.043)
Constant term	0.012 ***	−0.405 ***
	(0.007)	(0.424)
LR	2963.36 ***

Note: ***, **, * indicate significant at 1%, 5%, 10% levels of significance, respectively, and within () are standard errors.

); however, the detailed procedural explanation will not be reiterated in the subsequent discourse.

4.3. Discussion

While the preceding analysis confirmed the significant enhancement in ACEE through the improvement in QFL, it is crucial to acknowledge that this improvement is influenced by changes in farmers’ livelihood patterns. Therefore, this section delves deeper into the complex interactions between farmers’ diverse livelihoods and ACEE.

Table 6. Effects of farmers’ diverse livelihoods on ACEE.

Name of Variable	(5)	(6)	(7)
	ACEE	ACEE	ACEE
LNE	0.293 ***
	(0.109)
LAP		−0.853 ***
		(0.273)
LAL			0.027
			(1.078)
Control variable	Controlled	Controlled	Controlled
Area control	Controlled	Controlled	Controlled
Year control	Controlled	Controlled	Controlled
Constant	0.731	2.198 *	2.015
	(1.380)	(1.264)	(1.313)
Observations	589	589	589
R-squared	0.511	0.528	0.504

Note: ***, **, * indicate significant at 1%, 5%, 10% levels of significance, respectively, and within () are standard errors.

indicates that LNE positively correlates with ACEE, whereas LAP shows a negative correlation. Both relationships are statistically significant at the 1% level. These findings suggest that, at the national level, the rapid increase in the proportion of farmers’ wage income and the relative decline in the proportion of business income are closely linked to improvements in carbon emission efficiency in China’s agriculture, in line with the second and third hypotheses. The increase in wage income and decrease in business income prompts part-time farmers to place less emphasis on agriculture, reducing material inputs and subsequent carbon emissions [62]. Concurrently, the reduction in the rural labor force facilitates capital deepening and technological advancement in agriculture [63]. Migrant farmers are acquiring more knowledge and skills, enhancing their willingness to accept capital-intensive low-carbon technologies, such as new varieties, straw recycling, soil testing, and formulated fertilization, which reduces agricultural carbon emissions. Additionally, no significant relationship exists between LAL and ACEE. This may be due to varying effects across different regions, which do not manifest as significant at the national level.

4.4. Analysis of Regional Heterogeneity

To examine whether regional variations exist in the impact of QFL and their diverse livelihood strategies on ACEE, this study categorizes Chinese provinces into eastern, central, and western regions based on the division criteria of the National Bureau of Statistics (

Table 7. Effects of farmers’ diverse livelihoods on ACEE in the eastern, central, and western regions.

Name of Variable	(9)	(10)	(11)	(12)
	QFL—ACEE	LNE—ACEE	LAP—ACEE	LAL—ACEE
Eastern region	4.047 ***	0.668 ***	−0.784 ***	0.910
	(1.271)	(0.109)	(0.272)	(1.173)
Central region	1.517	1.083	0.293	11.592 ***
	(1.392)	(0.667)	(0.223)	(1.794)
Western region	4.434 ***	2.070	−2.762 ***	−15.987 ***
	(1.422)	(1.331)	(0.477)	(4.210)
Control variable	Controlled	Controlled	Controlled	Controlled
Area control	Controlled	Controlled	Controlled	Controlled
Year control	Controlled	Controlled	Controlled	Controlled

Note: ***, **, * indicate significant at 1%, 5%, 10% levels of significance, respectively, and within () are standard errors.

). The eastern region comprises Beijing, Tianjin, Hebei, Liaoning, Shanghai, Jiangsu, Zhejiang, Fujian, Shandong, Guangdong, and Hainan. The central region comprises Shanxi, Jilin, Heilongjiang, Anhui, Jiangxi, Henan, Hubei, and Hunan. Lastly, the western region comprises Inner Mongolia, Guangxi, Chongqing, Sichuan, Guizhou, Yunnan, Tibet, Shaanxi, Gansu, Qinghai, Ningxia, and Xinjiang.

The results shown in Table 7 indicate that the improvement in QFL in the eastern and western regions has a significant positive impact on the efficiency of agricultural carbon emissions, and the impact is larger in the western region, while there is no statistically significant relationship in the central region. The reason for this difference may be that the migration of agricultural labor from the western region to the eastern region and the local urban employment of eastern farmers have led to this, while the migration and local urban employment of agricultural labor in the central region is relatively limited. The reason for this difference may be that the improvement in farmers’ lives in the western region has significantly exceeded that in the eastern and central regions in recent years [64]. The significant improvement in the living standards of farmers in the western and eastern regions has led to an improvement in labor quality and the adoption of environmentally friendly technologies, which in turn have had a significant impact on the efficiency of agricultural carbon emissions.

From the perspective of diverse livelihoods among farmers, the relationship between these livelihood patterns and ACEE in eastern China mirrors the national trend. Specifically, LNE has a significantly positive correlation with ACEE, whereas LAP shows a significant negative correlation. Over the past two decades, a considerable influx of rural residents has been attracted to urban employment opportunities in the eastern region, with the floating population consistently comprising about half of the region’s workforce. From 2002 to 2020, the proportion of farmers’ wage income in the eastern region averaged 49.53%, higher than in the central and western regions. Conversely, the proportion of farmers’ business income was 36.99%, the lowest among the three regions. As a result, rural inhabitants in the eastern region have more favorable prospects for non-agricultural employment and thus command a larger share of wage income. Additionally, the eastern region benefits from stronger agricultural subsidies, leading to a higher share of property income for farmers compared to the central and western regions.

In the central region, LAL demonstrates a significant positive correlation with ACEE. This is attributable to the central region’s role as a major grain-producing area in China, with provinces like Jilin, Heilongjiang, Anhui, Jiangxi, Henan, Hubei, and Hunan making substantial contributions to national grain output. The adoption of large-scale farming operations following land transfers has notably improved grain production efficiency, thereby establishing a strong positive correlation between LAL and ACEE. Furthermore, data from the “Seventh National Population Census” reveal that nearly 60% of the central region’s urbanization level is below the national average. This suggests that agricultural labor migration in the central region remains relatively limited. Consequently, LNE and LAP do not show significant correlations with ACEE.

In the western region, both LAP and LAL show a negative correlation with ACEE. This indicates that many rural residents in the western region have sought employment opportunities elsewhere, mainly migrating to the eastern region. As a result, LNE does not significantly correlate with ACEE. A different correlation pattern between LAL and ACEE emerges in the western region compared to the central region. This contrast indicates that land transfers and scaled operations in the western region have not yielded an improvement in ACEE. The western region of China is characterized by severe land fragmentation. After transferring land, farmers in this area often rely on mechanized farming methods. However, the operational efficiency of agricultural machinery in the hilly area remains lower than that in plains, and energy consumption is higher. Consequently, while land transfer and other asset-leasing activities increase the use of agricultural machinery, they still have a negative impact on ACEE in the western region [65].

5. Conclusions

Based on panel data from 31 provinces and municipalities in China spanning from 2002 to 2020, this study examines the impact of QFL on agricultural carbon emission efficiency, and new explorations were conducted based on existing research. This paper explores the impact of regional differences by evaluating the differences in the intensity of the impact of different farmer livelihood types on agricultural carbon emission efficiency. The results of this study can provide targeted suggestions for achieving high efficiency of agricultural carbon emissions in the context of farmer livelihood transformation. The analysis yields the following conclusions: (1) The progressive improvement in QFL across China distinctly enhances agricultural carbon emission efficiency. (2) At the national level, increased LNE correlates with higher agricultural carbon emission efficiency, whereas LAP shows the opposite effect. (3) Regionally, the improvement in QFL in the eastern and western regions has a significant positive impact on the efficiency of agricultural carbon emissions, and the impact is larger in the western region, while there is no statistically significant relationship in the central region. Building upon the aforementioned analytical outcomes, this study presents the ensuing policy implications.

(1)

The government should recognize the positive relationship between QFL and agricultural carbon emission efficiency, and through relevant policies, reduce the urban–rural income gap while simultaneously improving agricultural carbon emission efficiency. This includes supporting the development of new industries in rural regions to increase employment and income for farmers. Authorities should also provide comprehensive agricultural technology training in areas such as soil analysis, optimized fertilization, water-saving irrigation, reduced input costs, and improved production efficiency. These policies have enhanced the level of rural human capital while achieving a synergistic improvement in agricultural carbon emission efficiency.

(2)

Given that LNE has a strong positive effect on agricultural carbon emission efficiency, the government can vigorously support the development of rural specialty processing industries and rural leisure tourism, creating more local employment opportunities and allowing rural residents to work and increase their income at their doorstep while simultaneously achieving high-efficiency agricultural carbon emissions. In addition, the government needs to align with market demands and rural realities to conduct practical skills training, enhancing the employment capabilities of rural residents in urban areas and thereby facilitating urbanization while simultaneously achieving low-carbon agricultural development.

(3)

Due to the regional differences in the impact of QFL and different livelihood patterns on agricultural carbon emission efficiency, different strategies should be provided for the eastern, central, and western regions of China. Given that LNE in the eastern region is the main driver of ACEE, the eastern region should continue to expand non-agricultural employment channels for rural residents while vigorously cultivating family farms. In addition, the government can learn from Jiangsu Province’s “Jiang Aifen Family Farm Model” to achieve a reasonable division of labor among family members and moderate-scale operations. By leveraging the rational allocation of production factors, it can achieve higher agricultural carbon emission efficiency. In the central region, LAL has the most positive impact on ACEE. The central region can further improve land transfer policies, eliminate the current obstacles in the land transfer process, promote large-scale land management, optimize planting structures, and promote low-carbon agricultural technologies. The government can learn from Heilongjiang Province’s “Renfa Cooperative Model”, advancing the coordinated progress of agricultural economic benefits and low-carbon benefits. Western regions can refer to Sichuan Province’s “Yufeng Crop Planting Professional Cooperative Model” to optimize agricultural social service methods, improve the model of cooperatives undertaking farming and planting, and strengthen the cooperative relationship between cooperatives and farmers, thereby preventing adverse impacts on agricultural carbon emission efficiency when farmers’ livelihoods transition to business-oriented and asset-oriented models.

In addition, this study has the following limitations. Since this study used provincial statistical data, it failed to empirically test the mechanism path of QFL affecting agricultural carbon emission efficiency. In the future, this study will attempt to investigate microdata and use mediation models and path analysis models to further explore the mechanism relationship between QFL and agricultural carbon emission efficiency.