Determinants of Farmland Abandonment Among Peasants in Scattered Villages: The Impact of Family Structure and Social Policies in Southern China

Abstract

With China’s urbanization process and changes in rural family structures, the abandonment of farmland in scattered villages within hilly mountainous regions is becoming an increasingly serious issue, restricting the improvement of land use efficiency. This study analyzes the basic characteristics and variations in abandoned farmland by conducting surveys and interviews with peasants in a scattered village in southern China. Using the Heckman two-stage model, we perform empirical analysis on the factors influencing farmland abandonment, addressing potential sample selection bias. The findings show the following: peasants with better health and higher education levels are more likely to transition to non-agricultural occupations which contributes to an increased abandonment of farmland. However, larger and more integrated land parcels, along with favorable farming conditions, help reduce abandonment. Additionally, rural land transfer and agricultural subsidies are important factors that enhance farmland utilization and mitigate abandonment. These results provide a reference for addressing the abandonment of farmland and improving both the farming environment and social policies in rural villages.

1. Introduction

Since ancient times, peasants have been regarded as the backbone of agricultural cultivation. However, changes in peasants’ farming behavior are increasingly influenced by external factors. With shifts in the natural environment and human society, the human–land relationship is evolving at an accelerated pace [1], particularly highlighted by the Land Use Land Cover Change (LUCC) [2], especially in farmland areas [3]. Land use plays a fundamental role in human society [4], with farmland serving as the material foundation for promoting balanced economic and social development [5]. It also acts as a powerful barrier for fostering green development of the environment [6] and is critical for ensuring the ability to produce staple food [7].

Farmland resources are closely linked to food security and regional stability, and, by extension, national stability [6,7,8], because agriculture is a fundamental industry for economic and social development. Peasants migrate to urban areas for economic benefits [9], better natural conditions [10], and various government policies [11], leading to the abandonment of marginal rural lands [12]. In Europe, Japan, and other developed regions, urbanization has caused a decline in rural population density, leading to a loss of rural labor, which is a primary cause of farmland abandonment [13]. External market stimulation [14] and the decline in agricultural product income have also played crucial roles in expanding land abandonment. Governments have implemented policies to support agricultural development [15], including increasing agricultural subsidies and improving production conditions, which have alleviated the abandonment of farmland to some extent.

In China, since the 21st century, urbanization [16] has diminished the attractiveness of cultivating land in scattered villages, especially in hilly and mountainous areas. The marginalization of farmland in these regions has become more pronounced [17], as the process in which the economic productivity of a certain land use state continuously decreases, while rising opportunity costs of farming [18] and difficulties in mechanization [19] have led to profound shifts in peasants’ social environments. Alternative rural land uses, such as urban expansion or industrial development, offer more lucrative income opportunities. Farmland located in remote areas with inconvenient transportation, low economic returns, and poor agricultural infrastructure has become increasingly abandoned by peasants [20], exacerbated by labor force separation and the expanding scale of abandonment, because the production inputs of peasants are insufficient to generate adequate economic returns, which affects land use efficiency. How to mitigate the phenomenon of farmland abandonment, effectively improve land use efficiency, and achieve stable agricultural production has become a key focus of current societal attention.

Scattered villages represent the traditional form of rural settlements [21], with peasants’ houses scattered near fields, mountains, forests, and lakes. These villages have no central location and are often far apart [22]. The abandonment of farmland can occur both implicitly, due to extensive land management by peasants [23], and explicitly, when peasants abandon agricultural production and allow the land to naturally revert to a non-cultivated state [24]. In Ruijing, located on the western foothills of the southern section of the Wuyi Mountains, scattered villages are widespread, and both forms of abandonment occur frequently.

This study focuses on Q Village, located in the Gannan Mountain Area of southern China. As a typical scattered village, Q Village faces challenges including poor transportation and economic stagnation. The ongoing urbanization process has led to a significant outflow of young local laborers. The sparsely populated land, high farming costs, and large amounts of idle farmland have directly contributed to farmland abandonment in the area, leading to low land use efficiency. For example, inconvenient transportation increases the cost of transporting agricultural products, while being located in mountainous areas raises the expenses of using agricultural machinery. Furthermore, government relocation projects aimed at poverty alleviation have accelerated the outflow of middle-aged laborers, exacerbating the issue. Q Village, in its mountainous location, serves as a representative case for examining the phenomenon of farmland abandonment in southern China.

Peasants are the primary actors in farming activities, and their willingness to farm greatly influences the extent of farmland abandonment [25]. In general, existing studies have mainly used geospatial analysis methods to analyze the causes and countermeasures of farmland abandonment at the macro level of the province or city. However, there are few studies that analyze the phenomenon of farmland abandonment in the scattered villages in southern China, which are very distinctive, from the perspective of small scales such as villages, especially from the perspective of peasants. Therefore, this study explores the basic characteristics and variations in abandoned farmland in scattered villages in hilly mountainous regions. Using questionnaires and interviews with peasants, the study applies the Heckman two-stage model to empirically analyze farmland abandonment and its influencing factors. The findings aim to enrich the theoretical understanding of farmland abandonment and offer practical recommendations for addressing abandonment issues in other scattered villages. Additionally, the study provides a reference for crafting relevant land protection policies for village governments and formulating effective social policies based on the characteristics of the family structure in scattered villages.

2. Methodology

2.1. Study Area

The study area is located at the border of southern Jiangxi Province and western Fujian Province, characterized by a predominantly mountainous and hilly terrain. This region is a typical example of a scattered village in mountainous areas. Based on the data provided by the local government, as shown in Figure 1, the total area of the village is 44.25 km², with 2.708 km² of farmland, including 0.722 km² of abandoned land. The per capita land area is 11,675.46 m², with an average per capita farmland area of 714.41 m², reflecting a dispersed settlement pattern.

According to the local statistical yearbook, the village has a total of 669 registered households, with a population of 3790. Of these, only 210 permanent residents live in 68 households. The village is divided into 30 natural village groups. Abandoned land in the area is spread out across various locations, with small, fragmented plots. The land quality is poor, with most areas being terraced, steep, and lacking adequate water sources. The low economic returns from agricultural production have led to significant population outflow. This is a typical characteristic of scattered villages in southern China. These factors, combined with remote locations and inconvenient transportation, contribute to the abandonment of farmland.

Q Village is a traditional agricultural community where the local economy relies primarily on agricultural production and migrant labor. Designated as a provincial demonstration village for rural revitalization, Q Village boasts a vast area with abundant forest and water resources. Its primary industries include the cultivation of white lotus, navel oranges, and the raising of cattle, pigs, and goats. During the 13th Five-Year Plan, Q Village was recognized as a poverty-stricken area. To assist in alleviating poverty, the government initiated relocation projects, resulting in a significant outflow of the population. Additionally, some villagers have migrated to the coastal areas of China for work. As a result, the village has experienced considerable population loss.

2.2. Data Source

The farmland parcel and residential area data used in this study were provided by the local natural resources department (http://www.ruijin.gov.cn/, accessed on 30 December 2024). The data on abandoned land parcels and the survey information were collected through a rural household survey conducted in Q Village, Ruijin City, Gannan Mountain Region, southern Jiangxi Province, China, in July 2022. Through questionnaires and interviews with each household, we gained insights into the individual, family, farmland, production, and policy characteristics of local peasants. The respondents were either residents of the village or engaged in agricultural activities there. Of the 669 registered residents, only 68 are permanent residents. The researchers distributed questionnaires to the heads of these households and conducted in-depth interviews to ensure data accuracy and reliability. Although the research sample aimed to encompass the village residents comprehensively, the voluntary nature of participation may introduce social desirability bias. To mitigate this, surveys were also conducted with the outflow population to identify the reasons behind the abandonment of farmland. The sample includes peasants of varying ages, household structures, and socioeconomic backgrounds, effectively reflecting the land abandonment situations across different groups.

The survey process involved using local land use data to define the scope of farmland and the location of residential areas within the village. We then focused on the abandonment and idleness of local farmland (shown in Figure 1). The questionnaire covered a variety of topics, including family characteristics (such as household members and agricultural inputs), farmland status (e.g., land area, crops, abandoned land), and personal information (e.g., agricultural subsidies, income, and expenditure). These questionnaires were tested with a sample to ensure accuracy. Finally, field surveys and in-depth interviews with peasants, along with semi-participatory evaluation questionnaires, were conducted to gather detailed information on the reasons for farmland abandonment in the village. The surveyors introduced the questions to the respondents based on the questionnaire and assisted them in answering.

2.3. Theoretical Basis and Research Hypothesis

2.3.1. The Influence of Individual and Labor Force Characteristics on Peasants’ Farmland Abandonment

Figure 2 summarizes all five constructs, the proposed hypothesis, and the research framework.

Hypothesis 1.

Younger peasants and those with higher educational levels are more likely to abandon their farmland.

Age is widely regarded as a significant factor influencing farmland abandonment [26]. As the head of household grows older, they become more experienced in agricultural production and more reliant on land for their livelihood, making them more likely to continue farming. However, with age, a decline in labor capacity often leads to neglect in land management, a reduction in cultivation capacity, and ultimately, the abandonment of land. Additionally, the health of the household head plays a critical role in agricultural production [27]. Peasants in good health are better able to maintain the physical strength required for daily agricultural activities, increasing the likelihood of expanding their farming operations. Conversely, a higher level of education may enable peasants to adopt more advanced production methods, improve mechanization, and implement intensive land management practices, all of which can expand agricultural production. However, due to the changes in choices brought about by the improvement in education levels and the competitive attractions between rural and urban lifestyles, peasants with higher education levels begin to have doubts about agricultural production when faced with the choice between farming and non-agricultural employment in urban areas. The urbanization process provides more employment opportunities, and the low income and labor intensity of agriculture cannot compete with the advantages of urban jobs. Additionally, the impact of the diverse lifestyle in modern cities may lead some peasants to shift away from agriculture, increasing the likelihood of farmland abandonment.

Hypothesis 2.

The migration of the agricultural population to urban areas increases household income but also raises the likelihood of farmland abandonment.

When the younger labor force moves to non-agricultural areas to settle and form new families, the remaining rural labor force typically continues agricultural work to sustain their livelihood [28]. The total population reflects the available labor force for agricultural production and the family’s daily livelihood needs. A larger population can provide greater human capital, which may expand agricultural activities. In contrast, a smaller population can reduce the burden of daily living but may struggle to scale up agricultural production, which increases the risk of land abandonment. In the mountainous areas of southern Jiangxi, where the ecological environment is fragile, it is challenging for peasants to boost household income through traditional agriculture [29]. Purely agricultural households typically have lower per capita income compared to households involved in non-agricultural labor [30]. Thus, lower household income, which primarily depends on agricultural activities, correlates with higher land utilization. In contrast, higher per capita income, often due to non-agricultural work, increases the likelihood of abandoning farmland.

2.3.2. Effects of Resource Endowment and Production Characteristics on Peasants’ Farmland Abandonment

Hypothesis 3.

Lower land fragmentation and higher satisfaction with land quality will reduce farmland abandonment.

The cultivated area is indicative of a household’s agricultural production capacity [31]. Larger land areas facilitate improved production efficiency, enabling peasants to engage in large-scale operations. This helps to increase the productivity per unit of land, achieve economies of scale, and thus enhance the overall output and efficiency of agricultural production. In China, agricultural production is predominantly organized on a small scale, with the peasant household as the primary operational unit. This small-scale approach, especially when contrasted with the employment opportunities available in urban areas, has evident drawbacks, including increased risks of farmland abandonment. The spatial distribution of farmland parcels [32] reflects the degree of fragmentation. High fragmentation can promote planting diversification and the optimal allocation of land and labor resources because fragmented plots often vary in soil type, slope, and microclimate conditions. However, it can also increase farming costs, reducing food production security and raising the risk of abandonment [33]. On the other hand, less fragmented land allows for larger-scale production and mechanized farming, improving efficiency, accelerating commercialization, and increasing land utilization, which in turn reduces abandonment risks.

Land quality satisfaction reflects peasants’ overall evaluation of farming conditions [34]. Adequate irrigation, accessible transportation, fertile soil, and improved land infrastructure contribute to favorable land conditions. Peasants are more likely to be satisfied with their land when these resources are available, which reduces the chances of land abandonment.

Hypothesis 4.

Increased crop rotation and shorter commuting distances can reduce the risk of farmland abandonment.

Crop rotation, a land use practice, helps regulate soil fertility and conserves the environment [35]. A higher rotation rate cannot only increase crop yields and peasants’ income but also reduce fertilizer use, improving soil quality and maintaining agricultural production. A low rotation rate, however, leads to nutrient depletion, which diminishes crop quality, disrupts the local ecological cycle, and contributes to land abandonment.

Commuting distance impacts farming convenience [36]. Longer commuting distances increase production and transportation costs, making it less profitable to expand production and more likely that farmland will be abandoned. Village abandonment indicators reflect the severity of this issue [37]. If abandonment is widespread within a village, it suggests poor agricultural conditions and high levels of land neglect. However, through the development of abandoned land through land transfers, peasants can lease idle or underutilized wasteland to professional growers or enterprises, enabling other peasants to continue cultivation and optimizing land resource utilization. This approach helps reduce the overall abandonment rate in the village.

2.3.3. The Influence of Policy Characteristics on Peasants’ Farmland Abandonment

Hypothesis 5.

Rural land transfer and agricultural subsidies can reduce farmland abandonment.

Government policies play a crucial role in supplementing resource endowments. Rural land transfer, which separates land ownership from land management rights, refers to peasants transferring or leasing the management rights of their land to others or enterprises while retaining ownership. This practice reduces land idleness caused by peasants’ limited management capabilities or insufficient labor. Peasants who acquire management rights can consolidate multiple plots, implement mechanized operations, and apply modern agricultural technologies, thereby improving production efficiency and land utilization and reducing the risk of land abandonment. When the land transfer rate is low, landowners who stop farming may leave their land abandoned, leading to the underutilization of farmland resources [38].

Agricultural subsidies are a form of government support for peasants’ production activities [39]. Broad coverage and higher subsidy amounts can motivate peasants to increase their agricultural production, helping to counteract land marginalization and slow the rate of land abandonment. In contrast, limited coverage and small subsidies reduce peasants’ production enthusiasm, which may encourage the non-agricultural transfer of labor and increase the likelihood of abandoning farmland.

The evaluation of grain purchase prices also influences peasants’ willingness to engage in agricultural production [40]. If peasants perceive the grain purchase price as low, their enthusiasm for grain cultivation decreases, making it harder to improve their living standards, leading to a greater tendency to abandon farming and impacting food security. On the other hand, a high evaluation of grain prices indicates stronger production enthusiasm, as higher income from grain farming can improve both the peasants’ livelihood and their willingness to continue farming, ultimately reducing land abandonment.

3. Variables and Research Approaches

3.1. Variable Selection

3.1.1. Dependent Variables

This paper adopts the definition of abandoned farmland proposed by current scholars, which identifies two key criteria: permanently idle farmland and land that no longer generates agricultural value [41]. The study aims to collect data on whether peasants abandon farmland through field surveys. The dependent variables in this study are whether peasants abandon land and the extent of the abandoned area. To measure land abandonment, the following binary approach is used: if peasants abandon their land, the value is recorded as “1”; otherwise, it is “0”. The extent of land abandonment is measured by the actual area of abandoned land (1 mu, approximately equivalent to 666.7 m²).

3.1.2. Exclusive Variables

The Heckman model uses exclusive constraint variables to address the endogenous issue in the selection model. These variables (Z) must strongly explain the first outcome variable (Y₁) to ensure that the inverse Mills ratio (IMR) can accurately detect and correct any bias. In the second stage, the exclusive variable (Z) influences the second outcome variable (Y₂) only indirectly, through its effect on the IMR, ensuring the model’s robustness.

In this study, the exclusive variable is whether peasants have considered recultivating their abandoned land. Whether peasants consider recultivating abandoned land directly reflects their attitudes and willingness toward land use, which influences their participation in certain agricultural projects. A higher willingness to recultivate makes peasants less likely to abandon their land. However, this willingness does not directly affect the decision to abandon land. If the outcome variable of the study is peasants’ income level or production efficiency, their willingness to recultivate abandoned land may not directly affect these outcomes but rather exert an indirect influence by shaping their decision-making behavior. Other factors, such as water shortages or poor infrastructure, can still lead to land abandonment, even if the peasants are inclined to recultivate [42].

3.1.3. Independent Variables

Based on the studies of Wu [43], we selected independent variables that encompass individual and household characteristics, farmland and production characteristics at the parcel level, as well as policy environment factors influencing peasants’ decision-making, as shown in

Table 1. Description of variables.

Type	Variable	Code	Description	Mean	S.D.
Dependent variables
Whether peasants abandon land		Y1	No = 0, Yes = 1	0.397	0.489
Abandoned cultivated area		Y2	Actual abandoned cultivated area (mu)	0.826	1.715
Exclusive variables
Whether peasants have considered recultivation		Z	No = 0, Yes = 1	0.582	0.493
Explained variables
Individual characteristics	Age	X1	Less than or equal to 18 = 1, 19–39 = 2, 40–59 = 3, 60 and above = 4	3.394	0.665
	Health	X2	Very poor = 1, Poor = 2, General = 3, Good = 4	3.678	0.635
	Education	X3	Primary school and below = 1, junior high school = 2, High school/Secondary school = 3, College and Above = 4	1.812	0.739
Household characteristics	Total population	X4	Total household population	4.973	1.721
	Household income per capita	X5	Annual family income/Total population	12,611.642	11,740.152
	Proportion of non-agricultural income	X6	Non-agricultural income/General income	46.769	26.730
Characteristics of farmland	Cultivated area	X7	Total area of household farmland	4.952	3.207
	Parcel of land	X8	Farmland area/Number of farmland blocks	0.730	0.343
	Satisfaction with land quality	X9	Very low = 1, Low = 2, General = 3, High = 4	2.501	1.056
Production characteristics	Crop rotation	X10	No = 0, Yes = 1	0.293	0.455
	Commute distance	X11	200 m and below = 1, 201–400 m = 2, 401–600 m = 3, Above 600 m = 4	1.809	1.010
	Village abandonment	X12	No = 0, Yes = 1	0.407	0.491
Policy characteristics	Rural land transfer	X13	No = 0, Yes = 1	0.418	0.493
	Agricultural subsidies	X14	No = 0, Yes = 1	0.710	0.454
	Evaluation of grain purchase price	X15	Very low = 1, Low = 2, General = 3, High = 4	2.155	0.974

.

3.2. Heckman’s Two-Stage Model

Heckman’s two-stage model effectively addresses both sample self-selection bias and the endogenous problems that arise from such bias in regression analysis. The model consists of two stages. The first stage involves the behavioral decision-making process, where the Probit model is used to predict the probability of farmland abandonment based on all observed values. In addition, the inverse Mills ratio (IMR) for each observation is calculated and incorporated into the estimation in the second stage, allowing for the correction of any endogenous bias caused by sample self-selection [44].

In this study, we apply Heckman’s two-stage model to empirically analyze farmland abandonment and its influencing factors in the mountainous regions of southern Jiangxi, China. In the first stage of the analysis, the model assesses whether a household has abandoned its farmland. The Probit model is employed for decision estimation at this stage, as outlined in the following expression:

$$Y_{1i}=\left\{\begin{array}{c}1,if Y_{1i}^{*}>0\\ 0,if{ Y}_{2i}^{*}\le 0\end{array}\right\}$$

(1)

$$Y_{1i}^{*}=\beta X_i+{\epsilon }_i$$

(2)

In Equations (1) and (2), $Y_{1i}^{*}$ represents the unobservable latent variable that indicates whether the peasants’ farmland is abandoned. $X_i$ represents the explanatory variables for both equations, while $\beta$ represents the estimated coefficients of the respective variables. ${\epsilon }_i$ is the random disturbance term and ${\epsilon }_i~N (0, {\sigma }^2)$. When $Y_{1i}$ = 1, it indicates that the peasants’ farmland is abandoned, and when $Y_{1i}$ = 0, it signifies that the farmland is not abandoned.

The missing inverse Mills ratio (IMR) in the regression equation, which corrects for the bias caused by sample self-selection, is expressed as the following:

$$\mathsf{\lambda }={\displaystyle \frac{\phi \left(\beta X_i/{\sigma }_0\right)}{\varnothing \left(\beta X_i/{\sigma }_0\right)}}$$

(3)

In Equation (3), $\phi \left(\beta X_i/{\sigma }_0\right)$ represents the density function of the standard normal distribution, while $\varnothing \left(\beta X_i/{\sigma }_0\right)$ represents the cumulative distribution function.

The second stage involves analyzing the extent of farmland abandonment, represented by the area of abandoned land. In this stage of regression, we incorporate the IMR calculated in the first stage as a correction parameter. The expression for this is as follows:

$$Y_{2i}=\beta x_i+\alpha \widehat{\mathsf{\lambda }}+{\epsilon }_i$$

(4)

In Equation (4), $Y_{2i}$ represents the explanatory variables in the second-stage regression, while denotes the area of peasants’ abandoned land. $\beta$, $\alpha$ are the parameters to be estimated, and $\widehat{\mathsf{\lambda }}$ represents the inverse Mills ratio function. If $\alpha$ passes the significance test, it indicates the presence of sample selection bias, thereby confirming the necessity of the Heckman model. Additionally, $x_i$ is a strict subset of $X_i$ from the first-stage regression, meaning that at least one variable in $X_i$ does not directly affect $Y_{2i}$, but it significantly explains the decision to abandon farmland.

4. Results

4.1. Demographic Profile of the Respondent

As shown in

Table 2. Descriptive statistics of individual characteristics.

Variable	Classification	Sample Size	Proportion (%)
Age	18 and below	4	1.19
	19–39	22	6.57
	40–59	147	43.88
	60 and above	162	48.36
Educational level	Elementary school and below	121	36.12
	Junior high school	164	48.95
	High school/vocational school	42	12.54
	College and Above	8	2.39
Proportion of household labor force (%)	0–25	71	21.19
	25–50	81	24.18
	50–75	111	33.13
	75–100	72	21.49
Farmland area (m2)	0–1666.66	55	16.42
	1666.66–3333.33	179	53.43
	3333.33–5000	51	15.22
	>5000	50	14.93
Parcel of land (m2)	0–166.66	6	1.79
	166.66–333.33	131	39.10
	333.33–500	56	16.72
	>500	142	42.39
Abandoned land area (m2)	0	202	60.30
	0–666.66	77	22.99
	666.66–1333.33	25	7.46
	>1333.33	31	9.25

, a total of 335 samples were collected. The majority of respondents are middle-aged and elderly peasants, with 48.36% being over 60 years old, and 85.07% having only a junior high school education or lower. The farmland area per household is 3300 m², with 69.85% of households cultivating less than 3333.33 m². The average land parcel size is 486.66 m², with most plots being smaller than 333.33 m², accounting for 40.90% of cases. Among the interviewed peasants, 39.70% reported abandoned land. The results indicate that the scale of individual farming is small, with fragmented fields, and abandonment is widespread.

The respondents’ average annual income per capita is relatively low, at 12,611.64 yuan, with non-agricultural income contributing a significant portion-up to 46.77%. Additionally, households with higher annual incomes tend to have a larger proportion of non-agricultural income.

4.2. Descriptive Analysis

Before performing regression analysis, tests for multicollinearity and variable correlation are conducted.

Table 3. Correlation matrix and VIF test table between variables.

	X1	X2	X3	X4	X5	X6	X7	X8	X9	X10	Z
X1	1.000
X2	−0.123 **	1.000
X3	−0.420 ***	0.093 *	1.000
X4	0.179 ***	−0.065	−0.121 **	1.000
X5	−0.259 ***	0.113 **	0.142 ***	0.067	1.000
X6	−0.391 ***	0.110 **	0.194 ***	0.095 *	0.571 ***	1.000
X7	0.047	−0.021	−0.171 ***	0.486 ***	−0.032	0.095 *	1.000
X8	0.085	0.047	−0.055	0.152 ***	−0.120 ***	−0.192 ***	0.244 ***	1.000
X9	0.110 **	0.010	−0.109 **	0.188 ***	0.039	0.004	0.172 ***	0.472 ***	1.000
X10	0.073	−0.046	−0.049	0.216 ***	−0.068	−0.032	0.302 ***	0.179 ***	0.204 ***	1.000
Z	−0.071	−0.049	0.063	0.043	0.080	0.091 *	−0.031	0.001	0.018	0.026	1.000
VIF	2.06	1.99	1.67	1.60	2.98	3.22	3.93	4.22	7.59	1.74	19.04

Note: ***, **, and * indicate significance at the levels of 1%, 5%, and 10%, respectively.

presents the correlation coefficients between the explanatory variables used in the first and second stages. Generally, if the absolute value of the correlation coefficient exceeds 0.7, it indicates a high linear correlation. The correlation matrix shows that all coefficients are below 0.7, suggesting no significant autocorrelation between the variables.

The Variance Inflation Factor (VIF) is used to assess multicollinearity. A VIF greater than 10 indicates serious multicollinearity, while a VIF between 5 and 10 suggests moderate multicollinearity. A VIF below five implies that multicollinearity is not a concern. According to the VIF analysis, the average VIF for the variables in the abandonment decision model is 3.50, indicating that multicollinearity is not a significant issue in the selected variables.

As shown in

Table 4. Heckman regression model results.

Y1 = Whether to Abandon	Abandonment Decision Model (H-1)		Abandonment Area Model (H-2)
Y2 = Abandoned Area	Coef.	S.E.	Coef.	S.E.
Whether to abandon			1.799 ***	(0.000)
Age	−0.191	(0.179)	0.021	(0.886)
Health	−0.321 ***	(0.008)	−0.179	(0.239)
Education	−0.148	(0.189)	0.269 **	(0.025)
Total population	−0.030	(0.572)	0.027	(0.596)
Household income per capita	−0.000	(0.122)	0.000 *	(0.061)
Proportion of non-agricultural income	0.009 **	(0.020)	−0.010 **	(0.023)
Cultivated area	0.112 ***	(0.000)	0.067	(0.115)
Field area	−0.880 ***	(0.003)	0.963 **	(0.020)
Satisfaction with land quality	−0.447 ***	(0.000)	0.174	(0.330)
Crop rotation	−0.320 *	(0.085)	0.501 **	(0.012)
Commute distance			0.283 ***	(0.000)
Village abandonment			−0.220	(0.179)
Rural land transfer			−0.335 *	(0.072)
Agricultural subsidies			0.021	(0.908)
Evaluation of grain purchase price			0.160 *	(0.087)
Whether peasants have considered recultivation	−0.331 **	(0.037)
IMR			−0.915 *	(0.077)
_cons	3.143 ***	(0.000)	−0.913	(0.363)
r2_a			0.472
F			18.490

Note: The parentheses are the standard errors corresponding to the estimated coefficients. ***, **, and * indicate significance at the levels of 1%, 5%, and 10%, respectively.

, the results from the first stage of the abandonment decision model (H-1) indicate that farmland characteristics significantly influence peasants’ decisions regarding land abandonment. Specifically, the cultivated area, field area, and land quality satisfaction—representing the farmland characteristics—passed the significance test at the 1% level. Additionally, the health status, representing individual characteristics, and the proportion of non-agricultural income, reflecting household characteristics, were significant at the 1% and 5% levels, respectively. The exclusive variable, whether peasants have considered recultivating their land, was significant at the 5% level. Therefore, factors such as land quality and the health status of peasant households significantly affect peasants’ decisions to abandon their farmland.

The results from the abandonment area model (H-2) in the second stage show that, in addition to farmland characteristics, household, production, and policy characteristics also significantly influence the extent of farmland abandonment. When combined with the findings from the abandonment decision model (H-1) in the first stage, it is evident that a smaller land parcel, lower household income per capita, and lower satisfaction with land quality are associated with a greater area of abandoned farmland.

4.3. Impact of Exclusive Variables on Farmland Abandonment

The results from the dependent variables in the abandonment decision model (H-1) reveal that whether peasants have considered recultivating their land has a significant negative impact on land abandonment, with a significance level of 5%. In other words, peasants who are willing to recultivate their land are significantly less likely to abandon it. This indicates that the selection of dependent variables is valid, and the Heckman two-stage model effectively addresses the endogenous bias caused by sample selection.

4.4. Impact of the Family Structure of Peasants on Farmland Abandonment

From the perspective of individual characteristics, the health status of peasants has a significant negative impact on land abandonment, with a significance level of 1%. Although age and education level do not directly influence the likelihood of abandoning land, education levels significantly increase the area of abandoned land among those who have abandoned their land, with a significance level of 5%.

Regarding household characteristics, per capita household income is a key factor influencing farmland abandonment. The proportion of non-agricultural income has a significant positive impact on land abandonment at the 5% level. For those who have abandoned their land, both per capita household income and the proportion of non-agricultural income significantly affect the area of abandonment, with significance levels of 1% and 5%, respectively.

4.5. Impact of Peasants’ Production on Farmland Abandonment

From the perspective of peasants’ farmland characteristics, cultivated area, field area, and land quality satisfaction are significant factors influencing farmland abandonment. These variables have a substantial impact on the area of abandoned land, with significance at the 1% level. Furthermore, the field area has a significant positive impact on the area of abandonment at the 5% level. The influence of field area on abandonment is evident in that larger field areas tend to have lower fragmentation, which facilitates more efficient farming and reduces the likelihood of abandonment. However, the average field area of the interviewed peasants is only 453.33 m², and the overall plot integrity is relatively low. For fragmented plots with poor farming conditions, peasants are more likely to abandon them. This finding supports Hypothesis 3, suggesting that greater land integrity reduces abandonment.

From the perspective of peasants’ production characteristics, both crop rotation and commuting distance significantly affect the extent of farmland abandonment, with positive impacts observed at the 5% and 1% levels, respectively. To improve land productivity, peasants often rotate or fallow less economically viable plots to restore soil fertility. Over time, land with poor farming conditions—such as high altitude, long commuting distances, and inadequate irrigation—tends to be abandoned, further contributing to the phenomenon of land abandonment. Additionally, the longer the commuting distance, the more time peasants spend traveling between their homes and farmland, which increases the likelihood of abandonment.

4.6. Impact of Rural Policy on Farmland Abandonment

From the perspective of rural policy characteristics, several factors influence peasants’ decisions regarding land abandonment, including agricultural subsidies, rural land transfer, and the evaluation of grain purchase prices. It is important to note that despite receiving agricultural subsidies, some peasants continue to exhibit high levels of land abandonment. This is primarily due to China’s current agricultural policies, with a relatively low number of subsidies, which are only 113 yuan per mu, and are distributed exclusively to landowners, while the actual cultivators of the land do not receive agricultural subsidies.

Rural land transfer has a significant negative impact on the area of farmland abandonment, with a significance level of 10%. This indicates that greater land transfer activity leads to higher land utilization rates and reduces the amount of idle land, thereby decreasing the degree of farmland abandonment. Conversely, the grain purchase price has a significant positive impact on the area of abandonment, at the 10% level. Although the grain purchase price has increased, it remains relatively low, which diminishes peasants’ enthusiasm for cultivating grain. This lack of motivation hinders improvements in their living standards and serves as a key factor discouraging peasants from continuing grain cultivation, which ultimately affects food security.

5. Discussion

This paper analyzes the relationship between farmland abandonment and peasants from the perspective of peasants using the Heckman two-stage model. However, one limitation of this study is that, due to the constraints of the sample data, the conclusions are applicable only to scattered villages in hilly regions. Further research is needed to determine whether these conclusions hold in other areas. Additionally, the farmland plot data and the peasant survey data have not been effectively integrated, and the underlying causes of abandonment for each specific plot require further analysis.

5.1. Effect of Family Structure

Peasants, as the primary unit of economic and social activity in rural areas, significantly influence the allocation of farmland resources through changes in various production factors [45], which can lead to land abandonment, thereby influencing the improvement of land use efficiency. In certain scattered villages in hilly regions of China, government poverty alleviation relocation projects have resulted in peasants moving out of their original villages, leaving land in remote areas unused and abandoned [46]. Furthermore, urbanization, along with the declining farming capacity of the aging population, leads to a reduction in farmland area, while the outflow of young people has a substantial impact on local farmland abandonment [36], which partially supports Hypothesis 1. The outflow of young people in these villages is significant, and although the remaining permanent population is highly dependent on land, aging reduces their farming capacity, thereby increasing the likelihood of abandonment.

Additionally, the highly educated young population tends to migrate for better employment opportunities, further escalating the risk of land abandonment, which aligns with Hypothesis 1. The lower education levels of the middle-aged and elderly labor force left behind in the mountainous areas limit their ability to adopt advanced farming techniques, making it difficult to scale up production. In addition, as peasants age, their physical ability to perform labor-intensive agricultural tasks diminishes, making it challenging to effectively maintain and cultivate their land, while the abandoned land is less likely to be utilized effectively [47]. Thus, the aging population contributes to the expansion of farmland abandonment, suggesting that Hypothesis 1 is not entirely accurate. Good health among peasants promotes non-agricultural labor migration, which affects their daily farming activities, ultimately leading to land abandonment. Conversely, poor health limits the physical strength required for farming, thereby exacerbating land abandonment.

Non-agricultural labor migration is an important factor contributing to land abandonment, which aligns with Hypothesis 2. In many developing countries, limited economic opportunities, inadequate infrastructure, and insufficient public services in rural areas have driven significant rural-to-urban migration as individuals seek better living conditions in cities. This migration often leads to labor shortages in agriculture and contributes to farmland abandonment. As non-agricultural income increases, peasants tend to leave their farmland idle, thus exacerbating land abandonment. This confirms that higher household income provides peasants with the time and resources to expand agricultural production, thereby reducing abandonment. Population size also influences the extent of abandonment. In households with larger populations, more young members tend to engage in non-agricultural labor, which increases the likelihood of land abandonment.

5.2. Effect of Peasants’ Production

In hilly regions, the location of farmland significantly influences abandonment decisions. Poor geographical conditions and low accessibility can increase farming costs, leading to higher abandonment rates for these parcels. Many studies have examined this issue using remote sensing data [48] and spatial sampling surveys [49]. Factors such as fragmented parcels, steep slopes, limited access to transportation, long commuting distances, and low land quality contribute to a higher likelihood of abandonment, consistent with Hypotheses 3 and 4. When peasants are dissatisfied with the land conditions, they are less likely to continue farming. The lack of irrigation, limited transportation access, poor soil fertility, and inadequate land infrastructure in mountainous areas result in lower satisfaction with land quality. When the input required to cultivate the land exceeds the output, peasants may reduce their farming efforts, ultimately leading to land abandonment. While larger cultivated areas are conducive to large-scale operations, the difficult cultivating conditions in mountainous areas push peasants to prioritize land with higher profits, further aggravating land marginalization and abandonment.

To reduce commuting and material transportation costs, peasants are also likely to abandon more distant land. This supports Hypothesis 4, indicating that shorter commuting distances which lower daily management costs, are associated with reduced abandonment. When some peasants abandon their land, others may farm it through land transfer, curbing the overall abandonment trend. However, if other peasants opt for non-agricultural employment, this may exacerbate abandonment.

In some developed countries, large-scale, modernized farming operations have achieved economies of scale, significantly reducing the influence of individual farmer decisions on land abandonment. These experiences indicate that farmers’ production decisions are largely influenced by external environments and policy frameworks. Therefore, the key to addressing the issue of farmland abandonment lies in strengthening farmland infrastructure construction to enhance land use efficiency. Improving irrigation systems, roads, and mechanized farming conditions can make land more suitable for efficient agricultural production. Particularly in mountainous areas, constructing farmland water conservancy facilities, paving field roads, and improving terrace construction can effectively reduce land abandonment. Additionally, optimizing farmland for mechanized operations and expanding the scale of mechanized production should be prioritized. In regions with complex terrain, special attention should be given to developing small-scale agricultural machinery suitable for farmers, thereby improving their production efficiency.

5.3. Effect of Rural Policy

Due to the decline in land resource utilization efficiency and the severe issue of land waste caused by farmland abandonment, the government has taken notice of this problem. Through the implementation of policies such as rural land transfer, land consolidation and reclamation, and the construction of high-standard farmland, the government is striving to enhance the efficiency of abandoned land use, promote sustainable agricultural development, and revitalize rural areas. As mentioned in this study, the land transfer policy allows the management rights of abandoned farmland to be transferred to other local peasants, thereby increasing land utilization efficiency and expanding agricultural production. This not only revitalizes rural productivity but also effectively reduces abandonment of high-quality land [50]. However, in cases of land transfer, if the actual cultivators of the land do not receive agricultural subsidies, their motivation to continue grain cultivation is diminished, leading to higher abandonment [51]. The incomplete distribution of agricultural subsidies, particularly in land transfer scenarios, means that cultivators may not fully benefit, thus undermining Hypothesis 5.

In response to this issue, the Chinese government has implemented measures at various levels to address land abandonment. According to the No. 1 Central Document released in 2025, the Chinese government has implemented several measures to strengthen farmland protection and utilization. The document emphasizes the need to strictly regulate the total amount of farmland, calls for the high-quality development of high-standard farmland, and proposes a classified approach to promoting the reclamation and utilization of abandoned farmland. Land reclamation is also one of the measures involving the restoration of damaged or degraded land to a usable state through engineering and technical interventions. These initiatives aim to revitalize existing land resources and prevent land from remaining idle and wasted. However, in the local context, weak infrastructure poses significant challenges to implementation, and the mountainous terrain may lead to risks such as soil erosion and other forms of ecological degradation.

Similarly to studies conducted in other mountainous areas of China, such as Chongqing, which has a higher average altitude [17], improving rural road network density and accelerating farmland transfer are considered key measures to address the issue of farmland abandonment. These studies also mention the impact of the development of non-agricultural employment on peasants’ abandonment of farmland [20,26]. However, the unique topographical features of mountainous areas also mean that some land may have no agricultural value, and in such cases, returning land to forestry can be an option. This is also different from the issue of farmland abandonment in plain areas. Specifically, this study focuses on scattered villages, where the impact of family structure and social policies is more pronounced in influencing whether farmers abandon their land.

As part of the 2020 policy to curb abandoned land, local governments have further subdivided the responsibility for managing abandoned land, with county agricultural departments overseeing abandoned land in towns, township cadres handling land in village groups, and village cadres responsible for managing land abandonment at the household level. Peasants’ subsidies play an important role in encouraging grain cultivation [52]. However, according to our survey, in areas where land transfer occurs, actual cultivators have not received subsidies, highlighting a need for improvement in policy implementation. That is, the grain subsidy should be distributed to the actual grain producers, as this would better encourage their enthusiasm for farming.

Many countries have accumulated valuable experience in addressing farmland abandonment. For example, the European Commission’s Common Agricultural Policy encourages farmers to continue cultivation and adopt sustainable farming practices by providing direct subsidies and ecological compensation [53]. In the United States, the Farm Bill offers a variety of support measures-including price supports, insurance subsidies, and environmental incentives-aimed at stabilizing farmers’ incomes and promoting efficient land use [54]. Therefore, when designing and implementing agricultural policies, government agencies can consider introducing diversified incentive mechanisms. These may include offering long-term rental compensation or ecological subsidies to encourage farmers to participate in land fallowing and ecological restoration programs [55]. At the same time, it is important to provide technical guidance and services to participating farmers [31]. Policymakers should balance the needs of food security with environmental protection and avoid one-size-fits-all policy enforcement.

6. Conclusions

This study focuses on scattered villages in southern Jiangxi and systematically analyzes the causes and mechanisms of farmland abandonment from the perspective of peasants, aiming to uncover the underlying factors contributing to land abandonment in mountainous areas.

The findings suggest that good health and higher education levels among peasants often lead to non-agricultural labor migration, which exacerbates the extent of farmland abandonment. Conversely, larger and more integrated land parcels, along with favorable farming conditions, contribute to a reduction in land abandonment. Additionally, the presence of rural land transfer policies and agricultural subsidies plays a crucial role in improving farmland utilization and mitigating land abandonment.