Can Land Transfer-In Improve Farmers’ Farmland Quality Protection Behavior? Empirical Evidence from Micro-Survey Data in Hubei Province, China

Abstract

Enhancing farmers’ behaviors toward cultivated land quality protection is a crucial support in achieving sustainable agricultural development and the national food security strategy. This study aims to investigate the impact of land transfer-in on farmers’ behaviors regarding cultivated land quality protection, explore the underlying mechanisms, and analyze group heterogeneity. To achieve this, the study empirically estimates the impact of land transfer-in on farmers’ behaviors in protecting cultivated land quality using micro-survey data from 743 households in Hubei Province, while addressing endogeneity and conducting robustness checks. The study further explores the mechanisms and heterogeneity of the effects of land transfer-in on farmers’ cultivated land quality protection behaviors. The results reveal that (1) land transfer-in significantly increases the likelihood of farmers taking actual actions to protect cultivated land quality and enhances their awareness, thereby promoting protective behaviors; (2) land transfer-in facilitates these protective behaviors mainly through income incentives and social network interactions, while rental constraints may have a suppressive effect; (3) full-time farmers, those with higher agricultural literacy, those who access agricultural knowledge online, and those affected by natural disasters are more likely to engage in proactive land quality protection behaviors after land transfer-in.

1. Introduction

Cultivated land quality is the core of China’s “trinity” protection of arable land [1]. A good quality of cultivated land is fundamental to achieving quality-driven agriculture, green agriculture, and rural revitalization [2]. However, the quality of cultivated land in China is increasingly deteriorating. Issues such as soil degradation, land infertility, over-cultivation, and excessive use of chemical fertilizers are widespread and have become key challenges hindering sustainable agricultural development [3,4,5]. According to data from the National Bureau of Statistics of China, in 2019, the country’s chemical fertilizer input reached 325.65 kg/hm², 1.45 times the internationally recognized upper limit of 225 kg/hm² [6]. The overall soil organic matter content is less than 50% of that of the same type of soil in Europe [7]. Therefore, the 2024 “Report on the Situation of Cultivated Land Protection” issued by the State Council emphasized the need to “strengthen the management of abandoned land, and promote the quick restoration of agricultural production through substitute planting and farming methods, ensuring that cultivated land is well protected and effectively utilized”. This policy orientation strengthens the demand for the intensive utilization of land resources, making land transfer—especially land transfer-in—an important means of incentivizing farmers to adopt cultivated land protection behaviors under institutional incentives. Land transfer-in not only changes the land use patterns of farmers but also influences their production decisions and farming methods [8]. Many farmers, under the traditional small-scale farming economic model, often tend to pursue short-term economic benefits due to the lack of stable land use rights, thereby neglecting the long-term protection of cultivated land quality [9]. With the advancement of land transfer, the scale of farm operations has expanded. Specifically, relevant studies have shown that with the advancement of land transfer, the scale of farm operations has continued to expand. Agricultural production methods have gradually shifted toward more intensive and large-scale practices, leading to corresponding changes in land use patterns. For example, some studies have found that land transfer promotes the consolidation and large-scale operation of farmland, thereby facilitating the optimization and efficiency improvement of land use practices [10]. Whether or not this leads to enhanced behaviors in protecting cultivated land quality has become an important research question.

Farmers are the direct participants in the protection of cultivated land quality, making the study of their behaviors in this regard particularly important. Current research on farmers’ cultivated land quality protection behaviors mainly focuses on two areas: first, the participants in land quality protection behaviors [11], and second, the various factors influencing farmers’ behaviors [12]. Regarding the study of the participants in land quality protection behaviors, the mainstream view holds that farmers are the central actors in land quality protection [13]. As direct users of the land, farmers’ actions are crucial for the protection of cultivated land quality [14]. In terms of the multiple influencing factors on farmers’ land quality protection behaviors, they are mainly divided into two categories. The first includes the internal factors of the behavior subjects, such as farmers’ individual characteristics [15,16,17], family operation characteristics [18,19], cognitive changes [20,21], and social network characteristics [22,23]. Specifically, social networks in rural China may exert dual effects under different contexts. On the one hand, as an informal institutional arrangement, social networks can foster and disseminate community norms related to conservation-oriented farming practices within villages [24]. For example, experienced large-scale growers or grassroots cadres may influence other farmers to adopt land-conserving measures—such as straw return, crop rotation and fallowing, and organic fertilization—through word-of-mouth, demonstration, and informal communication. On the other hand, in certain regions, social networks may reinforce reliance on high-intensity farming practices [25]. Particularly in contexts characterized by profit maximization and short-term land leases, growers may be more inclined to increase inputs of pesticides and chemical fertilizers or engage in frequent planting to boost yields, rather than prioritizing the long-term sustainable use of cultivated land. Therefore, the role of social networks is context-dependent and may create tension between enhancing intensive production efficiency and promoting sustainable farming norms. The second category, the external factors, includes items such as government incentives and supervision [26], land ownership certification [27,28,29], and agricultural technology promotion [30,31,32].

Research on the impact of land transfer on farmers’ cultivated land quality protection behaviors mainly focuses on the following aspects. The first is the relationship between land ownership and land quality. Relevant scholars have analyzed this from the perspective of land transfer types. For example, Gao Liangliang divides land transfer types into two categories—transfer from relatives and transfer from non-relatives—and suggests that land transferred from relatives tends to have better stability [33]. The second aspect is research on land transfer models and their relationship with cultivated land quality protection, with studies exploring the impact from the perspective of land transfer models [34]. For example, Zhang Jian (2019) [35] investigated the effects of two land transfer models—village collective organization and voluntary transfer by farmers—on farmers’ long-term agricultural investments. The study found that, compared to voluntary land transfers, those managed by village collective organizations offer better land tenure stability advantages. Third, some studies have paid attention to the impact of land lease duration on farmers’ behaviors related to cultivated land quality protection. Existing research indicates that the length of the lease term significantly affects farmers’ investment decisions, particularly in terms of soil quality protection. Longer lease durations enhance farmers’ sense of land tenure security and expectations of future returns, thereby increasing their willingness to invest in long-term conservation practices such as organic fertilization and soil improvement [36].

Existing studies have analyzed various factors influencing farmers’ cultivated land quality protection behaviors [12,14,17], as well as the impact of land transfer on these behaviors [5,8,9], providing empirical references for this study. However, research on the impact of land transfer-in on farmers’ cultivated land quality protection behaviors still requires further development. Compared to existing studies, this paper’s marginal contributions are as follows: First, it offers a new research perspective. Existing studies primarily focus on individual characteristics (such as age, education level) and external environments (such as policy incentives, agricultural technology promotion) that influence farmers’ land quality protection behaviors [16,26,29], but rarely explore how land transfer behavior, especially land transfer-in, affects farmers’ land protection decisions. Second, it expands the research mechanisms. This paper further investigates the impact of land transfer-in on farmers’ land quality protection behaviors through three mechanisms: income incentives, transfer cost constraints, and enhanced environmental protection awareness. This provides a new framework for understanding the pathways through which land transfer-in influences farmers’ land quality protection. Third, it improves the empirical estimation strategy. Farmers’ land transfer-in is a typical self-selection behavior [37], influenced by various internal and external factors. This paper employs multiple methods, including IV-2SLS, endogenous switching models, and replacing explained variables, to address endogeneity and perform robustness tests, ensuring the scientific and rigorous nature of the estimation results.

Based on the above content, the purpose of this study is to explore the impact of land transfer-in on farmers’ cultivated land quality protection behaviors, the mechanisms through which it operates, and group differences. The work undertaken to achieve this research goal is as follows: First, the study constructs an analytical framework for how land transfer-in affects farmers’ land quality protection behaviors. Second, using data from a micro-level survey of 743 farmers in Hubei Province, China, the study employs Probit models, Order-Probit models, and IV-2SLS models to estimate the impact of land transfer-in on farmers’ actual land quality protection actions and awareness, with corresponding endogeneity treatment and robustness checks. Third, the study investigates the mechanisms through which land transfer-in influences farmers’ land quality protection behaviors and explores group differences.

2. Theoretical Analysis and Research Hypotheses

The study on the impact of land transfer on farmers’ land quality protection behavior in this paper mainly focuses on two aspects: first, the direct impact effect, and second, the analysis of the mechanism. The research analysis framework is shown in Figure 1:

2.1. The Direct Effect of Land Transfer on Farmers’ Land Quality Protection Behavior

Cultivated land possesses the characteristics of a quasi-public good, with both farmers and the government sharing the costs of maintaining land quality. According to property rights economics, clearly defined and stable property rights can effectively incentivize economic agents to invest [38]. Property rights reduce uncertainty and facilitate transactions, thereby directing the investment behavior of rights holders toward improving efficiency and increasing output. Land transfer typically implies that farmers acquire more long-term operational rights through the land transfer market, which reduces the uncertainty associated with short-term leases. In cases where property rights are unstable, farmers may adopt short-sighted, profit-maximizing strategies such as intensive cultivation and excessive fertilization, aiming for immediate returns at the expense of long-term soil quality [39]. Conversely, when land tenure is relatively secure, farmers are better able to anticipate future returns and are thus more willing to take responsibility for land conservation. Stable land tenure not only strengthens farmers’ sense of ownership over the cultivated land but also encourages them to adopt more sustainable farming practices to safeguard the land’s future productivity [40]. Under China’s land tenure system, land ownership belongs to the collective, while farmers gain usage rights through land contracting. The land transfer system further grants farmers more flexible operational rights. In recent years, with the advancement of the “separation of three rights” reform in rural land, the stability of land tenure has gradually improved. After transferring land, farmers can obtain relatively secure operational periods, making them more willing to invest in soil improvement and land protection.

On the other hand, land inflow enables farmers to expand their operational scale, facilitating a transition from smallholder farming to large-scale, intensive agriculture [41]. This shift directly reduces per-unit production costs and enhances the economic viability of agricultural operations, thereby increasing farmers’ capacity and incentives to invest in the protection of cultivated land quality. Small-scale farmers often rely on traditional farming experience and lack the motivation and ability to adopt modern agricultural technologies [42]. In contrast, within a larger-scale farming context, farmers are more inclined to employ advanced agricultural machinery, smart irrigation systems, and soil testing-based fertilization techniques to improve production efficiency. These technologies not only increase agricultural output but also minimize soil degradation, thereby enhancing the feasibility of farmland protection. Based on the above, we propose Research Hypothesis 1.

H1: 

Land transfer can promote farmers’ land quality protection behavior.

2.2. The Mechanisms Through Which Land Transfer Affects Farmers’ Land Quality Protection Behavior

2.2.1. Incentive Mechanism of Benefits

Land inflow strengthens income incentives, making farmers more inclined to adopt sustainable farming practices and invest in the protection of cultivated land quality to ensure long-term profit maximization. The rationale for this hypothesis is as follows.

First, land inflow increases operational scale, raises marginal returns, and enhances incentives for land protection [43]. Under the traditional smallholder farming model, farmers typically operate on small plots of land, limiting production scale and resulting in low marginal returns. Consequently, they lack motivation for long-term investment in land quality. However, after acquiring additional land through transfer, farmers are able to expand their operational scale, dilute fixed costs per unit of land, and increase marginal returns, which encourages greater attention to soil quality in pursuit of long-term returns.

Second, the long-term return on investment from land transfer promotes sustainable farming. Farmers cultivating their own fragmented or insecurely leased land may prioritize short-term gains through practices such as over-fertilization or intensive tillage. In contrast, stable, long-term land tenure encourages farmers to adopt sustainable agricultural practices. With long-term operational expectations, farmers are more likely to implement soil conservation measures such as crop rotation, fallowing, and straw incorporation to prevent soil degradation and enhance long-term land productivity.

Third, land inflow promotes the adoption of agricultural technologies and improves land conservation behavior [44]. Land transfer facilitates the entry of more capable and professional farmers into agricultural production, who are more likely to adopt modern agricultural technologies that contribute to cultivated land protection. For instance, large-scale operations reduce the per-unit cost of agricultural machinery, making precision farming practices more accessible and reducing soil degradation through improved management techniques. Based on the above, we propose Research Hypothesis 2.

H2: 

Land inflow enhances farmers’ cultivated land protection behavior through an income incentive mechanism.

2.2.2. Rent Constraint Mechanism

From the perspective of land transfer costs, the key to understanding how land inflow affects farmers’ cultivated land protection behavior lies in how farmers adjust their farming practices and land management strategies in response to rental payments, with the goal of maximizing economic returns. This paper argues that rental pressure may reduce farmers’ efforts in protecting cultivated land quality. Specifically, after land is transferred, tenant farmers are required to pay rent, which may prompt them to adopt high-intensity production methods—such as excessive fertilization and over-irrigation—to increase per-unit output and offset rental expenses [45]. Such short-term oriented production practices often neglect the long-term health of the soil, leading to soil degradation and a decline in cultivated land quality.

This issue is particularly pronounced when rental pressure is high, as farmers tend to focus more on immediate profits and overlook soil conservation measures, thereby further burdening the land [46,47]. High rental costs may incentivize farmers to pursue high-input, high-output short-term strategies to recover rent quickly. While such approaches may improve yields in the short term, they can also result in long-term soil degradation and threaten the sustainable use of farmland. Therefore, excessive rental pressure may reduce farmers’ investment in land quality protection, exacerbate soil degradation, and undermine agricultural sustainability. Based on the above, we propose Research Hypothesis 3.

H3: 

Land transfer imposes rental cost constraints on farmers, thereby reducing their cultivated land protection behavior.

2.2.3. Social Network Interaction Mechanism

Land inflow often leads to the formation of closer social network relationships among farmers, particularly by increasing mutual assistance and information exchange. Through such interactions, farmers can acquire knowledge and experience related to cultivated land protection via their social networks—especially after land inflow, when they may face more decisions regarding land use, management, and conservation. The dissemination of information and sharing of experience through social ties can raise farmers’ awareness of land quality issues and encourage them to adopt effective measures to improve cultivated land quality [48].

Moreover, this paper argues that in regions where land transfer is relatively concentrated, farmers tend to develop stronger connections with other farmers and village collectives. These social networks not only help farmers access information on agricultural policies, subsidies, and loans, but also facilitate the diffusion and sharing of such policy-related information among peers. With increased access to policy support and subsidies, farmers may become more actively engaged in land conservation and quality improvement initiatives [49].

At the same time, as farmers expand their cultivated land through inflow, they may face greater demands for resources such as capital, labor, or technical support. Through their social networks, farmers can obtain resource support from relatives, friends, or village community members—such as labor assistance, financial loans, or agricultural technical guidance. This support not only alleviates the economic burden associated with land protection but also enhances their capacity to implement conservation measures. Based on the above, we propose Research Hypothesis 4.

H4: 

Land inflow enhances farmers’ social network interactions, thereby promoting their cultivated land protection behavior.

3. Materials and Methods

3.1. Study Area and Data

The data used in this study were obtained from a household production and livelihood survey conducted by the research team in Hubei Province in 2021. In this study, a stratified random sampling method was employed to ensure the representativeness of the sample and the reliability of the research findings. Stratified sampling is a method in which the population is divided into distinct strata based on specific criteria, and samples are randomly selected from each stratum. Specifically, in this study, we first divided Hubei Province into several strata based on geographic regions, taking into account varying levels of economic development, agricultural production characteristics, and natural environmental conditions. To better reflect the land transfer situation in different regions, eight representative survey locations were selected, including Honghu City, Jianli City, Shishou City, Jiangling County, Qianjiang City, Xiantao City, Gong’an County, and Tianmen City.

In each selected region, three townships were randomly chosen from each county (or city). Subsequently, three villages were randomly selected from each township, and based on characteristics such as the scale of land transfer, income level, and production methods of households in these villages, 10 to 15 farming households were selected for the survey. This process ensures that households from each stratum are adequately represented in the sample, allowing the survey results to reflect the actual conditions of agricultural households in different regions and types.

To ensure that the research findings accurately reflect the actual conditions of the land transfer system, the research team selected these eight regions, which are characterized by relatively typical levels of maturity in their land transfer markets. First, most of these regions are located in the core agricultural areas of Hubei Province, exhibiting typical agricultural production models and land transfer characteristics. For example, regions such as Honghu City, Jianli City, and Shishou City have relatively active land transfer markets, diversified agricultural production methods, and a more mature land transfer system. These areas have relatively well-established land transfer policies, strong governmental support, diverse market participants, and large-scale land transfers. As a result, these regions serve as strong representatives of the maturity of the land transfer market in Hubei Province and provide robust data to support the study of the impact of land transfer on farmers’ behavior. Second, the selected survey regions encompass different levels of economic development and agricultural production characteristics, which reflect regional variations in the maturity of land transfer markets. For instance, while counties such as Jiangling County, Qianjiang City, and Xiantao City have a solid agricultural foundation, their land transfer markets may not be as developed as those in Honghu City. This diversity in selection is beneficial for analyzing how the maturity of the land transfer market influences farmers’ land protection behavior and further reveals the different mechanisms through which the level of land transfer market development impacts farmers’ behavior. Finally, the geographic distribution of these regions is wide-ranging, covering both the core area of the Jianghan Plain and relatively peripheral agricultural areas. Such a sample selection allows for a comprehensive representation of the varying stages of land transfer market development within Hubei Province, thereby enhancing the generalizability and practical relevance of the research conclusions. The survey region is shown in Figure 2.

Prior to this survey, respondents were informed about the content of the survey and their consent was obtained. Respondents were fully informed about the content and methods of the survey. The survey covered multiple dimensions, including the farmers’ personal characteristics, family demographic structure, agricultural production and management conditions, income and expenditure, living conditions, social participation, and environmental perceptions, aiming to comprehensively reflect the production and living conditions of farming households. Additionally, the survey collected information on the basic conditions of production roads in the respondents’ villages, demographic characteristics, and agricultural technology promotion and services.

The survey was conducted through in-home interviews by trained enumerators, who also ensured data accuracy and reliability through field observations. A total of 743 valid questionnaires were collected. This study, based on the survey data, explores the impact of land transfer on farmers’ cultivated land protection behavior, its mechanisms, and group differences.

3.2. Methods

This study focuses on the impact of land inflow on farmers’ cultivated land protection behavior, where both land inflow and actual actions for land quality protection are binary variables. The study uses whether the farmer has transferred land as a proxy variable. If the farmer has transferred land, the land transfer variable is assigned a value of 1; otherwise, it is assigned a value of 0. Drawing on existing research approaches [50], this study employs a Probit model for estimation, which is specified as follows:

$${(Farmland\_Protect}_i=1)=\mathsf{\Phi }({\beta }_0+{\beta }_1{Farmland\_transfer}_i+X^{\prime }\gamma +{\theta }_{\rho }+{\epsilon }_i)$$

(1)

In this model, Φ denotes the cumulative distribution function (CDF) of the standard normal distribution, ${Farmland\_Protect}_i$ represents whether farmer $i$ participates in actual land quality protection actions (1 = Yes, 0 = No). ${Farmland\_transfer}_i$ indicates whether farmer $i$ rents land. $X$ is a set of control variables, including individual characteristics, family characteristics, and village characteristics; ${\theta }_{\rho }$ represents county (or city) fixed effects, which are included to control for unobservable heterogeneity across different regions that may influence farmland protection behavior. This study primarily focuses on the estimated coefficient ${\beta }_1$ of the explanatory variable ${Farmland\_transfer}_i$. When ${\beta }_1$ is significantly different from zero, it indicates that land inflow alters the probability of farmers’ participation in land quality protection actions.

The awareness of cultivated land quality protection is an ordered categorical variable, and this study uses the Ordered-Probit model for estimation. For farmland quality protection awareness, this study uses the subjective evaluation from the questionnaire item: “Has the household’s use of biopesticides, organic fertilizers, and green manures increased in recent years? (1 = Strongly disagree; 2 = Disagree; 3 = Neutral; 4 = Agree; 5 = Strongly agree)” as a proxy variable for the awareness of farmland quality protection.

In addition, this study also performs relevant endogeneity and robustness checks. Drawing on existing studies [51], we address endogeneity issues using the IV-2SLS method and the endogenous switching model. The estimation model for the IV-2SLS method is as follows:

$$\begin{gathered} P\left({Farmland\_transfer}_i=1|{IV\_Farmland\_transfer}_i,X^{\prime }\right) \\ =\mathsf{\Phi }(a_0+a_1{IV\_Farmland\_transfer}_i+X^{\prime }\gamma +{\theta }_{\rho }+{\epsilon }_i) \end{gathered}$$

(2)

$$\begin{gathered} P\left({(Farmland\_Protect}_i=1|\widehat{I{V\_Farmland\_transfer}_i},X^{\prime },\widehat{{\epsilon }_i}\right) \\ =\mathsf{\Phi }(b_0+b_1\widehat{{Farmland\_transfer}_i}+X^{\prime }\gamma +{\theta }_{\rho }+\rho \widehat{{\epsilon }_i}+u_i) \end{gathered}$$

(3)

In this model, ${IV\_Farmland\_transfer}_i$ represents the instrumental variable. This study uses the average level of land inflow in the village where the farmer resides as a proxy variable. $\widehat{{Farmland\_transfer}_i}$ is the predicted probability of land inflow from the first-stage regression; $\widehat{{\epsilon }_i}$ is the residual term from the first stage, which corrects for endogeneity issues. The parameter $\rho$ measures the impact of endogeneity; if the coefficient is significant, it indicates the presence of endogeneity.

This study uses the endogenous transformation model to construct counterfactual scenarios and estimates the average treatment effect (ATE) of actual land inflow on farmers’ cultivated land protection behavior. Drawing on existing studies [52], the average treatment effect on the treated (ATT) equation is as follows:

$$\begin{gathered} ATT=E\left({Farmland\_Protect}_{1i}|{Farmland\_transfer}_i=1\right)- \\ E\left({Farmland\_Protect}_{0i}|{Farmland\_transfer}_i=1\right) \end{gathered}$$

(4)

In Equation (4), $E\left({Farmland\_Protect}_{1i}|{Farmland\_transfer}_i=1\right)$ represents the cultivated land protection behavior of the sample farmers when they transfer land, and ${Farmland\_Protect}_{0i}$ represents the cultivated land protection behavior of the sample farmers under the “counterfactual” scenario when they do not transfer land.

Similarly, the average treatment effect for untreated farmers (ATU) of land inflow on their cultivated land protection behavior can be obtained as follows:

$$\begin{gathered} ATU=E\left({Farmland\_Protect}_{1i}|{Farmland\_transfer}_i=0\right)- \\ E\left({Farmland\_Protect}_{0i}|{Farmland\_transfer}_i=0\right) \end{gathered}$$

(5)

In Equation (5), $E\left({Farmland\_Protect}_{1i}|{Farmland\_transfer}_i=0\right)$ represents the “counterfactual” cultivated land protection behavior of the sample farmers when they have not transferred land, and $E\left({Farmland\_Protect}_{0i}|{Farmland\_transfer}_i=0\right)$ represents the cultivated land protection behavior of the sample farmers when they have not transferred land.

3.3. Variable Selection

3.3.1. Dependent Variable

The study argues that farmers’ cultivated land quality protection behavior should be distinguished into two aspects: actual protection actions and protection awareness. Actual protective actions refer to the specific behaviors undertaken by farmers in their agricultural practices. In this study, the measurement of actual cultivated land protection behaviors is primarily based on variables derived from farmers’ responses to survey questions regarding their production practices. These include whether they have reduced the use of chemical fertilizers and pesticides, implemented straw return, practiced crop rotation and fallow, or adopted drainage systems. These variables are typically constructed as binary indicators, reflecting whether the farmer has adopted the corresponding protective measures. Protection awareness, on the other hand, reflects farmers’ knowledge, attitudes, and values regarding cultivated land quality protection, embodying their subjective willingness. This includes whether farmers consider land quality protection important and whether they support practices like reducing the use of chemical fertilizers and pesticides or increasing organic fertilizer application. Although there is a certain correlation between the two, it cannot be simply assumed that awareness will always lead to action, or that the lack of awareness means no action will be taken. Therefore, distinguishing between the two helps to better understand the decision-making process behind farmers’ cultivated land quality protection.

Based on the approach of previous studies [53,54], this study identifies six types of cultivated land quality protection actions that farmers might actually participate in: applying organic fertilizer, using soil testing-based fertilization techniques, straw comprehensive utilization technology, water-saving irrigation technology, physical pest control techniques, and water-nutrient integrated technology. If a farmer participates in one or more of these actions, their actual protection behavior for cultivated land quality is recorded as 1; otherwise, it is recorded as 0. This binary approach helps simplify the variable structure, making the econometric model more operational and allowing for a clearer identification of whether farmers have participated in land quality protection measures.

As for protection awareness, this study uses the subjective evaluation from the questionnaire item “In recent years, has the use of biopesticides, organic fertilizers, and green manure increased for the farmer? 1 = strongly disagree; 2 = disagree; 3 = neutral; 4 = agree; 5 = strongly agree” as a proxy variable for land quality protection awareness.

3.3.2. Independent Variable

The explanatory variable in this study is land transfer. Drawing on existing research [55], the study uses whether the farmer has transferred land in as a proxy variable. If the farmer has transferred land, the land transfer variable is assigned a value of 1; otherwise, it is assigned a value of 0.

3.3.3. Control Variables

This study further controls for a series of variables that may affect both land transfer and farmers’ land quality protection behavior, drawing on previous research [15,18,33]. These variables are selected from three aspects: individual characteristics, family characteristics, and village characteristics. Individual characteristics include: gender, age, education level, health status, and experience as a village official. Family characteristics include: labor force size, dependency ratio, land fertility, land fragmentation, the degree of land road hardening, and non-farm income. Village characteristics include: village cultivated land area, per capita village income, and government training programs for the village. Among them, Conservation Awareness, Health, Soil_Fertility, and Social_network are ordinal categorical variables. The definitions of these variables and the descriptive statistical analysis are presented in

Table 1. Definition of variables and descriptive statistical analysis results.

Variable Category	Variable Name	Variable Description	Mean	Standard Deviation
Dependent variable	Conservation Behavior	Whether participated in farmland quality protection behavior: 1 = Yes; 0 = No	0.9502	0.2176
	Conservation Awareness	The usage of biological pesticides and organic fertilizers has been increasing in recent years: 1 = Strongly disagree; 2 = Disagree; 3 = Neutral; 4 = Agree; 5 = Strongly agree.	3.2005	0.9595
Independent variable	Farmland_Transfer	Whether land has been transferred: 1 = Yes; 0 = No.	0.7537	0.4311
Individual characteristics	Gender	1 = Male; 0 = Female.	0.6406	0.4801
	Age	Household head’s age	54.9878	8.5448
	Edu	Years of education received	7.9878	2.7168
	Health	Household head’s health status: 1 = Very unhealthy; 2 = Unhealthy; 3 = Average; 4 = Healthy; 5 = Very healthy.	4.3271	0.7801
	Leader	Whether the household head is a village official: 1 = Yes; 0 = No.	0.1184	0.3233
Household characteristics	Labor	Number of agricultural laborers in the household.	1.9246	0.61405
	Depend_Burden	The proportion of household population under 16 and over 60 years old (%).	0.1841	0.2419
	Soil_Fertility	Soil fertility of farmland: 1 = Very poor; 2 = Poor; 3 = Average; 4 = Good; 5 = Very good.	3.4333	0.7722
	Farmland_Plot	Actual number of plots.	3.1001	3.2641
	Farmland_Roads	The road pavement rate of cultivated land.	0.9428	0.1882
	Non-Farm_Income	Logarithm of non-agricultural income.	1.5264	1.1667
Village characteristics	Farmland_Village	Logarithm of the actual cultivated land area in the village.	6.0226	0.5386
	Village_Income	Logarithm of per capita income in the village	0.6040	0.3519
	Gov_Train	Whether the village organizes technical training by government departments. 1 = Yes; 0 = No	0.8534	0.3213
Mechanism variables	Agri_Income	The logarithm of agricultural net income	2.0819	0.8306
	Rent	The logarithm of the land transfer rent per acre	4.2205	3.0123
	Social_network	Communication with villagers about pesticide and fertilizer usage: 1 = Strongly disagree, 2 = Disagree, 3 = Neutral, 4 = Agree, 5 = Strongly agree.	2.8573	0.6697
Identification variables	IV_Farmland_Transfer	Average land transfer rate within the same village	0.7486	0.2272

.

As shown in Table 1, 75.37% of farmers participated in land transfer, indicating a high prevalence of land circulation within the study sample. Additionally, the actual actions of farmers in land quality protection are generally high, with a mean value of 0.9502. The mean value of protection awareness is 3.2005, suggesting that farmers’ awareness of land quality protection is generally above average. In other words, most farmers acknowledge the importance of land quality protection, although there are still some farmers with relatively low awareness.

4. Results

4.1. The Baseline Regression Results

This paper first analyzes the impact of land transfer on farmers’ farmland quality protection behaviors. The baseline regression results are shown in

Table 2. Estimation results of land transfer-in on farmers’ farmland quality protection behavior.

Variables	(1)	(2)	(3)	(4)
	Conservation Behavior	Conservation Behavior	Conservation Awareness	Conservation Awareness
	Probit	Probit	Order-Probit	Order-Probit
Farmland_Transfer	0.4712 ***	0.4624 ***	0.4070 ***	0.3483 ***
	(0.1713)	(0.1047)	(0.0893)	(0.0975)
Gender	0.2614	0.2485	0.1327	0.1413
	(0.2438)	(0.2435)	(0.0820)	(0.0874)
Age	−0.0279 ***	−0.0266 **	−0.0207 ***	−0.0192 ***
	(0.0101)	(0.0106)	(0.0048)	(0.0049)
Edu	0.2111 ***	0.1266 ***	0.2206 ***	0.1033 ***
	(0.0277)	(0.0290)	(0.0140)	(0.0142)
Health	0.4853 **	0.4194 **	0.3211 ***	0.2682 ***
	(0.1909)	(0.2009)	(0.0516)	(0.0537)
Leader	0.7833 **	0.8675 **	0.2229 *	−0.2540 *
	(0.3830)	(0.3943)	(0.1307)	(0.1395)
Labor		0.2589 **		0.1531 **
		(0.1246)		(0.0689)
Depend_Burden		−0.2095 ***		−0.2422 ***
		(0.0812)		(0.0566)
Soil_Fertility		−0.3072 ***		−0.2438 ***
		(0.1172)		(0.0546)
Farmland_Plot		−0.0578 *		−0.0194 *
		(0.0310)		(0.0111)
Farmland_Roads		0.0154 ***		0.0123 ***
		(0.0040)		(0.0021)
Non-Farm_Income		0.0170		0.1990
		(0.1356)		(0.1531)
Farmland_Village		0.1324		0.1459
		(0.1842)		(0.1772)
Village_Income		0.0611		−0.0283
		(0.2960)		(0.1220)
Gov_Train		0.0528 ***		0.1502 ***
		(0.0130)		(0.0325)
Constant	4.3633 ***	4.4120 ***	-	-
	(1.1356)	(1.2116)	-	-
Regional FE	Controlled	Controlled	Controlled	Controlled
Observations	743	743	743	743

Note: ***, **, * represent significance at the 1%, 5%, and 10% statistical levels, respectively, with the robust standard errors at the county level shown in parentheses.

. In Table 2, columns (1) and (3) represent the impact of land transfer on farmers’ actual farmland quality protection behavior and protection awareness, respectively, when only individual characteristic control variables are included. Columns (2) and (4) show the regression results when all control variables are included.

From Table 2, the Probit regression results show that land transfer has a significantly positive effect on farmers’ adoption of cultivated land quality protection practices (coefficient = 0.4624, significant at the 1% level). It is important to note that the coefficients in a Probit model reflect changes in the latent propensity variable and cannot be directly interpreted as percentage changes in probability. To address this, we further calculated the marginal effect based on the sample mean. The results indicate that land transfer increases the probability of farmers adopting cultivated land quality protection practices by approximately 4.37 percentage points on average. The coefficient of 0.3483 for the land transfer variable in the ordered Probit model indicates that land transfer significantly enhances farmers’ awareness of land protection. To provide a more intuitive explanation, we further calculated the marginal effects. The results show that land transfer increases the probability of selecting “strongly agree” by approximately 4.35 percentage points, while decreasing the probability of selecting “strongly disagree” by approximately 2.42 percentage points. This indicates that land transfer behavior can significantly enhance farmers’ farmland quality protection behaviors, confirming the validity of Hypothesis 1.

This paper argues that land transfer is usually accompanied by longer-term land management rights, which increases farmers’ willingness to make long-term investments in farmland quality protection. For example, measures such as land improvement, increased organic fertilizer application, and straw return are more likely to be adopted. Governments often provide agricultural subsidies, green technology training, and agricultural insurance to farmers engaged in large-scale operations, encouraging them to adopt scientific farmland protection concepts. For instance, policies promoting green planting, soil testing and fertilization, and crop rotation help enhance farmers’ protection awareness and action capabilities.

In terms of personal characteristics, the increase in age significantly reduces farmers’ actions and awareness regarding farmland quality protection, while improvements in education level, better health, and having a village cadre identity significantly enhance both their actions and awareness. This study suggests that as farmers age, their physical strength declines, making it more difficult to carry out labor-intensive protection measures such as deep plowing and returning straw to the field [40]. Additionally, the decline in cognitive abilities may affect their willingness to adopt new technologies and sustainable agricultural practices. Farmers with higher education levels are more likely to access and understand knowledge related to soil protection, such as scientific fertilization and soil testing-based fertilization, thereby improving their awareness of protection. Healthier farmers have more physical strength and energy to implement farmland protection actions, such as deep plowing and green manure planting, which require a certain amount of physical effort [42]. Village cadres are more likely to be exposed to government agricultural policies, environmental protection requirements, and agricultural subsidy information, which strengthens their understanding of the importance of farmland protection.

In terms of family characteristics and village environmental characteristics, an increase in the number of agricultural laborers, road hardening for farmland, and government technical training significantly enhance farmers’ farmland quality protection actions and awareness. On the other hand, an increase in family dependency ratio, soil fertility, and land fragmentation significantly reduces farmers’ actions and awareness of farmland quality protection. This study suggests that with more agricultural laborers, families have more labor resources available to implement time-consuming and labor-intensive farmland protection measures such as returning straw to the field, deep tilling, and crop rotation. After road hardening, agricultural machinery and vehicles can more easily access the farmland, improving the efficiency of mechanized operations, making farmers more inclined to adopt mechanized protection measures (such as straw returning, deep plowing, etc.) [43]. Technical training enables farmers to learn protective farming practices such as scientific fertilization, soil improvement, and crop rotation, reducing the damage caused by excessive use of fertilizers and pesticides. An increased dependency ratio means farmers need to balance other income sources, such as working away from home, leading to reduced time for agricultural activities, making it harder to implement protective farming practices. High soil fertility may reduce farmers’ reliance on modern agricultural technologies [45]. High fertility soils provide a more “relaxed” production environment for farmers, meaning they may not need to rely heavily on external agricultural technologies (such as chemical fertilizers, pesticides, and other modern agricultural inputs). Farmers might even believe that traditional farming methods are sufficient to achieve adequate yields. In such cases, they may lack sufficient motivation for or interest in adopting emerging protective farming practices, such as using biopesticides, applying organic fertilizers, or implementing crop rotation and other sustainable measures.

4.2. Endogeneity Treatment

Although the research design takes into account personal, family, and village characteristics that influence both farmland transfer and farmers’ land quality protection behaviors, there may still be other unobserved factors affecting farmers’ land quality protection actions. First, farmers’ land protection behavior and awareness may influence their decision to transfer land, suggesting a bidirectional causality rather than a one-way relationship. Specifically, farmers’ land quality protection actions and awareness may indeed influence their land transfer decisions. On one hand, if farmers have a strong awareness of land protection or have already adopted proactive protective measures, they may be more inclined to engage in land transfer in order to access more resources or policy support, thereby enhancing land protection. On the other hand, land transfer brings more resources (such as funding, technology, etc.), which may also increase farmers’ awareness of land protection, leading them to pay more attention to long-term soil quality and sustainable agricultural development. Second, there may be unobserved factors, such as agricultural experience and capabilities, that simultaneously affect both farmland transfer and land protection, leading to endogeneity issues. Finally, land transfer is typically a decision made by farmers based on various factors such as their economic situation, land resources, and policy support. Therefore, selecting only farmers who have participated in land transfer may lead to sample selection bias, as this could exclude those who have not participated in the transfer, potentially affecting the generalizability of the study results [53]. For instance, if farmers perceive the land they are transferring to have lower quality, they might be more likely to implement protective measures. Conversely, if the land is already of high quality, farmers might reduce their protective actions. This heterogeneity could affect the accuracy of the estimates.

4.2.1. IV-2SLS Method

To overcome the aforementioned endogeneity issues, this paper applies the IV-2SLS method and endogenous switching model for comprehensive treatment. First, the paper uses an instrumental variable approach for two-stage estimation.

Drawing on existing research ideas [33,53], we use the average level of land transfer within the same village as an instrumental variable. The village-level average land transfer reflects factors such as the overall atmosphere of land transfer, policy support, and market activity in the region, all of which may influence individual farmers’ land transfer decisions, thus satisfying the relevance requirement for the instrumental variable. Regarding exogeneity, we argue that individual farmers’ land quality protection behaviors are primarily influenced by factors such as their own land management status, farming experience, and economic expectations, rather than the village’s average land transfer level itself. Therefore, the village-level average land transfer is not directly influenced by individual farmers and only affects their land quality protection behavior indirectly through its influence on land transfer decisions, thereby satisfying the exogeneity requirement. The results based on the IV-2SLS method are shown in

Table 3. Endogeneity Treatment I: IV-2SLS Method.

Variables	(1)	(2)	(3)
	Farmland_Transfer	Conservation Behavior	Conservation Awareness
	First Stage	Second Stage
IV	0.8869 ***
	(0.0617)
Farmland_Transfer		0.1802 ***	0.8034 ***
		(0.0428)	(0.1862)
Constant	2.3921 ***	1.1302 ***	2.6977 ***
	(0.7202)	(0.1067)	(0.4647)
Controls	Controlled	Controlled	Controlled
Regional FE	Controlled	Controlled	Controlled
F Statistic	206.623
Wu-Hausman		13.7631	9.8596
p-value		0.0002	0.0017
Observations	743	743	743
R-squared	0.2785	0.1367	0.1312

Note: *** represent significance at the 1% statistical levels, respectively, with the robust standard errors at the county level shown in parentheses.

.

In Table 3, Column (1) represents the first-stage estimation results, while Columns (2) and (3) represent the second-stage estimation results for farmers’ actual farmland quality protection actions and protection awareness, respectively. The first-stage regression results show that the average level of land transfer in the same village is highly correlated with individual farmers’ land transfer decisions, and the F-statistic is 206.623, which is greater than the critical threshold of 16.38, indicating that the instrument is exogenous. The results of the Wu-Hausman test its p-value indicate that the instrument variable used does not have endogeneity issues in theory and can effectively separate from the dependent variable, thus supporting the validity of the instrument variable employed in our analysis. The second-stage results show that land transfer can increase the probability of farmers’ actual farmland quality protection actions by 18.02% and raise their protection awareness by 0.8034 dimensions. After controlling for endogeneity, the research conclusions are consistent with the baseline regression, indicating that the findings of this study are robust.

4.2.2. Endogenous Switching Model

To further examine the potential sample selection bias caused by non-objective factors, such as the household head’s agricultural production experience and ability to access social resources, this study uses an endogenous switching regression model to address the issue. The results are shown in

Table 4. Endogeneity Treatment II: ESR Method.

Variables	(1)	(2)	(3)	(4)	(5)	(6)
	Selection Equation	Outcome Equation (Conservation Behavior)		Selection Equation	Outcome Equation (Conservation Awareness)
	Whether Transfer	Transfer_Y	Transfer_N	Whether Transfer	Transfer_Y	Transfer_N
IV	0.2494 ***			3.5476 ***
	(0.0019)			(0.2722)
ρ1		−0.1424 ***			−0.3883 ***
		(0.0231)			(0.0138)
ρ0			−0.1595 ***			−0.2777 ***
			(0.0315)			(0.0159)
Log likelihood		−321.5014			−1290.1862
Wald chi2		220.01 ***			406.16 ***
LR test		8.34 *			9.65 *
Controls	Controlled
Regional FE	Controlled
Observation	743

Note: ***, * represent significance at the 1% and 10% statistical levels, respectively, with the robust standard errors at the county level shown in parentheses.

. In Table 4, Columns (1) and (4) represent the selection equation, Columns (2) and (3) show the test results for the actual protection action equation, and Columns (5) and (6) display the test results for the farmland protection awareness equation. The Wald test results are significantly positive, and the correlation coefficients ρ1 and ρ0 are significantly different from zero. These test results indicate that using the endogenous switching model is appropriate.

To further explore the net effect of land transfer on farmers’ actual farmland protection actions and protection awareness, the average treatment effect for the group involved in land transfer (ATT), the average treatment effect for the group not involved in land transfer (ATU), and the average treatment effect for all samples (ATE) were calculated. The results are shown in

Table 5. Endogeneity Treatment II: Treatment Effect Analysis.

Treatment Effect	ATT	ATU	ATE
Conservation Behavior	0.5924 ***	0.4817 ***	0.1107 ***
	(0.0132)	(0.0182)	(0.0148)
Conservation Awareness	0.8245 ***	0.5446 ***	0.2799 ***
	(0.0077)	(0.0179)	(0.0194)

Note: *** represent significance at the 1% statistical levels, respectively, with the robust standard errors at the county level shown in parentheses.

.

From Table 5, it can be seen that the ATT results show that farmers who have participated in land transfer experience greater improvements in actual farmland protection actions and protection awareness compared to those who have not participated, indicating that land transfer has a more significant impact on farmers who are already involved. The ATU results indicate that for farmers who have not participated in land transfer, if they were to participate, their actual farmland protection actions and protection awareness would still significantly improve, though the effect is slightly lower than that of the treatment group. The overall effect (ATE) shows that land transfer still has a positive impact on all farmers. Therefore, this further confirms that the conclusions of this study are reliable.

4.3. Robustness Check

In the robustness check, we verify the reliability of the results by replacing the dependent variables. Specifically, at the level of actual farmland quality protection behaviors, we replace the original farmland protection actions with fertilizer usage (kg/ha) and pesticide spraying frequency (times) to examine changes in farmers’ inputs and behaviors in practice. At the cognitive level, we replace the original protection awareness variable with perceived reductions in pesticide and fertilizer usage and the perceived adherence to pesticide and fertilizer application according to instructions. These two indicators can more specifically reflect changes in farmers’ awareness of environmental protection and scientific management in agricultural production, especially their attention to reducing the use of fertilizers and pesticides and following pesticide instructions. The robustness check results are shown in

Table 6. Robustness treatment: replace the dependent variable.

Variables	(1)	(2)	(3)	(4)
	Conservation Behavior		Conservation Awareness
	Fertilizer_Usage	Pesticide_Sprays	Perception_Reduction	Perception_Application
	OLS	OLS	Order-Probit	Order-Probit
Farmland_Transfer	−8.8851 **	−3.5059 ***	0.3439 ***	0.2090 **
	(4.0411)	(0.6907)	(0.1007)	(0.0964)
Constant	68.7619 ***	25.0309 ***	-	-
	(20.5509)	(3.4652)	-	-
Controls	Controlled	Controlled	Controlled	Controlled
Regional FE	Controlled	Controlled	Controlled	Controlled
Observations	743	743	743	743
R-squared	0.0676	0.1809	-	-

Note: ***, ** represent significance at the 1% and 5% statistical levels, respectively, with the robust standard errors at the county level shown in parentheses.

.

Columns (1) and (2) in Table 6 show the regression results for the replacement of actual farmland quality protection behaviors, while columns (3) and (4) present the regression results for the replacement of farmland quality protection awareness. After calculation, the average amount of base fertilizer applied by sample farmers is 269.32 kg/ha, and the average number of pesticide applications is 12.46 times. From Table 6, it can be seen that land transfer leads to a reduction in the average fertilizer usage by 8.8851 kg/ha and a reduction in pesticide application by 3.5059 times, both of which are statistically significant at the 1% level. This suggests that land transfer helps reduce the use of fertilizers and pesticides, enhancing farmers’ actual farmland quality protection behaviors. Land transfer is associated with an average increase of 0.3439 units in farmers’ perception of reduced use of chemical fertilizers and pesticides, which is statistically significant at the 1% level. Land transfer is also associated with an average increase of 0.2090 units in farmers’ perception of applying chemical fertilizers and pesticides according to instructions, which is statistically significant at the 5% level. This result indicates that, after replacing the proxy variable for soil quality protection awareness, the research conclusion remains robust.

4.4. Mechanism Test

To further explore how land transfer affects farmers’ land quality protection behavior, this paper analyzes three mechanisms: income incentives, land transfer rent constraints, and social network interactions. The regression results of land transfer affecting farmers’ land quality protection behavior through the income incentive mechanism are shown in

Table 7. Mechanism Test I: Income Incentive Mechanism.

Variables	(1)	(2)	(3)
	Conservation Behavior	Agri_Income	Conservation Behavior
	Probit	OLS	Probit
Farmland_Transfer	0.4624 ***	0.6144 ***	0.4229 **
	(0.1047)	(0.0500)	(0.1965)
Agri_Income			0.2072 ***
			(0.0352)
Constant	4.4120 ***	0.1416	4.7856 ***
	(1.2116)	(0.3649)	(1.1639)
Controls	Controlled	Controlled	Controlled
Regional FE	Controlled	Controlled	Controlled
Sobel test Z			4.452
(p-value)			(0.000)
Observations	743	743	743
R-squared	-	0.2809	-

Note: ***, ** represent significance at the 1% and 5% statistical levels, respectively, with the robust standard errors at the county level shown in parentheses.

.

Column (2) of Table 7 presents the OLS model estimation results of the impact of land transfer on agricultural net income. The results show that the coefficient of land transfer is 0.6144 and is significantly positive at the 1% statistical level. This indicates that land transfer significantly increases farmers’ agricultural net income, providing economic incentives and resource support for farmers.

Column (3) of Table 7 shows that the coefficient of agricultural net income is 0.2072 and is significantly positive at the 1% statistical level. The Sobel Test result is significant, indicating that the income incentive mechanism plays a significant mediating role in the impact of land transfer on farmers’ land quality protection behavior, confirming that Hypothesis 2 is valid. Land transfer typically means that farmers gain additional agricultural output income, increasing their disposable income. This allows farmers to have more resources to improve land quality, such as purchasing better agricultural inputs or adopting advanced agricultural technologies.

Table 8. Mechanism Test II: Rent Constraint Mechanism.

Variables	(1)	(2)	(3)
	Conservation Behavior	Rent	Conservation Behavior
	Probit	OLS	Probit
Farmland_Transfer	0.4624 ***	2.5325 ***	0.4248 ***
	(0.1047)	(0.1117)	(0.1439)
Rent			−0.1904 ***
			(0.0679)
Constant	4.4120 ***	−1.2897	4.7471 ***
	(1.2116)	(0.8766)	(1.1709)
Controls	Controlled	Controlled	Controlled
Regional FE	Controlled	Controlled	Controlled
Sobel test Z			3.548
(p-value)			(0.000)
Observations	743	743	743
R-squared	-	0.6352	-

Note: *** represent significance at the 1% statistical levels, respectively, with the robust standard errors at the county level shown in parentheses.

shows the estimated results of land transfer affecting farmers’ land quality protection behavior through the rent constraint mechanism. From columns (2) and (3) of Table 8, it can be seen that the regression coefficient of land transfer on rental costs is 2.5325, and it is significantly positive at the 1% statistical level, indicating that land transfer significantly increases farmers’ rental costs. Additionally, the regression coefficient of rental costs on land quality protection behavior is −0.1904, and it is significantly negative at the 1% statistical level, suggesting that stronger rent constraints weaken farmers’ land quality protection behavior.

Moreover, the Sobel Test results are significant, indicating that the rent constraint mechanism plays a significant mediating role in the impact of land transfer on farmers’ land quality protection behavior, confirming that Hypothesis 3 is valid. This paper argues that although land transfer generally helps to promote land quality protection, the negative impact of higher rents suggests that higher rent might weaken farmers’ willingness to protect the land. Higher rents make farmers more inclined to pursue short-term profits, which may lead to more exploitative use of the land and reduce long-term sustainable protection actions. For example, farmers with higher rents might prefer intensive farming rather than adopting protection measures that benefit soil quality.

We found that land transfer can influence farmers’ land quality protection behaviors through both the income incentive mechanism and the rental cost pressure mechanism. However, the income incentive mechanism primarily enhances farmers’ awareness and actions regarding land protection by increasing their economic benefits. For farmers, higher income may motivate them to invest more resources in protective farming technologies. Meanwhile, the rental cost pressure mechanism affects farmers’ land protection behaviors by increasing the cost of land transfer, which could suppress their investment in land quality protection, especially in areas with higher rental costs. These two mechanisms each exert an independent influence, affecting farmers’ behaviors through their respective pathways. Therefore, we treat them as independent influencing factors, considering them separately in the model without further joint testing.

Table 9. Mechanism Test III: Social Network Interaction Mechanism.

Variables	(1)	(2)	(3)
	Conservation Behavior	Social_Network	Conservation Behavior
	Probit	Order-Probit	Probit
Farmland_Transfer	0.4624 ***	0.2670 ***	0.4277 **
	(0.1047)	(0.1013)	(0.1753)
Social_network			0.3197 ***
			(0.1017)
Constant	4.4120 ***	-	4.3931 ***
	(1.2116)	-	(1.1350)
Controls	Controlled	Controlled	Controlled
Regional FE	Controlled	Controlled	Controlled
Sobel test Z			4.2542
(p-value)			(0.000)
Observations	743	743	743

Note: ***, ** represent significance at the 1% and 5% statistical levels, respectively, with the robust standard errors at the county level shown in parentheses.

shows the regression results of land transfer influencing land quality protection behavior through farmers’ social network interactions. From columns (2) and (3) of Table 9, it can be seen that the estimated coefficient for land transfer on farmers’ social network interactions is 0.2670, which is statistically significant at the 1% level. This indicates that land transfer promotes farmers’ engagement in agricultural production activities through social network interactions. From column (3) of Table 9, it can be seen that the estimated coefficient for social network interactions on farmland quality protection behavior is 0.3197, which is statistically significant at the 1% level. This indicates that the more frequent the social network interactions, the stronger the farmers’ farmland quality protection behavior.

Furthermore, the Sobel Test results are significant, indicating that the social network interaction mechanism plays a significant mediating role in the impact of land transfer on farmers’ land quality protection behavior, confirming that Hypothesis 4 is valid. This paper argues that after land transfer, farmers, through more frequent neighborhood social network interactions, are more likely to obtain knowledge and technologies related to land protection, such as reducing the use of fertilizers and pesticides and practicing scientific crop rotation, thus improving their protection behavior.

4.5. Heterogeneity Analysis

In this section, the paper further explores the heterogeneous effects of farmland transfer on farmers’ cultivated land quality protection behaviors across different groups.

Farmers are categorized into part-time and full-time based on whether the household head is engaged in non-agricultural work. Additionally, farmers are classified into high and low agricultural knowledge literacy groups based on their responses to two specific questions:

Question 1: Are you able to distinguish between biological and conventional pesticides and understand their differences in usage? (1 = Yes, 0 = No)

Question 2: Can you apply agricultural knowledge in daily farming to improve farming methods or increase crop yield or quality? (1 = Yes, 0 = No)

If the farmer answers “1” to both questions, they are identified as having high agricultural knowledge literacy; otherwise, they are classified as having low literacy.

Table 10. Heterogeneity Analysis I: based on different part-time farming situations and agricultural knowledge levels.

Variables	(1)	(2)	(3)	(4)
	Conservation Behavior
	Part-Time_Farming	Full-Time_Farming	High_Literacy	Low_Literacy
Farmland_Transfer	0.5086	0.3970 ***	0.7623 ***	0.1909
	(0.3932)	(0.1069)	(0.2830)	(0.2882)
Constant	11.5106 ***	5.0445 ***	2.8224 *	11.7052 ***
	(2.2225)	(1.2765)	(1.4790)	(2.3217)
Controls	Controlled	Controlled	Controlled	Controlled
Regional FE	Controlled	Controlled	Controlled	Controlled
Observations	207	536	374	369

Note: ***, * represent significance at the 1% and 10% statistical levels, respectively, with the robust standard errors at the county level shown in parentheses.

presents the results of group differences based on household head’s employment status and agricultural knowledge literacy.

As shown in Table 10, compared with part-time farmers, full-time farmers exhibit a stronger positive response in cultivated land quality protection behaviors following farmland transfer, with an impact coefficient of 0.3970, which is significantly positive at the 1% statistical level. Moreover, farmers with higher agricultural knowledge literacy show even stronger protective behaviors after farmland transfer, with an impact coefficient of 0.7623, also significantly positive at the 1% level. This study argues that full-time farmers rely on agricultural production as their main source of income, making them more inclined to prioritize the long-term protection of farmland. Meanwhile, high-literacy farmers possess better agricultural knowledge and technical skills, enabling them to more effectively leverage the changes brought about by farmland transfer to implement quality protection measures.

Secondly, farmers are divided into two groups based on whether they acquire agricultural information and technical knowledge via the internet: one group obtains agricultural knowledge online, while the other does not. In addition, based on whether the farmer has experienced natural disasters, they are categorized into a disaster-experienced group and a non-disaster-experienced group.

Table 11. Heterogeneity Analysis II: based on different internet usage purposes and disaster experience situations.

Variables	(1)	(2)	(3)	(4)
	Conservation Behavior
	Internet	Non_Internet	Experienced_Disaster	Non_Experienced_Disaster
Farmland_Transfer	0.6162 **	0.3589	0.6828 ***	0.3091
	(0.2920)	(0.2244)	(0.1538)	(0.1994)
Constant	7.5956 ***	5.0203 ***	12.3660 ***	5.3995 ***
	(2.2351)	(1.3865)	(1.7880)	(1.3655)
Controls	Controlled	Controlled	Controlled	Controlled
Regional FE	Controlled	Controlled	Controlled	Controlled
Observations	296	447	208	535

Note: ***, ** represent significance at the 1% and 5% statistical levels, respectively, with the robust standard errors at the county level shown in parentheses.

presents the results of the heterogeneity analysis based on whether the household head obtains agricultural knowledge online and whether the household has experienced natural disasters.

As shown in Table 11, compared to farmers who do not acquire agricultural knowledge via the internet, those who do exhibit more proactive behaviors in farmland quality protection after land transfer, with a coefficient of 0.6162, which is significantly positive at the 1% statistical level. Similarly, farmers who have experienced natural disasters are more likely to engage in farmland quality protection after land transfer, with a coefficient of 0.6828, also significant at the 1% level. This study argues that farmers who access agricultural knowledge online likely have better access to information and can stay up to date with the latest agricultural technologies and environmental protection practices. After land transfer, they are more capable of putting farmland protection concepts into practice. Furthermore, experiencing natural disasters gives farmers a more direct and urgent awareness of the importance of protecting soil quality, making them more inclined to prioritize long-term land sustainability and restoration.

5. Discussion

5.1. Discussion of Results

(1)

Land Transfer Significantly Improves Farmers’ Cultivated Land Quality Protection Behavior.

This paper argues that land transfer not only leads to an expansion of operational scale but also, to some extent, incentivizes farmers to pay more attention to the sustainable use of land and to actively adopt cultivated land protection measures. This finding is consistent with the theory of economies of scale. As the area of cultivated land under management increases, farmers typically adopt several common agricultural conservation practices to ensure the long-term sustainability of soil quality. These practices include applying organic fertilizers, returning crop residues to the field, and implementing crop rotation or fallowing. These measures are widely regarded as agricultural conservation indicators, with the primary goals of improving soil fertility, reducing soil erosion, and enhancing the sustainability of agricultural production [55]. Moreover, the increase in long-term expected income brought about by land transfer may also prompt farmers to adopt more environmentally friendly agricultural practices to secure income stability in the future [56]. This finding has important implications for the optimization of China’s current land transfer management system. It suggests that while encouraging large-scale farming, there is also a need to further improve the land management system to ensure the sustainable use of land.

(2)

Mechanisms Through Which Land Transfer Affects Cultivated Land Quality Protection Behavior.

This study further reveals three key mechanisms through which land transfer affects farmers’ behavior in protecting cultivated land quality: the income incentive mechanism, the social network interaction mechanism, and the rent constraint mechanism.

First, the income incentive mechanism significantly promotes farmers’ behavior in protecting cultivated land quality [57]. After land is transferred in, the scale of operation expands, and agricultural income increases, thereby enhancing farmers’ expectations for the long-term use of the land and making them more willing to invest in land quality protection [10]. This aligns with the view that “large-scale land management promotes sustainable agricultural development” [58], suggesting that appropriate land transfer practices can contribute to the advancement of green agriculture.

Second, the social network interaction mechanism plays a positive role in this process. During land transfer, farmers often come into contact with more agricultural practitioners and form tighter social networks. This not only facilitates the dissemination of agricultural knowledge but may also influence farmers’ decision-making through the “demonstration effect”, making them more inclined to adopt scientific practices for land protection [59]. This mechanism implies that, in promoting land transfer, strengthening agricultural technical training and social network building can further improve farmers’ sustainable farming capabilities.

However, the rent constraint mechanism may suppress land protection behavior. The study finds that when farmers have to pay higher land rents, they may prioritize short-term profits and reduce long-term investments—such as applying less organic fertilizer or shortening fallow periods. This suggests that excessively high land rents could undermine the positive effects of land transfer on land protection [36]. Therefore, policymakers should establish relevant rules and regulations for the land rental market to ensure that farmers can continue investing in land quality protection while pursuing economic returns.

(3)

Group Differences in the Impact of Land Transfer on Cultivated Land Quality Protection Behavior

This study further analyzes the significant differences in how different types of farmers respond to land transfer regarding cultivated land quality protection. Compared to part-time farmers, full-time farmers are more dependent on agricultural operations and pay more attention to the long-term productivity of the land [60]. They are more inclined to adopt sustainable agricultural practices to ensure the stability of future income [61]. Farmers with higher agricultural knowledge are more likely to recognize the importance of land protection and master scientific farming methods. After acquiring more land, they are more willing to use advanced agricultural technologies for land protection [62]. The internet has increasingly played a role in the dissemination of agricultural information, helping farmers access the latest agricultural technologies and sustainable farming methods. Farmers who obtain agricultural knowledge online are more likely to adopt more proactive land protection behaviors after land transfer [63]. Experiences with natural disasters increase farmers’ risk awareness, making them pay more attention to land quality protection in order to reduce the uncertainty of future agricultural production [64]. This finding suggests that environmental risks significantly influence farmers’ behavior, and future policies can utilize this feature by introducing risk compensation or insurance mechanisms to guide farmers to enhance land protection.

5.2. Research Contributions and Limitations

Compared to existing research, the value of this study lies in the following aspects.

Previous studies on land transfer and farmland quality protection have primarily focused on macro-level policy design and technological pathways, with limited systematic analysis from the micro-behavioral perspective of farmers. Macro-level research typically concentrates on policy frameworks, technological innovations, and agricultural structural adjustments, with insufficient in-depth analysis of individual farmers’ decision-making, motivations, and actions in farmland protection [65]. Therefore, this study takes a micro-behavioral perspective to explore the specific impact of land transfer on farmers’ farmland quality protection behaviors, particularly through mechanisms such as income incentives, rent pressure, and social networks, thereby providing a new perspective that enriches the research on farmland quality protection. This study, using original survey data from the Jianghan Plain, deeply explores the specific impact of land transfer on farmers’ farmland quality protection behavior, helping to understand the intrinsic driving mechanisms of farmland quality protection at the micro behavioral level.

Second, it innovatively identifies and verifies the mechanisms of action. This study incorporates income incentives, rent constraints, and social network interactions into a unified analytical framework and systematically tests their transmission paths through mediation analysis. In particular, the identification of the social network mechanism provides a new perspective for understanding the impact of “non-market factors” on farmers’ decision-making behavior, which offers valuable insights for future related research.

Third, it reveals the heterogeneity of land transfer impacts across different groups. This study highlights the differences in the impact of land transfer on farmland quality protection behavior among different groups of farmers, based on factors such as professional identity, agricultural knowledge literacy, information access channels, and disaster experience. It provides more targeted reference for future policy-making and helps achieve more precise policies that cater to local conditions and guide protection behaviors accordingly.

Although this study thoroughly explores the impact of land transfer on farmers’ land quality protection behavior, there are still some limitations. First, the research area is relatively narrow, as it is based only on eight counties in the Jianghan Plain. The limited sample range may affect the generalizability of the results. Second, the study does not consider the nature of the contracts involved in land transfer, such as the long-term or short-term nature of land transfer contracts, which may have different impacts on farmers’ land quality protection behavior. Finally, the study measures farmers’ land quality protection behavior primarily through the dimensions of action implementation probability and protection awareness. However, land quality protection behavior is a comprehensive and diverse practice. Future research will expand the survey scope, deepen the study of the land transfer process, and design more comprehensive scales to quantify land quality protection behavior.

6. Conclusions

6.1. Research Conclusions

Land quality is the foundation of sustainable agriculture, and its protection is not only crucial for food security but also directly affects the long-term stability of the ecological environment and rural economy. Based on a micro-level survey data of 743 households from eight counties in the Jianghan Plain, this study explores the impact of land transfer on farmers’ land quality protection behavior, its mechanisms, and group differences. The findings are as follows.

First, land transfer significantly increases the probability of farmers’ land quality protection actions by 46.24% and improves their awareness of land quality protection by 0.3483 dimensions. This conclusion remains reliable even after addressing endogeneity using the IV-2SLS and ESR methods, as well as conducting robustness checks by replacing the dependent variables. It indicates that land transfer significantly enhances farmers’ land quality protection behavior.

Second, land transfer significantly improves farmers’ land quality protection behavior through income incentive mechanisms and social network interaction mechanisms, while it significantly reduces farmers’ land quality protection behavior through rent constraint mechanisms.

Third, the impact of land transfer on farmers’ land quality protection behavior exhibits group differences. For full-time farmers, those with higher agricultural knowledge literacy, those who access agricultural knowledge online, and those who have experienced natural disasters, participation in land transfer leads to more proactive land quality protection behavior.

6.2. Policy Implications

Based on the above research conclusions and discussion of results, this study proposes the following policy implications.

First, optimize the land transfer system to enhance the stability of land management. The study found that land transfer can significantly promote farmers’ land quality protection behavior. It is recommended to further improve rural land transfer regulations, encourage long-term land leasing contracts, and ensure that farmers have longer land management periods to strengthen their incentives for long-term investment and land protection. Additionally, establishing a land transfer registration and filing system can ensure the standardization and traceability of land transfers, reducing short-term and unstable transfer phenomena.

Second, regulate land rent reasonably to alleviate short-term profit pressure. The study found that excessively high land rents may inhibit farmers’ investment in land quality protection, leading them to focus more on short-term profits. Therefore, a land rent regulation mechanism could be considered. On the basis of market pricing, local governments could provide appropriate subsidies or financial support in areas with high rent levels to ease farmers’ management pressures.

Third, strengthen income incentives to guide farmers in long-term land quality protection. Increase agricultural subsidies and provide specific financial subsidies to farmers who implement green farming practices (such as straw returning, organic fertilizer application, etc.), thereby enhancing their economic motivation for land protection. Encourage the development of green agricultural finance by guiding banks and other financial institutions to offer low-interest loans or green credit support to help farmers invest in soil improvement and long-term land protection.

Fourth, leverage social network effects to enhance farmers’ awareness of land protection. Promote the demonstration role of rural agricultural cooperatives and specialized growers by organizing experience exchange meetings among land transfer farmers. This can encourage farmers to learn advanced farming techniques and sustainable agricultural models.

Fifth, provide differentiated support to different types of farmers to improve policy precision. For example, strengthen agricultural technical training and promote low-cost, high-efficiency land protection techniques suitable for small and medium-sized farmers to enhance their practical operational capabilities. Promote the application of “smart agriculture” by developing mobile apps or WeChat mini-programs related to land quality protection, offering online consultation, remote monitoring, and other services.