Labor Reallocation as a Mediating Channel: Farmland Transfer and Household Financial Vulnerability in Rural China

Abstract

The reallocation of production factors, particularly labor, is central to understanding economic development and household welfare. This paper investigates how the transfer of farmland, a fundamental shift in factor endowment, affects rural household financial vulnerability, with a specific focus on the mediating role of labor mobility. While factor market liberalization is theorized to enhance efficiency, the micro-level pathways through which land transactions influence financial resilience remain underexplored. Utilizing a unique household survey dataset from Shaanxi Province, China, and employing ordered Probit model alongside propensity score matching (PSM), the impact of farmland transfer-out on the financial vulnerability of rural households is revealed. The results show that farmland transfer-out significantly reduces household financial vulnerability. Mechanism analysis confirms that this effect operates primarily by releasing surplus agricultural labor and promoting its shift into non-farm employment, thereby expanding both the sectoral and geographic scope of household labor supply. Heterogeneity analysis further reveals that the responsiveness of labor mobility to land transfer is more pronounced among households with older heads, higher human capital, and stronger social networks. However, the ultimate mitigating effect on financial vulnerability is consistent across diverse household types. These findings contribute to the literature on factor market integration and household finance in developing economies and offer direct policy implications for designing land institutions and labor policies that synergistically enhance rural economic resilience.

1. Introduction

In recent years, the in-depth implementation of China’s rural revitalization strategy and urban–rural integration and development strategy has led to a significant transfer of farmland (X. Chen & Su, 2025). This transfer is a key initiative to activate “sleeping assets” in the countryside and enliven the rural factor market. It has profoundly influenced the allocation of resources and the ability to cope with risks of rural households (Fang et al., 2024). By July 2024, the total area of contracted rural land management rights in China had surpassed 570 million mu (approximately 38 million hectares), accounting for 29.4% of the total cultivated land area and 3.9% of the total land area of the country1. It reflects that the market-oriented allocation of land elements in China has formed a large scale in the agricultural field, and the impact of farmland transfer on farmers’ production and operation and family development is becoming more and more significant. While this shift has created new development opportunities for rural households, it has also introduced significant challenges (Jin et al., 2025). In China, rural land is collectively owned, and farmers hold only contractual management rights (the right to use and operate farmland) rather than full private ownership. In accordance with the provisions stipulated within the Rural Land Contract Law and the Civil Code, the sale or conversion of agricultural land to non-agricultural use is strictly prohibited. This measure is implemented with the objective of safeguarding the cultivated land redline and ensuring the national food security. However, the law explicitly permits the transfer, lease and subcontract of land management rights, thereby allowing farmers to transfer their operational rights to other households or business entities within a fixed term, provided that the agricultural use of the land remains unchanged and the transfer does not exceed the remaining period of the original contract. This legal framework has been demonstrated to ensure the stability of rural land tenure, whilst also providing the institutional basis for the ongoing expansion of farmland transfer. This policy background provides a solid foundation for the transfer of farmland and the rational allocation of land and labour. Despite the fluctuations and occasional decline in farmers’ agricultural income in recent years, there has been an observed increase in their emotional attachment to the land. This phenomenon has had a significant economic and socio-cultural impact on rural transformation (J. Liu et al., 2023). The transfer of agricultural land, as a significant form of rural factor reallocation, has had a considerable impact on rural households’ livelihood strategies and household welfare (H. Chen et al., 2025). Furthermore, it has also led to a reconfiguration of the income structure and risk characteristics of households through the reservoir effect of labour mobility (H. Wang et al., 2024a).

As the process of agricultural land transfer accelerates, the financial risk exposure faced by rural households has gradually emerged (M. Jiang et al., 2023). This is evidenced by insufficient social security coverage, increased volatility of non-farm employment, and the intertwining of credit constraints and financial exclusion. These factors have made financial fragility a significant hidden danger that restricts the stable development of households (X. Wang & Fu, 2021). The concept of financial vulnerability, as defined by Lusardi et al. (2011), pertains to the likelihood of a potential financial crisis arising from a household’s incapacity to settle its debts in a timely or comprehensive manner. This construct essentially encompasses the extent of financial risk accumulation and the capacity to manage adverse shocks. Consequently, financial vulnerability serves as a comprehensive indicator of rural household financial risk, as articulated by Faulkner et al. (2019). The vulnerability of rural households is not only related to the well-being of farm households, but also to the sustainable development of the rural economy and the comprehensive implementation of the rural revitalization strategy (Y. Zhou et al., 2025).

Existing studies have demonstrated that farmers’ decisions regarding the transfer of farmland are influenced by a multitude of factors (J. Chen et al., 2022; Quan et al., 2024), including farmland preservation (Qiu et al., 2020), pension income (Zhu et al., 2022), the ageing of farmers (J. Liu et al., 2023), and rural collective action (H. Chen et al., 2025). The mechanism through which these factors influence farmers’ decisions is characterized by differentiation. Meanwhile, factors such as carbon taxes (Faiella et al., 2022), the digital poverty (Ma & Liu, 2025), and population aging (W. Li et al., 2025b) have been demonstrated to increase household financial vulnerability. Conversely, social networks (B. Chen et al., 2024), digital credit (H. Wang et al., 2025) and special additional deductions policies (Y. Li et al., 2025) have been shown to reduce household financial vulnerability. The extant literature on this subject reveals the key triggers affecting farmland transfer and household financial vulnerability of farm households from the internal and external levels of households (Mutumba et al., 2025; F. Su et al., 2023; Yuan et al., 2025). The studies provide important references to promote moderate-scale agricultural operation and rural financial stability. However, they do not reveal the impact of farmland transfer on rural household financial vulnerability, especially lacking in-depth exploration of labour mobility as the core transmission mechanism.

Indeed, the farmland transfer frequently instigates a sequence of labour mobility chain reactions among rural households (Xu et al., 2024). That is to say, the farmland transfer not only releases the land’s constraints on the labour force and engenders favourable conditions for the participation of rural households in non-agricultural employment (B. Su et al., 2018), but also effects changes to the structure of the household’s income and economic sources (Kuang et al., 2021), and enhances the household’s ability to cope with shocks and risk management, which then affects the financial vulnerability of households. Therefore, the central question guiding this study is whether labour mobility can be considered a pivotal conduit through which the transfer of agricultural land exerts its influence on the financial vulnerability of rural households. The responses to this question will facilitate a more profound comprehension of the genuine impact of the transfer of farmland on the financial vulnerability of rural households. Furthermore, they will assist in elucidating the emphasis placed on the promotion of moderate-scale agricultural management and the stabilization of household finance from the perspective of labour mobility (Lu et al., 2019).

The existing research mainly discusses the influencing factors of farmers’ farmland transfer and how to alleviate the financial vulnerability of rural households. However, there is a general lack of in-depth analysis of the key intermediate mechanism of labor redistribution, and it has not yet clearly revealed how farmland transfer can change the specific path of farmers’ income structure and risk resistance by optimizing labor allocation, which constitutes the core literature gap that needs to be filled in this study. Consequently, it is imperative to investigate the impact of farmland transfer on rural household financial vulnerability from the perspective of labour mobility and to analyse its heterogeneity across diverse groups of rural households. The purpose of this paper is twofold. Firstly, it will explore the impact of farmland transfer on rural household financial vulnerability. Secondly, it will construct a theoretical and empirical analysis framework of “farmland transfer-labor reallocation-household financial vulnerability” from the perspective of labor mobility. In other words, we regarded labor redistribution as the core transmission mechanism of farmland transfer affecting farmers’ financial vulnerability. The novelty of this perspective is reflected in the following aspects. On the one hand, it breaks through the limitation that existing research only focuses on direct effects, and reveals the micro-logic of farmland transfer affecting farmers’ financial stability from the perspective of factor allocation. On the other hand, combined with the realistic background of China’s agricultural land transfer and labor mobility, it provides a new empirical basis for the linkage between agricultural land system reform and rural financial development in developing countries. Based on this, we utilized primary survey data collected by the research team in Zhouzhi County, Shaanxi Province. We employed empirical approaches, including ordered Probit model, propensity score matching (PSM), and mediation analysis, to comprehensively evaluate the impact of agricultural land transfers on rural household financial vulnerability. Potential endogeneity concerns are addressed using instrumental variable two-stage least squares (IV-2SLS) and conditional mixed process (CMP) frameworks. The findings aim to inform policy efforts aimed at optimizing the allocation of agricultural land resources, improving rural households’ financial well-being, and fostering stable and sustainable rural economic development.

2. Literature Review and Research Hypothesis

2.1. Rural Transformation, Structural Change and Farmers’ Financial Resilience

As a fundamental starting point for optimizing the allocation of rural production factors, the transfer of farmland factors is intrinsically linked to the financial vulnerability and financial resilience of farmers’ families. The quality of rural transformation and the sustainable development of farmers have become significant issues within the academic community. The extant research focuses on the driving logic of rural transformation and structural change, the influencing factors and mitigation paths of farmers’ financial vulnerability, and the economic effects and mechanism of agricultural land transfer. This provides a theoretical foundation and research reference for this paper, which studies the impact of agricultural land transfer behaviour on farmers’ financial vulnerability. Concurrently, it provides a framework for the present study to identify research lacunae and emphasize academic contributions.

From the standpoint of the primary agent of rural transformation and agricultural structure reform, farmland transfer and labour mobility are the two pivotal catalysts. The two factors are promoted in conjunction, with the objective of effecting the reallocation of rural production factors. This joint promotion serves as a significant manifestation of rural transformation. The reform of the system of property rights in farmland provides institutional guarantees for the transfer of farmland. The reform of rural land contract law has been instrumental in promoting a transformative shift in the agricultural structure, by virtue of the removal of restrictions on the transfer of farmland. In regions exhibiting low levels of industrialization and economic complexity, the impact of farmland transfer on the diversified and complicated development of the agricultural industry is more significant (J. Zhou et al., 2026). The magnitude of farmland transfer has further promoted the scale and specialization of agricultural production and operation. Following the transfer of cultivated land, there has been a marked enhancement in the profit orientation of new agricultural management entities. As postulated by Zheng et al. (2026), the alteration in the price of farmland exerts a direct influence on the decision-making processes pertaining to production, the adoption of technology and the manner in which farming is conducted. Furthermore, it is posited that this phenomenon serves to catalyse a structural transformation in the domain of agricultural production.

Concurrently, the transfer of agricultural land also releases space for the non-agricultural transfer of rural labour force. The transfer of non-agricultural labour force has been demonstrated to have a significant positive correlation with the improvement of the ecological efficiency of cultivated land use. Furthermore, it has been shown to promote the optimization of the structure of agricultural production factors through the factor substitution effect of agricultural machinery. The threshold effect of agricultural land management scale also corroborates the necessity of agricultural land and labour transfer to promote large-scale land management (W. Li et al., 2025a). Moreover, the development of digital technology and infrastructure has become a significant catalyst for rural transformation and structural change. The proposed strategy aims to enhance the efficiency of rural labour, land and capital allocation through the facilitation of enhanced information acquisition, the strengthening of social networks, and the relaxation of credit restrictions. The transfer of agricultural land has been shown to form a two-way relationship with the development of digital agriculture. The transfer of agricultural land has been demonstrated to promote the development of digital agriculture by transferring agricultural labour and increasing digital capital investment. The digital capabilities of farmers have been demonstrated to encourage the large-scale operation of agricultural land. This is achieved by expanding the trading radius and improving production efficiency. As demonstrated in the relevant literature, the two entities under discussion promote the digital transformation of agriculture and rural areas (X. Zhou et al., 2025; Xiao et al., 2026; Jin et al., 2025). The extant research has clarified the fundamental role of farmland transfer in the process of rural transformation and structural change. Furthermore, the synergistic effect of farmland transfer with labour mobility and digital technology application has become the key driving force to promote the transformation of agricultural production structure and factor structure.

In consideration of the factors that influence farmers’ family financial vulnerability and the strategies for enhancing financial resilience, extant research has identified numerous pivotal variables, including digital factors, social capital, financial literacy, and risk shocks. These studies have elucidated the mechanism by which these factors impact farmers’ financial status within the context of rural transformation. With regard to digital factors, the effects of digital poverty and digital ability on farmers’ financial status are contradictory. The presence of digital poverty has been demonstrated to exacerbate household financial vulnerability by increasing financing restrictions, weakening social interaction and reducing future expectations (Ma & Liu, 2025). This effect is more significant in urban residents and households in the eastern region. Digital proficiency has been demonstrated to enhance familial financial well-being through the medium of enhanced information literacy, diversified livelihoods, and risk mitigation (M. Li & Zhang, 2025). It has been further demonstrated that digital proficiency plays a more protective role in the event of risk shocks, exerting a more pronounced effect on families with a low level of education and medium financial knowledge.

Digital capabilities have been demonstrated to enhance the development resilience of rural households through a variety of means. These include the acquisition of information, financial inclusion, income growth and asset accumulation (H. Wang et al., 2025). This is especially the case for vulnerable groups with limited human and social capital. Furthermore, the presence of digital infrastructure has been shown to increase the opportunity for farmland transfer through the ‘income effect’. It has also been demonstrated that digital infrastructure can promote the non-agricultural employment and financial literacy of farmers through the ‘security effect’. Ultimately, this can result in enhanced financial resilience of rural households (Deng et al., 2025). In terms of social capital and financial literacy, social networks have been shown to play a key role in helping families to avoid financial vulnerability and to provide support for family finance through a range of information channels. These channels include risk preference, financial literacy and job stability, which are particularly important in times of economic challenge (L. Zhang et al., 2026). Financial literacy has been identified as a key driving force for farmers’ financial resilience (C. Chen et al., 2024). The acquisition of formal credit has been demonstrated to have a positive regulatory effect on the role of financial literacy in promoting entrepreneurial resilience (Xiang et al., 2026). In contexts characterized by fragility, the dissemination of financial knowledge emerges as a pivotal factor in enhancing social and economic resilience (Potrich et al., 2025). Furthermore, extant research has identified the impact of farmers’ own characteristics and external risk shocks on household livelihoods and financial vulnerability. As X. Liu et al. (2024) and Bhatnagar et al. (2026) have demonstrated, a range of factors have a significant impact on farmers’ livelihoods. These include labour capacity, agricultural insurance, agricultural production losses and natural risks caused by climate change. This research provides a reference for exploring the mitigation effect of farmland transfer on farmers’ financial vulnerability from a risk perspective.

The extant literature on the correlation between farmland transfer and farmers’ family development resilience has, to date, touched upon the impact of farmland factor allocation on farmers’ development ability, albeit with a greater focus on the effect of farmland inflow. Nevertheless, there remains a paucity of research directly correlating farmland transfer and farmers’ financial vulnerability. Research has demonstrated that the inflow of farmland has the capacity to substantially augment the resilience of rural households. Notwithstanding the initial economic encumbrance that will accompany the inflow of farmland, it is hypothesized that this will, in the long term, promote the development of farmers through the return of the labour force and economies of scale. This effect is predicted to be more pronounced in non-vulnerable households with abundant water resources and strong collective consciousness (Lang et al., 2025). Farmland transfer is an institutional arrangement that has the potential to influence the transformation of agricultural structure and the development of digital agriculture. Furthermore, it can also have an indirect effect on farmers’ income and asset accumulation. This is achieved through the synergy with digital infrastructure and non-agricultural employment, which has the capacity to alleviate the financial vulnerability of farmers. However, the majority of extant research focuses on the agricultural production effect and rural transformation effect of farmland transfer. A paucity of discussion has been observed with regard to the internal mechanism through which farmland transfer affects the financial vulnerability of farmers’ households, with labour mobility serving as the core intermediary in this process. Furthermore, it does not fully integrate the background of rural transformation and structural change in order to analyse the heterogeneous impact of farmland transfer on farmers’ financial vulnerability according to different characteristics.

2.2. The Impact of Farmland Transfer-Out on the Financial Vulnerability of Rural Households

(1)

The theoretical analysis of farmland transfer-out inhibiting the financial vulnerability of farmers’ households

Drawing on the Lewis Dual Economic Structure Theory, this study contextualizes farmland transfer within the framework of China’s structural transformation. Lewis (1954) highlights that developing economies feature a traditional agricultural sector characterized by surplus labor and a modern industrial sector with higher wages. In China, the transfer of farmland management rights serves as a key mechanism to unlock the hidden surplus labor in the countryside. By facilitating the flow of labor from the low-efficiency agricultural sector to the high-efficiency non-agricultural sector, farmland transfer helps improve household welfare and manage financial risks, providing a theoretical basis for examining its impact on household financial vulnerability. In this context, the surplus labour in agriculture is continuously transferred to the industrial sector (Huang et al., 2023). It is therefore posited in this paper that the income of a household comprises two distinct components: agricultural income and non-agricultural income (Mat et al., 2012). It is evident that the agricultural production function is defined as follows: Y_a = f(L_a, K_a). In this model, Y_a denotes agricultural income, L_a is the number of laborers invested by the farm household in agricultural production, and K_a is the land, farm tools and other factors invested by the farm household in agricultural production. The non-agricultural production function is represented by Y_n = g(L_n, K_n). In this model, Y_n denotes non-agricultural income, L_n represents the number of laborers engaged by farmers in non-agricultural production, and K_n signifies the skills, capital and other factors invested by farmers in non-agricultural production. Notably, neither K_a nor K_n includes human capital, which is instead incorporated into the empirical analysis through control variables and heterogeneity tests.

It is assumed that farmers are faced with the dual choice of farmland transfer (FT = 1) and non-transfer (FT = 0), and their decision-making follows the principle of maximizing expected utility. The expected utility of farmer i in two states is defined as U_i1(X_i) and U_i0(X_i), where X_i is the individual and family feature vector that affects utility (such as age, health status, education level, cultivated land scale, etc.). The decision-making rules of farmers can be expressed as: U_i1(X_i) − U_i0(X_i) > 0, that is, when the expected utility increment brought by farmland transfer is positive, farmers will choose to transfer farmland; otherwise, it will not be transferred out. This decision-making framework constitutes the micro-theoretical basis for the subsequent empirical identification of this paper. It is hypothesized that farmers can obtain R income through farmland transfer-out, which is equal to the area of farmland transfer multiplied by the rent per unit area (Ji et al., 2025). At this time, the total household income of farmers is as follows: Y is defined as the total income of the rural households, which comprises Y_a, Y_n and R. It is hypothesized that the income risk faced by rural households obeys the normal distribution, and the variance ${\sigma }^2$ is representative of the degree of household financial vulnerability2.

It is evident that as the magnitude of the variance increases, the financial vulnerability of rural households concomitantly rises. Prior to the transfer of farmland, the distribution of the household labour force is delineated by the axes of agricultural ($L_a^0$) and non-agricultural ($L_n^0$), that is, ($L=L_a^0+L_n^0$). Therefore, the aggregate household income of agriculturalists is designated as $Y^0=f\left(L_a^0,K_a\right)+g\left(L_n^0,K_n\right)$, the income variance is ${\sigma }_0^2$, and ${\sigma }_0^2=p_a^0{\sigma }_a^2+p_n^0{\sigma }_n^2$, where $p_a^0$ and $p_n^0$ represent the ratio of agricultural income and non-agricultural income to the total household income, respectively. After transferring out of farmland, the farm household will obtain the farmland transfer revenue R, and will also reallocate labor resources to the non-agricultural sector (B. Su et al., 2018; X. Li & Huo, 2021; Xu et al., 2024). This will maximize the household income and minimize the risk. It can be demonstrated that, at this time, $L_a^1<L_a^0$, $L_n^1>L_n^0$ and $L=L_a^1+L_n^1$. The total income of the farm household is $Y^1=f\left(L_a^1,K_a\right)+g\left(L_n^1,K_n\right)+R$. In accordance with neoclassical economic growth theory, it can be seen that the agricultural sector, due to the restrictions imposed by natural conditions such as land, the rate of technological progress and capital accumulation is relatively slow, and labour productivity and income level are relatively low (Z. Jiang et al., 2025). By contrast, the non-agricultural sector is more likely to realize technological innovation and improve production efficiency. This is due to the accumulation of capital and technological progress. The result is an increase in output and income. Consequently, farmers who transfer-out of farmland and reallocate their labour resources to the non-farm sector will earn relatively higher output and income (Imai et al., 2015). It can be concluded that $g\left(L_n^1,K_n\right)-g\left(L_n^0,K_n\right)>f\left(L_a^0,K_a\right)-f\left(L_a^1,K_a\right)$, that is, $Y^1>Y^0$. It is important to note that agricultural production is highly vulnerable to natural factors and income fluctuations. In contrast, non-farm income tends to be more stable. Following the relocation of farmers to other areas, the proportion of unstable agricultural income in their household income structure is expected to decline, while stable non-agricultural income and farmland transfer income are projected to increase (S. Wang et al., 2025).

Let ${\sigma }_a^2$ and ${\sigma }_n^2$ denote the variances of agricultural and non-agricultural income, respectively, and ${\sigma }_1^2=p_a^1{\sigma }_a^2+p_n^1{\sigma }_n^2+p_R{\sigma }_R^2$, where $p_a^1,p_n^1$ and $p_R$ denote the ratios of agricultural income, non-agricultural income, and farmland turn-out gains to total income, respectively. Given the established relationship between $p_a^1<p_a^0$ and $p_n^1>p_n^0$, and the commonly observed stability of farmland transfer-out gains, therefore, ${\sigma }_R^2\approx 0$, ${\sigma }_1^2<{\sigma }_0^2$, and it can be posited that farmers seek to optimize the structure of household income following the transfer of farmland by adjusting the allocation of household labour resources (Cui et al., 2025). This adjustment is theorized to reduce household income risk, thereby mitigating the financial vulnerability of households (Bai et al., 2024). Based on the theory of Financial Vulnerability, it becomes evident that the unique source of income for farming households’ families can elevate their financial vulnerability. Conversely, the diversification of income sources has been demonstrated to assist in mitigating this vulnerability. The transfer of farmland has been demonstrated to facilitate the mobility of labour within agricultural households, thereby effecting a transition from a single-structure household economy dominated by agriculture, to a more diversified structure incorporating agriculture, non-agriculture, and income derived from the transfer of farmland (Song et al., 2024; Xu et al., 2024). This transition has been shown to enhance the household’s income, ultimately mitigating financial vulnerability (X. Li & Huo, 2021).

(2)

Literature review on the inhibition of rural households’ financial vulnerability by farmland transfer-out

In the context of the conventional agricultural production model, the economic well-being of agricultural households is predominantly contingent on the extent of agricultural output (Nzira & Chegere, 2025). However, agricultural production is susceptible to constraints imposed by numerous factors, including natural conditions and technological limitations, thereby impeding the attainment of expeditious income growth (Shen et al., 2025). Furthermore, the price of agricultural products is susceptible to fluctuations in market supply and demand, thereby exacerbating the instability of farmers’ household income (Gohin & Zheng, 2020). Following the relocation of farmers from agricultural land, there is an observed decline in their reliance on agricultural production, as well as a concomitant reduction in the risk of significant fluctuations in income due to natural disasters and market fluctuations in agricultural product prices. For instance, in the event of natural disasters such as droughts and floods, farmers who have not transferred their land may experience a reduction in crop yields or even crop failure, which can result in increased financial distress (Lin & Wang, 2024). Conversely, farmers who have transferred their land can increase their household incomes through other means, thereby reducing the risk of financial vulnerability.

Following the transfer of farmland, farmers have the option to obtain a direct form of compensation for the transfer. This compensation can be received in the form of rental income or other forms of compensation (L. Zhang et al., 2018). Furthermore, these funds can be invested in other production and business activities or investment areas. This process has the effect of enhancing the liquidity and efficiency of the use of household assets and narrowing the income gap between farming households (F. Zhang et al., 2025). When evaluating the creditworthiness of farm households, financial institutions typically accord greater significance to the comprehensive income sources of such households and their degree of stability (Borges et al., 2025). In summary, the transfer-out of farmland by farmers has been demonstrated to promote household wealth accumulation. This, in turn, has been shown to enhance household economic strength and risk resistance, thereby reducing the risk of falling into financial vulnerability due to insufficient or unstable income. The transfer-out of farmland has also been shown to effectively mitigate household financial vulnerability. Therefore, we puts forward hypothesis 1:

Hypothesis 1.

Farmland transfer-out positively alleviates the financial vulnerability of rural households.

2.3. Farmland Transfer, Labor Mobility and Financial Vulnerability of Rural Households

Farmland is the core and most important means of production for farmers, and farmland transfer is conducive to the realization of moderate and large-scale agricultural operations. In the context of agricultural mechanization and large-scale production, agricultural production efficiency has been significantly improved, but also released a large number of rural labor (Hao et al., 2025). Farmland transfer can promote the optimization of labor allocation for farmers, and promote the transfer of family labor to non-agricultural employment, which is conducive to becoming a key point for the growth of farmers’ income (Han et al., 2024). The fragmentation of farmers’ farmland has a detrimental effect on the scale of production, which is consequently limited, resulting in low productivity (Lu et al., 2018). In the context of agricultural labour, if the marginal gain from labour mobility exceeds the marginal gain from agricultural labour, given the allocated labour time, farmers will continually transfer their labour from agricultural production to non-agricultural employment (Kafle et al., 2026). In the process of labour transfer, farmers will elect to transfer the right to utilize agricultural land, thereby achieving a rational allocation of labour resources, until the marginal returns of labour allocation are equal in agricultural labour and non-agricultural labour (Y. Zhang & Lerman, 2025). When farm households opt to transfer their agricultural land, they can obtain property income, such as land rent, and also transfer surplus labour within the household to non-farm industries, thereby obtaining wage income. This enhances their ability to withstand financial vulnerability.

In comparison with agricultural production, non-agricultural industries have been shown to offer higher wage incomes, more stable sources of income and greater growth potential for rural households (Zhan et al., 2025). These incomes are more readily realizable and utilizable for emergency situations, in contrast to incomes from agricultural production, which have been found to be more challenging to liquidate expediently (Shen et al., 2025). This scenario affords agriculturalists greater adaptability in the face of uncertain market fluctuations, thereby mitigating the probability of financial distress arising from liquidity constraints (Yuan et al., 2025). Consequently, for farm households that have transitioned out of farmland, the decline in investment in agricultural production and the rise in non-farm employment opportunities can contribute to an augmentation in household liquid assets and a diminution in household indebtedness (R. Li et al., 2025), thereby decreasing household debt risk. Based on the above analysis, we proposes hypothesis 2.

Hypothesis 2.

Farmland transfer-out reduces the financial vulnerability of rural house holds by promoting labor mobility.

2.4. Research Review

Existing research has constructed an analytical framework of “rural transformation-allocation of production factors-resilience of farmers’ development,” clarified the core role of farmland transfer and labor mobility in rural structural change, and also identified multiple influencing factors of farmers’ financial vulnerability and resilience, but there are still the following research gaps: First, there is insufficient discussion on the direct causal relationship between farmland transfer and farmers’ financial vulnerability. The existing research focuses more on the economic effects of farmland inflow, and pays less attention to the effects of farmland transfer on releasing labor and alleviating farmers’ financial risks. Secondly, it does not fully reveal the intermediary mechanism of farmland transfer affecting farmers’ financial vulnerability, especially the lack of empirical test on the core transmission path of ‘farmland transfer→labor mobility→farmers’ financial vulnerability mitigation’. Third, under the background of rural transformation and structural change, the differential performance of the impact of farmland transfer on farmers’ financial vulnerability has not been analyzed in combination with the heterogeneity characteristics of farmers. Based on this, this paper takes rural transformation and agricultural structure change as the research background, takes farmland transfer behavior as the core explanatory variable, focuses on the transmission mechanism of farmland transfer→labor flow→household financial vulnerability, and combines the age, human capital and social network dimensions of the head of household to carry out heterogeneity analysis, aiming to make up for the existing research gap, provide empirical evidence for the optimization of rural production factor allocation and the prevention and control of farmers’ financial risks, and also provide policy reference for promoting the deep integration of rural transformation and rural revitalization.

3. Research Design

3.1. Data Sources

The data for this study were obtained from a specialized survey conducted by the research team in July 2024 in Zhouzhi County, Xi’an City, Shaanxi Province. Zhouzhi County functions as a traditional agricultural hub, with key industries including kiwifruit cultivation and grain production (Guo et al., 2022; J. Zhou et al., 2023). Its pronounced agricultural orientation, high degree of marketization in farmland transfer, and rural households’ strong dependence on agriculture for income render it a well-suited context for examining how farmland transfers influence rural households’ financial vulnerability. Meanwhile, as the capital of Shaanxi Province, Xi’an exhibits frequent cross-flow of urban and rural production factors. Zhouzhi County, an agricultural county under Xi’an’s administration, benefits from urban spillover effects, which have promoted substantial agricultural land transfers and diversified transfer arrangements. This setting aligns well with the research needs for measuring agricultural land transfers and labor mobility. Furthermore, situated in the Guanzhong Plain, Zhouzhi County has a large agricultural population but limited arable land per capita. Its proximity to Xi’an’s urban core and accessible transportation have fostered two distinct labor migration patterns: (a) seasonal out-migration for urban employment, with periodic return for farming activities, and (b) household labor splitting, where some members migrate on a long-term basis while others remain to manage farmland. This diversity in migration trajectories allows for differentiation between distinct mobility scenarios, thereby facilitating investigation of the causal pathways through which labor mobility interacts with farmland transfer and rural household financial vulnerability.

To ensure the representativeness and reliability of the survey data, the research team conducted systematic training for enumerators prior to fieldwork. A combined sampling approach, incorporating stratified, typical, and random sampling, was adopted for survey participant selection. First, following statistical sample size determination principles and taking into account the agricultural economic map of Zhouzhi County as well as the coverage of pilot policies, six representative townships/subdistricts (Erqu Subdistrict, Cuifeng Town, Louguan Town, Mazhao Town, Situn Town, and Yabo Town) were selected to ensure township-level representativeness and heterogeneity. Next, villages within each township were stratified by farmland transfer rate and agricultural production scale to ensure full tier coverage. Finally, based on village size and agricultural development status, 10 to 70 households were randomly chosen from each village for questionnaire surveys. This procedure helped avoid over- or under-representation of villages of different sizes, resulting in 778 individual survey returns. The survey focused on basic household characteristics, loan experience evaluations, policy implementation status, and future financing needs. After removing samples with significant missing data or logical inconsistencies, 746 valid household head questionnaires were retained, yielding a valid response rate of 95.89%.

3.2. Variables

3.2.1. Core Explanatory Variable

The key explanatory variable, farmland transfer, is a dummy variable that equals 1 if the household transferred out farmland, and 0 otherwise (Tan et al., 2022; Fang et al., 2024; H. Chen et al., 2025). In other words, if the area of cultivated land transferred out by a household is greater than zero, it is considered evidence of transfer behavior, and the variable is assigned a value of 1; otherwise, it is assigned 0. In the baseline regression, this transfer behavior serves as the core explanatory variable. For robustness testing, the total area of cultivated land transferred out by households is used as an alternative measure to replace the binary transfer indicator.

3.2.2. Dependent Variable

The dependent variable in this paper is the financial vulnerability of peasant households. The financial vulnerability of rural households can be considered a measure of their family risks, reflecting the ability of rural households to cope with risk shocks (Imai et al., 2015). It is evident that a strong risk response capacity exhibited by farmers’ families is associated with a diminished financial vulnerability. As previously outlined by Noerhidajati et al. (2021), this study utilizes a methodology that employs dummy variables to assess the financial vulnerability of rural households. The assessment encompasses three distinct dimensions: Liquidity, emergency response capacity and debt level (Faulkner et al., 2019; H. Wang et al., 2024b; Yuan et al., 2025). The level of household liquidity is measured by the difference between household income and expenditure. That is to say, it is determined by the farmer’s surplus after the deduction of expenditure over the past six months. In the absence of surplus, the value is assigned to 1; otherwise, the value is assigned to 0. Concurrently, the family emergency response capacity is measured by the difference between the financial assets of the farmer’s family and the household expenditure of the farmer’s family over the past three months. Should the difference between the two be found to be less than zero, this would indicate an inadequate emergency response ability on the part of farmers’ families, with a value of 1 assigned to this category, and a value of 0 assigned to the alternative. If the income level of the peasant household is insufficient to cover all of its liabilities, the household is designated as an over-debt family, with a value of 1 assigned to this category, and a value of 0 assigned to the alternative. The present study proposes an aggregation of the aforementioned indicators to derive a composite measure of financial vulnerability for rural households. The value range is from 0 to 3, which indicates an increase in the financial vulnerability of rural households.

3.2.3. Mechanism Variables

The mechanism variables encompass the scope of labour mobility and the mobility of rural households in regard to employment. In relation to the measurement of labour mobility, this paper considers farmers who engage in non-agricultural labour for a proportion of the year and agricultural labour for the remainder of the year to be labour mobility families (L. Li et al., 2022; H. Wang et al., 2024a). The value assigned to this category is 1, while the value assigned to the other category is 0. Concurrently, the flow range is divided according to the labour flow area of farmers. In the event that the farmer is engaged in non-agricultural labour outside the province, this is regarded as the flow outside the province, and the assignment is assigned a value of 1. Otherwise, the assignment is assigned a value of 0 (Lu et al., 2019; Xu et al., 2024).

3.2.4. Control Variables

In addition to the aforementioned variables, the financial vulnerability of rural households is also influenced by a number of observable variables (Noerhidajati et al., 2021; C. Li et al., 2025). It is evident that in order to enhance the precision and reliability of the model estimation outcomes, and to circumvent the estimation bias that arises from the exclusion of salient variables, we undertook the control of other factors that have the potential to influence the financial vulnerability of rural households (Singh & Malik, 2022). These factors encompass the personal characteristics of the head of household (age, gender, health status and education level), family characteristics (family business type, credit demand) and agricultural production and operation status (cultivated land scale, agricultural technology training). Furthermore, in order to control for the impact of regional financial environments on rural households’ financial vulnerability, we also incorporated a variable measuring the transportation conditions required to access financial institutions.

3.3. The Descriptive Statistical Results

Table 1. Descriptive statistics.

Variables	Abbrev	Variable Assignment	Full Sample	Experimental Group (A)	Control Group (B)	A-B
Farmland transfer	FT	Farmland not transferred out = 0, farmland transferred out = 1	0.12	1	0	1
Financial vulnerability	FV	No vulnerability = 0, low vulnerability = 1, medium vulnerability = 2. High vulnerability = 3	1.61	0.75	1.72	−0.97 ***
Labour mobility	labor	Unflowed labor force = 0, flow = 1	0.36	0.61	0.33	0.28 ***
Scope of labour mobility	scope	Intra-provincial flow = 0, inter-provincial flow = 1	0.12	0.25	0.11	0.14 ***
Age	Age	The age of the head of household (years)	52.79	49.70	53.20	−3.50 **
Age squared	Age2	Age square of head of household	2995.16	2698.41	3034.33	−335.92 **
Gender	Gender	Female = 0, male = 1	0.89	0.87	0.90	−0.03
Health status	Health	No labor ability = 1; have disease, can not do farm work = 2, can only do simple farm work = 3; can do most of the agricultural work = 4; complete labor ability = 5.	4.07	4.38	4.03	0.35 ***
Educational level	Edu	Never been to school = 1, primary school = 2, junior high school = 3, senior high school = 4. College degree and above = 5	3.04	3.40	3.00	0.40 ***
Livelihood strategy	Livelihood	Pure agriculture = 1; agriculture-based = 2, non-agricultural-based = 3; completely non-agricultural = 4	2.93	3.25	2.89	0.36 ***
Credit demand	Credit	In the next three years, no borrowing demand = 0, borrowing demand = 1	0.15	0.15	0.15	0.00
Cultivated land scale	Scale	Farmers’ actual cultivated land area (mu), take the logarithm	2.13	2.14	2.12	0.02
Agricultural technology training	Training	Not participated = 0, participated = 1	0.54	0.64	0.53	0.11 **
Financial transportation convenience	Transportation	Very inconvenient = 1, less convenient = 2, general = 3. More convenient = 4, very convenient = 5	3.82	4.01	3.80	0.21 **

Note: ** and *** respectively indicate that the mean difference in variables between the experimental group and the control group is significant at the level of 5% and 1%.

presents the descriptive statistics for the variables under study.

As shown, 12% of the sample households participated in agricultural land transfer (the experimental group), while the remaining 88% did not (the control group). To further clarify the real-world distinctions between these two groups, an independent-samples t-test was conducted using StataMP 18 (with the ttable2 command) to examine mean differences across all variables. From the perspective of financial vulnerability, the average financial vulnerability of the entire sample of farmers is 1.61, which is between low vulnerability and moderate vulnerability. The mean financial vulnerability of farmers in the experimental group who underwent farmland transfer was found to be 0.75, a figure which is closer to the ‘low vulnerability-no vulnerability’ interval. By contrast, the mean financial vulnerability of farmers in the control group who did not undergo farmland transfer was 1.72, a figure which is closer to the ‘low vulnerability-moderate vulnerability’ interval. The mean difference between the experimental and control groups was −0.97, indicating that farmers who have transferred their farmland have a significantly lower level of financial vulnerability than those who have not. This difference was found to be statistically significant at the 1% level. The evidence presented indicates that the transfer of farmland can potentially mitigate the financial vulnerability experienced by farmers.

The results also indicate that, except for gender, credit demand, and farmland size, mean differences for all other variables are statistically significant at the 1% or 5% level, suggesting systematic disparities between the experimental and control groups. Specifically, households in the experimental group generally exhibited lower financial vulnerability, higher labor mobility, younger household heads, better health status, and higher educational attainment compared to those in the control group. In terms of household livelihood characteristics, experimental group households were primarily engaged in non-agricultural activities, whereas control group households mainly relied on agriculture. Regarding agricultural production and management, no significant difference in cultivated land size was observed between the two groups, indicating comparable land endowments. Additionally, experimental group households demonstrated greater participation in agricultural technology training and perceived transportation to financial institutions as more convenient.

3.4. Empirical Strategy

Given that the financial vulnerability of the explained variable is an ordered categorical variable, this study employs the ordered Probit model to analyse the impact of farmland transfer on the financial vulnerability of farmers’ households. This regression serves not only to provide initial evidence on this relationship but also to inform the selection of covariates for the subsequent PSM analysis. This approach helps control for potential confounders during the matching process, thereby improving the reliability and accuracy of the estimated treatment effects. The ordered Probit model is specified as follows:

$$F{V}_i^{*}={\alpha }_0+{\alpha }_{1}F{T}_i+{\alpha }_2{X}_i+{\epsilon }_i$$

(1)

$F{V}_i^{*}$ represents the potential tendency of the financial vulnerability of rural households, which is an unobservable, continuous, latent variable, and the conversion relationship with the actually observable financial vulnerability, Fin_i satisfaction (2). In formula (2), Fin_i is an observable financial vulnerability of rural households, which is determined by $Fi{n}_i^{*}$ and threshold μ₁, μ₂, μ₃. Finally, the influence of core parameters is estimated by maximum likelihood method.

$$F{V}_i=\left\{\begin{array}{l}0,F{V}_i^{*}\le {\mu }_1\\ 1,{\mu }_1<F{V}_i^{*}\le {\mu }_2\\ 2,{\mu }_2<F{V}_i^{*}\le {\mu }_3\\ 3,Y_i^{*}>F{V}_3\end{array}\right.$$

(2)

In Equation (1), FT_i is a binary indicator of whether household i engaged in farmland transfer-out; X_i represents a vector of control variables; α₀ is the constant term; α₁ and α₂ are parameters to be estimated; and ε_i is the random error term. The primary coefficient of interest is α₁. Should α₁ be negative and statistically significant, it would suggest that farmland transfer-out contributes to reducing rural households’ financial vulnerability. Building on this baseline analysis and considering that farmers’ farmland transfer-out behavior may have self-selection bias, we further examined the impact of farmland transfer using PSM to enhance causal inference. The following core assumptions underpin this method: Firstly, the ignorability assumption must be considered. Once all observable confounding variables have been controlled, it can be assumed that farmers’ decisions regarding the transfer of farmland are independent of potential household financial vulnerability. That is to say, there is no unobservable missing variable affecting both farmland transfer decision-making and financial vulnerability. Secondly, the common support hypothesis posits the existence of an overlapping interval in the propensity score distribution between the treatment group (farmland transfer-out households) and the control group (non-transfer-out households). This is done in order to ensure that a control group with similar characteristics can be found for matching. Thirdly, the hypothesis of stable unit treatment value posits that the farmland transfer behaviour of a single farmer will not have a spillover effect on the financial vulnerability of other farmers. The treatment effect will only act on the farmers themselves. In accordance with the aforementioned assumptions, PSM has the capacity to accurately ascertain the average treatment effect (ATT) of farmland transfer behaviour on farmers’ household financial vulnerability.

The core principle of PSM is to match farmers who have transferred farmland (the treatment group) with those who have not (the control group) based on their propensity scores, thereby establishing comparable groups and improving the credibility of causal inference (Yin et al., 2024; J. Wang et al., 2025). PSM is adopted in this study to assess the impact of farmland transfer on rural household financial vulnerability for several reasons. First, farmland transfer is a voluntary and self-selected household decision (Y. Zhang et al., 2023), which leads to systematic differences in initial conditions between the treatment and control groups and may introduce selection bias. Additionally, since pre-transfer data on financial vulnerability are unavailable in practice, a simple comparison between the two groups would produce biased estimates. PSM mitigates self-selection issues without relying on strong functional form assumptions, parameter constraints, or specific distributions of the error term. It also does not require strict exogeneity of explanatory variables (Z. Jiang et al., 2025; Wu et al., 2025). In this study, PSM is applied to construct a matched sample of treatment and control households, allowing the effect of farmland transfer to be estimated under comparable observational conditions. The main procedure consists of the following steps:

(1)

Covariate Selection. Based on established theoretical and empirical literature, we include the aforementioned control variables in the model to satisfy the ignorability assumption and mitigate potential estimation bias.

(2)

Propensity Score Estimation. The propensity score reflects the conditional probability of a household engaging in agricultural land transfer. Given that land transfer behavior is a binary outcome, a Probit or Logit model is used to estimate the propensity scores. The model is specified as follows:

$$F{T}_i=1\left(\alpha +\beta X_i+{\epsilon }_i>0\right)$$

(3)

(3)

Propensity Score Matching. In empirical applications, the consistency of results across different matching algorithms is commonly taken as evidence of robustness. Accordingly, this study employs multiple matching methods, including k-nearest neighbor matching, caliper-based k-nearest neighbor matching, radius caliper matching, kernel matching, and Mahalanobis matching, to construct matched pairs between the treatment and control groups3.

(4)

Estimate the ATT. Based on the matched sample, the ATT is calculated to measure the difference in financial vulnerability between farmers who transferred farmland and those who did not, thus capturing the net effect of land transfer on rural household financial vulnerability. The ATT is defined as follows:

$$ATT=E\left(F{V}_{1i}\left|FT=1\right.\right)-E\left(F{V}_{0i}\left|FT=1\right.\right)=E\left(F{V}_{1i}-F{V}_{0i}\left|FT=1\right.\right)$$

(4)

In the above equation, FV_1i denotes the observed financial vulnerability of households in the treatment group, while FV_0i represents the counterfactual outcome, that is, the financial vulnerability these same households would have exhibited in the absence of farmland transfer. Since FV_0i is unobservable, it is approximated using matched control households identified via propensity score matching.

4. Empirical Results

4.1. Analysis of the Influencing Factors of Farmers’ Transfer-Out of Farmland

Given that farmland transfer is a binary choice variable for farmers, we employed both Probit and Logit models to estimate its determinants. The estimation results are presented in

Table 2. The factors of farmers’ farmland transfer-out and its impact on household financial vulnerability.

Variables	Explained Variable: Farmland Transfer-Out		Explained Variable: Rural Household Financial Vulnerability
	(1)	(2)	(3)	(4)
FT			−1.2115 *** (0.1064)	−1.0967 *** (0.1097)
Age	−0.0225 (0.0287)	−0.0439 (0.0527)		0.0038 (0.0196)
Age2	0.0002 (0.0003)	0.0004 (0.0005)		−0.0001 (0.0002)
Gender	−0.0989 (0.1930)	−0.2047 (0.3471)		−0.2462 ** (0.1258)
Health	0.1345 ** (0.0629)	0.2565 ** (0.1205)		−0.1550 *** (0.0458)
Edu	0.0559 (0.0719)	0.1220 (0.1359)		−0.2068 *** (0.0485)
Livelihood	0.2922 ** (0.0770)	0.5426 *** (0.1496)		−0.1369 *** (0.0524)
Credit	−0.1266 (0.1727)	−0.2156 (0.3223)		0.0652 (0.1078)
Scale	0.0518 (0.0946)	0.0874 (0.1778)		−0.1803 *** (0.0603)
Training	0.2299 * (0.1322)	0.4426 * (0.2539)		−0.2189 *** (0.0822)
Transportation	0.2040 *** (0.0822)	0.3851 *** (0.1548)		−0.0678 (0.0477)
Constant	−3.1903 *** (0.9375)	−5.8136 *** (1.7868)
Pseudo R2	0.0670	0.0674	0.0459	0.0899
Observations	746	746	746	746

Note: *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively. The number in parenthesis is Standard Error.

. As shown in columns (1) and (2), the results from both models are largely consistent in terms of the signs and statistical significance of the coefficients, indicating that our findings are robust. Regarding individual characteristics, health status exhibits a positive and statistically significant effect (at the 5% level) on the decision to transfer-out farmland. This suggests that farmers in better health are more likely to transfer-out their land, a finding that could be attributed to their greater capacity to pursue off-farm employment opportunities. In contrast, coefficients for variables such as age, age squared, and gender are statistically insignificant, implying that these basic demographic factors may not be decisive in the transfer-out decision within our sample.

Turning to production and operational characteristics, the type of operation (non-agricultural) shows a strong positive correlation (significant at the 1% level) with transfer behavior. This indicates that farmers engaged in non-agricultural activities demonstrate a higher willingness to transfer-out land, supporting the view that non-agricultural opportunities are a core driver of land transfer. Furthermore, participation in agricultural technology training has a marginally positive effect (significant at the 10% level). A potential explanation is that such training may enhance human capital, enabling farmers to secure higher off-farm income or making their land more attractive to larger-scale operators. Meanwhile, variables measuring loan demand and landholding size are not significant, suggesting that credit constraints and initial land endowments are not key binding factors in this context. In terms of external environmental characteristics, proximity to financial institutions has a significant positive impact (at the 1% level) on the probability of land transfer. This result implies that better access to financial services increases the likelihood of transfer, possibly because convenient finance provides necessary support for off-farm entrepreneurship or employment, thereby reducing the risks associated with exiting agricultural production.

4.2. The Results of the Impact of Farmland Transfer-Out on the Financial Vulnerability of Rural Households

As shown in Table 2, the ordered Probit regression results illustrate the effect of farmland transfer-out on rural household financial vulnerability. Without any control variables, farmland transfer-out showed a significant negative effect on financial vulnerability at the 1% level, suggesting that transfer behavior reduces vulnerability in the absence of other factors. After including the control variables, farmland transfer-out continued to exert a statistically significant negative influence on financial vulnerability at the 1% level. This indicates that the mitigating effect of farmland transfer-out on financial vulnerability is robust, underscoring its important role in reducing such vulnerability. This result can be explained by farmers’ ability to optimize resource allocation through farmland transfer, thereby reallocating labor and capital toward more profitable economic activities (H. Wang et al., 2024a; Ji et al., 2025). Such behavior is consistent with the dual objectives of diversifying economic risks and lowering household financial vulnerability. Notably, the relatively low Pseudo R² is common in micro-level household studies due to high unobserved heterogeneity and the multifaceted nature of financial vulnerability.

The regression results from column 4 indicate that among the control variables, gender, health status, educational attainment, livelihood strategic, cultivated land size, and agricultural technical training all demonstrate statistically significant effects on rural household financial vulnerability at the 1% or 5% level. This suggests that these factors meaningfully influence households’ financial vulnerability. Regarding gender, female-headed households tend to display more risk-averse financial behaviors (Han et al., 2024). After transferring farmland, these households show a stronger preference for depositing funds in banking institutions or other secure channels. While this strategy ensures stable returns, it offers limited potential for substantially improving household economic conditions, thus resulting in a relatively modest effect on mitigating financial vulnerability. Both health status and educational attainment exhibit statistically significant negative effects on financial vulnerability at the 1% level. This indicates that the likelihood of experiencing financial vulnerability decreases with better health and higher education levels of the household head. In practical terms, healthier household heads face lower medical expenditures, while those with higher educational attainment are better equipped to manage household assets and liabilities effectively, both mechanisms help reduce financial vulnerability.

From the perspective of livelihood strategic, this variable also exerts a statistically significant negative impact on household financial vulnerability at the 1% level. This indicates that as households shift from purely agricultural operations to diversified or specialized operations, their likelihood of experiencing financial vulnerability decreases. Such shifts contribute to agricultural risk diversification while simultaneously raising household income levels (H. Wang et al., 2024b), thereby reducing susceptibility to financial distress. Cultivated land size demonstrates a significantly negative relationship with financial vulnerability at the 1% significance level. Larger operational scale typically implies more stable agricultural production, which ensures consistent output and income (X. Li & Huo, 2021), consequently enhancing financial resilience. Furthermore, agricultural technical training shows a statistically significant negative effect on financial vulnerability at the 1% level. Households participating in such programs are better equipped with advanced production techniques and management methods, leading to improved agricultural productivity and quality. These enhancements ultimately raise agricultural income and reduce financial vulnerability.

4.3. PSM Estimation Results

The common support condition and covariate balance tests are essential to minimize matching bias, ensure that treatment effect estimates are not confounded by observed differences, and thus safeguard the validity of causal inferences (Olaoye et al., 2024; Yin et al., 2024). Therefore, before estimating the effect of agricultural land transfer on rural household financial vulnerability using PSM, we examined the common support domain and assesses covariate balance across the matched groups. Taking k-nearest neighbor matching (k = 4) as an example, after matching, the overlap range of propensity scores of farmers in the experimental group and the control group was significantly expanded, the covariate bias was significantly reduced and the standardized bias was less than 20%, indicating that the matching quality was high and there was no significant difference in the distribution of covariates between the two groups.

In PSM, the ATT serves as a causal indicator estimating the average effect of farmland transfer on financial vulnerability among participating households.

Table 3. PSM estimation results on the impact of farmland transfer on rural household financial vulnerability.

Match Methods	Experimental Group	Control Group	ATT	Standard Error	T Value
The k-nearest neighbor matching (k = 4)	0.7442	1.6250	−0.8808	0.0922	9.55
Caliper-based k-nearest neighbor matching (k = 4, Caliper = 0.01)	0.7619	1.5952	−0.8333	0.1366	6.10
Radius caliper matching (Caliper = 0.01)	0.7619	1.6286	−0.8667	0.0774	−11.20
Kernel matching	0.7442	1.6254	−0.8812	0.0756	−11.66
Mahalanobis matching	0.7471	1.5833	−0.8362	0.0802	−10.43
Mean	—	—	−0.8596	—	—

reports the ATT estimates obtained through five matching methods. The results show consistently negative and statistically significant effects at the 1% level across all methods, indicating robust matching performance. Moreover, the ATT values derived from different matching approaches are closely aligned, further confirming the reliability of the estimated mitigation effect of farmland transfer on financial vulnerability.

The last row of Table 3 presents the mean ATT value of −0.8596, indicating that farmland transfer reduces the average financial vulnerability of rural households by 0.8596 units. This suggests that, compared to non-transferring households, those engaging in farmland transfer experience a substantial reduction in financial vulnerability, effectively shifting from higher to lower vulnerability levels. These findings demonstrate that farmland transfer contributes to improving rural households’ financial resilience and reduces their susceptibility to financial distress.

4.4. Endogeneity Checks

The baseline regression may be subject to endogeneity concerns due to potential omitted variables, such as financial literacy (H. Chen et al., 2025) and government spending (Frangiamore & Matarrese, 2025), as well as possible reverse causality. To address these issues, we employed both an instrumental variable two-stage least squares (IV-2SLS) approach and a conditional mixed process (CMP) framework. The perceived difficulty of farmland transfer is used as an instrumental variable for households’ transfer decisions4, coded as 1 if farmers perceive local land transfer as difficult and 0 otherwise. The results of the endogeneity tests are presented in

Table 4. Endogenous discussion results.

Variables	IV 2SLS		CMP
	Stage I	Stage II	Stage I	Stage II
	FT	FV	FT	FV
Difficulty	−0.3898 *** (0.0229)		−1.9728 *** (0.1729)
FT		−0.7697 *** (0.1849)		−1.0967 *** (0.2757)
F value	33.77 ***
Endogenous test parameters			0.70
R2/Pseudo R2	0.3148	0.1977	0.4007	0.0899
Control variables	YES	YES	YES	YES
Observations	746	746	746	746

Note: *** indicate significance at the 1% level. Table 4 reports the estimated coefficients, with the values in parentheses being standard errors.

.

The endogeneity analysis shows that in both the instrumental variables (IV) approach and the CMP model, perceived difficulty of farmland transfer exerts a statistically significant negative effect on farmers’ land transfer decisions at the 1% level. This result confirms that farmers’ transfer behavior is meaningfully constrained by transfer difficulties, thereby satisfying the relevance condition of the instrumental variable. Moreover, the first-stage F-statistic reaches 33.77, well above conventional thresholds, indicating that the instrument is not weak. Under this specification, farmland transfer continues to show a significantly negative impact on rural households’ financial vulnerability. The CMP estimation shows that the endogeneity test parameter atanhrho_12 is statistically insignificant, suggesting no substantial endogeneity in the benchmark regression. All in all, after solved endogeneity problem, farmland transfer still maintains a pivotal negative effect on rural household financial vulnerability.

4.5. Robustness Checks

This study assessed estimation stability through four types of tests. Within the PSM framework, robustness was examined using k-nearest neighbor matching (k = 4) as a representative case. (1) Redefined the dependent variable assignment. On the one hand, a dummy variable for financial vulnerability was constructed, taking the value of 1 for households with medium-to-high financial vulnerability and 0 otherwise. On the other hand, households were reclassified into three categories: non-vulnerable (household financial margin ≥ 0), low vulnerable (financial margin < 0 and solvency ≥ 1), and highly vulnerable (all other cases). PSM estimation was then repeated under these changes (Column 1 and 2). (2) Reassigned the core explanatory variable. While the baseline regression measured farmland transfer behavior using a binary indicator, the robustness test in column 3 substituted this with the actual area of farmland transferred by households, after which PSM estimation was conducted again. (3) Performed a resampling test. Resampling methods do not require strong distributional assumptions and thereby enable direct simulation of the empirical distribution through repeated sampling. Accordingly, we performed 1000 resampling iterations to generate 1000 pseudo-samples. For each pseudo-sample, the k-nearest neighbor matching (k = 4) procedure was repeated and the corresponding ATT value was calculated, ultimately yielding 1000 ATT estimates (Column 4). (4) Changed regression models. Given that rural household financial vulnerability is an ordinal variable, the ordered Logit (Ologit, Column 5), and generalised ordered Probit (Eoprobit, Column 6) models, all suitable for analyzing ordinal outcomes, were substituted in the robustness tests.

The results of the robustness tests are presented in

Table 5. Robustness test results.

Variables	(1)	(2)	(3)	(4)	(5)	(6)
FT	−0.2238 *** (0.0581)	−0.6860 *** (0.0746)	−0.5076 *** (0.0908)	−0.9767 *** (0.0695)	−1.8840 *** (0.2293)	−1.0967 *** (0.1340)
Control variables	YES	YES	YES	YES	YES	YES
R2/Pseudo R2				0.1115	0.0891
Observations	746	746	746	746	746	746

Note: *** indicate significance at the 1% level.

.

As shown, regardless of the robustness strategy applied, the regression outcomes remain consistent with the benchmark findings. This confirms that the risk-reducing effect of farmland transfer is highly robust and not sensitive to model misspecification, data outliers, or other potential confounders, thereby accurately capturing the causal relationship between farmland transfer and reduced financial vulnerability.

4.6. Mechanism Analysis

Building on the established finding that farmland transfer alleviates rural household financial vulnerability, this section examines the theoretical mechanisms underlying this relationship. Theoretical analysis suggests that within the urban–rural dual economy, labor mobility has become an important channel for many farm households to diversify income sources and reduce financial vulnerability. To validate this theoretical pathway, we assessed whether farmland transfer facilitates labor mobility, which in turn may reduce households’ financial vulnerability.

The results of the mechanism test are presented in

Table 6. The results of the impact of farmland transfer on farmers’ labor mobility.

Variables	Labour Mobility		Scope of Labour Mobility
	(1)	(2)	(3)	(4)
FT	0.2609 *** (0.0504)	0.1659 *** (0.0481)	0.1159 *** (0.0316)	0.0950 *** (0.0312)
Control variables	NO	YES	NO	YES
R2/Pseudo R2	0.0252	0.1472	0.0227	0.0859
Observations	746	746	746	746

Note: *** indicate significance at the 1% level. The results exhibited in Table 6 are all the average marginal effects. The number in parenthesis is Standard Error.

. The findings indicate that farmland transfer exerts a positive and statistically significant effect on rural household labor mobility at the 1% level, regardless of whether control variables are included. This suggests that farmland transfer facilitates the reallocation of household labor between agricultural and non-agricultural sectors, thereby promoting labor mobility (B. Su et al., 2018). Specifically, after accounting for covariates, the AMEs of farmland transfer on labor mobility is 0.16595, implying that households engaging in land transfer experience a 16.59% higher probability of labor mobility compared to non-transferring households, all else equal. The analysis further reveals that omitting control variables would lead to an overestimation of this promotional effect. Similarly, with respect to the scope of labor mobility, farmland transfer demonstrates a significant expansive effect irrespective of covariate adjustment, though the magnitude is similarly moderated when other factors are controlled for. These results collectively affirm the positive role of farmland transfer in enhancing both the incidence and extent of labor mobility among rural households (W. Li et al., 2025a; Zhan et al., 2025).

In terms of magnitude, the average marginal effect of farmland transfer on the geographical scope of labor mobility is 0.0950 after controlling for covariates, indicating that land-transferring households experience an average increase of 9.5% in the probability of engaging in cross-provincial employment. This reflects a meaningful promotional effect of farmland transfer on the spatial extent of labor mobility. Two mechanisms help explain this phenomenon. First, farmland transfer releases households from cultivation-related spatial constraints, enabling laborers to seek off-farm employment without the need to balance agricultural activities (Imai et al., 2015). Second, rental income from land transfers helps alleviate liquidity constraints associated with migration, reducing financial barriers to long-distance mobility (L. Zhang et al., 2018).

Notably, the marginal effect on mobility scope is smaller than that on mobility incidence, suggesting that other factors, such as access to off-farm employment information, care giving responsibilities, and challenges related to urban integration, also shape the extent of labor reallocation. Thus, while farmland transfer significantly facilitates labor mobility (Xu et al., 2024), its role is contextual and mediated by broader socioeconomic conditions. In summary, by reallocating labor resources through land transfer, rural households can diversify income sources and strengthen risk resilience, thereby contributing to financial stability and reduced financial vulnerability. These findings support Hypothesis 2.

4.7. Heterogeneous Analysis

To more precisely quantify the heterogeneous effects of farmland transfer on rural household financial vulnerability and labor mobility, we applied the PSM method in its heterogeneity analysis. The heterogeneity analysis results are presented in

Table 7. Heterogeneity analysis results.

Variables	Age Heterogeneity		Human Capital Heterogeneity		Social Network Heterogeneity
	Young and Middle-Aged	Elderly	Low Human Capital	High Human Capital	No Family Tree or Genealogy	Have a Family Tree or Genealogy
FT→labor	0.1711 ** (0.0801)	0.1917 * (0.1127)	0.1395 (0.0884)	0.2195 ** (0.0959)	0.2240 *** (0.0797)	0.2895 *** (0.0911)
FT→FV	−0.8684 *** (0.1100)	−0.7417 *** (0.1793)	−0.7791 *** (0.1254)	−0.7622 *** (0.1345)	−0.8854 *** (0.1128)	−0.8026 *** (0.1448)
Control variables	YES	YES	YES	YES	YES	YES
Observations	462	284	519	227	363	383

Note: *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively. The results exhibited in Table 7 are all the average marginal effects. The number in parenthesis is Standard Error.

.

The first two columns of Table 7 present age-specific heterogeneity analysis regarding the impact of farmland transfers on household labor mobility and financial vulnerability6. The results indicate that regardless of whether households are headed by middle-aged/younger or older individuals, farmland transfers consistently promote labor mobility and reduce financial vulnerability without significant differential effects across age groups. This pattern suggests that farmland transfer facilitates access to relatively stable wage income through non-agricultural employment across household life stages (Imai et al., 2015; W. Li et al., 2025a), thereby enhancing resilience against economic shocks and mitigating financial vulnerability.

In terms of human capital heterogeneity, farmland transfers demonstrate stronger effects on labor mobility among high-human-capital households compared to low-human-capital households. This implies that farmers with greater human capital can more effectively leverage labor market opportunities and transition into non-agricultural employment after transferring their land. Meanwhile, farmland transfer effectively reduces financial vulnerability regardless of the household head’s human capital level, with no substantial disparities observed across human capital strata. This occurs because farmland transfer optimizes labor allocation and income structure across diverse household types, thereby generally enhancing economic stability irrespective of human capital endowment (Ji et al., 2025).

From the perspective of social network heterogeneity, farmland transfer demonstrates a stronger promotional effect on labor mobility among households possessing genealogical records or clan documentation. This suggests that such households are better positioned to leverage social network support after transferring their farmland, thereby enhancing their capacity to overcome geographical constraints in pursuing employment opportunities and achieving labor mobility. Meanwhile, as shown in the last two columns of Table 7, the beneficial effect of farmland transfer on reducing rural household financial vulnerability does not exhibit substantial heterogeneity across different social network levels. Consequently, regardless of household social network strength, farmland transfer consistently alleviates financial vulnerability without significant variation due to network disparities. This pattern can be attributed to the universal benefits households receive through rental income or compensation from land transfer, combined with income diversification through new employment channels. These mechanisms collectively enhance financial risk resilience, thereby attenuating the moderating role of social network differences.

5. Discussion

Financial vulnerability refers to a condition of elevated exposure to financial risk (Singh & Malik, 2022). At the household level, it manifests as the inability of rural households to adequately cope with unexpected shocks using their own savings, liquid assets, financial instruments, or labor reallocation (Faulkner et al., 2019; Noerhidajati et al., 2021; H. Wang et al., 2024a). The stability of rural households is fundamental to the harmony and development of rural society. Different from the views of H. Wang et al. (2024b), Y. Li et al. (2025) and Yuan et al. (2025), we believe that facilitating and supporting farmland transfer is a feasible way to reduce the financial vulnerability of rural households. In other words, reducing financial vulnerability enhances farmers’ capacity for self-development and risk resilience, thereby helping narrow urban–rural and inter-household inequality and advancing the goal of common prosperity (J. Wang et al., 2024; Y. Li et al., 2025).

As the most critical productive asset for farmers, land plays an irreplaceable role not only in generating household income, ensuring basic livelihood security, and promoting sustainable agricultural development, but also in shaping household labor allocation decisions (Kuang et al., 2021; Xu et al., 2024). Farmland cultivation demands can create a “lock-in effect” that restricts farmers’ labor mobility. However, farmland transfer has been shown to break this constraint, enabling the movement of labor from agriculture to non-agricultural sectors and from rural to urban areas (B. Su et al., 2018). Therefore, strategies that integrate farmland transfer with labor mobility can effectively overcome constraints to income growth caused by fragmented land management (Pierri et al., 2025), while also diversifying income sources and optimizing household asset portfolios (Ji et al., 2025). This, in turn, reduces financial vulnerability. This constitutes the core rationale for examining the impact of farmland transfer on rural household financial vulnerability through the lens of labor mobility.

Based on the theoretical and empirical findings, this study concludes that farmland transfer enables rural households to reallocate surplus labor, thereby promoting participation in non-agricultural employment, diversifying income sources, and ultimately reducing financial vulnerability. Therefore, strengthening guidance and support for farmland transfer will help optimize factor allocation, improve land use efficiency, and enhance labor mobility (Cui et al., 2025; Y. Zhang & Lerman, 2025). All of which contribute to expanding farmers’ income streams and mitigating financial vulnerability.

6. Conclusions and Policy Implications

Under the realistic background of rising financial vulnerability among rural households and the continuous expansion of farmland transfer scale in China, this paper examines the impact of farmland transfer behavior on rural household financial vulnerability using field survey data from Zhouzhi County, Shaanxi Province. It employs the ordered Probit model and PSM methods to analyze how farmland transfer-out alleviates financial vulnerability through the channel of labor mobility, and further conducts heterogeneity analysis across the dimensions of household head age, human capital, and social networks, aiming to provide empirical references and decision-making bases for mitigating rural financial risks, optimizing farmland transfer policies, and promoting stable income growth for farmers.

The main findings are as follows: First, factors including health status, engagement in non-agricultural operations, participation in agricultural technology training, and proximity to financial institutions exhibit statistically significant and positive effects on farmers’ farmland transfer decisions. Among these, access to financial services exerts the strongest influence. In contrast, individual characteristics such as age, gender, and education level, as well as land endowment, do not show statistically significant impacts on the transfer behavior. Second, farmland transfer demonstrates a significant negative impact on rural household financial vulnerability, effectively reducing the probability of households falling into financial distress. Third, labor mobility serves as a key mediating mechanism through which farmland transfer mitigates financial vulnerability. The transfer not only promotes labor reallocation but also expands the geographical scope of mobility, thereby strengthening household financial resilience. Furthermore, the effects of farmland transfer on both labor mobility and financial vulnerability exhibit heterogeneous patterns across households with different characteristics, including the age and human capital of the household head, as well as social network endowment. These findings offer important insights for designing targeted land and labor market policies to reduce financial vulnerability in rural areas.

The findings of this paper demonstrate that the transfer of farmland can have a substantial impact on the financial resilience of farmers’ households. It is therefore essential to prioritize the enhancement of the farmland transfer mechanism and the dismantling of transfer barriers, with a view to unleashing the poverty reduction and risk mitigation potential inherent in optimizing land factor allocation. To maximize benefits and mitigate potential risks, policymakers should prioritize the development and refinement of land transfer–oriented policies (Fang et al., 2024). Specifically, efforts should focus on promoting appropriately scaled agricultural operations, strengthening regional farmland transfer markets, and improving regulatory oversight to guide market development effectively (Yin et al., 2024; Hao et al., 2025). Concurrently, knowledge dissemination and training programs should be enhanced to improve farmers’ understanding of land transfer policies and raise their awareness of how such transfers can alleviate household financial vulnerability (Y. Zhang et al., 2023). With full respect for farmers’ voluntary participation, governments should provide necessary institutional and financial support to encourage land transfer (F. Su et al., 2023; H. Chen et al., 2025), thereby strengthening rural households’ financial resilience.

The positive role of agricultural land transfer in facilitating rural labor mobility and thereby reducing financial vulnerability should be emphasized in policy design (B. Su et al., 2018; Xu et al., 2024; Y. Zhang & Lerman, 2025). Specifically, governments should adopt a dual approach that simultaneously promotes farmland transfer and supports labor mobility (Lu et al., 2019; L. Li et al., 2022). In line with market demands and employment trends, targeted support, including skills training, entrepreneurship programs, and job placement services, should be provided to households that have transferred farmland (R. Li et al., 2025). Such support can be delivered through training workshops, learning resources, and partnerships with leading agricultural enterprises, helping alleviate farmers’ concerns about transitioning to non-agricultural employment after land transfer.

Furthermore, policy interventions should account for heterogeneity in household characteristics such as the age and human capital of the household head (J. Liu et al., 2023; J. Wang et al., 2025), as well as social network endowment (B. Chen et al., 2024). To address age-related disparities, diversified subsidy schemes for farmland transfer and re-employment platforms for older farmers should be established to encourage off-farm employment. In terms of human capital, improving rural education and vocational training will enhance the competitiveness of young and middle-aged farmers in non-agricultural sectors. Regarding social networks, information platforms should be developed to integrate and disseminate employment information efficiently. At the same time, financial products leveraging social networks could be introduced to improve credit access and strengthen financial resilience among rural households.

In summary, policymakers should establish an integrated policy framework that leverages farmland transfer to facilitate labor mobility and alleviate financial vulnerability (Tan et al., 2022). Complementary measures, such as non-agricultural skills training and portable social security systems, should be strengthened to help farmers secure stable off-farm employment (Xu et al., 2024; S. Wang et al., 2025). This creates a virtuous cycle in which labor mobility boosts income, and income growth reduces vulnerability, thereby fostering sustainable household development and stable rural economic growth. Furthermore, while the financial vulnerability-reducing effect of farmland transfer is consistent across household types, its influence on labor mobility is more pronounced among elderly households, those with higher human capital, and households with clan networks. Policymakers should therefore account for such heterogeneous effects and design targeted interventions to enhance the role of farmland transfer in promoting labor mobility among these groups.

However, this study may have some limitations. First, while the cross-sectional survey data reveal short-term correlations between farmland transfer and household financial vulnerability, they cannot capture the dynamic evolution or long-term effects of this relationship. Second, the research is geographically limited to Zhouzhi County in Shaanxi Province, and the sample size remains relatively small with homogeneous regional features. These constraints may affect the external validity of the findings. Future research could employ panel household tracking data and dynamic econometric models to more precisely identify the long-term mechanisms through which farmland transfer influences household financial vulnerability. It would also be valuable to expand the scope of study to include rural areas with varying levels of economic development and resource endowments and to conduct comparative regional analyses. Such efforts would enhance the generalizability and policy relevance of the conclusions, providing more robust empirical evidence for national-level farmland transfer policies and rural financial risk prevention.