The Sustainable Livelihoods Analysis Framework proposed by DFID (2022) aims to explore which livelihood outcomes can be achieved by different combinations of livelihood capital and livelihood strategies for the poor livelihood context. In the framework of sustainable livelihood analysis, household livelihood capital includes human, social, natural, physical, and financial capital, while livelihood strategies refer to the behavioral strategies employed by households to achieve sustainable livelihoods such as farmland circulation and crop diversification. Besides, livelihood outcomes mainly include multiple dimensions, such as increased income, poverty alleviation, welfare enhancement, and reduced vulnerability. The core of the sustainable livelihood analysis framework is to assess the influence of household livelihood capital and livelihood strategies on livelihood outcomes. The framework can be applied to different research subjects, such as individuals, households, villages, and even countries (Sun, 2020; Natarajan et al., 2022; Sun et al., 2022), thus serving as a good theoretical analysis tool for this paper to understand and assess the impact of farmland circulation on the financial vulnerability of farm households (Figure 2).

According to the sustainable livelihoods analysis framework, livelihood strategies can significantly affect livelihood outcomes (Dey and Haloi, 2019; Eshun et al., 2019; Sun, 2020; Chowdhury, 2021; Natarajan et al., 2022). Specifically, this study suggests that the livelihood strategy of farmland circulation can have an impact on the financial vulnerability of farm households' livelihood outcomes. In general, household financial vulnerability mainly includes aspects such as over-indebtedness and emergency savings, and its specific measures are mostly related to income, liabilities, and assets (Fuenzalida and Ruiz-Tagle, 2011; Loke, 2017; Noerhidajati et al., 2021). Previous studies have shown that the farmland property rights reform and transfer in rural China can not only effectively improve the income level of farm households (Peng et al., 2020; Yang et al., 2020; Huo and Chen, 2021), but also explore the transformation of the financial value of farmland (Liu and Li, 2015; Jiang et al., 2017). The transformation of the financial value of farmland can further increase the liquidity of household capital, improve the efficiency of household asset allocation, and strengthen the ability of farms to cope with various uncertain risks. In addition, farmland circulation can increase income and improve the efficiency of asset allocation while alleviating or even eliminating household indebtedness to a certain extent. Based on the above analysis, the study proposes the following hypothesis:

In the context of the rapid development of urbanization and industrialization in China, a great deal of rural labor has chosen to seek employment opportunities by moving from the agricultural sector to the non-agricultural sector (Deng et al., 2020; Zhang et al., 2021). In process of transferring rural labor, there are certain costs and expenses, which are known as “economic threshold” (Chen and Wang, 2017). Under the theoretical assumption of “the rational peasant” (Schultz, 1964; Popkin, 1979), rural labor will choose to transfer if the benefits are greater than the “economic threshold.” However, when farmland cannot be reasonably transferred, potential farmland circulation (i.e., farmers who are willing to transfer their farmlands) cannot receive its rental income and even has to bear the costs of searching for contractors in the market, especially during the crop ripening season when farmers who have already worked in cities have to pay additional travel and labor costs in order to return home to harvest their crops (Ren and Kong, 2016). These costs and expenses are the “economic threshold” that currently limits the smooth transfer of rural labor to the non-farm sector.

With the continuous improvement of the rural farmland circulation market, more and more potential farmers can successfully transfer their farmlands, thereby reaping the benefits of farmland circulation and lowering their “economic threshold” for transferring to the non-farm sector (Chen and Wang, 2017). The lowering of the “economic threshold” will further accelerate the transfer of rural labor to the non-agricultural sector. Therefore, farmland circulation can facilitate the transfer of rural labor, which has been confirmed by some scholars using micro-survey data (Yao et al., 2010; Zhao et al., 2016; Han et al., 2019). In the context of the imbalance between China's urban and rural dual economy, the transfer of rural labor has obvious “income effects” and “Internet effects.” In other words, labor transfer helps to promote the increase in household income and the popularization of Internet use among farm households (Liao et al., 2020; Guo and Wang, 2021; Zhou and Chen, 2022). Among other things, the use of the Internet can alleviate objective constraints on farmers' participation in financial markets, increase the availability of household finance, and improve the way farmers allocate their financial assets (Bogan, 2008). Since the measures of household financial vulnerability mainly include income, liabilities, and assets (Fuenzalida and Ruiz-Tagle, 2011; Loke, 2017; Noerhidajati et al., 2021), the “income effect” and “internet effect” of labor transfer can improve the financial vulnerability of households. Based on the analysis, the second hypothesis is as follows:

For farmlands with financial attributes, farmland circulation is a financial act of farmers (Jiang et al., 2017). At the same time, according to the sustainability analysis framework, farmland circulation is a livelihood strategy for farmers (Tian et al., 2022). Rational farmers who seek to maximize their benefits will weigh the possible costs and benefits before making any financial decision or livelihood strategies adjustment (Schultz, 1964; Popkin, 1979). Therefore, before decision-making on farmland circulation, rational farmers will actively collect and understand market price information through various channels to make themselves financially literate and thus achieve their livelihood goal of maximizing household income. In addition, the increase in income from farmland circulation provides an incentive for farmers to improve their financial literacy in order to allocate their capital more efficiently. Thus, both before and after the transfer, the financial decision and livelihood strategies of farmland circulation are conducive to the financial literacy of farm households.

Financial literacy refers to the combination of knowledge, awareness, attitudes, and skills necessary for a person to understand basic financial concepts and make the right financial decisions (Atkinson and Messy, 2011; Chhatwani and Mishra, 2021; Hamid and Loke, 2021; Hsu et al., 2021). Most studies have used “financial knowledge” and “financial literacy” interchangeably (Ansari et al., 2022). Increased levels of financial literacy are favorable for households' maintenance of healthy long-term savings and investments (Sivaramakrishnan et al., 2017; Kitamura and Nakashima, 2021). At the same time, according to behavioral finance, consumers are prone to heuristic bias, which leads to sub-optimal financial decisions (Chen et al., 2007; De Bondt et al., 2008; Barber and Odean, 2013). Financial literacy, with its moderating and controlling effects, can reduce the probability of making poor financial decisions, thereby minimizing the adverse effects of behavioral biases (Grohmann, 2018). In general, households with higher levels of financial literacy are more likely to make the right savings and investment decisions but less likely to be affected by behavioral biases and investment errors, avoiding falling into a debt crisis and suffering from financial vulnerability. Therefore, financial literacy can reduce household financial vulnerability, and some scholars have confirmed this finding using micro-survey data (Liu et al., 2020; Chhatwani and Mishra, 2021; Seldal and Nyhus, 2022). Based on the above discussion, the third hypothesis is as follows:

Broadly speaking, labor transfer can affect the financial literacy of farm households through the “social network effect.” The transfer of rural labor from the agricultural sector to the non-agricultural sector can broaden the radius of work and life and the scope of social interaction compared with that in a relatively closed village. That is to say, labor transfer can enable farm households to have a wider social network (Bai and Yuan, 2014). Enlarging the size of social networks has an information-sharing effect, making it easier for different groups to communicate, share, and receive more open and diverse market information and financial knowledge (Ma and Yang, 2011), thus contributing to the improvement of financial literacy. At the same time, according to the SLF, vulnerable smallholder farmers are susceptible to a variety of external risks (DFID, 2022). When labor transfer enters a broad social network, the probability of households falling into financial vulnerability can be reduced through financial assistance from other social network members, even if the farm household suffers an external risk shock or fails in its financial decisions.

Based on the above discussion and research hypotheses H2-H3, it can be inferred that labor transfer can influence household financial vulnerability by promoting the financial literacy levels of farm households. In other words, the two variables of labor transfer and financial literacy can act successively as indirect mechanisms through which farmland circulation affects the financial vulnerability of farm households. Therefore, the following research hypothesis is proposed in this study:

The heterogeneity of the impact of farmland circulation on household financial vulnerability may also result from the fact that households are different in terms of generation and region. For example, in the process of urbanization in China, the different upbringing of “first-generation farmers” (generally born before the 1980's) and “second-generation farmers” (generally born after the 1980's) has led to significant differences in their willingness to return to their hometowns and their perceptions of income, consumption, and savings (Liu and Wang, 2020), and hence in their degrees of financial vulnerability. Compared with “first-generation farmers,” younger “second-generation farmers” are more inclined to live in the city and can integrate smoothly into new social networks. This, coupled with a higher level of acceptance of financial management and the digital economy, has enabled “second-generation farmers” to reallocate their assets more rationally and effectively after receiving income from farmland circulation. In addition, uneven and inadequate development is the main contradiction in Chinese society, and the “uneven development” is mainly reflected in the uneven level of economic development between regions (Liu et al., 2022). Based on the above discussion, we have the fifth hypothesis is as follows:

In this study, data were collected from a large-scale nationwide household tracking survey conducted by the Survey and Research Center for China Household Finance from 2015 to 2019. The survey is carried out every 2 years and includes community (village), household, and individual questionnaires, covering household assets and liabilities, insurance and protection, income and consumption, demographic characteristics, employment, and other micro-level financial information. Thus, it can provide strong data support for this paper to explore the impact of farmland circulation on the financial vulnerability of farm households from a micro perspective. In accordance with the needs of the study, those farmlands owned by individuals were selected as the sample for this study. By matching and cleaning 2015, 2017, and 2019 data (mainly by eliminating outliers and missing values and shrinking the tails of income, consumption, and asset variables), 9,822 valid samples were finally obtained while ensuring that the balanced panel was satisfied.

Through Stata 17.0 analysis, Table 3 shows the impact of farmland circulation on household financial vulnerability estimated mainly by a random effects model, with possible heteroskedasticity addressed by clustering robust standard errors. The columns (1–3) in Table 3 were estimated by sequentially adding household head characteristics and household characteristics to the core independent variable farmland circulation. The estimation results demonstrated that the effect of farmland circulation on household financial vulnerability remained significantly negative as the control variables were added in turn. When all control variables were added, the value of the coefficient of farmland circulation was −0.167, p < 0.01 [with a coefficient of −0.167 (p < 0.01)] indicating that farmland circulation can mitigate household financial vulnerability, which initially validated hypothesis H1.

Other control variables were also found to be significantly related to household financial vulnerability. Firstly, household head characteristics, such as higher education, good health, and retirement security all reduced household financial vulnerability, with education affecting household financial vulnerability by reducing the probability of debt burden and stabilizing income (Zhang et al., 2020), while better health made households less vulnerable as these households were less exposed to diseases and less likely to suffer unexpected shocks such as medical expenses, which were reduced by with good medical coverage. Secondly, at the household characteristics level, the population size was negatively related to household financial vulnerability, whereas the child dependency ratio and the old dependency ratio significantly increased the probability of a household being financially vulnerable. The positive effect of the child dependency ratio on household financial vulnerability was nearly three times that of the old dependency ratio. The possible reason for this is that households with larger populations have more members in the labor force and thus higher income that dampens financial vulnerability, but a higher proportion of children implies higher developmental consumption such as education, which leads to a high household burden that significantly aggravates financial vulnerability.

To ensure the robustness of the above estimation results, robustness tests were conducted using, among others, replacement models and variable measures. Estimation was first carried out through the propensity score matching method, and then common support domain and balance tests were performed. Table 4 shows the average treatment effects of farmland circulation on household financial vulnerability obtained using the four matching methods. Overall, the values of similar average treatment effects obtained through different methods were relatively close, indicating the robustness of the analysis. More importantly, the data results demonstrate that farmland circulation has a significant mitigating effect on household financial vulnerability. For the purpose of this analysis, mean values were used for characterization. Specifically, after PSM counterfactual estimation, farmland circulation was found to significantly mitigate financial vulnerability, with a mean treatment effect of −0.155, indicating that farmland circulation reduces household financial vulnerability by 15.5% after addressing sample selection bias.

In addition to robustness testing through propensity score matching, this study also used the method of replacing the independence variable with the dependence variable for further testing. The financial vulnerability index was first classified as a discrete variable by mean, with values from 0 to 2, representing low, medium and high vulnerability, respectively and then empirically tested using a panel multi-valued logit model, and the results were shown in column (1) in Table 5. As the estimated coefficients of the logit model are not directly comparable in a statistically significant way, the values reported in the table are marginal coefficients. It can be found that after replacing the measure and estimation method of financial vulnerability, farmland circulation still has a negative impact on financial vulnerability. Specifically, with low vulnerability as the benchmark group, the effect of farmland circulation on high financial vulnerability was much higher than that of low vulnerability, indicating that farmland circulation can better mitigate high household financial vulnerability. In terms of control variables, the health status and medical coverage of the household head always have a negative effect on financial vulnerability; the marginal effect of the child dependency ratio on high vulnerability was higher than that of medium vulnerability, and the old dependency ratio has a decreasing effect on the probability of households falling into medium financial vulnerability but increases the probability of households falling into high financial vulnerability, indicating that an excessive demographic dependency burden was more likely to put households into high financial vulnerability status. In addition, column (2) in Table 5 was the estimated result of replacing the independence variables, and we replaced whether farmland is diverted with the area of farmland diverted for the regression, which was still found to have a mitigating effect on financial vulnerability. Even after FVI was decomposed into five specific time-level indicators, namely short-term, short to medium term, medium to long term, long-term and accident period, the impact of farmland circulation on household financial vulnerability in different periods was examined by using the truncated regression model successively (see Appendix A for the results), the results remained unchanged.

In addition, considering the endogeneity problem caused by the possible omission of variables in the baseline regression, instrumental variables were introduced into the model. The estimated results are shown in column (3) in Table 5, where the relationship between farmland circulation and financial vulnerability remained negative and significant at the 1% statistical level after using instrumental variables, proving that farmland circulation effectively reduces financial vulnerability. Research hypothesis H1 was further verified, and again, the direction of influence of the control variables remained consistent with the baseline regression results, also indicating that the findings of this study remain robust after overcoming endogeneity.

The estimation results of models (4–6) are shown in Table 6. According to column (1) in Table 6, the coefficient of farmland circulation was 0.211 and significantly positive at the 1% level, indicating that farmland circulation can promote labor transfer. Column (2) shows that the regression coefficients of farmland circulation and labor transfer on financial literacy were 0.169 and 0.353, respectively, indicating that both farmland circulation and labor transfer can significantly increase financial literacy. Column (3) shows that the estimated coefficients of farmland circulation, labor transfer, and financial literacy were all significantly negative, with the coefficient of farmland circulation being −0.094, using the financial vulnerability index as the independent variable. The results of the above three columns suggest that farmland circulation can indirectly reduce household financial vulnerability by promoting labor transfer or improving financial literacy in one direction or through the chain mechanism of “promoting labor transfer → improving financial literacy.”

Table 7 reports the results of the Bootstrap method test for the chain multiple mediating effects under the OLS model, and it can be found that the farmland circulation → labor transfer → financial vulnerability path has the highest effect value, with a coefficient of −0.044, which does not contain 0, indicating that the mediating effect was significant and H2 holds. Farmland circulation → financial literacy → financial vulnerability with a coefficient of −0.013, does not contain 0, indicating that the mediation effect was significant and H3 holds. Farmland circulation → labor transfer → financial literacy → financial vulnerability with a coefficient of −0.006, does not contain 0, indicating a significant mediating effect, H4 holds. This implies that there was a significant continuous mediation effect from labor transfer to financial literacy, i.e., farmland circulation can mitigate financial vulnerability through the chain mediation pathway of “promoting labor transfer → enhancing financial literacy.”

In summary, the hypotheses H1–H4 of this paper were tested and passed, thus obtaining a chain mediation model of the impact of farmland circulation on financial vulnerability. According to the estimation results in Table 6, the relationship between the variables and the coefficient of influence were shown in Figure 3, and the direct effect of farmland circulation on financial vulnerability is −0.094. Further analysis in conjunction with Table 7 reveals that the total mediation effect of farmland circulation on financial vulnerability is −0.063 and the direct effect is −0.094, so the total effect is −0.157. This gives a total mediating effect as a proportion of the total effect of 40.13%. Among the different types of mediation effects, the independent mediation effect of labor transfer is the largest, with an effect value of −0.044, accounting for 28% of the total effect, the independent mediation effect of financial literacy is the second largest (effect value of −0.013) and the impact effect of chain mediation is the lowest (effect value of −0.006), indicating that the pathway of farmland circulation is the most significant in mitigating household financial vulnerability by promoting labor transfer, and is the most important mechanism through which farmland circulation mitigates household financial vulnerability.

To capture the differences in the impact of farmland circulation on financial vulnerability, this section presents a heterogeneity analysis from two dimensions: household and regional characteristics. Table 8 presents the results of the financial vulnerability measurement for different household and regional characteristics of farmers, which show that there are significant differences in the levels of financial vulnerability of farmers by household and regional characteristics. In particular, the level of financial vulnerability of young “second-generation farmers” (with heads born after the 1980's) was significantly lower than that of “first-generation farmers” (with heads born before the 1980's), F = 32.51, p < 0.01. The level of financial vulnerability of households in the eastern region, which is characterized by a higher level of economic development than the central and western regions, was also significantly lower than that in the central and western regions, F = 59.97, p < 0.01.

Further results of the regression model were shown in Table 9. Specifically, the regression coefficient of farmland circulation on the financial vulnerability of households in the “first-generation farmers” is −0.119, which was statistically significant at the 1% level, indicating that the farmland circulation behavior of households in the “first-generation farmers” can significantly reduce household financial vulnerability. The estimated coefficient is −0.1 but was not statistically significant when the sample is the “second-generation farmers.” At the regional level, farmland circulation had a mitigating effect on household financial vulnerability in the eastern, western and central regions, with the largest effect in the east.

In this study, a new area of research, the potential relationship between farmland circulation and the financial vulnerability of households, was introduced to enrich existing literature. This study not only analyzed the direct relationship between farmland circulation and financial vulnerability but also explored the indirect impact mechanisms involved using the chain mediation model, by using the latest available Chinese household survey data containing 9,822 households for the period 2015–2019. This study showed that farmland circulation was negatively correlated with household financial vulnerability with a coefficient of −0.167 (p < 0.01), indicating that farmland circulation, as a self-selected behavior of farmers (He et al., 2022), can achieve the purpose of safeguarding household financial security and improving the current situation (Xu et al., 2020), which supports the classical theoretical assumption that “farmer are rational” (Schultz, 1964; Popkin, 1979), as well as the idea that “livelihood strategies influence livelihood outcomes” in the sustainable livelihoods analysis framework is also applicable in the Chinese context. At the same time, promoting farmland circulation is the core task of China's rural land system reform at the present stage (Peng et al., 2020). As a result, this finding may also provide micro-level evidence for formulating relevant policies and measures.

In terms of the indirect effects of farmland circulation on the financial vulnerability of farmers, this study shows that farmland circulation can indirectly reduce household financial vulnerability by affecting labor transfer and financial literacy. Specifically, the independent mediation effect of labor transfer was the largest, with an effect value of −0.044, accounting for 28% of the total effect, followed by the independent mediation effect of financial literacy (−0.013) and the lowest effect of chain mediation (−0.006). This confirms that the “asset allocation effect” of farmland circulation can facilitate labor transfer (Yao et al., 2010; Zhao et al., 2016; Han et al., 2019), the “social network effect” of labor transfer can enhance the financial literacy of farmers, and financial literacy can reduce the financial vulnerability of farmers (Liu et al., 2020; Chhatwani and Mishra, 2021; Seldal and Nyhus, 2022). Therefore, in order to give greater play to the role of farmland circulation in alleviating household financial vulnerability, the government should pay attention to labor transfer and financial literacy in the process of promoting farmland circulation.

To capture the differences in the impact of farmland circulation on financial vulnerability, this study discussed the differences in household and regional characteristics of farmers in groups. The results of the financial vulnerability measurement showed that older “first-generation farmers” households had significantly higher levels of financial vulnerability than younger “first-generation farmers” households (F = 32.51, p < 0.01). In this study and previous studies (Fuenzalida and Ruiz-Tagle, 2011; Loke, 2017; Noerhidajati et al., 2021), financial vulnerability was measured mainly by income, consumption, and other indicators. According to the household life cycle theory, the household life cycle has an impact on these indicators (Martins et al., 2011; Liu and Wang, 2020). Therefore, there are differences in the level of financial vulnerability across household life cycles. Further model regression results showed that farmland circulation significantly reduced the financial vulnerability of “first-generation farmers” households (coefficient = −0.149, p < 0.01), but the estimated coefficient for the “first-generation farmers” was −0.1 and not significant. The possible reason is that younger “second-generation farmers” households are less likely to engage in agricultural production, and whether or not farmland is circulation does not have a significant impact on their livelihood outcomes. By contrast, older “first-generation farmers” often miss out on the opportunity to work in urban areas by transferring labor because they cannot afford to give up agricultural production. As a result, the financial vulnerability of first-generation farmers is more significantly affected when farmland circulation takes place. This also suggests that it is “better late than never” for farm households to transfer their farmlands.

In terms of regional groupings, the level of financial vulnerability of households in the eastern region, which is at a higher level of economic development, was significantly lower than that in the central and western regions (F = 59.97, p < 0.01), and the mitigating effect of farmland circulation on the financial vulnerability of households in the eastern region was also the largest (coefficient = −0.149, p < 0.01). This difference may be explained by the fact that the eastern region has been leading the development of financial markets nationwide, with urban expansion contributing to greater market demand for farmland circulation, as well as the fact that farmers in the eastern region have greater access to employment opportunities and financial information when they move to cities, which has led to a significant increase in their levels of financial literacy. As a result, farmers in the eastern region have been able to increase their income through farmland circulation, thus allocating their assets more efficiently to reduce their household financial vulnerability. The estimated coefficients of labor transfer (coefficient = −0.222, p < 0.01) and financial literacy (coefficient = −0.101, p < 0.01) on financial vulnerability in the eastern region in the regression results in Table 9 also support this judgment from the side, with the coefficients in the eastern region being significantly higher than those in the central and western regions. In the context of China's commitment to addressing regional development inequalities (Guo et al., 2020; Qi et al., 2022), the findings of this study are of great reference value for government departments to take effective measures to avoid the consequences of further widening the regional development gap brought about by farmland circulation.

The strengths of this study include the use of large data, a representative sample, a wide range of socio-economic variables, and the examination of whether and how farmland circulation affects household financial vulnerability and whether their effects are heterogeneous according to differences in household and regional characteristics of farmers. However, this study is not free from limitations. Firstly, farmland circulation includes two types, namely, inflow and outflow. This paper only focused on farmland outflow. However, what is the impact of farmland inflow on the financial vulnerability of households? Will it get better or worse? This is an important issue that needs further study. Secondly, in this study, financial vulnerability indicators were constructed in terms of household survival and security, social capacity, debt burden, and unexpected shocks, whereas there is no space for a detailed discussion about the impact of farmland circulation on these dimensions. Therefore, further research could consider systematically analyzing the impact of farmland circulation on different dimensions of financial vulnerability of farm households. Thirdly, although stepwise regression and Bootstrap methods were used in this study to test the mediating effects of labor transfer and financial literacy on the relationship between farmland circulation and financial vulnerability, it cannot be denied that there may also be moderating effects of labor transfer and financial literacy or endogeneity problems due to reverse causality (Jia et al., 2020; Kim et al., 2022). This is also a direction for future research and discussion.