Grassroots autonomous organizations embedding, non-agricultural employment and farmer homestead withdrawal behavior: based on the research data of Qionglai City, Sichuan Province, China

In the context of the rapid rise of the village revitalization in China, the study examines the impact of grassroots autonomous organizations embedding level and farmer non-agriculture employment on the farmer homestead exit behavior. Based on 421 household surveys in Qionglai, this study obtained three results through empirical analysis. (1) The system embedding level of grassroots autonomous organization and the interests embedding level of grassroots autonomous organization have a significant positive impact on the farmer homestead withdrawal behavior, but the emotion embedding level of grassroots autonomous organizations has a significant negative impact on the farmer homestead withdrawal behavior. (2) The farmer non-agriculture employment level is beneficial for the farmer homestead exit behavior. (3) Further mechanism analysis shows that the non-agricultural employment level plays an inhibitory role in the positive influence of grassroots organizations to mobilize farmers to exit homestead. Therefore, it is suggested that when mobilizing farmers to voluntarily withdraw from the homestead, grassroots organizations should actively improve their embedding level, improve the construction of farmers’ social network, and accelerate the accumulation of farmers’ material capital.


Introduction
At present, the urban-rural dual structure generally exists in developing countries, which restricts the stable development of the economy and society, and further widens the gap between urban and rural development. To promote the integrated development of urban and rural areas, the reform of the factor market of the urban and rural dual structure must be deepened, among which the land factor market is the key point of the reform. In this regard, to promote the integrated development of urban and rural areas and realize rural revitalization, the Chinese government has decided to conduct rural homestead reform to improve the land factor market after deeply thinking about the current domestic urban-rural development gap and the problems left over from history in reform and opening up, providing Chinese wisdom for solving the problem of urban-rural dual structure in developing countries. Under the new pattern of integrated urban-rural development in China, it is inevitable requirements for comprehensively promoting rural revitalization and achieving common prosperity to deepen the reform of rural homestead, optimize the allocation of urban and rural resources, make economical and intensive use of rural homestead, and stimulate the potential driving force of rural development Zhou et al. 2020;Luo 2021). At present, there are many problems in the process of rural homestead utilization in China, such as low utilization level, prominent vacancy phenomenon, "one household, multiple homesteads," and poor exit of farmers, which seriously 1 3 restrict the development of the rural economy and society and the growing needs of farmers for a better life (Su et al. 2019). In reality, grassroots self-governing organizations are established at the most basic level of Chinese society and are directly in contact with the masses. They are voluntarily organized by the masses to manage their affairs according to their living areas, grassroots autonomous organizations in the system, interests, emotions, and other aspects of embedded in the grassroots people and are the most basic implementation subject of rural policy, which has a strong correlation with farmers and is one of the important forces affecting the withdrawal of farmers' homestead (Shi et al. 2022;Tang et al. 2020;Zhang et al. 2021). Then, how to give full play to the embedded characteristics of grassroots organizations to promote farmers to actively withdraw from the homestead? What is the mechanism of this pathway? In addition, nonfarm employment is an important driving force to promote the transfer of homestead resources; heterogeneous farmers with different levels of non-agriculture employment will have different choices for homestead exit. Therefore, will farmers' non-agricultural employment have an impact on the process of grassroots autonomous organizations mobilizing farmers to withdraw from the homestead? In conclusion, taking the reform of rural homestead in China as an example, analyzing the influence of the embedding of grassroots autonomous organizations on the withdrawal behavior of farmers' homestead, and exploring the adjustment effect of non-agricultural employment of farmers, to improve China's rural farmers homestead exit rate, promote China's land market reform, improve the grassroots autonomy organization homestead exit action mechanism, and improve the level of China's rural governance have important practical significance. Moreover, it provides a reference idea for the general developing countries to solve the dilemma of urban-rural dual structure. Perfecting the homestead system is an important way to realize rural revitalization in China (Luo 2021;Shi and Yu 2021;Wu et al. 2021). At present, many scholars have done extensive research on the rural homestead exit, mainly focusing on three aspects. First, from the perspectives of individual characteristics of farmers, family endowment, property rights condition, perceived value, social capital, right protection, risk perception, and farmers' differentiation, current scholars have studied the influence of farmers' homestead exit willingness and behavior and discussed the contradiction between farmers' homestead exit willingness and behavior (Shi and Yu 2021;Sun et al. 2019;Zhang and Xia 2021a). Second, based on the perspective of rural land property right system reform, poverty alleviation relocation, farmer welfare, and family life cycle, the existing literature respectively analyzes the impact mechanism and action mechanism of homestead withdrawal Gao et al. 2021). Third is to study the impact of homestead exit on agricultural ecological efficiency, agricultural production efficiency, and farmers' production and lifestyle Zhang and Xia 2021b). Related studies on the farmers' homestead exit are abundant, but there are still few studies analyzing the farmers' homestead exit behavior from the embedded perspective of grassroots autonomous organizations (village committees and villagers' groups). Limited literature mainly focuses on theoretical analysis and lack of empirical tests (Wang and Lu 2021;Yang et al. 2014).
The purpose of this study is to analyze the impact of the embedding level of grassroots autonomous organizations on the farmer homestead exit behavior. Given that the existing literature rarely discusses the relationship between the embedding level of grassroots autonomous organizations and the farmers' non-farm employment, this paper also tries to verify whether farmers' non-farm employment has a moderation effect between the embedding level of grassroots autonomous organizations and the farmers' homestead exit behavior and to further explore the mechanism of the above pathway.
The contributions of this study are threefold. First, this paper constructs the research framework of grassroots autonomous organizations embedding, non-agricultural employment, and farmers' homestead exit behavior. Second, the grassroots autonomous organizations embedding is divided into three aspects: institutional embedding, organizational embedding, and emotional embedding, and this paper empirically explores the influence of the grassroots autonomous organization embedded level on the farmer homestead exit behavior according to embedded theory. Third, considering the important role of farmer income level (especially non-farm income) in homestead withdrawal, this paper examines the moderating role of non-farm employment in the grassroots organizations' embedded level and the farmer homestead exit behavior. Moreover, by introducing instrumental variable, this paper effectively solves potential endogenous problems and increases the reliability of the conclusions.

Theoretical analysis
The impact of the grassroots autonomous organizations embedding level on the farmer homestead exit behavior Embeddedness theory is one of the core theories in the analytical framework of new economic sociology (Wang and Lu 2021), which was first proposed by Polanyi in 1944. The central idea is that individual economic behavior is closely linked to social relations, always embedded in non-economic factors such as customs (Granovetter 1984). Granovetter proposed the concept of "weak embeddedness." According to the new economic sociology, people's economic behavior not only is the pursuit of personal benefit but also is embedded in the social structure and affected by various social relations (Zhang et al. 2020). Zukin and Dimaggio extended the connotation of embeddedness theory and divided embedding into political embedding, cultural embedding, cognitive embedding, and structural embedding. At present, most Chinese scholars build an embedded index system based on the framework proposed by Zukin and Dimaggio, which is divided into political embedding, cultural embedding, cognitive embedding, and structural embedding. (Wang and Lu 2021). Embeddedness theory was initially applied to the analysis of business activities in China (Liang et al. 2005) and was subsequently further used to analyze the interaction between economic entities. In combination with the practice of contemporary Chinese social transformation, many scholars have tried to establish an analytical framework of embedded governance in line with China's social reality (Jiang 2021;Yu et al. 2021) and proposed that the realization of the expected goals of grassroots governments can be influenced by three embedding methods: system embedding, organizational embedding, and interest embedding. Based on this, this paper defines the grassroots autonomous organization as the grassroots village committee and the villagers' groups. Combined with the actual research situation and the existing embedding index division system, the three factors obtained from the factor analysis are named institutional embedding, organizational embedding, and emotional embedding. Among them, the institutional embedding of grassroots autonomous organizations refers to the standardization and legitimacy of grassroots autonomous organizations in the process of promoting farmers to exit the homestead, as well as the guarantee of farmers' right to know (Shi et al. 2022;Tang et al. 2020;Zhang et al. 2021). This paper adopts the structure, politics, and other indicators to express. The organizational embedding of grassroots autonomous organizations means that in the process of promoting homestead withdrawal, grassroots autonomous organizations fully consider the reasonable demands of farmers, protect the legitimate interests of farmers, and respect the will of farmers (Shi et al. 2022;Tang et al. 2020;Zhang et al. 2021). This paper uses benefit and cognitive indicators to express. The emotional embedding of grassroots autonomous organizations refers to the respect and protection of farmers' traditional culture, behavioral cognition, and rural feelings when implementing homestead withdrawal policies (Shi et al. 2022;Tang et al. 2020;Zhang et al. 2021). This paper adopts emotional and cultural indicators to express.
Grassroots autonomous organizations are the intermediary of transmitting policy information and ensuring people's livelihood between the grassroots government and the farmers (Liu and Li 2017). It is not only the propagandist of the government homestead withdrawal policy but also the implementer of the farmers' homestead withdrawal policy. The embedding of grassroots autonomous organization can make it better realize its function of transmitting information from above and uploading information from below . Village collective organizations are embedded in the village acquaintance society to conduct homestead management, and their behaviors are affected by the rural governance system and tasks (Liu and He 2018). The function of village collective organization plays a direct role in the decision of homestead withdrawal of farmers and then affects the farmers whether to homestead exit (Sun et al. 2020). Generally speaking, the leaders of grassroots autonomous organizations are farmers themselves. To improve the embedded level of grassroots autonomous organizations can analyze and deal with various problems in the process of mobilizing farmer homestead withdrawal from the perspective of farmers, and it is also conducive to dispelling the stereotype of farmers that grassroots autonomous organizations are only agents of the government and developers, eliminate the potential friction between grassroots autonomous organizations and farmers; thus, farmers' homestead withdrawal decisions will be affected (Zhou et al. 2018). Different embedded levels of grassroots autonomous organizations have different effects on farmers' homestead withdrawal.
When the grassroots autonomous organizations have a high embedding level in terms of system and interests, they can meet the requirements of farmers for the right to know the withdrawal policy and the right to participate in the withdrawal compensation standard formulation, thus safeguarding farmers' legitimate interests in the process of homestead withdrawal from the multi-interest game, and to improve the prestige of grassroots autonomous organizations, furthermore, farmers are more likely to accept the homestead withdrawal mobilization activities of grassroots autonomous organizations. When the grassroots autonomous organizations have a high embedding level in terms of culture and emotion, they can maximize meet the psychological and emotional demands of farmers, so that farmers are reluctant to leave the countryside and reluctant to homestead withdraw. In conclusion, the embedding of grassroots autonomous organizations has an impact on the farmers' homestead withdrawal behavior. The stronger the embedding level of grassroots autonomous organizations in terms of system and interests, the more likely farmers are to exit homestead. The stronger the embedding level of grassroots autonomous organizations in terms of emotion, the less likely farmers are to exit the homestead. Based on the above analysis, this paper puts forward the research hypothesis H1. H1: The system embedding level of grassroots autonomous organization and the interests embedding level of grassroots autonomous organization have a significant positive impact on the farmer homestead withdrawal behavior. The emotion embedding level of grassroots autonomous organizations has a significant negative impact on the farmer homestead withdrawal behavior. The overall impact of the embedding level of grassroots autonomous organizations on the farmer homestead withdrawal behavior is uncertain and needs to be further tested empirically.
The impact of the non-agriculture employment on the farmer homestead exit behavior Non-farm employment is a rational decision of farmers to maximize household total welfare effect, and it is an important driving force to promote the homestead resources circulation. The farmer's income level has a significant effect on the farmer's homestead withdrawal behavior (Sun et al. 2020). At present, the farmers' income structure shows a non-agricultural orientation, and the proportion of nonagricultural income in the total income is steadily increasing (Cheng et al. 2021;. It can be seen that the level of farmers' non-agricultural employment has an important impact on their homestead withdrawal behavior. Therefore, this paper first chooses a non-agricultural employment level index to study household non-agricultural employment. The difference in non-agricultural employment attributes will lead to the different directions and degrees of non-agricultural employment levels, while the non-farm employment distance can measure the stability of part-time employment and non-farm employment of farmers (Huang et al. 2022). Therefore, this paper further chooses the non-farm employment distance index to study the impact of non-farm employment on farmers' homestead exit behavior.
The farmer household non-agriculture employment level is one of the main reasons for the homestead exit behavioral decisions of farmers. This paper selects the farmer household non-agricultural employment rate to measure the farmer household non-agricultural employment level (Chang et al. 2021;Liu and Zhang 2018). Homestead withdrawal is a family choice behavior, which requires a unified opinion among family members on whether to withdraw (Sun et al. 2020). When the household non-farm employment rate is low, more family members are engaged in agricultural production, and the household income is relatively low, so the ability to deal with potential risks after exit is lower, and the willingness to exit the homestead is lower. When the household non-farm employment rate is high, the household income is relatively higher, the ability to deal with potential risks is stronger, and the family members have higher requirements for a convenient and comfortable living environment, which is conducive to the withdrawal of household homestead. Based on the above analysis, this paper puts forward the research hypothesis H2.
H2: The non-farm employment level has a positive effect on the farmer's homestead withdrawal behavior.
The non-agriculture employment distance is one of the main reasons for the homestead exit behavioral decisions of farmers. The part-time employment conditions in different employment places are different, which will lead to different support and limiting factors for the career development of the rural labor force and then affect farmers' choice of rural homestead withdrawal (Xu et al. 2022). The distance judgment standard of non-agricultural employment is within and outside the county (Xu et al. 2022). In this paper, the non-agricultural employment distance of rural household migrant laborers is divided into three types: near distance (in the local county), a little long distance (within the province), and remote distance non-farm employment (outside the province). For near distance non-agricultural employment farmers, their social networks such as clan blood and human relations serve as a link to drive farmers to help each other and disperse the non-agricultural employment risks they faced, seeking more opportunities for their non-agricultural employment and taking more measures to deal with potential risks; therefore, farmers are more likely to exit homestead . For a little long distance non-agricultural employment farmers, their social networks and professional skills are weaker, their material base is poorer, job and livelihood stability are less stable, and they face more potential risks, therefore, rural homestead as their final foothold and reserve assets, they choose to withdraw from the probability of homestead is small (Zhang and Xia 2021a). For remote distance non-agricultural employment farmers, long-term employment experience maintains a stable employment relationship, with relatively higher professional skills, higher career salary, and higher non-farm employment income, the accumulated material capital makes them pay more attention to the time and energy cost of dealing with the withdrawal of homestead, and do not pay much attention to the reserve assets value of the homestead, and their connection with the countryside is relatively low, the emotional connection is shallow (Sun et al. 2020). Therefore, farmers are more likely to exit the homestead. Based on the above analysis, there may be a "U"-shaped trend between farmers' non-farm employment distance and their homestead withdrawal decision behavior when other conditions are unchanged. So, this paper puts forward the research hypothesis H3.
H3: The non-farm employment distance has a U-shaped impact on the farmer homestead exit behavior.

Analysis of the moderating effect of non-farm employment on the embedding level of grassroots autonomous organizations and the exit behavior of farmers' homesteads
According to the above analysis, we show that the embedding level of grassroots autonomous organizations has a significant impact on the farmer homestead exit behavior. In the Chinese homestead withdrawal pilot practice, due to the limited amount of withdrawal compensation, farmers need to pay certain extra costs in the process of homestead withdrawal (such as the decoration of new housing and the purchase of furniture), which requires farmers to consider their payment capacity . With the improvement of the urban economic development level and the household registration system reform, farmers will consider more expected returns and risk expectations when making decisions on whether to exit homestead (Su et al. 2020). Farmers with different degrees of occupational structure differentiation and non-farm employment stability have different expected returns and ability to cope with risks, trust radius, and trust level also differ, so they have different acceptance degrees of the credibility and authority of grassroots autonomous organizations.
In terms of non-farm employment levels, when the farmer's non-farm employment level is low, the relationship between the farmer and the countryside is relatively close. At this time, the farmer's income level is relatively low and the trust radius is small; there is a high relationship and closeness between grassroots autonomous organizations and the low non-agricultural employment level farmer, the farmer has a strong trust and dependence on the grassroots autonomous organizations. Therefore, when considering whether to withdraw from homestead, they will be affected by grassroots autonomous organizations. When the farmer's non-farm employment level is high, the farmer's income level is high, and the trust radius is expanded, the farmers have low trust in the grassroots government. Therefore, the authority of grassroots autonomous organizations for farmers with high non-agricultural employment rates declined, whether the homestead exits or not reflects the independent decision of the farmers. In terms of non-farm employment distance, the longer the non-farm employment distance, the weaker the social network between the farmer and rural areas, and the smaller the dependence and trust of the farmer on grassroots autonomous organizations. Therefore, grassroots autonomous organizations have less impact on farmers' decisions to exit the homestead. Based on the above analysis, this paper puts forward the research hypothesis H4.
H4: Non-agricultural employment plays a moderating role in the embedded level of grassroots autonomous organizations and the farmer homestead withdrawal behavior. With the increase of the farmer non-farm employment level, the effect of grassroots autonomous organization embedded level on the farmer homestead withdrawal behavior will be weakened. With the length of the farmer non-farm employment distance, the effect of grassroots autonomous organization embedded level on the farmer homestead withdrawal behavior will be weakened ( Fig. 1).

Study area and data collection
The study selects Qionglai City, Sichuan Province, as a study area to analyze the effect of different grassroots autonomous organizations' embedded modes on farmer's homestead withdrawal decisions. The selection of the study area follows the principles of random sampling and (1) since 2008, Chengdu has been one of the pioneer demonstration points in the reform of the homestead property rights system. In 2013, Qionglai City issued the voluntary and compensated policy of withdrawal homestead, and in 2016, it became one of the pilot reforms of withdrawal homestead in Chengdu. By the end of 2018, 3520 people from 1189 Qionglai households had voluntarily withdrawn from their rural homestead, covering an area of 67hm 2 . Qionglai City homestead withdrawal has achieved remarkable results, forming a set of experiences that can be generalized; (2) Qionglai City is located in the southwest of Chengdu City, covering 1377 square kilometers. The municipal government is located in Linqiong Town, 65 kilometers away from the main urban area of Chengdu, the economic level is higher, and farmers' non-farm employment level is higher, so, it is suitable to study the level and distance of farm households' non-farm employment; (3) Qionglai City is the first batch of pilot areas of homestead reform in China, local grassroots autonomous organizations have a high embedded level in farmers and pay great attention to the production and life problems of farmers before and after the withdrawal of homesteading, so it is suitable to study the grassroots autonomous organizations' embeddedness. The selection process of the sample farmers is as follows: (1) the research data are derived from field survey data in Qionglai City, Sichuan Province, from September to October 2019. Combined with the theoretical framework and the actual situation of Qionglai City, the research group first designed the survey questionnaire, carried out a presurvey, and modified and improved the questionnaire for a formal survey; (2) the contents of the questionnaire included basic information of household farmers, household farmland and housing status, non-farm employment, and social security status, homestead withdrawal, etc. According to the location conditions, homestead withdrawal implementation and, economic development status, Qionglai City Sangyuan Town, Wolong Town, Linqiong Town, Qianjin Town, Baolin Town, Daozuo Township, and Tea Garden Township were selected as the investigation areas; (3) exited and unexcited farmers households are selected from the above areas by a random sampling method. Of 450 collected questionnaires, after processing the incomplete value, 421 questionnaires are valid with an effective rate of 93.55%, respectively, including a total of 248 exited households and 173 unexcited households; (4) the data were collected through face-to-face interviews to ensure the data quality. The questionnaire was filled out by the investigators according to the results of the interviewee answers, and a centralized test was conducted after the survey (Fig. 2).

Variable setting and descriptive statistics
Farmers' homestead withdrawal behavior is the explained variable in this paper, which is measured by the question "whether farmers exit homestead" in the questionnaire. It is a binary variable; "0" means that the farmer has not exited the homestead, and "1" means that the farmer has exited the homestead.
The embedded level of grassroots autonomous organizations is the core explanatory variable in this study. Concerning existing studies, Yu et al. (2021) divided embedding into political embedding, cultural embedding, cognitive embedding, and structural embedding . Jiang (2021) proposed three embedding methods: institutional embedding, organizational embedding, and interest embedding; this study constructs an embedded index system from three aspects: institutional embedding, organizational embedding, and emotional embedding. Further, combined with the reality that China's grassroots organizations mobilize farmers to homestead withdrawal behavior, this study selects 8 indicators from the above three aspects to depict the embedding of grassroots organizations; then the core explanatory variables are obtained by factor analysis; then the core explanatory variables are obtained by factor analysis. The 8 indicators and factor analysis results are shown in Table 1.
The non-farm employment level and the non-farm employment distance of the household head are moderating variables in this study. This study uses "the farmers' household non-farm employment rate" to measure the non-farm employment level (Zhou et al. 2017;Xiang and Le 2021;Liu et al. 2021;Chang et al. 2021;Liu and Zhang 2018). The farmers' household non-farm employment rate is the ratio of household non-farm employment number to the total household population. This study divides the non-agricultural employment distance into three types: near distance (in the local county), a little long distance (within the Sichuan province), and remote distance non-farm employment (outside the Sichuan province) (Chang et al. 2021).
The township size in which the investigated village is located is an instrumental variable in this study. This study may have omitted variables and reverse causality, resulting in data imbalance. To solve the endogeneity problem of the model, " the township size in which the investigated village is located" is taken as the instrumental variable of the core explanatory variable (Su et al. 2020). The reasons are as follows: (1) the township size in which the investigated village is exogenous to the embedded game between grassroots (2) grassroots autonomous organizations have the dual status of village leaders and grassroots government agents, village homestead withdrawal is related to township regional development planning; therefore, the level of grassroots autonomous organizations embedded in household homestead withdrawal will be affected by the township government and township development level. Township size is one of the important indicators to measure the township development level, so it is related to the grassroots autonomous organizations embedded level; (3) this paper uses the population number of the surveyed towns in the seventh National census data to measure their size. This paper mainly includes the interviewees' characteristics, family characteristics and homestead features, and policy features as the control variables.
The definitions of all variables introduced in this paper and the corresponding descriptive analysis are shown in Table 2.

Empirical models
The explained variables in this paper, whether farmer exit homestead, are both binary dummy variables. Therefore, the Probit model is a suitable choice for the following analysis, and the IV-Probit model is a suitable choice for the endogeneity test. We set the benchmark model as follows: where Y i represents the homestead withdrawal behavior of household i. X i represents the institutional embedding level, organizational embedding level, and emotional embedding level of grassroots autonomous organizations embedding level evaluated by the household i. U i indicates the farmers' household non-farm employment rate and the non-farm employment distance of household i. Z i represents a series of control variables, including the interviewees' characteristics, family characteristics and homestead features, and policy features. ζ i is random errors. α 0, β i, γ i, η i are the parameter of the model to be estimated.
To further verify whether non-farm employment has a moderating effect on the grassroots autonomous organizations embedding level and the farmer homestead withdrawal behavior, the interaction item between non-farm employment and grassroots autonomous organizations embedding level is introduced in the model. The construction model is as follows: (1) where X represents the grassroots autonomous organizations embedding level evaluated by the household i evaluated by the household i. XU i represents the interaction item between the grassroots autonomous organizations embedding level evaluated by the household i and the farmers' household non-farm employment rate of household i, and between the grassroots autonomous organizations embedding level evaluated by the household i and the non-farm employment distance of household i, respectively. Other letters have the same meaning as in (1).

The analysis of the grassroots autonomous organizations embedding level
The KMO statistic was 0.803. The χ 2 statistic of Bartlett's sphericity test was 853.137, indicating that there was a correlation along the 8 indicators, which was suitable for factor analysis. After factor rotation using the maximum variance method, three common factors were obtained from 8 indicators, and the cumulative explained variance after rotation was 64.893%. According to each factor score and variance contribution rate, the comprehensive index is calculated as follows: It can be seen from the above equation that the system embedding of grassroots self-governing organizations has the greatest influence on the grassroots self-governing organizations embedding level. Therefore, measures such as standardizing homestead exit policy and establishing prestige can significantly affect the embedment level. Based on this, the grassroots autonomous organizations embedding level and grouping statistics in the research area are calculated (Table 3).
As shown in Table 3, based on the perspective of farmers in the research area, the comprehensive index of the grassroots autonomous organizations embedding level is located between-1.6 and 1.2, 201 households thought the grassroots autonomous organizations embedding level was less than average, while 217 households thought it was higher than average. This shows, there is a big difference in the evaluation of the grassroots autonomous organizations embedding level among different farmers, and the maximum and minimum values are located in the organizational embedding level and the emotional embedding level, respectively. It shows that farmers pay more attention to the organizational embedding level of grassroots autonomous organizations and pay less attention to the system embedding and emotional embedding. It is analyzed from three embedding methods, in the institutional embedding level of grassroots autonomous organizations, there were 188 households (44.7%) of farmers believed that the system embedding level was lower than average, and 233 households (55.3%) thought that they higher than average. In the organizational embedding level of grassroots autonomous organizations, there were 185 households (43.9%) of farmers believed that the organizational embedding level was lower than average, and 236 households (56.1%) thought that they were higher than average. In the emotional embedding level of grassroots autonomous organizations, there were 245 households (58.2%) of farmers believed that the organizational embedding level was lower than average, and 176 households (41.8%) thought that they were higher than average.

Benchmark regression
The results from estimating Eqs. (1) and (2), the benchmark regression, are shown in Table 4. VIF values were all less than 10, indicating no collinearity problem. The LR test shows that the fitting equation is highly significant overall. Model 1 takes the grassroots autonomous organizations as an explanatory variable. Model 2 includes non-farm employment as explanatory variable. Model 3 also included 3 kinds of grassroots autonomous organizations embedding levels and non-farm employment as explanatory variables. In the three models above, whether the farmer exits the homestead is the explained variable. 1 3 The impact of grassroots autonomous organizations embedding level on farmer homestead withdrawal behavior. The results show that when controlling for other variables (Table 4), when the impact of the non-farm employment variable is not considered, the institutional embedding level and the organizational embedding level of grassroots autonomous organizations have a significant positive impact on the farmer homestead withdrawal behavior, while the emotional embedding level of grassroots autonomous organizations has a significant negative impact on the farmer homestead withdrawal behavior. The positive effect of institutional embedding and organizational embedding is much greater than the negative effect of the grassroots autonomous organizations' emotional embedding; therefore, the stronger the level of grassroots autonomous organizations embedded in farmers' groups, the more likely they are to carry out homestead planning for farmers' rights and interests and get the support and trust of farmers, and the more likely they are to mobilize farmers to exit homestead. It can be seen that the grassroots autonomous organizations embedding level has a positive and significant impact on the farmer homestead withdrawal behavior, that is, hypothesis H1 is valid.
When the impact of the non-farm employment variable is not considered, the institutional embedding level of grassroots autonomous organization has a significant impact on the farmer homestead withdrawal behavior and conforms to the expected hypothesis, the impact direction of grassroots autonomous organizations' organizational embedding level and emotional embedding level on the farmer homestead withdrawal behavior conforms to the expected hypothesis but fails the significance test. The possible reason is, the higher the non-agricultural employment rate of the farmer, the higher the non-agricultural employment income of the farmer; the farmer will consider the time cost and salary loss in the process of negotiating the compensation plan of homestead withdrawal. Therefore, the farmer non-farm employment level will offset the impact of the grassroots autonomous organizations organizational embedding level on the farmer homestead withdrawal behavior. The farther the non-agricultural employment distance, the weaker the connection between farmer and villages, and the smaller the influence of rural feelings and clan affection on the farmer. Therefore, the farmer non-farm employment distance will offset the impact of the grassroots autonomous organizations' emotional level on the farmer homestead withdrawal behavior. The impact of non-agricultural employment on farmer homestead withdrawal behavior. The results show that when controlling for other variables (Table 4), the nonagricultural employment level has a significant positive impact on the farmer homestead exit behavior. Homestead withdrawal behavior is a collective choice behavior of family members. The more members engaged in nonagricultural employment, the more likely farmers are to pursue their own development needs under the impact of the market economy, and the weaker their emotional connection to the countryside, so they are more likely to exit the homestead. Hypothesis H2 is valid. The coefficient of the first term of non-agricultural employment distance is negative, which is consistent with the research conclusion of other scholars. To further study the effect of nonagricultural employment distance on farmers' homestead withdrawal behavior, a quadratic term of non-agricultural employment distance was added to the model. The quadratic coefficient of non-agricultural employment distance is positive, and the non-agricultural employment distance has a "U-shaped" impact on the farmer homestead withdrawal behavior. Near distance non-agricultural employment farmers have better social capital, little long-distance employment farmers have better material capital, and they have more possible opportunities and measures to deal with risks. Therefore, they are more likely to exit the homestead than remote distance non-agricultural employment farmers with weak social relations and low skills; that is, hypothesis H3 is valid.

Endogeneity test
The instrumental variable method is based on the premise that not all explanatory variables in the model are exogenous, therefore, the Wald test was performed on the model first. The Wald test value was 3.70 (P=0.0543), which was significant at 1% statistical level. Therefore, there are endogenous problems in the model, and it is reasonable to use the instrumental variable method. According to the above analysis, the township size in which the investigated village is selected as the instrumental variable to meet the requirements of exogeneity and correlation. In the first stage regression results, the instrumental variable is significantly correlated with the grassroots autonomous organizations embedding level, and the F-statistic is greater than 10, indicating that there is no weak instrumental variable problem. The regression results of the second stage are shown in Table 5. The marginal effect of grassroots autonomous organizations embedding level in the Probit model results is 0.129, while the marginal effect of grassroots autonomous organizations embedding level in the IV-Probit model results is 7.417. This indicates that if the endogeneity problem is not dealt with, the impact of the grassroots autonomous organizations embedding level on farmer homestead withdrawal behavior will be underestimated. Similarly, the estimation results of the impact of the non-agricultural employment level and the non-agricultural employment distance on farmer homestead withdrawal behavior will also be underestimated. Therefore,

The moderating effect of the farmer non-agricultural employment
To verify whether there is a moderating effect of nonagricultural employment on the grassroots autonomous organizations embedding level and the farmer homestead behavior, the interaction item between the grassroots autonomous organizations embedding level and the farmer non-farm employment index was constructed, and the IV-Probit model was used for regression. The secondstage regression results are shown in Table 6. The results indicate that both variables passed the significance test when the grassroots autonomy organization embedding level and the farmer's non-farm employment level were simultaneously put into the model; then, the interaction item between the grassroots autonomous organizations embedding level and the farmers off-farm employment level was put into the model, and the interaction item had a significant negative impact on the farmer homestead withdrawal behavior. This indicates that the farmer non-farm employment level plays a restraining role in the positive effect of grassroots autonomous organization embedding level on the farmer homestead withdrawal behavior. The larger the non-farm employment population in the family, the more affected the family is by the market economy and rationalism, and the more likely it is to have a lot of interference and dissatisfaction with the withdrawal and mobilization of grassroots autonomous organizations striving to maximize collective interests. Therefore, it will play a restraining role in the positive impact of grassroots autonomous organizations embedding level on the farmer exit behavior. The interaction item between the grassroots autonomous organizations embedding level and the farmer off-farm employment distance was put into the model, and the interaction item had no significant effect on the farmer homestead exit behavior. Therefore, the off-farm employment distance did not have a moderating effect between the grassroots autonomous organizations embedding level and the farmers' homestead exit behavior. In conclusion, hypothesis H4 is partially confirmed.

Robust analysis
Considering that the selected model will have an impact on the regression results, the Tobit model is further selected for the robustness test in this paper (Table 7). According to the micro-data obtained from the survey, using the isometric grouping method, the farmer non-agricultural employment level between 0 and 25% is considered as a low non-agricultural employment rate, between 25% and 50% is considered as a little low non-agricultural employment rate, between 50% and 75% is considered as a little high non-agricultural employment rate, and between 75% and 100% is considered as high non-agricultural employment rate. For specific practices, first, only the impact of the grassroots autonomous organizations embedding level on the farmer homestead exit behavior was studied (Model 10). Secondly, according to the non-agricultural employment level of farmers, farmers are divided into four groups: a low non-agricultural employment rate (Model 11), a little low non-agricultural employment rate (Model 12), a little high non-agricultural employment rate (Model 13), and high non-agricultural employment rate (Model 14); then, comparing analyzes the moderating effect of non-agricultural employment level in the grassroots autonomous organizations embedding level and the farmer withdrawal behavior. The results are shown in Table 7. Without adding the farmer non-agricultural employment level, the institutional embedding level, organizational embedding level, and emotional embedding level of grassroots autonomous organizations all have significant effects on the farmer homestead withdrawal behavior at the 10% statistical level. When the non-agricultural employment level of farmers is added, only the grassroots autonomous organization institutional embedding level significantly affects the farmer homestead withdrawal behavior at different non-agricultural employment levels. It verifies the hypothesis that the farmer's nonagricultural employment level has a moderating effect on the grassroots autonomous organizations embedding level and the farmer homestead withdrawal behavior. Grouping analysis found that under the different farmer non-agricultural employment rates, the indirect effect of the grassroots autonomous organization institutional embedding level on the farmer withdrawal homestead behavior is less than the direct positive effect, and the impact of high nonagricultural employment rate is less than the impact of low non-farm employment rate. This indicates that the farmer's non-agricultural employment rate plays a restraining role in the positive impact of the grassroots autonomous organization embedding level on the farmer homestead withdrawal behavior, and the restraining effect is strengthened with the increase of the farmer's non-agricultural employment rate. It can be seen that the grassroots autonomous organization embedding level has a positive impact on the farmer homestead withdrawal behavior on the whole, and the farmer non-agricultural employment rate plays a moderating role between the grassroots autonomous organizations embedding level and the farmer homestead withdrawal behavior. This is consistent with the analysis results above, so the analysis results of this paper are robust.

Conclusions and policy recommendations
To deepen the homestead system reform and promote the farmers' voluntary and orderly withdrawal homestead is an inevitable requirement for the Chinese government to stimulate the potential driving force of rural development and achieve common prosperity. This study adopts the embeddedness theory to analyze the mechanisms by which the effect of grassroots organization embedding level on the farmer homestead withdrawal behavior. Based on 421 questionnaire data from households in Qionglai City, Sichuan Province. The Probit model and IV-Probit model were applied to examine the effect of grassroots organization embedding level on the farmer homestead withdrawal behavior and to examine the moderating of the farmer nonagricultural employment rate and the farmer non-agricultural employment distance. The main conclusions are as follows: first, the village committee and villager group strengthening the connection with farmers can promote the farmer's homestead withdrawal behavior. The village committee and the villagers group in the process of implementing the farmers homestead exit policy, guaranteeing farmers' right to know has a positive impact on the farmer homestead withdrawal behavior, fully considering the reasonable demands of farmers has a positive impact on the farmer homestead withdrawal behavior, respecting and protecting village traditional culture has a negative effect on the farmer homestead withdrawal behavior, and guaranteeing farmers' right to know has the greatest influence on the farmer homestead withdrawal behavior. Second, the farmer's non-agricultural employment has a significant impact on the farmer's homestead withdrawal behavior. The farmer's non-agricultural employment rate can promote the farmer's homestead withdrawal behavior. There is a U-shaped relationship between the quadratic term of the farmer non-farm employment distance and the farmer exit behavior. Third, the farmer non-farm employment rate has a moderating effect on the grassroots autonomous organizations embedding level and the farmer homestead withdrawal behavior, which hinders the mobilization of grassroots autonomous organizations to persuade the farmer to exit homestead. Under the moderating effect of the farmer non-farm employment rate, the grassroots autonomous organizations embedding level has different effects on the farmers' exit behavior with different non-farm employment rates; in addition, different embedding methods of grassroots autonomous organizations have different effects on different farmer non-agricultural employment rates. Finally, the hypothesis that the farmer non-farm employment distance plays a moderating role on the grassroots autonomous organizations embedding level and the farmer homestead withdrawal behavior has not been confirmed, which needs further analysis and testing. Based on the above conclusions, this study has several noteworthy areas for discussion. Firstly, the market-oriented reform of land factors is the key to the market-oriented reform of the urban-rural dual structure. Therefore, the experience and conclusions of China's government-owned residential land reform provide new ideas for developing countries to solve the problem of the urban-rural dual structure. Secondly, due to the clear land property rights system in Western countries, the concept of residential land exit does not exist. However, China's residential land reform can provide some reference for similar research in Western countries, such as land markets and land expropriation compensation. Finally, China has a vast territory, and there are significant differences in the elections and development of grassroots autonomous organizations in different regions. Limited by the research data, this study only takes the example of Qionglai City for research and draws relevant conclusions on the relationship and influence mechanism between the embedding of grassroots organizations, non-agricultural employment, and the exit of farmers' residential land. In the subsequent data research, the heterogeneity analysis of clan culture and government regulations in different regions is a good perspective for research.
Based on the above conclusions and discussion, to improve the ability of grassroots autonomous organizations to embed farmers and promote farmers to voluntarily and orderly exit homesteads, this paper puts forward the following suggestions: first, strengthen the relationship between grassroots autonomous organizations and farmers. Grassroots autonomous organizations should formulate specific action plans for the farmer homestead exit from the perspective of farmers, constantly reflect the voice of farmers, and improve the standardization and legitimacy of the institutional decisions in the process of withdrawing homestead. Second, continue to strengthen the development of grassroots autonomous organizations. Grassroots autonomous organizations should give full consideration to the reasonable demands of farmers, make them open to public plans and financial affairs, protect farmers' right to know about collective actions, and create a good rural social atmosphere. In the context of rural revitalization strategy. Grassroots autonomous organizations should actively develop and strengthen the collective economy, promote the village's economic development, encourage farmer households to obtain non-farm employment nearby, accelerate rural urbanization, and promote the citizenization process of farmer households. Third, strengthen skills training for farmers. To improve the ability of farmers to adapt to market competition, enhance the stability of farmers' jobs, and improve the level of farmers' household income. Finally, grassroots autonomous organizations should strengthen the publicity of homestead reform policies, improve farmers' awareness of homestead reform policies, and ensure the orderly implementation of homestead reform.