Introduction

China’s ongoing urbanization and industrialization are producing significant changes in human-land relationships in rural China. Many rural workers were relocating to urban areas (Long et al. 2012), causing the rural population to decrease from 790 million in 1978 to only 477 million in 2023 (National Bureau of Statistics of China 2024). However, rural housing land area increased by 14 million hectares between 1995 and 2014 (Gao et al. 2017). In 2018, the vacancy rate of rural housing land reached 20% (Kong et al. 2018), much of it laying idle or inefficiently managed (Song et al. 2021). China’s central government has explored policies to promote voluntary, compensatory withdrawal from rural housing land (WRHL) in certain areas to increase housing land use. The document “Opinions on further Promoting the Reform of the Household Registration System” was issued by the State Council of the People’s Republic of China in 2014, proposing that local governments carried out a WRHL pilot program to investigate the willingness of villagers who had settled in the cities and towns to withdraw from their rural housing land. Fifteen counties (including cities and districts) in China participated in such reform in 2015. In addition, the Chinese government carried out a new round of WRHL pilot reform in 105 counties and three cities from 2020 to 2022. In 2025, the Chinese government proposed to establish a method for rural villagers who had settled in cities and towns to withdraw from rural housing land voluntarily and with compensation. Broadly speaking, WRHL refers to the Chinese government’s policy of using compensation to encourage villagers who rely little on their farmland and have already moved to cities and towns or who own urban properties to voluntarily relinquish their idle rural housing land, including any housing built on it. After villagers relinquish their housing land, they can move into cities or concentrated residential areas built by village collectives (Chen et al. 2017; Liu et al. 2021).

Decreases in the rural permanent population and inefficient use of rural residential land have also been observed in developed countries in Europe and the Americas (Gkartzios and Scott 2010; Park et al. 2022), where rural residential land consolidation has become an important way to address these issues (Pašakarnis et al. 2013; Podhrázská et al. 2015). Some studies have found that formal institution, landowners’ perception of land consolidation, and social capital impact villagers’ decision to participate in residential land consolidation (Kupidura et al. 2014; Lisec et al. 2014). Before the implementation of the relevant policy, a small number of villagers did not support the rural residential land consolidation. They were worried about leaving the land where they had long lived and rural residential land consolidation could easily destroy the local landscape and ecological environment (Lisec et al. 2014). However, most villagers still believed that the benefits of residential land consolidation outweighed the implementation costs, and that residential land consolidation could improve their living environment and landscape (Kupidura et al. 2014). In addition, most villagers still had a positive and supportive view of rural residential land consolidation after the end of the program (Podhrázská et al. 2015). In the process of land consolidation, governments in European countries have established open and transparent institutional frameworks, actively encouraged villagers to participate in residential land consolidation, and striven to increase villagers’ satisfaction with consolidation (Demetriou et al. 2012; Lisec et al. 2014). Those countries’ residential land consolidation policies have effectively met the living needs of residents, improved their living environment, and promoted the development of the local rural economy (Brink and Molema, 2008; Dudzińska et al. 2018).

Although there are differences in social background and institutional environment, the consolidation of rural residential land in developed countries has provided experience for the implementation of similar WRHL reform in China. Because WRHL reform in China aims to improve the efficiency of rural land use and promote rural revitalization (Huang et al. 2018; Wang et al. 2023), its reform demands are similar to those of relevant policies in developed countries, and villagers are the main stakeholders (Lisec et al. 2014; Podhrázská et al. 2015; Song et al. 2021). In addition, as in developed countries, rural housing land in China can be rented and transferred, but not sold to urban residents. Thus, a study of WRHL reform in China and the factors that affect villagers’ participation can provide important insights for rural residential land consolidation in other parts of the world.

Rural housing land refers to the collective construction land allocated by the village collective to villagers for residential housing free of charge (Jiang et al. 2016; Kong et al. 2018; Zhou et al. 2020). As the important asset of Chinese villagers, rural housing land has asset income value and serves important functions in terms of social security and auxiliary farm production (Guanghui et al. 2016; Su et al. 2022). Therefore, the Chinese government is cautious regarding WRHL. On the one hand, the central government requires respect for the villagers’ choices to withdraw from their housing land without compulsion. On the other hand, the central government requires that WRHL reform be focused on rural villagers who have settled in cities and towns, that is, villagers who rely less on their farmland, have moved to cities and towns, or own urban properties (Chen et al. 2017; Liu et al. 2020; Cui et al. 2025). The specific policy content of WRHL reform can be found as Supplementary Table S2 online. Villagers who rely less on their farmland and own urban properties have a low dependence on the social security and auxiliary agricultural production functions of rural housing land, which can effectively reduce implementation cost and conflict risk from WRHL reform. Therefore, these villagers have been chosen as the primary target group of WRHL policy. However, the following questions remain: Will the target villagers choose to actually withdraw from their rural housing land? Are villagers with low farmland dependence more likely to choose to withdraw than those with high farmland dependence? Are villagers with multiple urban properties more likely to choose to withdraw than those who do not own urban properties? These three questions directly affect whether the expected objectives of WRHL policy can be achieved.

Much research has focused on the factors that influence villagers’ willingness to withdraw from rural housing land. Age, gender, education level of the household head, family economy, household size, rural housing floor area, building structure, land rights certification, rural housing land transport location, rural housing benefit expectations, WRHL compensation, hukou system, and rural social networks have been found to be the factors with the greatest impact (Liang et al. 2022; Shi et al. 2022; Gao et al. 2023; Xie et al. 2023; Xia et al. 2024; Cui et al. 2025; Tang et al. 2025). For farmland dependence and urban housing, prior studies have defined farmland dependence as the proportion of farm income to total household income (Duan et al. 2016; Li and Cai 2018), while the number of urban properties has generally been used to quantify urban housing (Wang and Gu 2016; Zhang and Fan 2020). Although prior studies have identified the proportion of farm income and urban housing as factors influencing villagers’ willingness to withdraw, these studies have not analysed how these two factors affect villagers’ willingness to withdraw, and lack the analysis of their impact mechanisms. For instance, Xia et al. (2024) used structural equation modelling to explore the factors influencing villagers’ decisions to withdraw from housing land in the Chengdu–Chongqing area, and found that the proportion of farm income to total household income had a significantly negative impact on villagers’ willingness. However, they did not explore the influence mechanism involved. Zhang and Fan (2020) used the propensity score matching method to examine the impact of the Chinese housing market on villagers’ willingness to withdraw, and found that owning urban housing was correlated with a larger likelihood of withdrawal, but their study similarly did not explore the underlying mechanism.

In summary, relevant studies to date have significant shortcomings. First, few studies have examined whether WRHL policy target groups actually choose to withdraw from the standpoint of policymakers. Second, the target groups of China’s WRHL policy are villagers who rely little on their farmland or own urban properties. Although some studies have paid attention to the impact of farmland dependence and urban housing on villagers’ choices to withdraw from rural housing land, they have not considered the interactive effect of farmland dependence and urban housing on villagers’ withdrawal choices nor the underlying mechanisms, particularly the mediating effect of expected value.

To determine the answers to the above three questions and compensate for the shortcomings of prior studies, this study investigated the Chengdu pilot program to examine villagers’ choices to withdraw from rural housing land. Chengdu was selected as the research unit because it was one of the areas chosen for urban-rural integration pilot reform, and there have been extensive explorations of WRHL policy in this region. The Pidu District in Chengdu was identified as one of fifteen WRHL pilot areas by the central government in 2015, and it was also chosen to be one of China’s new WRHL pilot reform areas from 2020 to 2022. Therefore, by analysing survey data from 1515 villagers in Chengdu, this study investigated the impact of farmland dependence and urban housing on villagers’ choices to withdraw from rural housing land, with the goal of determining whether WRHL policy target groups choose to withdraw and whether the expected objectives of WRHL policy can be achieved.

The key contributions of this study are as follows. First, this study provides a new perspective for investigating whether the expected objectives of public policy can be achieved, i.e., it examines WRHL policy target groups from the standpoint of policymakers to determine whether the expected policy objectives can be achieved. By doing so, it expands the realm of public policy evaluation and policy target group analysis. Second, although a few studies have focused on the impact of farmland dependence and urban housing on villagers’ choices to withdraw from rural housing land, they have not considered the specific influence mechanisms and mediation factors of those impacts. Specifically, existing studies have not provided an answer to explain how farmland dependence and urban housing affect villagers’ withdrawal choices, and what their influence mechanisms and mediation factors are. This study investigates the interactive effect of farmland dependence and urban housing on villagers’ choices to withdraw and the mediating effect of expected value, further discusses the underlying impact mechanisms, and thereby bridges the knowledge gaps in the existing literature. Third, this study discusses the relationships linking farmland dependence and urban housing to villagers’ withdrawal choices based on the theory of Homo economicus, i.e., it posits that villagers’ choices to withdraw from their rural housing land depends on a rational assessment of the difference between WRHL compensation and the expected income that could be derived from holding rural housing. Therefore, this study expands the application field of Homo economicus theory and strengthens its explanatory power in terms of WRHL reform. Fourth, by examining WRHL policy target groups, this study yields a surprising result: WRHL policy target groups are unwilling to withdraw from rural housing land because they prefer to hold rural housing land as value-added assets for future benefits, which is essentially land speculation of the policy target groups. These findings are also applicable to rural residential land consolidation, land transfer, and land acquisition in China and other regions worldwide, and can provide reference for the improvement and optimization of relevant land policies.

The remainder of this paper is organised as follows. Section “History of China’s withdrawal from rural housing land(WRHL) reform” introduces the history of China’s WRHL reform. Section “Theoretical framework and research hypotheses” introduces the research hypotheses and details the theoretical framework linking farmland dependence and urban housing to villagers’ choices to withdraw from rural housing land. Section “Data and methods” outlines the econometric model, the data, and the selected research variables. Section “Empirical results” presents the empirical results, and Section “Discussion” discusses the main findings in detail. Finally, Section “Conclusions and policy implications” provides the conclusions and relevant policy implications.

History of China’s withdrawal from rural housing land (WRHL) reform

WRHL policy proposal stage (1949–2014)

In the early days of the founding of the People’s Republic of China, the central government implemented a land reform movement to transfer the ownership and use rights of land, including rural housing land, from the feudal landlords to villagers (Zhou et al. 2020). After the reform movement, China’s villagers have the ownership and use rights of rural housing land. In 1962, the central government announced that the ownership of rural housing land belonged to collectives, that farmers have the right to use the rural housing land and the ownership of any housing built on it, and that the sale of rural housing land was prohibited. In 1982, the government proposed that the rural housing land vacated by villagers after relocation should be recovered by the village collectives.

With the development of China’s economy and the concomitant improvement of villagers’ incomes, some villagers began to move to cities and towns, leaving their rural housing land idle. At the same time, some villagers illegally occupied land to build houses in order to improve their original living conditions. In response, the central government required in 1995 that if the area of villagers’ rural housing land exceeded the local standard, or if the rural housing land was idle or the housing built on it lay collapsed or demolished for more than two years without being restored, then the rural housing land should be partially returned to, or uniformly reclaimed by the collectives. In order to prevent housing land laying idle or being used wastefully, the central government required in 2000 that the rural housing land of villagers leaving rural areas for towns should be replaced in a timely fashion. In 2004, the Ministry of Land and Resources of China first encouraged villagers to vacate redundant rural housing land. In 2008, the central government further proposed that local governments should provide rewards or subsidies to villagers who voluntarily vacated their rural housing land. In 2009, to promote withdrawal from rural housing land more effectively, the Ministry of Land and Resources of China proposed a mechanism for gradual withdrawal from excess rural housing land, and in 2011, it called for new research on the relevant withdrawal and compensation mechanism. After further policy research, in 2014, the central government finally proposed that local governments carried out a WRHL pilot program to investigate the willingness of villagers who had settled in cities and towns to withdraw from their rural housing land.

During this period, rural housing land in China went through a transition from villagers owning and using rural housing land, to village collectives owning rural housing land ownership while villagers held the land use rights, then to village collectives owning rural housing land ownership while encouraging villagers to withdraw from rural housing land use rights, and finally to village collectives owning rural housing land ownership while encouraging those villagers who had settled in cities and towns to withdraw from rural housing land use rights. The specific policy content of this stage can be found as Supplementary Table S1 online.

WRHL pilot exploration stage (2015–2025)

In 2015, the central government of China officially announced that WRHL reform would be carried out in fifteen pilot counties (including cities and districts), marking the beginning of the stage of WRHL reform pilot exploration. After a year of receiving feedback on pilot reform experience, in 2016, China expanded the scope of WRHL pilot program to thirty-three regions. In 2019, China revised the Land Management Law according to the reform experience of the various pilot areas, clearly stipulating that rural villagers who settled in cities and towns were allowed to voluntarily withdraw from rural housing land. At the same time, the government expanded the new round of WRHL pilot to 108 regions, for the pilot period from 2020 to 2022. In 2024 and 2025, the government of China proposed to establish a method for rural villagers who had settled in cities and towns to withdraw from rural housing land voluntarily and with compensation.

During the pilot exploration stage, China’s central government aimed to provide reference for the ongoing improvement of WRHL system through continuous pilot programs. In such pilot exploration, the central government requires that villagers in the pilot areas should not be forced to withdraw, and their consent must be respected. At the same time, WRHL policy target groups are villagers who have settled in cities and towns, i.e., villagers who have low dependence on farmland and own urban housing. The specific policy content of this stage can be found as Supplementary Table S2 online.

Theoretical framework and research hypotheses

Impact of farmland dependence on villagers’ choices to withdraw from rural housing land

Studies to date have generally calculated farmland dependence based on the proportion of the farm labour population to the total labour population and the proportion of farm income to the total income (Duan et al. 2016; Li and Cai 2018; Liu and Zhang 2022). Some scholars have argued that it is unreasonable to use the proportion of farm labour population as a measure of farmland dependence, as Chinese rural households generally have part-time employment (Liu et al. 2018), making it difficult to scientifically distinguish between farm employment labour and non-farm employment labour (Ju et al. 2016). Meanwhile, the Ministry of Agriculture of China used the proportion of farm income to total income to distinguish between non-farm villagers and full-time villagers (Ministry of Agriculture of China 1997). More importantly, farmland is an important means of production and livelihood for villagers, and thus their dependence on farmland resources is directly reflected in their livelihood choices (Duan et al. 2016). Specifically, the proportion of farm income generated by relying on farmland to the total household income reflects the contribution of farmland to a household’s livelihood status (Li and Cai 2018). Based on the above, this study defines farmland dependence as the proportion of farm income to total household income. In addition, this study uses the proportion of farm labour population to the total labour population to represent the farmland dependence variable in the subsequent analysis of the robustness of the empirical results.

Different types of villagers make significantly different behavioural choices (Wang and Kang 2023; Xia et al. 2024), which may explain why villagers with varying levels of farmland dependence make different choices in regard to withdrawing from rural housing land. Rural housing land provides social security, asset income, and auxiliary farm production (Zhao et al. 2019; Su et al. 2022). Therefore, different proportions of farm income to the total household income lead to different functional preferences of villagers for rural housing land, resulting in different perceptions of housing land value (Liu et al. 2020). The theory of Homo economicus holds that individuals act and make decisions in full accordance with the principle of maximizing income (Levitt and List 2008), and this implies that villagers make a rational decision regarding whether to withdraw from rural housing land in exchange for compensation. Specifically, the choice to withdraw from rural housing land will depend on the difference between WRHL compensation and the expected income to be derived from holding rural housing land (Liu et al. 2020; Gao et al. 2023).

Villagers with relatively low farm incomes are less dependent on farmland. They usually have a relatively stable non-agricultural income, and their households are relatively wealthy (Xu et al. 2019; Khanum et al. 2024). This makes them less dependent on the social security and auxiliary farm production aspects of their rural housing land and more interested in the asset income function of that land (Zhao et al. 2019). If WRHL compensation meets their expectations, these villagers will be more likely to choose to withdraw from their rural housing land. Conversely, villagers with a high proportion of farm income are highly dependent on their farmland. They do not have a stable non-farm income, and their economic situation is relatively poor (Marino et al. 2021). For them, the social security and auxiliary farm production characteristics of rural housing land are more important (Peng 2013; Su et al. 2022). Under the unified compensation standards of WRHL pilot policy, WRHL compensation usually focuses on asset income and the building value of rural housing land. The result is that the expected income from holding rural housing land is significantly greater than that from WRHL compensation. Therefore, a higher proportion of farm income implies that villagers are less likely to choose to withdraw from rural housing land (Fig. 1). We thus hypothesise as follows:

Fig. 1
figure 1

Mechanism by which farmland dependence affects villagers’ choices to withdraw.

Full size image

H1: When the proportion of farm income is lower, villagers are more likely to choose to withdraw from rural housing land.

Impact of urban housing on villagers’ choices to withdraw from rural housing land

The term “urban housing” encompasses the area devoted to urban housing and the number of urban properties. Some scholars have argued that defining urban housing by the number of urban properties accurately expresses urban housing’s functions as living space and a wealth asset (Wang and Gu 2016). Thus, most studies have used the number of urban properties to quantify urban housing (Wang and Gu 2016; Zhang and Fan 2020). This study accordingly defines urban housing as the number of urban properties owned by a household. In addition, this study uses the urban housing area to represent the urban housing variable in the subsequent analysis to test the robustness of the empirical results.

Because urban properties represent assets and wealth, villagers with urban housing are relatively wealthy, particularly when they own multiple urban properties (Sun et al. 2025). Although a few villagers may purchase urban housing through mortgage loans, villagers with multiple urban properties are usually wealthier than villagers without urban housing, as urban housing can be monetised, financed, leased, and used for other economic activities to generate additional income. Moreover, a few studies have verified that the total family income of villagers with urban housing is higher than that of villagers without urban housing (Li and Fan 2020; Guangcai et al. 2024). Thus, even if some urban housing is purchased using a mortgage, it can represent the villagers’ family property and wealth. Villagers who own multiple urban properties are less dependent on the social security and auxiliary farm production functions of their rural housing land and more focused on the asset income aspect. If WRHL compensation is greater than the expected income from holding housing land, they are more likely to choose to withdraw.

By contrast, villagers who do not own urban housing or who only own a small number of urban properties are less wealthy (Dai et al. 2020). They are more focused on the social security and auxiliary production aspects of their rural housing land. Rural housing land can be used to develop courtyard economies, such as raising poultry and growing vegetables, which can reduce their daily living costs. Therefore, villagers without urban housing are more concerned with the increase in living costs after WRHL, even with compensation (Huang and Chen 2022). Under the unified compensation standards of WRHL pilot policy, WRHL compensation usually focuses on the asset income and building value of rural housing land, meaning that WRHL compensation is less than the expected income of holding housing land. Consequently, these villagers are more likely to choose not to withdraw (Fig. 2). Hence, we formulate the second hypothesis as follows:

Fig. 2
figure 2

Mechanism by which urban housing affects villagers’ choices to withdraw.

Full size image

H2: When villagers own more urban property, they are more likely to choose to withdraw from rural housing land.

Interactive effect of farmland dependence and urban housing on villagers’ choices to withdraw

Hypothesis 1 posits that villagers with a low proportion of farm income are more inclined to engage in non-farm employment activities, are less dependent on farmland, and are in relatively wealthy families. Furthermore, in the context of China’s rapid urbanization and industrialization, most villagers with a low proportion of farm income are mainly engaged in non-farm employment in cities and towns, are more adaptable to urban work and lifestyle (Tang et al. 2020), and are more likely to purchase urban housing (Wang et al. 2020). In addition, an increase in the number of urban properties owned by villagers further reduces their proportion of farm income, because they can use urban housing for rent and investment, which in turn increases the likelihood that these villagers will purchase urban housing. Lastly, the impact mechanism by which urban housing affects villagers’ choices to withdraw from rural housing land positively impacts their choices to withdraw (Fig. 3).

Fig. 3
figure 3

Theoretical framework for the interactive effect.

Full size image

Conversely, villagers with a high proportion of farm income are highly dependent on their farmland. Their work primarily consists of farm production activities, and their family’s economic conditions are relatively poor. Their lives are strongly connected to the countryside. Consequently, villagers with a high proportion of farm income are less willing to buy urban properties. Their lack of urban housing further reduces their opportunities to engage in non-farm employment activities, thereby maintaining their farm production behaviour and negatively affecting their willingness to buy urban housing. Ultimately, the impact mechanism by which urban housing affects villagers’ choices to withdraw from rural housing land negatively impacts their choices to withdraw (Fig. 3). We thus hypothesise as follows:

H3: Farmland dependence influences the impact of urban housing on villagers’ choices to withdraw from rural housing land. i.e., farmland dependence moderates the relationship between urban housing and villagers’ choices to withdraw.

Data and methods

Study area and data

The data were obtained from a field survey of villagers in Chengdu conducted in August 2023. In 2016, the Chengdu government proposed that WRHL pilot areas be established in Wenjiang District, Pidu District, Qionglai City, Dujiangyan City, and Pengzhou City. Multistage sampling was used to ensure sample representativeness. First, Wenjiang District, Pidu District, Qionglai City, and Pengzhou City were selected based on their regional economic development (based on GDP per capita) and villagers’ income levels (Chengdu Bureau of Statistics 2023). Next, three sample towns in each district or each city were randomly selected based on their distance from the centre of the district or city, specifically from the suburbs, middle suburbs, and outer suburbs, for a total of 12 towns. Six villages in each town were randomly selected based on their distance from the town centre, specifically, two sample villages each were selected from the suburbs, middle suburbs, and outer suburbs. A total of 72 villages were included in the final sample (Fig. 4). Finally, based on the distance from the rural household to the village centre, 20–30 households in each village were randomly selected for questionnaire interviews. If the rural household selected for interview was not in the village and had moved to a city or town, we chose to conduct the interview by phone. It should be noted that the interviewed rural households were defined according to their hukou registration, an independent hukou was counted as a rural household. The field survey focused on household heads, who were the main income providers and decision-makers.

Fig. 4
figure 4

Location of the sample area.

Full size image

Control measures were implemented to avoid information bias and meet ethical standards (Plott and Zeiler 2005). First, a pre-survey was conducted to determine whether any questions needed revision. Second, all investigators were given training before the field survey to ensure that they provided respondents with accurate information regarding the questionnaire. Third, in the absence of local government personnel, the household heads were interviewed either face-to-face or by telephone. Finally, this study was performed in line with the principles of the Declaration of Helsinki, and ethical approval was granted by the researcher’s institution. Before the survey, every participant confirmed his informed consent regarding participation, data use, and research publication after being adequately informed of the research aims and assurance of anonymity. In addition, participants were assured that their participation was entirely voluntary, that the survey would be used solely for scientific research, and that all responses would be kept strictly confidential.

The questionnaire included information on the head of household, household economic status, rural housing land status, and villagers’ choices to withdraw from rural housing land (see Supplementary Questionnaire online). A total of 1515 valid questionnaires were collected, yielding an effective rate of 95.70% (Table 1).

Table 1 Sample area information.
Full size table

Variables and descriptive statistics

Variable definitions

The dependent variable was villagers’ choices to withdraw from rural housing land, as assessed by the survey question, “Would you choose to withdraw from your rural housing land?” The two independent variables were farmland dependence and urban housing. According to the theoretical framework analysis presented above, farmland dependence was calculated as the proportion of farm income to the total household income, while urban housing was calculated as the total number of urban properties owned by a household.

Ten control variables were used, divided into two categories: household characteristics and rural housing land characteristics. Household characteristics included the age, gender, education level of the household head, number of family members, and total family income (Zhang and Fan 2020; Liang et al. 2022; Xie et al. 2023). Similarly, villagers’ choices to withdraw from housing land were determined by the rural housing land characteristics, namely the floor area of the rural housing land, housing building structure, certification of housing land right, and location of the rural housing land (Shi et al. 2022; Gao et al. 2023; Xia et al. 2024). The driving time from the rural housing land to the nearest county centre and whether the rural housing land was near the main road were used to represent the rural housing land location (Table 2).

Table 2 Definitions of variables and descriptive statistics.
Full size table

Reliability and validity of the questionnaire

Cronbach’s alpha was used to assess the reliability of the questionnaire (Cronbach 1951). The higher the Cronbach’s alpha coefficient, the higher the reliability of the questionnaire (Bland and Altman 1997). The results showed that the Cronbach’s alpha coefficient of the questionnaire is 0.744, indicating satisfactory internal reliability. For the validity of the questionnaire, content validity was used because the questionnaire was not a complete Likert scale questionnaire (see Supplementary Questionnaire online) (Sireci 1998; Almanasreh et al. 2019). Content validity was ensured by the process of questionnaire development, and the questions about the research variables in the questionnaire referred to existing relevant studies (Li and Cai 2018; Gao et al. 2023; Xie et al. 2023). Meanwhile, a pre-survey was used to identify questions in need of revision. Overall, these measures ensure that the questionnaire has satisfactory validity.

Descriptive statistics

Table 2 presented the data characteristics of the variables. Of the villagers in the sample area, 41.65% owned urban properties. Regarding villagers’ choices to withdraw from rural housing land, 30.63% chose to withdraw. The average proportion of villagers’ farm income to the total household income was 36.10%. The villagers’ average age was 53.92 years, and 77.82% of the sampled villagers were male. The highest proportion of villagers had only a junior high school education (39.74%). The average annual household income was 60,539.26 CNY, and the average household size was 3.77. The average rural housing land floor area was 336.58 m2; combined with the household size data, this indicates a per capita housing land area of 89.28 m2. Most rural housing consisted of either brick and concrete structures (58.88%) or masonry-timber structures (35.31%). A total of 91.68% of the villagers had rural housing land property rights certificates, and 37.29% of the rural housing land was near the main road. The average traffic time from the villagers’ houses to the nearest county was 0.42 h.

Model specification

Binary logit model

The dependent variable, villagers’ decision to withdraw, is a binary categorical variable: either they choose to withdraw, or they do not. As required by the binary logit model, the dependent variable is binary and the two categories are mutually exclusive, making this model very suitable for analysing the probability of individuals’ choices and their influencing factors (Horowitz and Savin 2001). Therefore, this study used the binary logit model to analyse the impact of farmland dependence and urban housing on villagers’ choices to withdraw from rural housing land.

The latent variable \({T}_{i}^{* }\) is used, which is related to the variable \({S}_{i}\) as follows:

$${T}_{i}^{* }={\alpha }_{i}{S}_{i}+{\mu }_{i}$$
(1)

In Eq. (1), \({T}_{i}^{* }\) is considered a continuous, unobservable latent variable representing the villagers’ choices to withdraw from rural housing land or not. \({S}_{i}\) denotes the explanatory variable vector of villager i, and \({\alpha }_{i}\) represents the parameter vector to be estimated. Lastly, \({\mu }_{i}\) denotes the disturbance term, and i represents the number of villagers, i.e., i = (1, 2, …, 1515).

The relationship between \({T}_{i}^{* }\) and \({T}_{i}\) is as follows:

$${T}_{i}=\left\{\begin{array}{c}1,\,{\,T}_{i}^{* } > 0\\ 0,\,{T}_{i}^{* }\le 0\end{array}\right.$$
(2)

Once \({S}_{i}\) is determined, and considering the two-point distribution probability of \({T}_{i}\), it can be expressed as:

$$\left\{\begin{array}{ll}P\left({T}_{i}=1\left|{S}_{i},{\alpha }_{i}\right.\right)=F\left({S}_{i},{\alpha }_{i}\right)\,\\ P\left({T}_{i}=0\left|{S}_{i},{\alpha }_{i}\right.\right)=1-F({S}_{i},{\alpha }_{i})\end{array}\right.$$
(3)

In Eq. (3), \({T}_{i}\) denotes the villagers’ choices to withdraw, and its value is either 0 or 1. \(F({S}_{i},{\alpha }_{i})\) represents the link function connecting the dependent and independent variables. \(F({S}_{i},{\alpha }_{i})\) denotes the cumulative distribution function of the logistic distribution, meaning that the expression for \({T}_{i}\) is:

$$P\left({T}_{i}=1\left|{S}_{i}\right.\right)=\cap \left({\alpha }_{i}{S}_{i}\right)=\frac{\exp \left({\alpha }_{i}{S}_{i}\right)}{1+\exp \left({\alpha }_{i}{S}_{i}\right)}=\cap \left({\alpha }_{i0}+{\alpha }_{i1}{S}_{i1}+{\alpha }_{i2}{S}_{i2}+{\alpha }_{{ij}}{S}_{{ij}}+{\alpha }_{{it}}{S}_{{it}}+{\mu }_{i}\right)$$
(4)

In Eq. (4), \(\cap\) denotes the cumulative distribution function of the logistic distribution. Here, \({T}_{i}\) represents villagers’ choices to withdraw from rural housing land, \({S}_{i1}\) denotes farmland dependence, \({S}_{i2}\) represents urban housing, \({S}_{{ij}}\) denotes the control variable for household characteristics (i.e., age, gender, education, household size, and household income), and \({S}_{{it}}\) represents the control variable for rural housing land characteristics (i.e., rural housing land area, building structure, certification of right, main road, and traffic time from housing land to the county). The definitions of the independent and control variables are listed in Table 2. In addition, \({\alpha }_{i0}\) is a constant, with the estimated parameters being \({a}_{i1}\), \({a}_{i2}\), \({a}_{{ij}}\), and \({\alpha }_{{it}}\), while j and t represent the numbers of specific variables, with j = (3, 4, 5, 6, 7) and t = (8, 9, 10, 11, 12). Finally, \({\mu }_{i}\) and i denote the same notation as in Eq. (1).

Mediation effect regression

To further analyse the impact mechanism by which farmland dependence and urban housing affect villagers’ withdrawal choices, the stepwise regression method and the binary logit model were combined to explore the mediation effect. The specific form is as follows:

$${T}_{i}={\alpha }_{i}{S}_{i}+{\mu }_{1}$$
(5)
$${M}_{i}={b}_{i}{S}_{i}+{\mu }_{2}$$
(6)
$${T}_{i}={a}_{i}^{{\prime} }{S}_{i}+{c}_{i}{M}_{i}+{\mu }_{3}$$
(7)

In Eqs. (5–7), \({T}_{i}\) is the withdrawal choice of villager \(i\), \({S}_{i}\) is the independent variable, and \({M}_{i}\) is the mediator variable. \({\mu }_{1}\), \({\mu }_{2}\), and \({\mu }_{3}\) represent the disturbance term. \({\alpha }_{i}\), \({b}_{i}\), \({c}_{i}\), and \({a}_{i}^{{\prime} }\) are the impact coefficients of the variables. If \({\alpha }_{i}\) is significant, it indicates that the independent variable has an impact on villagers’ withdrawal choice. If \({b}_{i}\) is significant, it means that the independent variable has an impact on the mediator variable. If \({a}_{i}^{{\prime} }\) and \({c}_{i}\) are significant, it shows that the independent variable affects villagers’ choice to withdraw through the mediator variable.

Empirical results

Hypothesis test

Using Eq. (4), the binary logit model estimated the impact of farmland dependence and urban housing on villagers’ choices to withdraw from rural housing land. Model (1) in Table 3 illustrated the impact of farmland dependence and urban housing on villagers’ choices to withdraw for the entire sample. As shown in Model (1), without the control variables included, the effect of farmland dependence was significant at the 5% level, and the coefficient was positive, while the significance coefficient for urban housing was negative and at the 1% level.

Table 3 Impact of farmland dependence and urban housing on villagers’ withdrawal choices.
Full size table

Model (2) in Table 3 showed the regression results after addition of the control variables. The effect of farmland dependence was significant at the 5% level with a positive coefficient. In other words, when all other conditions were controlled, villagers with a lower proportion of income from farm production were less likely to choose to withdraw from rural housing land. From the perspective of marginal effect, if the farmland dependence variable increased by one unit, the likelihood of villagers choosing to withdraw increased by 0.108 units. This is contrary to Hypothesis 1 and the results of prior studies (Xia et al. 2024), possibly because villagers with a lower proportion of farm production income usually have better economic conditions and stronger information acquisition abilities, that can help them better understand the future WRHL policy changes. In this case, their expectations for the future value of rural housing land are increasing; considering the current WRHL compensation, they are more willing to hold rural housing land as value-added assets for a long time. The reasons for this are further analysed in the discussion section.

The coefficient of urban housing was negative and significant at the 10% level. i.e., villagers who owned more urban housing were less likely to choose to withdraw. From the perspective of marginal effect, if the urban housing variable increased by one unit, the possibility of villagers choosing to withdraw decreased by 0.017 units, contrary to Hypothesis 2 and the results of prior studies (Zhang and Fan 2020). This may be because villagers with more urban housing are relatively wealthy and more sensitive to information about future rural housing land reform policies. Therefore, they perceive a higher expected value of rural housing land and are more inclined to hold their housing land for future appreciation. The specific reasons are analysed in detail in the discussion section.

As the data were obtained from different villages, it is crucial to consider the possible impact of village heterogeneity on villagers’ choices to withdraw from rural housing land. Model (3) in Table 3 illustrated that farmland dependence significantly positively impacted villagers’ choices to withdraw at the 5% level. Conversely, urban housing negatively affected villagers’ choices at the 5% level. From the perspective of marginal effect, when village heterogeneity was controlled, the impact of the farmland dependence and urban housing on villagers’ choices was consistent with the marginal effect results of Models (1) and (2). In addition, if Rho remains under 0.059, that it taken to mean that the model estimate does not need to consider multiple levels (Gibbons and Hedeker 1994). Rho was 0.0107, indicating that village-level variance accounted for only a small proportion of the total variance in the sample data. In other words, although the data were taken from different villages, village heterogeneity had little impact on the villagers’ choices to withdraw, consistent with the results of Gao et al. (2017).

The other control variables in Table 3 appeared to significantly impact villagers’ choices to withdraw from rural housing land. The villagers’ age affected their choices to withdraw, in that older villagers were more likely to withdraw from rural housing land. Moreover, villagers were more reluctant to withdraw if their rural housing land was closer to the main road. Finally, longer traffic time from the rural housing land to the nearest town was linked to an increased likelihood of withdrawal.

PSM analysis

It should be noted that in addition to farmland dependence and urban housing, the unobserved characteristics of villagers and other environmental factors may also affect villagers’ choices to withdraw. Therefore, to solve the endogeneity problems caused by the self-selection of samples and omitted explanatory variables, the propensity score matching model (PSM) was used to further improve our empirical analysis. The algorithms of nearest neighbour matching (1:1), radius matching (caliper = 0.01) and kernel-based matching (bandwidth = 0.06) were used to match samples between the treatment and control groups.

Table 4 showed the average treatment effects of the three different matching algorithms and the standard errors. The results of the average treatment effect for the treated (ATT) showed that the coefficients of farmland dependence and urban housing were all significant. After controlling for problems of endogeneity, farmland dependence and urban housing were still found to affect villagers’ withdrawal choices, consistent with the results of the previous analysis.

Table 4 PSM analysis results.
Full size table

Robustness analysis

To evaluate the robustness of the results, the binary probit model was used to analyse the impact of farmland dependence and urban housing. Model (1) in Table 5 showed that farmland dependence significantly impacted villagers’ choices to withdraw. Conversely, urban housing had a significantly negative effect on villagers’ choices. Based on the regional economic development levels (measured as GDP per capita) and villagers’ income levels, Wenjiang District and Pidu District were classified as regions with better economic development (observations = 754), while Pengzhou City and Qionglai City were categorised as regions with general economic development (observations = 761). Based on these two samples, a binary logit model was used to analyse the impact of farmland dependence and urban housing on villagers’ choices to withdraw. Models (2) and (3) showed that the farmland dependence coefficient was significantly positive, while urban housing had a significant negative impact on villagers’ choices.

Table 5 Robustness test.
Full size table

In addition, 50% of the total samples (observations = 758) were randomly selected for further analysis. The urban housing variable represents the total building area of urban housing owned by the villagers. To avoid the possibility that the definition of the farmland dependence variable could be misleading (i.e., the proportion of farm income to total household income), the farmland dependence variable in this analysis was taken to represent the proportion of farm labour population to the total household labour population. The results of Models (4), (5), and (6) were consistent with those of previous studies, further proving the robustness of the results.

Interactive effect test

To examine the effect of the interaction of farmland dependence and urban housing on villagers’ choices to withdraw from rural housing land, the interaction variable for farmland dependence × urban housing was created. The regression results in Table 6 showed that the interaction term coefficient was significant at the 5% level, indicating an interactive effect of farmland dependence and urban housing on villagers’ choices to withdraw. From the marginal effect of Model (2), with village heterogeneity controlled, an increase of one unit in the interaction term variable increased the likelihood of villagers choosing to withdraw by 0.092 units. In other words, farmland dependence moderated the marginal effect of urban housing on villagers’ choices (Fig. 5). These results suggested that when farm income constituted a lower proportion of total income, the negative impact of urban housing on villagers’ choices to withdraw was stronger; i.e., villagers who relied less on farmland and had more urban properties were less likely to choose to withdraw from rural housing land. In addition, Rho was 0.0104, indicating that village heterogeneity had little impact on villagers’ withdrawal choices. The results of the interactive effects test indicated that the target villagers of WRHL policy were unwilling to choose to withdraw from rural housing land.

Table 6 Interactive effect of farmland dependence and urban housing on villagers’ choices to withdraw.
Full size table
Fig. 5
figure 5

Farmland dependence moderates the marginal effect of urban housing on villagers’ choices.

Full size image

Further analysis: Mediating role of expected value

The previous theoretical analysis indicated that the impact of farmland dependence and urban housing on villagers’ withdrawal choices both depended on a comparison between the expected income of rural housing land and the current compensation for WRHL. To further explore the specific influence mechanisms by which farmland dependence and urban housing affect villagers’ choices and to reveal the reasons why villagers with lower dependence on farmland and more urban housing are unwilling to withdraw, the villagers’ expected value of rural housing land was chosen as the mediating variable for analysis. The expected value item in the survey questionnaire was, “Do you think the current compensation standard for WRHL meets your expected value?”, with answers coded as 1= Yes, 0= No. The research results are shown in Table 7.

Table 7 Mediation effect regression and Bootstrap test results.
Full size table

As for farmland dependence, the results of the first stage model indicated that farmland dependence had a positive impact on villagers’ choices to withdraw from their rural housing land. The results of the second stage model showed that the impact of farmland dependence on expected value was positive and significant at the 5% level, i.e., the higher the villagers’ dependence on farmland, the more the compensation standard for WRHL conformed to their expected value. Conversely, the lower the villagers’ dependence on farmland, the higher the expected value they placed on their rural housing land. The results of the model in the third stage showed that, after controlling the expected value variable, the influence coefficient of farmland dependence on villagers’ choices to withdraw from rural housing land was positive and significant at the level of 10%. Meanwhile, the results of the third stage model also showed that the expected value had a positive and significant impact on the villagers’ choices after controlling for the variable of farmland dependence, indicating that the more the compensation standard for WRHL met the villagers’ expected value, the more likely they were to choose to withdraw.

With regard to urban housing, the results of the first stage model indicated that urban housing negatively affected villagers’ choices to withdraw from rural housing land. The results of the second stage model showed that the impact of urban housing on expected value was negative and significant at the 1% level, and thus the more urban housing a villager owned, the higher the expected value they placed on their rural housing land. In other words, the less urban housing owned by a villager, the more the compensation standard for WRHL conformed to their expected value. The results of the third stage model showed that, after controlling for the expected value variable, the impact coefficient of urban housing on villagers’ choices was negative and significant at the 5% level. The results of the third stage model also indicated that the expected value had a positive and significant impact on villagers’ choices to withdraw from their rural housing land after the variable of urban housing was controlled.

In addition, the expected value accounted for 0.1800 in the impact of farmland dependence on villagers’ withdrawal choices, and 0.2030 in the influence of urban housing on villagers’ withdrawal choices. The results of Bootstrap test showed that the 95% confidence interval did not contain “0”, indicating that the model estimation results are credible. The results indicated that farmland dependence and urban housing affected villagers’ withdrawal choices by influencing the expected value they placed on their land; specifically, expected value played a mediating role in the impact of farmland dependence and urban housing on villagers’ choices to withdraw from rural housing land. This also showed that the reason why the villagers with lower dependence on farmland and more urban housing were unwilling to withdraw from rural housing land was that they placed a higher expected value on their rural housing land. Compared with the current compensation for WRHL, they believed that holding on to their rural housing land could produce higher value returns in the future.

Discussion

The results of this study enhance our understanding of villagers’ choices to withdraw from rural housing land, and they are also applicable to other issues of rural residential land consolidation, land transfer, and land acquisition. The main findings of this study are as follows.

First, the research results indicated that lower dependence on farmland was linked to a lower likelihood of villagers choosing to withdraw from rural housing land, contrary to both the research hypothesis and the results of prior studies (Xia et al. 2024). The villagers’ lower dependence on farmland was related to a higher proportion of non-farm income. Compared to villagers who were more dependent on their farmland for income, less dependent villagers tended to be wealthier (Xu et al. 2019). Therefore, as these villagers had external income sources and were relatively rich, they were not interested in WRHL compensation and preferred instead to keep their rural housing land as long-term assets (Gao et al. 2023). Furthermore, because of their higher proportion of non-farm income and long-term non-farm employment, these villagers were able to obtain better information and were thus more sensitive to policy development trends (In the context of rural housing land system reform in China, this means they were more aware that villagers’ housing land rights will be strengthened in the future). Thus, they had higher expectations regarding the future value of their rural housing land. Conversely, villagers with higher dependence on farm income tended to have worse living conditions and economic status. They were less sensitive to the development of WRHL policies (Liu et al. 2020). Therefore, these villagers were more likely to choose to withdraw from rural housing land to obtain WRHL compensation and improve their living conditions. The field survey observed that most villagers who chose to withdraw had been engaged in farm production for a long time, came from relatively poor economic situations, had bad living conditions (such as inconvenient traffic or a lack of running water and natural gas), and lived in relatively dilapidated houses. Moreover, the empirical results of this study indicated that a higher education level of the household head and higher total household income were both linked to an unwillingness to withdraw. Although these two results were not statistically significant, they still demonstrated that better economic status was related to more reluctance to withdraw from rural housing land, which indirectly supported the claim of an effect of farmland dependence on villagers’ choices to withdraw.

Second, this study showed that villagers who owned a greater number of urban properties were less likely to withdraw from rural housing land, contrary to the second hypothesis and to the results of existing studies (Zhang and Fan, 2020). Urban properties represent assets and wealth (Dai et al. 2020), and thus villagers who own urban housing are relatively more affluent, especially those with multiple urban properties (Liao and Zhang, 2021). Therefore, these villagers were not sensitive to the wealth increment brought by WRHL compensation and were more likely to hold onto rural housing land to wait for future value appreciation. For example, rural housing land can be rented and transferred. Furthermore, in the context of China’s rural land system reform, rural construction land can be traded directly in the land market. Because rural housing land can be classified as rural construction land, villagers have high-value expectations for their housing land and tend to focus on potential long-term property income. Meanwhile, wealthy villagers also had a better understanding of policy information than poorer villagers and thus tended to have higher value expectations for their housing land, making them more reluctant to withdraw. For example, the government may further strengthen villagers’ rural housing land rights. In the field survey, we observed that villagers with no urban property chose to withdraw mainly because they were relatively poor, and thus they could use WRHL policy compensation immediately to improve their family’s economic situation and living environment. In addition, of the control variables in this study, the impact of household income on villagers’ choices to withdraw was notably negative. Although this result was not significant, it indicated that when household economic income was higher, the villagers were less willing to choose to withdraw, indirectly demonstrating the impact of urban housing on villagers’ choices to withdraw from rural housing land.

Third, although villagers who were less dependent on farmland and owned multiple urban properties are considered to be ideal WRHL policy target groups, the empirical results for the interactive effect indicated that these villagers were less likely to choose to withdraw from rural housing land. In other words, our research results were contrary to policy expectations. Based on the previous analysis, the reasons for this lie in the villagers’ expectations of the future value of their rural housing land and their desire to hold onto that land in expectation of higher returns. Therefore, WRHL reform policymakers need to stabilise the value expectations of villagers who hold idle rural housing land and provide diverse forms of compensation other than monetary compensation. In addition, the interactive results showed that farmland dependence moderated the effect of urban housing on villagers’ choices. When the proportion of farm income was lower, the negative impact of urban housing on villagers’ choices to withdraw was stronger. A lower proportion of farm income was also associated with a better family economic situation (Khanum et al. 2024). Hence, this finding suggested that villagers’ choices to withdraw from rural housing land were affected by their economic conditions. Thus, if the government blindly encourages villagers to withdraw without considering their economic situation, the desired policy effect may not be achieved.

Fourth, although some studies have examined the impact of farmland dependence and urban housing on villagers’ choices to withdraw from rural housing land (Xia et al. 2024; Zhang and Fan 2020), these studies have not revealed the specific impact mechanism by which farmland dependence and urban housing affect villagers’ choices. This study investigated the effect of the interaction of farmland dependence and urban housing on villagers’ choices to withdraw and the mediating effect of expected value, clearly answering the question of how farmland dependence and urban housing affect villagers’ withdrawal choices and what their mechanisms and mediating factors are, providing a powerful supplement to existing research and more policy application value. In addition, this study’s analysis of the impact of farmland dependence and urban housing on villagers’ withdrawal choices is based on the Homo economicus theory. Specifically, villagers’ choices to withdraw from their rural housing land depends on the difference between WRHL compensation and the expected income of holding the rural housing land. Although our results indicated that the target villagers were unwilling to withdraw from their rural housing land because they were engaged in a form of land speculation, this behaviour reflects a rational choice made by the target villagers in the current WRHL policy environment to maximise their own interests. This study has thus increased the applicability and explanatory power of the Homo economicus theory in the context of WRHL reform and expanded the field of application of the theory.

Fifth, the results of this study are also applicable to other land policies such as rural residential land consolidation, land acquisition, and land transfer in other regions of the world, both in developed and developing countries. In the implementation of such land policies in these regions, the group targeted by the policies constitutes an important stakeholder, which can determine whether the expected policy objectives can actually be achieved (Mbatha and Antrobus 2012; Podhrázská et al. 2015; Hansson et al. 2021). The question of how to encourage target groups to participate in the implementation of these land policies and prevent them from engaging in land speculation is a matter of great concern to policymakers. Although the characteristics of the target groups of rural residential land consolidation, land acquisition, land transfer, and other land policies in other regions of the world may not be exactly the same as those of China’s WRHL, the differences between the policy compensation and the target groups’ perceptions of expected value, as well as land speculation by the target groups, can affect policy implementation. Therefore, the results of this study and their policy implications can provide references and new insights for the improvement and optimization of relevant land policies in other parts of the world.

Conclusions and policy implications

Based on survey data from villagers in Chengdu, China, this study investigated the impact of farmland dependence and urban housing on villagers’ choices to withdraw from rural housing land to determine whether WRHL policy target groups would choose to withdraw. The research results indicated that villagers who relied less on their farmland were less likely to withdraw, as were those who owned a greater number of urban properties. In addition, farmland dependence moderated the effect of urban housing on villagers’ choices, while expected value played a mediating role in the impact of farmland dependence and urban housing on villagers’ choices. These results showed that the villagers who were the targets of WRHL policy were unwilling to choose to withdraw from their rural housing land, contrary to policy expectations. The main reason for this behaviour is that villagers place an expected value on their rural housing land than WRHL compensation, and thus prefer to hold onto their land as value-added assets to obtain future benefits, as a form of land speculation.

This study not only fills the gap in the existing research on WRHL, but also effectively expands the application of Homo economicus theory and strengthens its explanatory power by application to WRHL policy. This study’s findings also have significant policy implications since the policy target villagers’ choices to withdraw are crucial to the success of WRHL policy. In addition, whether in developing countries like China or in developed countries, the decision by the policy target groups to participate in policies similar to rural residential land consolidation is a rational choice based on maximizing their own interests. Therefore, the policy implications of this study also provide important references and inspirations for the implementation of relevant land policies in other developed and developing countries around the world.

First, the results of this study indicated that villagers with lower dependence on farmland and who owned multiple urban properties were unwilling to withdraw from rural housing land. The main reason for this is that they have higher expectations for the future value of their rural housing land and thus are more willing to hold on to rural housing land for higher returns. Therefore, policymakers must stabilise the value expectations of villagers holding housing land to weaken their desire to engage in speculation. For example, taxes could be levied on long-term idle rural housing land to increase the cost to villagers who are less dependent on farmland and own multiple urban properties. The transfer price mechanism for rural housing land should also be improved to clarify the reasonable value of rural housing land.

Second, rural housing land has multiple functions, such as asset income, social security, and auxiliary production. Therefore, a single monetary compensation cannot effectively meet the diversified demands of the villagers targeted by WRHL policy. Therefore, to encourage target groups to choose to withdraw from rural housing land, it is important to provide other diversified compensation methods in addition to monetary compensation, including housing compensation, social security compensation, continuing education, and skill training.

Third, farmland dependence moderated the effect of urban housing on villagers’ choices, i.e., villagers’ choices to withdraw were naturally affected by their economic conditions. Consequently, the government cannot simply encourage villagers to withdraw and must consider their economic situation. For example, the government should carefully investigate villagers’ family structure, economic situation, and risk capacity. Furthermore, the government should also encourage villagers who rely less on farmland and have more urban properties to transfer or rent out their rural housing land while ensuring information openness and transparency. Specifically, the government should build an information service platform for the transfer of idle rural housing land, standardise market transaction procedures, and encourage other social groups to participate in the utilisation of rural housing land.

This study has some limitations that point to directions for future research. First, this study did not distinguish whether the housing land was situated in a remote rural area or on the rural-urban fringe, nor did it distinguish whether the urban housing owned by villagers belonged to mortgage purchases. Taking these factors into consideration may make the research results more accurate. Furthermore, this study will make more sense if the threshold values of the proportion of farm income to total household income correspond to villagers’ different functional preferences for rural housing land. Future studies can further focus on the impact of differences in WRHL compensation methods on villagers’ withdrawal choices. Finally, as these results are based on the Chengdu pilot program, further research in other rural parts of China will provide more valuable evidence and insights.