How Migration Behavior Affects the Contracted Land Disposal Methods of Rural Migrants in China: An Analysis Based on the Perspective of Geographical Differences

Abstract

Migration leads to the separation of rural people from their contracted land and, thus, affects the human–land relationship, agricultural production, and food security in China’s rural areas. Using spatial autocorrelation analysis, GeoDetector analysis, and geographical weighted regression, the mechanism by which the migration behavior of the rural population impacts their contracted land disposal methods was analyzed in this study, and the spatial heterogeneity of this mechanism was further revealed from the perspective of geographical differentiation. The results of this study show that: (1) Chinese rural migrants exhibit pronounced geographical differences in the disposition of contracted land. The spatial distribution patterns of family operation, subleasing, and abandonment are different. (2) Migration behavior contributes to the differences in contracted land disposal methods across regions. Educational attainment, the number of cities migrated to, and the house purchase rate significantly contribute to the geographical differences among all three contracted land disposal methods. (3) The mechanisms by which migration behavior influences rural migrants’ land disposal decisions vary. The direction and degree of the influence of each factor on family operation are relatively consistent across regions. However, regarding the subleasing and abandonment of contracted land, the mechanisms by which migration behavior influences land disposal decision-making exhibit significant regional heterogeneity. The results of this study provide a useful reference for many countries to solve the problems of the human–land relationship and rural development.

1. Introduction

After China’s reform and opening up, when the Household Contract Responsibility System was implemented, surplus rural laborers were liberated and began to enter urban areas for work [1,2,3,4,5]. By the end of 2021, the total number of rural workers in China had reached 293 million [6]. Rural migrants, such as migrant workers and their accompanying relatives, have become the main component of China’s mobile population and one of the important productivity factors driving socio-economic development [7].

The essence of rural migration is the flow of human resources from rural to urban areas and from the primary sector to non-agricultural industries [8]. This process supports urbanization and industrialization in the areas that are migrated to and affects the human–land relationship in the areas that are migrated from [9]. Since land is immobile and cannot move across regions with the rural population, the migration of the rural population causes the human–land relationship in rural areas to exhibit the significant characteristics of leaving farming but not contracted land, moving to a city but not returning to contracted land, and quitting farming but not transferring contracted land [10,11]. Although the state has made efforts to promote various forms of contracted land transfer activities in recent years, due to the influence of terrain, capital, and market factors, new forms of agricultural operations have not been realized on a large scale [12,13]. The traditional smallholder production model still occupies the mainstream [14,15,16]. Against the background of declining agricultural comparative incomes, the phenomenon of “the old generation being unable to cultivate the land and the new generation being unwilling to cultivate the land” has gradually come to the fore, resulting in the rough management of contracted land and even the phenomenon of abandonment from time to time [17,18,19,20,21].

The Chinese government attaches great importance to the issue of contracted land management in the context of rural–urban migration, considering it an important issue in ensuring food security and promoting rural revitalization, and has emphasized its importance in several government documents [22,23,24]. Academics also conducted studies on this issue and concluded that contracted land disposal results from a combination of factors, including individual characteristics [25], resource endowment [26], livelihood differentiation [27,28,29], and social integration [30,31], which affect the behavioral choices of rural migrants. As a direct manifestation of labor mobility, migration behavior is the primary factor influencing contracted land dispositions [32]. Factors, such as migration distance [21], migration time [33], migration scale [34], and willingness to citizenize, all influence rural migrants’ contracted land dispositions to varying degrees [35].

These studies are important for analyzing the intrinsic link between rural migration and contracted land disposal. However, there are still some points to be added. First, most of the existing studies are conducted from a micro perspective, based on research data at specific locations, and their conclusions, being limited to specificity, may not reflect the perspective of the issue from the macro level. Second, the rural population’s migration behavior has regional characteristics [36]. Hence, the contracted land disposal behavior of rural migrants in different regions may differ, and the effect of migration behavior on land disposal decision-making may not be evenly distributed in space. Thus, this interaction needs to be analyzed from the perspective of geographical differences, which has been less touched on in previous studies.

Therefore, in this study, we relied on nationwide China Migrants Dynamic Survey (CMDS) data and used spatial autocorrelation analysis, GeoDetector analysis, and geographically weighted regression to analyze the mechanism by which the migration behavior of the rural population impacts their contracted land disposal methods, and we further revealed the spatial heterogeneity of this mechanism from the perspective of geographical differences in order to improve the previous research system and provide references for policy innovation in rural land system reform, rural revitalization, and food security.

2. Theoretical Analysis and Research Hypotheses

China is a public ownership country: the ownership of rural land belongs to all the rural populace. Under the Household Contract Responsibility System, rural land is allocated to farmers for cultivation with the family unit as the basis, known as “contracted land”. However, due to the generally low prices of agricultural products, rural populations increasingly seek employment elsewhere [2]. Within the context of a limited family labor force, various groups of rural migrants adopt different strategies to manage their contracted land. Under this background, this study aims to examine the relationship between rural population migration and contracted land disposal methods, based on the theoretical framework of behavioral economics (Figure 1), and puts forward corresponding research hypotheses.

2.1. Loss Aversion: The Impact of Human Resources Migration on the Contracted Land Disposal Methods

Resources are essential for securing behavioral outcomes, and the pursuit of maximizing personal or family interests is a fundamental principle when people make behavioral decisions [37]. Behavioral economics points out that individuals, driven by the goal of maximizing interests, need to develop suitable action plans based on resource constraints, on the one hand, while they also show significant aversion to potential losses, on the other hand, which results in corresponding loss avoidance behaviors [38]. Human capital embodies the primary resource of rural migrants [11]. Before migration, rural populations inevitably face the question of whether to invest their limited human capital in the agricultural or non-agricultural sectors, thus linking human capital to contracted land disposal methods [39]. Given the current substantial rural–urban income gap and low agricultural returns, farming often signifies a loss of opportunity benefits [40]. Consequently, rural migrants with varying human capital make different behavioral choices based on their degree of loss aversion.

For rural migrants who are older or lack job skills, the loss of opportunity benefits caused by farming is less significant due to the lack of job opportunities and the decrease in employment income, which makes them more likely to generate family operation behavior regarding land disposal. Conversely, younger or more educated rural migrants are more likely to secure higher economic incomes through external employment [41], amplifying the loss of the opportunity benefits generated by farming. To avoid losses, they are more inclined to generate land subleasing or abandonment behaviors. The resulting differentiated contracted land disposal methods reflect these varying levels of loss aversion.

Furthermore, the marital status and gender structure of migrants may also influence the family’s agricultural production decisions. In China’s traditional household division of labor, of male-as-breadwinner and female-as-housewife, women’s migration behavior is often subordinate [42,43]. Thus, whether a migrant is married and whether a female laborer is migrating often dictate whether a family is completely disengaged from agriculture, subsequently affecting the contracted land disposal method [44]. Based on the above analysis, this study proposes the following hypothesis:

Hypothesis 1:

Factors related to human resources migration, such as age and education attainment, significantly affect the contracted land disposal methods of rural migrants.

2.2. Status Quo Bias: The Impact of Migration Characteristics on the Contracted Land Disposal Methods

Behavioral economics suggests that individuals are influenced by status quo bias during decision-making. In the case of limited information, individuals tend to overestimate the costs and risks of unfamiliar behaviors in the decision-making process, which in turn prompts them to choose more familiar options, exhibiting a tendency to maintain their current state unchanged [45]. This cognitive bias also manifests during rural population migration. Rural migrants with a short migration time or lack of family migration experience still have a certain sense of discomfort regarding the change in working mode and living environment [46], which makes it difficult for them to give up the control of contracted land in the short term and makes it easier to keep the contracted land in the family. However, with the increase in migration time and number of cities migrated to, the discomfort of migration gradually diminishes [46]. Under the influence of a status quo bias, “maintaining migration” occupies a greater weight in the behavioral decision-making of rural migrants, reduces their dependence on rural contracted land, thus, more easily leads to land disposal methods, such as subleasing or abandonment.

Additionally, the distance of migration also has an impact on the contracted land disposal methods of rural migrants. In the case of short-distance migration within a province, rural migrants can continue to provide labor support for family agricultural production during busy farming seasons with the advantage of lower migration costs [47], making it easier to develop a land disposal method of family operation. In contrast, cross-provincial migration takes laborers away from their rural contracted land, significantly increasing the time and transportation costs of part-time behavior [48]. The data suggest that the farther the laborers migrate, the higher the likelihood they earn more off-farm income, resulting in a declining share of agricultural income in the household income structure and highlighting the comparative loss due to part-time production [47]. In this situation, cross-provincial migrants are more likely to reduce their agricultural input time to achieve continuity of work, resulting in the land disposal methods of subleasing or abandonment. Thus, it can be seen that under the influence of a status quo bias, there is a close relationship between the migration characteristics of rural migrants and their contracted land disposal methods. Based on the above analysis, this study proposes the following hypothesis:

Hypothesis 2:

Factors reflecting migration characteristics, such as migration time and the number of cities migrated to, significantly affects the contracted land disposal methods of rural migrants.

2.3. Endowment Effect: The Relationship between Integration Willingness and Contracted Land Disposal Methods

For rural people, contracted land is not just a means of production but also a personalized property that reveals their identity as farmers [49]. Behavioral economics points out that the endowment effect is one of the important characteristics of personalized property, i.e., individuals attach a subjective value to their own goods, so their value judgment of the goods is significantly higher than the actual market value [50,51]. The widespread “attachment to land” complex of the rural population is the result of the prominent endowment effect.

However, migration behavior causes changes in the working style and living environment, so the identity of rural migrants undergoes a transformation process from the place of departure to the place of immigration and from rural to urban [52]. In this process, the contracted land is an essential link between the migrant and the traditional rural society. The two have not only economic ties generated by factors, such as livelihood security or expected income, but also emotional ties generated by factors, such as kinship and countryside memories [53,54]. Therefore, the endowment effect of the contracted land is particularly obvious when the integration degree of the relocated place is low, as rural migrants tend to be more willing to retain control of the contracted land, so they can return to their hometown and restore their original rural identity when necessary [55]. However, when rural migrants settle or purchase houses in the place they have moved to, showing a higher intention of integration, their willingness to maintain the relationship with the rural contracted land is reduced, demonstrating the stripping from the original social relationship and rural identity, and the endowment effect of contracted land also decreases accordingly, which makes them more likely to choose to sublease or abandon their contracted land [56]. Thus, different integration intentions lead to different endowment effects of contracted land, which further affect the land disposal methods of rural migrants. Based on the above analysis, this study proposes the following hypothesis:

Hypothesis 3:

Factors reflecting the integration willingness, such as house purchase and settlement intention, significantly affect the contracted land disposal methods of rural migrants.

2.4. Reference Dependence: The Influence Mechanisms of Various Factors May Exhibit Geographical Heterogeneity

Behavioral economics puts forward the concept of reference dependence, which means that individuals set a reference point during decision-making and compare the expected behavioral outcomes against it to assess the effectiveness of their actions [45,57]. Due to environmental differences, different individuals may set varying reference points. Consequently, the same factors may have different impact effects on individual decision-making in different regional environments, as demonstrated in previous studies [58,59]. Presently, there are still considerable disparities in socio-economic development levels, industrial structures, agricultural conditions, and cultural traditions among different regions in China [17,18]. Rural migrants from different regions may establish different reference standards in the decision-making process of contracted land disposal. As a result, the impact mechanisms of various factors on contracted land disposal methods may not be consistent in the spatial distribution. Therefore, based on the idea of reference dependence, this study proposes the following hypothesis:

Hypothesis 4:

The impact mechanisms of migration behavior on the contracted land disposal methods of rural migrants exhibit geographical heterogeneity, with the same factors having different effects in different regions.

3. Research Methods, Data Sources, and Variable Construction

3.1. Research Methods

3.1.1. Spatial Autocorrelation Analysis

Spatial autocorrelation refers to the potential interdependence among elements in the same distribution area. Moran’s I is an index to measure it. When Moran’s I is significant, it indicates that the data are not uniformly distributed in space, and the results of the traditional econometric analysis may be biased and need to be analyzed using spatial econometric models. Its calculation formula is:

$$I=\frac{n}{S_0}\frac{{{\displaystyle \sum }}_{i=1}^n{{\displaystyle \sum }}_{j=1}^n{w}_{i,j}{z}_i{z}_j}{{{\displaystyle \sum }}_{i=1}^n{z}_i^2}$$

(1)

where I is the Moran index, n is the number of study areas, $z_i$ is the deviation of the attribute of region i from its mean, $w_{i,j}$ is the spatial weight between regions i and j, and $S_0$ is the aggregation of all spatial weights.

3.1.2. GeoDetector

GeoDetector uses the magnitude of the variance between different stratified data to determine differences in the spatial distribution of each variable, which in turn explains whether the independent variable causes the spatial distribution of the dependent variable. Since GeoDetector tests the correlation of variables in two-dimensional space, it has a more comprehensive explanatory range than the traditional one-dimensional linear model. It is calculated as follows:

$$q=1-\frac{{{\displaystyle \sum }}_{b=1}^n{N}_b{\sigma }_b^2}{n{\sigma }^2}$$

(2)

where q is the explanatory coefficient, n is the number of study areas, b is the data layer of the variable, $N_b$ is the number of cells in data layer b, N is the number of cells in the full domain, and ${\sigma }_b^2$ and ${\sigma }^2$ are the variance of the detected variable in data layer b and the full domain, respectively. Meanwhile, GeoDetector requires explanatory variables to be fixed-order data. This study stratifies each explanatory variable in this part with a 7-layer natural breaks classification method based on previous research experience [60].

3.1.3. Geographically Weighted Regression

The classical linear regression model is a global estimation of each variable. However, if the variables are spatial data, and there is spatial autocorrelation, the residual terms in the estimation process of the classical model are no longer independent of each other and may lead to biased estimation results [61]. In contrast, geographically weighted regression (GWR), based on the idea of local smoothness, constructs an independent equation for each cell in the study area and detects the degree of each variable’s role in each spatial location using local regression. The equation is

$$y_i=\beta \left(u_i,v_i\right)+{\displaystyle \sum }_{k=1}^n{\beta }_k\left(u_i,v_i\right)x_{ik}+{\epsilon }_i$$

(3)

where $y_i$ is the explained variable, $\beta \left(u_i,v_i\right)$ is the intercept, $x_{ik}$ is the kth explanatory variable of region i, $\left(u_i,v_i\right)$ is the spatial coordinate of region i, ${\beta }_k\left(u_i,v_i\right)$ is the intercept of the kth explanatory variable of region i, and ${\epsilon }_i$ is the random disturbance term.

3.2. Data Sources

The data used in this study were obtained from the 2017 CMDS conducted by the National Health Commission, which selected sample sites in 31 provinces (municipalities directly under the Central Government and autonomous regions) in mainland China. This survey focused on investigating the non-local household registered mobile population who had lived in the migrant destination for over a month and were over 15 years old. The survey collected 169,990 samples. After data screening, 71,494 rural household samples with contracted land were obtained for analysis in this study. The advantages of these data are that this survey covers mainland China and has a large sample size and that the information on contracted land disposal of the sample was accurately obtained in the survey, which makes it possible to analyze the contracted land disposal behavior of rural migrants and regional differences from a macroscopic perspective. The basic information of samples is shown in

Table 1. Descriptive statistical table for basic information of samples (N = 71,494).

Variable	Ratio	Variable	Ratio
Contracted land disposal methods		Gender
Family operation	77.97%	Male	57.35%
Subleasing	14.87%	Female	42.65%
Abandonment	7.16%	Cross-provincial migration	49.97%
Age (year)	Average = 37.90	Migration time (year)	Average = 12.09
15–25	12.09%	<3	11.20%
26–35	33.48%	3–5	13.37%
36–45	29.24%	6–10	23.58%
46–55	20.09%	>10	51.85%
56–65	3.93%	Number of cities migrated to	Average = 2.16
>65	1.18%	1	48.23%
Educational attainment	Average = 3.15	2	27.08%
Have not attended school = 1	3.45%	3	11.94%
Elementary school education = 2	18.71%	>3	12.75%
Middle school education = 3	49.19%	Parents’ migration experience	21.42%
High school/technical secondary school education = 4	19.18%	House purchase	25.08%
Junior college education = 5	6.63%	Willingness to transfer hukou	31.49%
College education = 6	2.70%	Willingness to stay and live	82.36%
Post-graduate education = 7	0.14%	Willingness to return to hometown	1.90%
Marriage	85.65%

.

3.3. Variable Construction

This study analyzes the impact of migration behavior on contracted land disposal methods from the perspective of geographical differences; therefore, each province in mainland China is used as the basic research unit, and the valid samples are grouped into each unit using the household registration place as the index. Among them, the effective sample sizes of Beijing, Tianjin, and Shanghai are 3, 27, and 1, respectively. Due to the small sample size, the above three regions are not included in the analysis to avoid statistical bias, and a total of 28 regions are included in the final study. In spatial econometric models, the explained variable is the contracted land disposal methods of rural migrants. Given the feasibility of the study and the sample distribution, the seven contracted land disposal methods listed in the questionnaire are grouped into three categories: family operation, subleasing, and abandonment. Meanwhile, based on the results of the previous theoretical analysis, 12 indicators related to migration behavior are selected as explanatory variables from three aspects: human resources, migration characteristics, and social integration intention (

Table 2. Variable settings and descriptions used in this study.

Variable Category	Variable Name	Variable Description
Contracted land disposal methods	Family operation rate	Three types of contracted land disposal, including farming by oneself, hiring someone to farm, and farming by relatives, are classified as family operation, and the ratio of this type of sample to the total samples in the region to which they belong is calculated.
	Subleasing rate	Three contracted land disposal methods, including subleasing to private individuals, subleasing to village collectives, and subleasing to enterprises, are classified as subleasing, and the ratio of this type of sample to the total samples in the region to which they belong is calculated.
	Abandonment rate	Ratio of the samples with abandoned contracted land to the total samples in the region to which they belong.
Human resources	Gender ratio	Ratio of the female samples to total samples in the region to which they belong.
	Age	Average age of the samples from various regions.
	Educational attainment	Average educational attainment of the sample in each region. Have not attended school = 1; elementary school education = 2; middle school education = 3; high school/technical secondary school education = 4; junior college education = 5; college education = 6; post-graduate education = 7.
	Marriage rate	Ratio of the samples with de facto marriage to the total samples in the region to which they belong.
Migration characteristics	Cross-provincial migration rate	Ratio of the cross-provincial migration samples to the total samples in the region to which they belong.
	Migration time	Average migration time of the samples of each region.
	Number of cities migrated to	Average number of cities migrated to for the samples of each region.
	Parents’ migration experience	Ratio of the samples withparents that have worked outside the hometown to the total samples in the region to which they belong.
Willingness to integrate	House purchase rate	Ratio of the samples that have purchased a house in the place of immigration to the total samples in the region to which they belong.
	Willingness to transfer hukou	Ratio of the samples willing to move their hukou to the place of immigration to the total samples in the region to which they belong.
	Willingness to stay and live	Ratio of the samples willing to stay and live in the place of immigration to the total samples in the region to which they belong.
	Willingness to return to hometown	Ratio of samples with the intention to return to their hometown to the total samples in the region to which they belong.

).

4. Result Analysis

4.1. Regional Differences in Contracted Land Disposal Methods of Rural Migrants

As shown in Figure 2, there are significant differences in the disposition of contracted land by rural migrants in different regions. Regarding family operation rates, more than 80% of rural migrants in Tibet, Yunnan, Guangxi, and Gansu in western China and Liaoning, Hebei, Shandong, Shanxi, and Henan near northern China retain contracted land for family farming. In contrast, the family operation rate in Ningxia is significantly lower than that in other regions of the country, with only 50% of rural migrant households still willing to continue to operate contracted land. Similarly, the family operation rate is low in Heilongjiang, Zhejiang, Jilin, Chongqing, and Jiangxi.

Regarding subleasing rates, rural migrants in Heilongjiang, Jilin, Xinjiang, and Inner Mongolia in northern China and in Jiangsu, Anhui, Zhejiang, and Jiangxi around the Yangtze River Delta have higher rates of subleasing contracted land. In contrast, subleasing rates in southwest China’s Chongqing, Yunnan, Guangxi, and Tibet and northwest China’s Shaanxi and Gansu are significantly lower than in other regions.

In terms of the abandonment rate, western provinces, such as Ningxia, Chongqing, Guizhou, Shaanxi, Sichuan, and Qinghai, as well as coastal provinces in southeastern China, such as Zhejiang, Hainan, Fujian, and Guangdong, have high rates of contracted land abandonment, while provinces in the northeastern and central regions have lower abandonment rates.

In general, rural migrants in China show obvious geographical differences in contracted land disposal. In the central provinces, the three northeastern provinces, and Xinjiang, where arable land resources are relatively abundant and farming conditions are relatively favorable, the production and asset functions of contracted land are better reflected. After the rural population migrates, their contracted land often continues to be used for agricultural production through family operating or subleasing. In contrast, due to the constraints of topography and climate, the problem of the fragmentation of arable land is relatively prominent in the western provinces, and the agricultural production conditions are relatively backward, resulting in an underdeveloped contracted land rental market. When rural migrants dispose of their contracted land, they generally have to choose to continue its cultivation by family members who are left behind or leave it abandoned. In addition, in the eastern provinces, the secondary and tertiary industries are more developed, non-agricultural jobs are more plentiful, the agricultural comparative income is relatively low, and farmers generally lack enthusiasm for agricultural production. Therefore, rural households prefer to invest their limited human resources in non-agricultural industries, resulting in a higher abandonment rate of contracted land in these areas.

4.2. Driving Role of Migration Behavior on Inter-Regional Differences in Contracted Land Disposal

Spatial autocorrelation among data is a sufficient condition for applying spatial econometric models. Therefore, this study used spatial autocorrelation analysis to explore the spatial autocorrelation of contracted land dispositions among regions. The results show that the Moran’s I indices of the family operation rate, subleasing rate, and abandonment rate in China are 0.580351, 0.467325, and 0.383576, respectively. All indices passed the significance test at the 0.01 level, indicating that the contracted land disposal methods of rural migrants in different regions have a certain autocorrelation in the spatial distribution, which should be further analyzed using spatial econometric models.

To eliminate the influence of unit differences and multicollinearity between variables on the model estimation results, the SPSS platform was used to standardize each variable with a Z-Score. A linear regression model was constructed with the three contracted land disposal methods as the dependent variables and the 12 explanatory variables mentioned above to test the multicollinearity among the variables. The results show that the variance inflation factor (VIF)values of the explanatory variables “gender ratio” and “age” were greater than five, and they were excluded from the rest of the empirical analysis due to the multicollinearity problem.

All of this shows the geographic differences in the contracted land disposal methods of rural migrants in China. To clarify whether migration behavior contributes to the differences, this study used the GeoDetector method to detect the driving relationships between the remaining 10 explanatory variables and the differences in contracted land disposal methods through factor detection and interaction detection. The results of the factor detection show that, within the significance threshold of 0.05, “educational attainment”,“number of cities migrated to”, and “house purchase rate” all significantly contribute to the geographical differences among the three types of contracted land disposal methods, and the intensity of the effects are all at a high level. In addition, the geographical differences between the rates of family operation and the abandonment of contracted land are also influenced by the “marriage rate”, “migration time”, “willingness to transfer hukou”, and “willingness to return to hometown”. In contrast, the “cross-provincial migration rate” and “willingness to stay and live” drive the geographical differences in the rates of contracted land subleasing. The “parents’ migration experience” does not significantly affect the differences in the geographic distribution of the contracted land disposal methods (

Table 3. Effect of migration behavior on inter-regional differences in contracted land disposal.

Contracted Land Disposal Methods	Family Operation Moran’s I = 0.580351		Subleasing Moran’s I = 0.467325		Abandonment Moran’s I = 0.383576
Factors	q Value	Significance	q Value	Significance	q Value	Significance
Educational attainment	0.682	0.003	0.592	0.015	0.586	0.012
Marriage rate	0.644	0.004	0.476	0.107	0.627	0.005
Cross-provincial migration rate	0.400	0.233	0.534	0.036	0.178	0.683
Migration time	0.580	0.017	0.542	0.068	0.605	0.010
Parents’ migration experience	0.205	0.586	0.219	0.530	0.158	0.694
Number of cities migrated to	0.758	0.000	0.504	0.035	0.747	0.000
House purchase rate	0.707	0.000	0.582	0.010	0.768	0.000
Willingness to transfer hukou	0.576	0.015	0.402	0.131	0.593	0.010
Willingness to stay and live	0.389	0.424	0.697	0.004	0.292	0.448
Willingness to return to hometown	0.564	0.031	0.321	0.276	0.657	0.005

). The interaction detection results show that all significant factors have a two-factor enhancement after pairwise interaction, i.e., the superposition of two significant factors explains the differences in contracted land dispositions better than any single factor.

Overall, the three contracted land disposal methods of rural migrants are all affected by factors belonging to human resources, migration characteristics, and integration willingness; however, since not all the factors can be significant in the model, Hypothesis 1, Hypothesis 2, and Hypothesis 3 can only be partially supported.

4.3. Influence Mechanisms and Regional Heterogeneity of Migration Behavior on Contracted Land Disposal Decisions

The geographically weighted regression model was further applied to investigate the influence mechanism of each significant factor on the contracted land disposal decisions of rural migrants and to analyze the spatial differences between the direction and intensity of each factor in different regions. The model fitting results show that the R² of the three models for family operation, subleasing, and abandonment are 0.648, 0.734, and 0.620, respectively. The adjusted R² are 0.533, 0.567, and 0.364, respectively, indicating that the factors have a good explanatory ability for the contracted land disposal behavior of rural migrants. The AICs of the three models are all at low levels, and the maximum conditions of all three models do not exceed 30, indicating that there is no local multicollinearity problem in the analysis process. The goodness of fit of each model is high, and the model analysis results are more reliable (

Table 4. GWR model estimation results.

RegressionCoefficients	Family Operation		Subleasing		Abandonment
Factors	Min	Max	Min	Max	Min	Max
Educational attainment	0.073	0.105	−0.712	0.385	−0.596	−0.163
Marriage rate	−0.180	−0.174	——	——	−0.282	0.322
Cross-provincial migration rate	——	——	−0.209	0.437	——	——
Migration time	−0.143	−0.139	——	——	0.012	0.450
Number of cities migrated to	−0.510	−0.502	−0.126	0.290	0.074	0.555
House purchase rate	−0.843	−0.823	0.335	0.900	0.270	0.653
Willingness to transfer hukou	0.029	0.043	——	——	−0.513	0.160
Willingness to stay and live	——	——	−0.202	0.388	——	——
Willingness to return to hometown	−0.143	−0.142	——	——	−0.248	0.026
Bandwidth	77.590		15.238		18.968
Residual sum of squares	10.564		7.972		11.408
Sigma	0.683		0.658		0.797
AIC	81.962		85.136		99.670
R2	0.648		0.734		0.620
Adjusted R2	0.533		0.567		0.364
Max number of conditions	4.200		8.698		8.550

). The results of the coefficient regression show that, except for family operation, the influence mechanisms of each factor on the two contracted land disposal methods of subleasing and abandonment show significant regional heterogeneity. The intensity and direction of the influence of each factor differ from each other in terms of the spatial distribution. This study used the ArcGIS platform to visualize the regression coefficients of the two models with a seven-layer natural breaks method, highlighting the role of each factor in the local geographic space.

4.3.1. Influence Mechanisms of Migration Behavior on Family Operation

For the family operation model, the minimum and maximum values of the regression coefficients of each factor are close to each other, and the direction of action is relatively consistent (Table 4). The results show that the influence mechanism of each factor on rural migrants’ contracted land disposal decisions, in terms of whether or not they choose to continue to operate the contracted land by the family, is highly homogeneous across the country with no regional differences. Hypothesis 4 is not supported in the family operation model.

The “house purchase rate” has the strongest effect on the family operation rate of contracted land, with the intensity of the effect around the 0.800 level. The effect direction is negative, in that a higher house purchase rate in the place of immigration is associated with a lower probability of keeping the contracted land under family operation. The “number of cities migrated to” is also an important factor influencing the decision of contracted land disposal, affecting the family operation rate of contracted land around the 0.500 level. The direction of the effect is also negative, in that the more cities rural migrants have moved to, the less likely they are willing to keep their contracted land. The “marriage rate”, “willingness to return to hometown”, and “migration time” all negatively influence the family operation rate of contracted land at the 0.100 level, indicating that rural migrants who stay away for a long time, have a low willingness to return to their hometowns, and are married are more willing to leave agricultural production. “Educational attainment” positively influences the family operation rate of contracted land, but the intensity of the influence is low, so it only has an auxiliary effect.

4.3.2. Influence Mechanisms of Migration Behavior on Subleasing

Regarding the subleasing model, the standardized residuals of the regression results in each region are within the reliability threshold of [−2.5, 2.5], indicating that the goodness of fit of each local regression equation was good during the operation of the model. The influence of various factors exhibits geographical heterogeneity, so Hypothesis 4 is supported in the subleasing model (Figure 3).

Among the influencing factors, the regression coefficients of “educational attainment” gradually increase from northeast to southwest in terms of the spatial distribution. The coefficients are positive in all regions, except Heilongjiang, Jilin, and Liaoning, suggesting that the more educated rural migrants are, the more inclined they are to exploit the asset function of contracted land and obtain rent by renting contracted land, thus putting more human resources into non-agricultural production. This influence mechanism is particularly strong in Xinjiang and Tibet but is weaker in Inner Mongolia and the eastern coastal provinces.

The regression coefficients of the “cross-provincial migration rate” increase from west to east, negatively affecting contracted land subleasing in Xinjiang and Tibet and positively affecting other regions. That is, the higher the rate of cross-provincial migration is, the higher the rate of rural migrants renting out contracted land is, and the strength of this effect mechanism is higher in Northeast China and is weaker in Yunnan, Guizhou, and Sichuan. Generally speaking, cross-provincial migration implies a relative increase in the migration distance, which increases the time and economic costs for migrants to take care of both work and agricultural production, prompting them to give up the direct operation of contracted land. Except for in Xinjiang and Tibet, the “number of cities migrated to” also positively affects contracted land subleasing behavior. The high values of the regression coefficient are mainly concentrated in Guangdong, Fujian, Zhejiang, Jiangxi, and other provinces in East and Central China, showing that the more cities rural migrants have lived in, the more they tend to transfer out contracted land in their hometowns.

The “house purchase rate” positively promotes the subleasing of contracted land nationwide, with a high degree of influence, and it gradually increases from west to east in terms of the spatial distribution. The higher the rate of rural migrants owning their own houses in the place of immigration is, the higher the probability of renting out contracted land is. The direction of the effect of the “willingness to stay and live” is not uniform, that is, it exhibits a positive contribution in eight provinces, including Xinjiang, Tibet, Qinghai, Yunnan, Guangxi, Guangdong, Fujian, and Hainan, while it negatively affects rural migrants’ contracted land subleasing behavior in the remaining regions.

4.3.3. Influence Mechanisms of Migration Behavior on Abandonment

There are also significant geographical differences in the influence mechanisms of the factors on the abandonment of contracted land, as shown in Figure 4. The standardized residuals are within the reliability threshold of [−2.5, 2.5] for all regions, except Ningxia. Hypothesis 4 is supported in the abandonment model.

“Educational attainment” has a negative effect on the abandonment of contracted land nationwide. Rural migrants with a higher education level exhibit lower odds of abandoning contracted land in their hometowns. This influence mechanism is stronger in Xinjiang, Inner Mongolia, Jilin, Liaoning, and North and East China but is weaker in the southwestern provinces. Generally speaking, with the improvement of education level, migrants are often better able to realize the harm caused by abandoning contracted land, and they are also better at activating various family assets to obtain higher economic benefits, thus avoiding the occurrence of idle contracted land. The direction of the effect of “marriage rate” is not uniform. The regression coefficient increases from southeast to northwest in the spatial distribution. In the provinces of East, Central, and South China, marriage reduces the chance of abandoning contracted land by rural migrants, while in other regions, it has a positive contribution.

Both the “migration time” and the “number of cities migrated to” positively influence contracted land abandonment behavior nationwide, showing that the longer rural migrants have been away and the more cities they have lived in, the higher their odds of abandoning contracted land. However, the influence mechanisms of the two factors are not consistent in the spatial distribution, with rural migrants in southeastern Chinese provinces, such as Zhejiang, Fujian, Guangdong, and Jiangxi, being more likely to be influenced by the duration of the migration. In comparison, the number of cities migrated to strongly affects rural migrants in northern provinces, such as Heilongjiang, Jilin, Liaoning, and Inner Mongolia. The “house purchase rate” also positively contributes to the abandonment of contracted land. The higher the rate of rural migrants owning houses in the place where they move to is, the higher the probability of their contracted land in their hometown remaining idle is. The high values of the regression coefficient are located in the southwestern provinces of China, such as Sichuan, Guizhou, and Yunnan, while the low values are dominated by the three northeastern provinces. In general, the time of migration and the number of cities migrated to represent the degree of rural migrants’ detachment from their past rural life. The purchase of a house in the place of immigration represents their willingness to integrate into the new living environment. The combination of the two is a process that pulls the identity of rural migrants from the place of departure to the place of immigration, thus reducing the status of contracted land in their psychological and economic activities.

The direction of the effect of the “willingness to transfer hukou” is not uniform. Its regression coefficient gradually increases from northwest to southeast. Rural migrants in provinces in East, Central, and South China, such as Zhejiang and Fujian, are more inclined to leave their contracted land idle after transferring their hukou to the place of immigration, while the opposite is true for migrants in western and northeastern provinces. The “willingness to return to hometown” has a negative effect on abandonment behavior, and the regression coefficient increases from west to east. In general, the abandonment of contracted land not only affects household income but also leads to land degradation, reducing agricultural productivity and economic output. Therefore, migrants’ willing to return to their hometowns tend not to leave their contracted land idle, to avoid the depreciation of their land assets.

5. Discussion

The large-scale and cross-regional migration of rural populations is a widespread social phenomenon in China today. Its essence is the reciprocal movement of labor forces between regions, which further leads to the reconstruction of rural land-use patterns. In this context, this study analyzes how migration behavior affects rural migrants’ contracted land disposal methods from the perspective of geographical differences. The study finds that there are significant regional differences in the contracted land disposal methods of rural migrants in China and that migration behavior contributes to these differences to some extent. In addition, migration behavior also influences rural migrants’ contracted land disposal decisions. However, the influence mechanisms differ from region to region.

The main contributions of this study are: First, the research perspective has been improved. Most previous studies were conducted from a micro perspective, taking a village or town as the study area and individuals as the study object, focusing on the interaction between individual or family characteristics and land-use patterns. This study takes a macro perspective and uses a large sample of data from 28 provinces in China to conduct a nationwide spatial analysis of the influencing mechanism between migration behavior and contracted land disposal methods, so the macro aspect of the issue can be reflected and can complement previous micro studies. Second, this study focuses on three types of contracted land disposal methods, including family operation, subleasing, and abandonment, instead of analyzing only one of them, as in previous studies. The theoretical analysis shows that different contracted land disposal methods are not isolated from each other but have some internal linkages. The behaviors of family operation, subleasing, and abandonment are the reflections of decreasing land endowment effects. This study incorporates various contracted land disposal methods into the analysis, thus extending the scope and content of previous studies. Finally, this study focuses not only on whether rural migrants’ contracted land disposal methods are influenced by migration behavior but also on the geospatial heterogeneity of this influence mechanism, which has been addressed less often in previous studies. The analysis of the spatial heterogeneity is helpful for carrying out targeted policy innovation in each region, for example, by determining which regions should increase education input to promote rural land transfer and which regions should carry out household registration system reform to restrain rural land abandonment.

In addition, the theoretical analysis framework, research methods, and findings of this study are relevant, not only for China but also for many countries in the world, to solve the problems of human–land relations or rural development. Many developing countries, such as Pakistan, India, and Ghana [62,63], are undergoing rapid urbanization, resulting in a massive loss of the rural population and a mismatch between human and land resources, thus leading to rural land lying idle and a reduction in the agricultural output, which pose a threat to food security, similar to the situation in China. This study analyzes the land use of rural migrants in different regions. Its results can help these countries more scientifically and rationally formulate urbanization policies according to different geographical locations and different population migration characteristics, to avoid the disorderly spread of urbanization and excessive encroachment on rural resources.

Moreover, developed countries are also facing the problem of rural decline. Sweden and South Korea have shown a negative population growth trend in recent years [64,65]. A shortage of human resources has led to a slow growth of agricultural output, a widening income gap between urban and rural areas, and a lack of rural development potential, resulting in a vicious circle of continuous rural decline, as the new rural population often leaves the countryside for urban development after receiving education. This study analyzes the relationship between rural migration and agricultural production methods, which can help these countries adjust their population migration policies and capital investment to conduct migration guidance and industrial layout in a more rational and orderly manner, thus promoting agricultural development and curbing the continuous decline of rural areas.

Finally, this study also has some shortcomings and problems to be solved. First, the contracted land disposal behavior of rural migrants may be influenced by various factors, including objective factors, such as family structure, economic conditions, geographic environment, and resource endowment, as well as subjective factors, such as the attitude, moral standards, occupational preferences, and development expectations of migrants and their groups. Limited by the subject, in this study, the mechanism by which migration behavior affects contracted land disposal methods was only analyzed from the aspects of human resources, migration characteristics, and integration willingness, without taking other factors into consideration. Second, in this study, it was found that the effect of migration behavior on land subleasing and abandonment is not uniform across regions. For example, educational attainment inhibits land subleasing in Heilongjiang, Jilin, and Liaoning, while it promotes subleasing in other regions. However, this study was not able to provide a detailed explanation for this phenomenon. The reason is that the human–land interaction system is a complex system involving many factors, such as the regional development level, industrial distribution, natural conditions, and cultural traditions, which lead to differences in the human–land relations in different regions. Such a complex system needs to be analyzed from multiple perspectives and disciplines, and the related research remains to be further explored. Finally, previous studies showed that there is a significant relationship between the contracted land disposal methods of rural migrants and the total of the household labor force [66,67]. Rural migrants are more likely to retain contracted land for family operating when there is more of the labor force left in their former rural residences. However, when the family labor force is insufficient, rural migrants often lease out the contracted land after migration or simply abandon it. Due to the fact that the CMDS did not collect household labor force information from respondents, this study did not consider this aspect in data analysis. This is also one of the regrets and deficiencies of this study. It is hoped that a complete study can be conducted based on more detailed data in the future.

6. Conclusions and Suggestions

Using CMDS data, this study analyzed the impact of migration behavior on the contracted land disposal methods of rural migrants, with the following main findings:

(1)

Chinese rural migrants exhibit more pronounced geographical differences in the disposition of contracted land. In the central provinces, the three northeastern provinces, Xinjiang, and other areas with relatively good agricultural conditions, contracted land often continues to be farmed through family operation or subleasing. In the western provinces, where agricultural production conditions are relatively poor, contracted land is generally left to be cultivated by family members or is abandoned. In the eastern provinces, where the secondary and tertiary industries are more developed, rural households tend to invest their limited human resources in non-agricultural industries, resulting in a high abandonment rate of contracted land in these areas.

(2)

Migration behavior contributes to the differences in contracted land disposal methods across regions. Educational attainment, the number of cities migrated to, and the house purchase rate significantly contribute to the geographical differences among all three contracted land disposal methods. The geographical differences between family operation and abandonment are also affected by the marriage rate, migration time, willingness to transfer hukou, and willingness to return to hometown. In contrast, the cross-provincial migration rate and willingness to stay and live influence the geographical differences in contracted land subleasing.

(3)

The mechanisms by which migration behavior influences rural migrants’ contracted land disposal decisions are not the same. Among them, the direction and degree of influence of each factor on contracted land family operation are relatively consistent across regions, with a high degree of homogeneity across the country. In terms of the subleasing and abandonment of contracted land, the mechanism by which the migration behavior influences land disposal decision-making exhibits significant regional heterogeneity, and the direction and intensity of each factor differ from region to region.

The migration of rural populations is a process of pursuing the improvement of their own and their family’s quality of life, and its essence is the reflection of issues, such as the “significant urban-rural development gap”, “imbalanced industrial distribution”, and “constraints on urban-rural dual structure”, on the field of population mobility. Therefore, in order to better coordinate the relationship between rural population migration and agricultural production, based on the research findings, this paper proposes the following suggestions. Firstly, the industrial layout should be optimized. The secondary and tertiary industries should be appropriately transferred to the main population move-out regions, increasing employment opportunities in these areas and shortening the migration distance of rural migrants, so agricultural production can be better taken into account, and the abandonment of contracted land can be reduced. Secondly, mechanized large-scale agricultural production should be encouraged in rural areas, ensuring that when rural migrants lose their willingness to farm, their contracted land can be effectively used by others. Finally, due to the existence of geographical differences, the same policies may not be appropriate across the country; thus, each region should formulate policies according to its own reality, to better solve the problem of rural land waste caused by the decrease in the rural population.

In addition, for the direction of future research, we propose the following two prospects: First, as found in this research, the same factors have different effects on the contracted land disposal methods of rural migrants in different regions. Future research could try to analyze the reasons for this phenomenon, to gain a deeper understanding of the human–land relationship system. Second, this research analyzes the effects of migration behavior on contracted land disposal methods, so future research could attempt to incorporate more indicators into the analysis framework, such as cultural factors, psychological factors, environmental factors, etc., and compare their effects, to identify more causes that affect the contracted land disposal methods of rural migrants.