The Impact of Forestland Tenure Security on Rural Household Income: Analysis of Mediating Effects Based on Labor Migration

Abstract

Although collective forest tenure reform (CFTR) has improved the legal tenure security of forestland, its impact on increasing farmers’ income is unsustainable. This study used a multiple linear regression model to empirically analyze data from 505 farmers in Jiangxi Province, examining the impact of legal, actual, and perceived tenure security on rural household income, and incorporating migration into the framework. The findings indicate that both actual and perceived tenure security have a substantial positive impact on the total rural household income and forestry income. However, it is worth noting that legal tenure security only has a positive effect on forestry income. Furthermore, outside-of-county labor migration can serve as a mediator for the income effects of actual and perceived tenure security. However, the mediating effect of intra-county labor migration is not considerable. The study found that the increase in income due to the security of actual tenure security is significant for the group of people who own less than 50 mu of forestland. However, both actual and perceived tenure security have a significant impact on income for the group of people who own more than 50 mu of forestland. The aforementioned findings indicate that, in the ongoing extensive advancement of collective forest right reform, it is crucial to prioritize the execution of forest reform policies at the local level and enhance farmers’ awareness and comprehension of said policies. In addition, the government should enhance the monitoring system for policy implementation and intensify efforts in publicizing these policies, in order to fully utilize the benefits of CFTR.

1. Introduction

Global society is presently grappling with issues pertaining to hunger and climate change, underscoring the imperative to advance the sustainable development of forested areas. Several studies have indicated that the reformation of collective forest rights can enhance the ecological environment, mitigate climate degradation, and decrease deforestation [1,2]. Hence, a considerable number of domestic and international scholars is interested in the reformation of collective forest rights. Foreign scholars primarily concentrate on the historical progression, current accomplishments, and forthcoming obstacles of the collective forest rights reform [3,4,5,6]. China has initiated three changes to enhance the utilization rate of collective forest areas by modifying the collective forest right framework. The third phase of the collective forest right reform, also referred to as the new round of collective forest tenure reforms (CFTRs), involves the transfer of the rights to use and manage forestland to rural households. The primary objective is to establish clear legal tenure for forestland by issuing forest tenure certificates, and subsequently improve the distribution of forestland resources [7]. Nevertheless, the problem of forest fragmentation has arisen during the reform process [8]. In the context of separating forestland possession and management, forest fragmentation can impede the progress of rural areas [9]. Forest fragmentation leads to increased operating expenses and reduced enthusiasm for farmers to manage forestland, which hampers the shift to large-scale operations [10]. Thus, in 2008, the Chinese government implemented policies such as forestry subsidies, logging quotas, forest rights mortgages, and other supportive policies to enhance the CFTR’s potential to generate income by establishing the clear tenure of forestland [11]. However, in reality, the CFTR has not successfully accomplished the intended policy objectives. The efficiency and effects of implementing the CFTR policy are influenced by factors such as regional topography, variations in forestry development, the initial level of forestry policy, and village autonomy. These factors contribute to differences between forestland tenure security and the expected results of the policy. The failure to popularize and implement the policy in tenure security will also result in a lack of perception and comprehension among farmers regarding tenure. Moreover, their skepticism toward the security of tenure will not promote positive business behavior among rural households. According to the survey published by Ma et al. (2015), while China has a high level of legal tenure security, the actual and perceived aspects of tenure security are inadequate [12]. Can CFTR-related policies enhance the legal, actual, and perceived aspects of tenure security? Can enhancing the security of forestland tenure successfully help to increase farmers’ income? Assuming the response is affirmative, which characteristic of tenure security primarily contributes to this income-enhancing effect? What are the specific methods by which this effect of increasing income is achieved? Addressing the aforementioned questions will assist in optimizing the CFTR and its accompanying policies based on local circumstances, while establishing a sustainable system to enhance the income of agricultural households in the long run.

When studying forestland tenure security, many scholars usually measure it by examining the proportion of forestland certificates owned [13]. However, this measurement only captures the actual aspect of tenure security. To provide a more comprehensive understanding, some scholars have introduced additional dimensions of tenure security and analyzed how these dimensions affect farmers’ behavior and if there are differences among them. From the perspective of farmers’ business behavior, some scholars propose that perceived tenure security can significantly enhance the frequency and intensity of farmers’ business practices [13]. According to certain scholars who study the behavior of rural labor migration, they have discovered that the perceived tenure security can encourage the migration of rural labor, whereas actual tenure security hinders labor migration. Furthermore, the impact of actual tenure security is less significant compared to the impact of perceived tenure security [14]. In their study, Ma et al. (2013) discovered that both the actual and perceived security of tenure have a favorable influence on farmers’ investment behavior [15]. However, the actual security of tenure primarily affects farmers’ investments in forestland construction and maintenance, whereas the perceived tenure security mainly impacts self-governed investments. Within China’s CFTR system, farm household income is a significant indicator of economic development in rural regions. Scholars are increasingly studying the connection between forestland tenure security, which is the central aspect of the CFTR, and rural households’ income. The majority of these scholars have a positive perspective, asserting that improving tenure security contributes to a higher rural household income [16]. According to Li et al. (2024), the granting of forest certificates can enhance total household income by facilitating the lease of forestland; however, the impact on forestry income is not substantial [17]. In contrast, Zhang et al. (2023) contend that ensuring the long-term title of forestland is crucial for enhancing total household income by facilitating specializations among farming households and encouraging investments in agriculture [18].

The level of labor migration has been increasing, and by the end of 2021, the cumulative figure of rural workers reached 292.51 million [19]. Off-farm employment has emerged as an essential means for farmers to generate income, and off-farm income has also become a significant source of income. An increasing number of rural households is allocating labor to off-farm industries, resulting in a phenomenon of rural population decline while simultaneously boosting the total household income. This hinders the forest rights reform from achieving the ultimate goal of increasing income, so more and more scholars are focusing on the role played by labor migration in the process of the impact of tenure security on income. Previous studies have extensively investigated the influence of tenure security on labor migration, although they have not yet reached a consensus. Some experts argue that enhancing the security of forestland tenure can encourage the migration of rural households. When title rights of forestland are unclear and rural households transfer their labor force, village officials are more likely to allocate the family’s land to other farmers in order to enhance land utilization efficiency, as the family size decreases [20]. Following the implementation of the CFTR, the establishment of clear title rights at the perceived level significantly decreases the risk of forestland loss when farmers seek employment opportunities and encourages labor migration. Yang and Ren (2020) substantiated the favorable impact of perceived tenure security on labor migration using data from collective forest regions in southern China [21]. Furthermore, improving the security of tenure for forestland decreases the risk of losing forestland once it has been rented out [22]. According to De et al. (2015), farmers might still receive rental income to offset the expenses associated with the labor migration procedure [23]. The enhanced security of tenure for forestland facilitates farmers in obtaining finance through forest rights mortgages, leading to the realization of production improvements [24]. Enhanced efficiency decreases the potential loss of choosing off-farm employment and enables the occurrence of labor migration [25]. According to Benjamin and Brandt (2002), increased forestland tenure security can incentivize labor migration among farmers who do not have financial access [26]. Conversely, there is a separate faction of scholars who maintain the contrary belief that ensuring the protection of forest title rights acts as a deterrent to rural labor migration. Xiao et al. (2022) discovered that a rise in the proportion of forestland certificates owned encourages farmers to actively engage in their businesses and allocate more labor resources to forested areas [27]. When farmers predict a decline for risk in expected income, they will increase their investments in capital, labor, and other resources. The increased allocation of labor and time by farmers toward forestry activities will impede the transition of labor to off-farm sectors [28]. Conversely, enhancing the security of tenure for forestland could promote the leasing out of forestland use rights. In their empirical investigation, Feng and Heerink (2008) discovered a negative association between forestland leases and the migration of labor among farmers [29]. The enhanced security of tenure for forestland will incentivize farmers to lease out forestland for the purpose of expanding forestland management size, but this will discourage the migration of rural labor.

While it is generally recognized that forestland tenure security may increase rural households’ income, there remains some debate regarding the specific aspect of income that will benefit from growth. The investigation into the mechanism of forestland tenure security as a means to enhance income remains highly significant, offering a crucial theoretical foundation for this study. After thoroughly examining the available literature, we identified certain deficiencies that still persist. Initially, the current research primarily assesses tenure security by examining the percentage of forestland certificates, which quantifies the extent of tenure security from a legal perspective. The CFTR aims to enhance the tenure security of farmers by providing forestland certificates, and subsequently influence the management practices of farmers in order to increase their income. Nevertheless, the perceived and actual tenure security for rural households significantly influence their management behavior. Merely assessing the security of tenure based on the number of forestland certificates issued at the legal level is insufficient [13,30]. Simultaneously, enhancing forestland tenure security may encourage labor migration and increase the number of farmers engaging in such a migration. This allows for the migration of labor from the forestry sector to off-farm industries, thereby achieving income growth through the “substitution effect”. Nevertheless, there is a scarcity of studies examining the differences in migration places of labor. This article examines the differences in income growth among farmers in Jiangxi Province in 2018, focusing on the legal, actual, and perceived aspects of forestland tenure security. Simultaneously, the analytical approach incorporates labor migration to examine the impact of labor migration place on mediating the income growth process of tenure security. This research makes significant contributions in several aspects. Firstly, it introduces an innovative method to differentiate forestland tenure security based on its legal, actual, and perceived dimensions. This comprehensive framework provides a more accurate representation of the varying levels of forestland tenure security. Secondly, this article differentiates from previous researchers by primarily using labor migration place as a measure of labor migration and further categorizes it into intra-county and outside-of-county labor migration. Third, this study replaces the measurement of perceived tenure security with the perception of tenure security among other farmers. This approach effectively addresses the issue of endogeneity between farmers’ perceptions of their tenure security and household income, thereby strengthening the validity of the conclusions. The structure of this paper is as follows: the first section presents the theoretical mechanism; the second section outlines the research design, including data sources, variable selection, and econometric model; the third section presents the estimation results, which includes the benchmark regression, mediation effect, robustness analysis, endogeneity test, and heterogeneity analysis; the fourth section is dedicated to the discussion; and the final section presents the conclusions and policy implications.

2. Theoretical Mechanism

2.1. Forestland Tenure Security and Rural Households’ Income

The CFTR has enhanced legal tenure security primarily by issuing forestland certificates. The establishment of legal security of tenure promotes the effective allocation of forestland resources and rural households’ labor resources [23]. At the level of forestland resource allocation, legal tenure security will encourage farmers to lease forestland [31]. Establishing a clear forestland tenure system can incentivize farmers to lease out unproductive forestland, while farmers who lease the rights to manage forestland can obtain a higher rental income [32,33]. According to Krul et al. (2020), farmers who rely on forestry will likely expand the size of their forestland [34]. Farmers expect reducing the total costs by expanding the size of the forestland, while simultaneously attaining extensive and concentrated outputs to enhance the efficiency of forestland production [35,36]. Furthermore, the establishment of legal tenure security can significantly decrease the probability of conflicts regarding forestland, hence decreasing the transaction costs [37] and promoting the leasing out of forestland from rural households to encourage the optimum use of rural forest resources. Regarding labor allocation, the recognition of forestland tenure at the legal level will incentivize farmers to allocate a portion the labor force to off-farm employment. An optimal forestland tenure framework could promote the action of rural labor and efficiently increase the supply of labor in rural areas [21]. The current disparity in wages between the primary industry and the secondary and tertiary industries remains present. Todaro’s theory of population mobility indicates that the difference in expected earnings will stimulate a continuous influx of extra rural labor into industries that provide greater comparative returns [38]. Furthermore, forestland tenure security includes both actual tenure security and perceived tenure security [39]. According to Ma et al. (2015), actual tenure security is mostly reliant on the effective implementation of laws and regulations concerning tenure at the local level [12]. Nevertheless, the enforcement of laws and regulations has been hindered by problems such as inaccuracies in forestland information entries and the inconsistent issuance of forestland certificates and felling certificates. These problems have led to a difference between the expected results and the real results [40]. On the contrary, farmers’ perceived tenure security is influenced by both legal and actual tenure security. In other words, farmers feel more secure about their forestland tenure when both legal and actual security groups are higher [14].

The statement above elucidates the connection between legal tenure security, actual tenure security, and perceived tenure security. It reveals that both actual tenure security and perceived tenure security are influenced by legal tenure security. Hence, the financial benefit of having legal tenure security also appears in various degrees of tenure security, though there may be differences in the rate of increase. Therefore, this study proposes the following hypothesis:

Hypothesis 1: 

The level of security in forestland tenure has a substantial and beneficial impact on the income of rural households.

2.2. The Intermediary Role of Labor Migration

Enhancing the security of forestland tenure will encourage rural households’ labor to migrate to off-farm industries, and the improved efficiency of these industries will increase the earnings of rural households, leading to a total increase in income levels. Nevertheless, various migration places will have distinct effects on income growth [41]. Referring to Li (2024), this study divides migration destinations into two categories: one is intra-county labor migration by rural households, and the other is inter-county labor migration by rural households [42]. The ongoing implementation of the rural revitalization strategy has resulted in labor migration, contributing to rural urbanization and reducing the economic development disparities among townships [43]. Consequently, the influence of intra-county labor migration on rural households’ income may not be substantial. Nevertheless, if labor migration occurs outside the county, enhancing the security of forestland tenure can mitigate the risk of losing forestland [22]. This has two benefits: on the one hand, farmers can lease forestland to generate a stable capital income. On the other hand, the rural labor force freed up by the leasing of forestland is typically directed toward economically prosperous regions, where they can earn higher wages compared to working in the township. Consequently, this leads to an increase in household wage income (Figure 1).

Hypothesis 2: 

The security of forestland tenure increases the total income of rural households by promoting the migration of labor.

Hypothesis 3: 

Forestland tenure security mainly promotes outside-of-county migration of labor and thus increases the total income of rural households, and the mediating effect of intra-county migration is not significant.

Figure 1. Mechanisms of forestland tenure security and rural households’ income.

Figure 1. Mechanisms of forestland tenure security and rural households’ income.

3. Research Design

3.1. Data Sources

The data utilized in this study originate from the field research and interviews carried out by the Collective Forest Rights Monitoring Research Group of Jiangxi Province during the summer break of 2019. The research focused on the forestland management methods of sample farmers in 2018 within the collective forest regions of Jiangxi Province. Stratified random sampling was employed in Jiangxi Province to select sample counties based on the distribution of forest resources and the level of regional socio-economic development. A total of 10 sample counties (municipalities) was chosen, representing 10.10% of the province’s total number of counties. These counties were evenly distributed in the north, center, and south of Jiangxi Province. Based on the per capita income of farmers and the growth of forestry in each township of the selected counties, two townships were chosen from each county, resulting in a total of 20 sample townships. In the designated townships, a selection process was carried out based on the forestry development status of each administrative village and its proximity to the township government. Each township chose 2–3 sample villages, and the total number of villages in each sample county (city) was limited to 5. Using the statistical data on farm household income obtained from the village roster, farm household income was ranked and equally divided into three sample groups (high, medium, and low). The team based on stratified sampling selected villages with better forestry development for the survey. This method of selecting a number of representative samples from the population is known as typical sampling. Therefore, this sampling followed a combination of typical sampling and random sampling principles. Random sampling was then conducted through face-to-face interviews within each sample group. Historical study data were analyzed by accessing public databases to determine the total standard deviation and acceptable margin of error. Based on the results of the analysis, a minimum sample size of approximately 488 was calculated to be required. Therefore, a study was conducted, involving a total of 505 farmers who were surveyed. Because the sampling was performed in representative urban areas to select sample farmers, and at the same time the combination of random sampling and stratified sampling was employed, the inclusion of representative data in this study improved the accuracy of its findings in capturing the broader context of Jiangxi Province.

Based on the above sampling principles, the group was divided into 10 groups to conduct the survey in different villages. The survey respondents were divided into two groups, households and villages, so the research team members mainly investigated the heads of households or farmers who knew their own family situation, while the team leaders mainly investigated the leaders of the villages, mainly the village chiefs and other village cadres who were familiar with the basic situation of the sample villages. To obtain the necessary data for the study, the research team created two types of questionnaires: the farm home questionnaire and the village questionnaire. The questionnaire for farmers primarily emphasizes the objective and cognitive elements of tenure security, information on family labor and employment, and the extent of comprehension and assessment of forestland resources and associated forestry policies. The village questionnaire primarily addresses the extent of tenure security at the legal level, the amount of forestry development in the selected villages, and the implementation of forestry policies within these villages. The project team members conducted field pre-surveys using preliminary questionnaires and outlines. Pre-researchers arrived in the research area 3–4 days before the research members arrived at their destination. Within each sample county, they randomly chose a village that they planned to research. Through pre-surveys, they gained an understanding of the basic situation of the research farmers and revised the questionnaires and interview outlines based on the findings. After the pre-research, they also conducted in-depth interviews with farmers to gain a better understanding of local forestland ownership and forestland management. And then, in July 2019, they officially started the formal research. During the actual research, the survey team members followed structured interviews. After each questionnaire was completed, it was checked by the researcher, and the research members checked each other’s. Finally, the lead teacher conducted a check. If any information was found to be missing or not in line with logical common sense, the researcher promptly made a return visit or telephone call back to the respondents to make additional corrections to the missing questions and erroneous contents. Through the above measures, the accuracy of the research data could be effectively improved.

For this research, we used a sample survey dataset with all individual identities removed, and it was approved and supported by the local Forestry Department and village officials.

3.2. Variable Selection

3.2.1. Independent Variable

While legal tenure security has been enhanced following the CFTR, there has not been an equivalent improvement in actual and perceived tenure security. This can be attributed to reasons such as village autonomy, transfer of information, and the specific characteristics of rural households. Consequently, the CFTR can result in various levels of changes in the legal, actual, and perceived aspects of forestland tenure security. This research aims to accurately assess the security of the tenure of forestland by developing a comprehensive scale that considers three groups. Legal tenure security refers to the duration of forestland management by the farmer under the Land Contract Law. Actual tenure security is determined by using variables such as the proportion of forestland certificates owned by the rural households, the challenges of obtaining logging indexes, and whether a contract has been signed. Perceived tenure security is based on the farmer’s comprehension of the “Three Rights Separation” and the expected adjustments made to forestland.

3.2.2. Dependent Variable

This study examines the changes in farm household income by analyzing both total income and forestry income as independent variables in the model. This study includes income as a variable in the regression model after using the logarithm transformation in the empirical analysis.

3.2.3. Mediating Variable

In this study, we refer to the studies of Xiao et al. (2020) and Liu et al. (2014), which classify labor migration as intra-county and outside of county [44,45]. Intra-county labor migration refers to the situation where rural households leave their hometown to work in the county town. Inter-county labor migration refers to rural households leaving their hometown to work in places outside the county town, which includes both working in areas within the same province but outside the county town and working in other provinces. This study utilizes the ratio of the number of workers who working outside the home to the total number of people in rural households as a more accurate indicator of the distribution of the labor force. It acknowledges that solely considering the number of people who have migrated may not provide an accurate representation of labor dynamics in these households. This study categorizes labor migration among farm households based on the variation in the distance between districts. It distinguishes between the proportion of work in the county and the proportion of work outside the county.

3.2.4. Control Variables

To enhance the validity of the empirical analysis in the study and minimize the volatility of the results caused by excluding other variables that could affect the income of farm households, this study incorporates a set of control variables across four dimensions: the personal attributes of the household’s head, the attributes of the farm household, the attributes of the forestland, and the characteristics of the village. This approach is based on the work of Xie et al. (2019) [46]. It is crucial to examine the personal traits of the head of the household, who plays a key role in making decisions on forest property management. Hence, variables such as gender, age, education level, employment, and whether the head of the household holds the position of village cadre are included to mitigate the influence of the head of the household’s individual characteristics. Forestland fragmentation was chosen as a means to control its impact on the process of forestry management. The characteristics of forestland play a crucial role in determining production and management decisions made by farmers. Aside from the aforementioned features, family characteristics and village characteristics are also significant control variables that impact the decisions of farmers [32]. The study includes three controlled variables at the household level: the number of individuals in the labor force, the occurrence of labor migration, and the decision to join farmers’ forestry professional cooperatives. At the village level, three variables are controlled: the percentage of labor migration, the percentage of village forestry income, and the village population (

Table 1. Variable description.

Variables	Variable Codes	Variable Definition	Variable Type	Mean	SD
Dependent variables
Total rural household income	lnincome	Total household income (CNY) logarithmized for the year before the research was conducted	Continuous variables	11.021	1.116
Forestry income	lnfincome	Forestry income (CNY) logarithmized for the year before the research was conducted	Continuous variables	4.304	4.280
Intermediary variable
Percentage of intra-county labor migration	Rin-county	Numbers for intra-county labor migration/off-farm household employment	Continuous variables	0.399	0.453
Percentage of outside-of-county labor migration	Rout-county	Numbers for outside-of-county labor migration/off-farm household employment	Continuous variables	0.589	0.454
Independent variables
Legal tenure security
Duration of forestland management	Duration	The mean period of management (in years) for the amount of forestland owned by rural households	Continuous variables	54.950	15.321
Actual tenure security
Proportion of forestland certificates owned	Rcer	Whether the forestland certificates are in your possession (yes =1; no = 0)	Binary variables	1.857	0.483
Level of complexity for the process of applying for logging	Logging	Has your household had any difficulty applying for logging targets? (Never applied = 2; yes = 1; no = 0)	Categorical variables	1.379	0.882
Execution of contractual agreements	Wcontract	Whether the rural household has a contract (1 = yes; 0 = no)	Binary variables	0.383	0.486
Perceived tenure security
Degree of understanding of the “Three Rights Separation” policy	Upolicy	Do you understand the “Three Rights Separation” policy? (yes = 1; no = 0)	Binary variables	0.681	0.471
Expectation of forestland restructuring	Expected	At present and in the future, does your household intend to lease forestland? (yes = 1; no = 0)	Binary variables	0.110	0.314
Evaluation of the “Three Rights Separation” policy	Evaluation	Do you think that the “Three Rights Separation” policy is good? (good = 3; fair = 2; bad = 1)”	Categorical variables	2.636	0.572
Evaluation of forest logging management policy	Satisfaction	How would you rate the current forest logging management policy? (satisfied = 3; fair = 2; dissatisfied = 1)	Categorical variables	2.375	0.685
Characteristics of the head of the household
Gender	gender	Sex of head of the household (M = 1; F = 0)		0.944	0.229
Age	age	Age of head of the household during year of survey (actual years)	Continuous variables	56.805	10.185
Educational level	edu	Educational level of the head of the household (elementary school and below = 1; middle school = 2; middle or high school = 3; college or bachelor’s degree or higher = 4	Categorical variables	1.899	0.823
Occupation	career	Farming = 1; farming and part-time work = 2; farming and side hustle = 3; permanent work outside the home = 4; regular-wage income = 5; other = 6	Categorical variables	2.825	1.921
Village cadres	carde	Whether the head of the household is a village cadre (yes = 1; no = 0)	Binary variables	0.392	0.500
Family characteristics
Number of laborers	numlabor	Number of family laborers (persons)	Continuous variables	2.712	1.529
Whether or not there is labor migration?	labortmig	Whether or not there is labor migration in the rural household (yes = 1; no = 0)	Binary variables	0.719	0.450
Membership in farmer forestry cooperatives	wcooperation	Are you a member of a farmers’ forestry cooperative? (joined = 2; yes but have not joined = 1; no local cooperatives = 0)	Categorical variables	0.280	0.624
Forestland characteristics
Forestland fragmentation	frafor	Household forestland area/number of forestland plots	Continuous variables	25.338	50.149
Village characteristics
Level of labor migration at the village level	Vmig	Number of permanent migrant workers in the village/total population of the village	Continuous variables	0.546	0.226
Level of village-level forestry income	Vfincome	Average per capita income from forestry in the village/average total per capita income	Continuous variables	0.197	0.187
Village population	Vnum	Logarization of the population of the village (persons)	Continuous variables	7.239	0.643
Instrumental variables
Understanding of the “Three Rights Separation” policy among other farmers in the village	Vupolicy	Are other farmers in the village aware of the “Three Rights Separation” policy? (yes = 1; no = 0)	Binary variables	0.364	0.482
Understanding of the “Three Rights Separation” policy by other farmers in the village	Vevaluation	Do other farmers in the village think that the “Three Rights Separation” policy is a good idea? (good = 3; fair = 2; bad = 1)	Categorical variables	2.268	0.555

Note: Intermediary variables: “Percentage of intra-county labor migration = Number of intra-county labor migration/off-farm household employment” and “Percentage of outside-of-county labor migration = Number of outside-of-county labor migration/off-farm household employment”.

).

3.3. Econometric Model

This study specifically examines the influence of forestland tenure security on rural households’ income within the context of the CFTR. Therefore, it is important to establish a clear cause-and-effect relationship between various aspects of forestland tenure security and rural household income. In order to understand whether there is a linear relationship between forestland tenure security and farm household income, this research used a multiple linear regression model. The linear regression model is able to fit the relationship between forestland tenure security and rural household income better through the OLS, and the coefficients are used to understand the degree of influence of forestland tenure security on rural household income. The specific model is developed as follows:

$$Y_i={\partial }_0+{\partial }_1{X}_i+{\partial }_2{Control}_i{+\epsilon }_i$$

(1)

In model (1), $Y_i$ represents the income of a rural household, $i$, which includes both the total income and the income earned from forestry activities. $X_i$ represents forestland tenure security, encompassing legal tenure security, actual tenure security, and perceived tenure security. Considering that different types of tenure security may interact with each other according to the theoretical analysis, this study referred to Liu et al. (2022) and incorporated legal tenure security, actual tenure security, and perceived tenure security separately into the regression model for stepwise regression [30]. ${Contral}_i$ represents the matrix of control variables in vector form. The coefficients ${\partial }_0,{\partial }_1$, and ${\partial }_2$ represent the values associated with the variables. ${\epsilon }_i$ represents a random error.

This research aims to investigate the impact of forestland tenure security on rural households’ income. To achieve this, the study incorporates off-farm employment as an intermediary variable in the analytical framework. Through empirical testing, this study examines the role of off-farm employment as an intermediary, and constructed a specific model as follows:

$$Z_i={\beta }_0+{\beta }_1{X}_i+{\beta }_2{Control}_i{+\epsilon }_i$$

(2)

$$Y_i={\gamma }_0+{\gamma }_1{X}_i+{\gamma }_2{Z}_i+{\gamma }_3{Control}_i{+\epsilon }_i$$

(3)

$Z_i$ represents the level of off-farm employment of the household, $i$, in the sample. This study measures the off-farm employment variable by calculating the proportion of labor migration from outside the county, taking into account the places of labor migration. The variables ${\beta }_0,{\beta }_1,{\beta }_2,{\gamma }_0,{\gamma }_1,{\gamma }_2,{\gamma }_3$ represent the coefficients. The remaining variables are established in the same manner as described previously.

The OLS model requires the dependent variable to be numerical or quantitative, and in this study, the total household income and forestry income of farmers, which serve as the dependent variables (Y), are both quantitative variables, hence the choice of the OLS model for the regression analysis. Due to the significant income disparities among farmers, income (Y) is log-transformed to reduce the problem of heteroscedasticity. For the small number of zero values in forestry income, the transformation ln (Y + 1) is applied. To ensure the validity and robustness of the regression model, this study conducted tests for multicollinearity and heteroscedasticity before performing the OLS regression analysis. It also considered the impact of autocorrelation, with the specific results as follows.

This paper employs the variance inflation factor (VIF) to test for multicollinearity. The results show that the VIF values of all variables are significantly less than the critical value of 10, and the average VIF value is maintained at a low level (approximately 1.20), indicating that there is no significant multicollinearity issue in the model. The explanatory variables exhibit a high degree of independence. To test for heteroscedasticity, this paper conducted the White test. The test results indicate that, in all models, the p-value of the White test is less than 0.05, firmly rejecting the null hypothesis of homoscedasticity, indicating that the model exhibits significant heteroscedasticity. To address this issue, this paper used robust standard errors in subsequent regression analyses to ensure the robustness of the estimation results. Since the data used in this paper are cross-sectional, the applicability of autocorrelation tests is limited. However, by adding the robust option, this paper has already considered potential within-group issues when calculating standard errors, thereby controlling for possible spatial autocorrelation to some extent. Therefore, in the context of this study, the impact of autocorrelation on the regression results is controllable.

4. Estimation Results

4.1. The Impact Forestland Tenure Security on Rural Household Income

Table 2. Regression results of forestland tenure security on rural households’ income.

Variables	Dependent Variables: lnincome			Dependent Variables: lnfincome
	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
Duration	−0.006 (0.004)	—	—	0.033 *** (0.013)	—	—
Rcer	—	0.190 * (0.108)	—	—	0.708 * (0.388)	—
Logging	—	−0.080 (0.064)	—	—	−0.506 ** (0.224)	—
Wcontract	—	0.052 (0.114)	—	—	0.667 * (0.402)	—
Upolicy	—	—	0.238 ** (0.110)	—	—	1.321 *** (0.394)
Expected	—	—	0.038 (0.165)	—	—	−0.875 (0.590)
Control	Yes	Yes	Yes	Yes	Yes	Yes
cons	8.515 *** (0.877)	8.853 *** (0.950)	8.436 *** (0.856)	9.310 *** (3.156)	11.420 *** (3.353)	8.039 *** (3.061)
R2	0.159	0.144	0.157	0.126	0.139	0.131

Note: *, **, and *** indicate significance at the 10%, 5%, and 1% statistical levels, respectively; values in parentheses are regression standard errors.

displays the regression results regarding the impact of forestland tenure security on rural household income. Model 1 did not identify any statistically significant impacts of duration of forestland on the total income of rural households. While the mention of a specific duration of forestland management in the law may encourage farmers to participate in such activities, it also limits their ability to allocate labor toward other off-farm activities that could potentially generate a higher income. As a result, legal tenure security does not have a significant impact on the total income of rural households in terms of total earnings. Model 2 captures the actual tenure security through the variable number of forestland certificates owned, level of complexity in the process of applying for logging, and execution of contractual agreements. The results indicate that the number of forestland certificates owned presents a notable and positive correlation with the total income of rural households. This result was statistically important at the 10% level of significance. On the contrary, the level of complexity in the process of applying for the logging and execution of contractual agreements did not have a significant impact on the total income of rural households. The increase in the percentage of forestland certificates helps farmers in efficiently distributing forestland and family labor resources, hence promoting the leasing of forestland and labor migration, ultimately leading to an increase in total income. Model 3 assesses the security of tenure at the perceived level using the degree of understanding of the “Three Rights Separation” policy and expectation of forestland restructuring. The findings indicate that, at a 5% significance level, an increase in rural households’ comprehension of the “Three Rights Separation” policy positively impacts the total income of rural households. However, the expectation of forestland restructuring did not have a significant impact on the total income of rural households. The understanding of the “Three Rights Separation” policy can enhance farmers’ comprehension of their rights at a subjectively perceived level. This, in turn, may improve farmers’ incentive to engage in commercial activities and ultimately lead to an increase in their income. Nevertheless, irrespective of farmers’ intentions to lease forestland, the immediate alteration in forestland management is unfeasible, thus the effect on total household income cannot be accurately depicted based on the data from a single year. The results mentioned above indicate that the tenure security of forestland can indeed enhance rural households’ income. However, it is important to point out that solely considering the legal dimension when assessing the security of tenure of forestland has its limitations. Specifically, the actual security of tenure, as opposed to the perceived security of tenure, plays a significant role in facilitating the increase in rural households’ income.

Table 2 illustrates the influence of forestland tenure security on the farmers’ forestry income. Model 4 demonstrates that duration of forestland management had a substantial impact on the increase in forestry income, with a significance level of 1%. Model 5 shows a statistically significant positive correlation between proportion of forestland certificates owned and forestry income, with this result passing the 10% significance level test; level of complexity in the process of applying for logging had a negative impact on forestry income, and this result was statistically significant at the 5% significance level. Farmers that sign contractual agreements usually achieve a higher forestry income. Model 6 demonstrates that the degree of understanding of the “Three Rights Separation” policy has a substantial impact on the increase in forestry income, with a significance level of 1%. Conversely, the expected forestland restructuring does not have a significant effect on forestry income. The results mentioned above suggest that legal tenure security, actual tenure security, and perceived tenure security all have a positive impact on the increase in rural households’ forestry income.

4.2. Mediation Effect

This research conducts a mediation effect test to further investigate the process by which forestland tenure security increases income. Improving the security of forestland tenure might alleviate the worries of farmers concerning labor migration and enable them to access extra income through off-farm employment opportunities. Nevertheless, the various places for labor migration will result in distinct consequences, namely an elevation in the psychological issues and transportation expenses incurred by farmers. Thus, this study examines whether intra-county and outside-of-county labor migration have a mediating impact on the increase in income resulting from actual tenure security and perceived tenure security.

4.2.1. The Mediation Effect of Outside-of-County Labor Migration

Table 3. Analysis of mediating effects of outside-of-county labor migration.

Variables	lnincome	Rout-County	lnincome	lnincome	Rout-County	lnincome
	Model 7	Model 8	Model 9	Model 10	Model 11	Model 12
Rcer	0.181 * (0.108)	0.044 ** (0.031)	0.038 * (0.118)	—	—	—
Upolicy	—	—	—	0.238 ** (0.110)	0.043 ** (0.040)	0.150 * (0.114)
Rout-county	—	—	0.214* (0.120)	—	—	0.210 * (0.119)
Control	Yes	Yes	Yes	Yes	Yes	Yes
cons	8.279 *** (0.872)	0.262 (0.420)	8.673 *** (0.969)	8.423 *** (0.853)	0.326 (0.409)	8.655 *** (0.942)
R2	0.154	0.077	0.071	0.157	0.076	0.075

Note: *, **, and *** indicate significance at the 10%, 5%, and 1% statistical levels, respectively; values in parentheses are regression standard errors.

displays the findings of the investigation on the mediating impacts of outside-of-county labor migration. Model 7, Model 8, and Model 9 examine the role of outside-of-county labor migration in mediating the relationship between actual tenure security and income growth. Model 7 demonstrates that proportion of forestland certificates owned had a significant impact on increasing the total income of rural households, with a confidence level of 10%. Model 8 determines that proportion of forestland certificates owned had the ability to enhance outside-of-county labor migration, and this result is statistically significant at a 5% level of significance. Model 9 provides evidence that the proportion of forestland certificates owned could effectively stimulate an increase in the total income of rural households by conducting tests on outside-of-county labor migration with a significance threshold of 10%. Models 10, 11, and 12 investigate the role of outside-of-county labor migration in moderating the relationship between perceived tenure security and income increase. Model 10 demonstrates that the degree of understanding of the “Three Rights Separation” policy effectively stimulated the increase in rural households’ total income, with statistical significance at the 5% level. Model 11 determines that the degree of understanding of the “Three Rights Separation” policy effectively enhanced outside-of-county labor migration, and this finding is statistically significant at a 5% level of significance. Model 12 demonstrates that the degree of understanding of the “Three Rights Separation” policy effectively enhances the increase in rural households’ total income by encouraging outside-of-county labor migration, as evidenced by a statistically significant test at the 10% significance level. The results mentioned above demonstrate that both the actual and perceived security of forestland tenure might contribute to the increase in the total income of rural households by enabling the migration of labor outside the county. The migration of labor from outside of the county can serve as a mediator in the process of increasing income through forestland tenure security. Forestland tenure security can decrease costs associated with labor migration and reduce the risk of forestland loss [22,23].

4.2.2. The Mediation Effect of Intra-County Labor Migration

Table 4. Analysis of mediating effects of intra-county labor migration.

Variables	lnincome	Rin-County	lnincome	lnincome	Rin-County	lnincome
	Model 13	Model 14	Model 15	Model 16	Model 17	Model 18
Rcer	0.181 * (0.108)	−0.050 (0.051)	0.037 * (0.118)	—	—	—
Upolicy	—	—	—	0.238 ** (0.110)	−0.025 (0.050)	0.158 * (0.115)
Rin-county	—	—	−0.187 (0.121)	—	—	0.210 * (0.119)
Control	Yes	Yes	Yes	Yes	Yes	Yes
cons	8.279 *** (0.872)	0.588 (0.422)	8.961 *** (0.981)	8.423 *** (0.853)	0.506 (0.412)	8.945 *** (0.954)
R2	0.154	0.077	0.065	0.157	0.075	0.069

Note: *, **, and *** indicate significance at the 10%, 5%, and 1% statistical levels, respectively; values in parentheses are regression standard errors.

displays the findings of the investigation on the mediating effects of intra-county labor migration. Model 13, Model 14, and Model 15 examine the role of intra-county labor migration in mediating the relationship between actual tenure security and income growth. Model 13 demonstrates that the proportion of forestland certificates owned had a significant impact on the increase in rural households’ total income, with a significance level of 10%. Model 14 determines that the impact of the proportion of forestland certificates owned on intra-county labor migration was not statistically significant. Model 15 provides evidence that, based on a test with a significance level of 10%, an increase in the proportion of forestland certificates owned continues to have a positive impact on rural households’ total income growth when intra-county labor migration is taken into account. However, the effect of intra-county labor migration on income generation is not statistically significant. Model 16, Model 17, and Model 18 investigate how intra-county labor migrations influences the perceived increase in income security associated with tenure. According to Model 16, there was a statistically significant relationship between improving the degree of understanding of the “Three Rights Separation” policy and promoting rural households’ total income growth at a significance level of 5%. Model 17 determines that there was no significant correlation between the degree of understanding of the “Three Rights Separation” policy and intra-county labor migration. Model 18 determines that there was a statistically significant positive relationship between the degree of understanding of the “Three Rights Separation” policy and intra-county labor migration and the increase in $lnincome$ at a 10% significance level. The aforementioned results suggest that the role of intra-county labor migrations as a mediator is not statistically significant, whether it is examined in terms of actual tenure security or perceived tenure security.

4.3. Robustness Analysis

4.3.1. Replacement of Core Independent Variables

To ensure the reliability of the results mentioned above, this study substitutes the indicator of perceived tenure security in the main independent variables, the test model. Simultaneously, in order to minimize the impact of outliers on the results of this study, the data are also subjected to a process of data reduction and re-regression. Referring to Liu et al. (2023), this section replaces the measure of perceived tenure security with the evaluation of farmers on the “Three Rights Separation” policy and the evaluation of forest logging management policy in the regression analysis [47]. The results of the regression analysis are presented in

Table 5. Impact of perceived tenure security on rural households’ income.

Variables	Dependent Variables: lnincome	Dependent Variables: lnfincome
	Model 19	Model 20
Evaluation	0.088 ** (0.054)	0.337 ** (0.177)
Satisfaction	0.034 (0.070)	0.305 (0.290)
Control	Yes	Yes
cons	9.252 *** (0.902)	9.088 ** (4.098)
R2	0.173	0.098

Note: **, and *** indicate significance at the 5%, and 1% statistical levels, respectively; values in parentheses are regression standard errors.

. Model 19 determines that there was a positive correlation between farmers’ satisfaction with the evaluation and an increase in rural households’ income. However, the impact of the evaluation of forest logging management policies on rural households’ total income is not statistically significant. Model 20 showed a substantial positive correlation between the evaluation of the “Three Rights Separation” policy and forestry income, as confirmed by passing the 5% significance level test. However, the impact of the evaluation of forest logging management policies on forestry income was determined to be insignificant. The results mentioned above validate the strength and reliability of the conclusion of the positive impact of perceived tenure security on rural households’ income.

4.3.2. Replacement Model

In order to enhance the reliability of the remaining results in the research, baseline regression was subjected to an additional test using a robust clustered standard error regression. The results of this regression are presented in

Table 6. Impact of forestland tenure security on rural households’ income-based on robust clustered standard error regression.

Variables	Dependent Variables: lnincome			Dependent Variables: lnfincome
	Model 21	Model 22	Model 23	Model 24	Model 25	Model 26
Duration	0.005 * (0.003)	—	—	0.033 *** (0.012)	—	—
Rcer	—	0.190 * (0.105)	—	—	0.708 * (0.388)	—
Logging	—	−0.080 (0.056)	—	—	−0.506 ** (0.224)	—
Wcontract	—	0.052 (0.110)	—	—	0.667 * (0.402)	—
Upolicy	—	—	0.238 ** (0.117)	—	—	1.321 *** (0.394)
Expected	—	—	0.037 (0.141)	—	—	−0.875 (0.590)
Control	Yes	Yes	Yes	Yes	Yes	Yes
cons	8.515 *** (1.022)	8.853 *** (1.190)	8.436 *** (1.052)	9.310 *** (3.156)	11.420 *** (3.353)	8.039 *** (3.061)
R2	0.159	0.144	0.157	0.126	0.139	0.131

Note: *, **, and *** indicate significance at the 10%, 5%, and 1% statistical levels, respectively; values in parentheses are regression standard errors.

. Model 21 determined that there was a positive relationship between duration and the increase in rural households’ total income. This result was confirmed by passing a significance level test at the 10% threshold. In Model 22, there was a strong positive connection between number of forestland certificates owned and rural households’ total income, and this association was statistically significant at the 10% level. However, there is no significant effect of level of complexity in the process of applying for logging and the execution of contractual agreements on the total income of rural households. Model 23 identified a statistically significant positive correlation between the degree of understanding of the “Three Rights Separation” policy and rural households’ total income at a significance level of 5%. However, the expectation of forestland restructuring did not have a meaningful impact on rural households’ total income. Model 24 demonstrates that duration was a statistically significant factor for the increase in forestry income at a 1% level of significance. Model 25 discovered a statistically significant positive correlation between proportion of forestland certificates owned and forestry income, with the results satisfying the 10% significance threshold. Additionally, level of complexity in the process of applying for logging had a significant negative impact on forestry income, with this result satisfying the 5% significance threshold. Furthermore, execution of contractual agreements typically produced higher levels of forestry income, and this result met the 10% significance threshold. The significance level met the criteria of the test. Model 26 provides evidence that the degree of understanding of the “Three Rights Separation” policy had a substantial impact on forestry income at a statistically significant level of 1%, whereas the expectation of forestland restructuring did not have a meaningful influence on forestry income. The conclusions obtained by substituting the model remain in agreement with the aforementioned findings, exhibiting no substantial disparity or even an increase in the degree of significance. This ensures that the conclusions formulated in this study are not influenced by chance.

4.3.3. Shrinkage Therapy

The presence of outliers in the sample data may result in a distortion of the results achieved in this study. To further confirm the strength of the preceding results, this study performed bilateral shrinkage at the 1% quantile for each of the three aspects of forestland tenure security and rural household income. The results are presented in

Table 7. Impact of forestland tenure security on rural households’ income.

Variables	Dependent Variables: lnincome			Dependent Variables: lnfincome
	Model 27	Model 28	Model 29	Model 30	Model 31	Model 32
Duration	0.005 * (0.003)	—	—	0.031 ** (0.012)	—	—
Rcer	—	0.182 * (0.096)	—	—	0.642 * (0.385)	—
Logging	—	−0.071 (0.057)	—	—	−0.494 ** (0.223)	—
Wcontract	—	0.011 (0.102)	—	—	0.587 (0.399)	—
Upolicy	—	—	0.180 * (0.100)	—	—	1.316 *** (0.393)
Expected	—	—	0.023 (0.148)	—	—	−0.842 (0.585)
Control	Yes	Yes	Yes	Yes	Yes	Yes
cons	9.092 *** (0.789)	9.483 *** (0.852)	9.015 *** (0.772)	9.046 *** (3.142)	11.00 *** (3.339)	7.681 ** (3.044)
R2	0.185	0.167	0.178	0.134	0.147	0.141

Note: *, **, and *** indicate significance at the 10%, 5%, and 1% statistical levels, respectively; values in parentheses are regression standard errors.

. In Model 27, duration positively influenced the increase in rural households’ total income, and this result is statistically significant at the 10% level. According to Model 28, an increase in the number of forestland certificates owned leads to an increase in rural households’ total income. This conclusion is confirmed by passing the 10% significance level test. However, level of complexity in the process of applying for logging and the signing of contractual agreements did not have a significant effect on the total income of rural households. Model 29 determines that the degree of understanding of the “Three Rights Separation” policy increased the total income of rural households based on the 10% significance level test. However, there was no significant association between expectation of forestland restructuring and rural households’ total income. Model 30 provides evidence that duration had a significant impact on the increase in rural households’ total income at a significance level of 5%. According to Model 31, the number of forestland certificates owned enhanced forestry income growth and this relationship was statistically significant at the 10% significance level. At the same time, level of complexity in the process of applying for logging had a negative impact on forestry income, and this effect was statistically significant at the 5% significance level. Lastly, the signing of contractual agreements did not have a statistically significant effect on forestry income. Model 32 demonstrates that, with a 1% level of significance, the degree of understanding of the “Three Rights Separation” policy made a considerable contribution to forestry income. However, it was found that the expectation of forestland restructuring did not have a significant effect on forestry income. The results above suggest that all three elements of tenure security enhancement contribute to the increase in rural households’ income. The significance level of this result is not substantially distinct from the results obtained previously, which strengthens the reliability of this paper’s conclusions.

4.4. Endogeneity Test

When farmers have a clear understanding of their rights, they experience a decrease in the risks associated with forestland management and expect a steadier income. This, in turn, enhances their enthusiasm to effectively manage forestland [48]. The increase in household income can be indicative of the farmers’ efficiency in managing forestland, thereby enhancing their understanding of the “Three Rights Separation” policy, which validates farmers’ rights based on the empirical evidence. The sample data chosen for this study may be affected by endogenous interference caused by measurement and associativity errors. To mitigate the potential interdependent relationship between variables and enhance the accuracy of the conclusions, this research employs the instrumental variable method to further address the issue of endogeneity. To prevent the interdependent relationship between farmers’ understanding of the “Three Rights Separation” policy and their own households’ income, this study replaces the measurement subject with other farmers in the village. Specifically, the understanding of other farmers regarding the “Three Rights Separation” policy is used as a substitute for the understanding of individual farmers regarding the same topic. Put simply, the understanding of the “Three Rights Separation” policy among farmers is substituted with the farmers’ own understanding of the “Three Rights Separation” policy as the instrumental variable. Furthermore, to enhance the reliability of the results, this paper incorporates the research conducted by Yang et al. (2020) and incorporates the assessment of other farmers in the village regarding the “Three Rights Separation” policy as an instrumental variable for the regression analysis [13]. The regression results are presented in

Table 8. Regression results of forestland tenure security on rural households’ income after accounting for instrumental variables.

Variables	Dependent Variables: lnincome
	Model 33	Model 34	Model 35	Model 36
Duration	−0.004 (0.004)	−0.007 (0.005)	0.004 (0.007)	−0.006 (0.008)
Rcer	0.161 (0.122)	0.205 * (0.117)	0.151 (0.123)	0.202 * (0.117)
Logging	−0.195 ** (0.079)	−0.108 (0.081)	−0.100 (0.131)	−0.095 (0.122)
Wcontract	0.042 (0.131)	0.001 (0.125)	0.003 (0.139)	−0.009 (0.124)
Vupolicy	0.686 * (0.473)	0.814 ** (0.466)	—	—
Vevaluation	—	—	1.950 ** (0.706)	0.809 ** (0.625)
Expected	0.081 (0.187)	0.074 (0.176)	0.029 (0.197)	0.071 (0.176)
Control	No	Yes	No	Yes
cons	11.490 *** (0.704)	9.274 *** (1.453)	10.180 *** (1.597)	9.024 *** (2.439)
R2	—	0.110	—	0.138

Note: *, **, and *** indicate significance at the 10%, 5%, and 1% statistical levels, respectively; values in parentheses are regression standard errors.

. Before using instrumental variables, it is imperative to assess the soundness of the chosen instrumental variables. The original hypothesis of the Wald exogeneity test, which examines endogeneity and weak instrumental variables in relation to the understanding of other farmers regarding the “Three Rights Separation” policy and the evaluation of the “Three Rights Separation” policy at the village level, is rejected with a significance level of 1%. This implies that there is a problem of endogeneity between farmers’ understanding of the “Three Rights Separation” policy and total household income. The instrumental variable tests conducted in Models 33–36 demonstrate that there is no issue of weak instrumental variables. Model 33 demonstrated a negative correlation between level of complexity in the process of applying for logging and rural households’ total income, but the understanding of the “Three Rights Separation” policy among other farmers in the village had a favorable impact on the total income of rural households. Model 34 identifies a favorable impact of number of forestland certificates owned and an understanding of the “Three Rights Separation” policy among other farmers in the village on rural households’ total income when taking control variables into account. The Model 35 demonstrates that an understanding of the “Three Rights Separation” policy by other farmers in the village had a positive correlation with rural households’ total income. Model 36 provides evidence that both the number of forestland certificates owned and understanding of the “Three Rights Separation” policy by other farmers in the village had a significant positive impact on rural households’ total income, even after accounting for control variables. In other words, it can be observed that both the actual security of tenure and the perceived security of tenure can still have a positive impact on the total income of farm households. This confirms the robustness and dependability of this study.

4.5. Heterogeneity Analysis

4.5.1. Size of Forestland Management

When analyzing the impact of forestland management size on rural households’ income, previous studies commonly employed regression analyses that incorporated forestland size either as a moderating variable or as a control variable. Forestland size is an endogenous variable, meaning that it can impact farmers’ income by allowing for economies of scale. As a result, farmers may be motivated to expand their forestland size of management and share the average cost when their income level increases. This study will introduce forestland management size as a grouping variable. For forestland management size less than or equal to 50 mu, $areatype$ will be assigned as 1. For forestland management size larger than 50 mu but less than or equal to 100 mu, $areatype$ will be assigned as 2. For forestland management size greater than 100 mu, $areatype$ will be assigned as 3. The results of the grouping regression are displayed in

Table 9. Differential analysis of the income-generating effects of tenure security for different forestland management size groups.

Variables	Areatype = 1			Areatype = 2			Areatype = 3
	Model 37	Model 38	Model 39	Model 40	Model 41	Model 42	Model 43	Model 44	Model 45
Duration	−0.004 (0.005)	—	—	−0.003 (0.007)	—	—	−0.012 (0.008)	—	—
Rcer	—	0.229 * (0.129)	—	—	0.295 (0.214)	—	—	0.016 (0.297)	—
Logging	—	−0.067 (0.090)	—	—	0.009 (0.122)	—	—	−0.249 * (0.135)	—
Wcontract	—	0.202 (0.162)	—	—	0.367 * (0.217)	—	—	0.244 (0.247)	—
Upolicy	—	—	0.046 (0.145)	—	—	0.419 * (0.222)	—	—	0.401 * (0.242)
Expected	—	—	0.217 (0.211)	—	—	0.027 (0.375)	—	—	−0.309 (0.366)
Control	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
cons	8.794 *** (1.288)	9.481 *** (1.400)	9.029 *** (1.259)	9.028 *** (1.634)	10.15 *** (1.732)	9.164 *** (1.585)	8.603 *** (2.101)	8.476 *** (2.241)	7.623 *** (2.044)
R2	0.247	0.236	0.242	0.158	0.181	0.176	0.216	0.223	0.213

Note: * and *** indicate significance at the 10% and 1% statistical levels, respectively; values in parentheses are regression standard errors.

. The analyses of Model 37, Model 40, and Model 43 indicate that the impact of duration of forestland management on farmers’ income does not exhibit a statistical significance when considering various sample groups based on the forestland operation area. According to the test on actual tenure security, Model 38 demonstrated a positive correlation between number of forestland certificates owned and rural households’ total income for the sample group of forestland management size less than 50 acres. This result is statistically significant at the 10% level of significance. Model 41 indicates that the signing of contractual agreements had a positive effect on the total income of rural households for the sample group of forestland management size ranging from 50 to 100 acres. This result also passed the 10% significance level test. Additionally, Model 41 shows that the signing of contractual agreements had a positive effect on forestry income for the sample group of forestland management size ranging from 50 to 100 acres, and this result is statistically significant at the 10% level of significance. Model 44 indicated a negative relationship between level of complexity in the process of applying for logging and rural households’ total income for the sample group consisting of a forestland management size larger than 100 acres. The variation in the results may be attributed to the varying methods utilized by farmers of different management sizes to generate income. This discrepancy also affects the significance of different dimensions used to measure actual tenure security, in spite of the fact that actual tenure security plays a role in increasing the total household income of rural households. According to Model 39, the impact of perceived tenure security on rural households’ total income was not statistically significant for the subset of individuals in the study group who had a forestland management size of 50 acres or less. Models 42 and 45 demonstrate that the degree of understanding of the “Three Rights Separation” policy had a substantial impact on rural households’ total income for the subset of farmers with a forestland management size over 50 acres. This finding is statistically significant at the 10% level of significance. In contrast, expectation of forestland restructuring did not have a significant effect on rural households’ total income.

4.5.2. Whether or Not Labor Migration Occurs

The data mentioned above indicate that households discover an increase in total income when labor migrations take place. However, several studies have revealed that rural households with lower levels of income are more likely to engage in labor migration [49]. The study aims to examine the variation in income improvement resulting from tenure security among different households in the sample. To achieve this, the study categorizes households based on whether or not labor migration occurs. For rural households without labor migration, the value $Labortransfer$ = 0 is assigned, while rural households with labor migration are assigned the value $Labortransfer$ = 1. The regression results are shown in

Table 10. Differential analysis of the income-enhancing effects of tenure security.

Variables	Labortransfer = 0			Labortransfer = 1
	Model 46	Model 47	Model 48	Model 49	Model 50	Model 51
Duration	−0.010 (0.008)			−0.003 (0.004)
Rcer		0.039 (0.123)			0.461 ** (0.221)
Logging		−0.229 (0.167)			−0.030 (0.066)
Wcontract		0.016 (0.266)			0.123 ** (0.091)
Upolicy			0.308 (0.262)			0.276 ** (0.120)
Expected			−0.119 (0.372)			0.074 (0.182)
Control	Yes	Yes	Yes	Yes	Yes	Yes
cons	9.441 *** (2.108)	10.58 *** (2.368)	9.529 *** (2.119)	9.593 *** (0.949)	9.462 *** (1.021)	9.379 *** (0.915)
R2	0.154	0.121	0.118	0.049	0.058	0.054

Note: ** and *** indicate significance at the 5% and 1% statistical levels, respectively; values in parentheses are regression standard errors.

. The results of Model 46, Model 47, and Model 48 suggest that there is no substantial impact of forestland tenure security on the total income of rural households in the sample, when labor migration is not considered. Model 49 suggests that duration did not have a statistically significant impact on rural households’ total income. Model 50 demonstrates that the number of forestland certificates owned and the signing of contractual agreements had a statistically significant impact on the increase in rural households’ total income at a 5% significance level. Model 51 demonstrated a substantial relationship between the degree of understanding of the “Three Rights Separation” policy and rural households’ total income, with a significance level of 5%. Within the sample group experiencing labor migration, there is a direct correlation between actual tenure security, perceived tenure security, and rural households’ total income. This result supports the hypothesis that labor migration plays a role in enhancing tenure security.

5. Discussion

This study examines the influence of three aspects of forestland tenure security on rural households’ income. The analysis utilized a multiple linear regression model with data from Jiangxi Province in 2018. Most scholars currently consider the possession of forestland certificates as a measure of forestland tenure security [27], or just investigate actual tenure security and perceived tenure security [30]. This paper introduces an innovative method to categorize forestland tenure security into three groups: legal tenure security, actual tenure security, and perceived tenure security. Through an empirical analysis, it is determined that legal tenure security does not have a significant impact on the total income of rural households. However, both actual tenure security and perceived tenure security have a significant positive effect on the total income of rural households. Furthermore, all three groups of forestland tenure security contribute to the increase in farmers’ forestry income. Forestland tenure security has a positive impact on farmers’ management of forestland [13]. However, both the actual and perceived groups of tenure security can affect farmers’ decision making and how they allocate resources, especially the proportion of off-farm income. These factors, in turn, result in changes in rural households’ income. This discovery aligns with the research conducted by Zhang et al. (2023) [18]. The above results show that it is not enough to focus only on the formulation and release of policies related to the security of tenure, but it is also necessary to pay attention to the implementation of policies in forestland tenure security, and to strengthen the supervision and control of the implementation of these policies. In addition, it is also worthwhile to enhance the publicity of the policy on forestland tenure, so as to improve the security of farmers’ tenure from the cognitive level, and to stimulate farmers’ enthusiasm for forestland management, which will help to guarantee the existing achievements of the CFTR and build a mechanism for farmers to increase their income.

This study emphasizes the mediating influence of labor migration, which is an important consideration in farmers’ decision-making process regarding labor resource allocation and forestland leases [45]. However, in contrast to previous researchers who primarily assessed the labor migration of rural households by examining the level of labor migration [50,51], this study categorizes the labor migration of rural households into two distinct categories: intra-county labor migration and outside-of-county labor migration. The analysis examines both perspectives of labor migration based on location. The results indicate that the impact of intra-county labor migration on income is not statistically significant. Additionally, non-farm employment can diversify the income sources of rural households and stimulate total income growth through the substitution effect. While attempting to achieve the ultimate goal of the CFTR, it is crucial to focus on the labor migration of rural households and provide guidance for them to move their labor to areas outside the county. A portion of the wages obtained by workers that is sent outside the county will be reinvested in rural areas, thus fostering the synchronized advancement of rural regions. Furthermore, backing the expansion of forestry will facilitate the reintegration of the labor force that has migrated to outside the county, thereby mitigating the scarcity of rural laborers.

This research examines the possibility of an interdependent relationship between rural households’ perceived tenure security and their income. To address the issue of endogeneity, we replaced the measure of perceived tenure security with instrumental variables. The results demonstrate that the perceived security of forestland tenure can continue to have significant effects on the income of rural households, even after substituting the measurement method. This supports the durability and reliability of the conclusions presented in the study.

Furthermore, this section delves into the advantages and disadvantages of data selection. Firstly, this study chose cross-sectional data from 2018 as the year when the CFTR policy was implemented in Jiangxi Province. The data of farmers for this year are picked to be representative and typical. Nevertheless, it is not possible to compare the data from a single year with previous years in order to determine if there have been any changes in the indicators of tenure security. Furthermore, it is not feasible to assess the resilience of the paper’s findings and investigate the enduring patterns using the data from later years. Furthermore, the 2018 data exhibit a significant temporal disparity from the present, indicating a noticeable delay. Due to the ongoing reform and innovation of forest rights legislations, it is necessary to gather additional data for different years for the regression analysis in order to determine if the conclusions of this research are still relevant to the current situation in the rural areas of China.

Moreover, there are certain viewpoints that have not been thoroughly examined in this study. This study examines the relationship between forestland tenure security and the income of rural households. Specifically, it focuses on the total income of rural households and forestry income as measures of changes in income levels. Nevertheless, in the present circumstance of the progressive expansion of the concept of rural revitalization, it is insufficient to solely concentrate on the income level of rural households. It remains essential to investigate the alterations in the income levels of these households in order to promote the establishment of a sustainable income-generating mechanism. To investigate the impact of and variation in tenure security on income level and structure, the team will include off-farm income and other sources of income in the upcoming study. Furthermore, this research introduces an innovative method to assess forestland tenure security by considering its legal, actual, and perceived groups in the analysis of independent variables. However, do variations in the effects of various aspects of tenure security on farmers’ income exist? Furthermore, which aspect of tenure security has a greater influence on different aspects of farmers’ income as a consequence of these variations? We lack the capacity to comprehend this. To gain a deeper understanding of the diverse effects of forestland tenure security on rural households’ income and to develop more precise forest rights laws, our team plans to conduct further extensive studies in the future.

6. Conclusions

This study establishes a theoretical framework that examines the relationship between forestland tenure security, rural labor migration, and rural households’ income. It empirically assesses the different impacts of forestland tenure security on rural households’ income using data from 505 farm households in collective forest areas in Jiangxi Province in 2018. This study employs a multiple linear regression model to examine the influence of three groups of forestland tenure security on the income disparities among rural households. Additionally, it investigates the mediating role of labor migration of rural households in the process of enhancing forestland tenure security to promote income growth. The research results are as follows:

(1)

In terms of legal tenure security, duration does not have significant effect on lnincome. However, for each additional unit of duration, the total income of rural households grows by 3.30%. In terms of actual tenure security, an increase of one unit in the number of forestland certificates owned is associated with a 19.00% growth in rural households’ total income and a 70.80% growth in forestry income. However, there is no significant relationship between level of complexity in the process of applying for logging and signing of contractual agreements with rural households’ total income. On the other hand, a one-unit increase in level of complexity in the process of applying for logging results in a 50.60% decrease in forestry income, and an increase of one unit in the signing of contractual agreements results in a 66.70% increase in forestry income. Regarding the perceived tenure security, an increase of one unit in the degree of understanding of the “Three Rights Separation” policy results in a 23.80% increase in rural households’ total income and a 132.10% increase in forestry income. However, the expectation of forestland restructuring does not have a significant impact on income. Based on the statistical data presented in this study, it is evident that a majority of farmers own forestland certificates. However, there is still a need to enhance the implementation mechanism for issuing these certificates. Simultaneously, it is imperative to expedite the reform of pertinent supportive policies, while concurrently conducting outreach and educational initiatives to better farmers’ understanding of tenure security and legal concepts. This will facilitate the collective improvement of tenure security at the legal, actual, and perceived levels.

(2)

Outside-of-county labor migration can act as a mediator in the number of forestland certificates owned and degree of understanding of the “Three Rights Separation” policy to achieve an increase in rural households’ income, but the mediating effect of intra-county migration is not substantial. The migration of labor may promote the integration of the CFTR results into rural households’ income and stimulate the economic advancement of rural regions. To enhance the income growth of rural households, it is advisable to incentivize the migration of labor from farming to off-farm sectors. This will help diversify the income sources available to farm households. Simultaneously, the results of this study indicate that the involvement of intra-county labor migration as an intermediary in the process of enhancing income through forestland tenure security is not substantial. This observation partly suggests that the disparities between rural and smalltown regions are diminishing, but there is still a gap compared to economically advanced areas. To achieve the ultimate objective of forest reform, it is imperative to provide guidance to farmers to relocate to more urbanized areas. Furthermore, it is crucial to consider the investment in human capital of agricultural households. Following migration, farm households mostly engage in manual labor, which might result in migrant workers struggling to satisfy their financial needs, especially in the face of increasing living costs. To construct a sustainable and consistent method of increasing income, it is crucial to focus on the occupational skills and education of farm households, as well as the development of a robust foundation for education in rural areas.

(3)

The study analyzes whether there are variations in the income impacts of forestland tenure security among different groups of farm households who have different forestland management size and labor migration scenarios, in order to analyze heterogeneity. This article shows that, among farmers with a land size of 50 mu and lower, the number of forestland certificates owned significantly enhances rural households’ total income in the sample groups categorized by forestland management size. The execution of contractual agreements has a significant effect on rural households’ total income among farmers with 50–100 acres of land. The level of complexity for the process of applying for logging has a significant negative effect on rural households’ total income in the sample group with a forestland management size of 100 acres or more. The positive impact on income of the degree of understanding of the “Three Rights Separation” policy is evident among all investigated farmers who own forestland measuring 50 acres or larger. This study shows that the beneficial impacts of the number of forestland certificates owned and the degree of understanding of the “Three Rights Separation” policy on rural households’ total income are only observed in households if labor migration has taken place within the sample group. Furthermore, unlike the regression results for the entire sample, the impact of the variable signing of contractual agreements on the total income of rural households is statistically significant for households that have experienced labor migration.