Poverty Alleviation Resettlement and Household Natural Resources Dependence: A Case Study from Ankang Prefecture, China

Abstract

In order to assess the degree to which China’s poverty alleviation resettlement (PAR) has been able to address the development conundrum of natural resources reliance and human welfare, it was necessary to investigate the effects of PAR on rural households with regard to their dependence on natural resources. This article evaluated households’ natural resources dependence in rural China by constructing a natural resources dependence index and empirically analyzing the effect of PAR on households by using household survey data from Ankang Prefecture, located in southern Shaanxi Province. The findings demonstrated that PAR could effectively decrease the dependence of households on local natural resources, thus safeguarding the natural environment. Moreover, there were noteworthy distinctions regarding households’ natural resources dependence. This research endeavored to complete the fusion of natural resource dependence and PAR at the household level, and then contemplated the policy implications of PAR on rural households’ dependence on resources, furnishing fresh information for future evaluations of nature conservation and development policies.

1. Introduction

Development-induced displacement and resettlement, referring to the intentional relocation of a population due to natural catastrophes, urban growth, impoverishment, depletion of natural resources or development projects, is a reality in numerous countries across the world [1,2,3,4,5]. Poverty alleviation resettlement (PAR) is a national rural development policy that employs resettlement as a means to address resource constraints and poverty-related issues in an ever-evolving China. It is viewed as one of the most effective avenues for the impoverished to escape poverty in the implementation of the targeted poverty alleviation (TPA) strategy [6]. The aim of PAR is to tackle the quandary of development that “the environment of a given locale cannot sustain its inhabitants”, thus striving to attain environmental conservation and poverty relief in spite of limited resources [7]. Poverty is inextricably linked to geographic and environmental conditions, particularly the spatial poverty traps of remote mountainous, arid and semi-arid regions [8,9]. Previous studies have examined the impacts of disaster resettlement programs on population problems, such as their influence on social transformation, means of subsistence, volition hesitancy and the eradication of poverty [10,11]. Rural households are both the primary project participants and a vital stakeholder group, and also the basic unit of consumption decisions and social production during the implementation of the PAR [7,12,13]. The influence of natural resources on relocated households is essential. Measuring and assessing the advantages and dependence on the natural resources of PAR households could assist in producing scientifically sound and practical decisions after relocation resettlement.

Natural resources serve as the main source of sustenance for households in these areas, which typically rely heavily on local resources due to their poor living conditions and lack of alternative means of income. Consequently, these households are left with no choice but to depend on natural resources to meet their basic requirements [14]. The varying dependence of relocated households on natural resources partially reflects the efficacy of the relocation plan. In order to assess the extent of a given household’s dependence on such resources, researchers developed the concept of natural resources dependence. The proportion of a household’s income stemming from natural resources in relation to their total income has often been utilized as a metric to gauge the reliance of households on natural resources [15]. Most of the extant research has gauged reliance on natural resources by calculating the ratio of income earned by households from natural-resource-related activities, such as forestry, agriculture, and stock farming, to the total household income [16,17,18,19,20]. According to the results of a study on Bangladesh by Mukul, S.A. et al., the contribution of natural resources to the household economy is high, ranging from 14% to 45% [21]. Stefano Balbi et al. [22] constructed a measure of overall reliance on natural resources and modeled the detected dependence patterns with the help of demographic variables such as gender, income, and family size. Additionally, natural resources have been shown to bolster households, create prospects for increased livelihoods and shield against risks [23]. Households’ natural resources dependence can be divided into direct and indirect dependence, or local and nonlocal dependence [24,25]. Direct use of natural resources is manifested when individuals and families collect items, such as firewood for cooking and warming, grass for animal fodder, or fish for sustenance. Indirect use of natural resources, besides, involves purchasing products derived from these resources, such as fish from the market or fabric made from cotton. This article is mainly concerned with the direct dependence of natural resources. Comprehending the extent to which households depend on forest resources is essential for ensuring sustainable management and long-term conservation of forests [26]. In addition, research has shown that income inequality is greater in countries with high resource dependence [27]. Furthermore, Li et al. [28] postulated that government guidance and follow-up support were of paramount importance, primarily concentrating on the restructuring of income from natural resources and reducing dependence upon them. Studies on households’ natural resources dependence focus on livelihood capital [18,19], its relationship with households’ poverty and management [29,30], measurement methods and evaluation system [31], and its influencing factors [15,20]. No study has yet linked China’s PAR to the natural resources dependence of migrant relocated households. In this article, we measured the degree of natural resources dependence of households in China’s PAR areas, attempts were made to elucidate the effects of PAR on the natural resources dependence of relocated households and provided new evidence for a deeper understanding of the ecological effects of relocation and resettlement.

Studies have shown that social excessive reliance on natural forests for conservation strategies in protected areas can have detrimental effects. This over-dependence can present significant obstacles to the achievement of conservation objectives [29,30]. A number of studies on natural resources dependence have concentrated on the macro level, exploring the connection between these resources and development [31]. Some studies incorporated livelihood capital into their analysis models at the micro level to measure the natural resources dependence of households [19,30]. For instance, households’ consumption patterns, natural resources dependence, and utilization of natural resources have all been adjusted to decrease their natural resources dependence and the environment [32,33]. Grasping households’ natural resources dependence is therefore critical for the sustainable management and lasting protection of these resources [26]. Furthermore, poverty is a crucial factor in determining dependency on natural resources [34]. Some studies have associated dependence on natural resources with livelihood capital; due to the distinctions between households’ livelihood capital and living conditions, the selection of livelihood adaptation strategies could change [16,35]. Furthermore, the study found that livelihood strategies dependent on natural resources include firewood collection, medicinal materials, plants, livestock feed, food and building materials [36].

Most of the research on natural resources dependence has been concentrated on the macro and environmental-protection spheres [37,38]. Several recent studies though have evaluated dependence levels on natural resources in different areas around the world [16,17,27,39]. Meanwhile, few micro-level studies have been conducted within the context of China’s PAR. In the context of PAR in China, studies at the micro level mainly focus on the households’ livelihood [35,39,40], factors affecting the relocation of households [41,42], and the disposal of agricultural land after relocation [43], etc., but there is no study that combines China’s PAR with households’ natural resources dependence. Li Cong et al. studied the impact of PAR on the dependency of households on ecosystem services. They found that there were significant differences in the dependency of ecosystem services among households with different relocation factors. The implementation of PAR can effectively reduce households’ dependence on ecosystem services. Features such as voluntary relocation, centralized settle, and the new-stage relocation also play a positive role in reducing households’ dependence on ecosystem services. At the same time, it also emphasizes the construction and guidance of relocated households’ self-development ability in the process of non-agricultural transformation. They built an appropriate model and formed a systematic theory in their research. However, the natural resources dependence of households studied in this article is similar to the ecosystem service dependency mentioned by Li Cong in terms of variable division and calculation methods. Therefore, based on previous studies, this article references the theory obtained by Li Cong and other scholars for research. Based on previous studies, this paper proposes the following hypothesis: the implementation of PAR can effectively reduce households’ dependence on natural resources, and there are differences in the degree of natural resources dependence among households with different relocation features. Features such as voluntary relocation, centralized settle, and the new-stage relocation will also reduce households’ natural resources dependence. The Ankang prefecture, located in Shaanxi Province, China, recently implemented a range of policy interventions, and it was deemed essential to conduct an extensive empirical study to examine the dependence of the resettlement sites. This article thus sought to address this gap. This study drew upon existing literature to examine two questions. First: What is the degree of the rural households’ natural resources dependence, in the context of China’s PAR? Second: What is the impact of China’s PAR on households’ natural resources dependence? To explore the answers to these questions, this article established a relatively effective and comprehensive evaluation index system for determining a households’ natural resources dependence. Subsequently, the level of natural resources dependence among households was calculated and the influence of migrants’ relocation on households’ natural resources dependence was evaluated through using the established regression model. Finally, the discussion and conclusions were presented.

2. Materials and Methods

2.1. Study Site

This research was conducted in Ankang Prefecture, situated in Shaanxi Province, to explore the resettlement process that occurred post disaster [28]. Ankang Prefecture is situated to the north of the main ridge of the Qinling Mountains and the north slope of the QinBa Mountain. It has 3 million permanent residents and is located in the hinterland of Qinba, which borders poverty-stricken areas. The convergence of regions that are prone to disasters, experience poverty, and have ecological significance has posed great challenges to local economic and social advancement, contributing to a highly fragile ecology, an intense degree of poverty, and a meager rate of urbanization. As a key element of the massive relocation program initiated in Shaanxi Province in 2011, the Ankang government has been devoted to addressing the subsistence and growth of the disadvantaged populace [39]. The relocation project involves about 2.4 million people in 28 prefectures in Shaanxi Province [12].

2.2. Data Collection

The data provided by the text were obtained from a survey on the livelihood of rural households in Ankang Prefecture in south China’s Shaanxi Province (for more information about the survey, consult Li et al. [28]). The three districts and counties selected in the survey were Ziyang County, Hanbin District, and Ningshan County, respectively; they were the focal points of the national poverty relief action. We adopted a convenient sampling design and administered questionnaire-based surveys. Taking into consideration the representativeness of the sample selection and the feasibility of the research program, we first chose three more typical centralized resettlement communities in Ziyang County, as well as four towns and eight administrative villages, in Hanbin District and Ningshan County. These had already embarked on ecology compensation projects such as the sloping land conservancy program. After that, we applied convenience sampling for these administrative villages and resettlement communities. Random choice of household owner means the randomness of the selection. Finally, members of households between the ages of 18 and 65 were randomly chosen for the survey. In the process of questionnaire survey, we took the adult head of the farm household (usually male) as the survey object. If the head of the household was not at home that day, the adult female (the spouse of the household owner) of the household was selected as the survey object. If both were not at home, other family members between 18 and 65 years old were selected for the survey. Generally, the family situation is not well understood enough to ensure the quality and completeness of the survey results, so it is not within the scope of our survey. The questionnaire includes five parts: basic information of households, land use, conversion of farmland to forest and family livelihood, family production and consumption behavior, relocation project, rural community participation and poverty alleviation through tourism. During the process to investigate the field, in order to guarantee the quality of the data, first, our investigators were all systematically trained and composed of experienced teachers, graduate students and undergraduate students. We then assigned group leaders to each survey team to monitor and adjust their operations at any time during the survey. Third, the group leader selected some of the completed questionnaires and reviewed them daily with these respondents. These results were returned to the investigator on the same day. After the investigation, the entire questionnaire was converted into numerical codes and entered into the dataset. In the end, we performed both logical and digital data cleaning procedures to further ensure the quality of the input data. This survey took the form of face-to-face interview, a total of 800 questionnaires were distributed, 670 were recovered with the recovery rate of 83.75%; After the data review and cleaning, 657 valid questionnaires were obtained with an effective rate of 98.06%, including 459 samples of relocated households and 198 samples of non-relocated households. Most of the households moved from the mountain area of the village and the mountain area of the village near the town, accounting for 80% of the total, less cross-town relocation.

2.3. Indicator Construction

Households’ natural resources dependence has typically been assessed by the proportion of their total gross income deriving from natural resources [15]. Vedeld et al. [44] performed a theoretical investigation of the earnings of households relying on natural resources, hypothesizing that environmental income was the rent or additional benefit acquired by households in the course of natural resources entering the first market link. In Soman et al.’s [45] exploration, the forest dependence index was characterized by calculating the mean value of the income dependence index, fuel dependence index, housing dependence index and drug dependence index. Studies also investigated the extent of households’ natural resources dependence in the economic sphere. Natural resources dependence is classified into three levels: income dependence, food dependence and energy dependence [15,18,20]. By referring to previous studies, we first divided natural resources dependence into three levels. The first was: (1) income dependence, determined by whether a households’ income sources included revenue from the utilization of natural resources. In this study, income resources dependent on natural resources were divided into three parts: agriculture, forestry, and aquaculture. The natural resources income was equal to the sum of the above three annual incomes divided by the total annual peasant household income (%) [46]. The second level was: (2) food dependence, natural resources utilization in the form of self-sufficiency of the household food supply. This was calculated as follows: the total annual household self-sufficiency of food expenditure was divided by the total annual household food consumption (%) (“the total annual household self-sufficiency of food expenditure” refers to household food self-provision) [20]. The third level was: (3) energy dependence, indicating the proportion of natural resources directly consumed to meet households’ energy needs. To calculate it, the household annual firewood collection income was divided by the household annual energy consumption expenditure (%) [47]. Finally, the total degree of dependence of the relocated households on natural resources was determined by calculating the average of three indices—income dependence, food dependence and energy dependence [45]. The higher the natural resources dependence index, the more dependent the households were on the corresponding natural resources [48]. To calculate this index, data, e.g., household income divided into gross income per capita, subsistence food income per capita, per capita annual total food consumption, income from firewood collection per capita, per capita annual energy consumption, etc., were obtained (see

Table 1. Households’ natural resources dependence.

Item (Unit)	Mean Value	Standard Deviation	Minimum	Maximum
Gross income per capita (yuan)	7785.379	13,504.670	0.000	125,075.000
Per capita income from natural resources (yuan)	999.865	4030.130	0.000	51,590.000
Income dependence (%)	0.190	0.261	0.000	1.000
Subsistence food income per capita (yuan)	496.003	2545.486	0.000	50,496.000
Per capita annual total food consumption (yuan)	2697.486	3517.663	0.000	60,000.000
Food dependence (%)	0.150	0.199	0.000	1.000
Income from firewood collection per capita (yuan)	134.257	414.516	0.000	7500.000
Per capita annual energy consumption (yuan)	657.294	854.361	0.000	13,333.330
Energy dependence (%)	0.177	0.280	0.000	1.000
Natural resources dependence (%)	0.1722	0.182	0.000	0.860

).

In addition, we calculated that the income dependence of non-relocated households was 0.281%, food dependence was 0.226%, energy dependence was 0.363%, and the natural resources dependence was 0.290%. Therefore, the dependence of households before relocation was high.

Relocation factors, representing different relocation scenarios, were the primary focus of the research, including whether it was part of the relocation of households and relocation features, the relocation features includes relocation type, resettlement mode and relocation time [35]. The control variables included livelihood assets, family features and geographical features.

Table 2. Setting and value of independent variables.

Variables	Variables Setting	Value
		Mean	Standard Deviation
Whether Relocated
Relocated-family	Relocated household takes 1; otherwise takes 0	0.700	0.459
Relocation Feature
Relocation type	Voluntary relocation takes 1; involuntary relocation takes 0	0.840	0.363
Settlement mode	Centralized settle takes 1; scattered settle takes 0	0.758	0.429
Relocation time	The new stage (2011 and after) takes 1; the early stage takes 0	0.689	0.464
Livelihood Assets
Natural capital
Per farmland area	The ratio of total land area to the total population (unit: mu a/person)	1.019	2.543
Per woodland area	The ratio of total forest area to the total population (unit: mu a/person)	4.251	20.583
Sloping land conservancy program	Yes takes 1, No takes 0	0.730	0.446
Physical capital
Per housing area	The ratio of housing area to total household size (%)	40.656	29.929
Per own assets	The ratio of productive tools to total household size (%)	1.002	0.536
Financial capital
Difficulty of borrowing	Total assets of the rural household are normalized 3 indicators are combined b	0.278	0.257
Social capital
Cooperatives participation	Yes takes 1, No takes 0	0.037	0.189
Number of village cadres	Number of village cadres in relatives and friends (persons)	0.500	1.447
Monetary help	Number of households available to provide support (persons)	3.972	5.306
Human capital
Education years of household members	The actual average years of education of the household membersin the survey year (years)	6.325	2.474
Per health status	Total health status divided by total household population (healthy for 1, generally healthy for 0.67, unhealthy for 0.33)	0.8333	0.219
Per skill level	The ratio of total skills to total household size (%)	0.170	0.232
Family features
Family members	Number of family members (persons)	4.500	1.608
Labor force ability	The number of workers in the household relative to the household size	0.739	0.220
Geographical features
Distance to the road	Distance between household’s house and main village road (within one mile for 1, two miles to five miles for 0.67, five miles outsides for 0.33)	0.981	0.088

^a 1 mu = 0.667 hm²; ^b 3 indicators including whether the household borrowed from the bank, whether the household borrowed from the government and whether the household borrowed from relatives and friends.

lists the values of variables and exact specifications.

Relocation types encompassed both voluntary relocation and involuntary relocation [49], among the 459 relocated households, 395 households had been relocated voluntarily, comprising of 86.06%. Voluntary relocation included poverty alleviation resettlement, ecological resettlement, and disaster avoidance relocation. Involuntary relocation included engineering resettlement and relocation caused by tourism development [7]. Among the relocated households, disaster avoidance immigrants are the most, followed by poverty alleviation immigrants and ecological immigrants. Relevant studies have found that the severity of disaster has a significant positive correlation with residents’ relocation intention [50]. Households can choose two resettlement modes: for centralized settle and scattered settle. In centralized resettlement, an entire village is turned into a new customized community facility; in scattered resettlement, residents are dispersed into smaller departments in their original resettlement sites [35,51]. The South Shaanxi Relocation Plan from 2011 and other relocation activities had been carried out previously. However, the allocation, magnitude, and backing were much less than those afforded during later periods [28]. Consequently, we referred to those who relocated after 2011 as belonging to the new stage, while those who relocated before the new policy was implemented were denoted as the early stage. This was done to differentiate the effects of policies between the two periods, and to evaluate the transformation of households’ ecological behavior. Livelihood capital is a significant influence on households’ willingness to relocate [52]. According to the British International Sustainable Livelihoods Analysis Framework proposed by DFID, the livelihood capital of households consists of five elements: natural capital, physical capital, financial capital, social capital, and human capital. Natural capital consists of per farmland area, per woodland area, and whether a household participated in the sloping land conservation program; this reflects the economic situation of households who own farmland and woodland areas [51]. The rural–urban relocation of labor force will lead to the change of households’ land use [52]. Previous studies have found that the sloping land conservancy program affected households’ participation and income in other livelihood activities by changing the form of factor input in agricultural and forestry planting activities [53]. Physical capital includes per housing area and per own assets; housing is often used to characterize material resources [54,55]. The ratio of productive tools to total household size represents the measure of the per own assets of the household, and is the ratio of the means of production, means of vehicles and durable goods of the household to the household population. The means of production will promote the efficient completion of production activities, and the quantity of durable goods reflects the life quality of the household [7,56]. Financial support is explained by the number of family members who can provide support, as financing from relatives or friends is the basic means of informal financing [57]. Financial capital includes the difficulty of borrowing and includes three indicators: whether the household borrowed from the bank, whether the household borrowed from the government, and whether the household borrowed from relatives and friends [39]. Social capital includes the participation of professional cooperatives, the number of village cadres in relatives and friends and the number of households available to provide support [39]. The number of households to seek help when in urgent need of large expenses represents the size of the households’ social support network, as loans from relatives and friends are a basic channel for informal financing [58]. In rural China, when households face risks and shocks, their social networks can act as a buffer, preventing them from exploiting their environment in a predatory manner [35]. Human capital includes the education years of household members, per health status, and per skill level [49]; the higher the standard of the household members’ education, the lower the natural resources dependence, as they have more profitable sources of income available [16]. Education provides improved and diversified employment opportunities, thereby reducing dependence on forests [29,57,59]. The average education years of household members is 6.32 years. The ratio of total skills to total household size represents the measure of the per skill level of the household. The per skill level of the household is 0.17. The more opportunities households have to receive skill training, the better it is for them to learn and master various skills and enhance their livelihood ability [53]. The more adept the household members are at non-farm work technology, the lower their degree of natural resources dependence [16]. Family features include family members’ labor force ability [35]. As household size accelerates, the number of household members who depend on natural products increases, and so the demand for natural resources increases [45]. The average age of the head of a household is 50.50 years old and the average family size is 4.50 years old. This paper selected the ratio of household labor force to the total family population, with the labor force being the number of household members aged 16–65 [35]. The average number of workers in the household is 3.27. The average of labor force ability is about 0.74. Geographic feature is indicated by the distance of housing from the main road in the village, which affects access to the market [54]. The mean distance of housing from the main road of the village is 0.98 miles. In rural areas, households situated closer to the main roads in the village have improved transportation options, in addition to more choices in production and lifestyle.

2.4. Analysis Method

The Tobit regression model was employed in the study to evaluate the effects of PAR on a households’ natural resources dependence. This model was suitable for dealing with the issue of a censored/restricted explanatory variable; if the data in the explained variable were censored/restricted, the Tobit model could be used to address the problem of inconsistent estimators caused by deletion. In addition, the dependent variable, the overall natural resources dependence, was right-censored. It fit the requirement of a Tobit model proposed by Tobin (1958). And by referring to the relevant studies of Li et al., we believed that Tobit model was adopted for data regression processing in this study, expressed in the following formula:

$$\begin{gathered} Natural Resources Dependence y=\left\{\begin{array}{l}1 if y_i^{\ast }>1, \\ y_i^{\ast } if y_i^{\ast }\le 1,\end{array}\right. \\ {y}_i^{\ast }=\beta x_i+{\mu }_i,{\mu }_i\sim N\left(0,{\sigma }^2\right), \end{gathered}$$

where $y_i^{\ast }$ is the latent variable, and x_i is the illustrated variable vector, namely, relocation factors (including whether relocated and relocation features), livelihood assets, family features, and geographical features. β is the parameter estimation coefficient, and µ_i is the perturbation term. Among them, relocation type, settlement mode, and relocation time, as important features of PAR, are included in the second stage of regression to investigate the impact of relocation features on the sample of relocated households. Referring to the research of Liu et al.’s [7,37], the three variables are usually included in the model as the main independent variables, and the influence of PAR is examined from different perspectives. As a reference, the three variables are also included in the regression model in this article.

Following previous studies by Li et al. [7], the regression method was as below: first, the variable “whether they belong to relocated households” and control variables were covered in the whole sample regression model to explore the impact of participating PAR on households’ income dependence, food dependence, energy dependence, and natural resources dependence. In the second step, to evaluate the impact of relocation features on their natural resources dependence, we used a sample of relocated households, where the three relocation features—that is the relocation type, resettlement mode, and relocation time—were involved in the model with the control variables. The control variables in the regression model included: livelihood assets, family characteristics, and geographical features.

3. Results and Analysis

3.1. Comparison of Rural Households’ Benefits from Natural Resources

In-depth information about the resettlement factors of survey households can be found in the research by Liu et al. [49].

Table 3. Incomes of relocated households, non-relocated households, and the total sample.

Income (Unit: Yuan)	Relocated Households		Non-Relocated Households		The Total Sample
	Mean (Standard Deviation)	Minimum: Maximum	Mean (Standard Deviation)	Minimum:Maximum	Mean (Standard Deviation)	Minimum: Maximum
Total Agricultural Income	668.401 (4498.597)	0:90,000	21,874.016 (4358.370)	0:40,000	1043.659 (4486.733)	0:90,000
Total Forestry Revenue	655.908 (3658.086)	0:64,000	2728.069 (2713.037)	0:23,270	678.444 (3389.250)	0:64,000
Gross Income of Aquaculture	1944.240 (15,582.330)	0:286,698	24,834.930 (24,844.680)	0:257,470	2799.696 (18,829.250)	0:286,698
Natural Resources Income	3265.602 (17,426.560)	0:286,698	27,057.935 (25,241.540)	0:257,950	4417.447 (20,175.660)	0:286,698
Amount of Self-contained Food	1012.551 (3369.906)	0:39,000	24,920.807 (22,146.650)	0:252,480	2196.463 (12,614.640)	0:252,480
Fuel Wood Collection	347.189 (1709.928)	0:30,000	21,019.44 (1481.006)	0:252,480	542.999 (1673.550)	0:30,000

shows that the non-relocated households received more income from natural resources than their relocated counterparts, with a difference of 3792.33 yuan in terms of gross income.

Moreover, the non-relocated households’ mean self-contained food income was 3908.26 yuan higher than the relocated households and the firewood collection income was 772.25 yuan higher. This shows that non-relocated households relied more on natural resources for income, food, and firewood than relocated households.

3.2. Comparison of Rural Households’ Natural Resources Dependence

Table 4. Comparison of households’ dependence on natural resources.

Indices	Whether Relocated		t-Test	Relocation Type		t-Test	Settlement Mode		t-Test	Relocation Time		t-Test
	Yes	No		Voluntary	Involuntary		Centralized	Scattered		New Stage	Early Stage
Income dependence	0.150	0.281	−5.283 ***	0.134	0.226	−2.471 **	0.142	0.189	−1.917 *	0.129	0.201	−2.803 ***
Food dependence	0.116	0.226	−6.302 ***	0.112	0.137	−0.995	0.110	0.153	−2.123 **	0.110	0.136	−1.401
Energy dependence	0.097	0.363	−11.039 ***	0.062	0.248	−4.715 ***	0.066	0.195	−4.268 ***	0.083	0.131	−1.914 *
Natural resources dependence	0.121	0.290	−10.904 ***	0.103	0.204	−4.570 ***	0.106	0.179	−4.001 ***	0.107	0.156	−2.974 ***

Note: *, **, *** indicate statistical significance at the level of 10%, 5%, and 1%, respectively.

presents a comparison of the level of natural resources dependence and its sub-indicators among households of different relocation features.

It was evident that relocated households had a lower level of dependence on natural resources than non-relocated households. Furthermore, voluntarily relocated households had lower natural resources dependence than those who were involuntarily moved. Additionally, households who were resettled in a centralized manner showed a lower dependence on natural resources than those who were scattered in their relocation. Finally, those households who moved more recently had a lower dependence on natural resources than those who moved earlier.

3.3. Determinants of Natural Resources Dependence

3.3.1. Effects of Relocation and Its Factors on Households’ Income Dependence

Based on the descriptive statistical comparison, the quantified income dependence was selected as the dependent variable; the influence of PAR on household income dependence was regressed. The results are listed in

Table 5. Analysis of the impact of relocation and its factors on the income dependence of households.

Variables	Model 1	Model 2	Model 3	Model 4
Whether Relocated
Relocated-family	−0.199 ***
Relocation Feature
Relocation type		−0.121 **
Settlement mode			−0.058
Relocation time				−0.075 *
Livelihood Assets
Per farmland area	−0.002	−0.005	−0.006	−0.004
Per woodland area	0.000	−0.001	0.000	0.000
Sloping land conservancy program	0.053	0.010	0.038	0.034
Per housing area	0.001 *	0.001	0.001 *	0.001
Per own assets	−0.047	−0.084 *	−0.071	−0.075 *
Financing channels	−0.108 *	−0.123	−0.117	−0.097
Cooperatives participation	0.234 ***	0.329 ***	0.294 ***	0.294 ***
Number of village cadres	−0.003	−0.001	−0.001	−0.002
Monetary help	0.004	0.007 **	0.007 **	0.007 **
Education years of household members	−0.009	−0.010	−0.006	−0.006
Per health status	0.011	−0.042	−0.075	−0.055
Per skill level	−0.059	−0.026	−0.045	−0.036
Family features
Family members	0.005	−0.006	0.002	0.003
Labor force ability	0.010	0.042	0.094	0.090
Geographical features
Distance to the road	−0.095	−0.173	−0.226	−0.226
Constant	−0.345 *	0.459	0.338	0.3337
R2	0.103 ***	0.074 *	0.067 *	0.071 *
Sample size	548	340	378	377

Note: *, **, *** indicate statistical significance at the level of 10%, 5%, and 1%, respectively.

.

After lost value and odd value analyses, 548 total samples were entered into the model. Model 1, based on the total sample, tests the effect of households participation in PAR on household income dependence. On this basis, to further analyze the influence of relocation features on income dependence, relocation type (Model 2), settlement mode (Model 3), and relocation time (Model 4) were included in the sample of relocated households.

On the basis of the total sample (Model 1), regression results emphasize that PAR significantly reduced households’ income dependence. Furthermore, the voluntary relocation and new-stage relocation significantly decrease household’s income dependence. In Model 2, involuntary relocation was shown to enhance income dependence; compared with households who relocate involuntarily, households who relocate voluntarily are more willing to accept relocation psychologically. Compared with the relocation type and relocation time, the effect of settlement mode on the income dependence of households is not very significant. Additionally, Model 4 revealed that the new relocation project noticeably decreased households’ income dependence, mirroring the government-initiated systematic progress. As for the control variables, cooperatives participation has the most significant influence on the income dependence. Monetary help has a great influence in Models 2, 3, and 4, while the other control variables have no significant influence on the dependent variables separately.

3.3.2. Effects of Relocation and Its Factors on Households’ Food Dependence

A regression interpretation of the influence of relocation factors on households’ food dependence was presented; the consequences are shown in

Table 6. Analysis of the impact of relocation and its factors on the food dependence of households.

Variables	Model 5	Model 6	Model 7	Model 8
Whether Relocated
Relocated-family	−0.201 ***
Relocation Feature
Relocation type		−0.051
Settlement mode			−0.090 **
Relocation time				−0.061
Livelihood Assets
Per farmland area	0.004	−0.001	−0.001	−0.000
Per woodland area	0.002 *	0.006	0.002	0.002 *
Sloping land conservancy program	0.089 ***	0.054	0.067	0.068
Per housing area	0.000	0.000	0.001	0.001
Per own assets	−0.043	−0.067	−0.100 **	−0.102 **
Financing channels	−0.101 *	−0.116	−0.111	−0.097
Cooperatives participation	0.097	0.093	0.029	0.041
Number of village cadres	−0.007	0.002	0.007	0.005
Monetary help	0.007 ***	0.010 ***	0.010 ***	0.010 ***
Education years of household members	−0.008	−0.018*	−0.014	−0.015 *
Per health status	0.150	0.249 *	0.184	0.208
Per skill level	0.009	0.063	0.005	0.013
Family features
Family members	−0.010	−0.011	−0.017	−0.014
Labor force ability	−0.038	0.012	0.030	0.032
Geographical features
Distance to the road	−0.09	−0.374 *	−0.339	−0.336
Constant	0.213	0.294	0.353	0.297
R2	0.160 ***	0.083 **	0.092 ***	0.085 **
Sample size	548	340	378	377

Note: *, **, *** indicate statistical significance at the level of 10%, 5%, and 1%, respectively.

.

This regression was based on the total sample (Model 5), which indicated that PAR reduced households’ food dependence. The centralized settle and the new-stage relocation observably impacted the relocated households’ food dependence. Model 7, centralized resettlement reduced households’ food dependence. Model 8, new stage relocation reduced households’ food dependence; this is because the households who participated the new-stage relocation had more diversified access to food materials. As for the control variables, per own assets and the education years of household members can significantly reduce households’ food dependence, and monetary help also has a significant impact on households’ food dependence. Because households have a fixed way to obtain food, some control variables, such as housing area, cooperatives participation, number of village cadres, skill level, family members and labor force ability, have little effect on households’ food dependence.

3.3.3. Effects of Relocation and Its Factors on Households’ Energy Dependence

In

Table 7. Analysis of the impact of relocation and its factors on the energy dependence of households.

Variables	Model 9	Model 10	Model 11	Model 12
Whether Relocated
Relocated-family	−0.568 ***
Relocation Feature
Relocation type		−0.186 ***
Settlement mode			−0.111 ***
Relocation time				0.045 *
Livelihood Assets
Per farmland area	0.002	0.001	0.000	0.001
Per woodland area	0.005 ***	0.004 **	0.003 ***	0.004 ***
Sloping land conservancy program	0.183 ***	0.074 ***	0.070 ***	0.073 ***
Per housing area	0.002 *	−0.000	0.000	0.000
Per own assets	−0.028	−0.009	−0.010	−0.011
Financing channels	−0.031	0.034	0.028	0.031
Cooperatives participation	0.182	0.221 ***	0.192 ***	0.213 ***
Number of village cadres	−0.017	−0.003	−0.003	−0.006
Monetary help	0.003	0.001	0.002	0.001
Education years of household members	0.016	0.003	0.006	0.006
Per health status	−0.312 *	−0.160 **	−0.170 **	−0.147 **
Per skill level	−0.157	0.059	−0.002	0.007
Family features
Family members	0.010	0.008	0.013	0.015 *
Labor force ability	−0.052	0.006	0.005	0.013
Geographical features
Distance to the road	−0.253	−0.257 **	−0.274 **	−0.271 **
Constant	0.491 *	0.490 ***	0.402 ***	0.309 **
R2	0.250 ***	−1.157 ***	−0.961 ***	−0.754 ***
Sample size	548	340	378	377

Note: *, **, *** indicate statistical significance at the level of 10%, 5%, and 1%, respectively.

, the regression interpretation of the impact of relocation factors on a households’ energy dependence is presented.

Model 9, total sample-based, revealed that PAR had a detrimental effect on the energy dependence of a household. The regression of the relocated households (Model 10) revealed that the voluntary relocation had a prominent impact on a households’ energy dependence. Furthermore, Model 11 showed that centralized resettlement reduced households’ energy dependence, and Model 12 indicated that relocation in the new stage reduced households’ energy dependence. Additionally, the area of woodland owned by households, sloping land conservancy program, and cooperatives participation had a significant effect on their ability to obtain energy from natural resources. Per health status and distance to the road played a positive role in reducing households’ energy dependence. Due to the limited channels for households to obtain energy resources from nature, variables such as their own assets, financial channels and skill level, family members, and labor force ability have no obvious influence on the energy dependence of households.

3.3.4. Effects of Relocation and Its Factors on Households’ Nature resources Dependence

Table 8. Analysis of the impact of relocation and its factors on the natural resources dependence of households.

Variables	Model 13	Model 14	Model 15	Model 16
Whether Relocated
Relocated-family	−0.241 ***
Relocation Feature
Relocation type		−0.154 ***
Settlement mode			−0.094 ***
Relocation time				−0.069 **
Livelihood Assets
Per farmland area	−0.000	−0.001	−0.002	−0.000
Per woodland area	0.002 **	0.003	0.002 **	0.002 **
Sloping land conservancy program	0.056 **	0.039	0.050 *	0.050 *
Per housing area	0.001 **	0.000	0.001 **	0.001 *
Per own assets	−0.038	−0.058 *	−0.059 *	−0.061 *
Financing channels	−0.068	−0.064	−0.061	−0.048
Cooperatives participation	0.138 **	0.247 ***	0.212 ***	0.224 ***
Number of village cadres	−0.008	0.003	0.003	0.001
Monetary help	0.004 **	0.006 **	0.006 ***	0.006 **
Education years of household members	−0.003	−0.007	−0.003	−0.003
Per health status	−0.053	−0.043	−0.068	−0.044
Per skill level	−0.037	0.039	−0.015	−0.006
Family features
Family members	−0.001	−0.000	−0.003	−0.006
Labor force ability	−0.014	0.046	0.065	0.067
Geographical features
Distance to the road	−0.143	−0.285 *	−0.324 **	−0.322 **
Constant	0.505 ***	0.519 ***	0.450 ***	0.395 **
R2	0.447 ***	0.264 ***	0.237 ***	0.217 ***
Sample size	548	340	378	377

Note: *, **, *** indicate statistical significance at the level of 10%, 5%, and 1%, respectively.

presents the consequences of a regression of the impact of relocation factors on households’ dependence on natural resources.

Model 13 revealed that PAR reduced households’ natural resources dependence. Model 14 further revealed that involuntary relocation encouraged household dependence on natural resources. Centralized resettlement, as seen in Model 15, reduced households’ natural resource dependence. Lastly, Model 16 indicated that relocation at the new stage reduced households’ natural resources dependence.

4. Discussion

PAR has striven to conserve natural resources and alleviate poverty, providing a viable method for reducing poverty and the effects of natural disasters. Our findings demonstrated that PAR could effectively reduce households’ natural resources dependence; the results also suggested that households with different relocation features had varying levels of natural resources dependence. Voluntary relocation, centralized settle, and new-stage relocation all helped to reduce the natural resources dependence of relocated households; this was corroborated by Li et al.’s [7] research. Among the relocation features, the relocation type had the most significant effect on the relocated households’ natural resources dependence, compared to their settlement mode and relocation time. It was observed that households which were relocated voluntarily had higher expectations and were more willing to adapt to the changes that the displacement entailed, i.e., to construct a new, resource-independent livelihood model [54]. Studies also have shown that relocated households’ dependence on ecosystem services was obviously lower than non-relocated households [7].

PAR has also assisted in diminishing the dependence of households on income, food, and energy. Through voluntary relocation, centralized settle, and new-stage relocation, the natural resources dependency index of relocated households was reduced. In comparison with income and food dependence, PAR had the most obvious inverse effect on a households’ energy dependence. Before relocation, households had to depend on local firewood [60]. On top of that, after relocation, infrastructure was enhanced, households’ livelihood activities grew more varied, agricultural production activities declined, and household income raised, allowing households to use clean energy in place of firewood consumption [45,61,62]. Pinpointing the factors that determine natural resources dependence will be crucial for long-term sustainable management [26]. Among the three relocation features, voluntary relocation had the most considerable influence on the income and energy dependence of relocated households, while centralized settle had the most critical impact on the food dependence of relocated households. Households with centralized settle had more significant market access and greater access to food and resources [7]. Households who willingly relocated had a more positive attitude and were more proactive in expanding their income sources [63].

Our research revealed that the closer a household was to a major road, the less dependent it was on natural resources. Furthermore, the health status of household members had some bearing on households’ natural resources dependence, with healthier households having more capacity and access to sources of energy other than firewood—as confirmed by Huang et al. [20,61]. Additionally, natural capital and physical capital were identified as significant factors impacting households’ dependence on natural resources. Nevertheless, the findings of Ren et al. [64] contradicted this conclusion, suggesting that a healthy labor force could earn income from natural resources and tended to rely on them. Additionally, research conducted by Duan et al. [19,61] revealed that the more forest land households possessed, the more firewood they were able to collect and the higher their energy dependence. Rural households with plenty of natural capital were also likely to engage in traditional agroforestry production activities; these could generate high livelihood risks and isolated income sources [55,64], further affecting the energy dependence of households. As for physical capital, self-owned assets significantly and negatively influenced the income dependence, food dependence, and natural resource dependence of relocated households.

5. Conclusions

According to the survey of poor households in the mountains of southern Shaanxi Province, this study compared and investigated different relocation features and evaluated the impact of PAR on households’ natural resources dependence from a micro perspective. The degree of households’ natural resources dependence differed in the study area; 18.97% of their income sources, 14.95% of their food sources, 17.74% of their energy consumption, and 17.22% of their ultimate natural resources. Our results showed that relocation and its features significantly negatively impacted a households’ natural resources dependence. Voluntary relocation reduced relocated households’ natural resources dependence. Centralized settle also reduced relocated households’ natural resources dependence. Furthermore, the new-stage relocation reduced relocated households’ natural resources dependence. The households which were voluntarily relocated preferred to actively respond to outside changes, seize new opportunities, expand livelihood channels, and complete the transformation to a non-farm livelihood mode faster. Centralized settle offered households the chance to enjoy more appropriate follow-up support, thus creating economic benefits through scale economy, policy spillover impacts to enhance their capacity to utilize external opportunities, and increased non-agricultural income. Measures and incentives for the current phase of the relocation policy significantly increased compared to previous stages. These findings could be extended to other contexts globally. The analysis of factors affecting rural households’ dependence on natural resources revealed that certain indicators of relocation features had a substantial impact on reliance on natural resources. Voluntary resettlement, centralized settle, and new-stage relocation significantly decreased the relocated households’ natural resources dependence. The distance to a major road, the health status of household members, farmland area, and their own assets also significantly impacted a households’ natural resources dependence.

The natural resources dependence of relocated households differed from that of non-relocated households. After relocation, the livelihood capital of households changed significantly, with a decrease in natural capital and human capital. Many households could not quickly adapt to the life and environment after relocation. It is necessary to increase investment in continuing education and training, constructing public facilities and employment assistance in relocated areas, and encouraging the relocated households to obtain more cultural services from their natural resources, such as eco-tourism and agricultural entertainment, rather than traditional farming and forest land. Moreover, promoting the equalization of regional public services, improving the efficiency of public service supply, realizing the effective allocation of resources in different regions, and alleviating the uneven income caused by the differences in geographical factors are crucial. In centralized resettlement mode, optimizing the efficiency of resource allocation at the same time, giving full play to the centralized resettlement under various resources agglomeration advantage, and improving households’ livelihood capital and accessibility to external resources at the same time help in focusing on low-income households. This will help to provide non-farm employment opportunities and skills training for the relocated households, assisting households in forming new income structures, achieving sustainable livelihoods after relocation, and reducing natural resources dependence. PAR could effectively solve the development plight, i.e., that natural resources have been unable to maintain the livelihood of local households. In this regard, PAR is a successful model that has solved the livelihood problems of households, to some extent, while reducing their dependence on natural resources and improving their well-being.

The theoretical significance of this study is as follows: First, to deepen the research on PAR policies, it is important to systematically evaluate and comprehensively consider the poverty reduction and ecological effects of PAR, and provide a feasible supplementary perspective for poverty-related issues. Second, a more effective and comprehensive evaluation index of households’ natural resources dependence is constructed, which provides a method to empirically test the impact of PAR on households’ natural resources dependence. Third, the innovative perspective of this study provides new ideas for alleviating natural resources dependence. The practical significance of this research is as follows: First, it is conducive to promoting the implementation of follow-up assistance for PAR, and promoting the effective connection between poverty alleviation and rural revitalization policies. Second, it is beneficial to systematically evaluate conservation and development projects and find improvement opportunities for how to alleviate resources dependence, so as to achieve the unity of the dual goals of conservation and development. Third, it is conducive to improving the pertinence and scientificity of the relevant policies and to provide information that is beneficial to the effective positioning of target groups for the protection and development projects in ecologically fragile areas, in order to serve the formulation and adjustment of relevant policies.

The study had a few limitations that should be taken into account. First, the economic indicators used to measure the relocated households’ natural resources dependence could have been limited, as they did not consider social and other factors. Second, using scientific methods to determine the weight of each index would be beneficial to efforts to obtain accurate data. Third, the survey was implemented in Ankang Prefecture, in southern Shaanxi Province, China, and the findings may not be applicable to other regions. As such, future change in household livelihood and natural resources dependence remain unknown. As for the direction of future research, we will improve the index system of households’ natural resources dependence in view of the deficiencies of this study, conduct follow-up investigation on the subsequent livelihood and natural resources dependence of relocated households, discover the changes of relocated households’ natural resources dependence, and improve the research results. To provide practical support and scientific and effective suggestions for the follow-up work and policy formulation of PAR.