Linking Natural Resource Dependence to Sustainable Household Wellbeing: A Case Study in Western China

Abstract

To reduce the threat of natural disasters, protect natural resources, and promote human wellbeing, Shaanxi Province, China has implemented the largest disaster resettlement project since 2011. It has moved 2.4 million people in three prefectures over 10 years. Using survey data from southern Shaanxi, China, this article measures sustainable household wellbeing (SHWB) and natural resource dependence in the context of disaster resettlement. It explores the differences in SHWB and natural resource dependence across different groups and relocation characteristics. To specifically analyze the effects of natural resource dependence on SHWB, ordinary least square (OLS) regression models were used to investigate their relationship. The results show that natural resource dependence shows significant positive correlation with SHWB. Meanwhile, the food dependence, energy dependence, and income dependence of relocated households show significant positive correlation with SHWB. Additionally, the SHWB of relocated households, voluntary relocation households, and centralized relocation households is significantly lower than local households, involuntary relocation households, and scattered relocation households. These findings have broader implications for rural communities in China and other developing countries, and are crucial for rural households to change the heavy dependence on natural resources and improve their wellbeing.

1. Introduction

The frequent occurrence of natural disasters brings vexing dilemmas and challenges to people in many ways, such as material, economic, and in terms of their living environment [1]. Although the displacement and relocation of residents should be the last resort [2], a growing number of resettlement projects have been launched [3,4]. Resettlement has already been an important approach for disaster prevention and risk reduction in countries around the world [5]. The Chinese government has completed the disaster resettlement project of millions of households, including Shaanxi Province, which in 2011 launched the largest resettlement project in China. The disaster resettlement project is a national development project led by the local government to address resource shortages and low wellbeing caused by frequent natural disasters [5,6]. Existing studies have analyzed the ecological and demographic aspects of disaster resettlement [7,8,9], but there is a lack of empirical studies discussing disaster resettlement aimed at reducing dependence on natural resources, and promoting their wellbeing. It is stated that disaster resettlement alleviates multidimensional poverty or achieves diversification of livelihood strategies to attain the goal of improving households’ wellbeing [10,11]. Similarly, disaster resettlement is conducive to improving households’ livelihood resilience or adaptive capacity [11,12], and the enhancement of resilience and adaptive capacity obviously improves sustainable household wellbeing. Scholars tend to analyze disaster resettlement from the perspective of livelihoods [12,13,14], while this article pays more attention to the impact of households’ natural resource dependence on sustainable household wellbeing in the case of disaster resettlement.

With the development of the theory of ‘natural capital’ [15], the relationship between natural resources and human activities has received widespread attention. The various products provided with natural resources and their functions in ecology and society have been widely recognized by scholars [16,17]. Natural resources significantly affect the improvement of human wellbeing [18]. Therefore, how to balance natural resources and human wellbeing is important. This means that the measurement of household income from natural resources should be key to scientific research and policy making [19]. Natural resource dependence is a commonly used indicator, that is, the proportion of household income derived from natural resources [20]. Previous studies have analyzed the reasons for the differences in household natural resource dependence, such as the labor force number, geographical location, educational background, and amount of cultivated land and forest land [21,22,23]. Additionally, there are extensive studies on measuring natural resource dependence, optimizing the evaluation system of natural resource dependence, or exploring the relationship between natural resource dependence and household livelihoods [24,25,26,27]. However, it should also be noted that excessive dependence on natural resources has an impact on the protection of the environment and the sustainability of resources [28]. Some scholars have proposed that environmental regulation can be used from a macro perspective to curb people’s uncontrolled use of natural resources [29,30]. While the research on how to reduce dependence on natural resources at the micro levels needs to be further supplemented, this article analyzes the dependence of natural resources from the perspective of rural households in China.

Understanding the relationship between natural resources and human wellbeing is the premise underlying the coordination of humans and nature and is crucial to solving the phenomenon of the ‘tragedy of the Commons’ [19]. However, research on this relationship lacks quantitative analyses and empirical examples. To clarify how natural resources affect human wellbeing, we need to understand wellbeing more accurately. In this article, wellbeing is defined as a measure of people’s perception of a good quality of life, and a state of human health, happiness, and prosperity. Human wellbeing is a comprehensive concept, and it is important to analyze its changes and establish measurement indicators [31]. There is much literature showing that wellbeing is multi-dimensional, and the indicators selected in different studies are not the same [32,33,34,35]. For example, Summers et al. [36] considered two aspects when selecting indicators of wellbeing: subjective and objective. Similarly, there is a large amount of literature examining subjective and objective wellbeing in different research contexts [36,37]. In practice, many indicators, methods, and frameworks have been used to study the wellbeing of rural households [38,39,40]. In recent years, influenced by the Millennium Ecosystem Assessment (MA) [41] and the 2030 Global Agenda for Sustainable Development, scholars have realized that human wellbeing is closely linked to ecosystems, and that improving household wellbeing and protecting natural resources are equally important. For instance, Peng et al. [42] developed the index of sustainable household wellbeing (SHWB), which measures household wellbeing and integrates sustainable household livelihoods. Not only is there innovation in the wellbeing index, scholars have also conducted extensive research on the relationship between ecosystem services (ES) and human wellbeing [37,39,40]. Using participatory mapping, Sarah et al. [43] analyzed the significant contribution of ES to farmers’ wellbeing in four semi-arid regions of West Africa. Li et al. [40] provided a new perspective for the study of ES and household wellbeing by constructing a coupling model in rural China. Extensive research is proving that natural resources and human wellbeing should be harmonized with the goal of sustainable development. In this context, clarifying the impact of natural resources on human wellbeing is critical to addressing the issue of rural household wellbeing derived from high natural resource dependence. Based on this, this paper proposes the following research hypothesis:

H1: The high dependence on natural resources significantly affects the improvement in SHWB.

H1.1: The high dependence of food on natural resources significantly affects the improvement in SHWB.

H1.2: The high dependence of energy on natural resources significantly affects the improvement in SHWB.

H1.3: The high dependence of income on natural resources significantly affects the improvement in SHWB.

Poverty eradication, health and wellbeing, and terrestrial ecology are some of the goals of the Sustainable Development Goals (SDGs) and the deep hopes of China. Through disaster resettlement, the Chinese government is trying to address the urgent problems of natural resource shortage and poor wellbeing of residents in some areas, achieving the win–win goals of resource protection and wellbeing enhancement [44]. Many studies have assessed the impact of disaster resettlement on sustainable livelihoods [11,12], and a few studies have attempted to explore the effect of disaster resettlement on the ES and natural resource dependence of rural households [9,25]. For example, Li et al. [40] found that relocation projects can reduce farmers’ dependence on ES. They are also conducive to the optimization of farmers’ income structure. The resettlement characteristics such as centralized resettlement and new stages of relocation also play a positive role in reducing farmers’ dependence on ES. Furthermore, Liu et al. [25] suggested that disaster resettlement could significantly reduce household natural resource dependence and help protect the natural environment. These studies illuminated subsequent research through scientific theory and appropriate modeling. This paper explores the impact of rural households’ natural resource dependence on SHWB, which not only complements disaster-resettlement-related studies at the household level but also enriches the studies on the relationship between natural resource dependence and wellbeing. The structure of the paper is arranged as follows: First, we constructed the SHWB index system from five dimensions with reference to MA and used the method of Peng et al. [42]: basic material needs for good life, i.e., the material basis to improve the quality of life; freedom of choice and action, i.e., households have their own or external advantages to face some choices and actions; health, i.e., the state of health of family members; social relation, i.e., the relationships’ network that households can draw upon when they need help; and security refers to whether the food, water, and shelter can meet the needs of the household. Then, we measured natural resource dependence from three aspects: the food dependence, energy dependence, and income dependence. Next, natural resource dependence and SHWB of different types and samples of rural households were compared, respectively. Finally, OLS regression was used to study the impact of natural resource dependence on SHWB in different samples and by introducing different relocation characteristics.

2. Materials and Methods

2.1. Study Area

The area studied in this paper was Ankang prefecture (Figure 1), situated in the south of Shaanxi Province, which is bounded by the Han River and divided into two regions: the northern Qinling Mountains and the southern Daba Mountains. Its landform presents the characteristics of the north and south mountains and the middle basins. In the land area, the proportion of mountain area is the largest, accounting for about 92.5% [9]. Over the years, there have been many natural disasters, which have not only caused huge economic losses, but also affected local economic and social development. In addition, the livelihoods of households are mainly dependent on agriculture and forestry. However, due to the limitation of cultivated land area and the forestry protection policy, the economic burden of households is heavy. As a result, Ankang prefecture struggles with the dual dilemmas of ecological fragility and typical poverty.

As the water source of China’s South-to-North Water Diversion project, Ankang prefecture needs to transport clean water to North China, so it undertakes the tasks of water source protection, soil conservation, and ecological construction [9,11]. Therefore, relevant departments have set up ecological functional areas, and widely implemented measures such as returning farmland to forest, building nature reserves, etc. This action has brought challenges to local economic development. To eliminate the threat of natural disasters, and fundamentally improve the wellbeing of households, the Shaanxi government launched a large-scale disaster resettlement project in May 2011.

2.2. Data Collection

Our data came from the rural household survey conducted by the research group of the Population and Development Institute of Xi’an Jiaotong University in Ankang. The investigators were composed of teachers, graduate students, and undergraduates from Xi’an Jiaotong University, Northwest University, and Xi’an University of Finance and Economics. This survey adopted convenience sampling for some administrative villages or resettlement communities, and randomly selected household members between the ages of 18 and 65 for interviews. Based on the representativeness and feasibility of sample selection, we finally selected 3 large-scale resettlement communities in Ziyang County and 8 villages in Hanbin District and Ningshan County. These areas not only participate in disaster resettlement projects, but there is an urgent need to reconcile the use of natural resources with the wellbeing of households. In the above resettlement communities, most households are relocated households, which makes the proportion of relocated households in the questionnaire larger than local households. Also, we found that small differences exist among local households in the survey, so local households in the questionnaire are representative. In the end, 670 questionnaires were issued, of which 657 were valid, including 459 for relocated households and 198 for local households.

From the perspective of household livelihood, the questionnaire mainly included the following four parts: (1) basic information of households, including household size, education level, dependency ratio, per capita income, non-agricultural income, etc.; (2) household livelihood capital, including natural, financial, social, material, and human aspects and involving more indicators; (3) natural resource dependence, which is reflected in the three aspects of food, income, and energy; and (4) information related to relocation, i.e., whether or not a household has been relocated, the relocation type, the relocation nature, etc. By categorizing different households and then proposing targeted improvement measures, it is easier to achieve the overall goal [45]. With reference to existing studies [12,40,46], this paper divided households into relocated households and non-relocated households (i.e., local households), divided the relocation nature into voluntary relocation and involuntary relocation, and divided the relocation type into centralized relocation and scattered resettlement. In addition, livelihood strategies were classified into pure farming type, non-farming type, and diversified livelihood type [47]. To improve the quality of field research, the research team was composed of experienced teachers and students. After the questionnaire was completed, we recorded it into a data set and carried out data processing.

2.3. Indicator Construction

2.3.1. Sustainable Household Wellbeing (SHWB)

Based on the research of Peng et al. [42] and the Millennium Ecosystem Assessment [41], we divided sustainable household wellbeing into five dimensions: (1) basic material needs for a good life; (2) freedom of choice and action; (3) health; (4) social relation; and (5) security. We then selected a series of indicators for each dimension according to previous literature (

Table 1. The index system of SHWB.

Variables	Indicator	Indicator Description	Weight
Basic material needs for a good life	Per capita net income	The actual amount of per capita net income (CNY)	0.166
	The number of durable goods	The number of durable goods owned by household (pieces)	0.098
	Non-agricultural income	The amount of household income excluding agricultural income (CNY)	0.087
Freedom of choice and action	Income diversity index	Diversity level of a rural household’s income (%)	0.078
	Distance from roads	Distance to the main road of the address (m)	0.073
	Train	Family members receive training (0 for no; 1 for yes)	0.071
Health	Medical expenses	Medical expenses as a proportion of total consumption (%)	0.068
	Self-rated health of members	Poor health, fair health, or good health	0.067
Social relation	Number of cadres	Number of cadres among relatives (persons)	0.060
	Get help	Number of households available to provide help	0.055
	Number of cooperatives	Number of types of cooperatives involved (types)	0.054
Security	Housing security	The housing area (m2)	0.052
	Water security	Whether has tap water (0 for no; 1 for yes)	0.043
	Food security	Crop yield per unit of cultivated land in the survey year (kg)	0.028

Note: 1 USD = 6.2284 CNY in 2015; Source: Authors’ survey from Ankang prefecture, Shaanxi Province, China.

).

For the rural households studied in this article, the dimension of basic material needs for a good life was mainly composed of per capita net income, the number of durable goods, and non-agricultural income. Income as an indicator is often used to measure household wellbeing [40]. We used per capita net income and non-agricultural income to explore the economic level of a household and the income brought with non-agricultural activities. The higher the income mentioned above, the higher the wellbeing of households. Moreover, household durable goods, such as furniture and electrical appliances, are also typical indicators to measure the quality of household life and the more durable goods a household has, the higher their quality of life [12]. The dimension of freedom of choice and action mainly indicates whether households can successfully realize non-agricultural livelihood transformation and have convenient transportation conditions [48]. In fact, disaster resettlement will partially change the situation of households based on agricultural income, giving households more freedom of choice and action [49]. Therefore, we adopted three indicators: income diversity index, distance to the main roads, and training. Income diversity index is defined with the diversity level of a rural household’s income. The higher the income diversity index, the higher the benefit to the SHWB. The closer the distance to the main roads, the more conducive to household travel, and households with more training have better employment choices than those without training [11,12,49]. The dimension of health was characterized by medical expenses and the self-rated health of members. Health affects the wellbeing of the household. High medical costs mean that family members are in poor health, and self-rated health is a subjective evaluation of their own health level [46]. Harmonious social relation mainly involves the breadth and validity of household social relation networks. The dimension of social relation included number of village cadres, monetary help, and cooperatives’ participation. If a household has village officials, access to monetary help, and participation of cooperatives, the household can reduce its risk in the face of external shocks [49]. The dimension of security refers to households that provide good shelter and sufficient food and water, so we used housing security, food security, and water security as representations [50].

2.3.2. Households’ Natural Resource Dependence

Natural resource dependence refers to the proportion of total household income derived from natural resources. Mukul et al. [27] found that natural resources can help families improve their economic level. Moreover, Rafael et al. [51] also argued that the difference in natural resource dependence is one of the causes of household income inequality. This paper divides natural resource dependence into three parts: food dependence, energy dependence, and income dependence (

Table 2. Households’ Natural Resource Dependence.

Variables	Description	Mean	Standard Deviation
Total dependence (%) (Natural resource dependence)	(Food dependence + energy dependence + income dependence)/3	0.133	0.163
Food dependence (%)	The proportion of self-sufficient food income to the family’s annual total food expenditure	0.053	0.156
Energy dependence (%)	The proportion of the firewood collection amount to the family’s annual energy consumption expenditure	0.177	0.280
Income dependence (%)	The proportion of the income from agroforestry and livestock to the family’s total income	0.167	0.253

).

Food dependence is defined as the portion of a household’s food source provided with natural resources, measured as the proportion of self-sufficiency of food expenditure in total household annual food consumption. Energy dependence refers to the direct consumption of natural resources in household energy consumption, as measured with the amount of firewood collected as a percentage of the total annual household energy consumption. Income dependence refers to income derived from the use of natural resources, measured as a share of total household income from agriculture, forestry, and livestock. Natural resource dependence is the average of the above three parts [52]. The lower the natural resource dependence index, the lower the rural household’s dependence on natural resources.

2.4. Econometric Model

2.4.1. Wellbeing Evaluation Model

We used a principal component analysis to calculate the SHWB index [42]. Before the principal component analysis, we used the range standardization to process the original data. In this way, the data of different units can be compared, and the result is in the range of 0–1. The calculation method is as follows:

Positive indicators:

$$X_{ij}=\frac{X_{ij}-min\left\{X_{1j},\dots ,X_{nj}\right\}}{max\left\{X_{1j},\dots ,X_{nj}\right\}-min\left\{X_{1j},\dots ,X_{nj}\right\}}$$

(1)

Negative indicators:

$$X_{ij}=\frac{max\left\{X_{1j},\dots ,X_{nj}\right\}-X_{ij}}{max\left\{X_{1j},\dots ,X_{nj}\right\}-min\left\{X_{1j},\dots ,X_{nj}\right\}}$$

(2)

where X_ij is the initial value of each indicator; it is the value of the jth indicator of the ith sample (i = 1, 2,…, n).

A principal component analysis makes the determination of indictor weights of different dimensions more reasonable and scientific. In this paper, a total of 14 indictors were analyzed with principal components (

Table 3. Principal component analysis results.

Factor	Eigenvalue	Proportion	Cumulative	Factor	Eigenvalue	Proportion	Cumulative
Per capita net income	2.327	0.166	0.166	Self-rated health of members	0.934	0.067	0.708
The number of durable goods	1.367	0.098	0.264	Number of cadres	0.842	0.060	0.768
Non-agricultural income	1.221	0.087	0.351	Get help	0.772	0.055	0.823
Income diversity index	1.091	0.078	0.429	Number of cooperatives	0.752	0.054	0.877
Distance from roads	1.024	0.073	0.502	Housing security	0.732	0.052	0.929
Train	0.997	0.071	0.573	Water security	0.595	0.043	0.972
Medical expenses	0.952	0.068	0.641	Food security	0.395	0.028	1.000

), from which 5 principal components were extracted. Finally, the value of SHWB was calculated as follows:

$$SHWB=w_1{F}_1+w_2{F}_2+w_3{F}_3+w_4{F}_4+w_5{F}_5$$

(3)

where w_1–5 is the score of the principal component; F_1–5 is the weight of the principal component.

2.4.2. Regression Analysis Model

In order to assess the impact of households’ natural resource dependence on sustainable household wellbeing, a multiple linear regression model was constructed, and the ordinary least squares (OLS) method was used. The formula is as follows:

$$Y={\beta }_0+{\beta }_1{X}_1+{\beta }_2{X}_2+\cdots +{\beta }_n{X}_n+\mu$$

(4)

where β₀ is the constant term, β₁, β₂…βn are the coefficients of regression, and µ is the random error term.

According to the existing literature [7,11,12], the regression in this paper included two parts: First, to study the impact of natural resource dependence on SHWB using different sample sizes. We divided the sample into relocated households, local households, and total sample. Second, whether a household was relocated, the nature of the relocation, and the relocation type were introduced into the whole sample to analyze the impact of natural resource dependence on SHWB under different relocation characteristics.

Table 4. Variables used in the regression analysis.

Variables	Description	Mean	Standard Deviation	Min	Max
Whether relocated	Relocated household = 1; otherwise = 0	0.699	0.459	0	1
Relocation nature	Voluntary relocation = 1; involuntary relocation = 0	0.840	0.363	0	1
Relocation type	Centralized resettlement = 1; Scattered resettlement = 0	0.758	0.429	0	1
Education level	Average years of education for household members	6.180	2.654	0	13.5
Household size	Number of household members (persons)	4.496	1.608	1	9
Dependence ratio	The proportion of children and elderly to the total workforce (%)	0.277	0.225	0	1
Phone charge	Household members last-month phone charge (CNY)	228.122	357.361	0	5000
Social support	The amount of money available for help (CNY)	61.072	185.960	0	4542
Experience	Types of experiences for household members (1. Village cadres; 2. Technician, teacher, or doctors; 3. Enterprise employees; 4. Soldiers; and 5. No experience above)	0.498	0.841	0	5
Loan	The possibility of loan (Definitely can = 1; More likely = 2; Generally = 3; Less likely = 4; and Certainly cannot = 5)	2.570	1.348	1	5

Note: 1 USD = 6.2284 CNY in 2015.

lists all the variables involved in the regression analysis.

Relocation characteristics included whether relocated, relocation nature, and relocation type. For the whether relocated characteristic, we divided the rural households into relocated households and local households. Among the valid questionnaires we collected, 459 households relocated, accounting for 69.863% of all households. The relocation nature was divided into voluntary relocation and involuntary relocation, of which a total of 395 rural households were voluntarily relocated, accounting for 86.057% of the relocated households. The households engaging in voluntary relocation may be driven by poverty, disaster avoidance, and severe ecological damage to the original site. Most of the households in this article relocated to avoid the threat posed by disasters. Zhou et al. [53] found that natural disasters and the intensification of damage increased residents’ desire to relocate. Relocation types were divided into centralized resettlement and scattered resettlement. Centralized resettlement areas are residential communities led by the government that implement centralized management of disaster-evacuated migrants [8]. Scattered resettlement is when households are placed into smaller departments [54]. The education level was the human capital; the higher the education level, the more job opportunities, and thus the higher the SHWB. The average years of education for sample households was 6.18. The average household size was 4.496, and the maximum was 9 people. Household size affects the consumption of natural resources by households [52]. The dependence ratio refers to the proportion of children and elderly to the total workforce. If the value of the dependency ratio is large, it will increase the economic burden of the family. Both phone charge and social support represent a household’s social capital. When households face external risks, social capital can help them ease the difficulties. In the questionnaire design, experience refers to 5 options (1. Village cadres; 2. Technician, teacher, or doctors; 3. Enterprise employees; 4. Soldiers; and 5. No experience above). Existing literature [11,12,25] has found that experience 1–4 is beneficial to family development, while most low-income families have no experience above. Finally, a loan is financial capital, which was mainly represented by the possibility of a loan.

3. Results and Analysis

3.1. Comparing the Natural Resource Dependence of Different Types of Households

Table 5. Natural Resource Dependence of different types of households.

	Whether Relocated			Relocation Nature			Relocation Type
Indices	Yes	No	t-Test	Voluntary	Involuntary	t-Test	Centralized	Scattered	t-Test
Total dependence	0.087	0.238	12.089 ***	0.075	0.161	5.082 ***	0.071	0.141	4.992 ***
Income dependence	0.131	0.252	5.758 ***	0.120	0.196	2.672 ***	0.124	0.152	1.185
Food dependence	0.032	0.101	5.259 ***	0.032	0.036	0.235	0.023	0.062	2.842 ***
Energy dependence	0.097	0.363	12.365 ***	0.072	0.252	6.091 ***	0.065	0.208	5.902 ***

Note: *** indicates that the t values are significant at the 1% significance level.

shows the natural resource dependence of different types of households. We found that the total dependence of the relocated households (0.087) was significantly lower than that of the local households (0.238). Among them, the income dependence, food dependence, and energy dependence of the relocated households were also significantly lower than those of the local households. Also, the total dependence of voluntary relocation households was lower than that of involuntary relocation households. The natural resource dependence of households that centralized resettlement was lower than those that scattered resettlement, including total dependence, food dependence, and energy dependence. This is because relocated households, especially those that voluntarily relocated and centralized resettlement, not only have more advantages in government and community support, but also need to actively adapt to the new community life and gradually carry out non-agricultural activities. In addition, relocation also means that rural households in the original location of the agriculture, forestry, and stockbreeding are affected, or even unable to continue.

3.2. Comparing the SHWB of Different Types of Households

As shown in

Table 6. Comparison of the level of SHWB under different relocation characteristics.

	Whether Relocated			Relocation Nature			Relocation Type
Indices	Yes	No	t-Test	Voluntary	Involuntary	t-Test	Centralized	Scattered	t-Test
Overall SHWB	0.280	0.308	5.387 ***	0.277	0.307	3.789 ***	0.274	0.304	4.858 ***
Income and physical needs	0.041	0.045	2.478 **	0.040	0.044	1.692 *	0.039	0.046	3.500 ***
Freedom of choice and action	0.107	0.127	6.140 ***	0.105	0.128	4.911 ***	0.103	0.123	5.196 ***
Health	0.081	0.076	−1.625	0.081	0.083	0.356	0.080	0.086	1.464
Social relations	0.041	0.044	1.601	0.042	0.037	−1.748 *	0.041	0.040	−0.510
Security	0.010	0.016	6.838 ***	0.009	0.015	5.245 ***	0.010	0.010	−0.336

Note: ***, **, and * indicate that the t values are significant at the 1%, 5%, and 10% significance levels, respectively.

, the overall SHWB of relocated households (0.280) was significantly lower than that of local households (0.308). Among them, the income and physical needs, freedom of choice, and action and security of relocated households were significantly lower than those of local households. The overall SHWB of voluntarily relocated households (0.277) was significantly lower than that of involuntarily relocated households (0.307). Among them, the income and physical needs, freedom of choice and action, and security of voluntarily relocated households were significantly lower than those of involuntarily relocated households. However, the social relations dimension associated with voluntary relocation was higher than that associated with involuntary relocation. At the same time, the overall SHWB of those that underwent centralized resettlement (0.274) was significantly lower than that of households that experienced scattered resettlement (0.304), which is mainly reflected in the two dimensions of income and physical needs and freedom of choice and action. As a positive intervention against the vicious cycle of natural disasters, disaster resettlement reduces the external risk to relocated households. The reason for the lower wellbeing of relocated households is that their living environment has undergone great changes, so they will go through a long period of economic reconstruction.

In addition, Figure 2 shows the density distribution of SHWB under different livelihood strategies. As can be seen from Figure 2, there are slight differences in the density distribution of the three livelihood types, and the curves are similar in shape. However, we also see that the curve for diversified livelihood type is relatively flat, meaning that the wellbeing of households in this type fluctuated more significantly. At the same time, the curves of pure farming type and non-farming type are steeper, meaning that the wellbeing of these households underwent tiny fluctuations.

3.3. Analysis of Natural Resource Dependence and SHWB

To further analyze the dependence on natural resources, a K−means cluster analysis was used to divide households into high dependence, medium dependence, and low dependence categories. The results show that the F statistic was 2274.55 and the significance level was 0.000, indicating a significant difference between the three groups. Further, the low dependence was 0.034, accounting for 65.906%, which was the largest among the three groups. In addition, medium dependence (0.249) and high dependence (0.500) accounted for 24.049% and 10.046%, respectively. The box plot is drawn with the three groups of natural resource dependence as the horizontal coordinate and the SHWB as the vertical coordinate (Figure 3a). As shown in Figure 3a, the medians of low and medium dependence are located in the center of the box, with equal distances between the upper/lower quartiles, indicating that the wellbeing level was relatively uniform. However, the median wellbeing of high dependence was inclined to the lower quartile, and the distribution was more skewed.

In order to discuss the relationship between natural resource dependence and SHWB, the first fitting graph analysis, the second fitting graph analysis, and the kernel density regression fitting graph analysis were used. The results of graph fitting are shown in Figure 3b. Natural resource dependence had an overall upward effect on sustainable household wellbeing, and this effect gradually strengthened with the increase in natural resource dependence. The kernel density regression fitting results show that the marginal effect of natural resource dependence on SHWB fluctuated to a certain extent after reaching a certain critical value.

3.4. The Influence of Natural Resource Dependence on SHWB

Taking the natural resource dependence (total dependence) and its three parts (food dependence, energy dependence, and income dependence) as the explanatory variable, and SHWB as the response variable, the influence of natural resource dependence on SHWB was analyzed. As shown in

Table 7. The impact of Natural Resource Dependence on SHWB in different samples.

Variables	Local Households				Relocated Households				Total Sample
	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6	Model 7	Model 8	Model 9	Model 10	Model 11	Model 12
Total dependence	0.170 *				0.352 ***				0.337 ***
Food dependence		0.117				0.157 **				0.192 ***
Energy dependence			0.009				0.200 ***				0.156 ***
Income dependence				0.126 **				0.117 ***				0.159 ***
Education level	0.022 ***	0.021 ***	0.021 ***	0.022 ***	0.016 ***	0.017 ***	0.016 ***	0.016 ***	0.020 ***	0.022 ***	0.021 ***	0.021 ***
Household size	0.023 **	0.022 *	0.024 **	0.022 *	−0.003	−0.001	−0.002	−0.002	0.006	0.006	0.007	0.005
Dependence ratio	0.000	0.003	0.018	0.012	−0.005	−0.009	−0.004	−0.010	−0.002	0.007	0.003	0.010
Phone charge	0.000	0.000	0.000	0.000	0.000 *	0.000	0.000 *	0.000	0.000	0.000	0.000	0.000
Social support	−0.000	−0.000	−0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000
Experience	0.038 **	0.037	0.034	0.033	0.007	0.004	0.003	0.008	0.019 *	0.016	0.017 *	0.018 *
Loan	−0.031	−0.030 **	−0.032 **	−0.031 **	−0.034 ***	−0.035 ***	−0.032 ***	−0.036 ***	−0.036 ***	−0.038 ***	−0.037 ***	−0.039 ***
Constant	−0.129	−0.089	−0.071	−0.118	−0.050	−0.026	−0.045	−0.027	−0.092 **	−0.058	−0.078 *	−0.068
R2	0.171	0.166	0.158	0.179	0.196	0.152	0.194	0.158	0.222	0.182	0.200	0.195
N	193	193	193	193	450	450	450	450	643	643	643	643

Note: ***, **, and * indicate that the t values are significant at the 1%, 5%, and 10% significance levels, respectively.

and

Table 8. The impact of Natural Resource Dependence on SHWB in different relocation characteristics.

Variables	Model 13	Model 14	Model 15	Model 16	Model 17	Model 18	Model 19	Model 20	Model 21	Model 22	Model 23	Model 24
Total dependence	0.250 ***				0.293 ***				0.327 ***
Food dependence		0.133 **				0.152 **				0.133 *
Energy dependence			0.094 ***				0.162 ***				0.185 ***
Income dependence				0.120 ***				0.092 **				0.110 ***
Whether relocated
Relocated households	−0.075 ***	−0.104 ***	−0.088 ***	−0.098 ***
Relocation nature
Voluntary relocation					−0.101 ***	−0.124 ***	−0.099 ***	−0.118 ***
Relocation type
Centralized resettlement									−0.036 *	−0.053 **	−0.034 *	−0.054 ***
Education level	0.018 ***	0.018 ***	0.018 ***	0.018 ***	0.015 ***	0.016 ***	0.015 ***	0.015 ***	0.016 ***	0.016 ***	0.016 ***	0.016 ***
Household size	0.007	0.007	0.008	0.006	−0.004	−0.003	−0.003	−0.003	−0.003	−0.001	−0.002	−0.002
Dependence ratio	−0.010	−0.008	−0.007	−0.005	0.000	−0.000	0.001	−0.003	−0.001	−0.003	−0.000	−0.004
Phone charge	0.000	0.000	0.000	0.000	0.000 *	0.000	0.000	0.000	0.000 *	0.000	0.000 *	0.000
Social support	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000
Experience	0.018 *	0.016	0.016	0.017 *	0.007	0.005	0.003	0.007	0.006	0.004	0.003	0.007
Loan	−0.034 ***	−0.035 ***	−0.035 ***	−0.036 ***	−0.036 ***	−0.037 ***	−0.035 ***	−0.038 ***	−0.033 ***	−0.034 ***	−0.032 ***	−0.035 ***
Constant	−0.020	0.028	0.004	0.016	0.059	0.101	0.061	0.096 *	−0.023	0.012	−0.019	0.011
R2	0.239	0.221	0.224	0.230	0.226	0.183	0.223	0.200	0.202	0.164	0.199	0.171
N	643	643	643	643	450	450	450	450	450	450	450	450

Note: ***, **, and * indicate that the t values are significant at the 1%, 5%, and 10% significance levels, respectively. “Whether relocated” takes local households as the reference group; “Relocation type” takes scattered resettlement as the reference group; and “Relocation nature” takes involuntary relocation as the reference group.

, models 1−12 divided all surveyed households into local households, relocated households, and total sample to deeply explore the impact of natural resource dependence on SHWB. At the same time, models 13−24 introduced the characteristics of whether relocated, relocation nature, and relocation type to further analyze the role played by relocation characteristics.

As shown in Table 7, model 1 shows that at the 10% level, total dependence was positively correlated with SHWB. In models 2 and 3, food dependence and energy dependence of local households were positively correlated with SHWB, respectively, but the positive correlation changes were not statistically significant. Model 4 shows that income dependence was positively correlated with SHWB at the 5% level. For relocated households, total dependence, food dependence, energy dependence, and income dependence all had significant positive correlations with SHWB, which are specifically shown in models 5–8. This result shows that relocated households can improve their wellbeing by making full use of natural resources and increasing their dependence on natural resources in terms of food, energy, and income. There are great similarities between the total sample and the relocated households. Models 9−12 show that at the 1% level, total dependence, food dependence, energy dependence, and income dependence of the total sample were significantly positively correlated with SHWB.

As shown in Table 8, relocation characteristics were introduced to further analyze the role of relocation on the impact of natural resource dependence on SHWB. The results show that relocated households and voluntary relocation were negatively correlated with SHWB at the level of 1%. Meanwhile, centralized resettlement also showed a significant negative correlation. Among the control variables, the education level was significantly positively correlated with SHWB, that is, the higher the average years of education of family members, the higher the SHWB. A loan has a significant negative effect on SHWB. A large loan means a heavy financial burden on the household, resulting in a low SHWB. In addition, experience was significantly positively correlated with SHWB in models 13 and 16. Telephone charges were significantly positively correlated with SHWB in models 17, 21, and 23.

To test robustness, we took SHWB as an ordered category, introduced control variables in Table 8, and used the Ordered Logit model for a regression analysis (

Table 9. (Robustness test) The impact of Natural Resource Dependence on SHWB in different relocation characteristics.

Variables	Model 13	Model 14	Model 15	Model 16	Model 17	Model 18	Model 19	Model 20	Model 21	Model 22	Model 23	Model 24
Total dependence	2.396 ***				2.799 ***				3.094 ***
Food dependence		1.160 **				1.401 *				1.138
Energy dependence			1.055 ***				1.459 ***				1.626 ***
Income dependence				1.128 ***				1.013 **				1.239 ***
Whether relocated
Relocated households	−0.614 ***	−0.888 ***	−0.701 ***	−0.836 ***
Relocation nature
Voluntary relocation					−0.931 ***	−1.163 ***	−0.964 ***	−1.058 ***
Relocation type
Centralized resettlement									−0.381 *	−0.548 ***	−0.386 *	−0.554 ***
Education level	0.150 ***	0.146 ***	0.148 ***	0.148 ***	0.147 ***	0.148 ***	0.147 ***	0.149 ***	0.148 ***	0.150 ***	0.149 ***	0.148 ***
Household size	0.051	0.053	0.064	0.048	−0.007	−0.001	0.004	−0.005	0.004	0.016	0.017	0.006
Dependence ratio	−0.006	−0.019	0.013	0.027	−0.008	−0.046	0.007	−0.033	−0.045	−0.108	−0.024	−0.078
Phone charge	0.000 *	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000 *	0.000
Social support	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000
Experience	0.106	0.073	0.082	0.094	0.033	0.010	0.001	0.034	0.028	0.002	−0.009	0.035
Loan	−0.330 ***	−0.332 ***	−0.328 ***	−0.339 ***	−0.369 ***	−0.383 ***	−0.367 ***	−0.388 ***	−0.338 ***	−0.345 ***	−0.334 ***	−0.354 ***
R2	0.021	0.019	0.020	0.020	0.019	0.017	0.019	0.017	0.018	0.014	0.017	0.016
N	643	643	643	643	450	450	450	450	450	450	450	450

Note: ***, **, and * indicate that the t values are significant at the 1%, 5%, and 10% significance levels, respectively. “Whether relocated” takes local households as the reference group; “Relocation type” takes scattered resettlement as the reference group; and “Relocation nature” takes involuntary relocation as the reference group.

) [55,56]. The results show that the coefficients and significance of the OLS model and the Ordered Logit model tend to be consistent, so the robustness of the research results can be proved.

4. Discussion

Disaster resettlement is the key strategy to improve wellbeing and to respond to natural disasters, as well as a solution to improving the livelihoods of local families [7]. Based on the disaster resettlement project of Ankang prefecture in southern Shaanxi province, China, this paper focuses on the impact of natural resource dependence on SHWB, which is also the core thought of this paper. This paper shows that the natural resource dependence of relocated households is significantly positively correlated with SHWB. Also, the food dependence, energy dependence, and income dependence of relocated households show significant positive correlation with SHWB. The higher the natural resource dependence, the higher the SHWB of the relocated households. This finding is similar to previous research affirming the benefits of natural resources to households’ wellbeing [17,27]. In our survey, respondents identified fuelwood, crops, and livestock as key factors in improving human wellbeing. Although households engage in non-farming activities after relocation, there is limited support for career development in resettlement communities to enable relocated households to achieve full employment and effectively improve household wellbeing [12]. Therefore, greater access to natural resources remains one of the reasons for the high wellbeing of households. This conclusion is supported by many scholars. Liu et al. [57] proposed that due to the low income of rural households, fuelwood contributes significantly to their wellbeing by effectively reducing their cooking and heating costs. The research has also confirmed that the reduction in the food supply and the ecological function of natural resources will significantly reduce income and health [58,59], which are important factors affecting SHWB.

Additionally, the results show that natural resource dependence varied between relocated households and local households. According to the field investigation, we found that relocated households were removed from the original natural environment on which they depended, which inevitably had an impact on the way they use natural resources. This is consistent with previous research showing that disaster resettlement plays a significant role in reducing household dependence on natural resources, alleviating ecological degradation, and changing household livelihood patterns [9,11,25]. Our results also show that the SHWB of relocated households, voluntarily relocated households, and households that underwent centralized relocation was significantly lower than that of local households, involuntarily relocated households, and households that underwent scattered relocation. This is because disaster resettlement brings many problems and challenges to the improvement in household wellbeing [7,12,46,49]. This is supported by Chen et al. [6] and Galarza-Villamar et al. [60], who found that post-disaster resettlement made it difficult for relocated residents to rebuild their livelihoods and increased their exposure to new risks compared with local households.

Relocation characteristics have significant effects on SHWB. The results show that relocated households, voluntary relocation, and centralized resettlement all have significant negative correlation with SHWB. This finding is supported with previous research showing that relocation increases the vulnerability of rural households and decreases household wellbeing [12,61]. Relocation will not only generate huge resettlement costs and increase the economic burden of relocated households [11,62] but also bring the destruction of natural and social capital, such as the reduction in arable land and forest land, or the separation from relatives and neighbors [9,12]. A few studies have shown that voluntary resettlement improves human wellbeing [63], but this study proves that voluntary resettlement in Ankang prefecture does have a negative impact on SHWB. Although voluntary relocation of households causes them to be more active in adapting to their new lives, their SHWB is lower due to low levels of education, narrow social networks, and difficulties in obtaining financial loans [64,65,66]. It is undeniable that centralized resettlement has improved the housing conditions and infrastructure of households [7]; however, after the end of the one-time relocation subsidy, some households still cannot avoid the risk of poverty [63]. This agrees with Lo et al. [62], who found that once the farmers moved to the centralized resettlement community, the government considered its mission had been accomplished and ignored the need for follow-up support [12,67].

There are clear limitations to our study. First, this study used sectional data and quantitative surveys. Therefore, if this research could be combined with qualitative analysis methods, it would be more comprehensive and specific. To address the problem of narrow data, we are considering a dynamic study of relocated households through a larger follow-up survey. Second, measurements of SHWB and natural resource dependence apply mainly to rural areas. It may not be possible to explain the effects of natural resource dependence on SHWB in other contexts. Especially in developed countries (for example, where firewood is not a major source of energy acquisition), it may take a completely different form. Finally, our research object was the rural area of southern Shaanxi, China, without considering the disaster resettlement situation in other parts of Shaanxi, which has certain regional limitations.

5. Conclusions

This paper focuses on the impact of rural household natural resource dependence on SHWB in the context of disaster resettlement. We seek to provide experience and evidence for research on natural resource dependence reduction and wellbeing enhancement at the household level in developing countries. Methodologically, this paper adopts the t-test, and introduces a density map, box plot, fitting plot, OLS regression analysis, and other methods to explore the effect of natural resource dependence on SHWB. The study shows that the natural resource dependence has a significant positive correlation with SHWB. Meanwhile, the food dependence, energy dependence, and income dependence of relocated households show significant positive correlation with SHWB. In addition, the natural resource dependence of voluntarily relocated households and those undergoing centralized relocation is lower than that of involuntarily relocated households and those undergoing scattered relocation. Meanwhile, the SHWB of relocated households, voluntarily relocated households, and households undergoing centralized relocation is significantly lower than local households, involuntarily relocated households, and those undergoing scattered relocation. These findings have broader implications for rural resettlement efforts in China and other developing countries, and are crucial for changing the heavy dependence on natural resources and enhancing the wellbeing of rural households.

To address the problem of high wellbeing generated with the high dependence on natural resources, policy makers and stakeholders need to work together. The government should give priority to the coordination of resource protection and wellbeing enhancement when formulating policies. It can help relocated households improve their lives by providing job information, conducting technical training, and expanding financing channels [7,12]. It is vital that relocated households achieve sustainable livelihoods, which also requires that relocated households actively adapt to the new living environment and accumulate livelihood capital to cope with various risks [12]. Although the results of this study may vary according to regional differences, it is worth affirming that improving rural household wellbeing is a long-term process [68]. We have tried to draw more attention to natural resource dependence and SHWB, and hope that more theoretical and empirical research will complement this topic in the future. Given the importance of reducing dependence on natural resources and improving the SHWB of rural households, scholars must continue to consider these related issues.