Reassessing Resettlement-Associated Poverty Induced by Water Conservancy Projects in China: Case Study of the “Yangtze to Huai River Inter-Basin” Water Diversion Project

Abstract

The displacement and resettlement-associated poverty caused by water conservancy projects (WCP) is a worldwide issue. Re-settlers are often impoverished for extended periods due to loss, difficult re-establishment, and insufficient compensation. Addressing poverty has become a worldwide concern, and accurate measurements of poverty remain a fundamental issue. Before 2020, the Chinese government used the absolute income method to measure re-settler poverty. However, this method reflected neither the overall income gap nor potential benefits of social development and poverty alleviation policies. Therefore, we used the relative income and multidimensional methods alongside the absolute income poverty method to measure the poverty in recently resettled households. Based on survey data from over resettled 1000 households we conclude that: (1) The remaining poor measured by the absolute poverty line were mainly caused by serious diseases, disabilities and loss of labor ability, which means they have no ability to be lifted out of poverty except through the bottom line of local governments. As a result, the absolute poverty line loses its distinction to poverty. (2) Rural re-settlers were more resilient to forced majeure because land guarantees employment and food supply, allowing households to avoid secondary livelihood destruction. (3) Income derived measurement of re-settler poverty masks the benefits of poverty alleviation and other socioeconomic aid programs. A few households showed improvements in child school attendance, child mortality, nutrition, cooking fuel, asset ownership, and social insurance following resettlement. (4) To reduce the multidimensional gap, government aid programs should focus on years of schooling (including training), nutrition, household savings, and household labor force rather than simply providing monetary assistance. At the same time, we suggest that the government adopt a variety of compensation methods, such as: sharing the benefits of water conservancy projects, industrial support and improving the bottom line guarantee.

1. Introduction

The displacement and resettlement of re-settlers as a result of water conservancy projects (WCP) such as dam construction or inter-basin water transfer often lead to poverty. Because such projects affect large areas, numerous resettlement issues occur, such as those observed in association with the resettlement of 50,000 people when the Itaipu Dam, the world’s largest hydropower station was built [1], and the relocation of approximately 120,000 people as a result of Aswan High Dam construction [2]. China is home to many mega hydropower projects that have led to the relocation of many millions of people, including those forced to resettle due to the construction of the Three Gorges Dam and the South-to-North Water Diversion Project [3,4]. Displacement and resettlement have huge impacts on people [5], with issues such as loss of farmland severely affects the household incomes impacting the livelihood of the dependents [6]. Furthermore, the low compensation offered in many cases means that resettled people lack sufficient capital to build new houses, restore their capacity for production, and live at pre-displacement levels [7]. Therefore, they often endure long-term poverty as a result of resettlement [8].

The poverty of WCP-induced re-settlers is an urgent problem for the government. Since most resettlements generally occur due to water conservancy projects, the responsibility to mitigate and alleviate the poverty caused by such projects lies with the government [9]; poverty alleviation includes both the duration over which a government is to perform its responsibilities and the amount of compensation offered [10]. This is especially important because long-term impoverishment can lead to social instability [11] and environmental degradation [12]. To reduce WCP-induced poverty, the World Bank provides developing countries with special financial and technical assistance for relocation projects, allowing the formulation of better resettlement policies and post-resettlement action plans [13]. Through the World Bank’s consistent efforts, these developing countries have gradually formed their own local compensation standards and allocation procedures [14,15,16], helping resettled people to alleviate and eliminate poverty by restoring their livelihoods [17,18].

In China, the poverty of the WCP-induced re-settlers has gradually become a matter of concern for the central government. China’s early WCPs failed to address the poverty and livelihood restoration problems faced by the re-settlers. Until 1985, 60% of reservoir-induced re-settlers lived in poverty [19]. To deal with the large scale of poverty induced by WCP, the State Council of China issued the first “Report on Quickly Dealing with Reservoir Resettlement Issues” in 1986. Several revisions led to the production of two important documents in 2006: (1) State Council Decree No. 471 (2006) on the land acquisition and resettlement compensation rules associated with large- and medium-scale hydraulic and hydropower projects and (2) Suggestions of the State Council No. 17 (2006) on the improvement of follow-on support for people affected by large- and medium-scale reservoirs. These decrees are aimed at preventing WCP-induced poverty by providing pre-resettlement compensation, resettlement subsidies, and follow-up support [20]. The most influential factor responsible for improvements in China’s WCP poverty problem over the last seven years, Xi Jinping’s “Precise (Targted) Poverty Alleviation” campaign, was aimed at lifting 70 million Chinese people above the poverty line by 2020, and although WCP-induced re-settlers have benefitted from this program [21], the poverty ratio of these people remains higher than that of the general population, and most remain in abject poverty [22,23]. Most re-settlers are poor because their former homes were located in remote rural areas, and resettlement sites are generally established in under developed regions. Additionally, lower rates of education and reliance on basic farming skills means that many struggle to transition to other livelihoods, if their land is reduced or they become landless following resettlement [24,25].

Currently, the poverty line in China is measured using the absolute income poverty method. The poverty line was initially set at a net income of 2300 RMB per year in 2011 (equivalent to US$ 1/day), and the figure is adjusted yearly according to the consumer price index (CPI) in each province [26]. However, compared with the World Bank poverty line (US$ 1.9/day), China’s standards are relatively low [27]. In addition, measuring poverty from the perspective of absolute income alone cannot reflect the overall income gap or the potential benefits of any social development or poverty alleviation policies [28]. Xu et al. (2019) recommended that China adopt the relative income poverty method to accurately measure the poverty of WCP-induced re-settlers, as this method is useful for comparing the overall income gap among different resettlement groups [29]. Wang et al. (2021) argued that the multidimensional poverty method could identify the main factors causing poverty and reflect the effects of policies [30]. However, no research currently addresses the best way to measure China’s WCP poverty using both methods in tandem; no comprehensive analysis of WCP-associated poverty using multiple measurement models has as yet been performed. To comprehensively understand the current poverty levels of WCP-induced re-settlers and advise governments on their justification of compensation levels and assistance programs, multiple measurement models are required to evaluate the poverty of WCP-induced re-settlers.

In this study, the “Yangtze to Huai River inter-basin” water diversion project (YtoH Diversion) was considered as a case study and multiple poverty measurement methods were used to comprehensively analyze and interpret the poverty status and causes of poverty in China. The study contributes to the literature surrounding poverty alleviation for WCP-induced re-settlers in two ways. First, a multidimensional poverty framework suitable for China’s WCP-induced re-settlers was constructed based on the Global Multidimensional Poverty Index (GMPI), which was published by the United Nations Development Program (UNDP) and Oxford University. Second, this framework was used to dynamically interpret the poverty status of re-settlers in the YtoH Diversion project.

The remainder of this paper is organized as follows. Section 2, based on a brief review of the poverty measurement related literature, provides a framework for multidimensional poverty analysis. The characteristics of the case study region, research methods, and data collection are described in Section 3. The results are presented in Section 4. The main research findings, proposes specific countermeasures for poverty alleviation under China’s current WCP-induced resettlement system are discussed in Section 5, and conclusions drawn in Section 6.

2. Literature Review and Multidimensional Poverty Framework

2.1. Poverty Measurement in the Literature

The most commonly used method for identifying and measuring poverty, the monetary approach, defines poverty as consumption (or income) below a certain line [31]. According to Foster’s poverty theory, the poverty line can be divided into absolute and relative poverty, which reflect “subsistence” and “basic needs”, respectively [32]. Monetary approaches include the income poverty method, Engel coefficient method, Martin method, and extended linear expenditure system method (ELES) [33]. However, the WCP-induced resettlement study for China uses only the income poverty method, which requires income data for quantification [34,35], and as an increasing number of people relocate from remote rural areas to suburban and urban areas, their consumption structures and employment options change, often shifting them from income-based to consumption-based poverty [30]. Therefore, some Chinese resettlement scholars, such as Wang et al., proposed the use of the Engel coefficient, Martin, and ELES methods to measure poverty status [35]. These methods fully consider all dimensions of consumption and thus more accurately measure poverty from the perspective of demand.

Recent studies have shown that the income level of WCP-induced re-settlers in China is sufficiently high to meet their minimum or basic needs and maintain them at or above the absolute poverty line following resettlement. However, owing to higher consumption at resettlement sites and reduced livelihood capital, their poverty characteristics have been observed to gradually change from consumption- to development-based [36,37]. Therefore, some scholars, including Wang and Ke (2009), began to develop a multidimensional poverty framework to measure the poverty of re-settlers [35], which was largely based on the GMPI framework developed by the UNDP and Oxford University and includes three dimensions: education, health, and assets [38]. However, Chinese scholars argue that indicators such as electricity, and improved sanitation and drinking water supply are not applicable to China’s WCP-induced re-settlers. Wang et al. (2021) added four production indicators to the GMPI: farmland quantity, quality, stable employment, and labor skills [30], while Xu et al. (2019) added a psychological dimension that included two indicators: development prospects and the willingness to return [29]. Some scholars have stated that a security dimension also needs to be considered, such as social insurance or employment training [39]. Therefore, the current multidimensional poverty indicator system is considered insufficient. In addition, indicators such as farmland quality and development prospects are not easily measured in practice, and the willingness to return is particularly subjective. Therefore, it is necessary to build a more comprehensive and practical indicator system for use with WCP-induced resettlement.

2.2. Multidimensional Poverty Framework

Based on the GMPI and existing research, a more comprehensive and practical multidimensional poverty index system was developed that comprises four dimensions: education, health, assets, and development, with six indicators: years of schooling (YS), child school attendance (CSA), child mortality (CM), nutrition (NU), cooking fuel (CF), and asset ownership (AO), which includes household savings (HS), social insurance (SI), and the household labor force (HLF). These indicators are measured in household units.

2.2.1. Education Dimension

The education dimension includes two indicators: YS and CSA. According to the GMPI, if no household member has completed at least five years of education, a household is considered ‘deprived’ in terms of this dimension, while attendance deprivation is assumed if a school-age child is not attending school up to the age at which they would complete class 8 [40]. Since the per capita years of education is on average 7.7 years in rural China [41] and even lower for WCP-induced re-settlers, 7 years was taken as the threshold for YS in this study. At the same time, taking into account the 9 years of compulsory education and labor laws in China, school-age children between the ages of 7 and 16 years who do not attend school are considered deprived in terms of attendance.

2.2.2. Health Dimension

The health dimension includes two indicators: CM and NU. The GMPI considers the CM rate to be severe if a child has died within a family within the last five years, while nutritional deprivation is assumed if at least one family member is undernourished [40]. The interpretation of mortality defined by the GMPI was used in this study. In terms of NU, combined with the “Guidelines for the Prevention and Control of Overweight and Obesity in Chinese Adults” that was officially issued by mainland China, this article considers a BMI less than 18.5 as malnutrition [42].

2.2.3. Asset Dimension

The asset dimension included two indicators: CF and AO. As defined by the GMPI, residents that use dung, wood, or charcoal for cooking are considered deprived in terms of CF, while asset ownership is considered deprived if residents do not own more than one of the following: a radio, TV, telephone, bicycle, motorbike, or refrigerator, and do not own a car or truck [40]. In this study, the indicator illustration of CFl in the GMPI was used directly. In terms of AO, on the basis of the GMPI and in combination with the “Water Conservancy and Hydropower Project Resettlement Supervision and Evaluation Regulations” (SL716-2015) in China, commonly owned belongings such as a TV, refrigerator, washing machine, air conditioner, electric fan, water heater, rice cooker, pressure cooker, induction cooker, microwave oven, and telephone were considered assets and households that owned less than three were considered deprived.

2.2.4. Development Dimension

In addition to the three dimensions of the GMPI, an additional development dimension was added that included three indicators: HI, SI, and HLF. Family savings are an important financial guarantee for the subsequent development of WCP-induced re-settlers [43]. As the livelihood monitoring cycle for projects in China is six years, this article considers a resettled household deprived when its savings are less than six times that of the locals. SI provides anti-risk security for re-settlers [44]. Currently, rural re-settlers can purchase new rural social endowment and cooperative medical insurances, whereas re-settlers in cities can purchase work injury, maternity, endowment, medical, and unemployment insurance. This study considers SI to be deprived if individuals do not have insurance coverage. In addition, the HLF is the basis of subsistence [45]; thus, a household was considered deprived if its labor force was less than 50% of the household population.

3. Materials and Methods

3.1. Research Region and Sampling

The Anhui Province section of the “Yangtze-Huai River Inter-Basin” Water Diversion Project, from which the original residents were relocated to designated sites by 2016 (see Figure 1), was considered in this study. Fieldwork was conducted over four consecutive years (2017–2020), with the first survey in January 2017 considering resettlement baseline and resident data up to 2016. Surveys were given to the same households every December from 2017 to 2020. Of the 2745 relocated households, our sample comprised 1098, leading to a sampling rate of 40% (see

Table 1. Sample selection.

Area	Location	Relocated Households	Sample Households	Geographical Coordinates
Urban	Liangting Community	158	63	117.157 E, 31.745 N
	Liuhe Community	1269	508	117.282 E, 31.638 N
	Zipeng Community	758	303	117.025 E, 31.758 N
Rural	Binguang village	404	162	117.325 E, 31.549 N
	Wuhe village	113	45	117.304 E, 31.568 N
	Changzhen village	43	17	116.902 E, 31.849 N
Total		2745	1098

). A stratified sampling method was used to select the participants, with 15% earning a low income, 20% a relatively low income, 30% a middle income, 20% a relatively high income, and 15% a high income. The basic properties of the samples were as follows.

• All resettled people were of official rural household registration status prior to resettlement, and were mainly engaged in agricultural production. The local government adopted a mixed resettlement model, with 79.6% resettled in urban and 20.4% in rural sites. Rural resettled people were continuously engaged in agricultural production, while urban resettled people had to find jobs in non-farming sectors.
• The rural resettlement sites were approximately 15 to 20 km away from their original home villages, while urban resettlement sites were approximately 8 to 10 km away.

3.2. Poverty Measurement Model

The absolute and relative income poverty, and multidimensional poverty methods were used to comprehensively measure the poverty status of the resettled people. Absolute and relative income poverty were calculated using the Foster–Greer–Thorbecke (FGT) indices, and multidimensional poverty was calculated using the A–F model.

3.2.1. FGT Indices

The FGT indices are often used to analyze income poverty [46] and measure the poverty headcount ratio, poverty gap, and income inequality. The index is derived by substituting different values of parameter α into the following equation:

$${\mathrm{F}\mathrm{G}\mathrm{T}}_{\mathsf{\alpha }}=\frac{1}{\mathrm{N}}\sum _{\mathrm{i}=1}^{\mathrm{H}}{\left(\frac{\mathrm{z}-{\mathrm{y}}_{\mathrm{i}}}{\mathrm{z}}\right)}^{\mathsf{\alpha }}$$

(1)

where $\mathrm{z}$ is the poverty line, $\mathrm{N}$ is the number of people comprising an economy, H is the number of people in poverty (those with income at or below z), and $y_i$ is the income of each individual, $i$. The formula reduces to the poverty headcount ratio (PHR) when $\alpha =0$, poverty gap index (PGI) when $\alpha =1$, and squared poverty gap index (SPGI) when $\alpha =2$. Among these, the PHR and PGI are the most commonly used poverty indicators [47]. To facilitate comparison with the calculated multidimensional poverty result, only the PHR and PGI were calculated in this study.

The Lorentz curve equation proposed by Villasenor and Arnold was used to calculate the FGT indices [48]. This equation is expressed by the following:

$$L\left(1-L\right)=a\left(P^2-L\right)+bL\left(P-1\right)+c\left(P-L\right)$$

(2)

where L is the cumulative share of income earned and P is the cumulative share of people or households from lowest to highest income (all poverty indexes in this paper are based on households). Parameters a, b, c are estimated while e, m, n, and r are obtained using the equations: $e=-(a+b+c+1)$, $m=b^2-4a, n=2be-4c, r=\sqrt{n^2-4m{e}^2}$. After obtaining the quadratic Lorenz curve, the corresponding FGT indices are calculated using the formula:

$$L\left(P\right)=-\frac{1}{2}\left(bP+e+\sqrt{m{P}^2+nP+e^2}\right)$$

(3)

Parameters $s_1$ and $s_2$ are calculated using $s_1=(r-n)/2mand{s}_2=-(r+n)/2m$, repsectively. PHR and PGI were calculated using the following formulas:

$$PHR=-\frac{1}{2m}(n+r\frac{(b+2z/\mu )}{\sqrt{{(b+2z/\mu )}^2-m}})$$

(4)

$$PGI=H-\left(\frac{\mu }{z}\right)L\left(H\right)$$

(5)

where $\mu$ is per capita net income and $z$ is the poverty line. The absolute poverty headcount ratio (APHR) and absolute poverty gap index (APGI) can be obtained when z denotes the absolute income poverty line. Subsequently, the relative poverty headcount ratio (RPHR) and relative poverty gap (RPGI) can be obtained when z denotes the relative income poverty line.

3.2.2. A–F Model

The A-F method, which has the advantages of being highly intuitive and suitable for policy analysis, was adopted for the assessment of multidimensional poverty [49].

Suppose there are n individuals in an economy. The poverty status of individual $i$ is measured using $m$ indicators. The value of the individual $i$ for each indicator $j$ is expressed as $g_{ij}$, with $g_{ij}=1$ if indicator $j$ of individual $i$ is deprived, and $g_{ij}=0$ otherwise. Setting the weight of indicator $j$ to $w_j$(${0<w}_j<1,{\sum }_{j=1}^m{w}_j=1$), the weighted score of individuals $i$ on all $m$ indicators can be represented by $c_i$, and $c_i={\sum }_{j=1}^m{w}_j{g}_{ij}$. The critical value $k$ ($0<k\le 1$) is used to compare the degree to which individual $i$ is deprived under $m$ indicators to determine their multidimensional poverty status, with individual $i$ regarded to be in multidimensional poverty if $c_i\ge k$. No threshold is assumed when $k=0$, otherwise, a threshold is included. Based on this method, the multidimensional poverty index (MPI) is obtained using:

$$MPI=\frac{1}{n}\sum _{i=1}^n{c}_i\left(k\right)$$

(6)

If $m=1,$ and $k\ne 0,$ the formula indicates the censored single factor poverty index (CSFPI):

$$CSFPI=\frac{1}{n}\sum _{i=1}^n{g}_i·I\left(c_i\ge k\right)$$

(7)

where, $I\left(·\right)$ is a threshold function for which the value is 1 when $c_i\ge k$ and 0 if $c_i<k$. In this study, the value of k was set to 0.3.

The MPI can also be divided into the multidimensional poverty headcount ratio (MPHR) and the multidimensional poverty gap index (MPGI).

$$MPI=\frac{q}{n}\times \frac{1}{q}\sum _{i=1}^n{c}_i\left(k\right)=MPHR\times MPGI$$

(8)

where q is the number of people that have fallen into multidimensional poverty.

The ratio CSFPI/MPHR can be used to measure the impact of a certain indicator on multidimensional poverty, which we define as the single factor impact index (SFII):

$$SFII=\frac{CSFPI}{MPHR}$$

(9)

For a certain indicator $i$, the larger the SFII, the greater its impact on multidimensional poverty.

3.3. Data Processing

3.3.1. Data Used in Parameter Estimation for the Lorenz Curve

In this study, k-means cluster analysis was used to divide the per capita net income of the sample households into ten levels from low to high and the cumulative share of people at all levels was calculated as an independent variable. The cumulative share of the net income earned was obtained simultaneously as the dependent variable. Based on the above process, 10 sample points (Pi, Li) (i = 1, 2, …, 10) were obtained. The Lorenz curve was obtained using Equation (2) for parameter estimation. In this study, SPSS 19 was used for cluster analysis and nonlinear regression.

3.3.2. Poverty Line and per Capita Net Income

When using FGT indices to measure the absolute and relative income poverty, the poverty line $z$ and per capita net income $\mu$ are required. The absolute income poverty line ${\mathrm{z}}_1$, which was set at 2300 yuan by China in 2011 [50], was adjusted according to the CPI of Anhui Province. The relative income poverty line ${\mathrm{z}}_2$ accounts for 30% of the median per capita household income of re-settlers and value μ is obtained from statistical analysis of the sampled data, as shown in

Table 2. Poverty line and per capita net income from 2016 to 2020 (unit: RMB).

Year	${\mathbf{z}}_1$		${\mathbf{z}}_2$		$\mathsf{\mu }$
	Urban	Rural	Urban	Rural	Urban	Rural
2016	3004	2961	3987	3713	8643	7930
2018	3104	3056	4242	4097	8386	8471
2020	3267	3224	4137	4059	8437	8202

.

3.3.3. Weight of Multidimensional Poverty Indicators

The indicators and definitions used in this study are presented in

Table 3. Definition and weight of multidimensional poverty indicators.

Dimension	Indicator	g = 1 When Occurred, Otherwise g = 0	Weight
Education	Years of Schooling (YS)	No household member has completed 7 years of education, i.e., graduated primary school.	1/8
	Child School Attendance(CSA)	School-age children between the ages of 7 and 16 do not go to school.	1/8
Health	Child Mortality (CM)	A child has died in the family within the last 5 years.	1/8
	Nutrition (NU)	At least one family member under the age of 60 has a BMI less than 18.5.	1/8
Asset	Cooking Fuel (CF)	Uses dung, wood, or charcoal for cooking.	1/8
	Assets Ownership (AO)	Own less than three of the following assets: TV, refrigerator, washing machine, air conditioner, electric fan, water heater, rice cooker, pressure cooker, induction cooker, microwave oven, and telephone.	1/8
Development	Household Saving (HS)	Household savings of re-settlers is less than 6 times that of the locals.	1/12
	Social Insurance (SI)	Not covered by any insurance.	1/12
	Household Labour Force (HLF)	Household labour force is less than 50% of the family population.	1/12

. The weight of each was calculated using an equal-proportion weighting method.

4. Result

By comparing the calculation results for absolute income, relative income, and multidimensional poverty, the poverty status of resettled people induced by the “Yangtze to Huai River inter-basin” water diversion project, a contemporary WCP, was comprehensively measured. The parameter estimation results and curves obtained using the Lorentz curve equation is presented in

Table A1. Lorenz curve regression results.

Area	Year	Parameter	B	SE	95% Confidence Interval		${\mathbf{R}}^2$
					Lower Bound	Upper Bound
Urban	2016	a	0.949	0.012	0.926	0.972	0.999
		b	−1.71	0.034	−1.776	−1.644
		c	0.101	0.019	0.064	0.138
	2018	a	0.851	0.019	0.813	0.888	0.999
		b	−1.545	0.043	−1.629	−1.461
		c	0.158	0.022	0.115	0.201
	2020	a	0.884	0.015	0.855	0.913	0.999
		b	−1.547	0.049	−1.643	−1.451
		c	0.12	0.023	0.075	0.165
Rural	2016	a	0.859	0.014	0.832	0.886	0.999
		b	−1.575	0.023	−1.619	−1.53
		c	0.139	0.018	0.104	0.174
	2018	a	0.809	0.01	0.789	0.828	0.999
		b	−1.416	0.029	−1.473	−1.359
		c	0.195	0.019	0.158	0.232
	2020	a	0.952	0.009	0.934	0.969	0.999
		b	−1.725	0.023	−1.769	−1.68
		c	0.078	0.012	0.054	0.101

and Figure A1 in Appendix A. The calculation results for the MPI and CSPI are presented in

Table A2. MPI of urban and rural re-settlers.

Area	Year	MPI	MH (%)	MPG (%)
Urban	2016	0.027	8.34	32.92
	2018	0.026	8.10	33.21
	2020	0.028	8.44	33.55
Rural	2016	0.054	13.34	40.67
	2018	0.040	10.36	38.72
	2020	0.034	9.16	37.59

and

Table A3. CSFPI of urban and rural re-settlers.

Index	Urban			Rural
	2016	2018	2020	2016	2018	2020
YS	5.95	4.59	3.30	8.54	7.54	7.04
CSA	2.00	3.10	4.33	3.52	3.52	4.52
CM	0.06	0.00	0.00	0.25	0.00	0.00
NU	6.72	6.46	3.75	9.55	8.29	7.04
CF	0.00	0.00	0.00	4.77	2.76	2.01
AO	1.10	0.65	0.58	7.29	5.53	3.77
HS	6.79	7.18	8.02	2.76	1.76	1.01
SI	5.04	4.52	4.98	11.81	10.05	9.05
HLF	6.59	6.92	8.34	9.30	9.05	7.29

, respectively, of Appendix A.

4.1. Absolute Income Poverty

Figure 2a indicates that the APHR of urban and rural resettled people declined slightly between 2016 and 2018, indicating that, on average, the income of resettled people measured by the APHR was slightly improved, with poverty alleviation rates of 0.13% and 0.29%, respectively. However, from 2018 to 2020, the APHR indicators for both urban and rural resettlement increased, indicating a return to poverty with rates of 4.59% and 2.60%, respectively. In general, the APHR of both resettled and rural households increased over the period from 2016 (rural versus urban: 2.26% versus 1.76%) to 2020 (rural versus urban: 4.57% versus 6.22%). A similar trend can also be observed in Figure 2b, which shows that the APGI of urban re-settlers was essentially the same as that of rural re-settlers from 2016 to 2018; however, the gap widened in 2020, when the APGI of urban resettled people increased rapidly.

4.2. Relative Income Poverty

It is apparent from Figure 3a that the RPHR of the urban resettled increased between 2016 and 2020, indicating that the overall income gap expanded during this period, while the rural resettled show an inverted “V” shape that indicates a widening overall income gap between 2016 to 2018 followed by a narrowing from 2018 to 2020. In general, the RPHR of both resettled and rural households increased from 2016 (rural vs. urban: 8.17% vs. 6.94%) to 2020 (rural vs. urban: 9.57% vs. 13.18%). An identical pattern for the RPGI can be observed in Figure 3b.

4.3. Multidimensional Poverty

Figure 4a indicates that the MPHR of urban resettled areas from 2016 to 2020 remained stable, with only slight fluctuations between 8.10% and 8.44%; however, the MPHR of their rural counterparts show a consistent decline from 13.34% in 2016 to 9.16% in 2020. These results indicate that the urban resettled population experienced a relatively stable MPHR, while the rural resettled population experienced a rapid reduction in MPHR. The multidimensional poverty of the resettled population thus generally improved. Similar trends can also be seen in Figure 4b, in which the MPGI is used to measure the poverty gap.

4.4. SFII Calculation

To identify the main factors that influence the poverty of the resettled population, the SFII was used to determine which of the key factors contribute more to multidimensional poverty. As shown in Figure 5a, YS, NU, and AO show a downward trend in urban resettled areas from 2016 to 2020, while CSA, HS, and HLF show an upward trend, and SI shows a fluctuating trend. By 2020, HS and HLF had a much greater impact on the multidimensional poverty of the urban resettled population, with SFIIs of 95% and 99%, respectively. It is therefore apparent that more attention should be devoted to HS and HLF if we are to further improve the multidimensional poverty of urban resettled populations in the future.

Figure 5b presents the SFII for rural resettlement between 2016 and 2020. CF, AO, and HS show decreasing trends, while YS, CSA, and SI show increasing trends, and NU and HLF show fluctuating trends. By 2020, YS, NU, SI, and HLF had the greatest impacts on multidimensional poverty in rural resettled areas, with SFII values of 76%, 76%, 98%, and 80%, respectively. Future alleviation efforts should therefore focus on YS, NU, SI, and HLF for these populations.

5. Discussion

Three different poverty indices were used together with the SFII to measure the poverty trends and status of the resettlement-case population. The main factors contributing to poverty within the resettled population were identified, and the consistency between the calculation results and the actual situation was discussed. The main conclusions were obtained by comparing the calculation results for absolute income poverty, relative income poverty, and multidimensional poverty.

5.1. Absolute Income Poverty Analysis

According to the current poverty line in China, ${\mathrm{z}}_1$, the APHR index shows an overall positive result for the first two years following resettlement, with a decline in the APHR for both urban and rural re-settlers. The values of APHR obtained in the study area (1.63% for urban resettled and 1.97% for rural resettled) were slightly better than the 2.2% obtained for Anhui Province in 2018 [51]. Our survey data show that poverty in these households is largely caused by serious diseases, disabilities, and loss of labor ability [52]. Alternatively, the remaining impoverished population has no capacity to be lifted above the poverty line via the bottom-line guarantee of the local government.

However, the APHR of both urban and rural resettlements increased between 2018 and 2020, with that of urban resettlement increasing faster than that of rural resettlement. Such a high rate of re-impoverishment indicated that the resettled remained vulnerable to poverty and that the urban resettled are more sensitive to this problem. One of the main explanations for this phenomenon is the impact of the China-United States trade war, which has greatly affected employment in China’s secondary and tertiary industries [53,54]. Despite the decline in agricultural sales within resettled rural areas during the trade war, the pressure of unemployment and lower income could be partially mitigated by having land as a basic guarantee. However, urban re-settlers suffer more from unemployment, resulting in a higher rate of return on poverty than those who have been resettled within the same timeframe.

5.2. Relative Income Poverty Analysis

The RPHR is influenced by the overall income gap. According to the relative income poverty line ${\mathrm{z}}_2$, the overall income gap of the urban resettled population expanded between 2016 and 2020, whereas that of the rural resettled population expanded between 2016 and 2018 and shrank between 2018 and 2020. These trends are consistent with the change in the Lorentz curve in Figure A1; that is, the Lorentz curve of urban resettlement constantly deviated from the line of equality from 2016 to 2020, while that describing rural resettled deviated between 2016 and 2018, and approached between 2018 and 2020.

The income gap of the urban resettled expanded from 2016 to 2020, which is consistent with China’s macro Gini coefficient [55]. Before 2018, the development of internet platforms such as Meituan, Alipay, JD.com, and Didi Taxi provided a large number of jobs for the tertiary industry [56], which greatly increased the income of urban re-settlers. However, the income of those engaged in secondary industries remained largely unchanged. Therefore, there was a notable increase in the income gap. However, the China-United States trade war has meant that the employment rate and income of both secondary and tertiary industries declined to varying degrees in 2019, which led to a lower increase in the income gap [57]. The income gap in the rural resettled populations also expanded from 2016 to 2018. Similar to the urban re-settlers, the main reason for this is that rural re-settlers who were engaged in tertiary industries also benefitted from the internet platforms. However, for the same reason, 68.53% of the rural resettled were forced to change their production from secondary and tertiary industries to agriculture [52]. Because the agricultural income gap is small and the income of secondary and tertiary industries has declined, the income gap of the rural resettled population narrowed between 2018 and 2020.

5.3. Multidimensional Poverty Analysis

The MPHR of the urban resettled remained stable from 2016 to 2020 (ranging from 8.10% to 8.44%); however, the MPHR of the rural resettled decreased substantially from 13.34% in 2016 to 9.16% in 2020. This indicates that multidimensional poverty has generally improved in the resettled population. Compared with the nationwide MPHR of 4% estimated for 2014 by Shen et al. [40], it is apparent that the deprivation associated with WCP-induced resettlement is more serious. At the same time, the much lower and more stable MPHR of the urban resettled compared to their rural counterparts indicates that the urban resettled, with official urban registration, are entitled to more material and social resources than their rural counterparts, which is consistent with the research findings of Shangguan et al. [39].

HS have been identified as one factor that contributes to the MPHR for urban resettlement. Relocating to urban areas means higher cash expenditure. The replacement of coal or wood with natural gas and the higher electricity consumption increase the cost of living. Other costs associated with transportation and improving the quality of life in urban areas further increases consumption. The most important extra cost is urban housing, which necessitates savings urban houses cost more than the resettlement compensation [58]. YS, NU, and SI have also been identified as important factors contributing to the MPHR of the rural resettled. The severe deprivation in terms of YS and NU is largely due to the relatively poor educational and material resources in rural areas [59]. The deprivation of social security occurs because rural resettlement focuses largely on short-term benefits. Since China’s new rural social endowment insurance requires that insurance premiums are paid until the age of 60 to receive a pension, the resettled are more inclined to use money for their near-term living expenses. This phenomenon is consistent with the research results of Banerjee [60]. The HLF has a greater impact on the MPHR of both urban and rural re-settlers and is the basis of the migrant’s livelihood, regardless of location.

6. Conclusions

This study used multiple poverty measurement methods to reassess and dynamically interpret the poverty status of China’s WCP-induced re-settlers. The following conclusions were obtained: (1) Absolute poverty analysis indicates that China’s current absolute poverty standards are out of date for WCP-induced re-settlers because poverty is not eliminated through the bottom-line guarantees of local government. The current absolute poverty line does not sufficiently represent the different experiences and needs of the resettled poor. (2) Through relative poverty analysis, we found that rural re-settlers are more resilient to force majeure, as witnessed during the recent pandemic. The guarantee of employment and food supply through land ownership allows re-settlers to avoid the secondary destruction of their livelihoods. (3) Comparison of the results for income poverty (both absolute and relative) with those of multidimensional poverty indicates that worsening income poverty is universal for the resettled, whereas multidimensional poverty has generally improved. Therefore, measuring poverty in terms of income alone masks the potential benefits of mitigation processes such as social development programs and poverty alleviation policies. Table A2 clearly shows that CSA, CM, NU, CF, AO, and SI for both urban and rural resettled children have all improved to varying degrees following resettlement, which is mainly due to better access to the relevant public, material, and information resources [61]. However, the higher MPHR for urban and rural resettlement indicates that multidimensional poverty could still be improved. (4) Comparison of the APGI, RPGI, and MPGI indicates a comparatively small gap between absolute and relative income poverty, whereas the gap associated with multidimensional poverty is much larger. Therefore, reducing the multidimensional poverty gap should be the focus of poverty alleviation in the later stages. Accordingly, analysis of the different included factors indicates that the YS, NU, HS, SI, and HLF are most important and should be targets for future poverty alleviation efforts. In addition, in order to improve the livelihood resilience and resist secondary disasters caused by force majeure, a stable source of income for re-settlers is also necessary. To this end, we suggest that the government adopt a variety of compensation methods, such as: sharing the benefits of water conservancy projects, industrial support and improving the bottom line guarantee. Therefore, subsequent studies should consider how to reduce multidimensional poverty of the re-settlers through the above compensation methods. At the same time, more indicators can be included in the measurement of multidimensional poverty to better reflect poverty status.