The contributions of livelihood resilience to reshaping adaptive capacity in rural China

More quantitative evidence is necessary on the link between livelihood resilience and livelihood adaptive capacity (LAC) in disaster resettlement. This study used 459 field research data collected from Ankang Prefecture, southern Shaanxi, China, examining how livelihood resilience influences adaptive capacity in the context of disaster-induced relocation. The resilience of rural household livelihood systems is described in terms of two components, general resilience, and specific resilience, which are quantified using the space vector method from systems engineering. The awareness, ability, and action framework is used to measure the LAC of rural households, and quantile regression is applied to explore the impact of livelihood resilience on LAC. Guided by the Sustainable Livelihoods Approach (SLA) and awareness, ability, action framework, we differentiate between cognitive, resource-based, and behavioral dimensions of adaptive capacity. The space vector method further reveals that individual adaptive capacity is reinforced by community-level resilience. It is found that: 1. livelihood resilience has a significant positive effect on high LAC levels, with the strongest effects observed at lower quantiles; as livelihood resilience increases, LAC also increases significantly. For rural households with low LAC levels, the impact is not significant; 2. general resilience and education have significant positive effects on all levels of LAC, with high levels being the most affected; 3. specific resilience has a significant negative effect on the lowest level of LAC only, and no significant effect on other levels. This study deepens our understanding of the relationship between livelihood resilience and LAC in the context of disaster resettlement, while testing the relationship between the two provides a methodological contribution to the study of disaster resettlement and community development.

1 Introduction

As a strategy to cope with natural disasters and related risks, the Chinese government has implemented large-scale disaster-related resettlement programs in recent years. In 2020, China completed the resettlement of nearly 10 million people. In Ankang Prefecture, Shaanxi Province alone, more one million people have been relocated since 2011. While many rural households have successfully moved out of impoverished and high-risk areas, they now face a range of challenges in their new settlements, including uneven resource endowments, disruption of social networks, and shifts in livelihood strategies. As a key social intervention, disaster resettlement is typically linked to development objectives such as poverty alleviation, ecological restoration, disaster risk reduction, and infrastructure construction (Rogers and Xue, 2015; Tran and Vu, 2020). However, most existing research emphasizes on development-oriented resettlement and resettlement induced by development, with relatively few studies examining the social and ecological effects specific to disaster-related resettlement (Chen et al., 2018; Lo and Wang, 2018; Deng et al., 2020). Disaster resettlement aims to mitigate the negative economic, social and livelihood experienced faced by people forced to relocate due to major natural disasters (Ma et al., 2024; Xu et al., 2022). Although scholars have analyzed resettlement from the dimensions of livelihood adaptive capacity (LAC) and population change, there has not yet been an operational, practical pathway detailing how to alleviate its adverse effects and support the reconstruction of livelihoods.

Notably, livelihoods—an important indicator of resettlement—are often neglected in existing disaster-resettlement research, yet they provide a crucial means of exploring the impacts of disaster resettlement (Liu et al., 2020b). Recently, some researchers tried to analyze this process from the micro level. For example, Zhao et al. (2024) found that the resettlement plan that avoids disaster risk may weaken the LAC of households through empirical research in Ankang, Shaanxi Province, which reveals that policy intervention may have unintended consequences. The limitations of policy intervention not only show the economic, but also profoundly affect the social and cultural structure. Boyacioglu et al. (2023)’s reveal that top-down spatial planning ignores local buildings and community identity, which will lead to identity fracture and weaken household’s sense of belonging and social cohesion in Behram Kalay in Turkey. This warns us that disaster resettlement policies must incorporate local cultural identity and social capital into core considerations while ensuring material livelihoods. While exploring the depth of disaster resettlement research, these studies further highlight the important value of putting livelihoods at the center of analysis, and provide direction for formulating more operational follow-up support policies. Therefore, based on the Sustainable Livelihoods Approach (SLA), this study conducts an empirical investigation in typical post-disaster resettlement communities in Ankang, Shaanxi Province, aiming to unpack how the multidimensional structure of livelihood resilience influences household adaptive capacity and to reveal its stage-dependent evolutionary mechanisms. The findings not only contribute to a deeper theoretical understanding of the “resilience–adaptive capacity” nexus, but also provide evidence-based support for designing phased and differentiated policies to enhance community resilience.

In recent decades, the theory of resilience has been deepened in the field of sustainable livelihood research. Among its concepts, livelihood resilience has attracted substantial attention as a central construct for understanding household’s capacity to maintain and enhance development in the face of disturbances. It generally denotes the ability of rural households not only to endure shocks but also to rebound to their pre-crisis state or even achieve improvement through adaptive actions when confronted with various crises. Tahiru et al. (2019) emphasized that focusing analyses on livelihood resilience effectively places people at the core of development, thereby addressing the fundamental question of who resilience serves. Improving livelihood resilience can help households cope with all kinds of sudden shocks and systemic uncertainty more calmly. However, effectively measuring livelihood resilience has long posed a challenge, and the academic community has therefore proposed a variety of conceptual frameworks and empirical methods. For example, the assessment framework developed by Speranza et al. (2014) is often used to analyze resilience mechanisms in livelihood systems; simultaneously, a series of quantitative methods based on index systems have also been widely used in specific situations such as climate change, post-disaster relocation and natural disasters (Sina et al., 2019; Dovie and Pabi, 2023). Most existing studies tend to rely on quantifiable proxy variables to characterize livelihood resilience, and construct corresponding evaluative indicators. However, Quandt (2018) argued that many theoretically coherent frameworks at the theoretical level often fail to achieve the desired results in practice, and these methods and indicators are frequently criticized. In response to these methodological debates, this study adopts the space-vector method from the field of systems engineering to evaluate livelihood resilience in a more scientifically grounded manner.

When discussing livelihood resilience, most of the literature analyzes it at the theoretical intersection of resilience and vulnerability. Household adaptive capacity is the core ability of rural household livelihood system to adjust, transform and develop in the face of external disturbances. It also serves as a bridge connecting the concepts of resilience and vulnerability. This ability shows its importance in different contexts such as environment, climate, and water resources management (Yang et al., 2022), however, its specific manifestations and constituent dimensions vary by region and developmental stage. Scholars have increasingly employed empirical and quantitative methods to study household adaptive capacity in various scenarios and have established corresponding research frameworks for adaptive capacity. This study draws on the European Regional Adaptability Framework (Acosta et al., 2013) and the Adaptability Framework for apple growers in Northwest China (Li et al., 2017). In these studies, household adaptive capacity is framed as consciousness, ability and action, and includes dimensions such as social capital, infrastructure and flexibility. However, due to regional differences, natural conditions, and development objectives, establishing a universal adaptive capacity assessment index remains challenging. Some studies have sought to integrate theories to enhance the universal applicability of adaptive capacity assessments. Peng et al. (2022) studied the impact of livelihood diversification limits in the context of China, providing a new perspective for understanding the formation boundary of household adaptive capacity in different regions; Bauer et al. (2022) emphasized the key role of local knowledge in the construction of adaptive capacity index system through the comprehensive evaluation of Bolivian indigenous communities. According to these researches, this study continues the three-dimensional framework of ‘awareness, ability, action’ established by the European Regional Adaptability Framework, draws on the ideas of Xu et al. (2022), and combines the SLA to screen variables closely related to household adaptive capacity, and constructs an adaptive capacity assessment path in line with the disaster resettlement situation. We ground our empirical design in the SLA, which identifies five core capital assets—natural, human, social, financial, and physical—that shape household resilience. In line with this framework: land and livestock ownership represent natural capital; education and health status capture human capital; social networks and trust reflect social capital; savings and credit access measure financial capital; and housing quality and infrastructure proximity indicate physical capital.

To date, research on household livelihood resilience and LAC in disaster-resettlement contexts has received limited scholarly attention, particularly regarding the link between these constructs at the household level. While numerous studies have examined how disaster resettlement affects household livelihood resilience (Liu et al., 2020a), fewer have examined the components of livelihood resilience that influence adaptive capacity. Moreover, in studying the livelihoods of disaster-affected migrants, the literature has yet to fully address the fundamental question of how livelihood resilience influences household adaptive capacity. The internal relationship between these two concepts, especially the mechanism of action, still lacks empirical evidence from the post-disaster relocation situation. The present study attempts to measure household livelihood resilience and its impacts on adaptive capacity in a disaster resettlement setting. Given the urgent need for empirical evidence on livelihood resilience, adaptive capacity, and settlement development, this work may contribute to bridging the gap. Building on the identified gap and our proposed conceptual framework, this study poses two research questions: How does livelihood resilience influence household adaptive capacity in rural communities affected by disaster-induced relocation? Does the strength of this relationship vary among households with different levels of adaptive capacity, and through which mechanisms might such heterogeneity arise? These questions guide our empirical analysis.

Our empirical analysis uses survey data from 459 relocated households in post-disaster resettlement communities in Ankang, Shaanxi Province. To address the gap, we introduce resilience thinking into the livelihood system of rural households and apply the space vector method from system engineering and the indicator-based framework of household adaptive capacity to the case of disaster resettlement. And we argue that the concept of livelihood resilience must be disaggregated. Drawing on the SLA and the awareness, ability, action framework, this study develops a multidimensional measure of livelihood resilience using the spatial vector method, and examines its impact on household adaptive capacity in post-disaster resettlement communities in Ankang. This approach structures our empirical analysis and facilitates interpretation of heterogeneous effects across the distribution of adaptive capacity. By incorporating quantile regression, we reveal how these effects vary along the distribution of adaptive outcomes, offering insights for phased and targeted resilience-building strategies.

2 Materials and methods

2.1 Study site: southern Shaanxi province, China

As a hazard prone region, Shaanxi province (Figure 1) has had major challenges in carrying out disaster mitigation, environmental protection, and livelihood improvement. According to news media reports, a variety of natural hazards and extreme weather climate events occur every year in this area, causing injuries and deaths and tremendous economic losses. To mitigate environmental concerns caused by these natural disasters and ecological degradation, and to protect the ecosystems and their services, the local government launched a disaster avoidance resettlement project in May 2011. Through this initiative, Shaanxi province has attempted to improve the living conditions of impoverished rural residents living in areas deemed unable to achieve sustainable livelihoods (Lo et al., 2016). According to official statistics, 408,000 local households (1,380,000 residents) in Shaanxi province have been resettled in the 13th Five Year Plan period (2016–2020). The local government displaced 3,160,000 residents between 2011 and 2020 at considerable cost. The total investment in housing building, physical infrastructure, basic public services, and local industrial development has exceeded RMB 132 billion. The resettlement project implemented in this region can serve as a typical attempt to build resilience and enhance adaptive capacity at the household scale.

Figure 1

Figure 1. Location of the study area.

2.2 Data collection

This research was based on questionnaire-based household survey data from a cross-sectional quantitative survey in Ankang Prefecture, one of the three prefectures of southern Shaanxi province. For this survey, our team employed a convenience sampling design with questionnaires for rural households and communities, rather than stratified sampling. There were also some unstructured informal individual interviews. Survey data were collected by professionally trained survey-takers through face-to-face interviews with the head of each household or adult household members. The survey recorded information about family demographic features, livelihood capitals and activities, household consumption and expenditure, as well as resettlement characteristics. The interviewers were trained to administer and recycle the structured questionnaires, while supervisors oversaw the interviewers’ collection process. In addition, supervisors randomly revisited some survey respondents to check the information quality, and sent the feedback to the interviewers for further improvement. After data collection, our team conducted logical and numerical checks, data entry tests, and privacy protection of whole respondents. The survey eventually collected a total of 657 valid questionnaires, with an effective rate of 98.06%–459 relocated and 198 non-relocated households. This study focuses only on the sample of 459 relocated households.

2.3 Measuring household livelihood resilience and adaptive capacity

2.3.1 Livelihood resilience

People have the capacity to protect themselves against unpredictable future risks through their livelihoods (Qin et al., 2021). Livelihood resilience entails that household’s livelihood strategies can cope with and manage the impact of shocks in order to adapt to changing conditions (Ma et al., 2024). A resilient socio-ecological system is one that has the capacity to continue to provide support for livelihoods when subjected to external shocks and perturbations. In the case of rural development, rural household’s livelihoods are the link between social and natural systems through the management of resource use at the individual, community, regional, global, or other scales (Kwazu and Chang-Richards, 2022). In household level socio-ecological systems, livelihood systems with high resilience and low sensitivity are the most robust, while the most vulnerable livelihood systems are the least robust.

Here, we take the resilience of livelihood systems to be the ability to respond to disturbances or shocks. Specifically, when the degree of perturbation exceeds the threshold of resilience, the household livelihood system will be transformed. Therefore, more resilient households have more difficulty in accomplishing this transition. Specific resilience is the resilience of certain specific parts of system to cope with specific shocks through variation of relevant control variables in the context of specific types of disturbances (Liu et al., 2025). In contrast, general resilience does not entail that certain parts of the system may exceed a threshold, or refer to the types of shocks the system must suffer; it describes the capacity of the system to cope with various uncertainties (Li et al., 2017). Household livelihood resilience can be characterized as the capacity of the rural household to cope with disturbances, successfully navigate changes, and maintain its livelihood system via a combination of livelihood activities, adaptive actions, and coping strategies. It may involve specific resilience, which entails response to an identifiable disturbance, and general resilience that can respond to various risks and shocks.

We have the following definitions: General resilience refers to household’s long-term, cross-contextual adaptive capacities accumulated under normal conditions, reflecting foundational assets such as human, social, and financial capital that support recovery from various shocks. Specific resilience shows context-specific preparedness for disaster risks, emphasizing rapid response and short-term recovery capabilities including emergency supplies, risk awareness, access to early warning systems, and expectations of institutional support. Although there is difference in mechanisms, these two forms of resilience are complementary rather than mutually exclusive: general resilience provides a stable baseline for adaptive capacity, whereas specific resilience enhances immediate coping capacity.

Following previous studies (Liu et al., 2025), we employ the space vector method from system engineering to evaluate livelihood resilience in terms of specific resilience and general resilience (Liu et al., 2024; Schlör et al., 2018; Simonsen, 2014). First, we simplify the household livelihood system by focusing on livelihood activities as the primary observational units, and a quantitative model of livelihood status through functional relationships. Based on the actual livelihood practices in the field of study, the key activities are incorporated into the assessment framework. Subsequently, both general resilience and specific resilience are measured by representing them as vectors in multidimensional space. Following established approaches in systems forecasting and decision analysis, the overall livelihood resilience of each household is estimated using the spatial vector method. This method is based on the livelihood activities of rural households. Due to limited space, we do not show steps of the formula and variable calculations here, And the specific measurement steps can be seen in the studies of Liu et al. (2025).

2.3.2 Livelihood adaptive capacity

The LAC refers to the response behavior exhibited by the social ecosystem in the face of both internal and external pressures. It represents household’s ability to anticipate and react to natural or perceived disturbances, mitigate their impact, and recover from the aftermath of these disturbances (Xu et al., 2022). Previous research on the adaptive capacity of rural households has mainly centered on their adaptation behaviors and strategic choices concerning climate change (Mayer et al., 2014). Although some studies have explored adaptive capacity in the context of resettlement (See and Wilmsen, 2020), there is little research utilizing quantitative methods to assess the adaptive capacity of rural households (Li et al., 2017; Bauer et al., 2022). Acosta et al. (2013) offer a conceptual framework for gauging regional adaptive capacity, and in the present study, we adopt their framework to evaluate the adaptive capacity of livelihood. Specifically, we establish an index of LAC for rural households based on awareness, ability, and actions (refer to Table 1). Through principal component analysis (PCA), we identify the main factors influencing LAC. Thirteen variables undergo PCA, resulting in the identification of five principal components, and using these we develop a formula to calculate the index of LAC (Equation 1). This formula incorporates the variance contribution of each principal component as a weight for the score value of the respective component.

$\text{LAC}={\mathrm{W}}_1{\mathrm{F}}_1+{\mathrm{W}}_2{\mathrm{F}}_2+\cdots +{\mathrm{W}}_{\mathrm{i}}{\mathrm{F}}_{\mathrm{i}}(1)$

here LAC represents the score of the rural household’s LAC; F_i is the score of the i-th principal component (i = 1, 2 … 5); Wi is the weight of the i-th (i = 1, 2 … 5) principal component, namely, the contribution of this principal com-ponent to the overall variance; the matrix of principal component score coefficients and the original normalized values of each indicator can be calculated for i = 1, 2, … , 5. The specific formula for LAC constructed in this study is Equation 2:

$\text{LAC}=0.1246{\mathrm{F}}_1+0.1235{\mathrm{F}}_2+0.1055{\mathrm{F}}_3+0.0941{\mathrm{F}}_4+0.0897{\mathrm{F}}_5(2)$

Table 1

Table 1. Index of LAC.

Our study conceptualizes livelihood resilience and household adaptive capacity as distinct yet interrelated constructs. Although both concepts are related to household’s ability to navigate change, they are different in scope and temporal orientation: livelihood resilience refers to the latent potential—the accumulated stock of capital (natural, human, social, financial, physical)—that enables a household to withstand, recover from, or reorganize aftershocks; by contrast, adaptive capacity denotes the active process—the awareness, skills, and actions—through which households adjust their livelihood strategies in response to new environmental and socioeconomic conditions.

2.4 Model construction

Building on the development of the LAC index for rural households, we study the influence of livelihood resilience on the adaptive capacity of these households. The initial step involves employing the ordinary least squares (OLS) regression analysis, which is expressed as follows (Equation 3):

$\text{LAC}={\mathrm{\beta }}_0+{\mathrm{\beta }}_1{\mathrm{F}}_r+{\mathrm{\beta }}_2\text{\hspace{0.17em}Average\hspace{0.17em}years\hspace{0.17em}of\hspace{0.17em}education}+{\mathrm{\beta }}_3\text{\hspace{0.17em}Voluntary\hspace{0.17em}relocation}+\mathrm{\mu }(3)$

where Fr, voluntary relocation, average years of education are explanatory variables; β₀ is the constant term, β₁, β₂ are the coefficients of the explanatory variables; μ is the random term.

The strategy of the regression is as follows: first, take the relocated households as a sample and include livelihood resilience, whether relocation was voluntary, and average years of education, respectively, to analyze their impacts on LAC; and then add three covariates of household size, dependence ratio, and livelihood diversity index for regression. The household size is population of household families (number), the dependence ratio is the proportion of children and the elderly in the labor force of rural households and the livelihood diversity index is number of types of livelihood activities. So, the regression formula becomes Equation 4:

$\text{LAC}={\mathrm{\beta }}_0+{\mathrm{\beta }}_1{\mathrm{F}}_r+{\mathrm{\beta }}_2\text{\hspace{0.17em}Average\hspace{0.17em}years\hspace{0.17em}of\hspace{0.17em}education}+{\mathrm{\beta }}_3\text{\hspace{0.17em}Voluntary\hspace{0.17em}relocation}+{\mathrm{\beta }}_4\text{\hspace{0.17em}Household\hspace{0.17em}size}+{\mathrm{\beta }}_5\text{\hspace{0.17em}Dependence\hspace{0.17em}ratio}+{\mathrm{\beta }}_6\text{\hspace{0.17em}Livelihood\hspace{0.17em}Diversity\hspace{0.17em}Index}+\mathrm{\mu }(4)$

The final regression is formed by adding the three variables of level of trust in those around you (called “Belief”) (Wang et al., 2025), likelihood of borrowing (called “Loan”) (Asante et al., 2025), and whether the household has suffered a natural disaster (called “Risk”), and the regression formula is as follows:

$\text{LAC}={\mathrm{\beta }}_0+{\mathrm{\beta }}_1{\mathrm{F}}_r+{\mathrm{\beta }}_2\text{\hspace{0.17em}Average\hspace{0.17em}years\hspace{0.17em}of\hspace{0.17em}education}+{\mathrm{\beta }}_3\text{\hspace{0.17em}Voluntary\hspace{0.17em}relocation}+{\mathrm{\beta }}_4\text{\hspace{0.17em}Household\hspace{0.17em}size}+{\mathrm{\beta }}_5\text{\hspace{0.17em}Dependence\hspace{0.17em}ratio}+{\mathrm{\beta }}_6\text{\hspace{0.17em}Livelihood\hspace{0.17em}Diversity\hspace{0.17em}Index}+{\mathrm{\beta }}_7{\text{Belief}+\mathrm{\beta }}_8{\text{Loan}+\mathrm{\beta }}_9\text{\hspace{0.17em}Risk}+\mathrm{\mu }(5)$

Traditional least squares regression provides an approximation to the conditional mean of the distribution (Peng et al., 2022), and is highly sensitive to extreme values. Quantile regression, rooted in median regression, offers an alternative approach. Median regression aims to minimize the sum of the absolute values of residuals, in contrast to OLS regression, which minimizes the sum of squared errors (Karimi Alavijeh et al., 2023; Tohidimoghadam et al., 2023). A significant advantage of quantile regression is its capacity to be estimated for various percentiles of the conditional distribution. Consequently, to mitigate the impact of extreme values on estimation outcomes and enhance the precision of model estimation, we also employ quantile regression to examine the impact of resilience on LAC. Building on OLS estimates, quantile regression models encompass all covariates outlined in Equation 5. We conducted these computations using Stata 16.1 statistical software.

Additionally, to visually examine the distribution of LAC and the potential nonlinear relationship between livelihood resilience and LAC, we employ kernel density estimation to calculate the probability density function of LAC, which helps identify subgroups such as low and high adaptive capacity households. Polynomial fitting is used to explore the functional form of the association. Specifically, a quadratic polynomial regression is applied to capture the overall trend, while a cubic polynomial regression is used to test for more complex inflection patterns. These graphical techniques are exploratory in nature and are intended to complement the quantile regression results by providing visual insights into heterogeneous response patterns.

3 Results

3.1 Descriptive statistics of variables

The relocation types of the sampled rural households are categorized as voluntary and involuntary relocations. Involuntary relocation of rural households is project-induced resettlement, while all other types of migrants are considered to be voluntary relocations. For those relocating from their own villages, the relocation is classified as short distance, whereas relocations from other sources are categorized as long distance. Table 2 illustrates the descriptive statistics for all variables. Table 3 illustrates the distinctions in livelihood resilience and LAC among various relocated rural households.

Table 2

Table 2. Descriptive statistics for all variables.

Table 3

Table 3. Descriptive statistics.

The results indicate that the LAC of short distance relocated rural households (−0.008) is significantly lower than that of long-distance households (0.052). Similarly, the livelihood resilience of short distance relocated rural households (0.203) is marginally lower than that of their long-distance counterparts (0.228). In terms of general resilience, short distance relocated rural households (0.065) exhibit significantly higher values than long distance relocated rural households (0.116). However, specific resilience (0.163) is slightly lower for short distance relocated households than for long distance relocated households (0.164). The LAC of voluntarily relocated rural households (−0.006) is slightly lower than that of involuntarily relocated rural households (0.022). Voluntarily relocated rural households also show significantly lower livelihood resilience (0.197) compared to involuntarily relocated households (0.253). In terms of general resilience, voluntarily relocated rural households (0.069) have slightly lower values than involuntarily relocated households (0.074). Notably, the specific resilience of voluntarily relocated rural households (0.153) is lower than that of others (0.225).

3.2 Fitted graph of livelihood resilience and LAC

Kernel densities were fitted with the horizontal axis as the years of relocation and whether to voluntarily relocate, and the vertical axis as LAC. The results of the fitting are shown in Figure 2. The peaks of short term relocated rural households appear first, followed by long term and medium term, indicating that the longer the relocation period, the higher the household’s LAC; the peak points of voluntarily relocated rural households and involuntarily relocated rural households are closer to each other, indicating that whether to voluntarily relocate has little effect on the LAC of rural households.

Figure 2

Figure 2. Graphical fitting of relocation information and adaptive capacity.

Figure 3 shows the primary fit, secondary fit, and kernel density fit with LAC on the horizontal axis and livelihood resilience and general resilience on the vertical axis, respectively. It can be seen that the primary fit of livelihood resilience and general resilience with LAC are all ascending straight lines; the secondary fit are all curves; the kernel density fit are all oscillating curves, and the kernel density fit curves of general resilience with LAC are multipeak curves, and as general resilience improves the highest peaks eventually appear; the kernel density fit curves of livelihood resilience with LAC are oscillating curves; when livelihood resilience is lower, LAC is also lower, and as livelihood resilience increases, LAC is slowly increasing, but the final peak is still lower than general resilience.

Figure 3

Figure 3. Graphical fitting of resilience and adaptive capacity.

3.3 Ordinary least squares average estimates

We initially estimated LAC using OLS regression models. Models 1, 4, and 7 include livelihood resilience, general resilience, specific resilience, and the average years of education of the households as variables. Subsequently, Models 2, 5, and 8 introduce additional covariates, namely, household size, dependence ratio, and livelihood diversity index. Further complexity is incorporated in Model 3, 6, and 9 by adding more covariates, namely, belief, loans, and risk. Tables 4–6 show the outcomes of OLS regressions with three different sets of control variables.

Table 4

Table 4. Ordinary least squares estimates.

Table 5

Table 5. Ordinary least squares estimates.

Table 6

Table 6. Ordinary least squares estimates.

3.4 Quantile regression

Across all results, livelihood resilience, general resilience, and the average years of education for households exhibit strong, significant, positive impacts on LAC. The effect of specific resilience on LAC is not found to be significant. Recognizing that relying solely on the OLS model might obscure the relationship between LAC, resilience, and education, we explore further this using quantile regression. The coefficients and the significance results of the quantile regression model are shown in Tables 7–9, using the full set of covariates from Tables 4–6. Table 7 shows two major results. First, livelihood resilience has a positive association with LAC at the 50th, 75th, and 90th percentiles, but it is not significant at low levels. General resilience has significant positive impacts upon LAC, which generally increases with increase in LAC. Specific resilience is positively correlated with livelihood resilience only at the 10th percentile. Second, the average years of household education has a significant positive effect on LAC for all percentiles of distribution. The OLS estimates show significant positive effects on livelihood resilience, general resilience, and the average years of education, which mask the trend we see when estimating conditional distributions based on quartiles of LAC.

Table 7

Table 7. Quantile regression results.

Table 8

Table 8. Quantile regression results.

Table 9

Table 9. Quantile regression results.

Tables 8, 9 show results for the other two models when the control variables are different, and the results are consistent with Table 7. Broadly, livelihood resilience, general resilience, and the average years of education are more strongly associated with increased LAC quartiles at higher quartiles, and specific resilience is more strongly associated with increased LAC at lower quartiles. We found some slight biases, but the overall trend remains the same as before. Figures 4–6 plot the quantile estimates of resilience and education in the first and second row, respectively. The first row shows the quantile regression plot including only resilience and education as independent variables when the dependent variable is LAC. The last row shows the quantile regression results for resilience and education that include all covariates.

Figure 4

Figure 4. Coefficient estimates obtained from quantile regression models for LAC (Voluntary relocation).

Figure 5

Figure 5. Coefficient estimates obtained from quantile regression models for LAC (Relocation distance).

Figure 6

Figure 6. Coefficient estimates obtained from quantile regression models for LAC (Relocation type).

4 Discussion

Livelihood resilience and adaptive capacity are important for rural household livelihood systems. The resilience level of a livelihood system determines its ability to handle and respond to stresses and shocks (Liu et al., 2020b), while adaptive capacity allows adjustments to environmental changes by mobilizing and utilizing capital developed due to resilience (Xu et al., 2022). Generally, stronger adaptive capacity provides households with more opportunities to improve their wellbeing, reduce vulnerability, and further enhance resilience (Qin et al., 2021; Tohidimoghadam et al., 2023). This correlation has been noted in prior studies. For instance, Tanner et al. (2015) observed that households exhibit greater adaptive capacity when they have improved access to capital assets. Tahiru et al. (2019) investigated the enhancement of adaptive capacity among Ghanaian farmers through skill development and knowledge building. Similarly, our previous studies have highlighted the impact of disaster resettlement on livelihood resilience and adaptive capacity (Xu et al., 2022; Liu et al., 2020a). However, these studies have seldom integrated the two paradigms and frameworks when discussing resilience and adaptive theoretical analytical frameworks. It is crucial for the adaptive capacity of household livelihood systems to cope with environmental changes and shocks, thereby enhancing resilience.

Livelihood resilience offers insights into the ability to recover from catastrophic disruptions from a livelihood perspective, encompassing mechanisms for adaptation, coping, and transformation in the face of disasters (Xu et al., 2022). It often constitutes a fundamental element of sustainable development strategies (Dovie and Pabi, 2023). In this study, livelihood resilience, particularly specific livelihood resilience, is closely linked to livelihood diversification. There is a general acknowledgment that the higher the degree of livelihood diversification, the greater the resilience and adaptive capacity of household livelihood systems (Melvani et al., 2020; Morris et al., 2009). However, our findings indicate that specific resilience has a significantly negative impact at lower levels of LAC when employing various control variables. This suggests that for households with low levels of LAC, the transformation of single livelihood activities into diversified livelihoods diverts household efforts and engenders risks in livelihood transformation (Karimi Alavijeh et al., 2023). The focus on diversification in livelihood activities overlooks the dilemma faced by farm households. Capital for non-agricultural development is de-rived from the already scarce assets and capabilities of rural households, imposing a burden on them and posing livelihood risks (Kwazu and Chang-Richards, 2022).

In contrast, livelihood resilience has significant positive impacts at higher LAC levels, and these impacts increase with the level of LAC in the higher quartiles (Tran and Vu, 2020). General resilience and education have significant positive effects on adaptive capacity at all levels, consistent with previous studies (Bauer et al., 2022). As relocated rural households are subject to various policy subsidies as well as welfare preferences (Li and Zander, 2020), their overall income level is increased significantly. This has solved the double dilemma of eco-environmental protection and social and economic development to a certain extent. Conversely, lower LAC has less of an impact on education than higher LAC. Although previous studies have suggested that the more vulnerable rural households need stronger human capital to cope with external risks (Peng et al., 2019), we find that the higher the education level of the relocated households, the more likely they are to accept new technologies and means to improve their adaptive capacity. Social capital yields greater returns at lower levels of adaptive capacity. Finally, education consistently enhances adaptive capacity across all levels, rendering it a life-course–effective lever for intervention. Local governments should therefore ensure access to quality schooling in resettlement areas and provide adult education opportunities as a sustained, cross-cutting strategy. This phased approach—“secure first, empower next, educate throughout”—is grounded in the empirical patterns revealed by our model and offers a clear, actionable roadmap for policymakers. Therefore, our results lend indirect support to the proposed mechanisms linking resilience to adaptive capacity. The significantly stronger effects of financial and physical capital at lower quantiles align with the resource-access pathway, suggesting that asset-poor households rely heavily on basic endowments to initiate adaptive actions. Meanwhile, the increasing importance of social capital and institutional participation among higher-capacity households resonates with the social leverage mechanism, where networks amplify individual agency into collective outcomes.

This study has some limitations. First, it constructed the livelihood resilience index using the space vector method and determined the indicator weights within the space vector framework by using the load values of principal component analysis. However, different analyses are needed to determine the accuracy of these results. Second, due to the limited survey data, our analysis took care to use quantile regression analysis, but this does not exclude further statistical improvement. Third, the factors affecting the resilience of rural household livelihoods are complex and interrelated (Liu et al., 2020a), and although we introduce relevant relocation characteristics, we still cannot include all potentially relevant variables. In addition, though we control variables such as average years of education in our regression models, unobserved household characteristics or self-selection into relocation programs may still generate indigeneity and selection bias. However, our findings provide a systematic assessment of the association between livelihood resilience and adaptive capacity through robust measurement approaches, and thus remain credible. Future studies could employ instrumental variable (IV) estimation to address indigeneity concerns. Finally, this study relies on a cross-sectional survey conducted in Ankang, southern Shaanxi, and uses convenience sampling to collect data. Although feasible given resource and time constraints, this approach may introduce systematic biases in demographic characteristics, regional distribution, or household livelihood strategies, thereby affecting external validity Furthermore, due to the single time point design, the observed associations cannot support causal inferences. Nevertheless, the primary objective is to identify patterns of association rather than establish causality This research provides preliminary evidence regarding how enhanced livelihood resilience contributes to the reshaping of LAC in rural China, laying the groundwork for future studies with greater representativeness and stronger causal identification. Potential causal mechanisms could be further examined through longitudinal cohort studies or intervention experiments.

5 Conclusion

Based on collected data from Ankang, Shaanxi Province, this study analyzes the relationship between livelihood resilience and adaptive capacity among rural households related to disaster resettlement using quantile regression. The findings show significant heterogeneity in resilience effects: overall livelihood resilience (Fr) and general resilience (Fa) have strong positive impacts at higher levels of adaptive capacity, but show limited or non-significant effects for households with low adaptive capacity. Notably, specific resilience (Fv) shows a significant negative effect only at the lowest quantile, suggesting that an excessive emphasis on disaster-specific preparedness may, paradoxically, hinder recovery among the most vulnerable households. This implies that in contexts of severe resource constraints, interventions focused solely on emergency readiness—such as drills or material stockpiling—may impose additional psychological or logistical burdens without sufficient support systems, thereby failing to translate into improved adaptive outcomes. On the other hand, general resilience (e.g., human capital accumulation) and education consistently show a positive increasingly effect in all levels, with stronger effects observed at higher quantiles, underscoring their role as foundational drivers of sustainable development. It explains why quantile regression results show stronger coefficients at lower quantiles of adaptive capacity—because for vulnerable groups, baseline resilience is the primary determinant of whether adaptive capacity can begin at all. By integrating empirical evidence with the analysis framework, the results of this article contribute to a more nuanced understanding of ‘resilience-adaptive capacity’ dynamics in post-disaster rural resettlement settings. In addition, building on SLA’s five capitals and the awareness, ability, action model, we propose a stage-dependent mechanism model, which outlines how each pathway operates differently across the adaptive capacity spectrum. General resilience supports early-stage awareness and ability building, whereas specific preparedness become critical in later-stage action and system-level change. By articulating both why and through which mechanisms livelihood resilience affects adaptive capacity, we advance a more nuanced, empirically testable understanding of the ‘resilience–adaptive capacity’ nexus in disaster-affected rural communities.

Previous studies have confirmed that decreasing vulnerability and enhancing resilience are all about better and more scientific adaptation (Xu et al., 2022; Acosta et al., 2013). The present study corroborates the findings of these previous studies. Efforts by relocating households to enhance adaptive capacity are part of a long-term process. Policymakers and stake-holders need to work together to address and cope with the challenges to improving livelihood capacity. As revealed by our quantile regression analysis, first, in the early stages of resettlement, policy efforts should prioritize securing basic livelihoods and promoting social integration, rather than imposing complex disaster-specific responsibilities. Our findings indicate that specific resilience has no significant—or even negative—effects at lower quantiles, suggesting that mandatory emergency drills or self-managed preparedness tasks may overwhelm vulnerable households. Instead, community support systems, mental health services, and stable employment placement can more effectively build foundational wellbeing. Second, once households reach moderate to high levels of adaptive capacity (e.g., above the 50th percentile), interventions should shift toward enhancing general resilience, such as vocational training, children’s education, and financial literacy programs. Moreover, we expect more theoretical and empirical studies focusing on this topic in the future to provide clear guidance on how to translate resilience thinking into practical implementation strategies on the ground. This will help to promote well-functioning livelihood systems for rural household livelihood systems in a sustainable way.

Data availability statement

The original contributions presented in the study are included in the article/supplementary material, further inquiries can be directed to the corresponding author.

Author contributions

LG: Writing – review and editing, Methodology, Conceptualization, Formal Analysis, Visualization, Software. QZ: Supervision, Writing – review and editing, Project administration, Funding acquisition. SZ: Formal Analysis, Visualization, Methodology, Writing – review and editing. XL: Methodology, Software, Visualization, Writing – review and editing. JL: Formal Analysis, Data curation, Writing – original draft. MF: Funding acquisition, Validation, Writing – review and editing, Supervision. WL: Writing – review and editing, Funding acquisition, Investigation, Writing – original draft, Conceptualization.

Funding

The author(s) declared that financial support was received for this work and/or its publication. This research was jointly supported by the National Natural Science Foundation of China (Grant No. 71803149; No. 72474173; No. 72503176; No. 72574182), the Social Science Foundation of Shaanxi Province (Grant No. 2023R010), the Innovation Capability Support Program of Shaanxi (Program No. 2025KG-YBXM-113), and the Morrison Institute for Population and Resource Studies at Stanford University.

Acknowledgements

The authors appreciate the support of the local government and the patient cooperation of the interviewees during the data collection.

Conflict of interest

The author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

References

Acosta, L., Klein, R. J. T., Reidsma, P., Metzger, M. J., Rounsevell, M. D. A., Leemans, R., et al. (2013). A spatially explicit scenario-driven model of adaptive capacity to global change in Europe. Glob. Environ. Change 23, 1211–1224. doi:10.1016/j.gloenvcha.2013.03.008

CrossRef Full Text | Google Scholar

Asante, B. O., Prah, S., Akutinga, S., Akusaki, E. T., and Ofosuhene, A. D. (2025). Climate resilience in the palm of a hand: digital financial inclusion and cocoa farmers’ adoption of climate smart agricultural technologies in Ghana. Future Bus. J. 11 (1), 213. doi:10.1186/s43093-025-00624-5

CrossRef Full Text | Google Scholar

Bauer, T., De Jong, W., Ingram, V., Arts, B., and Pacheco, P. (2022). Thriving in turbulent times: livelihood resilience and vulnerability assessment of Bolivian Indigenous forest households. Land Use Policy 119, 106146. doi:10.1016/j.landusepol.2022.106146

CrossRef Full Text | Google Scholar

Boyacıoğlu, D., Göçer, Ö., and Karahan, E. E. (2023). Exploring identity issues in development areas of vernacular rural settlements: a case study of Behramkale, Türkiye. J. Contemp. Urban Aff. 7 (1), 51–68. doi:10.25034/ijcua.2023.v7n1-4

CrossRef Full Text | Google Scholar

Chen, J., Yin, S., Gebhardt, H., and Yang, X. (2018). Farmers’ livelihood adaptation to environmental change in an arid region: a case study of the minqin oasis, northwestern China. Ecol. Indic. 93, 411–423. doi:10.1016/j.ecolind.2018.05.017

CrossRef Full Text | Google Scholar

Deng, Q., Li, E., and Zhang, P. (2020). Livelihood sustainability and dynamic mechanisms of rural households out of poverty: an empirical analysis of hua county, Henan Province, China. Habitat Int. 99, 102160. doi:10.1016/j.habitatint.2020.102160

CrossRef Full Text | Google Scholar

Dovie, D. B. K., and Pabi, O. (2023). Partial climatic risk screening, adaptation and livelihoods in a coastal urban area in Ghana. Habitat Int. 138, 102868. doi:10.1016/j.habitatint.2023.102868

CrossRef Full Text | Google Scholar

Karimi Alavijeh, N., Salehnia, N., Salehnia, N., and Koengkan, M. (2023). The effects of agricultural development on CO2 emissions: empirical evidence from the most populous developing countries. Environ. Dev. Sustain 25, 12011–12031. doi:10.1007/s10668-022-02567-1

CrossRef Full Text | Google Scholar

Kwazu, G. C., and Chang-Richards, A. (2022). A tool to assess livelihood preparedness for disasters: a study of kaikōura earthquake in New Zealand. Nat. Hazards 113, 745–766. doi:10.1007/s11069-022-05322-x

CrossRef Full Text | Google Scholar

Li, Q., and Zander, P. (2020). Resilience building of rural livelihoods in PES programmes: a case study in China’s loess hills. Ambio 49, 962–985. doi:10.1007/s13280-019-01236-4

PubMed Abstract | CrossRef Full Text | Google Scholar

Li, M., Huo, X., Peng, C., Qiu, H., Shangguan, Z., Chang, C., et al. (2017). Complementary livelihood capital as a means to enhance adaptive capacity: a case of the loess Plateau, China. Glob. Environ. Change 47, 143–152. doi:10.1016/j.gloenvcha.2017.10.004

CrossRef Full Text | Google Scholar

Liu, W., Li, J., and Xu, J. (2020a). Effects of disaster-related resettlement on the livelihood resilience of rural households in China. Int. J. Disaster Risk Reduct. 49, 101649. doi:10.1016/j.ijdrr.2020.101649

CrossRef Full Text | Google Scholar

Liu, W., Li, J., and Xu, J. (2020b). Impact of the ecological resettlement program in southern Shaanxi Province, China on households’ livelihood strategies. For. Policy Econ. 120, 102310. doi:10.1016/j.forpol.2020.102310

CrossRef Full Text | Google Scholar

Liu, W., Liu, J., Xu, J., Li, J., and Feldman, M. (2024). Examining the links between household livelihood resilience and vulnerability: disaster resettlement experience from rural China. Front. Environ. Sci. 11, 1340113. doi:10.3389/fenvs.2023.1340113

CrossRef Full Text | Google Scholar

Liu, W., Dou, B., Qian, L., and Zhao, X. (2025). Role of livelihood resilience in improving sustainable household well-being for disaster resettlement in ankang prefecture, Shaanxi Province, China. J. Mt. Sci. 22 (11), 4190–4205. doi:10.1007/s11629-024-9337-x

CrossRef Full Text | Google Scholar

Lo, K., and Wang, M. (2018). How voluntary is poverty alleviation resettlement in China? Habitat Int. 73, 34–42. doi:10.1016/j.habitatint.2018.01.00217

CrossRef Full Text | Google Scholar

Lo, K., Xue, L., and Wang, M. (2016). Spatial restructuring through poverty alleviation resettlement in rural China. J. Rural Stud. 47, 496–505. doi:10.1016/j.jrurstud.2016.06.006

CrossRef Full Text | Google Scholar

Ma, L., Zhang, Y., Li, T., Zhao, S., and Yi, J. (2024). Livelihood capitals and livelihood resilience: understanding the linkages in China’s government-led poverty alleviation resettlement. Habitat Int. 147, 103057. doi:10.1016/j.habitatint.2024.103057

CrossRef Full Text | Google Scholar

Mayer, A. L., Donovan, R. P., and Pawlowski, C. W. (2014). Information and entropy theory for the sustainability of coupled human and natural systems. E&S 19, art11. doi:10.5751/ES-06626-190311

CrossRef Full Text | Google Scholar

Melvani, K., Bristow, M., Moles, J., Crase, B., and Kaestli, M. (2020). Multiple livelihood strategies and high floristic diversity increase the adaptive capacity and resilience of Sri Lankan farming enterprises. Sci. Total Environ. 739, 139120. doi:10.1016/j.scitotenv.2020.139120

PubMed Abstract | CrossRef Full Text | Google Scholar

Morris, M. D., Popper, S. T., Rodwell, T. C., Brodine, S. K., and Brouwer, K. C. (2009). Healthcare barriers of refugees post-resettlement. J. Community Health 34, 529–538. doi:10.1007/s10900-009-9175-3

PubMed Abstract | CrossRef Full Text | Google Scholar

Peng, L., Xu, D., and Wang, X. (2019). Vulnerability of rural household livelihood to climate variability and adaptive strategies in landslide-threatened western mountainous regions of the three gorges reservoir area, China. Clim. Dev. 11, 469–484. doi:10.1080/17565529.2018.1445613

CrossRef Full Text | Google Scholar

Peng, W., Robinson, B. E., Zheng, H., Li, C., Wang, F., and Li, R. (2022). The limits of livelihood diversification and sustainable household well-being, evidence from China. Environ. Dev. 43, 100736. doi:10.1016/j.envdev.2022.100736

CrossRef Full Text | Google Scholar

Qin, Y., Shi, X., Li, X., and Yan, J. (2021). Geographical indication agricultural products, livelihood capital, and resilience to meteorological disasters: evidence from kiwifruit farmers in China. Environ. Sci. Pollut. Res. 28, 65832–65847. doi:10.1007/s11356-021-15547-1

PubMed Abstract | CrossRef Full Text | Google Scholar

Quandt, A. (2018). Measuring livelihood resilience: the household livelihood resilience approach (HLRA). World Dev. 107, 253–263. doi:10.1016/j.worlddev.2018.02.024

CrossRef Full Text | Google Scholar

Rogers, S., and Xue, T. (2015). Resettlement and climate change vulnerability: evidence from rural China. Glob. Environ. Change 35, 62–69. doi:10.1016/j.gloenvcha.2015.08.005

CrossRef Full Text | Google Scholar

Schlör, H., Venghaus, S., Märker, C., and Hake, J.-F. (2018). Managing the resilience space of the German energy system - a vector analysis. J. Environ. Manag. 218, 527–539. doi:10.1016/j.jenvman.2018.04.053

PubMed Abstract | CrossRef Full Text | Google Scholar

See, J., and Wilmsen, B. (2020). Just adaptation? Generating new vulnerabilities and shaping adaptive capacities through the politics of climate-related resettlement in a Philippine coastal city. Glob. Environ. Change 65, 102188. doi:10.1016/j.gloenvcha.2020.102188

CrossRef Full Text | Google Scholar

Simonsen, S. H. E. A. (2014). Applying resilience thinking seven principles for building resilience in social-ecological systems. Available online at: http://www.mendeley.com/research/applyingresilience-thinking-seven-principles-building-resilience-socialecological-systems/ (Accessed May 27, 2025).

Google Scholar

Sina, D., Chang-Richards, A. Y., Wilkinson, S., and Potangaroa, R. (2019). A conceptual framework for measuring livelihood resilience: relocation experience from Aceh, Indonesia. World Dev. 117, 253–265. doi:10.1016/j.worlddev.2019.01.003

CrossRef Full Text | Google Scholar

Speranza, C. I., Wiesmann, U., and Rist, S. (2014). An indicator framework for assessing livelihood resilience in the context of social–ecological dynamics. Glob. Environ. Change 28, 109–119. doi:10.1016/j.gloenvcha.2014.06.005

CrossRef Full Text | Google Scholar

Tahiru, A., Sackey, B., Owusu, G., and Bawakyillenuo, S. (2019). Building the adaptive capacity for livelihood improvements of Sahel Savannah farmers through NGO-led adaptation interventions. Clim. Risk Manag. 26, 100197. doi:10.1016/j.crm.2019.100197

CrossRef Full Text | Google Scholar

Tanner, T., Lewis, D., Wrathall, D., Bronen, R., Cradock-Henry, N., Huq, S., et al. (2015). Livelihood resilience in the face of climate change. Nat. Clim. Change 5, 23–26. doi:10.1038/nclimate2431

CrossRef Full Text | Google Scholar

Tohidimoghadam, A., PourSaeed, A., Bijani, M., and Eshraghi Samani, R. (2023). Towards farmers’ livelihood resilience to climate change in Iran: a systematic review. Environ. Sustain. Indic. 19, 100266. doi:10.1016/j.indic.2023.100266

CrossRef Full Text | Google Scholar

Tran, T. Q., and Vu, H. V. (2020). The pro-poor impact of non-crop livelihood activities in rural Vietnam: a panel data quantile regression analysis. Econ. Analysis Policy 68, 348–362. doi:10.1016/j.eap.2020.10.005

CrossRef Full Text | Google Scholar

Wang, L., Xu, J., and Zhu, D. (2025). Impact of social trust and personal constraints on farmers’ adoption of water-saving irrigation technology. J. Econ. Water Resour. 43, 60–69. doi:10.3880/j.issn.1003-9511.2025.05.008

CrossRef Full Text | Google Scholar

Xu, J., Liu, W., Li, J., Li, C., and Feldman, M. (2022). Disaster resettlement and adaptive capacity among rural households in China. Soc. and Nat. Resour. 35, 245–259. doi:10.1080/08941920.2022.2048329

CrossRef Full Text | Google Scholar

Yang, X., Li, X., Lu, K., and Peng, Z.-R. (2022). Integrating rural livelihood resilience and sustainability for post-disaster community relocation: a theoretical framework and empirical study. Nat. Hazards. doi:10.1007/s11069-022-05739-4

CrossRef Full Text | Google Scholar

Zhao, X., Li, J., and Liu, X. (2024). Does disaster resettlement reshape household livelihood adaptive capacity in rural China? Front. Sustain. Food Syst. 8, 1349532. doi:10.3389/fsufs.2024.1349532

CrossRef Full Text | Google Scholar