Measuring farmers' sustainable livelihood resilience in the context of poverty alleviation: a case study from Fugong County, China

As the COVID-19 pandemic enters its third year, its catastrophic impact on human lives and livelihoods and global efforts to achieve the Sustainable Development Goals (SDGs) are clear. Statistics show that the emergence of COVID-19 resulted in an increase in the global poverty level from 8.3% in 2019 to 9.2% in 2020, with about eight million additional workers falling into poverty. This is the first increase in extreme poverty since 1998 and the largest annual increase since 1990, setting global poverty reduction efforts back by about 3 years. SDG 1 (No Poverty) has been particularly impacted by multiple crises, and the pandemic halted years of progress toward eradicating extreme poverty. In recent years in China, with the country's layer-by-layer deployment, all rural poor people under the current standard have been lifted out of poverty, as have all 832 poverty-stricken counties, thereby completing the arduous task of eliminating absolute poverty and entering the post-poverty era. Thus, China's focus has shifted from poverty alleviation to stabilizing the outcomes of the poverty alleviation program and establishing a long-term mechanism aimed at alleviating relative poverty. However, the COVID-19 pandemic has presented a new challenge, especially in poverty-stricken mountainous areas, where farmers face not only economic, social, and environmental risks, as well as other risks commonly faced by poor areas but also due to illness, education, disability, accident, industrial project failure, unstable employment, and other specific risk factors for returning to poverty and leading to poverty. China is also facing other severe risks and uncertainties as a result of the epidemic, such as various impacts on migrant workers, agricultural production, children's education, medical treatment for the elderly, and people's future livelihoods and plans. In this context, poverty-stricken areas are exposed to various disturbances and shocks, which present serious obstacles to the survival and development of poor households, resulting in poor livelihood sustainability and persistent or repeated poverty for farmers. One reason is that the cultivation of farmers' livelihood resilience has not received sufficient attention. However, the key to lifting relatively poor groups out of poverty lies in changing the longstanding low status of farmers in areas with fragile ecologies and livelihoods, effectively improving their sustainable livelihood resilience and improving their ability to deal with risks and withstand external shocks, thereby reducing the likelihood of returning to poverty. This is the focus of programs aimed at alleviating poverty among poor mountainous households, and it also guides current rural poverty governance work, that is, boosting poverty alleviation and increasing farmers' income to enhance their sustainable livelihood resilience. Resilience first originated in the field of physics and was then applied to ecology. With the deepening of resilience theory and the expansion of its application, it is now possible to study farmers' sustainable livelihood resilience. The COVID-19 pandemic has provided a major setback to sustainable development, but has also stimulated new ideas that might help advance SDG policies. Therefore, introducing resilience theory into research on poverty alleviation and the sustainable livelihoods of households lifted out of poverty, and exploring the livelihood scenarios of farmers from the perspective of resilience can provide new insights and concepts regarding farmers' sustainable livelihoods. Some scholars have carried out exploratory research on the description and measurement of resilience systems from different perspectives and methods. However, many existing studies lack quantitative and positioning assessment methods and often neglect analysis of the mechanisms among components of resilient systems, limiting the practicality and relevance of results. Therefore, this study further examines resilience concepts and integrates them with traditional livelihood frameworks to conduct quantitative scientific measurement and analysis of farmers' sustainable livelihood resilience. In this context, the authors selected Fugong County, a once deeply impoverished county in Yunnan Province, China, and combined a sustainable livelihood research framework with resilience theory to clarify a descriptive framework of farmers' sustainable livelihood resilience, construct indicators and a model based on buffer capacity, self-organization capacity, and learning capacity, and classify resilience types to analyze spatiotemporal variation from 2015 to 2018 and propose improvement strategies.

The paper reviews resilience evaluation developments, noting the Resilience Alliance's general steps and various domain applications (post-disaster reconstruction, urban resilience, tourism destination resilience) using comprehensive evaluation, scenario simulation, and dynamic models. Household livelihood resilience research typically follows two paths: (1) analyzing resilience around key variables affecting livelihoods and (2) comprehensive evaluations based on traditional frameworks (e.g., social deprivation, vulnerability analysis, sustainable livelihoods) with multi-perspective indicators. Commonly used resilience dimensions include buffer capacity, adaptive capacity, and self-organization capacity; some studies also use income diversity to reflect vulnerability and adaptability. The Farmers' Distress Index is cited as a multidimensional measure focused on vulnerability; however, the authors argue resilience is better captured by buffer capacity, self-organization capacity, and learning capacity. Livelihood capital from sustainable livelihood analysis offers a reference for resilience measurement. Despite progress, no universally recognized resilience measurement paradigm exists; comprehensive indicator-based evaluations remain mainstream but often rely on qualitative descriptions, with limited quantitative and spatially explicit assessments and insufficient analysis of internal mechanisms among resilience components. The paper positions its contribution as integrating resilience characteristics into measurement, enhancing quantitative rigor, and examining interactions among components through coupling coordination and classification approaches.

Study area: Fugong County, Yunnan Province, China, a mountainous border county with seven townships and 57 administrative villages, characterized by rugged terrain, frequent natural disasters (e.g., landslides), and limited transportation and infrastructure. Data: 30 m digital elevation models, 30 m Landsat 8 OLI imagery, Points-of-Interest for infrastructure, administrative boundaries, road networks, targeted poverty alleviation data (2015–2018), and socioeconomic data from the geospatial data cloud, Google Maps, local poverty alleviation office, and statistical yearbooks. Preprocessing included data cleaning, image preprocessing, projection transformation, georeferencing, and topology checks. Framework: Extends the DFID sustainable livelihoods framework by explicitly incorporating resilience dynamics. The descriptive framework centers on livelihood risk (external disturbances such as natural disasters, policy changes, disease; and internal disturbances related to family structure and income sources), the farmers' livelihood system (natural, physical, human, social, financial capital affected by internal and external disturbances), and livelihood resilience (buffer, self-organization, learning capacities). These elements interact dynamically to maintain or transform the system under shocks. Indicator system: An index for sustainable livelihood resilience was built across three dimensions—buffer capacity (natural, human, physical, financial capital), self-organization capacity (public services, social help, life security), and learning capacity (learning paths, means of livelihood, cultural reserves)—with specific indicators and signs/weights (as detailed in Table 1 of the paper). Measurement model: A cloud-model-based multi-level fuzzy comprehensive evaluation model was used to quantify each dimension. Steps: (1) Entropy weight method to derive indicator weights; (2) Expert judgment via a nine-point Likert-type scale mapped into cloud model parameters (expectation Ex, entropy En, and super-entropy He) to build a fuzzy evaluation matrix; (3) Compute final fuzzy comprehensive score P = (Ex + 0.25En + 0.05He) * W for each indicator; (4) Use these as weights to aggregate standardized indicator data to obtain dimension scores (buffer Re, self-organization Ad, learning Tr) and overall resilience R = Re*w1 + Ad*w2 + Tr*w3 with equal weights (w1 = w2 = w3 = 0.333). Standardization removed dimensional differences. Coupling coordination analysis: To assess internal interactions among buffer, self-organization, and learning dimensions, a coupling coordination degree model was applied: C = [(U1*U2*...*Un) / ((U1 + U2 + ... + Un)/n)]^(1/2), then D = √C * T (T is the comprehensive value). The degree D was classified into strong obstruction, weak obstruction, weak promotion, and strong promotion types according to predefined intervals. Typology via decision tree: Based on resilience scores and coupling coordination degree, a decision tree classified villages/households into six development states: stable promotion, benign promotion, stagnation, mild recession, severe recession, and chaotic period, using interpretable rules shown in the paper’s figure and definitions table.

- Overall resilience trend (county level, 2015–2018): Mean resilience increased slightly from 0.421 (2015) to 0.450 (2018). Buffer capacity rose from 0.166 to 0.194, while self-organization capacity remained roughly stable (0.155 to 0.150) and learning capacity hovered around 0.100–0.110. Standard deviations increased modestly, indicating growing disparities. - Spatial patterns: Higher resilience concentrated in central Fugong; lower levels in southwestern and northwestern border regions. At village level, Shangpa, Shidi, and Chisadi consistently exhibited high resilience, while Mujiajia, Wawa, and Dadake were low. - Dimension-specific leaders/laggards: Buffer capacity highest in Buladi; lowest in Dadake. Self-organization capacity highest in Shangpa and Chisadi; lowest in Wawa, Puluo, and Bula. Learning capacity highest in Laomudeng, Bula, Maji, and Shangpa; lowest in Dadake, Weidu, Mujiajia, and Yaduo. - Coupling coordination: Average coupling coordination degree increased from 0.404 (2015) to 0.471 (2018). Top villages by coupling coordination included Shangpa, Chisadi, Shidi, Buladi, and Lazhudi; bottom included Dadake, Mujiajia, and Wawa. Central areas showed higher coordination; northern and southern regions lower, with strong/weak promotion types spreading over time from the center toward north and south. - Interaction insight: Buffer, self-organization, and learning capacities complement and co-develop; deficiency in any one dimension hinders overall resilience and coordinated development. Self-organization improved less than other dimensions, constraining coordination in some areas. - Resilience state types (decision tree classification): Proportion in stable promotion increased from 3% (2015) to 17% (2018); benign promotion from 7% to 14%. Mild and severe recession categories declined by 6% and 8% over the period, respectively. Chaotic category decreased from 31% (2015) to 15% (2018). Stagnation remained substantial, peaking at 36% in 2017. These shifts indicate progress due to assistance programs but persistent challenges for many households.

The study demonstrates that integrating resilience concepts with sustainable livelihoods enables a quantitative assessment of farmers’ capacity to withstand and adapt to shocks in a post-poverty-alleviation context. The rising resilience and coupling coordination, especially in central regions, suggest that targeted programs have strengthened resource endowments and internal synergies. However, uneven improvements—particularly in self-organization capacity—limit coordinated development in northern and southern border areas. The findings address the research question by showing that resilience depends on balanced development across buffer, self-organization, and learning capacities; shortcomings in any dimension propagate constraints across the system. For policy and practice, the results provide spatially explicit guidance to prioritize investments in infrastructure, public services, social governance, and learning channels, fostering multi-field, multi-actor collaboration to enhance internal coordination. The decision-tree-based typology supports differentiated interventions: consolidating gains for promotion groups, unlocking bottlenecks for stagnation groups, and comprehensive assistance for recession and chaotic groups to prevent relapse into poverty and move toward sustainable livelihoods.

The paper develops a descriptive framework and quantitative measurement system for farmers’ sustainable livelihood resilience, operationalized through a cloud-model-based fuzzy evaluation across buffer, self-organization, and learning capacities, coupled with a coordination analysis and a decision-tree typology. Empirically, Fugong County (2015–2018) shows gradual improvements in overall resilience (0.421 to 0.450) and coupling coordination (0.404 to 0.471), with central areas outperforming peripheral regions. High-resilience and highly coordinated villages (e.g., Shangpa, Shidi, Chisadi) contrast with persistently low-performing ones (e.g., Dadake, Wawa, Mujiajia). The share of households in promotion states increased, while recession and chaotic states declined, though stagnation remained sizable. Contributions include: integrating resilience characteristics into a sustainable livelihoods framework; providing a replicable indicator system and cloud-model-based evaluation; revealing internal interactions via coupling coordination; and delivering a practical typology for targeted policy. Future work should extend temporal coverage (including the COVID-19 period and beyond) and analyze mechanisms by which specific factors alter resilience, to strengthen theoretical understanding and guide precise, place-based interventions.

- Temporal scope: Data cover 2015–2018 only, potentially missing important dynamics, especially during the COVID-19 years. - Mechanism analysis: Due to space limits, the study did not examine in depth how specific influencing factors drive changes in resilience levels; future research should analyze these mechanisms and interrelationships among framework elements.