“I can migrate, but why should I?”—voluntary non-migration despite creeping environmental risks

The study addresses why many people exposed to environmental risks choose not to migrate, a phenomenon termed environmental non-migration. While environmental change is expected to increasingly influence human mobility, non-migration remains under-examined relative to migration. The paper defines environmental non-migration as staying despite environmental threats, and voluntary non-migration as the combination of the capability and aspiration to stay. Based on gaps in empirical and conceptual guidance, the study pursues two objectives: (i) to explore how societal and environmental factors drive non-migration decisions and how these relate to adaptive capacity and resilience; and (ii) to examine the relationship between non-migration and livelihood resilience and its spatial variability. Using five cyclone-affected communities in southwest coastal Bangladesh (Cyclone Aila, 2009), the authors distinguish voluntary and involuntary migrants and non-migrants, investigate past and future motivations, and develop a Livelihood Resilience Index (LRI) to analyze links between resilience and (non-)migration decisions.

The literature has often framed non-migrants as the inverse of migrants, overlooking the richness of non-migration as a distinct phenomenon. Psychosocial drivers—especially place attachment, social identity, and community belonging—emerge as key reasons for staying. Conceptual models such as the aspiration–capability framework suggest (non-)migration lies on a continuum from voluntary to involuntary, including ‘trapped’ populations lacking resources to realize migration aspirations and ‘voluntary non-migrants’ who choose to remain due to family ties, social networks, heritage, or livelihood strategies. The paper positions livelihood resilience as central to voluntary non-migration: individuals or households able to cope and adapt can sustain staying as an adaptive strategy. It clarifies linked concepts of vulnerability, resilience, and adaptive capacity, emphasizing context-specific social and environmental characteristics that shape (non-)migration decisions. The review highlights the need to integrate societal (social, economic, political, technical) and environmental (climate/weather, water, soils, biodiversity) dimensions to understand non-migration beyond economic drivers or simple push–pull models.

Analytical concept and operationalization: The authors propose a hierarchical analytical framework where individual and household decisions are embedded within community socio-environmental contexts. Non-migration is assessed through factors affecting livelihood resilience operationalized by measurable indicators. Four outcome categories based on aspiration and capability are conceptualized: voluntary non-migrants, involuntary non-migrants (trapped), voluntary migrants, and involuntary migrants. Factors and indicators span societal (social, economic, political, technical) and environmental (climate/weather, water, soils, biodiversity) dimensions, with example indicators such as social networks, place attachment, land/asset ownership, trust in institutions, infrastructure quality, seasonal patterns, water salinity and access, soil fertility, and biodiversity status. Data and study area: An empirical case study was conducted in five villages in southwest coastal Bangladesh, selected for differential cyclone exposure: two highly exposed (Padmapukur, Chakdah), two moderately exposed (Yaharpur, Vabanipur), and one less exposed (Panchakari). All villages were affected by Cyclone Aila (2009), providing a temporal baseline. Fieldwork occurred March–April 2018, using mixed methods: 38 in-depth interviews, 7 group discussions, and a structured household survey of 195 households administered via KoboCollect and analyzed in R. The questionnaire covered socioeconomics, demographics, place attachment, networks, associations/politics, hazard frequency, crop and seasonality, and environmental stressors (salinity, water stress, siltation, erosion). Livelihood Resilience Index (LRI): A total of 45 indicators across eight factors (four societal and four environmental) were used to construct a household-level resilience index. Steps: (i) data screening, outlier removal, multicollinearity checks, and Max–Min normalization to [0,1]; (ii) equal weighting via Simple Additive Weighting (SAW); (iii) aggregation to factor scores and then to an overall RI by averaging equally weighted factors (social, political, economic, technical, climate, soils, water, biodiversity). RI values were scaled to range from −1 (lowest resilience) to 1 (highest) following prior methods. Classification followed standard deviations around the mean and thresholds: RI < 0 (less resilient), 0–0.4 (moderate), and >0.4 (resilient. The analysis related RI to post-Aila migration behavior and future migration aspirations across villages.

• Approximately 40% of households reported at least one family member migrated outside their community for alternative income sources after Cyclone Aila (2009); nearly 50% in Padmapukur, 33% in Chakdah, and 24% in Panchakari. - About 94% of migrant households reported that their primary family member returned once the breached dam was repaired, indicating a strong tendency toward staying when local conditions improve. - Respondents were distributed into four categories (voluntary/involuntary migrants and non-migrants) based on aspirations and capabilities; many intended to migrate but could not (forced non-migrants), while others chose to stay despite risks (voluntary non-migrants). - Qualitative evidence shows strong drivers of non-migration in the societal dimension: place attachment; dense kin and social networks; perceived security and support; livelihood opportunities (agriculture, aquaculture, fishing); improved infrastructure (e.g., paved roads, transport); and concerns about uncertainty and costs in destination areas. - Environmental perceptions and adaptations also support staying: familiarity with local hazards, community-led embankment management, perceived better environmental quality (air, water, food) than cities, and normalization/adaptation to issues like salinity; for some, environmental changes (e.g., shrimp farming expansion, water quality) shape livelihood shifts rather than migration. - LRI construction and categorization revealed substantial variation across households and villages. While resilience relates to the voluntary nature of decisions once made, resilience alone could not predict whether a household stayed or migrated. - Future aspirations: nearly four out of five respondents expressed a desire to stay; among those who migrated after Aila, the vast majority preferred to stay in the future, and only about one-fifth wished to migrate. Among those who did not migrate after Aila, roughly 12% expressed a desire to migrate. Notably, many less-resilient households also preferred to stay, underscoring that aspiration and capability interact with resilience in complex ways. - Seasonal and temporary migration can serve as strategies to support long-term staying (voluntary non-migration), indicating that mobility and immobility are not mutually exclusive in adaptation strategies.

The findings support the conceptualization of environmental non-migration as a distinct, complex outcome shaped by intertwined societal and environmental factors. The analytic framework clarifies how aspirations and capabilities, mediated by livelihood resilience, contribute to four decision outcomes along a voluntary–involuntary continuum. Empirically, strong place attachment, social networks, and local livelihood opportunities underpin voluntary non-migration, while infrastructural improvements reduce perceived benefits of urban migration. Environmental adaptations (e.g., embankment strengthening, acceptance of salinity, resource use changes) can reduce migration pressure. The LRI is informative about the voluntary character of decisions but insufficient to predict staying or migrating on its own; psychosocial factors and event characteristics (e.g., fast-onset cyclones versus slow-onset salinization) remain decisive. The results imply that adaptation planning should not assume mobility as the default response to environmental risk; instead, policies must recognize and support voluntary non-migrants, including through livelihood support, infrastructure, and social services, while also addressing the needs of involuntary non-migrants who may be ‘trapped’.

The paper advances an analytical concept to study environmental non-migration by integrating societal and environmental factors of livelihood resilience and mapping outcomes through aspiration and capability. Operationalized with a Livelihood Resilience Index and a Bangladesh case study, the approach shows that resilience relates to the voluntary nature of staying or moving but does not by itself predict migration decisions. Non-migration is multi-faceted, context- and person-specific, and often involves selective, temporary mobility as part of a staying strategy. The study contributes a structured way to assess non-migration and highlights the importance of psychosocial drivers, community context, and hazard characteristics. Future research should: (i) conduct longitudinal studies to capture temporal dynamics of aspirations, capabilities, and resilience; (ii) examine multi-scalar influences (household, community, regional); (iii) integrate richer psychosocial measures (place attachment, identity, risk perception); and (iv) differentiate responses to fast- versus slow-onset environmental changes.

• Resilience index limitations: While LRI captures multiple livelihood dimensions, it cannot alone predict stay/migrate decisions; psychosocial and contextual factors remain under-measured. - Cross-sectional timing: Data collected nearly a decade after Cyclone Aila (2009) may be affected by recall and post-event changes; longitudinal evidence is needed to track evolving aspirations and capabilities. - Context specificity: Findings from five villages in southwest coastal Bangladesh may not generalize to other socio-ecological settings with different hazards, institutions, or cultural contexts. - Measurement constraints: Indicator selection, equal weighting (SAW), and normalization choices influence RI results; alternative weighting or modeling approaches could yield different patterns. - Event heterogeneity: The framework acknowledges fast- versus slow-onset events, but the single-case focus limits comparative inference across hazard types.