Modeling Climate Change and Its Impact on Migration Intentions: Mediation Through Sociocultural Impact in Gandaki Province, Nepal

Conceptual framework and hypothetical model

The hypothetical model (Fig. 1) suggests that migration intentions are influenced by sociocultural factors, considering both positive and negative aspects. It evaluates how climate change effects, impacts, and adaptation strategies shape migration intentions based on social exchange theory. The model shows that individuals balance perceived benefits (positive sociocultural impacts and effective adaptation) against costs (negative sociocultural impacts and adverse effects of climate change), deciding to stay if benefits surpass costs and migrating otherwise (Chang et al., 2020; Lübke et al., 2021; Zeng and Liao, 2022).

Fig. 1

Conceptual framework of a hypothetical model for migration intention caused by climatic factors and sociocultural impact.

Effect of climate change on Sociocultural migration intention

Nepal’s main climate-related impacts and disasters include increasing floods, landslides, soil erosion, and glacier lake outbursts (GLOFs). The Himalayan glaciers are retreating rapidly, fueling the growth of glacial lakes (Bajracharya et al., 2015) and increasing the risk of outbursts. Floods have destroyed crops, homes, and livelihoods, forcing people into temporary shelters, while droughts have reduced water availability, leading to poor harvests and low yields (Sherpa and Bastakot, 2021). Hazardous events like droughts, floods, and landslides are rising in Nepal, causing 3672 deaths from 2018 to 2024 (MoHA, 2024). This worsens water shortages, changes farming patterns, and increases food insecurity (Sudmeier-Rieux et al., 2012). Soil erosion, a natural process made worse by the complex mountainous terrain of the Himalayas, is especially severe in Nepal due to its intricate landscape (Raya, 2023). The diverse effects of climate change affect social and cultural dynamics related to migration and influence decisions to leave one’s home (Perch-Nielsen et al., 2008; Chen et al., 2021).

Therefore, the hypotheses related to climatic effects and sociocultural impacts are:

Climate change impact and migration intention

Climate change significantly influences migration in vulnerable regions through various pathways (Abir and Xu, 2019). Major triggers include agricultural decline, loss of biodiversity, rapid spread of invasive species, water shortages, prolonged droughts, and climatic disasters. Climate change also impacts social and cultural aspects. In Nepal, rising temperatures, unpredictable rainfall, and extreme weather reduce food crop yields (Joshi et al., 2011) and threaten ecosystems, jeopardizing rural livelihoods and increasing the desire to migrate (Gentle and Maraseni, 2012; Adhikari et al., 2018). Therefore, the hypotheses concerning the effects of climate change on the sociocultural aspects of migration and migration intentions are:

Adaptation and Migration Intention

In Nepal, adaptation is vital for reducing the effects of climate change, especially for smallholder farmers whose perceptions greatly influence their responses (Olana Jawo et al., 2023). Key strategies include crop and livestock insurance, adjusted planting schedules, and enhanced genetic quality of crops to increase productivity. However, unpredictable rainfall, glacial retreat, and landslides are increasingly disrupting agriculture, pushing communities to migrate when local adaptations fail (Gentle and Maraseni, 2012; Adhikari et al., 2018). Sociocultural factors such as gender roles and dependence on remittances influence migration patterns, with men often migrating while women handle household duties (Maharjan et al., 2020). Effective adaptation requires utilizing local knowledge and transforming traditional development methods (Regmi et al., 2013), emphasizing the need for integrated policies that address climate resilience and migration issues. Therefore, the hypotheses related to adaptation, migration intention, and sociocultural aspects of migration are:

Sociocultural impact of migration and intention

Migration has both positive and negative sociocultural impacts that influence migration intentions. On the positive side, it encourages cultural exchange, diversifies local economies, and strengthens transnational networks, boosting social resilience and community. (De Haas, 2014; Koshel and Yakovenko, 2020). remittances enhance household welfare, empower women, improve financial status, create educational opportunities, and support local adaptation, thereby reducing economic pressures (Maharjan et al., 2020). They facilitate technology transfer and promote shifts toward progressive mindsets. Conversely, migration can weaken traditional practices, lead to family disintegration, and increase workloads for those left behind, especially women (Gioli et al., 2014). It also contributes to rising crime, educational gaps, and the emigration of skilled individuals and young people, which may weaken community cohesion and cultural identity, deterring migration for those who value local ties (Adhikari et al., 2018). These dual effects influence migration intentions as individuals balance economic gains against sociocultural costs. Sociocultural factors such as migrants’ networks, community bonds, and remittances mediate the relationship between climate change and migration by directly affecting how environmental stressors influence migration decisions. Gray and Mueller (2012) explained that established migrant networks in urban areas mediate flood-related migration by providing information and economic support. Migrant communities maintain cultural ties through diaspora networks (Adger et al., 2019). Remittances from migrants act as a buffer, reducing immediate displacement while later enabling planned migration (Mallick and Schanze, 2020). Therefore, the hypotheses regarding sociocultural impact and migration intentions are:

Study Area

The study focused on four districts in Gandaki Province (Baglung, Myagdi, Mustang, and Lamjung) because of their different levels of vulnerability to climate change and high rates of indigenous migration. Myagdi and Baglung are extremely vulnerable (0.778–1.0), while Mustang and Lamjung are highly vulnerable (0.623–0.777) (MOFE, 2021). In terms of glacier lake outburst flood (GLOF) risk, Mustang is rated very high (0.911–1.0), Lamjung is high (0.767–0.910), and both Baglung and Myagdi are low (0.001–0.597) (MOFE, 2021). According to the combined exposure index, Lamjung is extremely vulnerable, Baglung and Myagdi are highly vulnerable, and Mustang has moderate vulnerability (MOFE, 2021). The study area map is shown in Fig. 2. These districts are mainly inhabited by indigenous communities, where migration is a common occurrence. Their indigenous populations and migration patterns make them suitable for purposive sampling.

Fig. 2

Map of the study area.

Population and Sample

The four districts examined have a combined population of 526,548, with Mustang experiencing positive growth, while Baglung, Myagdi, and Lamjung show negative growth (NSO, 2021). The study area spans 9,346 km² and features a diverse topography that ranges from upper tropical to alpine zones. A total of 493 respondents were selected from three urban municipalities and seven rural municipalities. The sample size was determined using the Kish (1965) formula.

Where C represents the coefficient of variation for the Likert item (ranging from 0.1 to 1), K denotes the number of constructs (from 1 to 10), D indicates the error (relative tolerance) (1 to 10%), and ρ is the correlation coefficient (between 0.1 and 0.7). Using Zα/2 as 1.96, C as 0.8, K as 10, ρ as 0.4, and D as 5%, the calculated value of n is 452. Including a 9% non-response margin results in a sample size of 493 (i.e., 109% of 452 = 493).

Data were collected using structured questionnaires that focused on various drivers of migration, as perceived by the respondents. The questionnaires utilized a five-point Likert scale and were administered through interviews. Data collection took place in two phases: the first from August 5, 2022, to September 23, 2022, and the second from November 23, 2022, to December 11, 2022.

Sampling design and data analysis

Four districts (Baglung, Myagdi, Mustang, and Lamjung) were selected purposively. Two districts (Baglung and Myagdi) from very high vulnerability and the other two districts (Mustang and Lamjung) from high vulnerability were chosen based on the vulnerability index (MoFE, 2021). From each of these districts—Baglung, Myagdi, and Lamjung—one rural municipality and one municipality were randomly selected, while four rural municipalities (out of five) were chosen from Mustang district. The household (HH) survey was conducted with the heads of the households. If the household head was unavailable, an adult member of the household was interviewed. This study adopted a quantitative, cross-sectional design, analyzing data using Smart-PLS software. The research employed PLS-SEM, an effective method for managing non-normal data and small sample sizes. The analysis involved two stages: first, the measurement model, which assesses the indicator-construct relationships; and second, the structural model, which tests the hypothesized relationships among variables.

Background Characteristics of Respondents

The demographic and socio-economic characteristics of 493 sample respondents from four districts of Gandaki province are presented in Table 1.

Table 1

Description of sample respondents (N = 493)

Table 1 indicates that over 60 percent of respondents were male, more than 80 percent owned their homes, and over 90 percent were married. Additionally, more than 65 percent had completed at least secondary education, and over 75 percent resided in permanent housing structures. These results emphasize the diverse socio-economic and demographic makeup of the study population.

To explore residents’ perceptions of various factors related to climate change and their link to migration intentions in the study area, we developed multiple measured (manifest) variables associated with different constructs. The six main constructs analyzed in this study include ‘Climate Change Effect,’ ‘Climate Change Impact,’ ‘Adaptation to Climate Change,’ ‘Positive Sociocultural Impact,’ ‘Negative Sociocultural Impact,’ and ‘Migration Intention.’ The original set included 53 indicators from the literature to assess these constructs. We retained only 29 indicators with strong factor loadings (greater than 0.6) in the measurement model. The selected indicators are detailed in Table 2.

Table 2

The indicators related to the constructs

Measurement Model Assessment

The research model was analyzed using the Partial Least Squares (PLS) method, implemented through Smart-PLS 3.0 software developed by Adger et al. (2013) and Ringle et al. (2022). Following the two-stage analytical approach recommended by Hair Jr. et al. (2014), the study assessed both the measurement and structural models. Additionally, the significance of the path coefficients was examined using the bootstrapping method, as suggested by Hair et al. (2022). The model’s reliability was tested with Cronbach’s alpha (α), Rho_A, and composite reliability, while convergent validity was evaluated through the Average Variance Extracted (AVE) presented in Table 3.

Table 3

Convergent validity and reliability

Table 3 examines the reliability and convergent validity of constructs related to climate change and migration. All six constructs show high internal consistency, with Cronbach’s Alpha, Rho_A, and Composite Reliability (CR) values above 0.7. The Average Variance Extracted (AVE) values for each construct exceed the 0.5 threshold, confirming acceptable convergent validity. While most indicators have strong loadings (> 0.7), a few (e.g., C04.4, F02a.7) show slightly lower loadings (0.6–0.7), indicating they are somewhat less reliable measures of their respective constructs. After assessing reliability and convergent validity, the next step is to determine if each construct represents a distinct concept compared to others, even when they are related. This is confirmed through the discriminant validity assessment, as shown in Table 4.

Table 4

Discriminant validity (Fornell-Larcker and HTMT)

Table 4 demonstrates discriminant validity, as the square roots of the Average Variance Extracted (AVE) values (shown as diagonal elements) exceed all inter-construct correlation coefficients (lower off-diagonal values). Furthermore, all Heterotrait-Monotrait (HTMT) ratios (upper off-diagonal values) remain below the 0.85 threshold, which further confirms that the constructs are distinct and supports the establishment of discriminant validity.

Structural Model Assessment

After evaluating reliability and validity, the next step was to develop the structural model to examine the relationships between exogenous and endogenous constructs. To assess the significance and robustness of these relationships, a bootstrapping method utilizing 5,000 subsamples was employed. This technique approximates the normal distribution of the data and provides t-values for factor loadings and path coefficients (Hair et al., 2021). The complete structural model, illustrating the relationships between the constructs and their associated indicators, is shown in Fig. 3 below.

Fig. 3

Structural model representing the link among constructs.

Table 5 shows the analysis of the structural model that tested eleven hypotheses about direct relationships between the constructs: climate change effects (CCE), climate change impacts (CCI), adaptation (ADPT), perception of positive sociocultural impact (SCPI), negative sociocultural impact of migration (SCNI), and migration intention (MI). The path coefficients (β), standard deviations (SD), t-values, and p-values were evaluated at a five percent significance level (α = 0.05).

Table 5

Results of the structural model path coefficient (direct relationships)

Table 5 shows that climate change strongly increases migration intentions (β = 0.420, p < .001), making it a major factor driving migration. While this effect has a modest but meaningful positive link to the socio-cultural benefits of migration (β = 0.113, p = .036), it slightly reduces the socio-cultural negative impacts (β = −0.101, p = .013). In contrast, climate change impacts do not have direct effects on either migration intentions (β = −0.066, p = .226) or socio-cultural benefits (β = 0.080, p = .131), but they significantly increase socio-cultural negative impacts (β = 0.608, p < .001), which then negatively affect migration intentions (β = −0.111, p = .017). Adaptation strategies positively affect socio-cultural benefits (β = 0.190, p < .001) but do not have a significant direct relationship with migration intentions (β = −0.014, p = .765) or socio-cultural negatives (β = 0.032, p = .410). Notably, socio-cultural benefits themselves are a positive predictor of migration intentions (β = 0.205, p < .001), indicating that climate change influences migration through both direct and socio-cultural pathways. These findings suggest that while climate change is a key driver of migration, its effects are mediated by socio-cultural factors, with adaptation strategies playing a limited but specific role in reducing negative outcomes. Further research could examine mediation effects and other contextual variables to better understand these complex relationships.

To test the mediation analysis on whether climatic factors indirectly influence migration intention through individuals’ perceptions of both positive and negative sociocultural impacts, we proposed six hypotheses, which are shown in Table 6.

Table 6

Results of mediation analysis (Indirect Relationship)

The mediation analysis provides key insights into how community residents view the connection between climate change and migration intentions. Two statistically significant pathways emerged from the data (p < .05), showing how the sociocultural effects of migration mediate this relationship. First, climate change impacts (CCI) were found to lower migration intention by increasing negative sociocultural consequences (β = −0.067, p = .023), reflecting how environmental damage to community structures and traditions can make people hesitant to leave. Second, effective adaptation strategies (ADPT) increased the likelihood of migration by promoting positive sociocultural impacts (β = 0.039, p = .004), demonstrating how measures that protect community benefits can encourage more voluntary movement.

Other tested pathways were not significant (p > .05), indicating that general climate change effects (CCE) do not influence migration intentions solely through sociocultural perceptions. This pattern aligns with Nepal’s specific context, where environmental changes are often assessed based on their immediate social impacts rather than as abstract phenomena. The findings are especially relevant in communities where climate impacts, such as land-slides or droughts, directly threaten social cohesion, while adaptation efforts like changing cropping patterns or migration support programs help maintain vital social networks.

These results suggest that interventions should aim to protect communities from climate-related social disruption while also developing adaptation programs that maintain the positive aspects of migration. The analysis emphasizes that for many Nepalese, migration choices are less driven solely by environmental factors and more influenced by how these factors affect their social networks and community connections.

The study found that climate change significantly affects sociocultural impacts and migration intentions. Specifically, climate change has a strong positive effect on exacerbating the negative sociocultural impacts of migration. Adaptation strategies notably enhance the positive sociocultural impacts. The analysis reveals that positive sociocultural impacts significantly influence migration intentions. Conversely, negative sociocultural impacts notably decrease migration intentions. These findings demonstrate how climate change affects migration decisions through its impact on sociocultural factors. Positive sociocultural impacts of migration mediate the relationship between adaptation to climate change impact and migration intention. Negative sociocultural impacts of migration mediate the relationship between the effects of climate change and migration intention.

The results show that climate effects and climate change impacts greatly influence migration intentions, aligning with earlier research (Black et al., 2011; Obokata et al., 2014). Participants who experienced extreme weather events like floods, landslides, and droughts were more likely to see migration as a practical adaptation strategy. This supports the wider literature on environmental migration, which suggests that climate-related stressors often worsen existing vulnerabilities, leading people to seek new livelihoods elsewhere (Adger et al., 2013). The study also highlights how perceived impacts of climate change, such as lower agricultural productivity and water shortages, influence migration choices, emphasizing climate change as a major factor driving rural-to-urban migration (Mallick and Schanze, 2020).

Adaptation to climate change has become a key factor influencing migration intentions. Participants who implemented adaptation measures, such as diversifying crops, securing insurance for livestock and crops, conserving water, and engaging in community-based disaster management, were less likely to have a strong desire to migrate. This supports the idea that effective adaptation measures can reduce migration (Smit and Wandel, 2006). However, the study also showed that when adaptation efforts are seen as inadequate or ineffective, migration may become more appealing. This highlights the importance of strengthening local adaptive capacities to prevent forced migration caused by climate change (Foresight, 2011).

The study’s mediation analysis shows that competing perceptions of sociocultural impacts influence migration intentions. Positive factors such as economic improvement, educational access, and cultural exchange encourage migration in accordance with the “migration as adaptation” framework (Black et al., 2011). Conversely, negative aspects like social dislocation, cultural erosion, and family separation act as deterrents, especially in collectivist contexts like Nepal (Hugo, 2011).

These findings emphasize the need for policymakers to create balanced strategies that boost local adaptive capacity through climate-resilient infrastructure and community-based solutions, while also addressing the sociocultural risks of migration with targeted support and initiatives to preserve culture.