Cumulative Climate Stress Drives Migration and Stratifies Mobility: Evidence from Ethiopia, India, Peru, and Vietnam

Cumulative Climate Stress Drives Migration and Stratifies Mobility: Evidence from Ethiopia, India, Peru, and Vietnam | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Cumulative Climate Stress Drives Migration and Stratifies Mobility: Evidence from Ethiopia, India, Peru, and Vietnam Héctor Cebolla Boado, Michael Lund This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-9405620/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted You are reading this latest preprint version Abstract Climate change is widely expected to influence human migration, yet most individuals remain immobile despite environmental stress, challenging event-based displacement models. Existing research often overlooks how repeated environmental shocks accumulate over time, reshaping mobility decisions. Here we analyze longitudinal data from Ethiopia, India, Peru, and Vietnam using fixed-effects models to distinguish between acute and cumulative climate stress. We find that contemporaneous shocks have weak effects on migration, whereas cumulative exposure strongly increases mobility. Socioeconomic resources buffer migration in the short term but enable it as stress accumulates, creating a temporal divergence where resource-poor households become trapped while wealthier ones retain adaptive flexibility. These findings suggest that migration under climate change is a process of cumulative vulnerability and stratified adaptation, highlighting the emergence of structurally trapped populations and challenging simplistic event-driven migration models. Scientific community and society/Social sciences/Sociology Scientific community and society/Social sciences/Climate change/Climate-change impacts Climate Shocks Cumulative disadvantage Trapped populations Life course Longitudinal analysis Figures Figure 1 Figure 2 Figure 3 Introduction Climate change is widely expected to intensify human mobility, a view consolidated in early global assessments (IPCC, 2007) and subsequent syntheses of the climate–migration nexus (Cattaneo et al., 2019; Rigaud et al., 2021). Yet empirical evidence accumulated over the past decade consistently shows that most populations exposed to environmental stress do not move, giving rise to the so-called (im)mobility paradox (Boas et al., 2019; Zickgraf, 2021). More recent work emphasizes that migration responses are heterogeneous and strongly mediated by socioeconomic and institutional conditions rather than directly triggered by environmental shocks (Black et al., 2011; Hunter et al., 2015; Cattaneo et al., 2026). However, this apparent inconsistency may reflect not the absence of an effect, but a mischaracterization of how environmental stress operates over time. Most empirical studies operationalize climate exposure as a contemporaneous shock (a drought, flood, or temperature anomaly) and estimate its immediate effect on migration (Bohra-Mishra et al., 2014; Cattaneo & Peri, 2016; Hoffmann et al., 2020). This approach implicitly assumes that mobility is a short-term response to discrete events. Yet environmental stress often unfolds cumulatively, through the progressive erosion of livelihoods, assets, and adaptive capacity (Hoffmann et al., 2020; Hunter et al., 2015). This perspective aligns with broader theories of cumulative disadvantage, in which repeated exposures compound over time to reshape life-course outcomes (Dannefer, 2003). Here we show that this temporal misspecification leads to systematically biased conclusions. Using longitudinal individual data from Ethiopia, India, Peru, and Vietnam, we distinguish between contemporaneous and cumulative exposure to environmental shocks within individuals over time. We find that contemporaneous shocks have weak and inconsistent effects on migration, whereas cumulative exposure produces large, monotonic increases in mobility. Models based solely on current shocks therefore underestimate the relationship between climate stress and migration. We further show that the role of socioeconomic resources reverses across these time horizons. In the short term, resources buffer mobility, allowing households to absorb shocks without relocating. As exposure accumulates, however, these same resources enable migration. This dynamic generates a structural divergence: resource-poor households become increasingly unable to move despite rising environmental pressure, while wealthier households retain adaptive flexibility. This pattern is consistent with theories of involuntary immobility and mobility traps (Carling, 2002; Nawrotzki & DeWaard, 2018; Schewel, 2020). Climate-related migration thus emerges not as a direct response to environmental events, but as a process of cumulative vulnerability that stratifies mobility across populations. Climate stress and migration Early theoretical accounts treated environmental degradation and extreme weather events as mechanical push factors expected to uproot large populations—the so-called "climate refugee" narrative (Rigaud et al., 2021 ). This deterministic view predicted a near-automatic relationship between climatic deterioration and displacement, projecting hundreds of millions of forced migrants by mid-century (Black et al., 2011 ). Empirical evidence has systematically failed to support these projections. Research across a range of geographic and climatic contexts consistently shows that most people remain in place even under severe environmental stress, and that when migration does occur it is rarely a simple, direct response to climatic exposure alone (Boas et al., 2019 ; Cattaneo et al., 2019 ; Zickgraf, 2021 ). This disconnection between theoretical expectation and empirical reality has prompted a substantial reorientation in the literature. A growing body of work now emphasizes that climate–migration linkages are mediated by economic, social, and institutional factors that condition the capacity and willingness to move (Black et al., 2011 ; Cattaneo et al., 2019 ; Hunter et al., 2015 ). Studies have documented non-linear relationships between temperature anomalies and migration (Bohra-Mishra et al., 2014 ), shown that effects vary substantially by income level and agricultural dependence (Cattaneo & Peri, 2016 ), and highlighted the importance of migration networks, institutional quality, and livelihood diversification as moderating factors (Hoffmann et al., 2020 ). What has emerged from this literature is a picture of substantial heterogeneity: environmental shocks do not trigger uniform mobility responses but interact with pre-existing socioeconomic and demographic conditions to produce divergent outcomes across individuals, households, and contexts (Nawrotzki & DeWaard, 2018 ). Cumulative vulnerability and the temporal dimension of climate stress. A significant limitation of much existing research lies in how the temporal dimension of environmental exposure is operationalized. As a result, existing estimates may be biased toward zero, as contemporaneous shocks simultaneously generate displacement and immobility effects that offset each other. By modeling migration as a function of contemporaneous shocks, this literature implicitly assumes that the relevant causal pathway runs from acute exposure to rapid mobility response. Yet livelihood degradation under climate change is often gradual and cumulative, operating through the progressive erosion of agricultural yields, water availability, and household asset stocks rather than through single catastrophic events (Hoffmann et al., 2020 ; Hunter et al., 2015 ). The concept of cumulative vulnerability captures this dynamic: repeated exposure to environmental stress compounds over time, progressively narrowing households' adaptive options and eventually shifting the cost-benefit calculus of migration (Bohra-Mishra et al., 2014 ). This perspective aligns with broader sociological theories of cumulative advantage and disadvantage, which posit that small initial differences can compound over time into substantial inequalities in life trajectories (Dannefer, 2003 ). As a result, approaches centered on discrete shocks may understate the role of sustained environmental pressure in shaping mobility decisions over time. Empirical evidence on cumulative climate effects on migration remains scarce, in part because most available data sources lack the longitudinal depth required to track both shock exposure and mobility outcomes across multiple periods for the same individuals. Cross-sectional studies and short panels are poorly suited to distinguishing between immediate displacement effects and the slower-acting consequences of repeated exposure. This limitation is particularly consequential given that cumulative processes may structure not only whether households migrate, but when they do so. This paper addresses this gap directly by exploiting five rounds of panel data and constructing an explicit measure of cumulative shock exposure that captures the accumulation of environmental stress over the observed life course. Immobility, trapped populations, and resources. Alongside the study of climate-driven migration, a parallel literature has developed around the conditions that prevent mobility even when environmental conditions deteriorate severely. Early work conceptualized immobility primarily as a sign of vulnerability or failure to adapt (Carling, 2002 ), but more recent scholarship has reframed it as a socially structured outcome that may itself represent a form of active, if constrained, adaptation (Schewel, 2020 ; Zickgraf, 2021 ). This reframing distinguishes between voluntary immobility—a deliberate choice to remain in place—and involuntary immobility, in which households facing deteriorating conditions lack the resources, networks, or capabilities required to migrate (Carling, 2002 ; Schewel, 2020 ). The concept of mobility traps captures the most severe form of involuntary immobility: situations in which precisely the environmental stress that should motivate migration simultaneously destroys the assets and liquidity required to finance it (Nawrotzki & DeWaard, 2018 ). Resource constraints operate as a double bind—households are both more exposed to environmental risk and less able to escape it through mobility. Empirical evidence for this mechanism remains limited, largely because documenting it requires data that simultaneously track asset shocks, household wealth, and migration decisions over time. Studies from sub-Saharan Africa have provided suggestive evidence of trapped population dynamics (Nawrotzki & DeWaard, 2018 ; Rigaud et al., 2021 ), but cross-national comparative evidence is scarce, and the interaction between the time horizon of exposure and the role of resources has not been systematically examined. In particular, while prior research has documented that socioeconomic resources condition migration responses, less attention has been paid to how their role may evolve across different stages of environmental exposure, potentially shaping not only the likelihood of movement but the capacity to translate accumulated stress into mobility. Taken together, these strands of research highlight the importance of both temporal dynamics and socioeconomic heterogeneity in shaping climate–migration linkages, but they have largely been studied separately. This paper integrates these perspectives by distinguishing between contemporaneous and cumulative exposure and by examining how the role of socioeconomic resources varies across these time horizons. In doing so, it shows that migration responses to climate stress are structured less by immediate reactions to isolated events than by the accumulation of environmental pressure over time, and that the effect of resources is not constant but reverses as exposure accumulates. This joint dynamic generates divergent mobility outcomes across households and provides a micro-level specification of the conditions under which mobility traps emerge. Although prior research has documented that resources condition migration responses, less attention has been paid to how their role evolves across different time horizons of environmental exposure. Building on this limitation, we conceptualize migration responses to climate stress as a dynamic process shaped by the interaction between the accumulation of environmental constraints and the changing role of resources over time. This framework yields four testable hypotheses. H1. Contemporaneous shocks and migration. Exposure to a contemporaneous environmental shock will have a weak or null effect on migration. Acute shocks simultaneously reduce household liquidity and increase coordination costs, such that their net effect on mobility is expected to be negligible or ambiguous. Households may absorb short-term shocks through asset drawdowns, credit, or informal transfers without resorting to geographic relocation, reflecting the short-term disruption generated by acute events. H2. Cumulative shock exposure and migration. Cumulative exposure to environmental shocks will be positively and progressively associated with migration. As the accumulation of stress over time erodes livelihood security and depletes coping reserves, the probability of both community-level and rural-to-urban migration is expected to increase in a graded, dose-response pattern consistent with cumulative vulnerability processes. H3. Socioeconomic resources as short-term buffer. Among households exposed to a contemporaneous shock, higher socioeconomic status will be associated with lower migration probabilities. Material resources allow better-off households to absorb acute environmental events without relocating, dampening immediate displacement responses. The buffering effect of wealth is expected to be observable as a negative interaction between contemporaneous shock exposure and household wealth, consistent with a short-term stabilizing role of resources. H4. Socioeconomic resources as long-term enabler and the mobility trap. In contrast to the short-term buffering effect, as cumulative shock exposure increases, the association between socioeconomic status and migration will turn positive and strengthen. While resource-poor households remain in place despite repeated environmental deterioration—lacking the means to finance or organize relocation—wealthier households retain the capacity to translate accumulated stress into migration. This asymmetry implies a positive interaction between cumulative shock exposure and household wealth and constitutes the core empirical signature of a climate-related mobility trap among the most resource-constrained. Taken together, these hypotheses imply that climate change does not uniformly increase mobility, but reorders who is able to move over time. Study context The analysis draws on four countries—Ethiopia, India, Peru, and Vietnam—that combine substantial exposure to climate variability with differing migration systems and socioeconomic conditions. All four contexts are characterized by a strong reliance on climate-sensitive livelihoods, particularly agriculture, but differ in the nature of environmental risks and mobility opportunities. Ethiopia represents a drought-prone, low-asset context where mobility constraints are likely to be binding. India combines monsoon variability with widespread distress migration. Peru exhibits high baseline mobility linked to climatic variability, while Vietnam reflects a flood-dominated environment with established internal migration systems. This cross-country variation is not merely descriptive but analytically central to the identification strategy. The four settings differ substantially in their climatic regimes, exposure to environmental shocks, and migration systems, providing independent sources of variation in both the intensity and temporal structure of climate stress. This heterogeneity allows us to assess whether the relationship between cumulative exposure and migration reflects a general demographic mechanism rather than context-specific dynamics tied to a particular environmental or institutional setting. A more detailed description of each country is provided in Appendix A.1. DATA: The Young Lives Study The analysis draws on data from the Young Lives longitudinal study, a panel survey conducted in Ethiopia, India (Andhra Pradesh and Telangana), Peru, and Vietnam. The study follows approximately 12,000 children from 2002 to 2016, providing repeated observations on household conditions and life trajectories over time. Its longitudinal structure makes it well suited to the analysis of processes—such as migration—that unfold dynamically. Young Lives tracks two cohorts recruited in 2002: a Younger Cohort (aged 6–18 months at baseline) and an Older Cohort (aged 7.5–8.5 years). Children were sampled from sentinel sites designed to capture variation across rural and urban areas, ecological zones, and population groups. While not nationally representative, the data span substantial socioeconomic and geographic heterogeneity within each country. The analysis uses data from Rounds 1 to 5 (2002–2016). We restrict the sample to children observed in all rounds, ensuring that each individual contributes four inter-wave transitions and allowing consistent measurement of cumulative exposure. Attrition is low (6.5% overall), though partly driven by migration, implying that estimates should be interpreted as conservative. Two migration outcomes are constructed from residential histories. The first is community migration, defined as any change in community between survey rounds. The second is rural-to-urban migration, defined for individuals of rural origin who move to an urban area between waves. The key explanatory variables distinguish between contemporaneous and cumulative exposure to environmental shocks. The contemporaneous measure is a binary indicator of whether the household reported experiencing at least one environmental shock (drought, flood, erosion, or frost) between survey rounds. The cumulative measure counts the number of prior periods in which such shocks were reported, excluding the current interval. This construction isolates accumulated exposure from contemporaneous shocks and enables a direct test of cumulative effects. Shock measures are self-reported and capture experienced environmental stress. Appendix A.2 shows that reported drought exposure exhibits persistence over time, supporting the use of cumulative measures. Appendix A.3 further demonstrates that self-reported droughts are strongly associated with objective precipitation deficits in most contexts, indicating that these measures reflect underlying environmental conditions rather than reporting bias. Socioeconomic heterogeneity is measured using the Young Lives household wealth index, a continuous measure based on assets, housing conditions, and access to services. We use this index to examine how the effects of contemporaneous and cumulative shocks vary across the wealth distribution through interaction terms. Table 1 reports descriptive statistics by country. Table 2 presents within-child variation in migration outcomes and shock exposure, corresponding to the identifying sample in the fixed-effects models. Table 1 Descriptive statistics by country Country N Community migration (%) Rural→Urban (%) Urban→Rural (%) Any shock (%) Cumulative shocks (mean) High education (%) Wealth index (mean) Ethiopia 11,146 12.42 2.70 1.14 26.32 0.30 8.77 0.32 India 11,406 11.01 2.58 1.02 13.96 0.18 17.04 0.49 Peru 10,171 29.52 3.25 1.35 13.21 0.14 42.70 0.52 Vietnam 11,358 11.09 2.30 0.21 10.82 0.14 30.48 0.56 Note: N refers to person-wave observations. Community migration = changed community between waves. Rural→Urban and Urban→Rural are directional transitions. Any shock = at least one environmental shock between waves (drought, flood, erosion, frost). Cumulative shocks = mean number of prior waves with at least one shock. High education = share with secondary or higher. Wealth index = mean value (0–1 scale). Table 2 Within-child variation in outcomes and shocks, and estimating sample Country Children (N) Migration varies (%) Shock varies (%) Identifying sample (%) Ethiopia 2,895 28.36 55.20 13.82 India 2,942 26.65 41.10 9.86 Peru 2,674 55.98 33.60 17.02 Vietnam 2,972 30.69 34.70 8.14 Note: Migration varies = share of children with at least one change in community migration status across waves. Shock varies = share of children with variation in shock exposure across waves. Identifying sample = share of children with within-child variation in both the outcome and the predictor (estimating sample for child fixed-effects models). Methods The empirical strategy exploits within-child variation over time to identify the effect of environmental shock exposure on migration decisions. While prior work relies on sub-national panel variation (Chen & Mueller, 2018 ), our design uses individual-level fixed effects, allowing for a stricter identification of within-household exposure to environmental shocks. The core identification assumption is that, conditional on child fixed effects, changes in shock exposure over time are not systematically correlated with time-varying unobservables that independently predict migration. Child fixed effects eliminate all stable sources of confounding at the individual and household level, including baseline socioeconomic status, risk preferences, location-specific factors, and time-invariant household characteristics, that would otherwise confound the estimated relationship between shocks and mobility. Because the models are estimated on inter-wave transitions (t to t + 1), temporal ordering between shock exposure and migration outcomes is ensured by design: shocks are measured during or prior to the inter-wave period, and migration is recorded at the subsequent survey round. We estimate linear probability models (LPM) of the following form: M i ₜ = α i + β₁ Shock i ₜ + β₂ CumShock i ₜ + γ X i ₜ + ε i ₜ where M i ₜ is a binary indicator equal to 1 if child i migrated between waves t and t + 1 , α i are child fixed effects, Shock i ₜ is the contemporaneous shock indicator, CumShock i ₜ is the cumulative prior shock count, X i ₜ is a vector of time-varying controls, and ε i ₜ is an idiosyncratic error term. Standard errors are clustered at the child level to account for serial correlation in the error term across waves for the same individual. We prefer the linear probability specification over logistic fixed-effects models because it produces marginal effects directly interpretable as percentage-point changes and avoids the incidental parameters problem that leads logistic FE to discard all children without outcome variation. Results are robust to complementary log-log and logistic fixed-effects specifications, available upon request. We present three nested model specifications for each outcome. Model 1 includes only the contemporaneous shock indicator, isolating its unconfounded within-child effect. Model 2 includes only the cumulative shock measure. Model 3 includes both simultaneously, which is the preferred specification: it allows each shock measure to be interpreted net of the other, cleanly separating the immediate displacement effect from the accumulated stress effect. The inclusion of both measures simultaneously is the critical design feature that allows us to test H1 and H2 jointly and to verify that the cumulative effect is not simply picking up a lagged contemporaneous effect. To test H3 and H4, we augment the preferred specification with interaction terms between both shock measures and the continuous household wealth index. The interaction model takes the form: M i ₜ = α i + β₁ Shock i ₜ + β₂ CumShock i ₜ + β₃( Shock i ₜ × Wealth i ₜ) + β₄( CumShock i ₜ × Wealth i ₜ) + β₅ Wealth i ₜ + γ X i ₜ ₜ + ε i ₜ where Wealth i ₜ is the time-varying household wealth index. The coefficients β₃ and β₄ capture the differential effect of contemporaneous and cumulative shocks respectively across the wealth distribution. A negative β₃ is consistent with H3 (wealth buffers the immediate displacement effect); a positive β₄ is consistent with H4 (wealth enables migration as stress accumulates). Predicted probabilities from this model are plotted across the wealth distribution separately for each level of cumulative exposure, producing the marginal effects shown in Figs. 1 and 2 . We additionally estimate an equivalent interaction model replacing the wealth index with a continuous measure of educational attainment to assess whether the heterogeneity patterns are driven by material resources or human capital. The comparison between wealth and education interaction results is theoretically informative: if the enabling mechanism is primarily financial, the SES gradient should be stronger for wealth than for education in the cumulative exposure models. A critical feature of the fixed-effects design is that identification relies exclusively on children who exhibit within-individual variation in both the predictor and the outcome. Children with no variation in shock exposure across waves — either always exposed or never exposed — contribute no identifying information for the shock coefficients and are effectively excluded from the estimation. Table 2 reports the share of children with variation in each variable by country, and defines the estimating sample accordingly. The identifying sample ranges from approximately 9.9% of children in India to 17.0% in Peru, reflecting differences in both migration rates and shock incidence across countries. These shares are sufficient for reliable estimation in the pooled models but limit precision in country-specific analyses, as discussed in the results. All models include wave fixed effects to absorb common temporal shocks — including macroeconomic trends, policy changes, or aggregate climatic conditions — that affect all children in the same period. Time-varying individual controls include age and its square, to capture non-linear life-course effects on migration propensity. Country fixed effects are absorbed by the within-country estimation structure in country-stratified models; in pooled models, country fixed effects are included explicitly. Results Table 3 presents the pooled fixed-effects estimates for community migration. The results indicate that migration responses are not driven by contemporaneous shocks, but by the accumulation of prior exposure. Consistent with this, the contemporaneous shock indicator is substantively small and statistically unstable across specifications. In the model including only the current shock, the coefficient is negative and marginally significant (− 0.008), and becomes essentially null (0.007) once cumulative exposure is included simultaneously. This pattern is inconsistent with a displacement interpretation of acute shocks and suggests that immediate environmental events do not, on average, trigger community-level migration. If anything, the slightly negative sign in the current-shock-only model points to a short-term immobilizing effect of acute stress, plausibly reflecting liquidity constraints following severe events. By contrast, the cumulative shock measure shows a strong and consistent association with migration, providing clear support for H2. Having experienced one prior shock period increases the probability of community migration by approximately 10.4 percentage points relative to households with no prior shock history. A second prior shock period produces an equally large additional increment (10.7 in the full model), indicating a roughly linear dose-response relationship between accumulated stress and migration probability. Crucially, these estimates change negligibly when the contemporaneous indicator is added, confirming that the cumulative and contemporaneous dimensions of shock exposure capture distinct processes rather than collinear variation. The pattern is stable across all model specifications. Table 3 Climate shocks and community migration (pooled, child fixed effects) Any shock Current shock Cumulative shocks Both -0.008+ 0.007 (0.004) (0.004) 1 prior shock 0.104*** 0.105*** (0.005) (0.005) 2 prior shocks 0.104*** 0.107*** (0.011) (0.011) Observations 44,081 44,081 44,081 R² 0.436 0.443 0.443 Note: Linear probability models with child fixed effects. Estimating sample restricted to children with variation in the outcome (n = 4,014 unique children; 44,081 person-wave observations). Standard errors clustered at the child level in parentheses. + p < 0.1, * p < 0.05, ** p < 0.01, *** p < 0.001. Table 4 extends the analysis to rural-to-urban migration, estimated on the subsample of children with rural origin (n = 915 with variation in the outcome; 27,570 person-wave observations). The results are broadly consistent with those for community migration and reinforce the central role of cumulative exposure, while also revealing an important distinction in the role of contemporaneous shocks across migration types. The contemporaneous shock indicator is positive and statistically significant both in isolation and in the full model, indicating that acute shocks do carry a displacement signal for rural-to-urban transitions. This pattern suggests that immediate environmental stress can trigger mobility when it directly undermines agricultural livelihoods. This does not contradict the weak average effect observed for community migration, but rather reflects the different nature of the decision under consideration. Rural-to-urban migration involves a more substantial reorganization of livelihoods and may therefore be more sensitive to acute disruptions in agricultural production than shorter-distance or lateral moves. Importantly, however, the contemporaneous effect remains substantially smaller than the cumulative effect, indicating that even in this domain, migration responses are primarily structured by accumulated exposure over time. The support for H2 is unequivocal. One prior shock period increases the probability of rural-to-urban migration by 6.6 percentage points, while two prior shock periods increase it by 9.2 percentage points. This graded, dose-response pattern is consistent with cumulative vulnerability dynamics: as environmental stress accumulates, it progressively erodes the livelihood foundations that anchor households to rural areas, eventually shifting the balance toward urban relocation. The increase from the 1- to 2-shock coefficient (from 6.6 to 9.2 pp) suggests that the relationship may steepen with repeated exposure, consistent with a threshold-like process in which each additional shock further depletes coping reserves. Table 4 Climate shocks and rural-to-urban migration (pooled, child fixed effects) Any shock Current shock Cumulative shocks Both 0.010** 0.019*** (0.003) (0.003) 1 prior shock 0.065*** 0.066*** (0.004) (0.004) 2 prior shocks 0.087*** 0.092*** (0.007) (0.007) Observations 27,570 27,570 27,570 R² 0.432 0.446 0.447 Note: Sample restricted to children with rural origin and variation in the outcome (n = 915 unique children; 27,570 person-wave observations). Standard errors clustered at the child level in parentheses. + p < 0.1, * p < 0.05, ** p < 0.01, *** p < 0.001. A potential concern is that shocks are measured at the household level, whereas some observed migration may reflect individual moves out of the parental home rather than household-level relocation. In particular, moves to another community may partly reflect individual moves out of the parental home rather than migration driven by climate shocks affecting the household as a whole. To address this, we conduct two checks. First, we re-estimate the pooled models separately for the younger and older cohorts, since younger children are less likely to leave the parental household independently. Second, we restrict the sample to observations in which the child remains in the original household. Neither check materially alters the results. Effects are somewhat stronger for the older cohort, consistent with greater autonomy over mobility at older ages, and restricting the sample to children who remain in the household produces only minor coefficient changes. Overall, these checks support the main findings. Full results are available upon request. Heterogeneity by socioeconomic status Figures 1 and 2 present the central heterogeneity results of the paper and identify the key mechanism linking climate stress, resources, and mobility. Taken together, they show that the effect of socioeconomic status reverses across time horizons, generating the empirical signature of a climate-related mobility trap. Figure 1 plots the marginal effect of contemporaneous shocks on community migration across the wealth distribution. The gradient is negative and statistically distinguishable from zero, indicating that higher-SES households are significantly less likely to migrate in response to an immediate environmental shock. This pattern provides clear support for H3 and is consistent with a short-term buffering mechanism: material resources—liquid assets, savings, access to credit, and diversified income sources—allow better-off households to absorb acute shocks without relocating. For these households, remaining in place constitutes a viable adaptive strategy. By contrast, lower-SES households exhibit a modest increase in migration propensity, suggesting that acute shocks more readily exceed their limited coping capacity. Note Marginal effects from interaction model. Shaded area = 95% confidence interval. Figure 2 examines the role of cumulative exposure and reveals a qualitatively different pattern. The transition from zero to one prior shock increases migration probability by approximately 9 percentage points, with no meaningful variation across the wealth distribution—the gradient is flat. In this initial stage, all households respond similarly to accumulated stress. However, the transition from one to two or more prior shocks produces a sharp and statistically significant positive wealth gradient: migration increases substantially among higher-SES households, while remaining flat or increasing only minimally among the poorest. This divergence marks a reversal in the role of resources and provides strong support for H4. Note Marginal effects from interaction model. Shaded area = 95% confidence interval. The contrast between these two stages is central. The first accumulation of stress appears sufficient to generate broadly shared incentives to move, but further accumulation reveals a constraint: the capacity to translate exposure into actual mobility depends on access to material resources. As environmental stress intensifies, better-off households retain the ability to respond through migration, whereas resource-poor households become increasingly unable to do so. The resulting divergence in mobility capacity constitutes the core empirical signature of a structural mobility trap, in which those most exposed to cumulative environmental risk are least able to exit through geographic relocation. Interactions with educational attainment (Figs. 3 and 4) reinforce this interpretation. While higher-educated households display a similar buffering response to contemporaneous shocks, the interaction between education and cumulative exposure is not statistically significant. This dissociation indicates that the enabling mechanism identified above is driven primarily by material resources rather than human capital. Education may shape aspirations and awareness of migration opportunities, but it is liquid wealth—rather than schooling—that enables households to finance and organize relocation under conditions of sustained environmental stress. Note: Interaction with continuous education variable. Marginal effects with 95% CIs. Country-level heterogeneity As a robustness and generalizability check, Tables 5 a and 5 b decompose the pooled results by country. Two observations deserve emphasis. First, the direction of the cumulative shock effects is consistent across all four countries and both migration outcomes — not a single country-outcome combination reverses the sign. This cross-national consistency is analytically important and indicates that the cumulative vulnerability mechanism is not an artifact of any single country's institutional context, agricultural system, or migration infrastructure, but reflects a more general demographic process operating across very different settings. Second, the magnitude of effects varies substantially, with Peru showing the strongest responses for both outcomes (0.181 pp for community migration and 0.179 pp for rural-to-urban migration at two prior shocks), followed by Vietnam, India, and Ethiopia in descending order for rural-urban flows. The contemporaneous shock results are more heterogeneous — positive and significant in Ethiopia for community migration, and in Ethiopia, India, and Peru for rural-urban migration — though generally smaller and less consistent than the cumulative effects. Table 5 a. Climate shocks and community migration by country (child fixed effects) Any shock Ethiopia India Peru Vietnam Pooled 0.043*** -0.001 -0.009 -0.020** 0.007 (0.007) (0.008) (0.014) (0.008) (0.004) 1 prior shock 0.083*** 0.121*** 0.105*** 0.102*** 0.105*** (0.008) (0.010) (0.016) (0.011) (0.005) 2 prior shocks 0.075*** 0.118*** 0.181*** 0.148*** 0.107*** (0.013) (0.025) (0.031) (0.032) (0.011) Observations 11,146 11,406 10,171 11,358 44,081 Note: All models include both any shock and cumulative shock indicators simultaneously. Standard errors clustered at the child level in parentheses. Country-level estimates should be interpreted cautiously given reduced within-FE sample sizes. + p < 0.1, * p < 0.05, ** p < 0.01, *** p < 0.001. Country-level estimates should be interpreted with caution given the reduced within-FE sample sizes at the country level, which limit statistical precision. We treat these results as descriptive evidence of directional consistency rather than country-specific effect sizes. Table 5 b. Climate shocks and rural-to-urban migration by country (child fixed effects) Any shock Ethiopia India Peru Vietnam Pooled 0.017** 0.009* 0.056*** 0.006 0.019*** (0.005) (0.004) (0.013) (0.004) (0.003) 1 prior shock 0.056*** 0.053*** 0.137*** 0.057*** 0.066*** (0.006) (0.006) (0.017) (0.007) (0.004) 2 prior shocks 0.085*** 0.054*** 0.179*** 0.071*** 0.092*** (0.009) (0.015) (0.028) (0.020) (0.007) Observations 7,170 8,480 2,826 9,094 27,570 Note: Sample restricted to children with rural origin. All models include both any shock and cumulative shock indicators simultaneously. Standard errors clustered at the child level in parentheses. + p < 0.1, * p < 0.05, ** p < 0.01, *** p < 0.001. Discussion This paper shows that migration responses to climate stress are not primarily driven by contemporaneous shocks, but by the accumulation of environmental exposure over time, and that the role of socioeconomic resources varies systematically across this process. Using longitudinal data across four countries, we find that contemporaneous shocks have weak and inconsistent effects on migration, whereas cumulative exposure produces large, monotonic increases in mobility. The central implication concerns the temporal specification of climate exposure. Much of the existing literature models migration as a response to discrete environmental events, implicitly assuming a short-run causal pathway from shock to mobility. Our results indicate that this approach is misspecified. Migration is more strongly associated with the accumulation of prior exposure than with isolated events, suggesting that the relevant unit of analysis is not the shock itself but the trajectory of environmental stress over time. Because contemporaneous shocks can simultaneously generate displacement and immobility effects, estimates based solely on current exposure are likely biased toward zero. By contrast, cumulative measures capture the progressive erosion of livelihoods that ultimately shifts migration from a constrained option to a feasible or necessary response. The second contribution concerns the role of socioeconomic resources. The results show that the association between resources and migration is not stable but reverses over the course of exposure. In the short term, wealth buffers migration: better-off households absorb shocks without relocating. As exposure accumulates, however, wealth becomes an enabling factor. Households with sufficient resources are able to translate accumulated stress into mobility, while poorer households remain in place despite rising environmental pressure. This reversal constitutes the empirical signature of a mobility trap. Resource-poor households are not less exposed to environmental stress, but are less able to respond to it through migration. As a result, the capacity to adapt through mobility becomes increasingly stratified. The households most exposed to cumulative risk are also those least able to exit deteriorating environments, while better-off households retain adaptive flexibility. These opposing dynamics are obscured in models that estimate average effects, where buffering and enabling mechanisms offset each other. The absence of a comparable pattern for education reinforces this interpretation. While education is associated with reduced migration following contemporaneous shocks, it does not condition migration under cumulative exposure. This suggests that the mechanism identified here operates primarily through material constraints rather than human capital. The consistency of these findings across Ethiopia, India, Peru, and Vietnam indicates that the mechanisms identified are not context-specific but reflect a broader demographic process linking environmental stress, resources, and mobility. Migration under climate change emerges as a process of cumulative vulnerability and stratified adaptation, rather than a direct response to discrete events. Several limitations should be noted. First, shock measures are self-reported and capture perceived exposure rather than objective climatic conditions, although the appendix provides validation against independent precipitation data. Second, restricting the sample to individuals observed across all waves introduces selection on residential stability. If attrition is positively correlated with both migration and shock exposure, the estimated effects should be interpreted as lower bounds. Third, unobserved time-varying factors may still affect both exposure and migration, although the fixed-effects design mitigates this concern. These findings have implications for both research and policy. Empirically, they suggest that studies relying on contemporaneous shock indicators or short panels systematically underestimate the climate–migration relationship by failing to capture cumulative dynamics. Substantively, they highlight that the key constraint is not only exposure to environmental risk, but the ability to respond to it. Policies aimed at facilitating climate adaptation through migration must therefore address the material constraints that prevent resource-poor households from moving. Without such support, environmental change is likely to increase not only mobility, but also immobility among the most vulnerable populations. As climate change intensifies, these dynamics are likely to become more pronounced, reinforcing the divergence between households able to adapt through mobility and those who remain structurally trapped in deteriorating environments. References Adil L, Eckstein D, Künzel V, Schäfer L (2025) Climate Risk Index 2026: Who suffers most from extreme weather events? Germanwatch Berlemann M, Tran TX (2020) Climate-Related Hazards and Internal Migration: Empirical Evidence for Rural Vietnam. Econ Disasters Clim Change 4:385–409 Black R, Bennett SR, Thomas SM, Beddington JR (2011) Migration as adaptation. Nature 478(7370):447–449. https://doi.org/10.1038/478477a Boas I, Farbotko C, Adams H, Sterly H, Bush S, van der Geest K, Hulme M (2019) Climate migration myths. Nat Clim Change 9(12):901–903. https://doi.org/10.1038/s41558-019-0633-3 Bohra-Mishra P, Oppenheimer M, Hsiang SM (2014) Nonlinear permanent migration response to climatic variations but minimal response to disasters. Proceedings of the National Academy of Sciences, 111 (27), 9780–9785. https://doi.org/10.1073/pnas.1317166111 Carling J (2002) Migration in the age of involuntary immobility: Theoretical reflections and Cape Verdean experiences. J Ethnic Migration Stud 28(1):5–42. https://doi.org/10.1080/13691830120103912 Cattaneo C, Peri G (2016) The migration response to increasing temperatures. J Dev Econ 122:127–146. https://doi.org/10.1016/j.jdeveco.2016.05.004 Cattaneo C, Beine M, Fröhlich CJ, Kniveton D, Martinez-Zarzoso I, Mastrorillo M, Schraven B (2019) Human migration in the era of climate change. Rev Environ Econ Policy 13(2):189–206. https://doi.org/10.1093/reep/rez008 Cattaneo C, Shayegh S, Albert C et al (2026) Broadening climate migration research across impacts, adaptation and mitigation. Nat Clim Chang 16:255–260. https://doi.org/10.1038/s41558-025-02545-1 Chen J, Mueller V (2018) Coastal climate change, soil salinity and human migration in Bangladesh. Nat Clim change 8(11):981–985 Dannefer D (2003) Cumulative advantage/disadvantage and the life course: Cross-fertilizing age and social science theory. Journals Gerontol Ser B: Psychol Sci Social Sci 58(6):S327–S337 Entzinger H, Scholten P (2022) The role of migration in enhancing resilience to climate change in the Vietnamese Mekong River Delta. Migration Stud 10(1):24–40 Gray C, Mueller V (2012) Drought and Population Mobility in Rural Ethiopia. World Dev 40(1):134–145 Hermans-Neumann K, Priess J, Herold M (2017) Human migration, climate variability, and land degradation: hotspots of socio-ecological pressure in Ethiopia. Reg Envriron Chang 17:1479–1492 Hoffmann R, Dimitrova A, Muttarak R, Crespo Cuaresma J, Peisker J (2020) A meta-analysis of country-level studies on environmental change and migration. Nat Clim Change 10(10):904–912. https://doi.org/10.1038/s41558-020-0898-6 Hunter LM, Luna JK, Norton RM (2015) Environmental dimensions of migration. Ann Rev Sociol 41:377–397. https://doi.org/10.1146/annurev-soc-073014-112223 IPCC, Birkmann J, Liwenga E, Pandey R, Boyd E, Djalante R, Gemenne F, Leal Filho W, Pinho PF, Stringer L, Wrathall D (2022) Poverty, Livelihoods and Sustainable Development. In: Pörtner -O, Roberts DC, Tignor M, Poloczanska ES, Mintenbeck K, Alegría A, Craig M, Langsdorf S, Löschke S, Möller V, Okem A, Rama B (eds) Climate Change 2022: Impacts, Adaptation and Vulnerability. Contribution of Working Group II to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change [H. Cambridge University Press, Cambridge, UK and New York, NY, USA, pp 1171–1274 IPCC (2007) In: Metz -B, Davidson OR, Bosch PR, Dave R (eds) L.A. Meyer Contribution of Working Group III to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge, United Kingdom and New York, USA Jha CK, Gupta V, Chattopadhyay U, Sreeraman BA (2018) Migration as adaptation strategy to cope with climate change: A study of farmers' migration in rural India. International Journal of Climate Change Strategies and Management Kaczan DJ, Orgill-Meyer J (2020) The impact of climate change on migration: a synthesis of recent empirical insights. Clim Change 158(3):281–300 Leal Filho W et al (2023) Assessing causes and implications of climate-induced migration in Kenya and Ethiopia. Environmental Science and Policy, p 150 Milan A, Ho R (2014) Livelihood and migration patterns at different altitudes in the Central Highlands of Peru. Climate Dev 6(1):69–76 Mueller V, Gray C, Hopping D (2020) Climate-induced migration and unemployment in middle-income Africa. Glob Environ Change 65:102187. https://doi.org/10.1016/j.gloenvcha.2020.102187 Nawrotzki RJ, DeWaard J (2018) Putting trapped populations into place: climate change and inter-district migration flows in Zambia. Reg Envriron Chang 18(2):533–546 Richa, Noy I, Sen S (2024) Extreme Events and Inter-state Migration in India: An Empirical Analysis. Migration Dev 13(2):289–311 Rigaud KK, de Sherbinin A, Jones B, Abu-Ata NE, Adamo S (2021) Groundswell Africa: A deep dive into internal climate migration in Senegal. World Bank Schewel K (2020) Understanding immobility: Moving beyond the mobility bias in migration studies. Int Migrat Rev 54(2):328–355. https://doi.org/10.1177/0197918319831952 Šedová B, Čizmaziová L, Cook A (2021) A meta-analysis of climate migration literature. CEPA Discussion Papers, 29. https://doi.org/10.25932/publishup-49982 Viswanathan B, Kumar KK (2015) Weather, agriculture and rural migration: evidence from state and district level migration in India. Environ Dev Econ 20:469–492 Wrathall DJ et al (2014) Migration Amidst Climate Rigidity Traps: Resource Politics and Social-Ecological Possibilism in Honduras and Peru. Ann Assoc Am Geogr 104(2):292–304 Zickgraf C (2021) Theorizing (im)mobility in the face of environmental change. Reg Envriron Chang 21(4):126. https://doi.org/10.1007/s10113-021-01849-4 Additional Declarations There is NO Competing Interest. Supplementary Files APPENDIXA.docx Cite Share Download PDF Status: Under Review Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-9405620","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":629941387,"identity":"a420cc87-aa4c-4eee-bf18-e9d4a3ef5bd0","order_by":0,"name":"Héctor Cebolla Boado","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAABD0lEQVRIie2RMUvDQBTH3xFIlkDWgw73Fe4mFYp+lStCu+QjiEQCcbGdr+hH6fBKIV2CroUKnotdHOIiHYL4YqqDHAE3h/sNx//g/Xjv3QF4PP8Q+ZM4MLQwbEN37VFY9l2DGsZ/V1Z9xR1HUb6zbwsQySBH1BcPIrm9trBvjgVEFByc3JTqyryAmt+VGnW5VeaxkmxacJXFlXQpcpOyPEbQFCTqcKslTyFgGddAwak8vXbK2ZfycU/KxAbQkCJ2zsHkJj504aSMCqSgZQBh2wXcg1VjNTfIaQXaZTQ7V4baLdtdijh1K+uVrWsc0kPlS1u/n4rETJ7tvrkUSbR2Dnbg119ge4Q99R6Px+Pp5xPSaGPh0Lbe0gAAAABJRU5ErkJggg==","orcid":"https://orcid.org/0000-0002-5804-8715","institution":"National Research Council (CSIC, Spain)","correspondingAuthor":true,"prefix":"","firstName":"Héctor","middleName":"Cebolla","lastName":"Boado","suffix":""},{"id":629941390,"identity":"7c850359-2cff-4985-9339-0119a4083c25","order_by":1,"name":"Michael Lund","email":"","orcid":"","institution":"National Research Council (CSIC, Spain)","correspondingAuthor":false,"prefix":"","firstName":"Michael","middleName":"","lastName":"Lund","suffix":""}],"badges":[],"createdAt":"2026-04-13 14:46:03","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-9405620/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-9405620/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":108740339,"identity":"dece7093-7c70-4083-bb90-9bbd44af690c","added_by":"auto","created_at":"2026-05-07 22:48:48","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":45748,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eMarginal effect of contemporaneous climate shock on community migration by wealth index (wi)\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eNote: Marginal effects from interaction model. Shaded area = 95% confidence interval.\u003c/em\u003e\u003c/p\u003e","description":"","filename":"image1.png","url":"https://assets-eu.researchsquare.com/files/rs-9405620/v1/61bb373729d7303a0b86e790.png"},{"id":108806508,"identity":"79f72efc-9436-4249-97ea-a3e3474e726a","added_by":"auto","created_at":"2026-05-08 15:28:47","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":75113,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eMarginal effect of cumulative climate shocks on community migration by wealth index (wi)\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eNote: Marginal effects from interaction model. Shaded area = 95% confidence interval.\u003c/em\u003e\u003c/p\u003e","description":"","filename":"image2.png","url":"https://assets-eu.researchsquare.com/files/rs-9405620/v1/dc473eabc9e14be18a5861e3.png"},{"id":108805950,"identity":"acf520f8-a829-4fa1-af3d-f7bdffe0b88a","added_by":"auto","created_at":"2026-05-08 15:27:16","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":113970,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eFigure 3 \u0026amp; 4. Marginal effects by education level (any shock and cumulative shocks)\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eNote: Interaction with continuous education variable. Marginal effects with 95% CIs.\u003c/em\u003e\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-9405620/v1/2ef09bd4a828d74255236bf7.png"},{"id":108810051,"identity":"fc7c53f5-78a8-4097-b0ea-96f8120cf28b","added_by":"auto","created_at":"2026-05-08 15:57:15","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":610275,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9405620/v1/8a749ca2-42ea-4f8e-b392-df5a10fef9b0.pdf"},{"id":108806151,"identity":"c63b5b04-ab0a-4643-b1ac-e3a6c025ae03","added_by":"auto","created_at":"2026-05-08 15:27:49","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":130217,"visible":true,"origin":"","legend":"","description":"","filename":"APPENDIXA.docx","url":"https://assets-eu.researchsquare.com/files/rs-9405620/v1/0df9419ecea791f638f83ad9.docx"}],"financialInterests":"There is \u003cb\u003eNO\u003c/b\u003e Competing Interest.","formattedTitle":"Cumulative Climate Stress Drives Migration and Stratifies Mobility: Evidence from Ethiopia, India, Peru, and Vietnam","fulltext":[{"header":"Introduction","content":"\u003cp\u003eClimate change is widely expected to intensify human mobility, a view consolidated in early global assessments (IPCC, 2007) and subsequent syntheses of the climate\u0026ndash;migration nexus (Cattaneo et al., 2019; Rigaud et al., 2021). Yet empirical evidence accumulated over the past decade consistently shows that most populations exposed to environmental stress do not move, giving rise to the so-called (im)mobility paradox (Boas et al., 2019; Zickgraf, 2021). More recent work emphasizes that migration responses are heterogeneous and strongly mediated by socioeconomic and institutional conditions rather than directly triggered by environmental shocks (Black et al., 2011; Hunter et al., 2015; Cattaneo et al., 2026). However, this apparent inconsistency may reflect not the absence of an effect, but a mischaracterization of how environmental stress operates over time.\u003c/p\u003e\n\u003cp\u003eMost empirical studies operationalize climate exposure as a contemporaneous shock (a drought, flood, or temperature anomaly) and estimate its immediate effect on migration (Bohra-Mishra et al., 2014; Cattaneo \u0026amp; Peri, 2016; Hoffmann et al., 2020). This approach implicitly assumes that mobility is a short-term response to discrete events. Yet environmental stress often unfolds cumulatively, through the progressive erosion of livelihoods, assets, and adaptive capacity (Hoffmann et al., 2020; Hunter et al., 2015). This perspective aligns with broader theories of cumulative disadvantage, in which repeated exposures compound over time to reshape life-course outcomes (Dannefer, 2003).\u003c/p\u003e\n\u003cp\u003eHere we show that this temporal misspecification leads to systematically biased conclusions. Using longitudinal individual data from Ethiopia, India, Peru, and Vietnam, we distinguish between contemporaneous and cumulative exposure to environmental shocks within individuals over time. We find that contemporaneous shocks have weak and inconsistent effects on migration, whereas cumulative exposure produces large, monotonic increases in mobility. Models based solely on current shocks therefore underestimate the relationship between climate stress and migration.\u003c/p\u003e\n\u003cp\u003eWe further show that the role of socioeconomic resources reverses across these time horizons. In the short term, resources buffer mobility, allowing households to absorb shocks without relocating. As exposure accumulates, however, these same resources enable migration. This dynamic generates a structural divergence: resource-poor households become increasingly unable to move despite rising environmental pressure, while wealthier households retain adaptive flexibility. This pattern is consistent with theories of involuntary immobility and mobility traps (Carling, 2002; Nawrotzki \u0026amp; DeWaard, 2018; Schewel, 2020). Climate-related migration thus emerges not as a direct response to environmental events, but as a process of cumulative vulnerability that stratifies mobility across populations.\u003c/p\u003e"},{"header":"Climate stress and migration","content":"\u003cp\u003eEarly theoretical accounts treated environmental degradation and extreme weather events as mechanical push factors expected to uproot large populations\u0026mdash;the so-called \"climate refugee\" narrative (Rigaud et al., \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). This deterministic view predicted a near-automatic relationship between climatic deterioration and displacement, projecting hundreds of millions of forced migrants by mid-century (Black et al., \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e2011\u003c/span\u003e). Empirical evidence has systematically failed to support these projections. Research across a range of geographic and climatic contexts consistently shows that most people remain in place even under severe environmental stress, and that when migration does occur it is rarely a simple, direct response to climatic exposure alone (Boas et al., \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e2019\u003c/span\u003e; Cattaneo et al., \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2019\u003c/span\u003e; Zickgraf, \u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e2021\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eThis disconnection between theoretical expectation and empirical reality has prompted a substantial reorientation in the literature. A growing body of work now emphasizes that climate\u0026ndash;migration linkages are mediated by economic, social, and institutional factors that condition the capacity and willingness to move (Black et al., \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e2011\u003c/span\u003e; Cattaneo et al., \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2019\u003c/span\u003e; Hunter et al., \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e2015\u003c/span\u003e). Studies have documented non-linear relationships between temperature anomalies and migration (Bohra-Mishra et al., \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e2014\u003c/span\u003e), shown that effects vary substantially by income level and agricultural dependence (Cattaneo \u0026amp; Peri, \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e2016\u003c/span\u003e), and highlighted the importance of migration networks, institutional quality, and livelihood diversification as moderating factors (Hoffmann et al., \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2020\u003c/span\u003e). What has emerged from this literature is a picture of substantial heterogeneity: environmental shocks do not trigger uniform mobility responses but interact with pre-existing socioeconomic and demographic conditions to produce divergent outcomes across individuals, households, and contexts (Nawrotzki \u0026amp; DeWaard, \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e2018\u003c/span\u003e).\u003c/p\u003e \u003cp\u003e \u003cem\u003eCumulative vulnerability and the temporal dimension of climate stress.\u003c/em\u003e A significant limitation of much existing research lies in how the temporal dimension of environmental exposure is operationalized. As a result, existing estimates may be biased toward zero, as contemporaneous shocks simultaneously generate displacement and immobility effects that offset each other. By modeling migration as a function of contemporaneous shocks, this literature implicitly assumes that the relevant causal pathway runs from acute exposure to rapid mobility response. Yet livelihood degradation under climate change is often gradual and cumulative, operating through the progressive erosion of agricultural yields, water availability, and household asset stocks rather than through single catastrophic events (Hoffmann et al., \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2020\u003c/span\u003e; Hunter et al., \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e2015\u003c/span\u003e). The concept of cumulative vulnerability captures this dynamic: repeated exposure to environmental stress compounds over time, progressively narrowing households' adaptive options and eventually shifting the cost-benefit calculus of migration (Bohra-Mishra et al., \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e2014\u003c/span\u003e). This perspective aligns with broader sociological theories of cumulative advantage and disadvantage, which posit that small initial differences can compound over time into substantial inequalities in life trajectories (Dannefer, \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2003\u003c/span\u003e). As a result, approaches centered on discrete shocks may understate the role of sustained environmental pressure in shaping mobility decisions over time.\u003c/p\u003e \u003cp\u003eEmpirical evidence on cumulative climate effects on migration remains scarce, in part because most available data sources lack the longitudinal depth required to track both shock exposure and mobility outcomes across multiple periods for the same individuals. Cross-sectional studies and short panels are poorly suited to distinguishing between immediate displacement effects and the slower-acting consequences of repeated exposure. This limitation is particularly consequential given that cumulative processes may structure not only whether households migrate, but when they do so. This paper addresses this gap directly by exploiting five rounds of panel data and constructing an explicit measure of cumulative shock exposure that captures the accumulation of environmental stress over the observed life course.\u003c/p\u003e \u003cp\u003e \u003cem\u003eImmobility, trapped populations, and resources.\u003c/em\u003e Alongside the study of climate-driven migration, a parallel literature has developed around the conditions that prevent mobility even when environmental conditions deteriorate severely. Early work conceptualized immobility primarily as a sign of vulnerability or failure to adapt (Carling, \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e2002\u003c/span\u003e), but more recent scholarship has reframed it as a socially structured outcome that may itself represent a form of active, if constrained, adaptation (Schewel, \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e2020\u003c/span\u003e; Zickgraf, \u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). This reframing distinguishes between voluntary immobility\u0026mdash;a deliberate choice to remain in place\u0026mdash;and involuntary immobility, in which households facing deteriorating conditions lack the resources, networks, or capabilities required to migrate (Carling, \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e2002\u003c/span\u003e; Schewel, \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e2020\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eThe concept of mobility traps captures the most severe form of involuntary immobility: situations in which precisely the environmental stress that should motivate migration simultaneously destroys the assets and liquidity required to finance it (Nawrotzki \u0026amp; DeWaard, \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e2018\u003c/span\u003e). Resource constraints operate as a double bind\u0026mdash;households are both more exposed to environmental risk and less able to escape it through mobility. Empirical evidence for this mechanism remains limited, largely because documenting it requires data that simultaneously track asset shocks, household wealth, and migration decisions over time. Studies from sub-Saharan Africa have provided suggestive evidence of trapped population dynamics (Nawrotzki \u0026amp; DeWaard, \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e2018\u003c/span\u003e; Rigaud et al., \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e2021\u003c/span\u003e), but cross-national comparative evidence is scarce, and the interaction between the time horizon of exposure and the role of resources has not been systematically examined. In particular, while prior research has documented that socioeconomic resources condition migration responses, less attention has been paid to how their role may evolve across different stages of environmental exposure, potentially shaping not only the likelihood of movement but the capacity to translate accumulated stress into mobility.\u003c/p\u003e \u003cp\u003eTaken together, these strands of research highlight the importance of both temporal dynamics and socioeconomic heterogeneity in shaping climate\u0026ndash;migration linkages, but they have largely been studied separately. This paper integrates these perspectives by distinguishing between contemporaneous and cumulative exposure and by examining how the role of socioeconomic resources varies across these time horizons. In doing so, it shows that migration responses to climate stress are structured less by immediate reactions to isolated events than by the accumulation of environmental pressure over time, and that the effect of resources is not constant but reverses as exposure accumulates. This joint dynamic generates divergent mobility outcomes across households and provides a micro-level specification of the conditions under which mobility traps emerge.\u003c/p\u003e \u003cp\u003eAlthough prior research has documented that resources condition migration responses, less attention has been paid to how their role evolves across different time horizons of environmental exposure. Building on this limitation, we conceptualize migration responses to climate stress as a dynamic process shaped by the interaction between the accumulation of environmental constraints and the changing role of resources over time. This framework yields four testable hypotheses.\u003c/p\u003e \u003cp\u003e \u003cem\u003eH1. Contemporaneous shocks and migration.\u003c/em\u003e Exposure to a contemporaneous environmental shock will have a weak or null effect on migration. Acute shocks simultaneously reduce household liquidity and increase coordination costs, such that their net effect on mobility is expected to be negligible or ambiguous. Households may absorb short-term shocks through asset drawdowns, credit, or informal transfers without resorting to geographic relocation, reflecting the short-term disruption generated by acute events.\u003c/p\u003e \u003cp\u003e \u003cem\u003eH2. Cumulative shock exposure and migration.\u003c/em\u003e Cumulative exposure to environmental shocks will be positively and progressively associated with migration. As the accumulation of stress over time erodes livelihood security and depletes coping reserves, the probability of both community-level and rural-to-urban migration is expected to increase in a graded, dose-response pattern consistent with cumulative vulnerability processes.\u003c/p\u003e \u003cp\u003e \u003cem\u003eH3. Socioeconomic resources as short-term buffer.\u003c/em\u003e Among households exposed to a contemporaneous shock, higher socioeconomic status will be associated with lower migration probabilities. Material resources allow better-off households to absorb acute environmental events without relocating, dampening immediate displacement responses. The buffering effect of wealth is expected to be observable as a negative interaction between contemporaneous shock exposure and household wealth, consistent with a short-term stabilizing role of resources.\u003c/p\u003e \u003cp\u003e \u003cem\u003eH4. Socioeconomic resources as long-term enabler and the mobility trap.\u003c/em\u003e In contrast to the short-term buffering effect, as cumulative shock exposure increases, the association between socioeconomic status and migration will turn positive and strengthen. While resource-poor households remain in place despite repeated environmental deterioration\u0026mdash;lacking the means to finance or organize relocation\u0026mdash;wealthier households retain the capacity to translate accumulated stress into migration. This asymmetry implies a positive interaction between cumulative shock exposure and household wealth and constitutes the core empirical signature of a climate-related mobility trap among the most resource-constrained.\u003c/p\u003e \u003cp\u003eTaken together, these hypotheses imply that climate change does not uniformly increase mobility, but reorders who is able to move over time.\u003c/p\u003e"},{"header":"Study context","content":"\u003cp\u003eThe analysis draws on four countries\u0026mdash;Ethiopia, India, Peru, and Vietnam\u0026mdash;that combine substantial exposure to climate variability with differing migration systems and socioeconomic conditions. All four contexts are characterized by a strong reliance on climate-sensitive livelihoods, particularly agriculture, but differ in the nature of environmental risks and mobility opportunities. Ethiopia represents a drought-prone, low-asset context where mobility constraints are likely to be binding. India combines monsoon variability with widespread distress migration. Peru exhibits high baseline mobility linked to climatic variability, while Vietnam reflects a flood-dominated environment with established internal migration systems.\u003c/p\u003e \u003cp\u003eThis cross-country variation is not merely descriptive but analytically central to the identification strategy. The four settings differ substantially in their climatic regimes, exposure to environmental shocks, and migration systems, providing independent sources of variation in both the intensity and temporal structure of climate stress. This heterogeneity allows us to assess whether the relationship between cumulative exposure and migration reflects a general demographic mechanism rather than context-specific dynamics tied to a particular environmental or institutional setting. A more detailed description of each country is provided in Appendix A.1.\u003c/p\u003e"},{"header":"DATA: The Young Lives Study","content":"\u003cp\u003eThe analysis draws on data from the Young Lives longitudinal study, a panel survey conducted in Ethiopia, India (Andhra Pradesh and Telangana), Peru, and Vietnam. The study follows approximately 12,000 children from 2002 to 2016, providing repeated observations on household conditions and life trajectories over time. Its longitudinal structure makes it well suited to the analysis of processes\u0026mdash;such as migration\u0026mdash;that unfold dynamically.\u003c/p\u003e \u003cp\u003eYoung Lives tracks two cohorts recruited in 2002: a Younger Cohort (aged 6\u0026ndash;18 months at baseline) and an Older Cohort (aged 7.5\u0026ndash;8.5 years). Children were sampled from sentinel sites designed to capture variation across rural and urban areas, ecological zones, and population groups. While not nationally representative, the data span substantial socioeconomic and geographic heterogeneity within each country.\u003c/p\u003e \u003cp\u003eThe analysis uses data from Rounds 1 to 5 (2002\u0026ndash;2016). We restrict the sample to children observed in all rounds, ensuring that each individual contributes four inter-wave transitions and allowing consistent measurement of cumulative exposure. Attrition is low (6.5% overall), though partly driven by migration, implying that estimates should be interpreted as conservative.\u003c/p\u003e \u003cp\u003eTwo migration outcomes are constructed from residential histories. The first is community migration, defined as any change in community between survey rounds. The second is rural-to-urban migration, defined for individuals of rural origin who move to an urban area between waves.\u003c/p\u003e \u003cp\u003eThe key explanatory variables distinguish between contemporaneous and cumulative exposure to environmental shocks. The contemporaneous measure is a binary indicator of whether the household reported experiencing at least one environmental shock (drought, flood, erosion, or frost) between survey rounds. The cumulative measure counts the number of prior periods in which such shocks were reported, excluding the current interval. This construction isolates accumulated exposure from contemporaneous shocks and enables a direct test of cumulative effects.\u003c/p\u003e \u003cp\u003eShock measures are self-reported and capture experienced environmental stress. Appendix A.2 shows that reported drought exposure exhibits persistence over time, supporting the use of cumulative measures. Appendix A.3 further demonstrates that self-reported droughts are strongly associated with objective precipitation deficits in most contexts, indicating that these measures reflect underlying environmental conditions rather than reporting bias.\u003c/p\u003e \u003cp\u003eSocioeconomic heterogeneity is measured using the Young Lives household wealth index, a continuous measure based on assets, housing conditions, and access to services. We use this index to examine how the effects of contemporaneous and cumulative shocks vary across the wealth distribution through interaction terms.\u003c/p\u003e \u003cp\u003eTable\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e reports descriptive statistics by country. Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e presents within-child variation in migration outcomes and shock exposure, corresponding to the identifying sample in the fixed-effects models.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eDescriptive statistics by country\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"9\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c8\" colnum=\"8\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c9\" colnum=\"9\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cem\u003eCountry\u003c/em\u003e\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eN\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eCommunity migration (%)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eRural\u0026rarr;Urban (%)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eUrban\u0026rarr;Rural (%)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eAny shock (%)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c7\"\u003e \u003cp\u003eCumulative shocks (mean)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c8\"\u003e \u003cp\u003eHigh education (%)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c9\"\u003e \u003cp\u003eWealth index (mean)\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cem\u003eEthiopia\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e11,146\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e12.42\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e2.70\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e1.14\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e26.32\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.30\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e8.77\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.32\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cem\u003eIndia\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e11,406\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e11.01\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e2.58\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e1.02\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e13.96\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.18\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e17.04\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.49\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cem\u003ePeru\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e10,171\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e29.52\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e3.25\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e1.35\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e13.21\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.14\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e42.70\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.52\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cem\u003eVietnam\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e11,358\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e11.09\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e2.30\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.21\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e10.82\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.14\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e30.48\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.56\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"9\"\u003e\u003cem\u003eNote: N refers to person-wave observations. Community migration\u0026thinsp;=\u0026thinsp;changed community between waves. Rural\u0026rarr;Urban and Urban\u0026rarr;Rural are directional transitions. Any shock\u0026thinsp;=\u0026thinsp;at least one environmental shock between waves (drought, flood, erosion, frost). Cumulative shocks\u0026thinsp;=\u0026thinsp;mean number of prior waves with at least one shock. High education\u0026thinsp;=\u0026thinsp;share with secondary or higher. Wealth index\u0026thinsp;=\u0026thinsp;mean value (0\u0026ndash;1 scale).\u003c/em\u003e\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eWithin-child variation in outcomes and shocks, and estimating sample\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"5\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cem\u003eCountry\u003c/em\u003e\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eChildren (N)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eMigration varies (%)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eShock varies (%)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eIdentifying sample (%)\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cem\u003eEthiopia\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e2,895\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e28.36\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e55.20\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e13.82\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cem\u003eIndia\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e2,942\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e26.65\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e41.10\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e9.86\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cem\u003ePeru\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e2,674\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e55.98\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e33.60\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e17.02\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cem\u003eVietnam\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e2,972\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e30.69\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e34.70\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e8.14\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"5\"\u003e\u003cem\u003eNote: Migration varies\u0026thinsp;=\u0026thinsp;share of children with at least one change in community migration status across waves. Shock varies\u0026thinsp;=\u0026thinsp;share of children with variation in shock exposure across waves. Identifying sample\u0026thinsp;=\u0026thinsp;share of children with within-child variation in both the outcome and the predictor (estimating sample for child fixed-effects models).\u003c/em\u003e\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e"},{"header":"Methods","content":"\u003cp\u003eThe empirical strategy exploits within-child variation over time to identify the effect of environmental shock exposure on migration decisions. While prior work relies on sub-national panel variation (Chen \u0026amp; Mueller, \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e2018\u003c/span\u003e), our design uses individual-level fixed effects, allowing for a stricter identification of within-household exposure to environmental shocks. The core identification assumption is that, conditional on child fixed effects, changes in shock exposure over time are not systematically correlated with time-varying unobservables that independently predict migration. Child fixed effects eliminate all stable sources of confounding at the individual and household level, including baseline socioeconomic status, risk preferences, location-specific factors, and time-invariant household characteristics, that would otherwise confound the estimated relationship between shocks and mobility. Because the models are estimated on inter-wave transitions (t to t\u0026thinsp;+\u0026thinsp;1), temporal ordering between shock exposure and migration outcomes is ensured by design: shocks are measured during or prior to the inter-wave period, and migration is recorded at the subsequent survey round. We estimate linear probability models (LPM) of the following form:\u003c/p\u003e \u003cp\u003e \u003cem\u003eM\u003c/em\u003e \u003csub\u003ei\u003c/sub\u003eₜ = α\u003csub\u003ei\u003c/sub\u003e + β₁\u003cem\u003eShock\u003c/em\u003e\u003csub\u003ei\u003c/sub\u003eₜ + β₂\u003cem\u003eCumShock\u003c/em\u003e\u003csub\u003ei\u003c/sub\u003eₜ + γ\u003cem\u003eX\u003csub\u003ei\u003c/sub\u003eₜ\u003c/em\u003e + ε\u003csub\u003ei\u003c/sub\u003eₜ\u003c/p\u003e \u003cp\u003ewhere \u003cem\u003eM\u003c/em\u003e\u003csub\u003ei\u003c/sub\u003eₜ is a binary indicator equal to 1 if child \u003cem\u003ei\u003c/em\u003e migrated between waves \u003cem\u003et\u003c/em\u003e and \u003cem\u003et\u0026thinsp;+\u0026thinsp;1\u003c/em\u003e, α\u003csub\u003ei\u003c/sub\u003e are child fixed effects, \u003cem\u003eShock\u003c/em\u003e\u003csub\u003ei\u003c/sub\u003eₜ is the contemporaneous shock indicator, \u003cem\u003eCumShock\u003c/em\u003e\u003csub\u003ei\u003c/sub\u003eₜ is the cumulative prior shock count, \u003cem\u003eX\u003csub\u003ei\u003c/sub\u003eₜ\u003c/em\u003e is a vector of time-varying controls, and ε\u003csub\u003ei\u003c/sub\u003eₜ is an idiosyncratic error term. Standard errors are clustered at the child level to account for serial correlation in the error term across waves for the same individual.\u003c/p\u003e \u003cp\u003eWe prefer the linear probability specification over logistic fixed-effects models because it produces marginal effects directly interpretable as percentage-point changes and avoids the incidental parameters problem that leads logistic FE to discard all children without outcome variation. Results are robust to complementary log-log and logistic fixed-effects specifications, available upon request.\u003c/p\u003e \u003cp\u003eWe present three nested model specifications for each outcome. Model 1 includes only the contemporaneous shock indicator, isolating its unconfounded within-child effect. Model 2 includes only the cumulative shock measure. Model 3 includes both simultaneously, which is the preferred specification: it allows each shock measure to be interpreted net of the other, cleanly separating the immediate displacement effect from the accumulated stress effect. The inclusion of both measures simultaneously is the critical design feature that allows us to test H1 and H2 jointly and to verify that the cumulative effect is not simply picking up a lagged contemporaneous effect.\u003c/p\u003e \u003cp\u003eTo test H3 and H4, we augment the preferred specification with interaction terms between both shock measures and the continuous household wealth index. The interaction model takes the form:\u003c/p\u003e \u003cp\u003e \u003cem\u003eM\u003c/em\u003e \u003csub\u003ei\u003c/sub\u003eₜ = α\u003csub\u003ei\u003c/sub\u003e + β₁\u003cem\u003eShock\u003c/em\u003e\u003csub\u003ei\u003c/sub\u003eₜ + β₂\u003cem\u003eCumShock\u003c/em\u003e\u003csub\u003ei\u003c/sub\u003eₜ + β₃(\u003cem\u003eShock\u003c/em\u003e\u003csub\u003ei\u003c/sub\u003eₜ \u0026times; \u003cem\u003eWealth\u003c/em\u003e\u003csub\u003ei\u003c/sub\u003eₜ) + β₄(\u003cem\u003eCumShock\u003c/em\u003e\u003csub\u003ei\u003c/sub\u003eₜ \u0026times; \u003cem\u003eWealth\u003c/em\u003e\u003csub\u003ei\u003c/sub\u003eₜ) + β₅\u003cem\u003eWealth\u003c/em\u003e\u003csub\u003ei\u003c/sub\u003eₜ + γ\u003cem\u003eX\u003csub\u003ei\u003c/sub\u003eₜ\u003c/em\u003e ₜ + ε\u003csub\u003ei\u003c/sub\u003eₜ\u003c/p\u003e \u003cp\u003ewhere \u003cem\u003eWealth\u003c/em\u003e\u003csub\u003ei\u003c/sub\u003eₜ is the time-varying household wealth index. The coefficients β₃ and β₄ capture the differential effect of contemporaneous and cumulative shocks respectively across the wealth distribution. A negative β₃ is consistent with H3 (wealth buffers the immediate displacement effect); a positive β₄ is consistent with H4 (wealth enables migration as stress accumulates). Predicted probabilities from this model are plotted across the wealth distribution separately for each level of cumulative exposure, producing the marginal effects shown in Figs.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e and \u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e.\u003c/p\u003e \u003cp\u003eWe additionally estimate an equivalent interaction model replacing the wealth index with a continuous measure of educational attainment to assess whether the heterogeneity patterns are driven by material resources or human capital. The comparison between wealth and education interaction results is theoretically informative: if the enabling mechanism is primarily financial, the SES gradient should be stronger for wealth than for education in the cumulative exposure models.\u003c/p\u003e \u003cp\u003eA critical feature of the fixed-effects design is that identification relies exclusively on children who exhibit within-individual variation in both the predictor and the outcome. Children with no variation in shock exposure across waves \u0026mdash; either always exposed or never exposed \u0026mdash; contribute no identifying information for the shock coefficients and are effectively excluded from the estimation. Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e reports the share of children with variation in each variable by country, and defines the estimating sample accordingly. The identifying sample ranges from approximately 9.9% of children in India to 17.0% in Peru, reflecting differences in both migration rates and shock incidence across countries. These shares are sufficient for reliable estimation in the pooled models but limit precision in country-specific analyses, as discussed in the results.\u003c/p\u003e \u003cp\u003eAll models include wave fixed effects to absorb common temporal shocks \u0026mdash; including macroeconomic trends, policy changes, or aggregate climatic conditions \u0026mdash; that affect all children in the same period. Time-varying individual controls include age and its square, to capture non-linear life-course effects on migration propensity. Country fixed effects are absorbed by the within-country estimation structure in country-stratified models; in pooled models, country fixed effects are included explicitly.\u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003eTable\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e presents the pooled fixed-effects estimates for community migration. The results indicate that migration responses are not driven by contemporaneous shocks, but by the accumulation of prior exposure. Consistent with this, the contemporaneous shock indicator is substantively small and statistically unstable across specifications. In the model including only the current shock, the coefficient is negative and marginally significant (\u0026minus;\u0026thinsp;0.008), and becomes essentially null (0.007) once cumulative exposure is included simultaneously. This pattern is inconsistent with a displacement interpretation of acute shocks and suggests that immediate environmental events do not, on average, trigger community-level migration. If anything, the slightly negative sign in the current-shock-only model points to a short-term immobilizing effect of acute stress, plausibly reflecting liquidity constraints following severe events.\u003c/p\u003e \u003cp\u003eBy contrast, the cumulative shock measure shows a strong and consistent association with migration, providing clear support for H2. Having experienced one prior shock period increases the probability of community migration by approximately 10.4 percentage points relative to households with no prior shock history. A second prior shock period produces an equally large additional increment (10.7 in the full model), indicating a roughly linear dose-response relationship between accumulated stress and migration probability. Crucially, these estimates change negligibly when the contemporaneous indicator is added, confirming that the cumulative and contemporaneous dimensions of shock exposure capture distinct processes rather than collinear variation. The pattern is stable across all model specifications.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eClimate shocks and community migration (pooled, child fixed effects)\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"4\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003e\u003cem\u003eAny shock\u003c/em\u003e\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eCurrent shock\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eCumulative shocks\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eBoth\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003e-0.008+\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.007\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e(0.004)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e(0.004)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cem\u003e1 prior shock\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.104***\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.105***\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e(0.005)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e(0.005)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cem\u003e2 prior shocks\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.104***\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.107***\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e(0.011)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e(0.011)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cem\u003eObservations\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e44,081\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e44,081\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e44,081\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cem\u003eR\u0026sup2;\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.436\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.443\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.443\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"4\"\u003e\u003cem\u003eNote: Linear probability models with child fixed effects. Estimating sample restricted to children with variation in the outcome (n\u0026thinsp;=\u0026thinsp;4,014 unique children; 44,081 person-wave observations). Standard errors clustered at the child level in parentheses. + p\u0026thinsp;\u0026lt;\u0026thinsp;0.1, * p\u0026thinsp;\u0026lt;\u0026thinsp;0.05, ** p\u0026thinsp;\u0026lt;\u0026thinsp;0.01, *** p\u0026thinsp;\u0026lt;\u0026thinsp;0.001.\u003c/em\u003e\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eTable\u0026nbsp;\u003cspan refid=\"Tab4\" class=\"InternalRef\"\u003e4\u003c/span\u003e extends the analysis to rural-to-urban migration, estimated on the subsample of children with rural origin (n\u0026thinsp;=\u0026thinsp;915 with variation in the outcome; 27,570 person-wave observations). The results are broadly consistent with those for community migration and reinforce the central role of cumulative exposure, while also revealing an important distinction in the role of contemporaneous shocks across migration types. The contemporaneous shock indicator is positive and statistically significant both in isolation and in the full model, indicating that acute shocks do carry a displacement signal for rural-to-urban transitions. This pattern suggests that immediate environmental stress can trigger mobility when it directly undermines agricultural livelihoods.\u003c/p\u003e \u003cp\u003eThis does not contradict the weak average effect observed for community migration, but rather reflects the different nature of the decision under consideration. Rural-to-urban migration involves a more substantial reorganization of livelihoods and may therefore be more sensitive to acute disruptions in agricultural production than shorter-distance or lateral moves. Importantly, however, the contemporaneous effect remains substantially smaller than the cumulative effect, indicating that even in this domain, migration responses are primarily structured by accumulated exposure over time.\u003c/p\u003e \u003cp\u003eThe support for H2 is unequivocal. One prior shock period increases the probability of rural-to-urban migration by 6.6 percentage points, while two prior shock periods increase it by 9.2 percentage points. This graded, dose-response pattern is consistent with cumulative vulnerability dynamics: as environmental stress accumulates, it progressively erodes the livelihood foundations that anchor households to rural areas, eventually shifting the balance toward urban relocation. The increase from the 1- to 2-shock coefficient (from 6.6 to 9.2 pp) suggests that the relationship may steepen with repeated exposure, consistent with a threshold-like process in which each additional shock further depletes coping reserves.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab4\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 4\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eClimate shocks and rural-to-urban migration (pooled, child fixed effects)\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"4\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003e\u003cem\u003eAny shock\u003c/em\u003e\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eCurrent shock\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eCumulative shocks\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eBoth\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.010**\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.019***\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e(0.003)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e(0.003)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cem\u003e1 prior shock\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.065***\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.066***\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e(0.004)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e(0.004)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cem\u003e2 prior shocks\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.087***\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.092***\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e(0.007)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e(0.007)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cem\u003eObservations\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e27,570\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e27,570\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e27,570\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cem\u003eR\u0026sup2;\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.432\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.446\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.447\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"4\"\u003e\u003cem\u003eNote: Sample restricted to children with rural origin and variation in the outcome (n\u0026thinsp;=\u0026thinsp;915 unique children; 27,570 person-wave observations). Standard errors clustered at the child level in parentheses. + p\u0026thinsp;\u0026lt;\u0026thinsp;0.1, * p\u0026thinsp;\u0026lt;\u0026thinsp;0.05, ** p\u0026thinsp;\u0026lt;\u0026thinsp;0.01, *** p\u0026thinsp;\u0026lt;\u0026thinsp;0.001.\u003c/em\u003e\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eA potential concern is that shocks are measured at the household level, whereas some observed migration may reflect individual moves out of the parental home rather than household-level relocation. In particular, moves to another community may partly reflect individual moves out of the parental home rather than migration driven by climate shocks affecting the household as a whole. To address this, we conduct two checks. First, we re-estimate the pooled models separately for the younger and older cohorts, since younger children are less likely to leave the parental household independently. Second, we restrict the sample to observations in which the child remains in the original household. Neither check materially alters the results. Effects are somewhat stronger for the older cohort, consistent with greater autonomy over mobility at older ages, and restricting the sample to children who remain in the household produces only minor coefficient changes. Overall, these checks support the main findings. Full results are available upon request.\u003c/p\u003e \u003cp\u003e \u003cem\u003eHeterogeneity by socioeconomic status\u003c/em\u003e \u003c/p\u003e \u003cp\u003eFigures \u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e and \u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e present the central heterogeneity results of the paper and identify the key mechanism linking climate stress, resources, and mobility. Taken together, they show that the effect of socioeconomic status reverses across time horizons, generating the empirical signature of a climate-related mobility trap.\u003c/p\u003e \u003cp\u003eFigure \u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e plots the marginal effect of contemporaneous shocks on community migration across the wealth distribution. The gradient is negative and statistically distinguishable from zero, indicating that higher-SES households are significantly less likely to migrate in response to an immediate environmental shock. This pattern provides clear support for H3 and is consistent with a short-term buffering mechanism: material resources\u0026mdash;liquid assets, savings, access to credit, and diversified income sources\u0026mdash;allow better-off households to absorb acute shocks without relocating. For these households, remaining in place constitutes a viable adaptive strategy. By contrast, lower-SES households exhibit a modest increase in migration propensity, suggesting that acute shocks more readily exceed their limited coping capacity.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003cstrong\u003eNote\u003c/strong\u003e \u003cp\u003e \u003cem\u003eMarginal effects from interaction model. Shaded area\u0026thinsp;=\u0026thinsp;95% confidence interval.\u003c/em\u003e \u003c/p\u003e \u003c/p\u003e \u003cp\u003eFigure \u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e examines the role of cumulative exposure and reveals a qualitatively different pattern. The transition from zero to one prior shock increases migration probability by approximately 9 percentage points, with no meaningful variation across the wealth distribution\u0026mdash;the gradient is flat. In this initial stage, all households respond similarly to accumulated stress. However, the transition from one to two or more prior shocks produces a sharp and statistically significant positive wealth gradient: migration increases substantially among higher-SES households, while remaining flat or increasing only minimally among the poorest. This divergence marks a reversal in the role of resources and provides strong support for H4.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003cstrong\u003eNote\u003c/strong\u003e \u003cp\u003e \u003cem\u003eMarginal effects from interaction model. Shaded area\u0026thinsp;=\u0026thinsp;95% confidence interval.\u003c/em\u003e \u003c/p\u003e \u003c/p\u003e \u003cp\u003eThe contrast between these two stages is central. The first accumulation of stress appears sufficient to generate broadly shared incentives to move, but further accumulation reveals a constraint: the capacity to translate exposure into actual mobility depends on access to material resources. As environmental stress intensifies, better-off households retain the ability to respond through migration, whereas resource-poor households become increasingly unable to do so. The resulting divergence in mobility capacity constitutes the core empirical signature of a structural mobility trap, in which those most exposed to cumulative environmental risk are least able to exit through geographic relocation.\u003c/p\u003e \u003cp\u003eInteractions with educational attainment (Figs.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e and 4) reinforce this interpretation. While higher-educated households display a similar buffering response to contemporaneous shocks, the interaction between education and cumulative exposure is not statistically significant. This dissociation indicates that the enabling mechanism identified above is driven primarily by material resources rather than human capital. Education may shape aspirations and awareness of migration opportunities, but it is liquid wealth\u0026mdash;rather than schooling\u0026mdash;that enables households to finance and organize relocation under conditions of sustained environmental stress.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"No\" id=\"Taba\" border=\"1\"\u003e \u003ccolgroup cols=\"1\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"1\"\u003e\u003cem\u003eNote: Interaction with continuous education variable. Marginal effects with 95% CIs.\u003c/em\u003e\u003c/td\u003e\u003c/tr\u003e \u003ctr\u003e\u003ctd colspan=\"1\"\u003e\u003cem\u003eCountry-level heterogeneity\u003c/em\u003e\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eAs a robustness and generalizability check, Tables\u0026nbsp;\u003cspan refid=\"Tab6\" class=\"InternalRef\"\u003e5\u003c/span\u003ea and \u003cspan refid=\"Tab6\" class=\"InternalRef\"\u003e5\u003c/span\u003eb decompose the pooled results by country. Two observations deserve emphasis. First, the direction of the cumulative shock effects is consistent across all four countries and both migration outcomes \u0026mdash; not a single country-outcome combination reverses the sign. This cross-national consistency is analytically important and indicates that the cumulative vulnerability mechanism is not an artifact of any single country's institutional context, agricultural system, or migration infrastructure, but reflects a more general demographic process operating across very different settings. Second, the magnitude of effects varies substantially, with Peru showing the strongest responses for both outcomes (0.181 pp for community migration and 0.179 pp for rural-to-urban migration at two prior shocks), followed by Vietnam, India, and Ethiopia in descending order for rural-urban flows. The contemporaneous shock results are more heterogeneous \u0026mdash; positive and significant in Ethiopia for community migration, and in Ethiopia, India, and Peru for rural-urban migration \u0026mdash; though generally smaller and less consistent than the cumulative effects.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab5\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 5\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003ea. Climate shocks and community migration by country (child fixed effects)\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"6\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003eAny shock\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eEthiopia\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eIndia\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003ePeru\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eVietnam\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003ePooled\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.043***\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003e-0.001\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-0.009\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003e-0.020**\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.007\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e(0.007)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e(0.008)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e(0.014)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e(0.008)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e(0.004)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e1 prior shock\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.083***\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.121***\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.105***\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.102***\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.105***\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e(0.008)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e(0.010)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e(0.016)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e(0.011)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e(0.005)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e2 prior shocks\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.075***\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.118***\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.181***\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.148***\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.107***\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e(0.013)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e(0.025)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e(0.031)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e(0.032)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e(0.011)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eObservations\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e11,146\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e11,406\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e10,171\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e11,358\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e44,081\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"6\"\u003e\u003cem\u003eNote: All models include both any shock and cumulative shock indicators simultaneously. Standard errors clustered at the child level in parentheses. Country-level estimates should be interpreted cautiously given reduced within-FE sample sizes. + p\u0026thinsp;\u0026lt;\u0026thinsp;0.1, * p\u0026thinsp;\u0026lt;\u0026thinsp;0.05, ** p\u0026thinsp;\u0026lt;\u0026thinsp;0.01, *** p\u0026thinsp;\u0026lt;\u0026thinsp;0.001.\u003c/em\u003e\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eCountry-level estimates should be interpreted with caution given the reduced within-FE sample sizes at the country level, which limit statistical precision. We treat these results as descriptive evidence of directional consistency rather than country-specific effect sizes.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab6\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 5\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eb. Climate shocks and rural-to-urban migration by country (child fixed effects)\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"6\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003eAny shock\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eEthiopia\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eIndia\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003ePeru\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eVietnam\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003ePooled\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.017**\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.009*\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.056***\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.006\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.019***\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e(0.005)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e(0.004)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e(0.013)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e(0.004)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e(0.003)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e1 prior shock\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.056***\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.053***\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.137***\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.057***\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.066***\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e(0.006)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e(0.006)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e(0.017)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e(0.007)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e(0.004)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e2 prior shocks\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.085***\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.054***\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.179***\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.071***\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.092***\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e(0.009)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e(0.015)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e(0.028)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e(0.020)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e(0.007)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eObservations\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e7,170\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e8,480\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e2,826\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e9,094\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e27,570\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"6\"\u003e\u003cem\u003eNote: Sample restricted to children with rural origin. All models include both any shock and cumulative shock indicators simultaneously. Standard errors clustered at the child level in parentheses. + p\u0026thinsp;\u0026lt;\u0026thinsp;0.1, * p\u0026thinsp;\u0026lt;\u0026thinsp;0.05, ** p\u0026thinsp;\u0026lt;\u0026thinsp;0.01, *** p\u0026thinsp;\u0026lt;\u0026thinsp;0.001.\u003c/em\u003e\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e "},{"header":"Discussion","content":"\u003cp\u003eThis paper shows that migration responses to climate stress are not primarily driven by contemporaneous shocks, but by the accumulation of environmental exposure over time, and that the role of socioeconomic resources varies systematically across this process. Using longitudinal data across four countries, we find that contemporaneous shocks have weak and inconsistent effects on migration, whereas cumulative exposure produces large, monotonic increases in mobility.\u003c/p\u003e \u003cp\u003eThe central implication concerns the temporal specification of climate exposure. Much of the existing literature models migration as a response to discrete environmental events, implicitly assuming a short-run causal pathway from shock to mobility. Our results indicate that this approach is misspecified. Migration is more strongly associated with the accumulation of prior exposure than with isolated events, suggesting that the relevant unit of analysis is not the shock itself but the trajectory of environmental stress over time. Because contemporaneous shocks can simultaneously generate displacement and immobility effects, estimates based solely on current exposure are likely biased toward zero. By contrast, cumulative measures capture the progressive erosion of livelihoods that ultimately shifts migration from a constrained option to a feasible or necessary response.\u003c/p\u003e \u003cp\u003eThe second contribution concerns the role of socioeconomic resources. The results show that the association between resources and migration is not stable but reverses over the course of exposure. In the short term, wealth buffers migration: better-off households absorb shocks without relocating. As exposure accumulates, however, wealth becomes an enabling factor. Households with sufficient resources are able to translate accumulated stress into mobility, while poorer households remain in place despite rising environmental pressure.\u003c/p\u003e \u003cp\u003eThis reversal constitutes the empirical signature of a mobility trap. Resource-poor households are not less exposed to environmental stress, but are less able to respond to it through migration. As a result, the capacity to adapt through mobility becomes increasingly stratified. The households most exposed to cumulative risk are also those least able to exit deteriorating environments, while better-off households retain adaptive flexibility. These opposing dynamics are obscured in models that estimate average effects, where buffering and enabling mechanisms offset each other.\u003c/p\u003e \u003cp\u003eThe absence of a comparable pattern for education reinforces this interpretation. While education is associated with reduced migration following contemporaneous shocks, it does not condition migration under cumulative exposure. This suggests that the mechanism identified here operates primarily through material constraints rather than human capital.\u003c/p\u003e \u003cp\u003eThe consistency of these findings across Ethiopia, India, Peru, and Vietnam indicates that the mechanisms identified are not context-specific but reflect a broader demographic process linking environmental stress, resources, and mobility. Migration under climate change emerges as a process of cumulative vulnerability and stratified adaptation, rather than a direct response to discrete events.\u003c/p\u003e \u003cp\u003eSeveral limitations should be noted. First, shock measures are self-reported and capture perceived exposure rather than objective climatic conditions, although the appendix provides validation against independent precipitation data. Second, restricting the sample to individuals observed across all waves introduces selection on residential stability. If attrition is positively correlated with both migration and shock exposure, the estimated effects should be interpreted as lower bounds. Third, unobserved time-varying factors may still affect both exposure and migration, although the fixed-effects design mitigates this concern.\u003c/p\u003e \u003cp\u003eThese findings have implications for both research and policy. Empirically, they suggest that studies relying on contemporaneous shock indicators or short panels systematically underestimate the climate\u0026ndash;migration relationship by failing to capture cumulative dynamics. Substantively, they highlight that the key constraint is not only exposure to environmental risk, but the ability to respond to it. Policies aimed at facilitating climate adaptation through migration must therefore address the material constraints that prevent resource-poor households from moving. Without such support, environmental change is likely to increase not only mobility, but also immobility among the most vulnerable populations.\u003c/p\u003e \u003cp\u003eAs climate change intensifies, these dynamics are likely to become more pronounced, reinforcing the divergence between households able to adapt through mobility and those who remain structurally trapped in deteriorating environments.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eAdil L, Eckstein D, K\u0026uuml;nzel V, Sch\u0026auml;fer L (2025) Climate Risk Index 2026: Who suffers most from extreme weather events? Germanwatch\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBerlemann M, Tran TX (2020) Climate-Related Hazards and Internal Migration: Empirical Evidence for Rural Vietnam. Econ Disasters Clim Change 4:385\u0026ndash;409\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBlack R, Bennett SR, Thomas SM, Beddington JR (2011) Migration as adaptation. Nature 478(7370):447\u0026ndash;449. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1038/478477a\u003c/span\u003e\u003cspan address=\"10.1038/478477a\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBoas I, Farbotko C, Adams H, Sterly H, Bush S, van der Geest K, Hulme M (2019) Climate migration myths. Nat Clim Change 9(12):901\u0026ndash;903. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1038/s41558-019-0633-3\u003c/span\u003e\u003cspan address=\"10.1038/s41558-019-0633-3\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBohra-Mishra P, Oppenheimer M, Hsiang SM (2014) Nonlinear permanent migration response to climatic variations but minimal response to disasters. \u003cem\u003eProceedings of the National Academy of Sciences, 111\u003c/em\u003e(27), 9780\u0026ndash;9785. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1073/pnas.1317166111\u003c/span\u003e\u003cspan address=\"10.1073/pnas.1317166111\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCarling J (2002) Migration in the age of involuntary immobility: Theoretical reflections and Cape Verdean experiences. J Ethnic Migration Stud 28(1):5\u0026ndash;42. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1080/13691830120103912\u003c/span\u003e\u003cspan address=\"10.1080/13691830120103912\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCattaneo C, Peri G (2016) The migration response to increasing temperatures. J Dev Econ 122:127\u0026ndash;146. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.jdeveco.2016.05.004\u003c/span\u003e\u003cspan address=\"10.1016/j.jdeveco.2016.05.004\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCattaneo C, Beine M, Fr\u0026ouml;hlich CJ, Kniveton D, Martinez-Zarzoso I, Mastrorillo M, Schraven B (2019) Human migration in the era of climate change. Rev Environ Econ Policy 13(2):189\u0026ndash;206. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1093/reep/rez008\u003c/span\u003e\u003cspan address=\"10.1093/reep/rez008\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCattaneo C, Shayegh S, Albert C et al (2026) Broadening climate migration research across impacts, adaptation and mitigation. Nat Clim Chang 16:255\u0026ndash;260. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1038/s41558-025-02545-1\u003c/span\u003e\u003cspan address=\"10.1038/s41558-025-02545-1\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChen J, Mueller V (2018) Coastal climate change, soil salinity and human migration in Bangladesh. Nat Clim change 8(11):981\u0026ndash;985\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDannefer D (2003) Cumulative advantage/disadvantage and the life course: Cross-fertilizing age and social science theory. Journals Gerontol Ser B: Psychol Sci Social Sci 58(6):S327\u0026ndash;S337\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eEntzinger H, Scholten P (2022) The role of migration in enhancing resilience to climate change in the Vietnamese Mekong River Delta. Migration Stud 10(1):24\u0026ndash;40\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGray C, Mueller V (2012) Drought and Population Mobility in Rural Ethiopia. World Dev 40(1):134\u0026ndash;145\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHermans-Neumann K, Priess J, Herold M (2017) Human migration, climate variability, and land degradation: hotspots of socio-ecological pressure in Ethiopia. Reg Envriron Chang 17:1479\u0026ndash;1492\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHoffmann R, Dimitrova A, Muttarak R, Crespo Cuaresma J, Peisker J (2020) A meta-analysis of country-level studies on environmental change and migration. Nat Clim Change 10(10):904\u0026ndash;912. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1038/s41558-020-0898-6\u003c/span\u003e\u003cspan address=\"10.1038/s41558-020-0898-6\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHunter LM, Luna JK, Norton RM (2015) Environmental dimensions of migration. Ann Rev Sociol 41:377\u0026ndash;397. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1146/annurev-soc-073014-112223\u003c/span\u003e\u003cspan address=\"10.1146/annurev-soc-073014-112223\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eIPCC, Birkmann J, Liwenga E, Pandey R, Boyd E, Djalante R, Gemenne F, Leal Filho W, Pinho PF, Stringer L, Wrathall D (2022) Poverty, Livelihoods and Sustainable Development. In: P\u0026ouml;rtner -O, Roberts DC, Tignor M, Poloczanska ES, Mintenbeck K, Alegr\u0026iacute;a A, Craig M, Langsdorf S, L\u0026ouml;schke S, M\u0026ouml;ller V, Okem A, Rama B (eds) Climate Change 2022: Impacts, Adaptation and Vulnerability. Contribution of Working Group II to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change [H. Cambridge University Press, Cambridge, UK and New York, NY, USA, pp 1171\u0026ndash;1274\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eIPCC (2007) In: Metz -B, Davidson OR, Bosch PR, Dave R (eds) L.A. Meyer Contribution of Working Group III to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge, United Kingdom and New York, USA\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJha CK, Gupta V, Chattopadhyay U, Sreeraman BA (2018) Migration as adaptation strategy to cope with climate change: A study of farmers' migration in rural India. International Journal of Climate Change Strategies and Management\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKaczan DJ, Orgill-Meyer J (2020) The impact of climate change on migration: a synthesis of recent empirical insights. Clim Change 158(3):281\u0026ndash;300\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLeal Filho W et al (2023) Assessing causes and implications of climate-induced migration in Kenya and Ethiopia. Environmental Science and Policy, p 150\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMilan A, Ho R (2014) Livelihood and migration patterns at different altitudes in the Central Highlands of Peru. Climate Dev 6(1):69\u0026ndash;76\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMueller V, Gray C, Hopping D (2020) Climate-induced migration and unemployment in middle-income Africa. Glob Environ Change 65:102187. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.gloenvcha.2020.102187\u003c/span\u003e\u003cspan address=\"10.1016/j.gloenvcha.2020.102187\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNawrotzki RJ, DeWaard J (2018) Putting trapped populations into place: climate change and inter-district migration flows in Zambia. Reg Envriron Chang 18(2):533\u0026ndash;546\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRicha, Noy I, Sen S (2024) Extreme Events and Inter-state Migration in India: An Empirical Analysis. Migration Dev 13(2):289\u0026ndash;311\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRigaud KK, de Sherbinin A, Jones B, Abu-Ata NE, Adamo S (2021) Groundswell Africa: A deep dive into internal climate migration in Senegal. World Bank\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSchewel K (2020) Understanding immobility: Moving beyond the mobility bias in migration studies. Int Migrat Rev 54(2):328\u0026ndash;355. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1177/0197918319831952\u003c/span\u003e\u003cspan address=\"10.1177/0197918319831952\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eŠedov\u0026aacute; B, Čizmaziov\u0026aacute; L, Cook A (2021) A meta-analysis of climate migration literature. CEPA Discussion Papers, 29. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.25932/publishup-49982\u003c/span\u003e\u003cspan address=\"10.25932/publishup-49982\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eViswanathan B, Kumar KK (2015) Weather, agriculture and rural migration: evidence from state and district level migration in India. Environ Dev Econ 20:469\u0026ndash;492\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWrathall DJ et al (2014) Migration Amidst Climate Rigidity Traps: Resource Politics and Social-Ecological Possibilism in Honduras and Peru. Ann Assoc Am Geogr 104(2):292\u0026ndash;304\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZickgraf C (2021) Theorizing (im)mobility in the face of environmental change. Reg Envriron Chang 21(4):126. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s10113-021-01849-4\u003c/span\u003e\u003cspan address=\"10.1007/s10113-021-01849-4\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"nature-portfolio","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"","title":"Nature Portfolio","twitterHandle":"","acdcEnabled":false,"dfaEnabled":false,"editorialSystem":"ejp","reportingPortfolio":"","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"Climate Shocks, Cumulative disadvantage, Trapped populations, Life course, Longitudinal analysis","lastPublishedDoi":"10.21203/rs.3.rs-9405620/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-9405620/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eClimate change is widely expected to influence human migration, yet most individuals remain immobile despite environmental stress, challenging event-based displacement models. Existing research often overlooks how repeated environmental shocks accumulate over time, reshaping mobility decisions. Here we analyze longitudinal data from Ethiopia, India, Peru, and Vietnam using fixed-effects models to distinguish between acute and cumulative climate stress. We find that contemporaneous shocks have weak effects on migration, whereas cumulative exposure strongly increases mobility. Socioeconomic resources buffer migration in the short term but enable it as stress accumulates, creating a temporal divergence where resource-poor households become trapped while wealthier ones retain adaptive flexibility. These findings suggest that migration under climate change is a process of cumulative vulnerability and stratified adaptation, highlighting the emergence of structurally trapped populations and challenging simplistic event-driven migration models.\u003c/p\u003e","manuscriptTitle":"Cumulative Climate Stress Drives Migration and Stratifies Mobility: Evidence from Ethiopia, India, Peru, and Vietnam","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-05-07 22:48:43","doi":"10.21203/rs.3.rs-9405620/v1","editorialEvents":[],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"nature-communications","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"NCOMMS","sideBox":"Learn more about [Nature Communications](http://www.nature.com/ncomms/)","snPcode":"","submissionUrl":"https://mts-ncomms.nature.com/","title":"Nature Communications","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"ejp","reportingPortfolio":"Nature Communications","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"e2b766f1-82bf-4402-be5a-f1fbc01326cc","owner":[],"postedDate":"May 7th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[{"id":67042156,"name":"Scientific community and society/Social sciences/Sociology"},{"id":67042157,"name":"Scientific community and society/Social sciences/Climate change/Climate-change impacts"}],"tags":[],"updatedAt":"2026-05-07T22:48:44+00:00","versionOfRecord":[],"versionCreatedAt":"2026-05-07 22:48:43","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-9405620","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-9405620","identity":"rs-9405620","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}