Trapped at Home: Climate Stress More Likely to Displace the Poor than Displace
Yves here. Perhaps readers will beg to differ, but I found this study crazy. It makes two assumptions that I find to undermine its validity. Its express purpose is to challenge long-standing studies, such as those conducted by researchers at the US War College in the early 2000s, that climate change will produce mass migration, for example, through flooding in regions most exposed to sea/sea level rise such as Bangladesh and Sri Lanka. Response from NASA:
Analysis is pointless. First, it looks at gradual changes in temperature, the kind of thing that might lead those who had migrated to the US Sunbelt to move back north, as opposed to the effects of climate change that threaten housing or livelihoods. Although it was an internal migration, the US Dust Bowl of the 1930s led 3.5 million from the Great Plains. In addition to climate change directly producing displacement of people through massive floods or droughts, climate change also creates conflicts over resources, such as water, or fuels ethnic conflicts.
Second, and I’m not making this up, they take legal immigration. Did they fall asleep during the migration caused by war, such as the exodus from Syria in 2014 and 2015, or those in Africa where important values tried to get out of the way of the chaos? Disaster-driven mass or low-level migration is often illegal, witness the migration of Myanmar.
Third, the fact that well-to-do people move and thus can get out of the way of the adverse effects of climate change early on (when the impacts are small) does not indicate that if the effects become severe, that low-income groups will not move out of necessity.
By Céline Azémar, Rodolphe Desbordes, Markus Eberhardt, and Eric Neumayer. Originally published by VoxEU
Climate-driven migration is widely expected to drive millions across international borders. Using monthly bilateral flows for 127 developing countries, this column finds the opposite: temperature shocks reduce migration rather than increase it, because climate stress reduces income and poor households cannot finance expensive international transport. In this context, the ‘climate refugee’ is the exception, and hardship often locks poor people in place rather than moving them.
Editors’ note: This column is based on CEPR Discussion Paper 21634 “Stranded at home: Climate shocks, income, and migration“.
The ‘climate refugee’ has become a policy-making framework. The Institute for Economics and Peace (2020) estimates that environmental threats put 1.2 billion people at risk of displacement by 2050, and security and transport organizations are increasingly treating large-scale cross-border flows as operational projections. The research record is very cautious. In its dedicated assessment, the IPCC (2022) judged the evidence for climate and migration to be mixed, with flows potentially increasing or decreasing depending on the context.
Part of the disagreement is the data problem. A series of bilateral migration long enough to study slow climate trends, constructed from the census stock of Özden et al. (2011), recorded at five- or ten-year intervals, are too coarse to separate the changing climate from everything else associated with it. In an extreme view, the confounders associated with the tropical climate – among them income of origin, migration policy, diaspora networks, and the business cycle at both ends of the migration tunnel – cannot be established, a limitation that has been common for a long time from the literature related to climate and income (Dell et al. 2014, Deshardt 2014 and E2berdes 20).
In a recent paper (Azémar et al. 2026), we revisit the question using monthly data. The migration flow is taken from Chi et al. (2025), who converted the location histories of nearly three billion Meta/Facebook users into a monthly panel of cross-border measures. We limit the sample to 127 developing countries and 180 territories, from 2019 to 2022. The monthly frequency is what makes our work possible: we compare each corridor by itself, we strip everything that is specific to the destination couple during a certain year, everything that is related to the country’s calendar of the year, and every period of global shocks and common global epidemics. the folding of the walk. What remains is the climate anomaly within the corridor: the deviation of the country’s temperature (or precipitation) in a given month from its seasonal norm. And it is this residual variation that determines the transport response.
Warmer-Than-Normal Climate Linked to Less Boundary Movement
Our results argue against the climate refugee issue. A warmer-than-normal year locally (in a developing country) is followed by less cross-border movement, not more. One degree of abnormal temperature reduces bilateral flows by about 5.7% over the next 12 months, while actual rainfall has no discernible effect.
The estimates survive many robustness tests: some fixed effects, dropping the largest migration corridors, regrouping to have a climate related to the area, some specifications to control the epidemic or remove it completely, the exercise of leaving one place, and the placebo test that rearranges the calendar time of each climate effect and disappears. The same result, and the same negligible effect on rainfall, reappears in a completely different data source: annual arrivals for 1990–2019 constructed from three decades of IPUMS census microdata covering 177 origins.
The average effect across countries hides a sharp asymmetry (Figure 1). Dividing the origin from the long-term baseline temperature, the response is close to zero in the cooler part of the origin and large and negative in the already warming part, where a one-degree anomaly reduces immigration by about 16%. The impact is concentrated at the very point where the income-temperature relationship rises the most, and at the point where further warming is expected to decrease the most. Dividing the sample by income leaves the answer unchanged, so this is not a story about poor origins alone: the effect holds across the developing world.
Figure 1 Impact of one degree of real temperature anomaly on international transport
Notes: Impact of one degree of temperature anomaly on migration, cumulative over twelve months, for all developing country origins and origins stratified by long-term baseline temperature average (1990–2018). Estimated Poisson-pseudo-maximum-likelihood for bilateral monthly flows, 2019–2022.
The source: Azémar et al. (2026).
Temperature Shocks Income, Makes Moving Cheaper
Why does the heat keep people at home? Crossing the border is expensive. Visa fees, travel, rental costs, and increased reserves are required by the immigration regimes of previously wealthy countries. A family that loses income cannot simply substitute for an expensive international trip; may lose a means of financial support. This is the resource-blocked immobility trap of Black et al. (2013) and Benveniste et al. (2022), and there is a clear implication: a temperature that depresses transport should depress income.
We show that they do so using two different data sets. In a sub-national GDP per capita grid (Rossi-Hansberg and Zhang 2025), an annual temperature rise of one degree reduces income by about 2%, consistent with the large-scale climate literature (Burke et al. 2015). A monthly panel of satellite night lights, built for the same countries, returns the same negative sign of confusing weather in the monthly frequency at which the transit response is identified. Read together, the two elasticities suggest that each 1% decrease in climate-induced migration is associated with a 2.6% decrease in migration. When budgets are tight, moving becomes out of reach.
If the binding force is financing migration rather than such a temperature, any shock that destroys income and spending power should similarly inhibit migration. We show that it does. Including a monthly index of internal conflict risk (Aizenman et al. 2026) in our regression models, an increase in conflict risk also reduces cross-border migration and nighttime lights of origin, while leaving the climate coefficients unaffected. The climate response is not a subtle consequence of conflict, and the barrier is general rather than climate specific.
The bilateral structure of the transport data allows for one additional test. A rich place is more expensive to get to, so the budget limit should be tighter on routes to rich countries. We confirm this by finding a significantly stronger migration response in corridors to high-income areas. The previously large diaspora, which reduces travel costs and in some routes to the US has been found to transform climate shocks into immigration pressures (Mahajan and Yang 2020), does not produce such a boost here when local income is accounted for. This points to an income barrier rather than a network effect shaping climate-induced international migration.
The ‘Climate Refugee’ Policy Assumptions Are Wrong
These results are directly consistent with how climate change is predicted. A common predictor multiplies the global population’s climate risk by the positive elasticity of immigration. In poor countries, hot origin where it is responsible, that figure has the wrong sign, so it overestimates migration near the borders. The population most exposed to climate damage is often those who cannot finance international migration, so in the near term climate stress is more likely. wet it systematic migration from poor areas rather than sending a wave of refugees to distant rich countries.
These results are worrying. When the main welfare benefit from migration accrues to those left behind, through remittances (di Giovanni et al. 2015), income shocks that pre-empt departures also predict the remittances on which the real economy relies. Cross-border movement is a form of adaptation, and a barrier that tightens as conditions worsen withholds that adaptation from the people innocent of the warming that drives it. Whether those who cannot leave the country instead migrate within it is a boundary that our border data cannot detect, and the next question is obvious.
In the current evidence, the climate refugee is different from the rule. Difficulties of this nature tend to trap poor people in place rather than move them, and a policy debate built on counterintuitive assumptions is addressing the wrong problem.
See the original post for clues
