Climate-induced migration is increasingly recognized as one of the central adaptation challenges of a warming world. Yet most of the economics literature has focused on permanent relocation as the primary margin of adjustment. This digest brings together papers that complicate that picture: by asking not just how much migration climate change induces, but what kind, and whether the households most exposed to climate stress can actually afford to move at all.
Temporary Migration
In my Das & Sanyal (2025), we study the role of temporary migration — shorterterm moves where workers maintain residential ties to their origin — as a distinct adaptation channel under heat stress in India. The distinction from permanent migration matters both quantitatively and qualitatively: India counted an estimated 13.6 million short-term migrants in a single year against 97 million permanent migrants over an entire decade, and temporary migrants are systematically poorer, less educated, and more likely to come from disadvantaged communities. Using the Consumer Pyramids Household Survey, a large-scale panel repeated three times a year, matched to ERA-5 weather data, we find that a one-degree rise in mean daily temperature during the summer growing season raises an individual’s probability of temporary out-migration by 6.1% (18% increase over the mean).
Modeling Climate Mobility
The reduced-form result establishes the behavioral response, but it cannot speak to what happens under widespread, simultaneous climate change, where all locations are affected and spatial spillovers are crucial. To investigate this general equilibrium problem, we build a multi-location, dual-sector spatial equilibrium model in which households choose between staying, migrating temporarily, or migrating permanently. A novel feature of the model is that destination amenitiesmdeteriorate with the share of temporary migrants in the local labor force – not because of general crowding, but because temporary migrants are systematically undercounted in the administrative systems that plan for housing, water, and sanitation, generating a specific under-provisioning of public services. We calibrate this externality so that a 10-percentage-point rise in the temporary migrant share degrades local amenities by 1.26.
Policy Evaluation
Simulating climate change under the IPCC’s SSP5-8.5 scenario, we find that restricting temporary migration generates a welfare cost of -2.73%, larger than the -2.12% cost of restricting permanent migration, even though permanent migration is the channel most models and policies treat as the dominant adjustment margin. We then compare two cost-equivalent policy responses: in-place adaptation that
restores productivity in climate-affected origin areas, and friction-reduction policies that remedy the administrative under-provisioning of services for temporary migrants at destinations. The friction-reduction policy delivers more than three times the welfare gain of the in-place adaptation policy, though at a modest cost to aggregate output. The two policies serve different goals- in-place adaptation restores output; friction-reduction improves welfare. Specific interventions that map to friction-reduction policies, such as migrant registration systems like India’s E-Shram portal or the Affordable Rental Housing Complexes scheme, can thus be seen as attractive candidates for cost-effective climate adaptation spending. The nested choice structure in our model, which distinguishes staying, temporary migration, and permanent migration as distinct decisions rather than a binary stayor-go, builds on Imbert et al. (2025) and Rai (2023). Rai finds that temporary flows
matter 1.3 times more than permanent migration for average welfare in India and five times more for the poorest 10% of Indian districts. This is a finding consistent with our own, though his model does not incorporate the amenity externality we identify.
Migration Constraints
Our India evidence establishes that heat stress raises temporary out-migration. Kafle et al. (2023) find the opposite relationship in a different setting. Using longitudinal data from Uganda’s nationally representative LSMS-ISA survey across seven waves between 2009 and 2020, and exploiting within-household variation with two-way fixed effects, they show that self-reported weather shocks like drought reduce the probability of a household having at least one temporary migrant by 2.2 percentage points, with parallel reductions in the number of migrants and total migrant months. The negative effect holds for both the bottom 40% and middle 40% of households by consumption expenditure, but disappears entirely for the wealthiest 20%. Among poor households, weather shocks reduce the value of agricultural production and agricultural revenue, depleting the cash needed to finance migration.The paper also finds that persistent weather shocks over many years are positively associated with migration, suggesting that while a single shock can trap the poorest households in place, repeated climate stress eventually forces displacement even among those with the fewest resources.
Heterogeneous Responses
Benveniste (2025) offer a methodological and empirical complement to both papers above, though one that focuses on migration broadly rather than temporary migration specifically. Combining causal inference methods with cross-validation techniques applied to global migration data spanning both internal and crossborder flows, they show that allowing weather effects to differ by age and education improves out-of-sample predictive performance by a factor of five or more compared with models that estimate a single average effect. The implication is that studies documenting small or ambiguous population-average climate-migration relationships may be obscuring large, opposing effects across demographic groups that offset each other in the aggregate. Their projections indicate that climate change’s effects on future cross-border migration will be an order of magnitude larger for most demographic groups than average estimates suggest, but pointing in different directions for different populations.
Key Lesson
Together, these papers suggest that the welfare stakes of climate-induced mobility turn critically on context: on baseline migration capacity, on whether the relevant climate shock depletes or preserves the resources needed to move, and on how well destination institutions are equipped to serve the populations that do arrive.