By Jonas Vestby, Andreas Forø Tollefsen, Halvard Buhaug
Over the past decade, academics and policy actors alike have raised awareness to climate change as a potential major driver of human mobility. Despite undisputable connections between extreme weather events and short-term displacement, empirical evidence linking broader categories of climatic conditions with migratory outcomes remains mixed, contrasting the widespread but contentious notion of “climate refugees”.
Within this context, the authors find that
- Methodological choices have a substantial impact on the estimated effects of temperature and precipitation on international migration flows. Methodological choices are likely an important cause for the variance of estimated effects reported in published literature.
- Measurement uncertainties – unaccounted for in published literature – have substantial impact on our estimates. We know less about the magnitude and consistency of climate effects on international migration than what published estimates lead us to believe.
- It is not clear that we are able to identify causal effects of climatic exposure on international migration flow using the gravity model with fixed effects. Particularly, the assumption of treatment non-interference seems to be violated. No published article directly controls for spatio-temporal/network dependencies. We are unable to find an econometric approach that does this in an adequate fashion.
- Evaluating predictive performance is a recommended supplement or alternative to causal inference, particularly in observed studies where assumptions in causal inference are difficult to evaluate. Published results exploring the predictive power of climatic variables on international migration flow show that such variables are poor predictors.
- Most international migration occurs between countries that have fairly similar climatic conditions (e.g., neighboring countries).