We present unique survey data on the migration predictions of 400 households in two extremely climate exposed unions of coastal Bangladesh. We have four main findings. First, despite having prospects no better than many low-lying pacific islands, few households in our two locations expect to relocate elsewhere over the coming five-year period. Second, to the extent that households predict they will move in the near future, they believe that fast onset events such as cyclones will be a main reason - not slow changing environmental factors like increasing soil salinity. Third, household migration predictions correlate non-linearly with household assets; the poorest and the richest households are the most likely to move. Fourth, results from an embedded discrete choice experiment suggest that the poor are more likely to migrate in scenarios where their wages are low, while the rich are more likely to migrate in scenarios where their earnings are high. One possible interpretation of these results is that the poor expect to migrate because and when they have to, while the rich expect to migrate because and when they can. Our discrete choice experiment confirms that households expect to move if there is considerable destruction of property from fast onset events, but not due to gradual erosion of environmental conditions. In sum, our results suggest that households in climate exposed regions to a limited extent perceive migration as an adaptation strategy to climate change.


The authors thank Rune Jansen Hagen, Katrin Millock and Solomon Walelign for helpful suggestions and comments.

Ivar Kolstad

Associated Research Professor

Sosina Bezu

Associated Senior Researcher