Extreme weather events are increasingly common as a result of climate change. Yet little is known about how exceptional climate shocks affect the lives of those most vulnerable to them, or about the barriers they face to moving out of harm's way. In this project, we study the effects of the flooding in Pakistan, which has affected 33 million households and left one third of the country under water.
We compile a panel dataset on a random sample of 5,000 ultra-poor rural households, tracking their migration, assets, employment, and welfare outcomes at (up to) monthly intervals over 3-5 years. Using the data, we document the effects of a catastrophic flood, including costs of adaptation, by comparing the post-flood outcomes of 3,000 flood-affected households to those of 2,000 flood-unaffected households. Given that flooding was rainfall driven and entirely unanticipated, we consider this comparison to be the causal effect of exposure to floods on welfare, migration, and occupational change.
We also test for three barriers that may explain limited adaptation to climate change through migration and occupational change: (i) physical distance to migration destinations; (ii) presence of strong village networks; and (iii) ownership of sector- or location-specific immobile assets. Unforeseen floods shock these factors in plausibly random ways, allowing us to unpick the impact of frictions.