Panel Paper: Climate change and internal migration in Brazil: the role of geography and road infrastructure

Saturday, November 4, 2017
Ogden (Hyatt Regency Chicago)

*Names in bold indicate Presenter

Claire Brunel, American University and Maggie Liu, Georgetown University


The rise in temperatures as well as the increased frequency of droughts and floods affect productivity in climate-sensitive sectors, notably agriculture. As a result, households whose income depends on agricultural output will experience income shocks, especially rural households in poor countries that are more vulnerable to climate change. Internal migration represents an important channel through which households can cope with these income shocks. Faced with declining revenue, households have an incentive to relocate to areas either less affected by climate change or less dependent on climate-sensitive sectors. However, recent empirical work suggests that while middle-income households do relocate as a result of climate change, poor households do not (Peri and Cattaneo 2015). One explanation could be that high migration costs act as a significant impediment to relocation for poor households facing environmental challenges. Using the state of road infrastructure as a proxy for migration costs, we estimate the extent to which these costs affect environmental migration within Brazil.

The analysis exploits exogenous variation in temperatures and precipitation rates across 137 mesoregions in Brazil and examines the response of long-run bilateral migration flows between 1980 and 2010. The empirical evidence is based notably on a novel road dataset we constructed by digitizing historical maps of road infrastructure, including all roads ranging from dirt roads to multi-lane highways. We combine the road data with geospatial data on climate factors as well as geo-referenced bilateral migration data drawn from decennial censuses. This work examines the increase in the proportion of migrants between mesoregions which accounts for both push and pull factors of migration, as well as the individual decision to migrate for the 1.2 million migrants in our dataset. Our results suggest that migration costs act as a significant deterrent of climate-driven relocation. As the road infrastructure improved overtime, more households moved across mesoregions when facing increased variability of temperature and precipitation, especially poor households relative to middle- or high-income households. In addition, the increase in mobility is particularly strong for households whose main activity is agriculture, though the effect remains significant for manufacturing industries as well, consistent with the literature on how temperatures affect productivity.