Climatic shocks and internal migration - Evidence from 442 million personal records in 64 countries

Abstract

This paper examines whether and how climatic shocks influence individual migration decisions. We use census microdata across 64 countries over the period 1960 to 2012, covering 442 million individual records, combined with geo-referenced temperature and precipitation data summarised for each origin and destination administrative unit. Migration is identified when an individual changed a place of usual residence one, five or ten years ago to a new major administrative unit in the same country. Given an exceptionally large number of observations, we apply a two-step approach to analyse the relationship between exposure to climatic shocks and migration. First, we use random forest models to uncover that in many countries climatic shocks are as important as better-known individual-level covariates in determining migration decisions. This observation serves as a yardstick for the second step of our analysis. For a subset of countries, where rainfall shocks play an important role in migration, we compare internal migration patterns across time by examining whether a region experiencing positive or negative rainfall shocks observed higher or lower migration. We find that negative rainfall shocks suppress outmigration particularly for low-income countries. The opposite is true for positive rainfall shocks whereby migration is found to increase, especially for lower-income countries. Our finding supports the liquidity constraint argument whereby adverse climatic conditions can disrupt migration financing and consequently suppress ability to migrate.