Le CIRED félicite Mélanie Gittard (PSE, CIRED) qui a obtenu une mention spéciale dans le cadre du « Prix FAERE du meilleur article de jeunes économistes » pour son papier « Climate Variability, Migration and Population in Kenya » lors de la 8ème Conférence annuelle de la FAERE à Grenoble.
Résumé (en anglais)
Over the past decades, East Africa has faced repetitive climate extremes and changes in precipitation trends, driving modi cation of demographic patterns. This paper studies the effects of past climate variability on inter-district migration and intra-district population, at the level of the 3715 Kenyan sub-locations. This study contributes to the micro-oriented literature on climate-induced migration, by decompressing history, using long term, local, precise and representative data. We match, over twenty years, population data at the sub-location level, from three exhaustive administrative censuses (1989, 1999, and 2009), with high spatial and temporal resolution precipitation and temperature data sets from the Climate Hazard Center (CHIRPS/CHIRTS-ERA). It is also a methodological contribution, advocating for the use of relevant climatic indicators (and data) and local demographic effects in order to understand the impacts of climate variability on internal migration. A particular attention is devoted to the definition and analysis of climate variability and changes in precipitation trends over the country.
At the district level, a yearly panel on bilateral migration is built thanks to retrospective questions, and permanent migrations are distinguished from return/seasonal movements thanks to micro-level data (the literature pointing to different magnitude of effects between temporary and permanent movements. The results suggest that climate events act as push factors on inter-district migration rather than pull factors, but with no statistical significance, showing the necessity to estimate migration locally. The main limitation of this first analysis is the spatial resolution, increasing the risk of concomitance between climatic and economic variables, and erasing an intra-district migration which should be important in the hypothesis of short-distance movements. Thus, a second analysis is made
at a much finer scale. A decadal panel fixed effect model, at the sub-location level, is used in order to understand the incidence of climate variability on migration behaviors and sub-location’s population changes and to tackle endogeneity issues. In order to estimate the magnitudes, and demographic decomposition of the induced migration, a demographic record of migration, according to gender, economic activity, age brackets is built. The results suggest that an additional dry rainy season over the decades implies a decrease of -2 percentage points (p.p) of the decadal population growth rate, and that the induced migration is about 48 % female (so 52 % male),95 % of the working population with its own business (such as agricultural holdings, proxy for individuals involved in agricultural activity) and 92 % in working age (no effects on inactive population and infantile mortality). Climate migrants have attended at least primary education, while the population from the low end of the skill distribution significantly stay in affected areas (in line with a poverty trap story). A Difference-in-Difference identifies the effect in West-Center of the country, and mainly borne by rural sub localities where pastoralism in the main sector of activity.