One of the major unresolved questions in the study of vulnerability to climate change is how human migration will respond in low and middle-income countries. The present study directly addresses this lacuna by using census data on migration from 4 million individuals from three middle-income African countries over a 22-year period. We link these individuals to climate exposures in their origins and estimate climatic effects on migration using a fixed-effects regression model. We show that climate anomalies affect mobility in all three countries. Specifically, mobility declines by 19% with a 1-standard deviation increase in temperature in Botswana. Equivalent changes in precipitation cause declines in migration in Botswana (11%) and Kenya (10%), and increases in migration in Zambia (24%). The mechanisms underlying these effects appear to differ by country. Negative associations between precipitation anomalies, unemployment, and inactivity suggest migration declines may be due to an increased local demand for workers to offset production risk, while migration increases may be indicative of new opportunities in destinations. These country-specific findings highlight the contextually-specific nature of climate-migration relationships, and do not support claims that climate change is widely contributing to urbanization across Africa.

气候变化脆弱性研究中尚未解决的主要问题之一是低收入和中等收入国家的人类移民将如何应对。本研究利用 22 年来来自三个中等收入非洲国家 400 万人的人口普查数据，直接解决了这一缺陷。我们将这些人与其起源地的气候暴露联系起来，并使用固定效应回归模型估计气候对移民的影响。我们表明气候异常影响这三个国家的流动性。具体而言，博茨瓦纳的气温每升高 1 个标准差，流动性就会下降 19%。降水量的同等变化导致博茨瓦纳（11%）和肯尼亚（10%）的移民减少，而赞比亚（24%）的移民增加。这些影响背后的机制似乎因国家而异。降水异常、失业和不活动之间的负相关表明，移民减少可能是由于当地对工人的需求增加以抵消生产风险，而移民增加可能表明目的地出现了新的机会。这些针对具体国家的调查结果强调了气候与移民关系的具体情况，并不支持气候变化广泛促进整个非洲城市化的说法。