Weather Anomalies, Crop Yields, and Migration in the US Corn Belt

We investigate the influence of weather anomalies on net migration in the Eastern United States using a county-level panel for the period from 1970 to 2009. There is a significant weather-migration relationship in the Corn Belt, but not outside of it. We present evidence that weather affects migration through its influence on agricultural productivity using an instrumental variables approach. Our preferred model uses the seasonality of the sensitivity of corn yields to extreme heat over the growing season, which peaks during corn flowering, as instrument. The reduced-form estimate of the migration response to extreme heat closely mirrors the seasonality of corn yield. Therefore, our IV approach provides an unbiased estimate of the responsiveness of outmigration to yield unless other determinants of migration, such as a direct preference for weather, perfectly align with the pattern of corn flowering. This is unlikely given that the exact dates of corn flowering vary from year to year. Our estimated semi-elasticity ranges from -0.3 to -0.4 depending on the chosen time trend, i.e., a one percent change in yields leads to an opposite 0.3-0.4 percentage point change in the net migration rate. The migration response is strongest for young adults and not significant for senior citizens. We cordially acknowledge the contribution to this paper by Alan Krueger before he started his government position as the chairman of the Council of Economic Advisers. We appreciate comments from participants at the Norface Migration Conference at University College London. All remaining errors are ours. ♣ School of Economics, Shanghai University of Finance and Economics and Department of Economics, Chinese University of Hong Kong. Email: [email protected] ♠Department of Geosciences andWoodrowWilson School, Princeton University. Email: [email protected] ♥ School of International and Public Affairs (SIPA) and Earth Institute, Columbia University. Email: [email protected]. We investigate the effect of weather variability on migration patterns of the U.S. population, especially through its impact on agricultural productivity. The associations among changes in climatic conditions, agricultural productivity, and human migration have been most vividly illustrated by the famous “American Dust Bowl,” one of the greatest environmental catastrophes in U.S. history. In the 1930s, exceptional droughts (Schubert et al. 2004), amplified by human-induced land degradation (Cook, Miller & Seager 2009), greatly depressed agricultural productivity in the Great Plains and led to large-scale and persistent net outmigration from those regions. Between 1935 and 1941, around 300,000 people migrated from the southern Great Plains to California (McLeman 2006). Hornbeck (2009) compares counties with different levels of soil-erosions in the Great Plains, and finds that the 1930s Dust Bowl generated persistent population loss in the following decades. In addition, the overall decline in population did not occur disproportionately for farmers, but had ramifications beyond the agricultural sector. This suggests a general economic decline that extends beyond the direct effect on agriculture. Many other businesses in agricultural areas, e.g., banking and insurance, are directly linked to the agricultural sector as they serve the agricultural community. Hornbeck (2009) argues that the economy mainly adapted through outmigration, not adjustment within the agricultural sector or increases in industry. The “American Dust Bowl” happened under very different conditions from today’s. It overlapped the Great Depression and a lack of credit may have limited the local capacity for adaptation. Since then, the American agricultural sector has undergone immense changes. On the one hand, it is much more mechanized and uses great amounts of chemical fertilizer and pesticides. As a result, it now accounts for a much smaller part of the overall economy and a smaller fraction of the population directly depends on agricultural outcomes. On the other hand, better communication and transportation networks may make the present generation of Americans more mobile. In either case, one might expect today’s relationship between migration and agricultural productivity to be different from the 1930s. To assess the possible magnitudes of migration flows under future climate change, it is necessary to base empirical work on more recent experience, which we do in this paper. In particular, we examine whether net migration rates over five year intervals between 1970 and 2009, defined as the fraction of people leaving a county net of new arrivals and deaths, are related to contemporaneous observed weather variations. We find a significant relationship in counties of the Corn Belt (which include all Midwestern states and Kentucky), but not outside the Corn Belt. We show that the main mechanism for the observed weather-migration relationship in the Corn Belt is through agricultural productivity, and