Climate Change and Civil Violence
The manifestations of climate change can result in humanitarian impacts that reverse progress in poverty- reduction, create shortages of food and resources, lead to migration, and ultimately result in civil violence and conflict. Within the continent of Africa, we have found that environmentally-related variables are either the cause or the confounding factor for over 80% of the civil violence events during the last 10 years. Using predictive climate models and land-use data, we are able to identify populations in Africa that are likely to experience the most severe climate-related shocks. Through geospatial analysis, we are able to overlay these areas of high risk with assessments of both the local population's resiliency and the region's capacity to respond to climate shocks should they occur. The net result of the analysis is the identification of locations that are becoming particularly vulnerable to future civil violence events (vulnerability hotspots) as a result of the manifestations of climate change. For each population group, over 600 social, economic, political, and environmental indicators are integrated statistically to measures the vulnerability of African populations to environmental change. The indicator time-series are filtered for data availability and redundancy, broadly ordered into four categories (social, political, economic and environmental), standardized and normalized. Within each category, the dominant modes of variability are isolated by principal component analysis and the loadings of each component for each variable are used to devise composite index scores. Comparisons of past vulnerability with known environmentally-related conflicts demonstrates the role that such vulnerability hotspot maps can play in evaluating both the potential for, and the significance of, environmentally-related civil violence events. Furthermore, the analysis reveals the major variables that are responsible for the population's vulnerability and therefore provides an opportunity for targeted proactive measures to mitigate certain classes of future civil violence events.
Effect of Climate-Induced Change in Crop Yields on Emigration: The Case of Mexico
Researchers have suggested several channels through which future global warming could trigger mass migration across country borders. This paper examines one of them by focusing on the effect of climate- induced crop failures on out-migration. Using data from Mexico, we identify and estimate elasticity of emigration with respect to changes in crop yield, which sheds light on the possible magnitudes of migrant flows for other areas of the world under different climate change scenarios. We choose Mexico as the study object as it is by far the largest migrant-sending country, with an estimated number of emigrants living in the United States to be well over 10 million. In addition, over 20% of Mexico population directly relies on the agricultural sector, which is heavily dependent on climate. For example, the prolonged drought from 1996 to 1998 in northern Mexico resulted in mass crop failures and the death of livestock. Historically, farmers have been using emigration as an adaptation strategy to cope with crop yield reductions. We first examine the relationship between corn yields and climate variables for the period of 1980-2000, using state-level data. We find significant positive effects of annual precipitation and annual average temperature, but a negative effect of summer temperature on corn yields. The effects of both annual and summer temperatures are also nonlinear. Our analyses of other crops such as wheat yield very similar results. Using Mexico Census micro data, we calculate the number of emigrants from each state for the periods of 1990-1995 and 1995-2000. We then regress changes in the number of emigrants on changes in crop yields, instrumented by changes in temperatures and precipitation. Our preferred specification gives an elasticity of -4, which suggests that a 25% reduction in crop yields would double the number of emigrants. The null hypothesis of no effect is rejected at the 5% significance level.