Three Essays on Climate and Development
Abstract
My thesis studies the impact of weather shocks in developing countries. Each chapter of my thesis focuses on a different outcome variable that may be affected by weather shocks: migration, agricultural GDP, and education. The first two are analyzed in the context of Eastern Africa, with data from Kenya and Tanzania, and use data spanning several decades. The third studies a specific tropical storm in Guatemala using data for several years before and after the storm.
My first chapter addresses two questions: What is the inter-regional migration response to weather in Kenya and Tanzania? And does this response vary depending on geography, socio-economic factors, and the time period under consideration? Using a combination of census, satellite, and climate model data, this relationship is estimated with a gravity model of regional bilateral migration. Results show that geography is a significant and robust source of heterogeneity. A region's baseline level of aridity can change the direction of the effects of rainfall measures. Warm arid regions experience higher out-migration as a response to ideal temperatures and are also the regions that are most influenced by weather variables. Most regions show a trend of restricted migration in the short-run in response to bad weather but increased migration over longer periods of time. The spatial and temporal heterogeneity found in this chapter builds a foundation for understanding some of the contradictions found in previous literature.
The next chapter uses the same climate model data as Chapter 1, as well as nighttime lights satellite data as a proxy for GDP, to study the relationship between weather shocks and agricultural total factor productivity (TFP) in Kenya and Tanzania. It focuses on finding spatial heterogeneity in the relationship to identify which areas are most sensitive to shocks and how the reactions differ by physical geography. Results show that ideal temperatures for crops cause a significant boost to TFP growth and extreme temperatures decrease TFP growth. Spatial heterogeneity exists between geographic zones. Warm arid regions have positive TFP growth after extreme rainfall conditions, warm humid regions see a decrease in output with extensive dry periods, and cool arid regions have a different reaction to temperature from other regions. The regional differences found in this chapter show that averaging weather shocks at the national level does not fully capture the relationship between weather and TFP growth.
The third chapter is co-authored with Fidel Perez Macal and examines how Tropical Storm Agatha affected test scores for high school students in Guatemala. Using student and school surveys and satellite rainfall data, we match test scores for math and reading with a full set of personal and educational controls and continuous, non-linear measures of exposure to Agatha. Panel data techniques allow us to estimate this relationship while controlling for school-level fixed effects. We find that exposure to Agatha significantly reduced test scores and that the main mechanism through which this happened was health vulnerability. Agatha affected students heterogeneously, most notably for vulnerable students who suffered disproportionately and had lingering effects years after the disaster. Ultimately, this paper shows that disasters, education, health, and inequality are inextricably linked and that the true cost of a disaster is highly complex.