Climate Change and Inequality in the U.S.: Sociological Analyses of Big Data

Abstract

Anthropogenic climate change is altering the severity, frequency, and spatial patterning of extreme weather in the United States. Sociologists have argued that the resulting disasters exacerbate social inequality, yet several empirical limitations have left unanswered questions about generalizable causal effects and broader theoretical questions about concepts of vulnerability and risk. The empirical chapters in this dissertation address these limitations by using advanced statistical techniques on large-scale data sources. In particular, the dissertation focuses on disparities in two outcomes related to post-disaster community and individual well-being: neighborhood demographics and birth outcomes. In the first empirical chapter, I investigate the effect of disaster exposure on community-level changes in population demographics. Despite scholarly consensus that post-disaster population change is driven by differential vulnerability to hazards, contradictory evidence remains regarding both the direction and magnitude of the effect of natural hazards on local demographic change. I adjudicate between differing findings using data on the paths of 1,016 severe tornadoes linked to block-group-level population data to estimate the effect of disasters on net demographic change. I find that severe tornadic activity results in no net change in population size but compositional changes, whereby affected neighborhoods become Whiter and socioeconomically advantaged. Moderation models show that the effects are exacerbated for already wealthy communities and that a federal disaster declaration does not mitigate the effects. Given tornadoes’ exogeneity and locals’ inability to make residential decisions based on tornado risk, I argue that my empirical case allows me to isolate the demographic elements constitutive of social vulnerability, and I interpret the findings as evidence of a displacement process by which economically disadvantaged residents are forcibly mobile, and economically advantaged and white locals rebuild rather than relocate. In the second empirical chapter, I interrogate how political institutions, through the delivery and functioning of federal aid for housing, exacerbate post-disaster inequality. Prior research documents aggregate differences in disaster aid amounts. However, they tend to focus on aid types available to local governments and businesses with neighborhood-level data, missing key forms of assistance aimed at household recovery and lacking individual-level data to understand embedded processes. Using household-level data on every application for federal disaster aid after 13 hurricanes from 2005-2016 linked to local estimates of peak wind speed and flood depth from hurricane modeling systems, I estimate racial and socioeconomic disparities in access to FEMA aid. Results demonstrate that key institutional features—the conditions of eligibility and sufficiency, burdens of proof, and bureaucratic interactions with state inspectors—combine in a stepwise process to funnel permanent repair resources to homeowners in whiter communities, while temporary rental aid is granted disproportionately to households in communities of color. Disparities in aid amounts across communities then structures unequal socioeconomic growth during recovery. An approach to disasters as the realization of climate risks incorporates insights from political sociology and social stratification to elucidate how the state intervenes to produce inequalities through institutional processes. In the final empirical chapter, I examine how in utero exposure to extreme heat impacts birth weight. While some evidence suggests that heat negatively impacts newborn health, existing studies tend to analyze data from a singular study site over a narrow time frame. I use near-population-level data on live births to U.S.-born White, U.S.-born Black, and Mexican-born Hispanic mothers for two birth cohorts—1970-1974 and 2005-2009—to estimate how effects vary across maternal race and access to key adaptation resources, namely residential air conditioning. Results show that exposure to heat in the last pregnancy month lowers newborn birth weight, on average, but no effect is detected for immigrant Hispanic mothers—the group with the highest exposure due to spatial and temporal nature of risk. Moreover, I demonstrate that in the 1970s cohort, negative effects of hot minimum temperature on birth weight were both significant and meaningful in places with low residential AC coverage, a relationship that has likely been attenuated by increased AC coverage in American households by the 21st century.