Date of Award

Summer 8-1-2020

Degree Type


Degree Name

Doctor of Philosophy (PhD)



First Advisor

Tim Sass

Second Advisor

Dan Kreisman

Third Advisor

Garth Heutel

Fourth Advisor

Dajun Dai


Coal ash accounts for one third of industrial water pollution in the United States. In Chapter 1, I assess the relationship between coal ash surface water discharges and three relevant outcomes: surface water quality, municipal system water quality, and fetal health indicators from a birth certificate database in North Carolina. Identification relies on geographic variation in downstream status of monitoring sites and municipal water intake locations, plant closures or conversions, and the relative quantity of coal ash released over time. I find that coal ash releases are associated with higher conductivity and pH in both downstream surface waters and municipal water supplies sourced from these waters. Water systems affected by coal ash tend to have more Safe Drinking Water Act violations for disinfectant byproducts, inorganic chemicals, and health-based violations. I quantify the costs of coal ash water pollution with respect to fetal health and home sales. Exploiting variation arising from mothers' moves, I find that a newborn potentially exposed to coal ash water pollution is 1.7 percentage points more likely to have low birthweight compared to an unexposed sibling. I conclude by estimating how a legislative act mandating drinking well testing affected home sale prices in regions around coal ash plants. After the act, sale prices of homes within 1 mile of coal ash ponds declined by 12-14%, or over $37,000.

Chapter 2 investigates how school-age children are affected by diesel emissions from school buses. Diesel emissions from school buses expose children to high levels of air pollution; retrofitting bus engines can substantially reduce this exposure. Using variation from 2,656 retrofits across Georgia, we estimate effects of emissions reductions on district-level health and academic achievement. We demonstrate positive effects on respiratory health, measured by a statewide test of aerobic capacity. Placebo tests on body mass index show no impact. We also find that retrofitting districts experience significant test score gains in English and smaller gains in math. Our results suggest that engine retrofits can have meaningful and cost-effective impacts on health and cognitive functioning.

Chapter 3 explores farm-to-school policies. School meal provision represents one of the largest food markets in the country. In 2015, 42,000 schools serving 23.6 million students engaged in farm-to-school nutrition sourcing policies. Yet, little is known about how much school systems actually source their food locally or about the average relationship between farm-to-school policy adoption and local sourcing of school food. I link 17 years of school district nutrition expenditures across the state of Georgia to a unique commodity-by-county survey of agricultural revenues to assess how much school systems source food from within their county and neighboring counties. I then incorporate four years of survey-based information on district farm-to-school policies to test how farm-to-school programs differentially impact local sourcing patterns. Identification comes from spatiotemporal variation in school district adoption of a farm-to-school policy and variation in expenditures associated with the community eligibility provision of the Healthy Hunger Free Kids Act. Results suggest that as much as $966M of school nutrition expenditures flow to producers within the same county. Of this, perhaps as much as $680M, or 0.6% of all agricultural revenues in the state from 2001-2017, are associated with adoption of farm-to-school policies by school districts.