Email: @email
Advisor: Christine L. Crago (chair); committee: Jamie T. Mullins, Matt Woerman
Title of Job Market Paper: “Data Centers and Transmission Constraints” (with Christine L. Crago and Matt Woerman)
Abstract: The rapid growth of artificial intelligence (AI) and cloud computing has made data centers one of the fastest-growing sources of electricity demand in the United States. We construct a spatial dataset linking data centers and electricity generator nodes through the transmission network and estimate the effect of data center expansion on electricity grid congestion from 2018 to 2025. Leveraging variation in generator nodes’ exposure to data center capacity and an instrumental variable strategy, we find that data center expansion affects congestion at both nearby nodes and more distant nodes. A 100 MW data center connected near a generator node increases the node’s congestion cost by $6.5/MWh, and the effect declines with distance.
Email: [email protected]
Advisor: Yongjoon Park and Lucy Wang (co-chairs); committee: Christopher Boone
Title of Job Market Paper: “Natural Gas Transition, Air Pollution, and Public Health: Evidence from Turkey’s Infrastructure Rollout”
Abstract: This paper estimates the health effects of Turkey’s transition from solid fuels to natural gas, driven by the staggered rollout of natural gas distribution networks across provinces. Using data on actual natural gas consumption for all 81 provinces from 2011 to 2019, I exploit variation in both the timing and the intensity of residential adoption. The main analysis uses a dynamic difference-in-differences estimator for continuous treatment, complemented by an instrumental variables strategy that leverages weather-driven variation in heating demand. A 10 m³ per-capita increase in residential natural gas consumption (equivalent to about 2.6 gallons of heating oil) reduces circulatory mortality by 3.4%, respiratory mortality by 4.7%, and hospital admissions by 3.4%, with effects accumulating over about two years. Back-of-the-envelope calculations based on the IV estimates imply roughly 3,500 fewer pollution-related deaths and 4 million fewer healthcare visits nationally in a year. The same increase lowers annual average PM10 concentrations by about 1.6 μg/m³. In contrast, industrial natural gas consumption shows no detectable health effects and much smaller effects on air pollution, consistent with households replacing coal and wood burned close to where people live. Relative to prior work, the paper uses actual consumption data to capture the dose-response nature of fuel switching, extends the analysis from mortality to morbidity, separates residential from industrial use, and traces the full pathway from cleaner fuel to cleaner air to better health.
Email: @email
Advisors: Jamie T. Mullins and Matt Woerman (primary advisors); committee: Christine L. Crago
Title of Job Market Paper: “Who Pays When a Coal Plant Closes?”
Abstract: Between 2010 and 2018, 525 U.S. coal-fired generators representing approximately 74.6 GW of nameplate capacity retired, removing major taxpayers from some host communities. I study who bears the resulting fiscal loss by linking generator-level Energy Information Administration records to county and school-district finances, together with plant-level assessment records from three states. Identification uses staggered difference-in-differences, matched stacked designs, and a shift-share instrument based on the Mercury and Air Toxics Standards. Plant-class taxable value falls by about 68 log points within three years of retirement. In Ohio, the host school district’s total tax base falls by 19 log points, while higher tax rates replace only 14 percent of the lost revenue. Nationally, county property-tax revenue falls by about 4.5 percent after a full plant exit, but does not decline after coal-to-gas conversions that keep the site on the tax rolls. County spending and school-district revenue do not measurably fall. The fiscal cost is therefore real but modest relative to local budgets and is absorbed primarily through adjustments on the revenue side.
Email: @email
Advisor: Jamie T. Mullins and David A. Keiser (co-advisors)
Title of Job Market Paper: “Targeted Interpretability for Estimating Policy Parameters: Evidence from U.S.
Gasoline Demand”
Abstract: Economists usually settle the trade-off between interpretability and flexibility for the whole model at once. I argue for settling it part by part, an approach I call targeted interpretability: keep interpretable only the parameter that policy uses, and let the data determine the rest, including nonlinearities and interactions, as the partially linear model estimated by double/debiased machine learning allows. Interpretability where policy does not need it can be costly: linear controls bias the estimate of that parameter, even in large samples, whenever their approximation errors in the instrument and outcome equations are correlated, and this bias is testable. In a new weekly U.S. state panel, the approach yields a gasoline price elasticity of −0.155, against −0.512 with linear controls. Under the Ramsey rule, −0.512 implies a gasoline tax of 28 to 70 percent of the price, whereas −0.155 implies no upper limit, so the estimates call for different policies.
Email: @email
Advisor: Christine L. Crago (chair); committee: Yongjoon Park, Matt Woerman
Title of Job Market Paper: “Subsidies, Cost Trends, and Pass-Through: Evidence from the Residential Solar Market”
Abstract: Clean energy subsidies are one of the main instruments in climate policy. Their values are designed to fall as technology adoption grows, but costs generally change over the same horizon, confounding the estimates of pass-through. I study two Massachusetts programs that pay homeowners per kilowatt-hour based on energy produced. To overcome timing challenges, I employ a novel identification strategy: estimating pass-through from differences in the subsidy across roofs that the same installer equips at the same time, because the subsidy depends on each roof’s generation. I find that buyers receive about 85 percent of the subsidy under both designs, whereas a comparison over time gives a biased estimate. Installers raise prices only slightly for roofs that earn larger subsidies, which in this setting bounds the markup from roof-specific price discrimination at 2–7 percent of price.