Research
Job Market Paper
Investment and the Transfer of Power: Dynamic Effects of Transmission in Electricity Markets
Renewable resources are essential for energy transition, but unequal geographic distribution limits nationwide adoption. Long-distance transmission offers a potential solution by connecting renewables-rich areas to high-demand markets. I examine how expanded transmission capacity impacts investment decisions by generators. I develop and estimate a dynamic model of generator behavior consisting of a short-run optimal dispatch problem incorporating line losses and zonal transmission constraints and a long-run dynamic game for capacity investment. Using data from ISOs in the Eastern Interconnection and ERCOT from 2018-2023, I find that upgrading all interzonal transmission to modern HVDC standards would reduce average wholesale prices by 6%, generating approximately $4 billion in annual welfare gains – roughly one-third of total congestion costs. Even partial upgrades (50% of lines) achieve over 5% price reductions. Regional effects are heterogeneous: the Northeast experiences the largest price declines while the Midwest sees modest increases. In the long run, adding major transmission projects such as the Grain Belt Express increases solar and wind adoption in the Eastern Interconnection by over 10% by 2050. I use these estimates to evaluate proposed policies including the Inflation Reduction Act and the Bipartisan Infrastructure Law and find that transmission and investment subsidies are complements in encouraging transition.
Publications
Offshore Horizons: HVDC Wind Farms – Exploring Techno-Economic Dimensions
High Voltage Direct Current (HVDC) technology is a cornerstone of efficient Offshore Wind Farm (OWF) power transmission. This review examines HVDC OWF connections through four interlinked dimensions: economic considerations, connection topologies, converter designs, and technical modeling. It begins with an in-depth economic analysis, evaluating cost-effectiveness, reliability, and market dynamics, focusing on investment, operational costs, and lifecycle expenses. Building on this foundation, the review explores various collection and transmission architectures, highlighting their technical trade-offs, and evaluates power converter designs for efficiency, reliability, and offshore adaptability. Finally, advanced modeling and simulation techniques are reviewed to optimize system performance, enhance reliability, and balance computational efficiency. Together, these insights provide a holistic framework for sustainable and economically viable offshore wind energy transmission systems.
Working Papers
Non Linear Dividend Taxation and Shareholder Disagreement
In 2020, nearly 50% of U.S. dividend income faced a different marginal tax rate than the top bracket, with the tax wedge from differential dividend taxation fluctuating significantly since 1983. This has unintentionally fueled shareholder disagreement over firm investment levels. Using ISS data, we show that such disagreement is more pronounced for investment and financial policies than for issues like director elections. To explore these dynamics, we develop a heterogeneous agent, general equilibrium model where differential dividend taxation generates shareholder conflict over firm decisions. We illustrate how voting can resolve this while preserving price-taking behavior and competitive, forward-looking investment. Changes in the median shareholder – driven by both shocks and capital dynamics – amplify capital fluctuations, increasing investment and stock price volatility.
Industrial Policies in a Market with Environmental Externalities: Evidence from the Chinese Solar Panel Industry
Monocrystalline silicon (c-Si) technology has dominated in the Solar Photovoltaics (PV) manufacturing and the production costs have declined significantly over the past decade. This study investigates the role of learning-by-doing (LBD) in driving this cost reduction and forming the technological pathway of the industry, and its interaction with input price and government subsidy policies.
AI Efficiency and United States Power Quality and Emissions
Data centers now consume 4.4% of U.S. electricity – more than double their 1.9% share six years ago – yet AI’s effects on grid stability remain poorly quantified. We exploit staggered Large Language Model releases as a natural experiment, using difference-in-differences to estimate how model training and inference affect local power grids. AI activity significantly degrades power quality and increases fossil fuel generation near data centers, with high-intensity periods adding 0.5-1 outages per year. We find an increase in fossil fuel power demand from AI equivalent to approximately 47,000 household-years during inference. We construct counterfactuals examining alternative AI development trajectories. A 1% increase in model parameters raises local power demand by 0.15% post-release. Fully implementing on-site generation would improve power quality by 0.17 reliability events per month – potentially making AI a net positive for grid reliability.
Integrating Biophysical and Economic Models to Assess U.S. Agricultural Resilience
Agricultural supply chains are vulnerable to disruptions across economic and biophysical systems. We develop an AI-powered tool that integrates economic models (GTAP, VAR) with biophysical models (APSIM) to analyze supply chain shocks, enabling policymakers and market participants to assess cross-disciplinary impacts through natural language queries.
Econ Gym: A Generalized Framework for Solving Dynamic Structural Models in Economics
This paper introduces Econ Gym, a general-purpose computational framework designed for articulating, solving, and sharing problems in dynamic structural economics. Inspired by the modular architecture of OpenAI Gymnasium, Econ Gym wraps dynamic games and heterogeneous-agent models in a standardized Gym-compatible environment, enabling researchers to define state spaces, transition rules, reward functions, and constraints in a rigorous and extensible manner. The framework integrates three essential components: (1) a dynamic environment API compatible with reinforcement learning paradigms, (2) an equilibrium solver for computing rational expectations equilibria, and (3) a calibration and estimation backend supporting simulated method of moments, GMM, and maximum likelihood estimation. Experimental results show that GPU-enabled JAX implementations achieve significant speedups over traditional NumPy-based solvers, deep learning approaches scale substantially better than value function iteration as state dimensionality grows, and transfer learning reduces training time for new model specifications.
Commodity Markets LLM-based Futures Forecasting
Commodity markets are inherently more physically-driven and information based than equities markets. New transformer-based architectures have revolutionized the practice of natural language processing. However, many commodities predictions trained on news sources suffer from a lack of validity due to an inability to separate fact from opinion. Our paper solves this problem by utilizing government reports commonly published in commodity markets to train a transformer-based model on a form of text ground truth and then fine-tuning the model on news reports.
Taxation with Debt Default
Works in Progress
A Structural Model of Data Center Procurement of Electricity with Impacts on Reliability, Emissions, and Prices
Agentic AI Modeling for Rapid Analysis of Chokepoint Crises Along Strategic and Economic Dimensions: A Case Study of the Closure of the Strait of Hormuz
The February 2026 closure of the Strait of Hormuz – through which approximately 20 million barrels of daily oil flow and 22% of global liquefied natural gas trade transit – has produced a systemic supply shock of a scale not seen in decades. We propose an agentic AI framework that orchestrates heterogeneous domain models – spanning water resources, oil, LNG, helium, fertilizer, agriculture, shipping, and macroeconomic general equilibrium – through a structured scenario analysis pipeline.
Agricultural Supply Chain Resilience: Using AI to Integrate Economic and Biophysical Models
The International Determinants of Shareholder Disagreement
Robust PCA and Limit Uniqueness in High-dimensional Global Games with Strategic Complementarities
I prove that supermodular games whose payoff matrices have rank one admit no better response cycles, and hence satisfy the generalized ordinal potential property. Combined with characterizations of limit uniqueness, this implies that limit uniqueness holds for rank-one supermodular global games. I also establish, via a transversality argument, that for sufficiently low rank r, generic payoff matrices on the rank-r manifold avoid better response cycles. Finally, we show that when players compete simultaneously in many independent markets governed by the same latent payoff structure, Robust PCA on the stacked observations delivers the “intermediate regime” of vanishing but nonzero residual noise required for the contagion mechanism of global games.