Decarbonization and Electrification in the Long Run

Stephen P. Holland, Erin T. Mansur, Andrew YatesSSRN 2022 · DOI

TL;DR

This paper develops a long-run equilibrium model of the U.S. electricity sector to show how the power grid will transform when capacity is free to enter and exit. The authors reveal that while short-run intuition often fails—such as storage potentially decreasing renewable capacity—strategic policies like daytime EV charging and high carbon pricing ($150/ton) can achieve near-total decarbonization.

What Problem Does This Paper Address?

Most energy policy analysis focuses on the "short run"—how to best use the power plants we already have. However, as we look toward 2050, almost every power plant currently operating will likely be retired and replaced.

The authors address the "long-run" challenge: if we were to build the grid from scratch today, accounting for the intermittency of wind and solar, the potential of batteries, and the massive new load from electric vehicles (EVs), what would it look like? They specifically tackle the "hysteresis" problem—the idea that our current grid is a series of historical accidents and legacy investments—and provide a framework for the "end goal" of a decarbonized economy.

Methods

The researchers constructed an analytically tractable long-run competitive equilibrium model. Key features include:

  • Endogenous Capacity: Unlike short-run models, this model allows for the entry and exit of five generation technologies (solar, wind, nuclear, combined-cycle gas, and peaker gas) and battery storage.
  • Planner’s Problem: They solve a constrained optimization problem that maximizes social surplus across 8,760 hourly periods (a full year).
  • Intertemporal Storage: The model optimizes battery usage across hours, accounting for the state of charge and capacity limits.
  • Theoretical Proofs: They derive several "lemmas" and "results" that prove counterintuitive possibilities, such as a carbon tax actually increasing total electricity consumption if it induces enough low-cost clean technology entry.

Datasets

The model is calibrated using high-resolution data from 2019:

  • Hourly Demand & Renewable Availability: Data for 13 EIA electricity regions covering the contiguous U.S.
  • Cost Estimates: Baseline capital and marginal costs are based on the EIA’s Annual Energy Outlook 2021 for capacity entering service in 2026.
  • EV Data: County-level vehicle miles traveled (VMT) from the EPA MOVES model and temperature-adjusted electricity consumption rates.

Key Insights and Ideas

  • The Storage Paradox: We often assume more storage is always better for renewables. However, the authors show that cheaper storage can actually decrease renewable capacity. Why? Storage favors the technology with the lowest long-run average cost. If natural gas is cheaper than renewables on an average basis, storage helps gas plants run more efficiently, potentially crowding out wind and solar.
  • The Nuclear/Renewable Flip: For modest decarbonization goals, renewable subsidies are the most cost-effective. But for "ambitious" goals (near-zero emissions), nuclear subsidies become more efficient because nuclear provides reliable, dispatchable carbon-free power that renewables struggle to match without massive, expensive storage.
  • Negative Incremental Emissions: In the long run, adding new demand (like EVs) can actually reduce total grid emissions. This happens if the new load occurs during hours that induce the entry of so much renewable capacity that it offsets fossil fuel generation in other hours.

Key Results

  • Carbon Pricing: A carbon price of $150 per ton essentially eliminates carbon emissions from the electricity sector in the long run.
  • EV Charging Profiles: This is the "make or break" factor for electrification.
  • If 100% of U.S. car miles are electrified and charged at night (Convenience profile), sector emissions increase by 23-27%.
  • If charged during the day (Carbon Minimizing profile), EVs could actually decrease electricity-sector emissions by inducing a massive build-out of solar capacity.
  • Transmission: Expanding transmission yields massive benefits, but only if renewable costs fall significantly. Connecting the wind-rich Midwest to the East Coast is identified as a high-value priority.

Limitations and Future Work

  • Cost Uncertainty: The results are highly sensitive to capital cost projections for 2026 and beyond, which are inherently speculative.
  • Simplified Transmission: While the model looks at 13 regions, it assumes no transmission congestion within those regions, which may overstate the ease of integrating renewables.
  • Static Demand Elasticity: The model uses a small, constant price elasticity (-0.15). In reality, demand response might become much more dynamic as "smart grid" technologies mature.
  • Transition Path: The paper describes the "destination" (the long-run equilibrium) but does not model the "journey" (the year-by-year transition and the costs of retiring plants early). Exploring this transition is a noted area for future research.

Summary updated

BibTeX
@misc{holland2022decarbonization,
  author = {Stephen P. Holland and Erin T. Mansur and Andrew Yates},
  title = {{Decarbonization and Electrification in the Long Run}},
  year = {2022},
  doi = {10.2139/ssrn.4122811},
  howpublished = {SSRN},
}
Figure