Retirement Decisions for Natural Gas and Coal-Fired Plants in ERCOT

This paper estimates a dynamic discrete choice model of retirement for natural gas and coal-fired generators in ERCOT, treating exit as an irreversible optimal stopping problem under forward-looking expectations. All 45 observed exits in the estimation panel occur at strictly positive operating profit, consistent with retirement as a forward-looking continuation-value decision rather than a decision triggered by low current-period margins. Operating profit is constructed at unit resolution from 15-minute spark and dark spreads combining real-time nodal settlement prices and fuel-cost indices, with ancillary service revenue recovered from the ERCOT 60-Day SCED record and pro-rated fixed operating costs subtracted. Operating profit enters the flow payoff positively and significantly, and generator age enters negatively, separating cohorts of different vintage. Wind and solar shape retirement primarily through the operating-profit transition kernel rather than the current-period flow payoff. Their persistence in the kernel transmits into continuation values, which dominate the effect of renewable trajectories on exit decisions. A forward projection built from the April 2026 ERCOT Generator Interconnection Status report produces 18.6 GW of thermal retirement by 2035 against a 73.4 GW starting fleet, dominated by age-driven exit rather than renewable-driven margin pressure. Doubling the zonal wind share adds 0.5 GW of retirement by 2035, and doubling the zonal solar share adds 1.9 GW. The structural estimates provide a reference point for policymakers and market planners assessing how the dispatchable thermal fleet is likely to evolve under continued renewable buildout. 

Dynamics of Renewables in Electricity Markets (with a view towards ERCOT)

This paper extends the analytical framework of Green and Léautier (2015) in two directions. First, we derive a complete set of marginal impact results under a Feed-in Premium, an extension identified as an open problem in that work. Because renewables curtail output in the worst-priced states under a premium, their per-MW market value falls more slowly with penetration than under a fixed-price Feed-in Tariff. Second, we adapt the framework to ERCOT, which uses an Operating Reserve Demand Curve (ORDC) to set scarcity prices in lieu of a forward capacity market. The principal theoretical result is that the ORDC does not directly affect the rate at which renewable per-MW values fall. Entry of conventional capacity competes the common scarcity increment away. A calibration to ERCOT data quantifies the cross-regime comparisons. Nuclear capacity exits at approximately 80 GW of wind under both the Feed-in Tariff and the ERCOT design, while the premium retains nuclear across the full simulated range. The combined premium-ERCOT case is the policy-relevant one, since the federal Production Tax Credit shares the premium-on-top-of-price structure. 

Generator Investment and Entry in ERCOT 

This paper estimates a discrete choice model of generator investment across ERCOT's four load zones over 2011–2024. Developers choose among five technologies (wind, solar, combined-cycle gas, open-cycle gas, and battery storage), grouped into dispatchable and non-dispatchable segments. A discrete choice model of generator entry that includes battery storage alongside thermal and renewable technologies is new to this literature. Operating profit is endogenous to the equilibrium electricity price and is instrumented with national fuel-cost shifters, rival characteristics, and a Hausman-type instrument. The estimation yields three results. First, entry composition responds strongly to annualized fixed costs. Second, federal subsidies shift composition toward the subsidized technologies. Third, we find no detectable entry response to measured energy margins at annual frequency, consistent with the multi-year lag between the commitment decision and commercial operation. Gas-fired entry falls below what the measured payoffs predict. Entry composition over this period is explained by fixed costs and subsidies.