What will be the future of U.S. electricity production?

In “Decarbonization and electrification in the long run” (National Bureau of Economic Research, Working Paper 30082, May 2022), Professors Stephen P. Holland, Erin T. Mansur, and Andrew J. Yates create a mathematical model to predict how the United States will produce its electricity in the future and the amount of carbon emissions that production will require. To determine the longrun balance of electricity supply and demand, the authors estimate the marginal and capital costs for five electricity-production technologies, namely, solar, wind, nuclear, combined cycle gas, and combustion turbine gas. Alongside these marginal costs, the authors estimate electricity storage costs and the capacity limits of wind and solar technologies. Combined with hourly data on the demand for electricity, the production side inputs can be used to determine the socially optimal level of electricity production and the mix of electricity-producing technologies.

The authors’ model predicts the U.S. energy sector to produce 30 percent less carbon emissions in 2019 than it did in reality. The reasons for this discrepancy are that the model excludes coal-based energy production and that the model predicts a higher electricity cost than the real cost of electricity in 2019. The break from reality does not imply that the authors model is wrong but that the longrun socially optimal mix of electricity-producing technologies given the current costs and constraints is different from the U.S. current mix.

Holland, Mansur, and Yates then model how taxes and reduction in capital costs may change the amount of electricity produced and the mix of technologies used. The authors consider a carbon tax, as well as reduced capital costs for solar and wind power (individually as well as together), nuclear power generation, and battery storage and electricity transmission technologies.

As the size of the tax on carbon increases in the authors model, the share of solar and wind production increases up to a point. After that point, the share of solar and wind decreases. This nonlinearity is because at some point the tax is enough to offset the high capital costs of nuclear reactors. The model also predicts that a carbon tax may not increase electricity prices in the long run since it would induce more production from low-cost technologies.

The authors find that reducing capital costs for each technology (either through a subsidy or a technical breakthrough) increases that technology’s share of electricity production and electricity production itself. In the case of solar, wind, and nuclear production, this decrease in costs also means lower overall carbon emissions. However, when the capital costs of batteries (which are commonly believed to synergize with intermittent technologies such as wind and solar) are reduced, there is neither a major increase in the share of intermittent technologies nor a major decrease in carbon emissions. The authors have a similar finding with electricity transmission, such that even if they model ideal transmission (in essence, the same price for everywhere in the United States except Alaska and Hawaii), wind and solar sources do not generate substantially more electricity.