U.S. Energy Information Administration logo
Skip to sub-navigation

Analysis of Carbon Fee Runs Using the Annual Energy Outlook 2021

Release date: November 17, 2021

Executive Summary

Our analysis indicates that the carbon fee levels presented in this report would initially reduce the levels of U.S. energy-related carbon dioxide (CO2) emissions. Emissions would decrease the most in the first 5–10 years but have significantly less effect beyond that period (Figure ES-1). The three carbon fee cases we examine assume economy-wide implementation of fees that start at $15.14, $25.23, and $35.33 per metric ton of CO2 in 2023 and grow by 5% each year thereafter up through 2050. These cases are labeled as the $15, $25, and $35 carbon fee cases for simplicity.

In the highest-priced $35 Fee case, we project 23% less U.S. energy-related CO2 emissions in 2050 than in the AEO2021 Reference case, which assumes no changes in current laws and regulations and current views on the rate of technology improvements. In the $35 Fee case, we project a reduction of 19% in U.S. energy-related CO2 emissions by 2050 compared with 2020. The Reference case projects an increase of more than 5% in CO2 emissions between 2020 and 2050.

Figure ES-1. Total CO2 emissions for the Reference case and carbon fee cases
figure data

The most immediate CO2 reductions come by replacing coal-fired generation in the electric power sector with natural gas or non-emitting renewables. These reductions occur for all modeled fee cases. Other pathways to emissions reductions, such as reducing CO2 from bulk chemical production and from significantly greater penetration of alternative-fueled vehicles, require higher fees than we explored.


The majority of emissions reductions that occur in response to carbon fees come from the U.S. electric power sector, amounting to 82% of the difference in projected CO2 emissions in 2050 when comparing the $35 Fee case with the AEO2021 Reference case in 2050.

Fuel substitution is relatively inexpensive in the electric power sector in the short term as coal is displaced by natural gas. In the medium to long term, more renewables and energy storage capacity leads to further CO2 reductions in the carbon fee cases. The majority of the projected capacity additions come from solar photovoltaics.

In addition, not only is less nuclear power capacity retired in the carbon fee cases relative to the AEO2021 Reference case, but new capacity is added. More natural gas capacity is retired and less new capacity is added in the carbon fee cases compared with the Reference case, although natural gas-fired plants with carbon capture and storage (CCS) technology become cost competitive before 2050.

Carbon fees increase the contribution of renewable generation in these cases as renewables, such as wind and solar, become more cost competitive relative to fossil fuels. The carbon fee cases also increase the economic viability of energy storage facilities used to help manage the variation of intermittent renewable resources.


The effects of carbon fees on transportation-related CO2 emissions are much more limited. In our modeling, a $35 carbon fee reduces U.S. energy-related CO2 emissions from transportation by 6% in 2050 relative to the Reference case.

In the $35 Fee case, CO2 emissions from light-duty vehicles decline 11% in the United States by 2050, 7 percentage points more than the 4% decline in the Reference case. By 2050, CO2 emissions from freight trucks decline by 2% in the $35 Fee case, when they had increased by 7% by 2050 in the Reference case. Air travel emissions are relatively unaffected in these cases.

Despite 9% annual growth in electricity sales to the transportation sector during the projection period in the $35 Fee case, petroleum remains the most consumed transportation fuel.


Natural gas is the predominant fossil fuel consumed by the U.S. industrial sector. Electricity purchases play a relatively smaller role in this sector. Although emissions decline in the carbon fee cases relative to the Reference case, emissions begin increasing after 2040 as gross output continues to rise with increasing economic growth.

The response to carbon fees varies by industry. For example, energy-intensive industries without readily substitutable alternative fuels, such as the bulk chemicals industry, show a relatively small reduction in CO2 emissions in response to carbon fees.

Modeling background

The carbon fee cases discussed in this paper assume economy-wide implementation of a $15.14, $25.23, and $35.33 carbon fee (2020 dollars per metric ton of CO2) starting in 2023. These fees increase by 5% in real terms per year. Emission fee revenues are distributed back to consumers through lump-sum payments, keeping the policy revenue neutral. We made these projections using the National Energy Modeling System (NEMS), which uses a market-based approach, subject to current regulations and standards. For each fuel and consuming sector, NEMS balances energy supply and demand, accounting for economic competition across the various energy fuels and sources. The projection period in NEMS currently extends to 2050.

The current NEMS is limited in its ability to run deep decarbonization scenarios. Although we regularly undertake significant NEMS model development, we have not been able to invest sufficiently to fully model deep decarbonization or net-zero emissions scenarios, which would require the model to better represent fuels and technologies such as:

  • Biofuels
  • Carbon capture, transport, and storage
  • Advanced electrification
  • Hydrogen deployment

We are working to assess our energy modeling capabilities and to develop a plan to modernize and integrate the modeling platforms and tools used to produce our flagship energy outlooks and forecasts.

Until we can make those changes, we are committed to doing the work we can to illuminate the potential effects of climate policy options.

See full report