The press release has a big number in it. On September 14, EPA announced it had repealed most of the greenhouse gas requirements the Biden administration set for power plants, an action the agency "projects will save $310 billion."
So I went and read how they got it. The answer is in the rule itself, on page 172, in one sentence: "Note that SAGE does not account for the effects of changing environmental quality as a result of this action."
SAGE is the model that produced the $310 billion. And in a rule about what comes out of power plant smokestacks, changing environmental quality means, first of all, the air you breathe. So we are going to ask the question this show always asks about a saving this size. Who pays for it?
The Receipt
On September 14, 2026, EPA Administrator Lee Zeldin signed a final rule titled "Partial Repeal of the Carbon Pollution Standards for Fossil Fuel-Fired Electric Generating Units." It throws out most of what the 2024 Carbon Pollution Standards required: the guidelines covering existing steam power plants that burn fossil fuels, the carbon capture standards for coal plants undertaking a large modification, and the carbon capture standards for new base load combustion turbines. The efficiency standards the 2024 rule set for new turbines survive, for now.
What EPA posted is the prepublication version. It isn't in the Federal Register yet, and the rule says it takes effect 60 days after it is.
Now the money. The preamble reports three savings figures for 2026 through 2047, in 2024 dollars, discounted to 2025. Compliance cost savings for the power sector come to $160 billion at a 3 percent discount rate and $95 billion at 7 percent. Real resource cost savings come to $280 billion and $180 billion. And the economy-wide social cost savings, estimated in a model called SAGE, come to approximately $310 billion.
The $310 billion is the number in the release, and it is the biggest of the three. It is also the one that comes with the sentence I quoted up top.
Why It Costs You
EPA's own analysis says what the repeal does to the air. It just doesn't put a price on it.
The Regulatory Impact Analysis, the technical document behind the rule, projects that power plants will put out 406 million more metric tons of carbon dioxide in 2035 than they would have with the 2024 standards in place, and 533 million more in 2040. It projects 375 thousand more tons of sulfur dioxide and 191 thousand more tons of nitrogen oxides in 2035. The RIA itself describes sulfur dioxide and nitrogen oxides as precursors to fine particle pollution, and nitrogen oxides as a precursor to ground-level ozone.
EPA ran the air quality models, too. For 2035 through 2045, the RIA says, the modeling shows "broad areas" of increased ozone and fine particle pollution across the eastern U.S. under the repeal, compared with keeping the 2024 standards. To be fair to the model, it also projects that compared with 2022, ozone and fine particle levels fall in most locations either way. The repeal means some areas may see them fall more slowly.
Then the analysis stops. The RIA lists the health effects EPA has historically counted for fine particles, including heart attacks, asthma symptoms, emergency room visits, stroke, lung cancer and lost workdays, and the ozone section carries its own list. After each list comes the same sentence: "The EPA did not quantify or monetize the benefits or disbenefits associated with changes in the incidence of the listed health effects for this rule."
The children's health section of the preamble is plainer still. It says the 2024 standards were expected to cut pollution and "some of the benefits of reducing these pollutants would have accrued to children," and that this repeal "is expected to decrease the impact of the emissions reductions" the 2024 standards were projected to deliver on those benefits. Then it says "the EPA does not quantify the health effects of air pollution in this final rule."
So when the release tells you the repeal saves $310 billion, here is what that number leaves out: the asthma, the emergency room visits, the kids. None of it has a dollar sign in this rule, which means none of it was subtracted from the saving.
How the Machine Works
Here is how a saving gets that big without the air in it.
The compliance number is what the power sector would have spent. The real resource number takes out transfers like tax credits, because the preamble says the credits would have paid for resources that are still a cost to society. SAGE takes it one step further and follows those avoided costs through the rest of the economy, into the industries that use electricity and the households that pay for it.
What SAGE does not do, by EPA's own description in the RIA, is "estimate changes in emissions." A model that can't see the smoke can only count what you save by not cleaning it up.
The second gear is a decision. The preamble says EPA's past analyses may not have shown the full range of uncertainty around the health benefits of cutting fine particles and ozone, and that the science is still evolving, especially at low concentrations. So EPA is "refraining from providing primary estimates" of those health effects, and says it will keep quantifying the emissions until it is "confident enough in the modeling to robustly monetize those impacts."
One model that doesn't count the air, and one choice not to count it any other way. That is the machine.
The third gear is carbon. Citing an executive order, the RIA says EPA did not monetize the effects of the added carbon dioxide either.
And this repeal is only the first step. In a separate supplemental proposal signed the same day, EPA proposes that Section 111 of the Clean Air Act "does not authorize the EPA to regulate emissions from power plants in response to global climate change concerns." The proposal says that reasoning would require repealing all of the 2015 and 2024 greenhouse gas standards for these plants.
The Strongest Version of the Other Side
EPA's legal case deserves to be read in its own words, not the release's, so here it is from the preamble.
Start with the statute. As the preamble describes section 111(a)(1), EPA picks the best system of emission reduction that, taking into account cost, energy requirements, and health and environmental impacts beyond air quality, "has been adequately demonstrated." EPA now concludes that 90 percent carbon capture fails that test for existing long-term coal plants. It says earlier projects that failed to reach 90 percent capture are not a sufficient basis to call the technology demonstrated, that the capture, pipeline and storage infrastructure doesn't currently exist, and that it is significantly unlikely to be in place in time for the January 1, 2032 compliance date. It reaches the same conclusion about carbon capture for new base load combustion turbines.
Then the Supreme Court. The preamble describes West Virginia v. EPA as a ruling that Section 111 does not authorize a regulatory scheme that "cap[s] carbon dioxide emissions at a level that will force a nationwide transition away from the use of coal to generate electricity." EPA says the 2024 standard for medium-term coal plants, burning 40 percent natural gas alongside the coal, is a kind of generation shifting that ruling doesn't allow, and that the 2024 rule read the case too narrowly. The release puts it more bluntly: the 2024 standards went past EPA's authority by demanding controls that weren't adequately demonstrated, "effectively forcing plants to retire," and it cites West Virginia v. EPA by name.
On the air, EPA doesn't deny the tradeoff. The preamble acknowledges that the 2024 standards would also have cut particulate matter, sulfur dioxide and nitrogen oxides, says those cuts did not factor into its decisions about carbon dioxide, and adds that even if they had, "the costs and energy impacts of the controls should be weighted more heavily than those forgone reductions."
On the model, the RIA says EPA's Science Advisory Board stated that a general equilibrium model's blind spot on benefits "does not invalidate the use of CGE models to estimate costs." On the health estimates, the RIA says a 2024 Science Advisory Board review repeatedly recommended that EPA get better at showing how uncertain its fine particle health estimates are.
And on your power bill, the RIA projects average retail electricity prices 5.8 percent lower in 2035 and 2.6 percent lower in 2045 than with the 2024 standards. The preamble adds that its cost savings don't account for benefits such as electric grid reliability that cannot be monetized.
That is a serious case. Here is the answer, and it comes from the same documents.
Almost every one of those arguments is about whether the 2024 standards were lawful and achievable. None of them changes what the $310 billion is. The same RIA paragraph that quotes the Science Advisory Board says estimates from SAGE are "a partial representation of the total effects of regulation on the economy." Whether the law allowed the 2024 rule is a question for the courts. What the saving counts is a question for you, and EPA answered it on page 172.
Uncertainty about the size of a health effect is a reason to show a range. The RIA describes EPA doing exactly that in past analyses, with sensitivity analyses and alternate concentration-response assumptions. This time the saving got a number and the health effects got none.
As for the bill, the same RIA table that shows prices falling in 2035 shows them 0.7 percent higher in 2030. A lower bill in 2035 is a real thing. So is the air in 2035. You were only handed one of them with a price on it.
What You Can Actually Do
The final rule is not open for comment. The supplemental proposal is the one still in play, and it is the one that would repeal everything left.
Comment on the proposal. The proposal takes comments in docket EPA-HQ-OAR-2025-0124, which is also the final rule's docket. It names regulations.gov as EPA's preferred method, and it also accepts email at a-and-r-docket@epa.gov with the docket ID in the subject line. Comments are due 45 days after the proposal is published in the Federal Register. It hasn't been published yet, so there is no calendar date to give you.
Aim the comment at the right question. The proposal says, "We are not reopening that final rule," and it isn't taking comment on the final rule's issues except where a commenter thinks the final rule matters to the question of statutory authority. So the comment that counts is on the proposal's question: whether Section 111 lets EPA regulate power plant emissions over climate change at all. If the uncounted health costs matter to you, you can say so there, and point EPA to page 172 of its own final rule.
Sign up for the hearing. The proposal sets a virtual public hearing 14 calendar days after it is published, with pre-registration for speakers closing 12 calendar days after publication, through EPA's rule page. EPA's news release says the hearing comes 15 days after publication. Go by the proposal, and check the rule page once it posts details.
Watch for the date. The final rule says EPA will post the Federal Register version at that same rule page once it is published. That is when the clocks start: 60 days until the final rule takes effect, and 45 days to comment on the proposal once the proposal is published.
Read page 172 yourself. The final rule is a free PDF. The SAGE sentence sits in the paragraph that begins "Changes in resources used by the power sector."
The So-What
A saving that doesn't count what it costs isn't the whole bill. It is a bill with the other column left blank.
EPA did the modeling. It mapped where the air gets worse, and it wrote down that some of what is lost would have gone to children. Then it put a price on the saving and no price on the rest. Three hundred ten billion dollars, and the air isn't in it. Bless its heart. Ask for the other column.
Come back for the rest
We do this every week. The official framing on top, the receipts underneath, every figure traced back to the document it came from so you can go open it yourself and check my work. Pull up a chair, pour something cold, and let us send you the receipts before the next quiet robbery clears the courthouse. Subscribe to The Long Pour. The free newsletter where the week's little robberies and the one big one land in your inbox together.
Y'all come back, now.
The Receipts
Every claim above traces to a document you can open yourself, with the access date on each line.
- Partial Repeal of the Carbon Pollution Standards for Fossil Fuel-Fired Electric Generating Units (EPA, final rule, prepublication version signed September 14, 2026, Docket EPA-HQ-OAR-2025-0124, RIN 2060-AW55, PDF pp. 1-2, 6, 10-14, 70, 171-173, 178-179; accessed September 16, 2026): what was repealed, the three savings figures, the SAGE sentence and EPA's legal case.
- EPA Finalizes Repeal of 2024 Power Plant Regulations, Delivering $300+ Billion in Savings (EPA news release, September 14, 2026; accessed September 16, 2026): the $310 billion projection, the release's legal summary and its hearing timing.
- Regulatory Impact Analysis for the Final Partial Repeal of the Carbon Pollution Standards (EPA-452/R-26-003, September 2026, Tables 1-3, 3-3 and 3-4, sections 1, 3.4, 4.1 to 4.2 and 5.2; accessed September 16, 2026): projected emissions, prices, air quality modeling and the unquantified health effects.
- Rescission of the Greenhouse Gas Findings for Fossil Fuel-Fired Power Plants and Repeal of Regulations for Power Plant Greenhouse Gas Emissions (EPA, supplemental proposal, prepublication version signed September 14, 2026, RIN 2060-AX00, PDF pp. 1-4; accessed September 16, 2026): the proposal's scope, docket, comment period and hearing timing.
- Greenhouse Gas Standards and Guidelines for Fossil Fuel-Fired Power Plants (EPA rule page; accessed September 16, 2026): where EPA posts the rule documents and hearing registration.