If you are converting metered kWh into tonnes of CO₂, the factor you use matters more than the precision of your meter. And right now that factor is harder to pin down than it should be, for three reasons: EPA publishes more than one, they disagree, and the agency has stopped updating them.
The current figures
The most recent reviewed dataset is eGRID2023 Revision 2, released 12 June 2025. Its summary tables were produced 27 March 2025. Total output emission rates, CO₂e:
| eGRID subregion | lb CO₂e/MWh | kg CO₂e/MWh (derived) | vs US average |
|---|---|---|---|
| NYUP — Upstate New York | 242.8 | 110.1 | 0.31× |
| NYCW — NYC & Westchester | 865.7 | 392.7 | 1.12× |
| NYLI — Long Island | 1,189.3 | 539.5 | 1.54× |
| US average | 770.9 | 349.7 | 1.00× |
The kg/MWh column is our conversion from the published pounds, not an EPA-published value.
The spread inside one state is the first thing worth noticing. Upstate New York is roughly a third of the US average; Long Island is roughly one and a half times it. A portfolio with sites in both is working with factors that differ by nearly 5×. Using a single state-level or national number across that portfolio will produce a materially wrong inventory, in a direction that depends on where the load sits.
The conflict
Two live EPA documents give different numbers for New York City.
- eGRID2023 Revision 2 — 865.7 lb CO₂e/MWh for NYCW
- EPA GHG Emission Factors Hub, dated January 2025 — 974.7 lb CO₂/MWh for NYCW
That is roughly 11% apart, and both documents are current and downloadable from EPA.
Two things are going on. The Emission Factors Hub edition dated January 2025 was built on the original eGRID2023, before Revision 2 restated it. And the two figures are not even the same metric: one is CO₂e, covering CO₂, CH₄ and N₂O; the other is CO₂ only. They are not interchangeable and the gap should not be split.
Our position: use eGRID2023 Revision 2, state the vintage, and never write "EPA's factor for New York City" without naming the document it came from. A reviewer who can see which file you used can check you. A reviewer who cannot has to take it on trust, and an 11% discrepancy is large enough to matter in a filing.
EPA has stalled, and third parties are filling the gap
There is no eGRID2024. There is no 2026 Emission Factors Hub. As of October 2026, Revision 2 of eGRID2023 remains the newest reviewed EPA dataset.
This has produced two traps:
"eGRID2024" figures are circulating. EPA's eGRID methodology is open source, so third parties have run it forward on newer generation data and published the results under the eGRID name. Those numbers may well be reasonable. They have not been through EPA review, and the people publishing them generally say so if you read far enough. They are not an EPA dataset and should not be cited as one.
A "2026 GHG Emissions Factor Hub" exists and is not EPA's. A document under that title was published on 11 June 2026 by Cornerstone — a private organisation. It is a commercial product, not an agency publication. The name is close enough to EPA's that it is easy to cite by mistake.
The practical implication for anyone writing a sustainability report, an ESG disclosure or a carbon inventory: any page or spreadsheet that says "the latest EPA factors" without a vintage label is already a liability. Label the file, label the revision, and put a recheck date on it.
Natural gas, and a derived number to label
For stationary combustion of natural gas, the EPA Emission Factors Hub dated January 2025 gives:
> 53.06 kg CO₂ per mmBtu
A detail that catches people out: the Hub publishes no per-therm value. The commonly quoted 5.306 kg CO₂ per therm is arithmetic — one therm is 0.1 mmBtu — and should be labelled as derived rather than presented as an EPA figure. It is correct arithmetic. It is just not something EPA published.
A practical labelling convention
Whatever you adopt internally, the convention that survives review looks like this:
- Name the dataset and its revision: eGRID2023 Revision 2, not eGRID
- Name the release date: released 12 June 2025
- Name the subregion, not the state: NYCW, not New York
- Say whether the figure is CO₂ or CO₂e
- Mark derived conversions as derived — kg from lb, per-therm from per-mmBtu
- Set a recheck date, because EPA will eventually publish again and the factor will move
That is six lines in a footnote and it is the difference between an inventory a reviewer can verify and one they have to question.
Why a metering company cares about this
Measurement and conversion are different problems, and the second one is where most carbon numbers go wrong. A building can have excellent interval data on every circuit and still report a materially incorrect emissions figure because it applied a national average factor, or an outdated one, or a CO₂ factor where CO₂e was required.
Better metering raises the ceiling on how good a carbon number can be. It does nothing about the factor. Both have to be right.
Emergent Metering supplies the measurement hardware — electric submeters, current transformers, gas and water meters, concentrators and gateways. The conversion conventions above are published because they are genuinely difficult to get right and the sourcing is not obvious, not because hardware solves them.
Frequently asked questions
What is the current EPA emission factor for New York City electricity? eGRID2023 Revision 2, released 12 June 2025, gives 865.7 lb CO₂e/MWh for the NYCW subregion, which covers New York City and Westchester. Note that the EPA Emission Factors Hub dated January 2025 gives 974.7 lb CO₂/MWh for the same subregion — a different vintage and a different metric.
Is there an eGRID2024? No. EPA has not published eGRID2024. Figures circulating under that name are third-party runs of EPA's open-source methodology and have not been through agency review.
Why do two EPA documents disagree? The January 2025 Emission Factors Hub was built on the original eGRID2023, before Revision 2 restated it. The two figures also differ in scope — one is CO₂e, the other CO₂ only.
What is the CO₂ factor for natural gas? 53.06 kg CO₂ per mmBtu, per the EPA Emission Factors Hub dated January 2025. The frequently quoted 5.306 kg per therm is derived from that figure rather than published by EPA.
Can I use a single factor for a multi-state portfolio? Not without introducing error. eGRID subregion factors vary by nearly 5× within New York State alone. Portfolio inventories should apply the subregion factor for each site's grid location.