Explore residual mixes

What’s in them, and who gets to take credit. Every clean megawatt-hour mints a certificate; whoever retires the certificate owns the claim. This page shows what the certificate registries actually left unclaimed in 2023 — fuel by fuel — and how far those pools sit from the residual mixes eLCI publishes.

The machinery

Certificate — Renewable Energy Certificate (REC)
One megawatt-hour of generation, as a tradable claim on its attributes — fuel, location, vintage, emissions. The energy and the claim are sold separately: the electrons go into the pool, the certificate goes to whoever pays for it.
Registry (tracking system)
The ledger that mints, transfers and retires certificates so each one can be claimed exactly once. There is no national registry — the US is a patchwork of regional systems, one per market footprint, each with its own platform and its own idea of what to publish.
Retirement (settlement)
Using a certificate. A retired certificate is consumed by its holder’s claim and leaves the pool permanently. Settlement is the registry’s season of retiring, reserving and exporting — in New England it runs after June 30 of the following year, and whatever survives it becomes the residual.
Renewable Portfolio Standard (RPS) compliance
State law requiring utilities to retire certificates for a share of their sales. This is the largest consumer of certificates in most regions — claims made on everyone’s behalf by statute, not by choice.
Voluntary claims
Certificates retired by choice — corporate renewable procurement, green tariffs, certified retail products. Smaller than compliance in most regions, but concentrated in the fuels buyers want: cheap plains wind above all.
Claim rate
The share of a pool’s certificates that got retired, reserved or exported. New England 2023: 39.5%. New York: 23.2%. PJM: 17.6%. This single number is what the size of the residual gap tracks.

The mixes

System mix
Everything generated in (or, in New York’s accounting, delivered to) the region, by fuel — the pool before anyone claims anything. Certificates count megawatt-hours, so a registry’s system mix should and does track physical generation closely.
Residual mix — the registry construction
The system mix minus every settled certificate: compliance, voluntary, reserved and exported alike. It is what remains available to someone who holds no certificates — and in claim-heavy regions it is a very different animal from the system mix, because the clean attributes leave first.
Residual mix — as eLCI publishes it
In the measured regions, eLCI’s residual sits 0.8–2% above its ordinary mix — consistent with subtracting only a small voluntary-market slice, not all settlement. The release does not state which subtraction it performs; that question is on this project’s upstream list. The two constructions must not be used interchangeably.
Location-based accounting
Prices your electricity at the physical mix of your grid — generation or consumption framing — and ignores certificates entirely. Everything else on this site is location-based.
Market-based accounting
Prices your electricity at the certificates you hold; whatever load is not covered by certificates defaults to the residual mix. Under this lens the residual is not a curiosity — it is the number most consumers are actually on, because most consumers hold nothing.
Import attributes
Registries assign unclaimed imported power the source region’s residual attributes, where physical flow tracing assigns the source region’s average. That single choice puts 2.13% coal in New York’s registry mix — a state with no operating coal plant — and is the main reason registry and flow-traced numbers disagree even before any claims are subtracted.

The whole picture, in two panels

Each blue dot is a balancing authority: eLCI’s consumption mix on the horizontal axis, its residual on the vertical. If the residual were just the mix, every dot would sit on the dashed y = x line — and most sit barely above it. The orange marks are what the certificate registries say the residual actually is.

eLCI residual vs consumption, 59 balancing authorities

y = x2002004004006006008008001000100012001200YAD: consumption 4.5, residual 4.5 (1.000×)AZPS: consumption 570.6, residual 598.2 (1.048×)AECI: consumption 603, residual 673.3 (1.117×)AVA: consumption 242.9, residual 251.2 (1.034×)BANC: consumption 229.5, residual 240.9 (1.049×)BPAT: consumption 160.3, residual 166 (1.036×)CISO: consumption 244.2, residual 254.3 (1.041×)HST: consumption 348.1, residual 354.6 (1.019×)TPWR: consumption 73.7, residual 76.3 (1.035×)TAL: consumption 466.7, residual 477.1 (1.022×)SCEG: consumption 419.9, residual 427 (1.017×)DUK: consumption 312.3, residual 317.5 (1.017×)FPC: consumption 510, residual 520.3 (1.020×)CPLE: consumption 264.4, residual 269.3 (1.019×)CPLW: consumption 78.9, residual 79.6 (1.008×)EPE: consumption 469.1, residual 523.2 (1.115×)ERCO: consumption 384.2, residual 482.5 (1.256×)FMPP: consumption 552.1, residual 564.2 (1.022×)FPL: consumption 348.1, residual 354.6 (1.019×)GVL: consumption 751.6, residual 745.6 (0.992×)GRID: consumption 483.8, residual 483.8 (1.000×)GRIF: consumption 494.9, residual 494.9 (1.000×)ISNE: consumption 290.7, residual 292.9 (1.008×)IPCO: consumption 314.7, residual 375.3 (1.192×)IID: consumption 213.5, residual 218.1 (1.022×)JEA: consumption 584.6, residual 589.6 (1.008×)LDWP: consumption 480.2, residual 514.6 (1.072×)LGEE: consumption 940.6, residual 951 (1.011×)MISO: consumption 512.7, residual 530.7 (1.035×)GWA: consumption 21.2, residual 21.2 (1.000×)WWA: consumption 13.8, residual 13.8 (1.000×)NEVP: consumption 411.9, residual 434.2 (1.054×)NYIS: consumption 228.1, residual 229.9 (1.008×)NWMT: consumption 718, residual 753.9 (1.050×)PJM: consumption 385.3, residual 393.5 (1.021×)DOPD: consumption 0.9, residual 0.9 (1.000×)PACE: consumption 722.2, residual 908 (1.257×)PACW: consumption 120.6, residual 128.8 (1.068×)PGE: consumption 289.2, residual 304.1 (1.052×)PSCO: consumption 416.2, residual 425.7 (1.023×)PNM: consumption 257, residual 276.1 (1.074×)CHPD: consumption 1.9, residual 1.9 (1.000×)GCPD: consumption 56.4, residual 58.3 (1.033×)PSEI: consumption 311.8, residual 322.9 (1.036×)SRP: consumption 307.1, residual 314.6 (1.024×)SCL: consumption 73.5, residual 76.1 (1.035×)SEC: consumption 655.7, residual 658.9 (1.005×)SC: consumption 748.9, residual 757.8 (1.012×)SEPA: consumption 117.6, residual 119.3 (1.014×)SOCO: consumption 456.3, residual 465.6 (1.020×)SWPP: consumption 438.7, residual 567.4 (1.294×)SPA: consumption 278.9, residual 315.1 (1.130×)TEC: consumption 460.7, residual 470.9 (1.022×)TVA: consumption 332.4, residual 343.3 (1.033×)TEPC: consumption 666.5, residual 694.3 (1.042×)TIDC: consumption 379.9, residual 398.1 (1.048×)WALC: consumption 424.2, residual 447 (1.054×)WACM: consumption 853.7, residual 1012.7 (1.186×)WAUW: consumption 141.8, residual 178.7 (1.260×)GVLPACESWPPISNE: registry residual 585.9–619.3ISNENYIS: registry residual 403.5NYISPJM: registry residual 467.5–468.7PJMMISO: registry residual 588.2MISOERCO: registry residual 466.3ERCOCISO: registry residual 197CISOeLCI consumption mix, kg CO₂e/MWhresidual, kg CO₂e/MWh

Blue: eLCI’s own residual (grid basis) against its consumption mix — most sit just above the y = x line. Orange: the registry-grounded residual at the same consumption-mix position (bands span the two import treatments; the hollow square is the indicative California analog). Hover a point for its values.

Where the registry evidence sits, geographically

YAD: eLCI residual ÷ consumption 1.000×AZPS: eLCI residual ÷ consumption 1.048×AECI: eLCI residual ÷ consumption 1.117×AVA: eLCI residual ÷ consumption 1.034×BANC: eLCI residual ÷ consumption 1.049×BPAT: eLCI residual ÷ consumption 1.036×CISO: eLCI residual ÷ consumption 1.041×HST: eLCI residual ÷ consumption 1.019×TPWR: eLCI residual ÷ consumption 1.035×TAL: eLCI residual ÷ consumption 1.022×SCEG: eLCI residual ÷ consumption 1.017×DUK: eLCI residual ÷ consumption 1.017×FPC: eLCI residual ÷ consumption 1.020×CPLE: eLCI residual ÷ consumption 1.019×CPLW: eLCI residual ÷ consumption 1.008×EPE: eLCI residual ÷ consumption 1.115×ERCO: eLCI residual ÷ consumption 1.256×FMPP: eLCI residual ÷ consumption 1.022×FPL: eLCI residual ÷ consumption 1.019×GVL: eLCI residual ÷ consumption 0.992×GRID: eLCI residual ÷ consumption 1.000×GRIF: eLCI residual ÷ consumption 1.000×ISNE: eLCI residual ÷ consumption 1.008×IPCO: eLCI residual ÷ consumption 1.192×IID: eLCI residual ÷ consumption 1.022×JEA: eLCI residual ÷ consumption 1.008×LDWP: eLCI residual ÷ consumption 1.072×LGEE: eLCI residual ÷ consumption 1.011×MISO: eLCI residual ÷ consumption 1.035×GWA: eLCI residual ÷ consumption 1.000×WWA: eLCI residual ÷ consumption 1.000×NEVP: eLCI residual ÷ consumption 1.054×NYIS: eLCI residual ÷ consumption 1.008×NWMT: eLCI residual ÷ consumption 1.050×PJM: eLCI residual ÷ consumption 1.021×DOPD: eLCI residual ÷ consumption 1.000×PACE: eLCI residual ÷ consumption 1.257×PACW: eLCI residual ÷ consumption 1.068×PGE: eLCI residual ÷ consumption 1.052×PSCO: eLCI residual ÷ consumption 1.023×PNM: eLCI residual ÷ consumption 1.074×CHPD: eLCI residual ÷ consumption 1.000×GCPD: eLCI residual ÷ consumption 1.033×PSEI: eLCI residual ÷ consumption 1.036×SRP: eLCI residual ÷ consumption 1.024×SCL: eLCI residual ÷ consumption 1.035×SEC: eLCI residual ÷ consumption 1.005×SC: eLCI residual ÷ consumption 1.012×SEPA: eLCI residual ÷ consumption 1.014×SOCO: eLCI residual ÷ consumption 1.020×SWPP: eLCI residual ÷ consumption 1.294×SPA: eLCI residual ÷ consumption 1.130×TEC: eLCI residual ÷ consumption 1.022×TVA: eLCI residual ÷ consumption 1.033×TEPC: eLCI residual ÷ consumption 1.042×TIDC: eLCI residual ÷ consumption 1.048×WALC: eLCI residual ÷ consumption 1.054×WACM: eLCI residual ÷ consumption 1.186×WAUW: eLCI residual ÷ consumption 1.260×SWPP 1.29×PACE 1.26×ISNE 2.02–2.13×eLCI 1.01×NYIS 1.77×eLCI 1.01×PJM 1.22×eLCI 1.02×MISO 1.15×eLCI 1.04×ERCO 1.21×eLCI 1.26×CISO 0.81×eLCI 1.04×eLCI residual ÷ consumption: 0.99×1.3×

Dot shade: eLCI’s residual ÷ consumption ratio, which concentrates its uplift in the wind belt (SWPP 1.29×, PACE 1.26×). Rings: the registry-verified regions, labelled with the registry residual ÷ eLCI consumption ratio (dashed ring = indicative California analog). The registry evidence puts the real uplift in the claim-heavy Northeast, not the wind belt.

Why can a residual sit below the mix it came from?

Subtracting a claim moves the residual away from the average, in the direction opposite to what was claimed. Wind, solar, hydro and nuclear carry life-cycle factors far below any pool’s average, so claiming them pushes the residual up — that is the normal case. But the arithmetic runs the other way when the claimed generation is dirtier than the pool average in life-cycle terms. The one blue dot below the line is GVL (Gainesville): a third of its generation is wood biomass, whose eLCI factor there (1,182 kg CO₂e/MWh) exceeds the pool average (751.6) — biomass burns carbon now and books its reabsorption over decades, so life-cycle accounting prices it like a fossil fuel with a discount, not like wind. Claim that biomass and the leftover pool is cleaner per megawatt-hour, 0.99× the consumption mix. The same arithmetic explains the California analog sitting below eLCI: the CRS/Green-e construction subtracts all specified sales — including specified natural-gas contracts — and removing above-average gas lowers what remains. A residual below the average is not an error; it is a statement about what got claimed.

What’s in the pools — 2023, fuel by fuel

For the three regions whose registries publish by-fuel data: system mix against residual mix, share of each pool. Watch what settlement removes: in New England the registry’s nuclear goes from a fifth of the pool to four certificates out of 66 million, and gas swells to 73%. In PJM, nuclear attributes go largely unclaimed and stay put. Who buys what differs by region — and that difference, not geography, is what shapes each residual.

New England — NEPOOL GIS

Residual from the registry; system side from eGRID 2023 generation, because NEPOOL publishes its system mix only as an aggregate.

Natural gas
56.1%
73.0%
Nuclear
22.7%
0.00%
Imports
0.00%
15.6%
Oil
0.38%
10.2%
Hydro
8.00%
0.15%
Biomass
4.12%
0.03%
Solar
3.62%
0.58%
Wind
3.27%
0.02%
Other
1.55%
0.19%
Coal
0.20%
0.30%

New York — NYGATS

Both mixes as published. NYGATS folds imports in at certificate attributes, which is why coal appears in a state with no operating coal plants.

Natural gas
49.8%
63.0%
Nuclear
21.7%
27.5%
Hydro
18.0%
2.45%
Solar
3.38%
0.06%
Other
2.26%
2.83%
Coal
2.13%
2.70%
Wind
1.76%
0.34%
Oil
0.78%
0.99%
Biomass
0.13%
0.10%

PJM — GATS

Both mixes as published by the registry, generation months January–December 2023.

Natural gas
44.2%
53.6%
Nuclear
33.6%
26.6%
Coal
14.9%
17.4%
Wind
3.55%
0.00%
Imports
0.00%
2.01%
Solar
1.38%
0.00%
Hydro
0.96%
0.00%
Other
0.76%
0.04%
Oil
0.27%
0.33%
Biomass
0.32%
0.00%

Worked example: New England, 2023

From the registry’s published pool to a residual emission rate, every step shown.

  1. Start with the pool. NEPOOL GIS issued 110,003,745 certificates for the 2023 trading quarters — one per megawatt-hour generated in or imported into New England.
  2. Let settlement run. 39.47% of those certificates were retired for Renewable Portfolio Standard compliance, claimed voluntarily, reserved or exported. They are gone — their attributes belong to their holders. What remains is 66,499,000 certificates.
  3. Look at what’s left. The registry publishes the unsettled pool by fuel: natural gas 72.96%, imports 15.58%, oil 10.18%, solar 0.58%, coal 0.30%, everything else under 0.2% — and nuclear at 4 certificates of 66.5 million. That is not 4%: it is four individual certificates, four megawatt-hours. New England’s nuclear plants minted roughly 23 million certificates for 2023, and every one of them except four was retired, reserved or exported by whoever held it — so nuclear, a fifth of the system mix, is 0.00% of the residual. The same happened to hydro, wind and solar. New England’s clean generation is essentially fully claimed.
  4. Price each fuel. Using this project’s per-fuel eLCI factors (ISNE-specific where published, US fallback otherwise): gas 443.3, oil 1,883.1, coal 1,206.4, solar 65.2 kg CO₂e/MWh.
  5. Take the weighted average. The primary unknown here is the imports: 15.5792% of the pool is a single line labelled Import System Mix, with no fuel breakdown, so no factor can be applied to it fuel by fuel. Both equations below handle that slice differently and bracket it. Every other term is the same: the fuel’s share of the whole pool times its factor.

    Equation 1 — imports removed from the pool

    fuelshare of pool× factor= term
    natural gas0.729626443.3323.44
    oil0.1018391,883.1191.77
    solar0.00583965.20.38
    coal0.0030051,206.43.63
    other0.0018891,648.73.11
    hydro0.00152210.00.02
    biomass0.0003201,255.50.40
    wind0.00016838.30.01
    nuclear0.00000017.30.00
    sum of terms522.78

    522.78 ÷ (1 − 0.155792) = 522.78 ÷ 0.844208 = 619.3 kg CO₂e/MWh

    The shares in the terms are shares of the whole pool, imports included, so they sum to 0.844208, not 1. Dividing by 0.844208 rescales the sum to an average per megawatt-hour of the 84.42% that has a factor — without that division, the import slice would silently count as zero emissions. Terms are rounded for display; the total uses full precision.

    Equation 2 — imports kept, given the eLCI US consumption factor

    522.78 + (0.155792 × 405.3) = 522.78 + 63.14 = 585.9 kg CO₂e/MWh

    Same nine terms, plus a tenth for the imports at the US-average consumption factor. Every share of the pool now has a factor, the shares sum to 1, and there is nothing left to divide by.

    The true number lies between the two: imports into New England are almost certainly not US-average (Québec hydro pulls cleaner, but a certificate-aware treatment would assign unclaimed imports the source region’s residual, which pulls dirtier). The pair brackets that unknown at 585.9–619.3.

  6. Compare. eLCI publishes New England’s residual generation mix at 324.5 — just 0.8% above its ordinary generation mix (321.9). The registry’s residual is 81–91% higher, because the registry subtracts everything that was claimed and eLCI evidently does not.

The same procedure, adapted to what each registry publishes, produced every row in the table below. Where a registry publishes a system mix by fuel, applying the same factors lands within 2.4–4.0% of eLCI’s corresponding mix — so the residual gaps are differences of definition, not of arithmetic.

Region by region — registry residual vs eLCI

All 2023, all in kg CO₂e/MWh, registry pools given this project’s per-fuel factors. The gap tracks the claim rate: the more of a pool that is settled, the further the registry’s residual sits above eLCI’s. Texas is the exception that proves the rule — eLCI’s residual already carries a 1.26× uplift there, and the reconstruction lands slightly below it.

RegionRegistryRegistry residualeLCI residualGapPool claimedSystem check
New England (ISNE)NEPOOL GIS585.9–619.3324.5+81 to +91%39.5%—
New York (NYIS)NYGATS403.5229.9 (consumption frame)+75.5%23.2%+41.6%*
PJMPJM GATS467.5–468.7394.0+18.7 to +19.0%17.6%+4.0%
MISOM-RETS (claim rate)588.2552.9+6.4%~8.3%+2.8%
Texas (ERCO)ERCOT REC program466.3482.5-3.4%16.5%+2.4%
California (CISO)CRS/Green-e CAMX — indicative197.0217.6-9.5%n/a—

System check: the registry’s (or eGRID’s) full system mix given the same factors, against eLCI’s corresponding mix — it bounds factor-choice and fuel-mapping error. *New York’s check measures the import-attribution difference described under Definitions, not mapping error. MISO applies the M-RETS registry-wide claim rate because retirements are not published by state — the weakest reconstruction here. Texas retirements are published only in aggregate and are spread proportionally across renewable fuels. Ranges span two import treatments.

So who gets to take credit?

Whoever retired the certificate — and only them. If a wind farm’s certificates were retired for a state mandate or bought by a corporate buyer, that wind is spoken for. A consumer with no certificates is, under market-based accounting, on the residual mix — and the registries say that mix is 73% gas in New England, 63% gas in New York, and holds zero unclaimed renewables in PJM.

The practical consequence for a practitioner: an unlabelled “grid mix” factor silently takes credit for renewables someone else already claimed. If your study uses market-based accounting and your site holds no certificates, the registry’s residual — not the average, and in the measured regions not eLCI’s residual — is the defensible number. In the claim-heavy Northeast, that choice moves the electricity factor by a factor of 1.2 to 2.1.

The registries

Every dataset used above, at its source. Coverage is uneven by design: each system decides for itself what the public gets to see.

NEPOOL GIS ↗

New England (ISNE)

All-generation registry. Public reports publish the residual mix by fuel per trading quarter; the system mix is published only as an aggregate.

PJM GATS ↗

PJM (Mid-Atlantic)

All-generation registry. Residual and system mixes both public, by fuel and month window.

NYGATS ↗

New York (NYIS)

All-generation registry run for NYSERDA. Publishes system and residual mixes by fuel and year — with imports folded in at certificate attributes.

M-RETS ↗

Midwest (MISO and beyond)

Certificate registry with public issued/retired statistics by fuel and year — but no state or balancing-authority dimension, so claims cannot be placed geographically from public data.

ERCOT REC Trading Program ↗

Texas (ERCO)

Renewables-only program. Retirements are published as annual aggregates in ERCOT's report to the Public Utility Commission of Texas — not by fuel.

2023 annual report (PUCT filing) ↗

WREGIS ↗

Western Interconnection

Publishes no public certificate data — the app is login-walled throughout. The Western residual is a coverage gap.

CRS / Green-e residual mix ↗

West (indicative)

The only published Western residual rate. Subtracts all specified sales under California disclosure rules — a different construction from the eastern registries.

California Power Source Disclosure ↗

California

Through the 2024 labels, unspecified power carries a fixed default (944 lbs CO₂e/MWh) rather than a calculated residual; a calculated one begins with the 2025 label.

Registry pools: NEPOOL GIS, PJM GATS, NYGATS public reports (2023); ERCOT REC program annual report via the Public Utility Commission of Texas; M-RETS public certificate statistics. System-side denominators for Texas and MISO: eGRID 2023. Per-fuel factors and published residuals: built from EIA-930 · eLCI · eGRID

Disclaimer. Academic project. US Government data used responsibly; outputs used at your own risk, with no responsibility or liability accepted. Opinions are my own and reflect no position of my employer.