Skip to content

Gas counterfactual

An audit trail for the orange "gas alternative" line on the CfD-vs-gas chart.

The question

"If we'd kept the gas fleet we already had instead of paying for renewables, what would the same electricity have cost?"

This is a policy-level counterfactual, not a market-clearing simulation. We are not modelling what wholesale prices would have done in the absence of renewables — that depends on gas supply elasticity, demand response, and interconnector flows that are beyond the scope of this analysis. We are asking: given the gas prices and carbon prices that actually occurred, what would it have cost to generate the same MWh from a CCGT fleet?

The formula

For each day, compute an implied £/MWh for gas-generated electricity:

gas_alt_£/MWh  =  fuel  +  carbon  +  O&M

fuel    =  gas_price_p_per_kWh × 10 / 0.55
carbon  =  UK_ETS_£_per_tCO2 × (0.18290 / 0.55)
O&M     =  £5 / MWh   (existing fleet, capex sunk)

Each daily £/MWh is multiplied by the actual daily CfD generation (MWh) and summed to monthly, then cumulative, totals.

Term Source Value
Gas price ONS System Average Price of gas Daily, p/kWh thermal
CCGT efficiency BEIS Electricity Generation Costs 2023, Table ES.1 55%
Gas CO₂ intensity DESNZ 2024 UK Government GHG Conversion Factors (gross CV) 0.18290 tCO₂/MWh thermal
Carbon price EU ETS (2018–20) → UK ETS (2021+), annual average (OBR) £13–£73/tCO₂
O&M BEIS Electricity Generation Costs 2023, Table ES.1 £5/MWh

Constants last audited against primary sources: 2026-04-22. Full provenance (source URL, basis, retrieval date, next-audit trigger) is recorded on each constant's docstring in counterfactual.py. Grep ^Provenance: across the source tree to enumerate every audited constant.

The × 10 / 0.55 converts p/kWh of gas (thermal) into £/MWh of electricity: the × 10 is a unit conversion, and dividing by 0.55 accounts for the heat lost in combustion — 1.82 kWh of gas burned per 1 kWh of electricity delivered.

Why existing fleet, not new-build

The UK CCGT fleet was largely built 1995–2012. By the time the CfD scheme started paying generators (2015) most of that capex was already sunk. The policy choice wasn't "build renewables or build gas"; it was "pay for renewables or keep running the gas plants we already had." For that comparison, only operating costs (fuel, carbon, O&M) are relevant — capital costs are bygones.

Using £5/MWh (BEIS Table ES.1 — fixed ~£3/MWh + variable ~£2/MWh for operational H-class CCGT) captures this. A new-build figure (£20/MWh with capex + finance) would be appropriate for a different counterfactual — "what if we'd built new gas plants instead of wind and solar?" — but that isn't the question the chart is answering.

Sensitivity

All figures are cumulative totals over Jan 2018 – Apr 2026 (the window where both the LCCC CfD data and the gas counterfactual are available) against £28.5bn of actual CfD electricity cost (wholesale + levy) over the same window. Numbers are generated from the current code via compute_counterfactual(...) — see the snippet at the end of this page.

Scenario O&M (£/MWh) Carbon tax Gas alternative Premium vs CfD
Default (existing fleet) 5 UK ETS £14.4bn £14.1bn
Fuel + carbon, no O&M 0 UK ETS £13.5bn £15.1bn
Fuel only (no carbon, no O&M) 0 — £10.9bn £17.6bn
New-build (capex + O&M) 20 UK ETS £17.3bn £11.2bn
No carbon tax, existing O&M 5 — £11.9bn £16.6bn

Including the earlier 2016–17 CfD generation (no gas counterfactual available for that period) brings the full-window CfD total to £29.1bn — the headline "~£29bn over ten years" figure on the home page. The £0.7bn difference is real CfD spending but cannot be set against a like-for-like gas comparator.

The choice of O&M and carbon-price assumptions moves the cumulative gas alternative within a £11–17bn band. The actual CfD cost (£28.5bn) sits comfortably above that band in every scenario: the premium is robust to plausible assumption changes.

Caveats

These are limits of the model, not reasons to distrust the direction:

  • Gas supply elasticity. We assume gas would have been available at the observed wholesale prices even with ~150 TWh/year of extra gas demand (the renewable generation that would not exist in the counterfactual). In reality extra demand would likely push gas prices up, which would make the gas alternative more expensive than the line shows — the premium shown here is if anything a lower bound on the 2021–23 crisis years.

  • Fleet age and retirements. Some of the 1995–2012 fleet has retired (Killingholme 2015, Cottam 2019) or had mid-life refurbishment over 2015–26. Using a flat £5/MWh O&M slightly understates long-run costs for a fleet aging past 30 years.

  • Thermal efficiency. 55% is representative of modern H-class CCGT. Older F-class plants run closer to 50%, which would raise fuel cost by ~10%. Using the best-case efficiency makes the gas alternative look cheaper — again making the shown premium conservative.

  • Carbon price coverage. 2018–20 used EU ETS (UK was still in scheme). UK ETS prices from 2021 onward. Pre-2018 carbon costs are omitted because the UK Carbon Price Floor was much lower and CfD generation volumes were small, so the effect is negligible.

Reproducibility

The full calculation lives in src/uk_subsidy_tracker/counterfactual.py.

Key constants:

  • CCGT_EFFICIENCY = 0.55
  • GAS_CO2_INTENSITY_THERMAL = 0.18290 (tCO₂/MWh thermal, DESNZ 2024 gross CV)
  • CCGT_EXISTING_FLEET_OPEX_PER_MWH = 5.0
  • CCGT_NEW_BUILD_CAPEX_OPEX_PER_MWH = 20.0
  • DEFAULT_CARBON_PRICES (dict keyed by year; 2018–2020 EU ETS, 2021+ UK ETS)

Each constant carries a full Provenance: block in its docstring. To list all audited constants:

grep -rn "^Provenance:" src/ tests/

To reproduce the chart:

uv run python -m uk_subsidy_tracker.plotting.subsidy.cfd_vs_gas_cost

To run a sensitivity scenario, call compute_counterfactual() with overrides:

from uk_subsidy_tracker.counterfactual import (
    compute_counterfactual,
    CCGT_NEW_BUILD_CAPEX_OPEX_PER_MWH,
)

# new-build instead of existing fleet
cf = compute_counterfactual(
    non_fuel_opex_per_mwh=CCGT_NEW_BUILD_CAPEX_OPEX_PER_MWH,
)

# strip carbon tax entirely
cf = compute_counterfactual(carbon_prices={})