UK domestic electricity and gas: record why ONS 04.5.1/04.5.2 is preferred over DESNZ × QEP, reconcile the two, and declare the price basis PE-UK should project from #1113

Revised after review (6 Oct 2026): following juaristi22's review:

• Corrected a false statement. The body said no ruling was recorded. In fact the current implementation contract is in the gates.json notes, including her 15 Sep 2026 ruling on UK #890 PR-S: uprate the LCFS and ETB donors, condition on vehicles, draw NTS bus use, rake bus and energy to vendored facts, price energy at the FY2024-25 cap #927. The issue is now split into "already documented" and "still to decide or validate".
• Added the producer's certified electricity fit (£26.5199bn against £26.5220bn, −0.008%).
• The ~83 TWh implied-volume figure is now labelled as a diagnostic, with a requirement to recompute it per household.
• Replaced "already tracked in What price basis do energy outputs carry after the Q2 2026 anchor? policyengine-uk-data#485" with separate deliverables for Microcosm (declare reference years and basis) and PE-UK (project from them consistently). Added the audit finding from Drill 2: mock Autumn Budget setup, 5 October (MOCK DATA, do not merge) uk-autumn-budget-dashboard-2026#7 that neither a £35bn base nor blanket CPI fixes this. Linked The temporary 0% VAT on domestic electricity in Great Britain (1 October 2026 to 31 March 2027) is not modelled, and VAT on domestic energy cannot be changed on its own policyengine-uk#2166.

Title

UK domestic electricity and gas: record why ONS 04.5.1/04.5.2 is preferred over DESNZ volume × QEP price, reconcile the two, and declare the price basis PE-UK should project from

Body

Summary

The Microcosm UK national file carries £26.5–26.9bn a year of domestic electricity spending (electricity_consumption, 2024–2027). That matches the bound target, ONS Consumer Trends COICOP 04.5.1 for CY2025 (£26.52bn). It sits about 10% below what the build's own energy stage produces from published volume and price: the DESNZ Energy Trends domestic volume at DESNZ Quarterly Energy Prices (QEP) average prices paid, about £29.4bn. Gas shows the same conflict, and a bigger one (+24.6%).

How the build handles this is already documented (see below). What is missing is the reason ONS is preferred, a reconciliation of the two sources, and a declared price basis that PE-UK can project from.

The figure that prompted this check, about £35bn including VAT, has no support in any official source and is not a valid calibration target.

Already documented: the current implementation contract

The gate notes for the lcfs_consumption energy_rake check in packages/microcosm-build/src/microcosm/build/uk/gates.json already state:

• The stage sets each fuel's level from the DESNZ Energy Trends domestic volume over the four FY2024-25 quarters, priced at QEP, and checks it at stage time.
• The calibrated frame is held to the bound ONS 04.5.1 (electricity) and 04.5.2 (gas) spending rows.
• The gap between the stage's DESNZ × QEP level and ONS spending is a recorded published-source residual (UK #890 PR-S: uprate the LCFS and ETB donors, condition on vehicles, draw NTS bus use, rake bus and energy to vendored facts, price energy at the FY2024-25 cap #927 receipts, Part T), not a gate.
• juaristi22's ruling of 15 Sep 2026 on UK #890 PR-S: uprate the LCFS and ETB donors, condition on vehicles, draw NTS bus use, rake bus and energy to vendored facts, price energy at the FY2024-25 cap #927 moved the energy check to before calibration.

So the precedence (ONS binds; DESNZ × QEP is a stage-time level and diagnostic) and the residual handling are documented. This issue does not ask for them to be re-decided from scratch.

The selected release's producer diagnostics (PE-UK 2.100.0 / core 3.32.5) confirm the fit: electricity expenditure £26.5199bn against the £26.5220bn target (−0.008%). Source: selected-build provenance and diagnostics in PolicyEngine/uk-autumn-budget-dashboard-2026#7 (comment of 6 Oct 2026). This validates the recorded expenditure fit under the producer runtime. It does not show agreement with independent volume measures.

Still to decide or validate

1. A written rationale for the target choice. Record next to price_domestic_energy (in the contract notes or sources.yaml) why ONS 04.5.1/04.5.2 is preferred over DESNZ volume × QEP price, and who approved it. If the preference changes after step 2, state the new choice:

• (a) keep ONS as the level and record the implied-volume shortfall as a known limitation in the release notes, so per-kWh costings are warned;
• (b) make DESNZ × QEP the level and keep the ONS rows as diagnostics; or
• (c) keep both and publish the disagreement with its size so users can bracket.
  Each option should leave the stage and the solver agreeing on the level.

2. Reconcile the two sources before choosing between them. The stage-time DESNZ result and the final weighted ONS result answer different questions. Reconcile:

• reference period: FY2024-25 against CY2025;
• coverage: meters against households;
• regional and payment-method tariffs, and standing charges;
• the price basis, including VAT treatment.

3. Treat the implied-volume figure as a diagnostic. At QEP FY2024-25 prices, the file's electricity spend implies roughly 83 TWh, against 93.0 TWh in Energy Trends (about −11%). That uses one national-average conversion, which cannot establish the exact size of any shortfall. Before publishing a measured volume shortfall, recompute implied kWh using each household's regional tariff and fixed charge and the final weights. Do the same for gas.

4. Declare the price basis, and have PE-UK project from it. This is the projection-contract defect found in juaristi22's audit (PolicyEngine/uk-autumn-budget-dashboard-2026#7). Microcosm prices energy at QEP FY2024-25 and calibrates to ONS CY2025 expenditure. PE-UK's uprating_indices.yaml leaves energy out of dataset projection because it assumes the data are already at Ofgem Q2 2026 (FY2026-27) prices. That assumption comes from the legacy policyengine-uk-data build. It does not hold for a Microcosm file, which therefore carries 2024-25/2025 prices into every later year. Two deliverables:

• Producer (Microcosm, this issue): publish with the data the expenditure, price and volume reference years, and the tariff and VAT basis, for electricity and gas.
• Consumer (PE-UK, needs a named owner and its own issue): project energy from the declared reference years consistently, instead of assuming a fixed later price anchor.

Raising the base to the mock £35bn, or restoring blanket CPI uprating of energy, would not fix this contract. PolicyEngine/policyengine-uk-data#485 is related legacy-pipeline work. A change there does not by itself change Microcosm's exported H5 or PE-UK's projection contract.

Native electricity-VAT policy (the temporary 0% rate from 1 October 2026) is separate PE-UK work in PolicyEngine/policyengine-uk#2166. Choosing an electricity calibration total will not implement it.

5. Rebuild and certify. Keep the source discrepancy and its assumptions recorded until a decision is made. Then rebuild, recalibrate and certify the chosen model/data pair. The selected release's certification covers PE-UK 2.100.0 / core 3.32.5, not the newer drill runtime automatically.

Numbers (weighted aggregates)

Related

• UK #890 PR-S: uprate the LCFS and ETB donors, condition on vehicles, draw NTS bus use, rake bus and energy to vendored facts, price energy at the FY2024-25 cap #927 (merged): introduced price_domestic_energy, measured the +10.7% / +24.6% residual, and holds the 15 Sep 2026 ruling quoted in gates.json.
• Bind DfT, HMRC, DESNZ, Ofgem and ORR facts for bus fares, road fuel, domestic energy and rail support (#790 ruling: bind facts and calibrate) #890 (closed): item F.
• Drill 2: mock Autumn Budget setup, 5 October (MOCK DATA, do not merge) uk-autumn-budget-dashboard-2026#7: audit with the selected-build diagnostics and the projection-contract finding.
• The temporary 0% VAT on domestic electricity in Great Britain (1 October 2026 to 31 March 2027) is not modelled, and VAT on domestic energy cannot be changed on its own policyengine-uk#2166 (open): native electricity-VAT policy, separate work.
• electricity_consumption and gas_consumption declare CPI uprating but are never uprated policyengine-uk#1859, Energy uprating starts from the data year, not the price-level year the build stores policyengine-uk#1867 (closed), and 2.100.1: removed energy from dataset projection.
• What price basis do energy outputs carry after the Q2 2026 anchor? policyengine-uk-data#485 (open): price basis in the legacy pipeline. Related, but it does not cover Microcosm or the PE-UK consumer side.

Sources

• ONS, Consumer Trends, COICOP 04.5.1 electricity and 04.5.2 gas, current prices.
• DESNZ, Energy Trends section 5, table 5.5: domestic electricity consumption.
• DESNZ, Quarterly Energy Prices / annual domestic energy price statistics, table 2.2.4: average variable unit cost and fixed cost paid, including VAT.
• Ofgem, Energy price cap unit rates and standing charges, and Changes to the energy price cap, 1 April to 30 June 2026.

🤖 Generated with Claude Code