Re-derive road fuel volumes from the October 2026 fuel-duty forecast #2720

gov.economic_assumptions.yoy_growth.obr.road_fuel_volume and the petrol and diesel spending proxies derived from it are still on the OBR's March 2026 fuel-duty receipts forecast. The October 2026 EFO refresh leaves them there because the derivation can't be reproduced from what's on record.

What's stored. The comment for 2025 to 2030 says: "OBR March 2026 fuel-duty receipts forecast, net of non-road fuel receipts, divided by fiscal-year average petrol/diesel duty rates." Commit 5e45253 added the values and no script.

Why it doesn't reproduce.

• Receipts Table 3.8 (cash basis) puts fuel duties at £24.359bn in 2024-25 and £24.242bn in 2025-26. That is a 0.5% fall, with the duty rate flat at 52.95p all year.
• The stored 2025 volume growth is −1.99%.
• So the netted-off non-road receipts, or the base year, differ from anything in the receipts tables. The fuel-duty supplementary release is the likely source, but it isn't recorded which edition or rows were used.

October 2026 inputs.

• Receipts Table 4.8 fuel duties, £bn: 24.559 (2025-26), 23.95, 25.713, 26.581, 26.267, 25.551, 24.432 (2031-32).
• Rates: 52.95p to 30 June 2027, then 60.59p (EFO ¶3.19), then RPI each April. This is the regenerated petrol_and_diesel.yaml in the October EFO PR.

Done means:

• a script under policyengine_uk/utils/ derives the series from a named receipts table and supplementary release;
• it reproduces the stored March values (or explains the difference);
• it is rerun on October, extending the series to 2031;
• the petrol and diesel spending proxies are re-derived with it (test_road_fuel_volume_uprating.py checks them);
• the economic assumptions README is updated.