ON THURSDAY the government published three things at once: an evaluation of a retrofit scheme, a record for solar, and the quarterly generation statistics. Nothing in them is concealed. Every caveat below is the departments’ own. What changes between the press line and the evidence is the unit of account: measure fuel poverty by the property’s EPC band and it falls, measure it by what the household pays and it doesn’t.

START WITH the retrofit evaluation. This is not the department marking its own homework: DESNZ commissioned IFF Research, Technopolis and BRE, gave the work to a scrutiny panel including the National Housing Federation and the Social Housing Regulator, and published what came back. Wave 1 of the Social Housing Decarbonisation Fund put 31,693 measures into 16,056 English social homes for £381m, of which £159m was grant, and the evaluators metered the result against 37,245 matched control properties to find the treated homes saving an extra 1,079.9 kWh of gas a year. Worth £97, and a fifth of a tonne of CO₂.

Take the public money alone and set the landlords’ £222m aside. That is £9,903 a home to take £97 a year off a gas bill [RL calculation]. Price the carbon over thirty years and the grant buys abatement at about £2,200 a tonne, against the £273 central value in the government’s own appraisal guidance — five times the top of its sensitivity band [RL calculation]. Every omission in that sum runs the government’s way, since it counts no warmth, no damp remediation and no carbon from the solar or clean-heat measures the evaluators could not track, and it holds against a scheme that was never appraised on carbon alone and did buy real capability, with installers taking accreditations and firms widening what they could fit. That capability then delivered 21% fewer homes than planned, at 32% more each, with clean heat installations cut from 1,936 to 1,344. The purchase shrank as the price rose.

And the scheme reports a triumph. Ninety-seven per cent of fuel-poor households were taken out of fuel poverty, because England measures fuel poverty by EPC band, so moving the certificate to C removes the household from the statistic whatever it pays. The evaluators say so themselves, adding that reaching EPC C ‘may have not led to a reduction in energy bills’. Bills did fall — £97, on strong evidence. What should trouble you is the sample. That 97% is 284 households, and it is reported as a national outcome.

THE SOLAR record has the same shape and a harder answer. DESNZ’s first regional breakdown counts nearly 172,000 installations since January, 28,000 in July, ‘one installation every 2 minutes’. The unit is the installation. A balcony kit and a solar farm each count once. The capacity data was published the same day in ET 6.3, one click from the release’s own notes to editors, and it shows GB solar going from 21.67 GW in December to 22.56 GW in July — 895 MW in seven months, an annualised 1,534 MW [RL calculation]. The Clean Power 2030 plan wants 45 to 47 GW by 2030, which needs about 5,100 MW a year. In a record year, Britain is building at under a third of the rate its own plan requires.

The quarterly statistics landed alongside, reporting that renewables provided 50.8% of electricity from major power producers in April to June, up 7.6% on the year, while two lines down the same table the low-carbon share falls 0.9 points to 67.2% and the fossil share rises a point to 32.2%, because nuclear output dropped 12% over the same period. One quarter proves nothing, and the series excludes embedded generation. But the shape is the argument. Renewables passed half, and the grid got dirtier, because the firm plant retiring is being replaced by weather.

TWO subsidy lines closed the week. The Green Gas Support Scheme forecast reached £190.82m against £98.42m a year earlier, up 94%, on 33 applications each carrying a fifteen-year tariff. That is a headroom figure, not expected spend, and it triggered no cut. The geometry is what to watch: the degression threshold flattens after 2027 while the forecast nearly doubles annually, so on this trajectory the scheme meets its own trigger during 2027, which is a dated prediction anyone can check. The levy funding it is charged per meter. It falls heaviest per unit on whoever uses least.

Then the visible relief. From 1 October VAT comes off household electricity, worth about £45 a year, funded — DESNZ says so plainly — ‘through the cancellation of the digital ID programme’, for one financial year only. The cap still rose 4%, and that 4% already has the tax cut netted into it, so the underlying rise is larger. Ofgem’s stated cause is gas.

Five documents, one habit. Where the state picked the unit, the answer came out flattering. Where it published the full arithmetic, the answer is £9,903 of public money a home to save £97 a year, and a solar build rate at under a third of what the plan needs. Both came out of the same building on the same day. Only one of them will be quoted.