British industry makes a quarter more than it did in 2005, using a third less electricity

Two of the government’s own statistical series seem to tell opposite stories about deindustrialisation. Read together, they show which sectors decoupled and which actually declined

Meridian Trade Advisory. 16 August 2026

On 30 July, DESNZ published the Digest of UK Energy Statistics. It records British industry consuming 76.7 TWh of electricity in 2025, the lowest level since 1983 and a third below the 116 TWh consumed in 2005.

The Office for National Statistics publishes a different series. Its Index of Production records manufacturing output in 2025 at roughly a quarter above its 2005 level, in real terms.

Read the first alone and the story is decline. Read the second alone and the story is growth. Both are the government’s own numbers, and both are true. The interesting question is how.

One table, two series

Put the two series side by side, sector by sector, and the answer appears. The electricity column is from DUKES. The output column is from the Index of Production. The third column is derived from the first two.

The decouplers

Start with the extreme case. Vehicle manufacturing produces 84 per cent more real value than it did twenty years ago, using 55 per cent less electricity. Its electricity intensity, the power needed per pound of output, has fallen by three quarters.

Chemicals and food tell the same story in a milder key: output up strongly, electricity down. Across manufacturing as a whole, Britain makes a quarter more than it did in 2005 with a third less electricity. The intensity of the whole sector has fallen by almost half.

Even where output is flat, the pattern holds. Paper and printing produce about what they did in 2005, on more than a third less electricity. Electrical equipment, the same.

Whatever else is true about British industry, the idea that falling electricity use simply tracks falling production is not. In most of the economy, the two have separated.

The pockets where they have not

Not every row decouples. Machinery output is down 14 per cent since 2005, and its electricity consumption is down 43 per cent. There, part of the fall is contraction rather than efficiency.

And then there is steel. Electricity consumption in iron and steel has fallen 66 per cent, the steepest decline of any sector. The fuel data shows what sits underneath: blast furnace gas burned for industrial generation fell 58 per cent over the same period, which is the signature of furnace capacity closing, not of processes becoming more efficient. The electricity fall in steel is mostly the story of a smaller industry.

So both popular accounts are wrong, and the table shows where. “Britain does not make things any more” fails in the aggregate, where output is up a quarter. “It is all efficiency” fails in the pockets, where electricity fell because the work left.

What this changes

For twenty years, industrial electricity consumption has been read as a proxy for industrial health. This table says it no longer is one. A sector can halve its electricity use while nearly doubling its output, and one did.

That matters anywhere electricity statistics are used to stand in for industrial reality. It matters, for example, in the government’s new industrial energy scheme. The BICS sectoral consumption estimates apportion national energy statistics down to individual sectors using three different bases, one of which is gross value added. In a decoupled sector, value added and electricity move in opposite directions, so the bases disagree. That is consistent with what the published estimates show: motor vehicles, the most decoupled sector in the table, carries one of the widest ranges in the scheme at 990 to 1,400 GWh, while sectors where the statistics agree are tightly bounded. That last step is an inference from the published ranges rather than a published fact, but it is a testable one, and it suggests the scheme’s uncertainty is concentrated exactly where decoupling is deepest.

Caveats

Three, because they matter.

Output here is real gross value added, not units. Vehicle production in units has fallen substantially over the past decade, and that story, which is about model mix and value per vehicle as much as volume, is a different article.

The intensity column is implied rather than measured. The two series use different sector definitions, electricity consumption includes self-generation, and the pairing for steel is the loosest: the electricity figure covers iron and steel, while the output index covers basic metals, a wider group.

And none of this says British manufacturing is healthy. It says something narrower and more useful: the health cannot be read from the electricity meter.

Sources

Electricity consumption figures are from the Digest of UK Energy Statistics 2026, Table 5.2 (industrial consumption split between the public distribution system and own generation) and Table 5.4 (autogeneration fuel use), published by DESNZ on 30 July 2026. Output figures are from the ONS Index of Production time series, release of 16 July 2026: manufacturing total and divisions 24 to 29, chained volume measure, seasonally adjusted, annual averages of monthly indices. Intensity figures are derived from the two series as described in the text. The BICS sectoral consumption estimates and their apportionment methodology were published by DBT on 8 July 2026.

If you think I have made an error, tell me and I will correct it.

Anthony Salasidis is the founder of Meridian Trade Advisory, which advises UK manufacturers on industrial energy costs, including BICS eligibility, pro-rating and evidence. anthony@meridiantradeadvisory.co.uk