"Britain has no industry": a sentence about thirty particular years
Nov 01, 2026
13 min read
9 views
Britain built the first factories, and the story goes that it then stopped caring about them. "Britain has no industry" gets repeated often enough to work like a fact. The question here came out of reading James Dyson's memoir and being surprised that manufacturing seemed to be held in such low regard in the country that started the industrial revolution. So: what do the public statistics actually support? Something narrower than the slogan, and more interesting.
Two warnings first. Whether manufacturing is looked down on is a question about attitudes, and no statistic here touches it. And the quantitative answer, for the period people usually have in mind, is not new: it was published a decade ago and more by Leif van Neuss, by Robert Rowthorn and Kenneth Coutts, and by Dani Rodrik, with bigger tables than the comparison below. What follows reproduces their result from two other public databases, says where every number comes from, and then points at the part that gets said less often.
The measure, and what it does not show
The standard measure of deindustrialisation is manufacturing's share of total employment: of everyone working in the country, what fraction works in manufacturing. It is used because it is comparable across countries and goes a long way back.
Being a share, it falls if manufacturing shrinks and also if everything else grows faster, and it says nothing about output. A country can employ a much smaller fraction of its workers in manufacturing and still make more goods than before: Britain's share fell by about twenty points between 1970 and the late 2000s while its real manufacturing output grew, by a factor of 1.28 or 1.90 depending on which official series you take, both of which are set out below. It also cannot see outsourcing: when a factory stops employing its own cleaners, designers and lorry drivers and buys those services instead, measured manufacturing employment falls with no change in what comes out of the gate. This article has no correction for that.
1970 to the late 2000s: Britain is the outlier, and this is known
On employment share over roughly 1970 to 2007, Britain's fall is the largest among advanced economies on every source tried. Leif van Neuss published the ranked table for 18 countries using EU KLEMS data, writing that "the UK seems to be the country in which manufacturing experienced the sharpest decline", with the share going "from 33.2% to 11.8% (-21.4%)". Robert Rowthorn and Kenneth Coutts, in an evidence paper for the UK Government Office for Science, put it as plainly as it can be put: "It has experienced the largest decline in manufacturing employment of any advanced economy," with a fall of 18.6 percentage points between 1973 and 2008. Dani Rodrik, drawing on the same GGDC database this article uses, wrote that "In Great Britain, to cite an example from the other end of the spectrum, deindustrialization has been both more rapid and thorough", with manufacturing's employment share falling "from a third in the 1970s to slightly above 10 percent today" and its real value added going "from around a quarter of GDP to less than 15 percent".
Those last two are exactly what our own GGDC run gives: a five-year-average employment share of 31.1% in 1968-72 against 11.2% in 2005-09, and a real manufacturing share of 25.4% against 15.0%, real meaning adjusted for price change, as against nominal, which uses the prices of the day (and against the GGDC sector total rather than GDP proper, a distinction taken up below). Reproducing a published number to the decimal place is a good outcome; calling it a finding would not be. Three databases, three samples, same answer, in percentage points of total employment on five-year averages. Each line gives EU KLEMS 1970 to 2007 (van Neuss), then GGDC 1968-72 to 2005-09 (ours), then OECD STAN 1971-73 to 2005-09 (ours):
United Kingdom: -21.4, -20.0, -19.0
Belgium: -17.4, not in GGDC, -17.0
Netherlands: -12.7, -13.2, -10.8
United States: -12.5, -12.5, -11.6
France: -10.6, -12.4, -11.9
Japan: -7.5, -7.8, -8.9
Britain is first in all three. But the runner-up is close enough that "Britain alone is off the chart" overstates it: Belgium is two points behind on STAN, four on EU KLEMS.
Rowthorn and Coutts add the detail that makes this worth explaining rather than reciting. They fit a model of deindustrialisation to advanced economies, with terms for productivity growth, shifting domestic demand and trade, and find Britain's fall "considerably larger than predicted by the econometric equations". The leftover runs 6.4 to 6.6 percentage points depending on the specification, much the largest unexplained difference in their table. So the size of Britain's fall is established. Why it was that large is not.
Since the early 1990s, Britain is ordinary
Stretch the window forward and the picture changes. Taking the OECD's 2025 STAN edition, which covers many more economies than the older databases, and re-aggregating it ourselves to compare 1991-95 with 2019-23. Each line gives the change in percentage points, then the level in 2019-23:
Korea: -8.8, 16.1%
Luxembourg: -8.6, 6.8%
Ireland: -8.5, 11.2%
Belgium: -8.2, 10.1%
United Kingdom: -8.2, 7.5%
Portugal: -7.9, 14.6%
New Zealand: -7.7, 9.3%
Japan: -6.9, 14.9%
Germany: -6.5, 16.7%
United States: -5.6, 7.8%
Britain is fifth of the 30 economies with data, in a group so tight that the gap from first to seventh is about one percentage point. The list rounds Belgium and Britain to the same -8.2; carried to two decimals on the same series, Belgium's fall is 8.23 points and Britain's 8.16, so the ordering holds but the gap between fourth and fifth is under a tenth of a point and should carry no weight. At the same two decimals Britain's 2019-23 level is the fourth lowest, with the Netherlands third at 7.50% against Britain's 7.52%, a difference of two hundredths of a point. The bottom of that ranking is tighter still: Luxembourg 6.8%, Australia 7.2%, the Netherlands 7.5%, Britain 7.5%, Norway 7.8%, the United States 7.8%.
So the honest version of "Britain has no industry" is a statement about roughly 1970 to the mid 2000s. That part is the published result, and it belongs to van Neuss, to Rowthorn and Coutts, and to Rodrik. What the re-aggregation above adds is the later window: over the past thirty years Britain has deindustrialised at about the rate of other rich countries, and sits in a cluster of them at a low but not uniquely low level. That cluster includes the United States, which is not ordinarily characterised in those terms.
Three ways to get a different number
Anyone who checks these figures against another source will get different ones. Three reasons, worth knowing before trusting any single number on this topic.
The denominator may not be the whole economy. The GGDC 10-Sector Database, widely used for this question, sums ten sectors, and that sum is not GDP. Dividing it by the OECD's total-economy value added gives 0.79 for Britain in 1970-72 and 0.75 in 2005-09, against 1.06 and 1.01 for the United States. The database deliberately excludes imputed rent, the value national accounts assign to owner-occupiers housing themselves, because there is no employment to match it. Its sources-and-methods document says such rents "frequently increase output in this sector by 50 per cent or more without any labour input equivalent", the sector being financial and business services, and that "this percentage varies over time and across countries". The consequence is concrete: manufacturing's share of nominal value added, meaning value added counted in the prices of the day, in Britain in 1970-72 is 36.0% against the ten-sector sum and 24.9% against the whole economy. The first number is not wrong, it is just not a share of the economy, and setting it beside a US figure computed on a near-complete denominator compares two different things.
Real output depends heavily on which series you pick. This ranking is published as well, in the same place: van Neuss's appendix 2 puts British real manufacturing value-added growth at 0.6% a year over 1970 to 2007, the lowest of his 18 countries, with Denmark next at 1.0%. What follows is not that result but how far it moves with the source. British manufacturing's real value added from 1970 to 2007 grew by a factor of 1.28 on GGDC and 1.90 on OECD STAN. Both are official: GGDC values output at the prices of a single year, 2005, while STAN re-bases year by year and chains the results into a volume index. We cannot say which is right. Over the full period our GGDC run agrees with van Neuss's ordering, Britain lowest of the four main comparators (Britain, the United States, France and Japan) at 1.28 with France next at 2.49, while STAN puts Britain third lowest of thirteen at 1.90, above Norway at 1.58 and Denmark at 1.83 and below France at 2.31. On both it grows more slowly than the British economy as a whole (2.44 on the same STAN series). But split the period and even the direction moves: on STAN, British manufacturing grows least of the four up to 1989 (1.08, with France next at 1.52) and second fastest from 1990 to 2007 (1.77, behind the United States at 1.91), while on GGDC it stays lowest throughout, though for 1990 to 2007 only barely, at 1.11 against Japan's 1.16. Anyone who tells you British manufacturing output did or did not grow is reporting a series, not a fact.
Vintages matter, and 1.28 is a 2014 vintage. In June 2021 the Office for National Statistics published the indicative effect of moving UK national accounts to double deflation, which strips price change out of a sector's outputs and its inputs with separate price indices instead of applying one index to the difference between them, and reported that "In the pre-crisis period annual average growth for the manufacturing sector was 3.1% in BB21 compared with 0.3% in Blue Book 2020". That is a revision of about 2.8 percentage points to an average annual growth rate over 1998 to 2007, a decade inside the window above, published years after the GGDC release used here. Note what it is not. It is not a tenfold revision of a level, and it is no reason to swap current UK figures into a cross-country table, because swapping the UK series inside a database built to measure every country the same way breaks the comparison the table exists to make. It is a reason to name the vintage, which is why this article does, and the same caveat applies to the EU KLEMS figures in the published literature.
A fourth point sits on top of these. Switch from employment to nominal value-added share and Britain stops standing out: on our STAN run from 1970-72 to 2005-09 its fall of 14.0 points is the largest of 18 economies, but Australia (-13.6), Japan (-13.2) and Belgium (-13.1) are all within a point. Van Neuss's EU KLEMS table has Britain at -21.7 against Australia at -14.0 and Japan at -12.9, a much wider gap, and ranks Britain second behind Luxembourg. Both can be true, since the sources and samples differ. That is the same lesson again, not a contradiction.
What these numbers cannot settle
There is a long argument about why, with a cultural explanation on one side and historians of industry and technology on the other. This article takes no side and will not summarise the books, because they were not read for it. Doing so would mean passing on other people's summaries as though they were the sources, which is how most popular accounts of this debate get assembled.
What can be said comes from two things that were read. David Edgerton, arguing against what he calls declinism, describes the position he attacks as the view that Britain is "an anti-scientific, anti-technological and anti-industrial culture", and Martin Wiener in particular as having "claimed to find that British elite culture was dominated by a literary yearning for the rural", leading to "the denigration of industry, science and technology". Edgerton's own verdict: "what is certain is that Britain has been neither particularly anti-science nor anti-industrial". That is a critic's account of his opponent, not Wiener's own, and should be read as such.
Second, Robert Rowthorn and Ramana Ramaswamy at the IMF treated deindustrialisation as general rather than British, with manufacturing's employment share across industrial countries falling "from about 28 percent in 1970 to about 18 percent in 1994". Their model attributes most of that decline to forces internal to rich economies, with north-south trade explaining "less than one-fifth" of it. Whether Britain's unexplained 6.4 to 6.6 points are cultural, financial, imperial or an artefact of measurement is not something an employment share can decide.
What would change the picture
Three things would. A correction for outsourcing, to say how much of Britain's early and steep fall is reclassification rather than lost production. A reconciliation of the GGDC and STAN real series for Britain, where the gap opens after 1995, the same point at which GGDC's European data switches from EU KLEMS to WIOD, a switch its own sources-and-methods document sets out and a coincidence we noticed and did not explain. And regional figures, because a national share hides the difference between the old industrial areas and London, and the slogan is mostly about the former.
What remains unknown is the thing the question started from. The numbers can say Britain's manufacturing employment fell further than its peers' between 1970 and the mid 2000s, which van Neuss, Rowthorn and Coutts, and Rodrik established and this article only reproduces from other databases, and that it has been unremarkable since, which is the part that gets said less often. They cannot say why it fell that far, and they cannot say anything at all about how the work came to be regarded.
Sources and how to check
Every input is public and free. The two re-aggregations in this article were computed from the files below on 2026-11-02; the GGDC file has SHA-256 0a99de029c82fef8e68d7841c93a7ec1b9671f0b5e59c7094d0bbbab61c08e24.
GGDC 10-Sector Database, 2014 release (the page asks that use be cited to Timmer, de Vries & de Vries, 2015, and warns that "the aggregate results are not fully consistent with the national accounts aggregates"):
https://www.rug.nl/ggdc/productivity/10-sector/
GGDC 10-Sector Database: Contents, Sources and Methods, January 2015, by de Vries, de Vries, Gouma, Pahl and Timmer, 126 pp. The imputed-rent passage quoted above is footnote 5 on page 6; the switch of the European data from EU KLEMS to WIOD in 1995 is in its section on the USA and Europe:
https://www.rug.nl/ggdc/docs/10sector_sm_jan2015.pdf
OECD STAN Database for Structural Analysis, 2025 edition, via the public SDMX API. Employment shares here use persons engaged. This dataset also carries hours and full-time-equivalent series under the same measure code, and mixing them gives meaningless shares, which is worth knowing before anyone reruns this:
https://sdmx.oecd.org/public/rest/data/OECD.STI.PIE,DSD_STAN@DF_STAN_2025,/A..C+_T.B1G+EMP..?format=csvfilewithlabels
That query is printed in full on purpose. It is the whole point of citing a public statistics service: anyone can pull the same series and recheck every STAN number in this article, including the extra decimals quoted for the close ranks.
Van Neuss, L. (2016), "The Economic Forces Behind Deindustrialization: An Empirical Investigation", University of Liege working paper, 50 pp.; published as "Globalization and deindustrialization in advanced countries", Structural Change and Economic Dynamics 45 (2018), 49-63. The ranked tables are appendices 1 and 2 of the working paper:
https://hdl.handle.net/2268/200764
Rowthorn, R. & Coutts, K. (2013), "De-industrialisation and the balance of payments in advanced economies", Future of Manufacturing Project Evidence Paper 31, Government Office for Science:
Rodrik, D. (2015), "Premature Deindustrialization", NBER Working Paper 20935, February 2015; published as "Premature deindustrialization", Journal of Economic Growth 21(1), 1-33:
https://www.nber.org/system/files/working_papers/w20935/w20935.pdf
Rowthorn, R. & Ramaswamy, R. (1999), "Growth, Trade, and Deindustrialization", IMF Staff Papers 46(1), 18-41:
https://www.imf.org/external/pubs/ft/staffp/1999/03-99/pdf/rowthorn.pdf
Office for National Statistics (2021), "Indicative impact of a new framework including double deflation on industry volume estimates of GDP: Blue Book 2021", released 28 June 2021:
Edgerton, D. (1996), "Myths of decline", Prospect, August 1996:
https://www.prospectmagazine.co.uk/essays/55114/myths-of-decline
The reading note the question came from, in Japanese:
https://note.com/kei_book/n/n25139e178abd
How this was checked
The van Neuss, Rowthorn and Coutts, Rodrik, Rowthorn and Ramaswamy, ONS and Edgerton quotations above were read in the original documents, not in summaries of them. So was the GGDC material: the imputed-rent footnote and the 1995 change of European source were read in the January 2015 sources-and-methods PDF linked above, and the attribution requirement and the consistency warning on the database's own page were read there. The GGDC and OECD STAN figures were recomputed from the raw files on the date given. One source we wanted could not be read: the 1997 IMF working paper by Ramaswamy and Rowthorn is posted as a scanned image with no text layer, so the 1999 journal article is cited in its place. The books in the cultural-decline debate were not read, which is why this article declines to characterise them.