Industrial Britain's workers were paid in cloth and spent on bread

Whenever somebody argues that artificial intelligence will enrich owners years before it enriches workers, a seventy-year stretch of British history is called as a witness.¹² The episode has a name, Engels' pause, given it by the economic historian Robert Allen; it points at Friedrich Engels, whose first book described the Manchester of the 1840s and appeared in Leipzig in 1845.¹² From roughly 1770 to 1840 output per British worker rose by close to a half while the real wage barely moved, the profit rate doubled, and the share of profits in national income expanded at the expense of labour and land.¹ Nicholas Crafts, using national accounts that did not exist when Allen wrote, says it is not one, and the shape of his disagreement is more useful than the verdict.

Neither side disputes that output outran wages. What is disputed is what the gap licenses, and that turns on a choice: there is more than one real wage. A real wage is a money wage divided by a price index, and the index is a decision. Deflate by the prices of what a worker buys, mostly bread in 1820, and you have the consumption wage: what an hour of work fetched at the shops. Deflate by the prices of what the worker helps produce, cloth and iron and coal, and you have the product wage, what an hour of labour cost an employer in units of his own output. Crafts states the division of labour between them plainly, that consumption earnings gauge living standards while product earnings should be the basis for deriving trends in factor shares.³ Factor shares, the split of national income between labour, capital and land, are governed by what labour costs in units of output, not what a wage buys at the baker's.

Allen built his case on the consumption wage. Using the Crafts and Harley output estimates, output per worker rose about 46 per cent between 1780 and 1840, against real wage growth of about 12 per cent on Charles Feinstein's series or 18 on Allen's own deflator.⁴ From 1840 to 1900 output per worker rose about 90 per cent and real wages by more, which is why it is called a pause.⁴

Before Crafts, the quarrel was about the deflator, and it ran twenty years. Feinstein's 1998 reconstruction of earnings and prices put the improvement in an average working-class family's standard of living at under 15 per cent between the 1780s and the 1850s.⁵ Gregory Clark's 2005 series was far more optimistic, finding real wages trendless before 1800.⁶ Lindert and Williamson had earlier put a blue-collar index at 50 in 1819 and 100 in 1851.⁷ Allen rebuilt the index from the strongest components of both and landed much closer to Feinstein.⁸ Among economists deflating by the cost of living, the pessimists had the better of it.

Crafts does not reopen that fight. He keeps the cost-of-living work and changes which wage is set against productivity. On his estimates real product wages grew more slowly than output per worker before 1800 and faster afterwards, and through the first half of the nineteenth century faster than consumption earnings: by 1820 consumption earnings were up about 12 per cent on 1770 and product wages about 16. By 1840 real output per worker stood 43.9 per cent above its 1770 level and real consumption earnings 27.1 per cent above theirs.³ Labour's share, on this accounting, barely moved, because real product wages grew at about the same rate as labour productivity over the medium term.³

The wedge between the two wages is a movement in relative prices, a shift in the rate at which the goods a worker makes exchange for the goods he eats, and Crafts attributes the long-run divergence to exactly that, manufactures becoming cheaper relative to food.³ The spinning frame made cloth cheap and did nothing for the price of wheat, which was most of a labourer's budget. Several forces kept food dear: a population growing faster than its farmland, two decades of war with France, and a tariff written for landowners. The Importation Act of 1815 shut foreign wheat out until the domestic price reached eighty shillings a quarter, and the Corn Laws stood until 1846.⁹ A worker in 1830 was paid out of goods whose price was falling and spent the money on goods whose price was not.

So the episode is strong evidence for a proposition quite different from the one it is summoned to support. Crafts states his own conclusion bluntly: economists looking for an historical example of rapid labour-saving technological progress seriously hurting labour's share will have to look elsewhere.³ The agent is the price of bread rather than the machine, and the instrument that mattered was a tariff repeal rather than a factory act. That matters now, because when wages lag measured productivity two diagnoses are available. One is distributional: the division of the product has moved against labour. The other is a relative-price diagnosis: labour's claim on output is intact, but the goods workers buy have grown dear relative to the goods they make, with housing the obvious modern candidate for the role wheat played. The two are distinguishable. Under the first the product wage lags productivity too; under the second it tracks productivity while the consumption wage falls behind, and the remedy lies in the supply of the dear necessity rather than the division of a firm's surplus.

Three objections deserve answers, and the first cuts against Crafts. His labour-share series is not an independent measurement: it scales a base-year share by the ratio of real product wages to output per worker, using the GDP deflator estimates of Broadberry and co-authors.¹⁰ So the finding that labour's share barely moved and the finding that product wages tracked productivity are one finding in two hats, and both rest on a price deflator for eighteenth-century national output. Second, the gap is real even on Crafts's numbers: a generation receiving under three-fifths of the gain it produced would not have been consoled to learn the shortfall was arithmetically a relative price. Crafts's answer is that against a population surge, slow real wage growth may be a good outcome.¹¹ Third, and best, is Ricardo's. If food stayed dear because land was scarce and protected, the gain the workers did not get still went to an owner, and that owner was the landlord.¹³ The quarrel about profits may have been aimed at the wrong proprietor.

There is a rule here. A real wage is two numbers and a choice, so ask what the wage was deflated by and what it is set against. A consumption wage placed beside output per worker measures something real, and does not measure the division of the product. The best-known historical warning that technology shrinks labour's share turns out, on the most recent accounts, not to be an instance of it. That is no reassurance about machines. It is a reminder that what squeezed British workers for two generations showed up in the price of a loaf, and that ending it took a political fight about a tariff.

Notes

1Robert C. Allen, "Engels' pause: Technical change, capital accumulation, and inequality in the british industrial revolution," Explorations in Economic History 46, no. 4 (October 2009), pp. 418-435, for the coining of the term and for the abstract's statement that the real wage stagnated while output per worker expanded, that the profit rate doubled, and that the share of profits in national income expanded at the expense of labour and land. Not opened; the environment's network policy blocked every attempt to reach the document, so the abstract and the quoted passages come from search results. Publisher page, NYU Scholars record and the author's own copy

2Friedrich Engels, Die Lage der arbeitenden Klasse in England (Leipzig: Otto Wigand, 1845), written between September 1844 and March 1845; the authorised English translation by Florence Kelley Wischnewetzky appeared in New York in 1887. Sources disagree about the date of the first London edition and it is not relied on here. Marxists Internet Archive text and a catalogue account of the book

3Nicholas Crafts, "Slow real wage growth during the Industrial Revolution: productivity paradox or pro-rich growth?", Oxford Economic Papers 74, no. 1, pp. 1-13, for the distinction between real consumption earnings as the measure appropriate to living standards and real product earnings as the basis for factor shares; for the findings that product wages grew more slowly than output per worker before 1800 and faster afterwards, that product wages grew faster than consumption earnings across the first half of the nineteenth century, that by 1820 consumption earnings had risen about 12 per cent since 1770 against about 16 per cent for product wages, and that by 1840 real GDP per worker was 43.9 per cent above its 1770 level against 27.1 per cent for real consumption earnings; for the conclusion that labour's share changed little because real product wages grew at a similar rate to labour productivity in the medium term; for the attribution of the long-run divergence to manufactures becoming cheaper relative to food; and for the statement that those seeking an historical example of rapid labour-saving technical progress seriously harming labour's share must look elsewhere. Sources date the article to both 2021 and 2022 for the same issue. Not opened. Publisher page, the CAGE working paper version and the Warwick repository copy

4For Allen's figures of output per worker rising about 46 per cent between 1780 and 1840 on the Crafts-Harley estimates, against real wage growth of about 12 per cent on Feinstein's series and about 18 per cent on Allen's own deflator, and for output per worker rising about 90 per cent between 1840 and 1900 while real wages rose by more. These figures were returned by search results summarising Allen's paper and the surrounding literature rather than read from his tables, and accounts differ over whether the window opens in 1770, 1780 or 1790. Allen's own copy of the paper and an encyclopedia account of the episode

5Charles H. Feinstein, "Pessimism Perpetuated: Real Wages and the Standard of Living in Britain during and after the Industrial Revolution," The Journal of Economic History 58, no. 3 (September 1998), pp. 625-658, for new estimates of nominal earnings and the cost of living from 1770 to 1870, for the adjustments for unemployment, dependants and the costs of urbanisation, and for the finding that the standard of living of the average working-class family rose by less than 15 per cent between the 1780s and the 1850s. The journal later published a correction to the article. Not opened. Publisher page and a bibliographic record

6Gregory Clark, "The Condition of the Working Class in England, 1209-2004," Journal of Political Economy 113, no. 6 (December 2005), pp. 1307-1340, for the optimistic series and for the finding that real wages were trendless before 1800 as a Malthusian account would predict. The characterisation of Clark's data as placing unskilled wages at a historic high when Engels wrote comes from commentary on the paper rather than from Clark's own words, and the paper was not opened. Bibliographic record and Our World in Data's republication of one of Clark's series

7Peter H. Lindert and Jeffrey G. Williamson, "English Workers' Living Standards during the Industrial Revolution: A New Look," Economic History Review 36, no. 1 (February 1983), pp. 1-25, for new real wage estimates covering 1755 to 1851 and for the blue-collar real wage index standing at about 50 in 1819 and 100 in 1851, implying a near doubling in thirty-two years. The index values are quoted from an encyclopedia essay citing the paper rather than from the article itself, which was not opened. Economic History Society issue listing and the Concise Encyclopedia of Economics essay reporting the index values

8Robert C. Allen, "Pessimism Preserved: Real Wages in the British Industrial Revolution," University of Oxford Department of Economics Discussion Paper 314 (2007), for the component-by-component comparison of Feinstein's and Clark's price and real wage indices, for the judgement that some of Clark's innovations are improvements while many of his changes degrade the index, and for the finding that a new index built from the strongest components sits much closer to Feinstein's than to Clark's and contradicts Clark's optimistic view. Not opened. Bibliographic record and a copy hosted in Thomas Piketty's teaching files

9For the Corn Laws running from 1815 to 1846, for the Importation Act 1815 barring foreign wheat until the domestic price reached eighty shillings a quarter, and for repeal by the Importation Act 1846 under Robert Peel. Sources disagree about whether and when the eighty-shilling price was actually reached, and that sub-claim is not relied on here. An encyclopedia account, a second encyclopedia account giving the statute references and the Victorian Web on the 1815 threshold

10For Crafts's method of estimating labour's share by scaling a base-year share by the ratio of real product wages to output per worker, using the GDP deflator estimates of Broadberry and co-authors (2015), which is the basis of the objection that his factor-share result and his product-wage result are not independent of one another. The method description comes from search results quoting the working paper's section on factor shares; the paper was not opened. CAGE working paper and the paper's bibliographic record

11Nicholas Crafts, "Understanding productivity growth in the industrial revolution," The Economic History Review 74, no. 2 (2021), pp. 309-338, for the finding of a gradual acceleration of productivity growth rather than a take-off, and for the argument that slow real wage growth in the face of a surge in population growth may be read as a reasonably good outcome. Not opened. Publisher page and the Warwick repository copy

12For the use of Engels' pause in current argument about artificial intelligence and wages, on both sides: a labour economist's reading that the pause was just that, a pause, and that productivity gains let employers afford higher pay; and the contrary reading that firms retaining gains through cost reduction and margin expansion could shrink labour's share further. These are commentary rather than research and are cited only as evidence of how the episode is being used. CIPD commentary and an investment commentary making the pessimistic case

13David Ricardo's account of rent, for the objection that food kept dear by scarce and protected land transfers the gain to landowners rather than to workers, which relocates rather than removes the distributional question. No single reachable edition is cited here because the point is a standard summary of Ricardo's theory of rent rather than a quotation; a library catalogue record of the Principles stands in its place. Catalogue record for Ricardo, On the Principles of Political Economy and Taxation (1817)