Industrial Revolution English Workers Li

English Workers' Living Standards during the Industrial Revolution: A New Look Author(s): Peter H.
Lindert and Jeffrey G. Williamson Reviewed work(s): Source: The Economic History Review, New Series, Vol. 36, No. 1 (Feb., 1983), pp. 1-25 Published by: Wiley on behalf of the Economic History Society Stable URL: http://www.jstor.org/stable/2598895 . Accessed: 03/03/2013 17:10
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp
.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.
Wiley and Economic History Society are collaborating with JSTOR to digitize, preserve and extend access to The Economic History Review.
http://www.jstor.org
This content downloaded on Sun, 3 Mar 2013 17:10:37 PM All use subject to JSTOR Terms and Conditions
THE
E C N1 anI"S
ECONOMIC
HISTORY
REVIEW
SECOND $ERIES, VOLUME
XXXVI, No.
i,
FEBRUARY
i983
SURVEYS AND SPECULATIONS, XVII
English During
Workers' Living Standards Revolution: the Industrial A New Look*

AND
By PETER H. LINDERT
JEFFREY G. WILLIAMSON
of fresh the IndustrialRevolution'deservesrenewalwith the appearance data or new perspectives. This paper mines an expandingdata base and emergeswith a far clearerpicture of workers'fortunes after I750. While optimistsand pessimists can both draw support from the enterprise, the case emergeswith the greaterneed for redirection and repair.The pessimists' evidence gainswereevenbiggerafterI 820 thanoptimists suggeststhatmaterial hadpreviously well-beingis expanded claimed,even if the conceptof material
* This article is part of a larger research project on 'British Inequality since i670', supported by grants from the US National Science Foundation (SOC76-8o967, SOC79-0936i, SOC79-o6869) and the US National Endowment for the Humanities (RO-26772-78-i9). The authors gratefully acknowledge the able research assistance of George Boyer, Ding-Wei Lee, Linda W. Lindert, Thomas Renaghan, Ricardo Silveira, Kenneth Snowden and Arthur Woolf, as well as the helpful comments of G. N. von Tunzelmann, Stanley L. Engerman, two anonymous referees, and seminar participants at the University of California (Berkeley, Davis, UCLA), Harvard University, Northwestern University and the University of Wisconsin. Readers are referred to the fuller display of evidence in the discussion paper, 'English Workers' Living Standards during the Industrial Revolution: A New Look', September i980, available either from the Department of Economics, University of California, Davis, 956i6 USA (Working Paper Series No. I44) or from the GraduateProgram in Economic History, University of Wisconsin, Madison, 53706 USA. Hereafter this paper is cited as 'DP'.
1 The historical who wanta full bibliography couldbeginwith literature is too vastto cite here. Readers in RealWages,I750-i850', EconomicHistoryReview, 2nd sources citedbelowandin M. W. Flinn, 'Trends series,XXVII, 3 (I974), pp. 395-4I3; A. J. Taylored., The Standard of Living in Britain in the Industrial duringthe Industrial and P. K. O'Brien,'IncomeDistribution Revolution (I975); StanleyL. Engerman Revolution',in R. C. Floud and D. N. McCloskeyeds., The Economic History of Britain since I700 clashis that betweenT. S. Ashton For heatedeloquence,the best twentieth-century (Cambridge, i98i). and the Historians by Historians',in F. A. von Hayek ed. Capitalism ('The Treatmentof Capitalism i968)). (The Making of the English WorkingClass (Harmondsworth, (Chicago, I954)) and E. P. Thompson
T he politically charged debate over workers' living standardsduring
PETER
H. LINDERT
AND JEFFREY
G. WILLIAMSON
factors.Althoughthe pessimistscan still to includehealthand environmental find deplorabletrends in the collectiveenvironmentafter i820, particularly risinginequalityand socialdisorder,this articlesuggeststhat their case must be shiftedto the period I750-i820 to retainits centralrelevance. I Which occupationsand social classes are of the greatestrelevanceto the debate?It seems unlikelythat we would get full agreement from the participants, but therearea few groupswhosefortuneshavebeen of primeconcern, of livingdebateandto the contemporary standard both to the historical debate overThirdWorldgrowthand distribution.2 Following establishedconventionsin the literature,each group listed in Table I refersto adult male employees:the self-employedand permanently unemployedare excluded. Our lowest earningsgroupconsistsof hired farm who representthe bottomtwo-fifthsof all workers.Next come the labourers, and theirnear-substitutes, non-farm commonlabourers a low-skilled"middle efforts and sizable wage gains have group". Artisans, whose organizational caused them to be singled out as the "labourfall roughly aristocracy",3 betweenthe 6oth and 8oth percentilesin the overalldistribution of earnings. "Bluecollar"workersinclude each of these groups,and define"the working class"most closely, at least within the debateover living standards.4 The list is completedby the additionof a diversewhite-collar group. These "class"rankingschangedlittle acrossthe nineteenthcentury,at least betweenI827 and I 85 I. However,since the relativegrowthof groupincomes wasrarelythe sameoverthe centuryfollowingI750, eachwill be documented in the sectionswhich follow. Furthermore, laterin this paperwe shallexplore just how much of the real wage trends for the blue collar labourercan be explainedby shiftsinto higherpaidworkandhow muchby wagegainsamong all blue collarworkers.Table I simplyestablisheswho the workerswere and where they fit in the size distributionof earningsin the early nineteenth century. II Quantitative judgementson workers'living standardshave alwaysbegun with time series on rates of normalor full time pay.5 This was certainlythe
2
On the debateoverthe 'bottom40 percent'in the ThirdWorld,see H. B. Chenery et al. Redistribution
Journal of and Development: A SurveyArticle', with Growth (Oxford,I974); W. R. Cline, 'Distribution DevelopmentStudies, II (I975), pp. 359-400; M. Ahluwalia,'Inequality,Poverty,and Development', Journal of Development Economics,2 (I976), pp. 307-42; andSimonKuznets,Growth,Populationand Income Distribution(New York, I979). 3 See T. S. Ashton, An Economic History of England: The i8th Century (I955), ch. VII; idem., 'The Ix Journal of Economic History, Supplement Standard of Life of the Workersin England, I790-i830', in A. J. Taylored., The Standard of Living; E. J. Hobsbawm, LabouringMen (New as reprinted (I949), York, i964), esp. chs. I5 and i6; and HaroldPerkin, The Origins of ModernEnglish Society, I780-1880
(i969), pp. I3I, I43, 395-7, 4I74 On the changing Asa Briggs,'The Language of nuancesof the term'working class',see in particular
in EarlyNineteenth-Century History "Class" England',in Asa Briggsand J. Savilleeds. Essaysin Labour and R. J. Morris,Class and Class Consciousness in the IndustrialRevolution, I780-i850 (I979). to earningsor 'pay'as thoughthese represented 5We follow past authorsin referring all pre-transfer is valid for Englishworkersbeforethis income, either gross or net of direct taxes. This simplification ownedtheirown homesor significant of other amounts century.Onlya tiny shareof blue-collar employees non-human property,and only a tiny sharepaid any directtaxes. The heavierindirecttaxes- excises, - werereflected in the pricesand rentsworkers paid, which importduties,and the localrateson property below. aremeasured
(i967),
LIVING
STANDARDS
Mean Positions in Table I. Adult-MaleEmployeeClassesand TheirApproximate EarningsRanks for England and Wales theNineteenth-Century
Occupational Class "Representative" Mean-wage Series used here (iL) farm labour
(2L) non-farm common labour
Approximate Mean-wage Percentile Positions in the EarningsRanks

1827 r85i
(i) Farm Labour
i3th 38th
I4th 35th 5oth 5Ist 58th 62nd 77th 63rd 7Ist 7Ist 62nd 8oth 8ist iooth 8oth 79th 70th 94th
(Bottom 40%) (2) Middle Group
(5L) police and guards
(6L) colliers (3) Artisans

("Labouraristocracy")
55th
62nd 67th 77th 74th 75th 78th 65th 87th goth 95th 88th 86th 75th 89th
(5H) cotton spinners (2H) shipbuilding trades (3H) engineering trades (4H) building trades (6H) printing trades
(I)+(2)+(3)
(4) Blue-CollarWorkers =
(5) White-Collar Employees
(3L) messengers and porters (4L) other government low-wage (iH) government high-wage (7H) clergy (8H) solicitors and barristers (9H) clerks (ioH) surgeons and doctors (i iH) schoolmasters (U2H) engineers, surveyors and other professionals
(6) All Workers = (4)+(5) Notes and Sources:The sources for the group earnings averages are discussed in Section II below, and at greaterlength, in Jeffrey G. Williamson, 'The Structure of Pay in Britain, I7Io-I9I9', in P. Uselding ed. Research in EconomicHistory, 7 (i982). The overall earnings distributions for i827 and i 85I on which these groupmeans are ranked are reported in Jeffrey G. Williamson, 'Earnings Inequality in Nineteenth-Century Britain',Journal of EconomicHistory, XL (I980), pp. 457-75. These size distributions refer to employee earnings only, excluding incomes from property, selfemployment, pensions or poor relief.
by Bowleyand Wood, Gilboy, starting point for the pioneeringcontributions Phelps Brown and Hopkins, and others. We also begin in the same way, addingseveralnew pay seriesalongthe way. An essentialfirst step is to select appropriate annualpay rates. Most pay seriesare constructedfrom daily or weekly rates, and we still have only the the average numberof daysor weeksworked sketchiest evidencedocumenting per year. It seems sensible to exploit the normalor full-timepay rates first, trends and then turn to clues about unemploymentor underemployment (SectionV) to infer movementsin true annualearnings.Daily and weekly up to a 52-weekyear, usingvariousestimates normal pay ratesare aggregated of normaldays per week in differentoccupations.6These annualearnings
6 The choice of numbersof weeks per yearis arbitrary and matterslittle to what follows. ArthurL. in the United Kingdom in the Bowleythoughtthat six weeks was the average'lost time' per year (Wages onlyif the numberof Nineteenth Century (Cambridge, i900), p. 68). The choiceof weeksper yearmatters in trueinvoluntary andnot justdue unemployment weeks'lost'variedgreatlyovertime,due to movements valuingtheirtimeaboutthe samein andout of work.We to marginal shiftsin employment ratesby persons doubt that the work year shifted in ways alteringthe conclusionsof this paper, to judge from the exploration of trendsin normal unemployment evidencein Sectionv belowand from M. A. Bienefeld's chs. 2, 3. annual industrial hours:Working Hours in British Industry (I972),
PETER H. LINDERT AND JEFFREY G. WILLIAMSON 4 figuresgenerallyexcludepaymentsin kind, but this rule is violatedfor farm labourers,whoselargein-kindpaymentshave been included. in Table2. These seriesreflect Eighteennominalpay seriesaredocumented literature on wagerates. a numberof additionsandrevisionsto the time-series additions,thoughnot the mostcrucialto the conclusions The mostconspicuous below, are the service occupations(Series 3L, 4L, iH, and 7H to I2H inclusive). With the exception of clergy and teachers, our view of serviceoccupationpay leans heavily on the public salary figures reportedin the AccountsandPapers "AnnualEstimates" (printedin the Houseof Commons'
Table
2.
I755-i85I:
Occupation
(iL)
Estimates of Nominal Annual Earnings for Eighteen Occupations, L's) AdultMales,England and Wales (in current
I755
I7-I8 20-75
I78i
2I.09 23-I3
I797
30-03 25-09
i8o5
40-40
i8io
42-04
i8I5
40-04
i8i9
39-05 4I-74
i827
3I.04
i835
30-03 39-29
i85I
29-04
farm labourers
(2L) non-farm common labour (3L) messengers & porters (4L) other government low-wage (5L) police & guards (6L) colliers
36-87 69-43
52-48
33-99 28-62
25-76
33-54 46-02 48-08 24-37

I04-55
57-66 46-77
47-04
43-94 76-0i 57-47 67-89 6g-22 I76-86

55-25
43-94 80.69 60-22 69-34 57-82 I95-I6

59-20 94-9I
43-65 84-39
59-01
8I-35 6o.60 69.I8 50-37 2I9-25

57-23 92-7I
87-20
58-70
44-83 88-88 66-45 53-62 55-44 234-87 64-I2 84-05 66-35 58-64
74-72
22-94 78-9I 38-82 43-60

30-5I
47-79
I33-73
5I-26 64-99 I5I-09

5I-32
62-95 54.6I 222-95 62-22 80.69 66-35 58-5o

70-23
63.33 56-4I 270-42 62-74 77-26

59-72
(IH) government high-wage (2H) shipbuilding trades

(3H) engineering trades (4H) building trades (5H) cotton spinners (6H) printing trades (7H) clergy (8H) solicitors and barristers (9H) clerks
45-26 50-83 35-57

4I-93 54-03
5I-7I 58-o8 40-64

47-90
75-88
55-30
88-23 66-35
78-2I 79-22
35-96 46-34 9I-90 23I-00 63-62 62-02 I5-97

I37-5I
65-I8
7I-II
66-35 67.60
79-22
63-02 67-60
7I-I4
64-56
70-23
66-6I 238-50 I65-00 I35-26 I74-95

43-2I
I82-65 242-67 IOI-57 88-35 I6-53

I70-00
266-42 340-00 I50-44 2I7.60

43-2I
283-89 447-50 I78-II 2I7.60

5I-I0
272-53 447-50 200-79 2I7.60

5I.I0
266-55 447-50 229-64 2I7.60 69-35 326-43
254-60 522-50 240-29 I75-20 69-35 265-7I
258-76 II66-67 269.II 200-92 8I.89 398-89
267-09 I837-50 235-8I 200.92 8I.Ii

479-00
(ioH) surgeons & doctors

(iiH) schoolmasters
(I2H) engineers & surveyors
I90-00
29I-43
305-00
337-50
Sources and Notes: From Williamson, 'The Structure of Pay', Appendix Table 4. Some of these occupations need no elaboration. Those that do are explained as follows: (4L) - watchmen, guards, porters, messengers, Post Office letter carriers, janitors; (iH) - clerks, Post Office sorters, warehousemen, tax collectors, tax surveyors, solicitors, clergymen, surgeons, medical officers, architects, engineers; (2H) shipwrights; (3H) - fitters, turners, iron-moulders; (4H) - bricklayers, masons, carpenters, plasterers; (6H)
- compositors.
from I797 onwards).This is a rich sourcefor consistenttime series on welldefinedoccupations.Annualearningsarereportedtherefor largenumbersof category,spanningthe whole earningsdistriemployeesin each occupational group.The key butionover age, tenure, and skill withina given occupational issue underlyingtheir use is whethertrendsin public "civil service"salaries replicatedtrends in the same privatesector occupations.Elsewherewe have offeredevidenceconfirmingthe correlation,at least for the nineteenthcentury.7
I78I.
Service-sector pay, again for public posts, is also availablefor I755 and For the latteryearwe havefiguresreportedto the House of Commons.8
of Payin Britain,I7I0-I9I I'. 'The Structure 7See JeffreyG. Williamson, Appointedto Examine,Take, and Statethe Public Accountsof the United Reportof 'Commission Papers,I782 and I786. Kingdom',Houseof Commons
8
LIVING
STANDARDS
estimates supply figures for I755, though for fewer John Chamberlayne's employees and departments than is true at the later dates.9 These eighteenth-century publicpaydatamust, of course,be treatedwith care,since in the upperechelons. 10 a trulybaroquepaymentssystemprevailed For clergy and schoolmasters,we have made use of private pay series. meanannualearnings(includingthe rentalvalueof the vicarage) Clergymen's can be estimatedfor the greaterpart of the nineteenthcenturyby using The ClericalGuide and Ecclesiastical and The ClergyList. A random Directory sampleof 550 clergymen, from all patronagesources (royal, ecclesiastical, and private),yields theirpay for I827, I835, and I85I. For earlier university, benchmark yearswe had to use public pay ratesfor clergy, splicingthese on to the privateseriesat i827. This procedureseems to have yieldedplausible backto I755, judgingby the similarity payseriesfor the averageclergyman in trendbetweenpublic and privateclergysalaries from i827 on. had low monthlycash earnings,both becausemuch of their Schoolmasters incomewas in kind (rents and fuel), and becausethey often receivedsupplefees and holidaybonuses.We haveassumedthat incomein kind was mentary a stableshareof total income, so that the twelve-month cash-income seriesin Table2 accurately reflectstrendsin totalincome.For I755-I835, ourestimates relyon schoolmasters' earningsin severalCharitySchoolsin Staffordshire and The I85I figurerefersto civilianschoolmasters Warwickshire. in public pay, in the AnnualEstimates.11 as reported Fornineteenth-century andbuildingcrafts(Series2H to 6H manufacturing inclusive), the well-known estimates of Bowley and Wood suffice.12The available series on eighteenth-century artisans'pay referonly to the building crafts, but Gilboy's data on these crafts offer the advantageof regional diversity. To give proper weight to the well-known regional variancein nominalwages, and to the shift in eighteenth-century populations,we have constructed an earningsaverage for buildingcraftsmen thatreflectschangesin "regionalmix" between I755 and I797.13 The result is a steeper rise in
9 John Chamberlayne, Magnae Britanniae Notitia, Or the PresentState of GreatBritain, I7th ed.
(I 755).
date back to the i68os. We are indebtedto David Earlier editions, begun by Edwin Chamberlayne, almanacs. foralerting us to the Chamberlayne Galenson system.Oftenextremely 10Forexample, headsandhightitledclerkswerepartof a patronage department his staffof clerks.We hadto maintain out of whichthe recipient weregrosssalaries salaries highreported haveignoredthe payof all officialsfor whichthis seemedto be the practice. a portionof officials,for example,received Customs earnings. surelyunderstated In othercases,salaries the taxescollectedin additionto the reportedincomes.These were excludedfrom our estimates.Also andheraldic wereofficialsfor whomthe statedstipendswerebut partialpoliticalside-payments excluded For example,in an earlieredition of Magnae Britanniae Notitia (i694 ed. p. 238), Edwin perquisites. Herald's listed the Lancaster pay as only ?26 I3S. 4d. per annum.The Heraldin this case Chamberlayne was GregoryKing. Were this his only income, King would have been no betterpaid than a common a messenger or a porter. seaman, serieson to otheravailable 11 For a fullerdiscussionof all schoolmaster pay series, with comparisons 'The Structure of Pay'. benchmark dates,see Williamson, in the 12 Bowley,Wages articlesthat appeared and the seriesof Bowley-Wood in the UnitedKingdom,
Journal of the Royal Statistical Society between i898 and i906.
13 For the Londonareawe used an unweighted Greenwich averageof wage seriesfromWestminster, and Maidstone.The London area series is then combinedwith series from six Hospital,Southwark Thesecounty theNorthRiding,andLancashire. Devon, Somerset, Gloucestershire, counties: Oxfordshire, in PhyllisDeaneandW. A. weightsreported population usingthe regional estimates arecombined earnings 2nd ed. i969), Table24, p. I03. Cole,British EconomicGrowth, i688-i959 (Cambridge,
6 PETER H. LINDERT AND JEFFREY G. WILLIAMSON earningsin the buildingtradesup to I797 than that reportedby Brownand Hopkins, whose seriesreferredto southernEnglandonly.14 remain:colliers,non-farmcommon Threevery largeunskilledoccupations labourers,and farm labourers.The colliers'earningsfiguresrefer to undergroundminingby adultmales. The I85I figureis derivedfromWood'swage series.15The i835 figure is from Bowley, as are the i8i0-i9 estimates,the latter referringto southernScotland.We have also used figuresfor I755 to to colliers' i805 inclusive,and (again)i835 fromAshtonand Sykes, referring and in Derbyshire.16 dailywageratesin the northerncounties,in Lancashire into Thesediverseestimatesarelinkedtogetherat variousdatesandconverted sketchedabove.Non-farmcommon ratesusingthe procedures annualearnings earningsare based on two sources. For the period I797-I85I, we labourers' have accepted the Phelps Brown-Hopkinsestimates for labourersin the buildingtrades.For I755-97, their estimateshave been set asidein favourof series based on Gilboy data for building labour, using the a multi-regional sameproceduredescribedfor buildingcraftsabove. For farmlabourers,the figuresare based on Bowley'swages for a "normalwork week", I797-I85I variation(but takingaccountof both income in kind and seasonalwage-rate 17 Fifty-two"normal weeks"arearbitrarnot seasonalemploymentvariation). ily assumedin constructingan annualfull-time series. The I78I figurealso of the figuresfor Surrey, average relieson Bowley,but hereit is an unweighted Kent, Hertfordshire,Suffolk, Cumberland,and Monmouth,spliced to the national series at I797. The I755-8i estimates are constructedfrom raw earningsdatacollectedby Rogers.18 earnings. our best interimview of trendsin occupational Table2 represents It is confined to ten benchmarkyears simply because the data are more for these years. abundant Table 3 reportsaveragefull-timeearningsfor the six groupsidentifiedin overour eighteen SectionI. The employmentweightsused in the aggregation
14 XXII of BuildingWages',Economica, E. H. PhelpsBrownand SheilaV. Hopkins,'SevenCenturies (I955),
pp. 205-6.
15As reproduced in Brian Mitchell and Phyllis Deane, Abstractof BritishHistoricalStatistics(Cambridge,

I97I).
16 Bowley,Wages; T. S. Ashtonand J. Sykes, The Coal Industryin the EighteenthCentury(Manchester, I929). 17 A. L. Bowley,'The Statistics Journal of Wages', of Wagesin the UnitedKingom.Parti. Agricultural
the Royal StatisticalSociety, LXI (i898).
for Kent, Essex, Dorset,Nottinghamfor the periodI790-i840 arealso available estimates Alternative of LivingControversy, 'The Standard and Suffolkin T. L. Richardson, Hampshire, shire, Lincolnshire, nominaldailywagesfor of Hull, I977), Pt. ii. Richardson's (unpub.Ph.D. thesis, University I790-i840' the I790s thando Bowley's showsomewhat lesssteeprisesacross daylabourers' agricultural 'fullyemployed The discrepancy may reflectthe morerapidrisein wagesin the north,an areagivenits averages. national averages conformrather In any case, the Bowleyand Richardson due morefully in the Bowleyaverages.
closely between i8o5 and i840.
18 JamesE. ThoroldRogers,A History of Agriculture and Prices in England, vii (I902). Onceagain,we between for Englandand Walesthat reflectsshiftsin population average have tried to build an earnings by the paucityof century,this timefor farmlabour.The taskis complicated regionsacrossthe eighteenth is hardlydefinitive.Ouraverage wagefor southernEnglandis average time seriesdata,and the resulting and Gloucester, the only two countiesfor which Rogerssupplies a weightedaverageof Cambridgeshire The northis represented only by Brandsby, continuous dailywage seriesfor adultmale farmlabourers. averages forthislocation is large.The northern andsouthern of observations thoughthe number Yorkshire, from Deane and Cole, wherethe 'south'is definedby the twenty estimates wereweightedby population andNorfolk.The 'north' Cambridge, Oxford,Northampton, countiesincluding,or southof, Gloucester, dailywageis then linked and the threeRidings.The resultingaverage in this case consistsof Lancashire estimate. with the I78i annualearnings
LIVING
STANDARDS
occupationsare very rough. Those for I8II and earlierare based on work published,while those for later yearsare basedon manipulations previously 19Table3 revealsthe earnings earlycensusdataon occupation. of the imperfect by differentclassesof workers.The varietyis striking.In historyexperienced commonlabourers thelatterhalfof the eighteenthcentury,farmandnon-farm especially.From workers,the labouraristocracy gainedgroundon higher-paid i8I5 to the middle of the nineteenthcentury, on the other hand, the gap Farm wages betweenhigher-and lower-paidworkerswideneddramatically. saggedbelow, while white-collarpay soaredabove, the wages for all other ourresultswith earlier seriesthathaveshaped Table3 alsocompares groups.20 past impressionsof wage trends and played a key role in Flinn's recent andcontrast.Where The new andold seriesexhibitboth conformity survey.21 they diverge, we stand by the new series as improvements,and urge other The major scholarsto harvest additionalwage series from the archives.22
Table 3. Trends in Nominal Full-Time Earnings for Six Labour Groups, Comparedwith ThreePrevious Series, I755-i85i
(i85I
(I) (2)
= Ioo) (4) (5) (6)
vs. Year Farm Labourers
Bowley's Farm Labourers
Middle Group
(3) Phelps Brownvs. vs. Hopkins Labour Building Labourers Aristocracy
Tucker's London Artisans
All Blue Collar
White Collar 2I.62 26.42
All Workers
I755
I78i I797
59.i6
72-62 I03-4I I39-I2 I44.76 I37-88 I34-47 I06.89 I03-4I I00-00 ioo-8 II2-3 I00.0
42.95
48.5
50.86
69.8
5I.05
75-5
93-9
54.88
72.92
i805
i8io i8I5 I8I9 I827 I835 i85i 52 weeks' earnings in I85I:
98-89
IIO-95 I05-55 99-4I 98-89 96-98 I00-00
57-6 66.7 83-3

97-0 97-0
57.38 64.86 79-44

92-03 95.28
69.8 8i.o 87.0

I05-6 II2-I I03-3 I05-I 98-9 100.0
59.64
74-42
38.62 46.62
32-55
96.58
I07.8i io6.i8 IOI.84 97-59 94-II 100-00
38.88
43.0I 46-55 50-77 55-09 75-03 I00-00
58.97 75-87
84.89 85.30 84.37 83.II 88.77 I00-00
9I 92 97.0 97-0 93-55 88-68 97-0 00.0 100-00
?29-04
?29-04
?52-95
?42-90
?75-I5
n.a.
?52-62
?258-88
?75-5I
Sourcesand Notes: The indices are aggregated from the finer groups listed in Table i, using wage series from Table 2 and employment weights. The employment weights for I755-I8 I5 draw on Lindert, 'English Occupations, I670-I8 i i', Table 3, while those for i8i5-i85i are derived from censuses. The derivations of the employment weights are described in DP, Appendix A. For the three previous series, see Bowley, Wagesin the UnitedKingdom,table in back; Phelps Brown and Hopkins, 'Seven Centuries', Rufus S. Tucker, 'Real Wages of Artisans in London, I729-I935', Journal of the AmericanStatistical Association, 3I (1936), pp. 73-84. The conversion of the Phelps Brown-Hopkins series from daily to annual wages assumed 3I2 working days a year.
19 The occupational to contemporary numbersfor i8 Ii and earlierare estimated,with comparisons i670-i8 i I', Journalof estimates by Massieand Colquhoun,in PeterH. Lindert,'EnglishOccupations,
EconomicHistory, XL (i980), pp. 685-7I2.

201 For moredetailson these distributional of Pay', andidem., 'The Structure changes,see Williamson, J. Econ. Hist., XL (i980), pp. 457-76. in NineteenthCentury Britain', 'Earnings Inequality 21 M. W. Flinn, 'Trends in RealWages'. 22 Bernard craftsmen, building wageratesforbuilding hasassembled a newserieson Midlands Eccleston ('A Surveyof WageRatesin Five MidlandCounties,I750labourers, estateworkersand roadlabourers with our findingsis Table 3, for of Leicester,I976). Consistent I834', unpub.Ph.D. thesis, University seriesrisingtoo slowlybetweenI755 and Ecclestonfinds the PhelpsBrown-Hopkins commonlabourers i 8 I 5, andagrees aswell. Eccleston seriesmissesthe slightpostwar deflation thatthe PhelpsBrown-Hopkins
PETER H. LINDERT AND JEFFREY G. WILLIAMSON 8 on, our choice by, but not conditional conclusionsof this paperarereinforced of these new nominalpay series.23
III indicesto deflate Severalscholarshaveattemptedto constructcost-of-living particular attracted has to i850 I790 such nominalearningseries. The period Gayeroffered by are those cited often most attention.The four price indices pioneerThese Tucker.24 and Rousseaux, (GRS),Silberling, Rostow-Schwartz ing effortscan be criticizedon threefronts:(I) the underlyingprice data;(2) the commoditiesincluded in the overallindex; and (3) the budget weights appliedto eachcommodityprice series. pricesareused by GRS, Silberling,andRousseaux.GRS,in fact, Wholesale who, in manycases, chose not to used wholesalepricescollectedby Silberling use them. Rousseauxalso borrowedfrom Silberling,as well as from Jevons lists "issuedby chief sourcewas the PriceCurrent and Sauerbeck.Silberling's Tucker's men".25 of business for the use London severalprivateagenciesin institutions: London -three by chief sources were the contractprices paid Greenwich,Chelsea, and Bethlem Hospitals. Other writers have criticized these series for relying on wholesaleand institutionalLondon prices, rather than on retail prices actuallypaid by workers'familiesacrossEnglandand however,wholesalepricesarea fairproxyfor As Flinn has argued,27 Wales.26 pricesover the verylong term. In most cases, thereis no alternative consumer anyway.An exceptionis clothing,for whichwe haveused a GRScotton-textile Londonprices, leadingto exportprice seriesinsteadof Tucker'sinstitutional trend between I790 and of cost-of-living the a slightly more optimisticview The commoditiesincluded in the cost-of-livingindex also need revision. potatoes. especially commodities, We haveaddedmorerelevantworking-class have in the GRS series, included raw materials, Some irrelevantindustrial rent. of house is the addition most change important been removed.But the partof the cost-of-living,28 Whilethe classicindicesall omittedthis important
or the present betweenI755and i8 I5 thanPhelpsBrown-Hopkins forcraftsmen findsfasterwageadvances resultsserveto whichalso showfasternominalgainsthanTucker'ssluggishseries.Eccleston's estimates, nominal wagegainswereconsiderably by pastwriters: suggested contrastalready a geographic emphasize about andnorththanin Londonandthe south,at leastup to i8I5. Pastimpressions in themidlands greater nominalwage gains by relyingtoo the late eighteenthcenturyand the war yearshave underestimated series. heavilyon southern of the points made in this section, see DP, Section4 and AppendicesB 23 For fullerdocumentation and C. 24 A. Gayer, W. W. Rostow,and A. J. Schwartz,The Growth and Fluctuationsof the British Economy, I790-1850 (Oxford,I953); N. J. Silberling,'BritishPricesand BusinessCycles, I779-i850', Review of
de fond de l'jconomieanglaise Economicsand Statistics, 5 (I923), pp. 223-6i; P. Rousseaux, Les mouvements
i850.
in London'. (Louvain,I938);andTucker,'RealWagesof Artisans 25 Silberling, British Prices, p. 224. 26 T. S. Ashton, 'The Standard of Life of the Workersin England',p. 48; Deane and Cole, British Flinn, 'Trendsin RealWages',p. 402. or morerecentperiods: slightlyshorter covering forpartsof England Houserentshavebeenmeasured Econ. duringthe NineteenthCentury', of Livingin the BlackCountry see G. J. Barnsby,'The Standard
27
EconomicGrowth, p. I3.
28
Hist. Rev., 2nd ser. XXIV (I97I), pp. 220-39; and R. S. Neale, 'The Standard of Living, I780-i844: A andClassStudy',Econ. Hist. Rev. 2nd ser. xix (i966), p. 6o6, givingrentsfor Bath, i8I2-i844. Regional
LIVING
STANDARDS
ours includes a rent series based on a few dozen cottages in Trentham, While the database is narrow,it Staffordshire (just outsideStoke-on-Trent). The rent series doesapplyto a housingstock of almostunchanging quality.29 implies that the cost of housing (at a fixed location)rose relativeto other the Industrial Revolution,thus offeringsomenew consumer itemsthroughout to the pessimists.30 support indexshoulduse commodity weightswhichreflect Finally,the cost-of-living All workers' budgetsshares.Past seriesdo not fully satisfythis requirement. inputs, andothersare exclude anyweightfor housing,someincludeindustrial simplyvague about their weights. One set of workers'householdbudgets stemsfromthe pioneeringwork of Davies and Eden on the ruralpoor in the Anotheris a miscellaneous groupof urbanworkers' lateeighteenth century.31 The urban from the late and nineteenth centuries.32 eighteenth early budgets workers' budgetsreveala lowersharespent on food, and a highersharespent on housing,thando the ruralpoor studiedby Daviesand Eden.
29 We wereable to hold the qualityof the cottages virtually constantby (a) splicingtogethersubseries Trentham thatfollowed fixedsets of cottagesand (b) conducting hedonicrentregression tests on detailed the impactof cottagequalitiesand cottagesurveysof i835, i842, and i849. The regressions quantified of the tenantson the rent charged.It turnedout that the rentsfetchedby the best and worst attributes cottages differed verylittlefor giventypesof tenants.See DP, Appendix C. items. Perhaps the It has not been possibleto pursuethe issue of qualityvariation for otherconsumer of clothingand beddingrose, and perhapsthe qualityof meat declined,in waysnot revealedby quality of such possiblequalitydriftsis doubtfulgiven whatwe know about relevance prices.The quantitative households. expenditures amongworkers' If, forexample,the qualityof meatfell by halfbetweenI780 and of meat i850, thehidden extracostto workers wouldstillbe only * 50 X *I I I = 5 -5% since* I I I is the share in the average expenditures budget(DP, AppendixB). The truenet driftin qualitywasalmostsurelyfar lessthanthis. 30 The importance of addingrents, andof replacing institutional priceswithmarketpricesfor clothing, urban" canbe seenfromthe followingcalculations, using"southern budgetweights(see Table4 and DP, Section 4 andAppendix B):
Cost of Living
Percentagechangeover the period

I790-i8I2 96-4 i812-1850 I790-1850
WithTucker's institutional - 56-o -I3.0 clothingprices,andwithoutrents 8i 0 -62.3 -3I.7 Withexportpriceof clothing,withoutrents -57.6 -20.6 Withexportpriceof clothing,with rents(Table4, "BestGuess") 87.2 revisionsis to tip the trendtowardoptimism(towards declining The net effectof the two cost-of-living increase livingcosts) but the inclusionof rents by itself adds eleven per cent to the net cost-of-living between I790 and i85o. Readersshould be warned, however, that the small Trenthamsamplemay give too pessimistican impression abouttrends.Acrossthe nineteenthcenturythe Trentham serieshas the sametrendsas two of Living',p. 236; and H. W. Singer,'An Indexof UrbanLandRents urban series(Barnsby,'Standard andWales,i845-I9I3', Econometrica,9 (I941), p. 230). If ruralcottagerents andHouseRentsin England seriesoverstates the rise in a national rosemoreslowlyacrossthe nineteenth century,then the Trentham fromlow- to high-rent rentindex usingfixedlocational average residential weights.(As for the migration seriesrisesmuch locations, see SectionVI below).Fromaboutthe I770S to aboutthe i840s, the Trentham 'Standard of Living',pp. 245-8; andSirJamesCaird, faster thantwo otherruralseries(T. L. Richardson, in trendis so greatas to imply in I85o-I85i (New York, i967), p. 474). The difference EnglishAgriculture weretakenas a national anunreasonably average rapidrise in urbanrentsif Trentham (rural-and-urban) Revolution eraandthe nineteenth seriesrosefaster index.So forboththe Industrial century,theTrentham rents. thanthe mostlikelytrendsin national residential 31 Rev. DavidDavies, The Case of Labourers in Husbandry(Bath,I795) andSirFrederick MortonEden, TheState of the Poor (I797), II and III. PhelpsBrownand Hopkinsalsoused budgetweightsfromEden, withouthouserents(PhelpsBrownandHopkins,'SevenCenturies', though pp. 296-3I4). 32 Five urban budgets for I795-i845 are presentedby J. Burnett,A History of the Cost of Living householdbudgetfor Bathin i83I. (Harmondsworth, i969). Neale (Bath, pp. 597-9) gives a labourer's in London',p. 75) ventured two non-farm Tucker household budgetsas averages ('RealWagesof Artisans of someunderlying studies'budgets.
IO
PETER
H. LINDERT
AND JEFFREY
G. WILLIAMSON
set of budget weightscould mattera great Choosingthe most appropriate deal. Goods and servicesare consumedin differentproportions by northern and southernhouseholds,by the ruraland urban, or by the poor and rich. Cost of living trendscould differ acrossclassessimplybecauseof differences in budget weights, as happenedoften in Americanexperience.33 This possibilitywaspursuedwith fourseparate cost-of-living indicesusingweightsfrom the rural north, rural south, urban north and urban south. As it happens, pricesmovedin such a waythatthe choiceof weightsmattered verylittle. The reasonis that the net rise in the priceof food relativeto manufactures, which would have impoverishedthe rural poor more than the better-paidurban workers, was offset by the equally impressiverelativerise in house rents, which took a greatertoll on urbanhouseholds.The analysisbelow continues to use southernurbanweights, but we now know that the choicemakeslittle difference. cost of livingindexis displayed The resulting"best-guess" in Table4. From I788/92 to I820/26, our index falls midwaybetween optimists(GRS, Rousseaux, Silberling) and pessimists (Phelps Brown and Hopkins, Tucker). Between i820/26 and I846/50, our index is more optimistic, showing a somewhatbigger drop in living costs than any of the past indices.34For the century as a whole, the "best-guess"index supports the middle ground betweenthe optimistand pessimistextremes. IV Deflating the nominal full-time wage series from Table 3 by the cost of living index in Table 4 yields the real wage trends in Table 5 and Figure I below.The resultssupportMichaelFlinn'sconclusion that"therearerelatively few indicationsof significantchangein levels of realwageseitherway before I8Io/I4".35 For lateryears,however,Table 5 offerssomerevisions.Flinn was
Pricesand UrbanInequalitysince i820', 7. Econ. Hist., xxxvi JeffreyG. Williamson,'American pp. 303-33. See also Jeffrey G. Williamsonand Peter H. Lindert, American Inequality; A History (New York, i980), ch. 5. Macroeconomic 34To wit: Per cent changein prices
33
(I976),
I788-92
i809-iS
i820-26
to
I809-IS
to
I820-26 -3I.2 -24.5
to
I846-50 -I6-7 - I0-0 - i6.4 - I9-4 -I0-5
-26-0
Silberling Tucker
Rousseaux
74. I
85.2
-34.8
65-7
Gayer-Rostow-Schwartz PhelpsBrown-Hopkins
Table 4, 'Best Guess'
84.6
72-5
-30-7 -23.5 -27.3
(See Flinn, 'Trendsin RealWages',p. 404.) 35 Ibid., p. 408. There would be clearersigns of deterioration betweenabout i8oo and i820 if the as couldbe wereaddedto the overallaverages, cottonworkers of weavers andothernon-spinning earnings wageratesfor all cottonworkers deflator,the Bowley-Wood done from i8o6 on. Using our "best-guess"
(Mitchell and Deane, Abstract, pp. 348-9) yield the following real wage indices: i8o6, 78-62; i8io, 66.57; i8I5, 75.43; i8I9, 54.67; i827, 66-96; i835, 78.62; and i85I, ioo-oo. Compared with blue collar earnings fell sharply fromi 8o6to i8i9, but keptpacethereafter. of all cottonworkers in Table5, theserealearnings
weaversmay not have sufferedany furthernet lossesafter i820. Bowley's Even the famoushandloom in the Manchester in theNineteenth Century, opp. p. weavers area(Wages dataon pieceratesfor handloom
LIVING
STANDARDS
II
into a period of of all real wage improvements struckby the concentration only a dozen years of deflationbeginning around I8I3. Table 5 does not conformwith Flinn's view.36 There was general real wage improvement
Table 4. A "Best-Guess"Cost-of-LivingIndex, I 78I-1850, Using SouthernUrbanExpenditureWeights
(I850
=
IOO)
COL
Year
I78I
I782 I783 I784 I785 I786 I787 I788 I789 I790 I79I I792 I793 I794 I795 I796 I797 I798 I799 I8oo I8oI I802 I803 I804
COL
Year
I805
i8o6 I807 i8o8 I809 i8io i8ii I8I2 I8I3 I8I4 I8I5 i8i6 I8I7 I8I8 I8I9 I820 I82I I822 I823 I824 I825 I826 I827
COL
Year
I828
I829 I830 I83I I832 I833 I834 I835 I836 I837 I838 I839 I840 I84I I842 I843 I844 I845 I846 I847 I848 I849 I850
Index
ii8.8
I9-3 I2I-9 II8-4 II2-3 I09-6 II2.5 II5-9 I22-3 I25-9 2-2 II8-3 I27-3 I30-7 I53-8 I59-5 I38-8 I36-9 I55-7 207-I 2I8-2 I60.9 I56-8 I60-2
Index
I86-7
I78.5 I69.I I80-5 204-9 2I5-4 204-5 235-7 230-0 203-3 I82.6 I92-I I97.5 I92-4 I82-9 I70-I I55.5 I39.8 I46-0 I54-6 I62-3 I44-4 I40-9
Index
I43-2
I43-9 I4I-3 I4I-3 I33-9 I24-7 II7.6 II2.8 I26-4 I29-2 I38-3 I42-3 I38-4 I33-3 I23-4 I09-6 II4.5 12-0 II6-4 I38-0 II0.9 IOI-2 I000
Source: DP, AppendixB.
between i8io and i8I5, and a decline between i8I5 and i8i9,
after which
therewas continuousgrowth. After prolongedwage stagnation,real wages, by the evidencepresentedhere, nearlydoubledbetween I820 and measured hadannounced.37 increasethaneven past"optimists" I850. This is a farlarger
by i832. i19) implya realwagegainof I5-3 percentbetweeni8i9 and i846, withmostof the gainachieved Allcottonweavers,handloom plus powerloom, gainedan apparent 58 o percentfromi8i9 to i850. Even havea pessimistic trendbias.Theyfailto reflect of weavers risingproductivity thepiecerateseriesprobably weavers whosepayseemedto plummetbefore of givenageand sex, and the dwindling groupof handloom by womenand children,as adultmalesfled to betteri820 appears to have been increasingly dominated Weavers (Cambridge, paying trades (DuncanBythell,TheHandloom i969), pp. 50-I, 6o-i). Onwomenand children's earning,see SectionVII below. 36 Flinn'sdatingof the real-wage 'Trends by G. N. von Tunzelmann, upturnhas also been questioned in RealWages,I750-i850, Revisited',Econ. Hist. Rev., 2nd ser. XXXII (I979), pp. 33-49,esp. p. 48. 37 The closestapproach centuryis the guarded to the presentfindingfor the firsthalfof the nineteenth by] about25 percentbetweeni 8ooand i824 and conjecture by DeaneandColethat"realwages[improved
over 40 per cent between i824 and i850" (BritishEconomicGrowth, pp. 26-7.)
whetherthe apparent of an earlier upturnafterI 820 is not dependent Somereaders drafthavewondered on our use of the i827 and i835 benchmarks insteadof nearbyyears. Some preferto follow pessimist of i842-3, whileotherschoosethe peak-price tradition the depression yeari839. Yeteventhese by stressing
I2
PETER
H.
LINDERT
AND JEFFREY
G. WILLIAMSON
Figure I. AdultMale AverageFull-Time EarningsforSelectedGroupsof Workers, I 755-i85i, at Constant Prices

?30()?200()
White-collar Employees
?258-88
Artisans
? 61 94
-
?7515
?5295
?50?4()0 ?30 ?25. 17 ?20 -
42)30( -
Middle Group ? 29-04 Farm Labourcrs
19 01-1
I
1755
~~~~~I I
1781
!1--11
'27
-I
1835
l
1851
1797 1805 10'15 19
It is also large enough to resolve most of the debate over whether real wages improved during the Industrial Revolution. Unless new errors are discovered or a host of new declining wage series added, it seems reasonable to conclude that the average worker was much better off in any decade from the I830s on than in any decade before I820. The same is true of any class of worker in Table 5. Why has this announcement not been made before? One might have expected it from any of several devout optimists. The answer lies partly in the steady accumulation of data. Yet past findings have also been muted by the belief that trends in real full-time earnings of adult males failed to measure trends in workers' true "living standards". Each time a recent writer has come close to announcing the pOSt-I820 improvement, the report has been disarmed by a confession of ignorance regarding trends in unemployment and in "qualitative" dimensions to life: perhaps health became poorer, work discipline more harsh and degrading, housing more crowded, and social injustice more outrageous, and perhaps these more than cancelled any improvement workers might have gained from rising real wages. These important issues dominate the remainder of this paper. V Time and again the unemployment issue has brought discussion of trends
extreme choices do not remove the post-I820 gains, as evident from these available real-wage data and unemployment estimates: 1819 1839 1843 1835 i8Si 80-04 I00 00 Real wage index, farm labourers: 73.52 9i.67 I03.9I 68. I7 Real wage index, middle group: 54 35 85 97 I00.00 88o50 Estimated EMS unemployment n.a. 3.9% rate (Section VI below): 3.9% 2.2% Io.0% In the real-wage trough year I839 fewer workers were denied income by unemployment. The depression year I843 was a time of high real wages, thanks to cheap provisions. Neither of these extreme benchmarks looks as bad as i8i9.
LIVING
STANDARDS
I3
in workers' living standards to a halt: lacking national unemployment data before I85I, how can the real wage series be trusted as indicators of annual Table 5. Trendsin Real Adult-maleFull-Time Earningsfor Selected Groupsof
Workers, I755-i85i
Artisans
56.29
48.30
Benchmark Year
I755
I78I
Farm labourers
65-46
6i-I2
Middle Group
47-54
46.19
All Blue Collar

56.50
50-I9
White Collar
23.93
22-24
All Workers
42-74
39.24
I797 I805
I8Io
74-50
74-5I 67-2I
52.54 52.96
5I.54
46-73
42.55 42-73 52.I8
53.6I
5I-73 50-04
23.45
20-82 I9.97
42.48 40-64
39.4I
I8I5 18I9 I827
75-5I 73-52
I835
I85I
75.86 9I.67
I00-00 I78I-I
57.8I 54-35 70-I8
50-26
85.97
II00-o
66.39 78-62
100.00
58.I5 55.68 69.25
25.49 27.76 39-10
46.7I 46.I3 58.99
83.43
IQOQO
66-52
100-00
78.69
I00-00
Percentage Change,
Mostpessimistic
"Best guess"
Mostoptimistic
3I.6% 63.6% I07-0%
85 I, under three sets of cost-of-living weights and price assumptions: 75.I% 68.o% 6I.88% 294-5% I03-7% II6-5% 99.2% 349.6% I07-0% I54.8%
I75-3% I64-2% I54-4% 520-3% 220-3%
Sourcesand Notes: The indices in the upper panel use the data in Tables 3 and 4, as does the row of "best guess" estimates in the lower panel. The most pessimistic and most optimistic variants are based on relatively unrealistic cost of living indices, selected as extreme cases from i6 alternatives. The most pessimistic used a cost of living index combining northern urban expenditure weights with Tucker's institutionalclothing prices and Trentham cottage rents, while the most optimistic used an index combining northernrural weights with export clothing prices and no rents. Again, we prefer the "best guess" index, combining southern urban weights with export clothing prices and Trentham rents. The I755 figures are derived by relying on the Phelps Brown-Hopkins index to extend our I78I-I850 series (Table 4) backwards.
earnings? Into this empirical vacuum Hobsbawm has injected fragmentary hints about unemployment in the industrial north, suggesting that the depression of I84I-3 was "almost certainly the worst of the century".38 Yet no conceivable level of unemployment could have cancelled the neardoubling of full-time wages and left the workers of the I840S with less than their grandfathershad had. Such a cancellation of gains would require that the nationalunemployment rate would have had to rise from zero to 50 per cent, or from Io per cent to 55 per cent-jumps which even the most ardent pessimist would dismiss as inconceivable. Even in the I930S, unemployment was less than a quarter of the labour force.The I840S lacked the availability of unemployment compensation, the sharp drop in output, wages and prices, as
38 Hobsbawm, Labouring Men, p. 74. We have checked Hobsbawm's discussion of unemployment against the materials he drew from Finch, Adshead, Facts and Figures, Ashworth, and the Leeds Town Council. In all cases we found the primary materials shaky enough to make them unreliable even as testimony on purely local unemployment, let alone as national averages. The sources repeatedly counted persons not fully employed in a particular trade as unemployed, a procedure that ignores the widespread shifting of individual workers between sectors over the year. Thus a worker employed 40% of the time as a carpenterand 50% of the time in harvesting and assorted odd jobs is simply counted as a carpenter who can find work only 40% of the time (and who may also be counted as an underemployed harvest worker). Many of the sources include as unemployed those who have left for other towns or America. Some, especially Ashworth, ignore newcomers who came to town recently and found jobs, while taking a very generous definition of the unemployment of those previously at work. In one case, Hobsbawm (p. 75) cites figures showing that about i I% of the town of Leeds had averageweekly incomes of I i4pence as evidence that "I5-20 per cent of the population of Leeds had an income of less than one shilling per head per week", a conclusion that ignores the obvious difference between a group average and a group upper bound. Two final difficulties: all of the sources were designed to influence Parliament with pleas of special distress, and at no time does Hobsbawm compare these scraps from the i840s with similar materials for earlier periods.
I4
PETER
H.
LINDERT
AND JEFFREY
G. WILLIAMSON
well as the fall in the investment share that accompanied record jobless rates ninety years later. We can be more precise about the extent to which the unemployment issue has been overstated in the standard of living debate. We have a number of clues about early nineteenth-century unemployment in Britain that have yet to be exploited. Let us examine these, focusing on the controversial period I820-50, beginning with the non-agriculturalsector before tackling the knottier problem of agricultural underemployment. We can put an upper limit on non-agricultural unemployment in the I850s by starting with the share of engineering, metal and shipbuilding union membership who were out of work: in I 85 I, 3 9 per cent were out at any one time, and the average was 5 2 per cent for i85I-9. This sector had all the attributes to suggest that unemployment would exceed economy-wide rates: early unionization, an unemployment insurance scheme, and business cycle sensitivity typical of all capital-goods industries. Indeed, from I 85 I to World War I the unemployment rate in the engineering-metals-shipbuilding sector (EMS) fell below overall unemployment for insured workers in only two years, both of them boom years. Between I923 and I939, the EMS unemployment rate exceeded that of all insured workers by far.39 Thus, the 3.9 and 5-2 per cent EMS figures clearly overstate unemployment for the non-agricultural sector as a whole. These figures establish upper bounds on the extent to which non-agricultural unemployment could have worsened. How much worse could non-agricultural unemployment have been in the "hungry forties" than in the I850s? That unemployment history can be approximated by appealing to the behaviour of other variables. We know that unemployment varies inversely with output over the business cycle. Furthermore, EMS unemployment must have been closely tied to the share of capital formation in national product. There was also a tight nonlinear relationship between unemployment and wage rate increases in Britain between i862 and I957, according to A. W. Phillips's classic study of the Phillips Curve.40Aside from the influence of these three variables, one might also suspect that the structure of the economy drifted over time in a way that shifted the unemployment rate. These propositions can all be tested for the second half of the nineteenth century. If they are successful, then they can be used to predict nonagricultural unemployment back into the I830s. Regression analysis can sort out the determinants of the unemployment rates in engineering-metals-shipwhere building(I85I-I892) the EMS = UEMS unemployment rate (a I per cent rate measured as
"1*.0");
GNP Ratio = the ratio of current nominal gross national product at factor cost to its average level over the immediately preceding five years; I/GNP = the share of gross domestic capital formation in gross national product at factor cost; w = the rate of change from the previous year in the wage rate for shipbuilding and engineering (a I per cent rise is "0oi"); and
Mitchell and Deane, Abstract, pp. 64-7. A. W. Phillips, 'The Relationship between Unemployment and the Rate of Change in Money Wage Rates in the United Kingdom, i862-I957', Economica, xxv (I958), pp. 283-99.
39 40
LIVING
STANDARDS
I5
Time = the year minus I85I. The regression results on annual data for the United Kingdom are (with standarderrors of coefficients in parentheses):41 I5 (GNP Ratio) - I68o02 (I/GNP)-7I.37W UEMS= 32*96-I6 (7.85) (63.81) (2I.83) -288 45(w)2-0*0032 (Time)+o I48 (Time)2
(I2I.80) (0-0036) (O-I5I)
UEMS= 5.78, SEE= 2-50, R2 = .543, F = 91ii, d.f. = 35. The results confirm that EMS unemployment was lower when GNP was on the rise, when investment was a higher share of national product, and when engineering and shipbuilding wage rates were rising. The results also fail to reveal any other structural drift over the second half of the century: the coefficients on the time variables are statistically insignificant. The regression can now predict EMS unemployment rates for the 1840s and late I83OS.42 It would be unwise to make any predictions earlier than this, given the Poor Law reform of I834 and other structural changes in the earlier years. For the period I837-50, the equation generates the following estimates: Boundsfor UEMS= Point estimate? two standard errors Period Estimate of UEMS I837-I839 i840-I850 2 70% 4-4I% 0-7.70% 0-9-4I% (4.44-I4.44%)
(two worst years:

I842-I843)
(9.44%)
The overall rate of non-agriculturalunemployment was probably lower than these estimates. Of the different sectoral output series available for the I830s and I840s,43 only brick output showed as bad a slump in the early forties as did shipbuilding, the latter reflected in the EMS unemployment figures. It is not at all clear that the slump of the early forties was the "worst of the century". The available evidence make it no worse than the slumps of the late I870s or mid-I 88os. More important to the standardof living debate, industrial depression might have been as bad in the immediate post-war years (I8I4-I9)
41 The EMS unemployment rates are from ibid., pp. 64-5; the GNP ratio and I/GNP are from Phyllis Deane, 'New Estimates of Gross National Product for the United Kingdom, I830-I9 I4', Review of Income and Wealth,XIV (i968), pp. I04-5; the rate of nominal wage increase in engineering and shipbuilding is the Bowley-Wood series from Mitchell and Deane, Abstract,pp. 348-5I. 42 Serialcorrelationwould imply that the text is too generous in setting an upper bound on unemployment in the I84os and late I830s. The Durbin-Watson statistic was I-40, suggesting the possibility of serial correlation.A first-orderCochran-Orcutttransformationwas performed (rho = 0 30). The altered regression had a lower standard error of estimate (2-33), but still had a Durbin-Watson statistic of only I-59, far enough from 2-00 to encourage the suspicion of continuing serial correlation. In the spirit of seeking overestimates of likely unemployment, we have reverted to the original equation, unadjusted for serial correlation, instead of pursuing successive iterations to push the Durbin-Watson statistic up toward 20oo. This yields inefficient estimates, overstating the standard error of the unbiased estimates presented here. (The fact that serial correlation also causes underestimation of the standard errors of the coefficients has little bearing here, since we seek accurate predictions rather than significance tests on coefficients.) 43 For sectoral output series, see Mitchell and Deane, Abstract, passim, and Sidney Pollard, 'A New Estimate of British Coal Production, I750-I850', Econ. Hist. Rev., 2nd ser. XXXIII (I980), pp. 2I2-35.
i6
PETER
H.
LINDERT
AND JEFFREY
G. WILLIAMSON
as in the early forties, given that the earlier wage-price deflation was far more severe. We conclude that non-agricultural unemployment was not exceptionally high in either the I 840s or the I 85os, and even if it did rise after I 820, that unlikely event could have had only a trivial impact on workers' real earnings gains. have affected the unemHow might employment conditions in agriculture ployment trends for the economy as a whole? Darkness is nearly total on this front. Seasonal unemployment was, of course, a serious problem throughout the eighteenth and nineteenth centuries.44 To guess when under-employment reached crisis proportions, we can be guided by literary evidence, grain yields, and the terms of trade. The literary signs of distress were strongest during the harvest failures of the I790S and in the twenty years after Waterloo.45 PostNapoleonic wheat yields were trendless from i8i5 to I840, and then rose.46 The terms of trade shifted drastically against agriculture only twice in the century surveyed here-by about 20 per cent against wheat from c. I770 to C. I780 and by about IO per cent against agricultural products from i8I2-I4 to I 822-24.7 The common denominator emerging from this evidence is that the early postwar period, especially the decade I8I5-24, probably witnessed exceptional unemployment in agriculture, followed by overall improvement to I 850. All of this evidence suggests two plausible inferences: first, that unemployment among workers listing non-agricultural occupations was less than 9*4I per cent in the I840s and i85os; and second that unemployment among agriculturalworkers was no worse in the I840s or I85Osthan around 1820. We also know that the share of the British labour force engaged in agriculture dropped from 28.4 per cent for I82I to 217 per cent for I85I.48 This information is sufficient to demonstrate that the (alleged) net rise in unemployment could not have exceeded 7*37 per cent, and it may well have fallen.49 The trend in unemployment thus could not have detracted greatly from the improvement in workers' real wages, and it may even have contributed to their improvement. Furthermore, even a pessimist's reckoning of the influence of unemployment overstates its relevance by assuming that time spent unemployed has no value as either leisure or non-market work.
44 C. Peter Timmer, 'The Turnip, The New Husbandry, and the English Agricultural Revolution', QuarterlyJournalof Economics, LXXXIII (i969), pp. 375-95; E. J. T. Collins, 'Migrant Labour in British Agriculture in the Nineteenth Century', Econ. Hist. Rev., 2nd ser. XXIX (I976), pp. 38-59. 45 E. L. Jones, Agriculture and the Industrial Revolution (New York, I974), ch. Io; T. L. Richardson, 'The Agricultural Labourer's Standard of Living in Kent, I790-i840', in D. Oddy and D. Miller eds. The Making of the ModernBritish Diet (I976), pp. I03-i6; and E. J. Hobsbawm and G. Rude, Captain Swing (New York, i969). 46 Jones, Agriculture, ch. 8. 47 Deane and Cole, British EconomicGrowth, p. 9i; Mitchell and Deane, Abstract, ch. xiv. 48 Deane and Cole, British EconomicGrowth, p. I42. 49 The proof runs as follows. Let the 0 superscript denote I820, and the I superscript denote I850. Further, let the a subscript refer to agriculture and n subscript to the rest of the economy. Let 'a' be the share of the labour force in agriculture, and U the rate of unemployment. The national unemployment rate. (no sectoral subscript) is linked to the sectoral rates by definition as follows: UO= a0U~a+(i -a0)Un?and U1 =a'U+(i -a')Ul Since all is *284 and a' = .2I7, the net change in the unemployment rate becomes
U'-U0
= (-2I7Ua-*284Uoa)+-783UA-.7I6Un.
Given that Un<.094I, and that UO?3Ul,it follows that only the second of these right-hand terms can be positive, and its value must be below *783 X-094I = 7-37%. This exercise for Britain can be repeated with the same results for England and Wales.
LIVING
STANDARDS
I7
VI Thus far we have taken the orthodox path by focusing solely on adult male purchasing power. Yet questions about work and earnings by women and children have always been lurking in the wings throughout the standard of living debate. The rise of their employment in mills and mines is deplored as much today as it was during the public outcry prior to the Factory Acts. The increasing dependence of working-class families on the earnings of children, and the shifting of both children and single young women from the authority of fathers to the discipline of capitalists, are thought to have undermined traditionalfamily roles and fathers' self-esteem.50 Such social side-effects will be treated as part of the larger issue of urban and industrial disamenities in Section VIII below. This section will address the prior question of how trends in the earning power of women and children compared with those in men's real wages. What data there are suggest the tentative conclusion that the earning power of women and children marched roughly in step with the real wage of unskilled labouringmen, a group with whom women and children tended to compete as substitutes. The evidence for this view begins with data on the ratios of women's and children's pay relative to that of adult male labourers, both within sectors and overall. The best single indicator of the relative earning power of a woman's or a child's time is the ratio of their hourly rate to that of men, averaged over all seasons and including home commodity production. Yet such hourly rates are seldom given in historical sources before I850. We thereforegather clues on some close substitute measures. Table 6 presents the best available evidence on two measures that should serve to bracket trends in the value of women's and children's time. One is the weekly earnings ratio. The other is the ratio of hourly hired-labour wage rates, based on data that do not reflect differences in hours worked per week. Working women may have gained ground on unskilled men during the century from I750 to I850. Our gleanings of data on relative weekly earnings hint as much, both for women within rural areas and for a shifting rural-urban average. Yet hints about relative hourlywages (the second column in Table 6) warn that we cannot be sure that there was any upward trend in the true relativevalue of women's time. It may simply be that the relative weekly hours of those working rose, though this seems unlikely. Our tentative conclusion is that the relative earning power of women did not decline. It may have stayed the same, or it may have risen. The relative earning power of children probably stayed the same above the age of fifteen or thereabouts, while declining slightly for younger children. This conclusion is derived mainly from the data on relative weekly earnings in Table 6. These data again may differ in trend from the true average time values because of changes in relative weekly hours, but relative hours would have had to have risen at an implausibly high rate to reverse the main conclusions advanced here. Thus, it appears that an employment-weighted average of the wage rates of
50 N. J. Smelser, Social Change in the Industrial Revolution (Chicago, English WorkingClass, ch. io.
I959),
chs. ix-xi; Thompson,
I8
PETER
H. LINDERT
AND JEFFREY
G. WILLIAMSON
Table 6. Relative Earnings of English Working-ClassWomenand Children, as Fractionsof thosefor Full-Time Adult Male Labourers,I 742-i890
(i -000 = earnings or wage rate of full-time adult male labourers)
Data context (a) i8th-centuryrural: Brandsby, Yorkshire, I742-5I: Corfe Castle, Dorset, I790: English rural poor, I787-I796: (b) igth-centuryrural: Rural England and Wales, I833: Norfolk-Suffolk, I838: (c) i9th-centurytown: Textiles in towns, England and Wales, I833: Birmingham, I839:
Employedwives/women Weekly Hourly earnings wage rate

<.488a
*202 .I34 .224 .284
EmployedChildren (unweightedboy-girlaverages) Weekly Hourly Ages earnings wage rate not given
ageI3/age I7 age I3/age I7
9-I5 yrs. over io yrs.
<-488
>-202 >-I34 *224--365 >-284
<-495a
*253/-330 .i96/-2I4 .I30 .I72
<-495 >-253/>-330
>-i96/>-2I4
.I30--253 >-I72
.47I .327b
>-47I >-327 >-350
age I3/age age I3/age

I4-I8
I7 I7
*233/-4I3 *I53/-230
>-233/>-4I3 >-I53/>-230 > 230
Manchester, I839:
Industrial workers' wives and children, England and Wales, I889I890:
yrs.
.407 >-407 age I3/age I7 *I85/-345 >-I85/>-345 Sourcesand Notes: a Relative daily wages, not weekly. For this reason, and because the hired-labour data are in this case biased toward the peak-demand season more for women and children than for men, the "Weekly" (here, daily) wage ratio overestimates the hourly ratio more than for later observations. b63 working women, ages 2I-70. Two different concepts of an "employed" state are being applied here. The averages for weekly earnings are for persons employed either at or away from home for a significant part of the year. This concept of employment is close to the broad definition of labour force participation. The average hourly wage rates refer to persons when hired outside the home, thus excluding any unemployment. and Prices in England, Brandsby hired women and children: J. E. Thorold Rogers, Historyof Agriculture VII, pp. 499-509. Men: io,657 days at 9-779d., all seasons; women, 2,06i15 days at 4.772d.; boys: I,948 days at 5 I94d.; girls, I22.5 days at 4.486d. Similar ratios for womens' earnings are given for I770 in Arthur Young, A Six Month's Tourof the North of England ( I 77 I). Corfe Castle, Dorset: all wives, 20-59, and children with stated occupations including outwork; from the detailed I790 census at the Dorset Record Office. 26i families. English rural poor: I72 poor families with children, calculated from Davies and Eden data as described in Peter H. Lindert, 'Child Costs and Economic Development', in Richard A. Easterlin, ed. Population and EconomicChange in Developing Countries(Chicago, ig80), pp. 5-79. Rural England and Wales, I833: Poor Law Commission, Rural Queries,in House of Commons, Sessional Papers, I833, drawing on parishes giving explicit answers to questions about earnings. Similar wage rates for women and children are quoted for I843 in House of Commons, Special Assistant Poor Law Commissioners, Reportson the Employmentof Womenand Childrenin Agriculture (I843). Norfolk-Suffolk farms, I838: 64 couples without children, I20 families with one child over IO, from J. P. Kay, 'Earnings of Agricultural Labourers in Norfolk and Suffolk', J3ournal of the Royal Statistical Society, I (I838), pp. I79-80. Textiles in towns, England and Wales, I833: weekly earnings for i,864 children at age I3, I,434 children at age I7, and 972 women at ages 26-30 from Dr James Mitchell's report in Factory Inquiry Commission, Supplementary Report of the Central Board . . . as to the Employment of Children in Factories, in House of Commons, Sessional Papers, I834, vol. xix. The other factory inquiry volumes of that era are also a rich source for earnings profiles by age and sex. Birmingham, I838: 'Contributions to the Economic Statistics of Birmingham', by a local Subcommittee, Journal of the Royal Statistical Society, 2 (I840), p. 44I. Manchester, I839: hourly wage rates from several industries, in David Chadwick, 'On the Rate of Wages in Manchester and Salford . . . I839-59', Quarterly3Journal of the Statistical Society, XXIII (I860), pp. I-36. The comparisons used here were biased against our argument somewhat by choosing lower-paid groups of women and children and better-paid adult male labourers. The I 849 ratios from the same source are similar. Industrial workers' wives and children, I889-I890: 857 industrial workers' wives and I59 families with one child over IO, fron US Commissioner of Labor, Sixth and SeventhAnnual Reports, pt. III (Washington, I89I and I892).
LIVING
STANDARDS
I9
women and children together would have advanced about as fast as those for This adult male farm labourers in the course of the century I750-I850. conclusion would not be overturned by noting the decline in overall labourforce participationof both women and children.51An optimist might interpret this decline as voluntary, that is as showing that the implicit purchasing power (the "shadow price") of time spent away from work rose faster than the observedwage rates. A pessimist might counter that the trend was involuntary, that women and children were being thrown into involuntary unemployment faster than adult males. The latter position is hard to sustain. Aside from "protective" hours legislation from I833 on, no institutions compelled employers to pay women and children wage rates that were increasingly above the opportunity cost of their time out of work. The real wage gains documented for men in Table 5 were not achieved at the expense of women and children. As for the perceived disamenities of having women and children shift their work from home toward factories, these become part of the net disamenities appraisedin Section IX below. Since wage rates differ between occupations, migration from low-paid to high-paid employment can raise average wages for the working class as a whole, even if wage rates do not change for any one occupation. How much of the wage gains up to I850 were due to such mobility-induced wage drift? The "blue collar" and the "all worker" wage series in Tables 3 and 5 already capturethis mobility effect. It can be factored back out by applying fixed wage rates to shifting employment numbers. Detailed calculations described elsewhere show that occupational mobility contributed less than 5-3 per cent to the rise in average full-time earnings of all blue-collar workers between I78I and I85I, using either Laspeyres or Paasche weights.52 Virtually all of the apparent wage gains were gains within occupations, not wage drifts due to changing occupational weights. Migration from low-wage to high-wage regions can also raise wages for the average worker even if wages fail to rise in any one region. That nominal wages varied widely across English counties well into the nineteenth century has been well known at least since Arthur Young's late eighteenth-century tours. Elsewhere we have analyzed the causes of the regional differences in wage rates for agricultural labour.53It appears that real wage gaps persist even after regional adjustments for variation in cost of living and disamenities are made. Furthermore, it has long been appreciated that labour drifted from the low-wage south to the high-wage north during the Industrial Revolution.54
51 On rates of labour-force participation, see DP, Section 3.4, and the sources cited in Table 6. 52 For all workers, the Paasche measure of the mobility effect was as high as I7 2 per cent between I78I and I85 I, but this was again a very small share of the overall real wage gains. These are improvements over the estimates discussed in DP, Section 6, due to slight revisions in the employment weights. Some additional shift toward higher-paying occupations after 8 iI has eluded our measures. We could include only those occupations that supplied pay proxies, and the excluded groups (about a third of the labourforce) shifted from a low-skill mix dominated by domestic servants to a high-skill mix dominated by new occupations created by the Industrial Revolution. More complete measures would show faster average wage gains between i8ii and I85I. 51 See DP, Section 7. 54 In this section, 'migration' is defined as changes in the labour force distribution across regions. There is, of course, an extensive literature which debates the demographic source of the population and labour force shift to high-wage areas over the period which bounds the standard of living debate. This section is not concerned with whether the observed shift can.be attributed to natural increase (and whether to birth or death rate differences) or to actual migration.
20
PETER
H. LINDERT
AND JEFFREY
G. WILLIAMSON
Since the average "blue collar" and "all worker" wage series in Tables 3 and 5 already include most of these regional migration effects, it is now a simple matter to explore their quantitative impact on real wage gains up to I850. It appears that regional migration contributed very little to the real wage gains for the average English worker after the I780s. True, some periods registered far larger regional migration gains than others: I84I-5I was a decade of impressive gains through migration to high-wage regions; and the late eighteenth century records some small positive gains.55 But over the period I78I-I 85 I as a whole, regional migration contributed less than 3.6 per cent to the observed real wage gains. It was real wage gains within regions that mattered most. VII Most of us care more about people's consumption per lifetime, rather than just per year. Conclusions based on trends in real earnings per annum risk grave error if the workers' length of life deteriorated. Alternatively, any improvements in the workers' length of life implies that standard of living gains are understated in Table 5, especially if sickness and longevity are inversely correlated. The standard of living debate commonly fails to confront the issue of lifetime incomes. There are exceptions to this characterizationof the debate, and surely the most famous is Engels' original broadside on The Conditionof the WorkingClass in England, which indicted industrial capitalism for killing workers with unhealthy conditions: ... societyin Englanddailyand hourlycommitswhatthe working-men's organs, as socialmurder. . . its deed is murderjust with perfectcorrectness,characterise as surelyas the deed of the single individual;disguised,maliciousmurderagainst which none can defend himself, which does not seem what it is, becauseno man sees the murderer....56 To link workers' early deaths with rising capitalism, Engels drew on several available estimates showing that mortality was worse in the cities than the countryside, appalling in Liverpool and Manchester, and worst in the most crowded neighbourhoods of the same cities. A host of studies has since confirmed this spatial pattern of mortality. Two pitfalls await anyone trying, as Engels did, to infer from such patterns that working-class longevity diminished as the Industrial Revolution progressed. First, it is important to remember that the working classes were not neatly segregated from the upper classes by place of residence. Liverpool was no closer to being purely working-class than was England in the aggregate. On the contrary, census figures show that the socio-occupational group most concentrated in the unhealthy cities was the bourgeoisie, although most bought their way out of the worst sections by paying higher rents in the more salubrious sections of the same unhealthy cities. The rise of centres of bad health gives no more support to the pessimists' view of working-class conditions
55 But hardly of the size suggested by the qualitative literature. See, for example, Deane and Cole, British EconomicGrowth, ch. III. 56 F. Engels, The Conditionof the WorkingClass in England, translated from the i845 German edition, with an introduction by E. J. Hobsbawm (St. Albans, I974), pp. I26, I27.
LIVING
STANDARDS
2I
than the nineteenth-century decline in national mortality gives to the optimists' position. Thanks to the energies of William Farr and a few other scholars, a fairly clear picture of adult male working-class life expectancy can be put together towards the end of the Industrial Revolution. As we have shown elsewhere, life expectancies did not differ dramatically across occupations for England and Wales as a whole. The occupational differences were far smaller than the spatial differences, though the estimates may be slightly biased toward uniformity by migration effects and by the omission of paupers from the availablefigures.57We are warned that occupational differentials were unlikely to have been as great as stark contrasts between Manchester and rural Norfolk would suggest. The other pitfall is mistaking a snapshot for a motion picture. Past observers have inferred that mortality must have worsened for city-dwellers (and for the workingclass) since mortality was higher in fast-growing cities. Such inferences are unwarranted.In fact, mortality improved (receded) in the countryside and most cities, and did so by enough to improve national life expectancy from about i8oo on. The exceptional cities were Liverpool and Manchester, where crude death rates stayed about the same from I 8oI to I85I, while age-adjusted death rates appear to have risen only slightly up to I820 and declined slightly to mid-century.58 In short, while we cannot yet determine trends in working-class infant and child mortality, it can be shown that any tendency of the rising industrial centres to lure adult workers to an early death was fully offset, or more than offset, by the lengthening of life both in the cities and the countryside.59 VIII Where in our measures is the degradation and demoralization associated with the long rigid hours spent at mind-numbing work for an insensitive avaricious capitalist? The disruption of traditional family roles? The noise, filth, crime, and crowding of urban slums? Until we can devise ways of weighing these varied dimensions against material gains, we cannot answer questionsabout living standards, but rather only questions about real earnings. In judging the importance of these quality of life factors, we must avoid imposing twentieth-century values on early nineteenth-century workers. We must avoid both facile indifference to past suffering and excessive indignation at conditions that seem much more intolerable now. The workers of that era
57 See DP, Table 20. 'Migration effects' could bias the life expectancy measures toward uniformity because persons whose health had been damaged by high-mortality environments (e.g., mining or Manchester) may have migrated to healthier environments to no avail, dying there (e.g., recorded rural labourers)and raising rural mortality rates for misleading reasons. 58 On national life expectancy, see E. A. Wrigley and R. Schofield, The Population History of England, i54i-i87I: A Reconstruction (Cambridge, i98i), esp. Table A3. I. For crude death rates by county and city, see Deane and Cole, British EconomicGrowth, p. I3I; and House of Commons, Sessional Papers, i80i-02, vols. VI and VII, i849, vol. XXI, and i865, vol. XIII, pp. I4ff. For London, we have calculated the following infant mortality rates as percentages of live births, using the bills of mortality plus the third English life table up to i830 and annual reports of the Registrar General thereafter: I729-39 - *3746, I750-9.3322,
I780-9-
.2406,
i800-09- -i864,
i820-30-
59See DP, Table 2i and the accompanying text.
.I428, i85i-6o-
.I548.
22
PETER
H. LINDERT
AND JEFFREY
G. WILLIAMSON
must themselves be allowed to reveal how much a "good" quality of life was worth to them. How is their voice to be heard? Most of them lacked the right to vote. Some protested violently, while most remained silent, but we cannot infer a majority view from either the outcries or the silence. The most satisfactory clue to a worker's view is his response to living and working conditions that offered different rates of pay and different qualities of life. Employers in the rising industries and cities had to pay workers more to attract a growing share of the labour force. How muchmore is a valuable clue to the importance which the workers attached to quality of life. To err on the side of exaggeration on so sensitive an issue, let us contrast the pay of unskilled common labourers in two extreme settings in the late I830s: cotton and metal mills in Manchester in I839 versus year-round farm labour in Norfolk and Suffolk in I837.60 Among the better paid unskilled in Manchester were labourers in metal mills and adult male warehousemen in cotton mills. These groups received I 8 shillings a week, or ?4I 4 for a normal 46-week year. Other common labourers generally got less. In healthy East Anglia, a single adult male farm labourer averaged ?25. I in I837. The extreme Faustian "Satanic mill" wage premium was thus 65 per cent. At some human cost of moving, labourers could choose to suffer the disamenities of Manchester work, even if it looks like a Hobson's choice to us today. And choose they did, in significant numbers. The census of I 85 I shows that every county in England, and even the rural districts of every county, attracted persons born in every other county. Workers also shuttled between the industrial and agricultural sectors on a seasonal basis.61 They were free to move, and many of them knew enough about conditions in other sectors if they cared to consider moving. The possibility of migration implies that the wage premium offered in Manchester cannot understate the importance of urban-industrialdisamenities and living costs to those workers who were actually near the margin in choosing between Manchester and East Anglia. Indeed, it may well overstate this importance not only because the continued flow toward the cities implied a true gain for movers, but also because we have selected two extreme cases for contrast. Manchester was as "bad" a city as any (except perhaps Liverpool), and farm workers in East Anglia were paid a good deal less than their counterparts in the rural north. The Manchester-East Anglia "wage gap" must be viewed as an upper bound on what it took to induce rural labour to modern industry in the second quarter of the nineteenth century. This gap can be viewed as compensation for two closely related costs: the payment necessary to compensate workers (in their own eyes) for a lower quality of life, and the payment necessary to compensate workers for the higher urban cost of living (e.g. expensive housing).
60 The earnings data come from David Chadwick, 'On the Rate of Wages in Manchester and Salford, and the Manufacturing Districts of Lancashire, i839-i859', Journal of the Royal Statistical Society, Series A, xxiii (i86o), pp. I-36, and J. P. Kay, 'Earnings of Agricultural Labourers in Norfolk and Suffolk', T. Royal Stat. Soc., I (i838), pp. I79-83. Both estimates posit some partial unemployment. This took the form of lower winter earnings in the Norfolk-Suffolk farm data and a 46-week year for Manchester. 61 On migration countercurrents, see also E. G. Ravenstein, 'The Laws of Migration', T. Royal Stat. Soc., Series A, XLVIII (i885), pp. i67-235, esp. pp. i87-9.
LIVING
STANDARDS
23
The 65 per cent gap in the I83os and I850s is wide, but how much long-run deteriorationdoes it reveal? Very little, as it turns out. First, it is a contemporaneous comparison between sectors. Workers did not choose between the living conditions of I780 and those of I850. Yet workers' migration behaviour in the nineteenth and twentieth centuries shows that they would have chosen the conditions of i85o overwhelmingly, not only because real wages rose over time but also because the quality of life was improvingwithin both the urban and the rural sectors over time. We infer this mainly from trends in mortality, the best single proxy for the disamenities associated with demoralization, bad health, and low resistance to disease. As we have seen in Section VII, mortality was declining within the countryside and within the cities (though it was only stablein Liverpool and Manchester). Thus, disamenities were declining within both sectors, an improvement we have not measured in monetary terms. What of the impact of population shift from low-density rural areas of high quality of life to high-density urban areas with unpleasant disamenities? For this implicit spatial migration taking place between generations, the full 65 per cent gap might be applied. But it could apply only to that share of the population making the implicit move. An overestimate of this fraction is the rise in the share living in urban places, a measure that generously assumes that all cities were as bad as Manchester. This fractional shift is only I4.89 per cent from I78I to I85I ,62 making the disamenities and living-cost increases for the average worker only .I489X.65 = 9-7 per cent. To repeat, even this is a generousoverestimate. Elsewhere, one of the present authors has analyzed the urban-industrialdisamenities and cost of living effects in greater detail, and found that the former was probably under 2.5 per cent and the latter was probablyunder 3-3 per cent around the I840s.63 Many workers did suffer from the higher disamenities and living costs in the rising industrial centres, and many others resisted them by resolving to live and work elsewhere. Yet the value implicitly put on these human costs by the marginal workers, who actually moved into the urban-industrial centres, was not large enough to cancel even a tenth of blue collar workers' real wage gains. Nor should this surprise us. The development process inevitably finds workers and policymakers more intent on basic purchasing power in the hungry early phases of development, as the history of environmental quality in Manchester, Pittsburgh, Manila, and Sdo Paulo clearly shows. Ix We now have a much broader empirical basis for inferring how workers fared in the century between I750 and I850. Our tentative findings can be summarized most easily by focusing on the relatively data-rich Industrial Revolution era from I 78 I to I 85 I.
62 Urban populations for i80I-5I were taken from Mitchell and Deane, Abstract, pp. 24-7. Deane and Cole (British Economic Growth, p. 7) suggested that the urban population share in the mid-eighteenth century was about I5-i6 per cent, and we extrapolated between I75i and i8oi using this figure for the formerdate. 63 See Jeffrey G. Williamson, 'Urban Disamenities, Dark Satanic Mills, and the British Standard of Living Debate', J. Econ. Hist. XLI (i98i), and 'Was the Industrial Revolution Worth It? Disamenities and Death in i9th Century British Towns', Explorationsin EconomicHistory, i9 (i982), pp. 22I-45.
24
PETER
H. LINDERT
AND JEFFREY
G. WILLIAMSON
Table 7 collects our best estimates of workers' experiences, beginning with our "best-guess" real wage trends and proceeding through several long-needed adjustments. Table 7 suggests impressive net gains in the standardof life: over 6o per cent for farm labourers, over 86 per cent for blue-collar workers, and over I40 per cent for all workers. The hardships faced by workers at the end of the Industrial Revolution cannot have been nearly as great as those of their grandparents. The great majority of these human gains came after I820. Pessimists must retreat to the pre-i820 era, where workers' net gains look as elusive in this paper as in past studies of single occupations. Optimists might feel a temptation to proclaim victory by shifting attention away from the classic Industrial Revolution era, but this inference seems premature. After all, the best measures of industrialization are still so uncertain for the years before I83I that the timing of the Revolution itself is debatable. Furthermore, the separate influence of the French Wars has yet to be factored out of the longer-run development experience. Pessimists in tune with larger intellectual currents of the i960s have called for less attention to conventional measures of real income and more to the environmental aspects of the Industrial Revolution-all of those appalling health conditions and social injustices with which workers were affronted. The shift is certainly welcome. Yet the results presented here suggest that nineteenth-century environmental influences on health-such as crowding, infection, and pollution-could hardly have lowered average quality of life over time. Research inspired by the pessimists' allegations about the Industrial RevTable 7. Revised Measures of English Workers'Standard-of-Living Gains,
I78i-I85I
Sourceof Improvement
i. Real full-time earnings
Overall improvement, I78i-i85I Farm All Blue Labourers Collar Workers

63.6% 99-2%
All Workers
I54.8%
(<I7-2%)
2.
("Best guess", Table 5) ia. Due to occupational change (DP, Section 6) ib. Due to regional migration (DP, Section 7) ic. Residual: real wage gains within occupations and regions Diminished by an "upper-bound" rise
(0) (<3.6%)
(>6o.o%)
(<5-3%) (<3.6%)
(>90.3%)
(<3.6%)
(>I34-0%)
in unemployment, or VI above)
<7.4%
(Section
63.6%
>9I.8% >88-5%
>86-o%
>I47-4%
3. Diminished by the shift toward higher
urbanlivingcosts, or <3-3%*
4. Diminished by urban-industrial disamenities, or <2-5%* 5. Augmented by adult mortality gains, which were not negative (Section VIII)
63.6%
63.6%
>I44-I%
>I4I.6% >I4I.6%
>63.6%
>86.o%
Notes: *These figures are taken from DP, Section 8. Readers preferring the estimates in Section IX above may wish to substitute the 9-7 per cent figure for Rows 3 and 4 together in the "blue collar" and "all workers" columns.
LIVING
STANDARDS
25
olution will now have to shift to social injustice and social disorder. Reported crime, alcoholism, protest, and illegitimacy appear to have been on the rise in the first half of the nineteenth century. In addition, rising material inequality seems to characterize the period, and these trends were likely to have influenced social disorder. We know from related research, for example, that common labourers failed to experience gains as rapid as those for better-paid employees, as even Table 7 shows in comparing three working-class aggregates. Earnings inequality statistics suggest the same between the I82os and the i88os. Our preliminary results on probated wealth and land rents also suggest rising inequality between upper and lower-middle classes in the course of the first half of the nineteenth century. Inequality between workers and the very poor may also have widened: we need far better information on the material condition of those below the working classes-vagrants and disabled paupers-than the large literature on poor relief has yet been able to supply. These issues, not trends in absolute living standards, are likely to mark the future battleground between optimists and pessimists about how workers fared under nineteenth-century British capitalism. University of California, Davis University of Wisconsin,Madison

Industrial Revolution English Workers Li

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Industrial Revolution English Workers Li

Uploaded by

Copyright:

Available Formats

English Workers' Living Standards during the Industrial Revolution: A New Look Author(s): Peter H.

SURVEYS AND SPECULATIONS, XVII

Workers' Living Standards Revolution: the Industrial A New Look*

T he politically charged debate over workers' living standardsduring

On the debateoverthe 'bottom40 percent'in the ThirdWorld,see H. B. Chenery et al. Redistribution

Approximate Mean-wage Percentile Positions in the EarningsRanks

(i) Farm Labour

(Bottom 40%) (2) Middle Group

(5L) police and guards

(6L) colliers (3) Artisans

33-54 46-02 48-08 24-37

43-94 76-0i 57-47 67-89 6g-22 I76-86

43-94 80.69 60-22 69-34 57-82 I95-I6

8I-35 6o.60 69.I8 50-37 2I9-25

22-94 78-9I 38-82 43-60

5I-26 64-99 I5I-09

62-95 54.6I 222-95 62-22 80.69 66-35 58-5o

63.33 56-4I 270-42 62-74 77-26

(IH) government high-wage (2H) shipbuilding trades

45-26 50-83 35-57

5I-7I 58-o8 40-64

35-96 46-34 9I-90 23I-00 63-62 62-02 I5-97

66-6I 238-50 I65-00 I35-26 I74-95

I82-65 242-67 IOI-57 88-35 I6-53

266-42 340-00 I50-44 2I7.60

283-89 447-50 I78-II 2I7.60

272-53 447-50 200-79 2I7.60

266-55 447-50 229-64 2I7.60 69-35 326-43

254-60 522-50 240-29 I75-20 69-35 265-7I

258-76 II66-67 269.II 200-92 8I.89 398-89

267-09 I837-50 235-8I 200.92 8I.Ii

(ioH) surgeons & doctors

(I2H) engineers & surveyors

15As reproduced in Brian Mitchell and Phyllis Deane, Abstractof BritishHistoricalStatistics(Cambridge,

the Royal StatisticalSociety, LXI (i898).

= Ioo) (4) (5) (6)

vs. Year Farm Labourers

Bowley's Farm Labourers

(3) Phelps Brownvs. vs. Hopkins Labour Building Labourers Aristocracy

Tucker's London Artisans

All Blue Collar

White Collar 2I.62 26.42

57-6 66.7 83-3

57.38 64.86 79-44

69.8 8i.o 87.0

9I 92 97.0 97-0 93-55 88-68 97-0 00.0 100-00

EconomicHistory, XL (i980), pp. 685-7I2.

Percentagechangeover the period

-30-7 -23.5 -27.3

Source: DP, AppendixB.

Figure I. AdultMale AverageFull-Time EarningsforSelectedGroupsof Workers, I 755-i85i, at Constant Prices

?50?4()0 ?30 ?25. 17 ?20 -

Middle Group ? 29-04 Farm Labourcrs

1797 1805 10'15 19

All Blue Collar

I8I5 18I9 I827

57.8I 54-35 70-I8

58.I5 55.68 69.25

25.49 27.76 39-10

46.7I 46.I3 58.99

3I.6% 63.6% I07-0%

(two worst years:

chs. ix-xi; Thompson,