7/27/2019 8419964 a W Phillips the Relation Between Unemployment Xxx

http://slidepdf.com/reader/full/8419964-a-w-phillips-the-relation-between-unemployment-xxx 1/18

The Suntory and Toyota International Centres for Economics and Related Disciplines

The Relation between Unemployment and the Rate of Change of Money Wage Rates in theUnited Kingdom, 1861-1957Author(s): A. W. PhillipsSource: Economica, New Series, Vol. 25, No. 100 (Nov., 1958), pp. 283-299Published by: Blackwell Publishing on behalf of The London School of Economics and Political

Science and The Suntory and Toyota International Centres for Economics and RelatedDisciplinesStable URL: http://www.jstor.org/stable/2550759

Accessed: 08/11/2008 10:54

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at

http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless

you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you

may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at

http://www.jstor.org/action/showPublisher?publisherCode=black .

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed

page of such transmission.

JSTOR is a not-for-profit organization founded in 1995 to build trusted digital archives for scholarship. We work with the

scholarly community to preserve their work and the materials they rely upon, and to build a common research platform that

promotes the discovery and use of these resources. For more information about JSTOR, please contact support@jstor.org.

The London School of Economics and Political Science, Blackwell Publishing, The Suntory and Toyota

 International Centres for Economics and Related Disciplines are collaborating with JSTOR to digitize,preserve and extend access to Economica.

http://www.jstor.org

7/27/2019 8419964 a W Phillips the Relation Between Unemployment Xxx

http://slidepdf.com/reader/full/8419964-a-w-phillips-the-relation-between-unemployment-xxx 2/18

1958]

The Relation Between Unemploymentand the Rateof Change of Money Wage Rates in the

United Kingdom, 1861-19571By A. W. PHILLIPS

I. HYPOTHESIS

Whenthe demandfor a commodityor service s high relatively othesupplyof it we expect hepriceto rise,the rateof risebeinggreaterthe greater he excessdemand. Converselywhen the demand s lowrelativelyo the supplywe expectthepriceto fall, the rateof fallbeinggreater he greater he deficiencyof demand. It seems plausiblethatthisprinciple houldoperateas one of the factorsdetermininghe rateof changeof money wagerates,whichare the priceof labourservices.Whenthe demand or labour s highand thereareveryfewunemployedwe shouldexpect employers o bid wage ratesup quite rapidly,eachfirmand each industrybeing continuallyempted o offera little abovethe prevailingrates to attractthe most suitablelabour from otherfirmsand industries. On the otherhand it appears hat workersare

reluctant o offertheirservicesat less than the prevailing ateswhenthe demand or labour s low andunemployments highso thatwagerates all only veryslowly. The relationbetweenunemploymentndtherateof changeof wagerates s thereforeikelyto behighlynon-linear.

It seemspossiblethata secondfactorinfluencinghe rateof changeof money wage ratesmightbe the rate of changeof the demand orlabour,and so of unemployment. Thus in a year of risingbusinessactivity,with the demandfor labour increasingand the percentageunemployment ecreasing, mployerswill be biddingmorevigorously

for the servicesof labourthantheywould be in a year duringwhichthe averagepercentageunemploymentwas the same but the demandfor labourwas not increasing.Converselyn a yearof fallingbusinessactivity,with the demandfor labour decreasingand the percentageunemploymentncreasing, mployerswill be less inclined o grantwageincreases,and workerswill be in a weakerpositionto pressfor them,than they would be in a year duringwhich the average percentageunemploymentwas the same but the demandfor labour was notdecreasing.

A thirdfactorwhichmay affectthe rate of changeof money wageratesis the rate of changeof retailprices,operating hroughcost oflivingadjustmentsn wagerates. It willbe arguedhere,however, hatcost of living adjustmentswill have little or no effecton the rate ofchangeof money wage rates except at times when retail prices are

1 This studyis part of a widerresearchprojectfinancedby a grantfrom the FordFoundation. The writerwas assisted by Mrs. MarjoryKlonarides. Thanks aredue to ProfessorE. H. PhelpsBrown,ProfessorJ. B. Meadeand Dr. R. G. Lipseyfor commentson an earlierdraft.

283 A

7/27/2019 8419964 a W Phillips the Relation Between Unemployment Xxx

http://slidepdf.com/reader/full/8419964-a-w-phillips-the-relation-between-unemployment-xxx 3/18

284 ECONOMICA [NOVEMBER

forced up by a very rapid rise in import prices (or, on rare occasions inthe United Kingdom, in the prices of home-produced agriculturalproducts). For suppose that productivity is increasing steadily at

the Iate of, say, 2 per cent. per annum and that aggregate demand isincreasing similarly so that unemployment is remaining constant at,say, 2 per cent. Assume that with this level of unemployment andwithout any cost of living adjustments wage rates rise by, say, 3 percent. per annum as the result of employers' competitive bidding forlabour and that import prices and the prices of other factor servicesare also rising by 3 per cent. per annum. Then retail prices will berising on average at the rate of about 1 per cent. per annum (the rate ofchange of factor costs minus the rate of change of productivity).

Under these conditions the introduction of cost of living adjustments inwage rates will have no effect, for employers will merely be givingunder the name of cost of living adjustmentspart of the wage increaseswhich they would in any case have given as a result of their competitivebidding for labour.

Assuming that the value of imports is one fifth of national income,it is only at times when the annual rate of change of import pricesexceeds the rate at which wage rates would rise as a result of com-petitive bidding by employers by more than five times the rate of

increase of productivity that cost of living adjustments become anoperative factor in increasing the rate of change of money wage rates.Thus in the example given above a rate of increase of import prices ofmore than 13 per cent. per annum would more than offset the effectsof rising productivity so that retail prices would rise by more than3 per cent. per annum. Cost of living adjustmentswould then lead toa greater increase in wage rates than would have occurred as a resultof employers'demand for labour and this would cause a furtherincreasein retail prices, the rapid rise in import prices thus initiating a wage-price spiral which would continue until the rate of increase of importprices dropped significantly below the critical value of about 13 percent. per annum.

The purpose of the present study is to see whether statistical evidencesupports the hypothesis that the rate of change of money wage ratesin the United Kingdom can be explained by the level of unemploymentand the rate of change of unemployment, except in or immediatelyafter those years in which there was a very rapid rise in import prices,

and if so to form some quantitativeestimate of the relation

betweenunemployment and the rate of change of money wage rates. Theperiods 1861-1913, 1913-1948 and 1948-1957 will be consideredseparately.

II. 1861-1913

Schlote's index of the average price of imports' shows an increase of12 5 per cent. in import prices in 1862 as compared with the previous

I W. Schlote, British Overseas Tradefrom 1700 to the 1930's, Table 26.

7/27/2019 8419964 a W Phillips the Relation Between Unemployment Xxx

http://slidepdf.com/reader/full/8419964-a-w-phillips-the-relation-between-unemployment-xxx 4/18

1958] UNEMPLOYMENTAND MONEY WAGE RATES 285

year, an increase of 7 *6 per cent. in 1900and in 1910,and an increase of7 0 per cent. in 1872. In no other year between 1861 and 1913 wasthere an increase in import prices of as much as 5 per cent. If the

hypothesis tatedaboveis correct he risein importprices n 1862mayjust have beensufficient o startup a mildwage-price piral,but in theremainder f the period changes n import priceswill have had littleor no effect on the rateof changeof wagerates.

U 0

cf.

E? 2 -

0 0~~~~

cl)

? -2

. ~~~~~~~~~~*

.-40 1 2 3 4 6 7 8 9 10 11

Unemptoyment, %.

Fig,t. 1861 1913

A scatterdiagramof the rateof changeof wagerates and the per-centageunemploymentor the years 1861-1913s shownin Figure1.During this time there were 61 fairly regulartrade cycles with anaverageperiodof about 8 years. Scatterdiagrams or the years ofeach tradecycleareshown n Figures2 to 8. Eachdot in the diagramsrepresentsa year, the average rate of changeof money wage ratesduringthe year being given by the scale on the verticalaxis and the

averageunemployment uringthe yearby the scaleon the horizontalaxis. The rate of changeof money wage rateswas calculated romtheindex of hourlywageratesconstructed y PhelpsBrownand SheilaHopkins,' by expressing he first centraldifferenceof the index foreachyearas a percentagef the indexforthesameyear. Thusthe rateof change or 1861 s taken to be half the differencebetween he indexfor 1862andthe index for 1860expressedas a percentage f the index

IE. HI.Phelps Brown and SheilaHopkins, 'TheCourse of Wage Rates in FiveCountries,1860-1939, OxfordEconomicPapers,June, 1950.

7/27/2019 8419964 a W Phillips the Relation Between Unemployment Xxx

http://slidepdf.com/reader/full/8419964-a-w-phillips-the-relation-between-unemployment-xxx 5/18

286 ECONOMICA [NOVEMBER

10

@i 10X.

*-,Curve fitted to 1861-1913 data

6-

IV

4 64 63

c 2: 2- t-t 2 >

~~~~~62

-2z ~~~~~~~~~67

I I I * a I I I I I B

0 1 2 3 4 5 6 7 8. 9 10 11

Unemployment, %.

Fig.2. 1861 - 1868

2 Curve fitted to 1861-1913 data

72

6-673

c)03: 4lZ'% 70

C

0 Unmpoyen,74.

Fig.3. 18681879 ~ ~ ~ 7

7/27/2019 8419964 a W Phillips the Relation Between Unemployment Xxx

http://slidepdf.com/reader/full/8419964-a-w-phillips-the-relation-between-unemployment-xxx 6/18

1958] UNEMPLOYMENTAND MONEY WAGE RATES 287

- Curve fitted to 1861-1913 data

6-

4-

C 2-0E 62 86

64 657~~~9

o -4-0 1 2 3 4 5 6 7 8 9 tO 11

Unemployment, %.

Fig.4. 1879-1 886

Sz 8*- Curve fitted to 1861-1913 data

,, 6- \

01

E 4

-3 0

018

t -2-

o -4I I

, , , , , ,I

0 1 2 3 4 5 6 7 8 9 10 11

Unemptoyment, %.

Fig.4a. 1879-1886, using Bowley's wageindex for the years 1881 to 1886

7/27/2019 8419964 a W Phillips the Relation Between Unemployment Xxx

http://slidepdf.com/reader/full/8419964-a-w-phillips-the-relation-between-unemployment-xxx 7/18

288 ECONOMICA [NOVEMBER

Ct1:

10-

-f1C Curve fitted to 1861-1913 data

+' 6 \

06

0)3 4 \9

o 90E 87 86

-00 92 930) -2

0 1 2 3 4 5 6 7 8 9 10 I

cr. Unemptoyment, %.

Fig.5. 1886- 1893

># 10

*\ Curve fitted to 1861-1913 data

6-

(U4

0 0

no4- \

E0

@ ~~~~~012 os 04 94 93

X-2-

o -4J0 1 2 3 4 5 6 7 8 9 10 11

Unemployment, 1.

Fig.6. 1893 -1904

7/27/2019 8419964 a W Phillips the Relation Between Unemployment Xxx

http://slidepdf.com/reader/full/8419964-a-w-phillips-the-relation-between-unemployment-xxx 8/18

1958] UNEMPLOYMENT AND MONEY WAGE RATES 289

*- Curve titted to 1861-1913 data

1 4-, 2 06 0

2-\0~ ~ ~~~~0

E 0809

o4 07 - _0 07 04

C -2 -u

o - I

,, 0 1 2 3 4 5 6 7 8 9 10 11

x JUnemployment, .

Fig. 7. 1904 - 1909

L

> 10-

*- Curve fitted to 1861-1913 data

6-

0)U 4

12

2- 11

o 13E 10

0'E

09

c))

C -2-

0 1 2 3 4 5 6 7 B 9 10 t1

r JUnempLoyment, %.

Fig.8. 1909 -1913

7/27/2019 8419964 a W Phillips the Relation Between Unemployment Xxx

http://slidepdf.com/reader/full/8419964-a-w-phillips-the-relation-between-unemployment-xxx 9/18

290 ECONOMICA [NOVEMBER

for 1861, and similarly for other years.' The percentage unemploy-ment figures are those calculated by the Boardof Tradeand the Ministryof Labour2 from trade union returns. The corresponding percentage

employment figuresare quoted in Beveridge, Full Employmenitn a FreeSocietp, Table 22.

It will be seen from Figures 2 to 8 that there is a clear tendency forthe rate of change of money wage rates to be high when unemploymentis low and to be low or negative when unemployment is high. Thereis also a clear tendency for the rate of change of money wage rates atany given level of unemployment to be above the average for that levelof unemploymentwhen unemploymenits decreasingduringthe upswingof a trade cycle and to be below the average for that level of unemploy-

ment when unemployment is increasing during the downswing of atrade cycle.

The crosses shown in Figure 1 give the average values of the rate ofchange of money wage r-atesand of the percentage unemployment inthose years in which uInemploymentay between 0 and 2, 2 and 3, 3 and4. 4 and 5, 5 and 7, and 7 and 11per cent. respectively(the upper boundbeing included in each interval). Since each interval includes years inwhich unemployment was increasing and years in which it wasdecreasing the effect of changing unemployment on the rate of change

of wage rates tends to be cancelled out by this averaging, so that eachcross gives an approximation to the rate of change of wages whichwould be associated with the indicated level of unemployment ifunemployment were held constant at that level.

The curve shown in Figure 1 (and repeated for comparison in laterdiagrams) was fitted to the crosses. The form of equation chosen was

y + a = bxcor log (y + a) = log b + c log x

where y is the rate of change of wage rates and x is the percentageunemployment. The constants b and c were estimated by least squaresusing the values of y and x corresponding to the crosses in the fourintervalsbetween 0 and 5 per cent. unemployment, the constant a beingchosen by trial and error to make the curve pass as close as possible tothe remaining two crosses in the intervals between 5 and 11 per cent.unemployment.3 The equation of the fitted curve is

y + 0900 = 9*638x-1394or log (y + 0 900) = 0.984 - 1*394 log x.

1The index is apparently ntendedto measure the averageof wage rates duringeach year. The first central difference s therefore the best simple approximationto the averageabsoluterate of changeof wage rates duringa year.and the centraldifference xpressedas a percentageof the index number s an appropriatemeasureof the averagepercentagerateof changeof wage rates duringthe year.

2 Memoranda upon British and Foreignt Trade and IndutstrialConditions (SecondSeries) (Cd. 2337), B.P.P. 1905, Vol. 84; 21st Abstract oJ Labour Statistics, 1919-1933 (Cd. 4625), B.P.P. 1933-34, Vol. 26.

3At first sight it might appearpreferableto carryout a multiple regressionof

y on the variablesx and drx However, owing to the particular orm of the relation

7/27/2019 8419964 a W Phillips the Relation Between Unemployment Xxx

http://slidepdf.com/reader/full/8419964-a-w-phillips-the-relation-between-unemployment-xxx 10/18

1958] UNEMPLOYMENTAND MONEY WAGE RATES 291

Consideringhe wage changes n individualyearsin relationto thefittedcurve,the wageincrease n 1862(see Figure2) is definitely argerthan can be accounted or by the level of unemployment nd the rate

of change of unemployment,and the wage increase n 1863 is alsolarger hanwouldbe expected. It seems hat the 12*5 percent. increasein importpricesbetween1861 and 1862 referred o above (and nodoubtconnectedwith the outbreakof the Americancivil war) was infact sufficient o have a real effect on wage rates by causing cost oflivingincreases n wageswhichwere greater han the increaseswhichwouldhave resulted romemployers'demand or labour and that theconsequentwage-price piralcontinued nto 1863. On the other handthe increasesn import pricesof 7 6 percent. between 1899 and 1900and again between1909 and 1910 and the increaseof 7 0 per cent.between1871 and 1872do not seemto have had any noticeableeffecton wage rates. This is consistentwith the hypothesisstated aboveaboutthe effectof rising mportpriceson wagerates.

Figure3 and Figures5 to 8 show a veryclearrelationbetween herateof changeof wageratesand the level and rate of changeof unem-ployment,'but the relationhardlyappearsat all in the cycle shown inFigure4. The wageindex of PhelpsBrownandSheilaHopkinsfrom

whichthe changes n wagerates were calculatedwas based on Wood'searlier index,2 which shows the same stability during these years.From 1880we havealsoBowley's ndex of wagerates.3 If the rate ofchange of money wage rates for 1881 to 1886 is calculatedfromBowley'sindex by the same method as was used before,the resultsshownin Figure4a are obtained,givingthe typicalrelation betweenthe rate of changeof wage ratesand the level and rate of changeofunemployment. It seems possible that some peculiaritymay haveoccurredn theconstruction f Wood's ndexfor theseyears. Bowley's

indexfor theremainderf theperiodupto 1913givesresultswhicharebroadlysimilarto those shownin Figures5 to 8, but the pattern s

betweeny and x in the presentcase it is not easy to find a suitable linear multiple

regressionequation. An equation of the form y+ a = bxc+ k (1 dx would

probablybe suitable. If so the procedurewhich has been adopted for estimating

the relationthat would hold betweeny and x if were zero is satisfactory,since itdt

1 dx

can easilybe shown that . is uncorrelatedwithx or with any powerof x pro-vided that x is, as in this case, a trend-freevariable.

1 Since the unemployment iguresused are the averagesof monthly percentages,the firstcentraldifferences againthe best simple approximation o the averagerateof change of unemploymentduringa year. It is obvious from an inspection ofFig. 3 andFigs. 5 to 8 thatin eachcyclethere s a close relationbetween hedeviationsof the points fromthe fittedcurveand the firstcentral differences f the employmentfigures, houghthe magnitudeof the relationdoesnot seemto have remained onstantover the whole period.

2 See Phelps Brown and Sheila Hopkins, loc. cit., pp. 264-5.3A. L. Bowley, Wagesand Incomein the UnitedKingdom ince 1860, TableVII,

p. 30,

7/27/2019 8419964 a W Phillips the Relation Between Unemployment Xxx

http://slidepdf.com/reader/full/8419964-a-w-phillips-the-relation-between-unemployment-xxx 11/18

292 ECONOMICA [NOVEMBER

rather less regular than that obtained with the index of Phelps Brownand Sheila Hopkins.

From Figure 6 it can be seen that wage rates rose more slowly than

usual in the upswing of business activity from 1893 to 1896 and thenreturned to their normal pattern of change ; but with a temporaryincrease in unemployment during 1897. This suggests that there mayhave been exceptional resistance by employers to wage increases from1894 to 1896, culminating in industrial strife in 1897. A glance atindustrial history' confirms this suspicion. During the 1890's therewas a rapid growth of employers' federations and from 1895 to 1897there was resistance by the employers' federations to trade uniondemands for the introduction of an eight-hour working day, which

would have involved a rise in hourly wage rates. This resulted in astrike by the Amalgamated Society of Engineers, countered by theEmployers' Federation with a lock-out which lasted until January1898.

From Figure 8 it can be seen that the relation between wage changesand unemployment was again disturbed in 1912. From the monthlyfigures of percentage unemployment in trade unions2 we find thatunemployment rose from 2 8 per cent. in February 1912 to 11.3 per

cent. in March,falling back to 3-6 per cent. in April and 2

-7 per cent. inMay, as the result of a general stoppage of work in coal mining. If

an adjustmentis made to eliminate the effect of the strike on unemploy-ment the figure for the average percentage unemployment during 1912would be reduced by about 0 8 per cent., restoring the typical patternof the relation between the rate of change of wage rates and the leveland rate of change of unemployment.

From a comparison of Figures 2 to 8 it appears that the width ofloops obtained in each trade cycle has tended to narrow, suggesting a

reduction in the dependence of the rate of change of wage rates on therate of change of unemployment. There seem to be two possibleexplanations of this. First, in the coal and steel industries before thefirst world war sliding scale adjustmentswere common, by which wagerates were linked to the prices of the products.3 Given the tendencyof product prices to rise with an increase in business activity and fallwith a decrease in business activity, these agreements may havestrengthened the relation between changes in wage rates and changesin unemployment in these industries. During the earlier years of the

period these industries would have fairly large weights in the wageindex, but with the greatercoverage of the statistical material availablein later years the weights of these industries in the index would bereduced. Second, it is possible that the decrease in the width of the

loops resulted not so much from a reductionin the dependence of wage

1 See B. C. Roberts,TheTradesUnionCongress, 868-1921,ChapterIV,especiallypp. 158-162.

2 21st Abstractof LabourStatistics,1919-1933, loc. cit.8 I am indebtedto ProfessorPhelpsBrownfor pointing this out to me.

7/27/2019 8419964 a W Phillips the Relation Between Unemployment Xxx

http://slidepdf.com/reader/full/8419964-a-w-phillips-the-relation-between-unemployment-xxx 12/18

1958] UNEMPLOYMENTAND MONEY WAGE RATES 293

changes on changesin unemployment s from the introductionof atime lag in the responseof wage changesto changesin the level ofunemployment, ausedby the extensionof collective bargainingand

particularlyby the growthof arbitrationand conciliationprocedures.If sucha time agexistedn the lateryearsof theperiod he wage changein anyyear shouldbe related,not to averageunemploymenturingthatyear, but to the averageunemploymentagged by, perhaps,severalmonths. This would have the effect of moving each point in thediagramshorizontallypart of the way towards the point of the pre-cedingyearand it can easilybe seen that this would widen the loopsin the diagrams. This fact makes it difficultto discriminate t allcloselybetweenthe effectof time lags and the effect of dependence f

wagechangeson the rate of changeof unemployment.

ILl. 1913-1948

A scatterdiagramof the rate of changeof wageratesandpercentageunemploymentor the years 1913-1948s shownin Figure9. From1913to 1920the series used are a continuationof those used for theperiod1861-1913. From 1921to 1948the Ministryof Labour'sndexof hourlywage ratesat the end of Decemberof each year1has been

used, the percentage hangein the index each year being taken as ameasureof the averagerateof changeof wageratesduringthat year.The Ministryof Labour's igures or the percentageunemploymentnthe United Kingdom2have been used for the years 1921-1945. Forthe years 1946-1948 he unemploymentigureswere taken from theStatisticalYearbooksf the InternationalLabourOrganisation.

It will be seen fromFigure9 that therewas anincrease n unemploy-mentin 1914(mainlydueto a sharprisein the threemonthsfollowingthe commencement f the war). From 1915to 1918unemployment

waslow andwageratesroserapidly. The costof livingwas alsorisingrapidlyandformalagreementsor automatic ost of livingadjustmentsin wageratesbecamewidespread, utit is not clearwhether he cost oflivingadjustments erea real factor n increasingwageratesorwhethertheymerelyreplacedncreaseswhichwould in any case haveoccurredas a resultof the high demandfor labour. Demobilisationbroughtincreasedunemploymentn 1919 but wage rates continued to riserapidlyuntil 1920,probablyas a resultof the rapidlyrising importprices,whichreached heirpeakin 1920,andconsequent ost of livingadjustmentsn wagerates. Therewas thena sharp ncrease n unem-ployment rom 2 6 per cent. n 1920to 17 0 percent. in 1921,accom-paniedby a fall of 22 2 percent. in wagerates n 1921. Partof the fallcanbe explainedby the extremely apid ncreasen unemployment, uta fall of 12 8 per cent.in the cost of living,largelya resultof fallingimportprices,was no doubtalso a majorfactor. In 1922unemploy-mentwas 14* percent.andwagerates ellby 19*1percent. Although

'Ministry of LabourGazette,April, 1958, p. 133.'Ibid., January,1940 and subsequent ssues.

7/27/2019 8419964 a W Phillips the Relation Between Unemployment Xxx

http://slidepdf.com/reader/full/8419964-a-w-phillips-the-relation-between-unemployment-xxx 13/18

294 ECONOMICA [NOVEMBER

32- Curve fitted to 1861 -1913 data

28-1

24-

17

2 0

o 16

116

8 15 0 1 1 6 1 0 2

0~~~~~~4

Unemployment %.

Fig. 9 1913-1948

unemployment was high in this year it was decreasing, and the majorpart of the large fall in wage rates must be explained by the fall of17*5per cent. in the cost of living index betwveen 921 and 1922. Afterthis experience trade unions became less enthusiastic about agreementsfor automatic cost of living adjustments and the number of these

agreements declined.From 1923 to 1929 there were only small changes in import prices

and in the cost of living. In 1923 and 1924 unemployment was highbut decreasing. Wage rates fell slightly in 1923 and rose by 3*1 percent. in 1924. It seems likely that if business activity had continuedto improve after 1924 the changes in wage rates would have shown theusual pattern of the recovery phase of earlier trade cycles. However,the decisionl to check demand in an attempt to force the price leveldown in order to restore the gold standard at the pre-war parity of

7/27/2019 8419964 a W Phillips the Relation Between Unemployment Xxx

http://slidepdf.com/reader/full/8419964-a-w-phillips-the-relation-between-unemployment-xxx 14/18

1958] UNEMPLOYMENT AND MONEY WAGE RATES 295

sterling prevented the recovery of business activity and unemploymentremained fairly steady between 9 7 per cent. and 12- 5 per cent. from1925 to 1929. The average level of unemployment during these five

years was 10 94 per cent. and the average rate of change of wage rateswas - 0 *60 per cent. per year. The rate of change of wage rates calcu-lated from the curve fitted to the 1861-1913data for a level of unemploy-ment of 10 94 per cent. is - 0*56 per cent. per year, in close agreementwith the average observed value. Thus the evidence does not supportthe view, which is sometimes expressed, that the policy of forcing theprice level down failed because of increased resistance to downwardmovements of wage rates. The actual results obtained, given thelevels of unemployment which were held, could have been predicted

fairly accura ely from a study of the pre-war data, if anyone had feltinclined to carry out the necessary analysis.

The relation between wage changes and unemployment during the1929-1937 trade cycle follows the usual pattern of the cycles in the1861-1913 period except for the higherlevel of unemployment through-out the cycle. The increases in wage rates in 1935, 1936 and 1937 areperhaps rather larger than would be expected to result from the rate ofchange of employment alone and part of the increases must probablybe attributed to cost of living adjustments. The cost of living indexrose 3*1percent. in 1935,3 0 per cent. in 1936and 5 *2 per cent. in 1937,the major part of the increase in each of these years being due to therise in the food component of the index. Only in 1937 can the rise infood prices be fully accounted for by rising import prices; in 1935 and1936it seems likely that the policies introduced to raise prices of home-produced agricultural produce played a significant part in increasingfood prices and so the cost of living index and wage rates. Theextremelyuneven geographicaldistributionof unemployment may also

have been a factor tending to increasethe

rapidity of wage changesduring the upswing of business activity between 1934 and 1937.Increasesin import prices probably contributed to the wage increases

in 1940 and 1941. The points in Figure 9 for the remaining war yearsshow the effectiveness of the economic controls introduced. After anincrease in unemployment in 1946 due to demobilisation and in 1947due to the coal crisis, we return in 1948 almost exactly to the fittedrelation between unemployment and wage changes.

IV. 1948-1957

A scatter diagram for the years 1948-1957 is shown in Figure 10.The unemployment percentages shown are averages of tbp monthlyunemployment percentages in Great Britain during the calendar yearsindicated, taken from the Ministry of Labour Gazette. The Ministryof Labour does not regularly publish figures of the percentage unem-ployment in the United Kingdom; but from data published in theStatistical Yearbooks of the International Labour Organisation it

7/27/2019 8419964 a W Phillips the Relation Between Unemployment Xxx

http://slidepdf.com/reader/full/8419964-a-w-phillips-the-relation-between-unemployment-xxx 15/18

296 ECONOMICA [NOVEMBER

appearsthat unemploymentn the United Kingdomwas fairly con-sistentlyabout0 1percent.higher han hat n GreatBritain hroughoutthis period. Thewageindexusedwas the index of weeklywage rates,

published monthly in the Ministry of Labour Gazette, the percentagechange duringeach calendaryear being taken as a measure of theaveragerate of changeof money wage rates during the year. TheMinistrydoes not regularlypublishan index of hourly wage rates ;1butanindexof normalweeklyhourspublishedn theMinistry f Labour

11

10

lo-l

>' 9 |-Curve fitted to 1861-1913 data

S 8 56

@52

06

57

c 5o 50E '

0 3 53

fo

-c

O 2

ci:

0

0 1 2 3 4. 5

Unemployment, 1.

Fig. 10. 1948-1957

Gazetteof September1957shows a reductionof 0-2 percent. in 1948

andin 1949andan averageannualreductionof approximately04 percent.from1950to 1957. Thepercentage hangesn hourlyrateswouldthereforebe greaterthan the percentagechangesin weeklyrates bythese amounts.

It will be argued ater that a rapidrise in importprices during 1947led to a sharp ncreasen retailprices n 1948whichtended o stimulatewage increasesduring 1948,but that this tendencywas offset by the

1 An index of hourly wage rates coveringthe years consideredin this section is,however, givenin the Ministryof LabourGazetteof April, 1958.

7/27/2019 8419964 a W Phillips the Relation Between Unemployment Xxx

http://slidepdf.com/reader/full/8419964-a-w-phillips-the-relation-between-unemployment-xxx 16/18

1958] UNEMPLOYMENT AND MONEY WAGE RATES 297

policy of wage restraint introduced by Sir Stafford Cripps in the springof 1948 ; that wage increases during 1949 were exceptionally low as aresult of the policy of wage restraint ; that a rapid rise in import prices

during 1950 and 1951 led to a rapid rise in retail prices during 1951and 1952 which caused cost of living increases in wage rates in excessof the increases that would have occurred as a result of the demand forlabour, but that there were no special factors of wage restraint orrapidly rising import prices to affect the wage increases in 1950 or in

.51

, 10 -

-<--Curve fitted to 1861-1913 dataL),.

'? 1 R ~~~55

tm \ ~~~~526\

c 5\E 50

c 4 \o48

c\

3o 353

O- 92

O 2 3 4 5

Unemployment, of**

Fig. 11. 1948-1957, with unemploymentlagged 7 months

the five years from 1953 to 1957. It can be seen from Figure 10 that

the point for 1950 lies very close to the curve fitted to the 1861-1913

data and that the points for 1953 to 1957 lie on a narrow loop aroundthis curve, the direction of the loop being the reverse of the direction

of the loops shown in Figures 2 to 8. A loop in this direction could

result from a time lag in the adjustment of wage rates. If the rate of

change of wage rates during each calendaryear is related to unemploy-ment lagged seven months, i.e. to the average of the monthly per-centages of unemploymentfrom June of the preceding year to May ofthat year, the scatter diagram shown in Figure 11 is obtained. Theloop has now disappearedand the points for the years 1950 and 1953

7/27/2019 8419964 a W Phillips the Relation Between Unemployment Xxx

http://slidepdf.com/reader/full/8419964-a-w-phillips-the-relation-between-unemployment-xxx 17/18

298 ECONOMICA [NOVEMBER

to 1957 lie closely along a smooth curve which coincides almost exactlywith the curve fitted to the 1861-1913 data.

In Table 1 below the percentagechanges in money wage rates duringthe years 1948-1957are shown in column (1). The figures in column (2)are the percentage changes in wage rates calculated from the curvefitted to the 1861-1913 data corresponding to the unemployment per-centages shown in Figure 11, i.e. the average percentages of unemploy-ment lagged seven months. On the hypothesis that has been used inthis paper, these figures represent the percentages by which wage rateswould be expected to rise, given the level of employment for each year,as a result of employers' competitive bidding for labour, i.e. they

represent the demand pullelement in wage adjustments.

TABLE I

(1I (2) (3) (4)Change in Demand Cost Change inwage rates pull push import prices

1947 .. .. .. .. .. 20-11948 .. .. 3.9 35 . 71 . 106

1949 119 . 41 i 2.9 j4.1

1950 .. . 46 . 44 . 30 . 26*51951 . .10-5 . 5.2 . 9.0 .23*3

1952 .. . 6.4 . 45 . 93 .11i

1953 -3o 3-0 3.0 -4.8

1954 .. .. 4.4 .. 4.5 . 1.9 . 5.01955 .. 69 . 68 . 4.6 . 191956 .. 7.9 . 80 . 4.9 . 3.81957 .. 5.4 5... 2 . 3D s -7.3

The relevant figure on the cost side in wage negotiations is the per-

centage increase shown by the retail price index in the month in whichthe negotiations are proceeding over the index of the correspondingmonth of the previous year. The average of these monthly percentagesfor each calendar year is an appropriate measure of the cost pushelementin wage adjustments,and these averages' aregiven in column (3).The percentage change in the index of import prices2 during each yearis given in column (4).

From Table 1 we see that in 1948 the cost push element was consider-ably greater than the demand pull element, as a result of the lagged

effect on retail prices of the rapid rise in import prices during theprevious year,and the change in wage rates was a little greater thancouldbe accounted for by the demand pull element. It would probably havebeen considerably greater but for the co-operation of the trade unionsin Sir Stafford Cripps' policy of wage restraint. In 1949 the costelement was less than the demand element and the actual change in

ICalculated from the retail price index published in the Monthly Digest of Statistics.The figure for 1948 is the average of the last seven months of the year.

2 Board of Trade Journal.

7/27/2019 8419964 a W Phillips the Relation Between Unemployment Xxx

http://slidepdf.com/reader/full/8419964-a-w-phillips-the-relation-between-unemployment-xxx 18/18

1958] UNEMPLOYMENTAND MONEY WAGE RATES 299

wagerateswas alsomuch ess,no doubtas a resultof thepolicyof wagerestraintwhich is generallyacknowledged o have been effective in1949. In 1950 the cost elementwas lower than the demandelement

and the actual wage changewas approximately qual to the demandelement.Importpricesrose very rapidlyduring1950and 1951as a resultof

the devaluationof sterling n September1949 and the outbreakof theKoreanWar n 1950. Inconsequencehe retailprice ndexroserapidlyduring 1951 and 1952so that the cost elementin wage negotiationsconsiderably xceeded he demandelement. The actualwage increasein each year also considerably xceededthe demandelement so thatthese two yearsprovidea clear caseof cost inflation.

In 1953thecostelementwasequalto the demandelementand in theyears 1954 to 1957 t waswell belowthe demandelement. In each ofthese years the actual wage increasewas almost exactly equal to thedemand element. Thus in these five years, and also in 1950, thereseemsto havebeenpuredemand nflation.

V. CONCLUSIONS

The statisticalevidence n SectionsTIto IV above seems in general

to support he hypothesis tated n SectionI, thatthe rateof changeofmoney wagerates can be explainedby the levelof unemployment ndthe rate of change of unemployment, except in or immediately afterthose years in which there is a sufficiently rapid rise in import prices tooffset the tendencyfor increasingproductivity o reducethe cost ofliving.

Ignoringyearsin whichimportpricesrise rapidlyenoughto initiatea wage-price piral,which seemto occurvery rarelyexceptas a resultof war,andassuminganincrease n productivity f 2 percent.peryear,it seems from the relationfittedto the data that if aggregatedemandwerekept at a value whichwould maintaina stable level of productpricesthe associated evel of unemploymentwould be a little under2~ per cent. If, as is sometimesrecommended, emandwerekept ata value which wouldmaintain tablewagerates the associated evel ofunemploymentwouldbe about51percent.

Becauseof the strongcurvature f the fittedrelation n the regionoflow percentageunemployment,herewill be a lower averagerate of

increaseof wagerates f unemployments heldconstantat a given evelthan therewill be if unemployments allowedto fluctuateabout thatlevel.

These conclusionsare of coursetentative. Thereis need for muchmore detailed research into the relations between unemployment, wagerates,pricesandproductivity.

The London School of Economics.