Professional Documents
Culture Documents
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 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.
Royal Statistical Society and Blackwell Publishing are collaborating with JSTOR to digitize, preserve and
extend access to Journal of the Royal Statistical Society. Series A (Statistics in Society).
http://www.jstor.org
J. R. Statist.Soc. A (1994)
157, Part 2, pp. 253-270
SUMMARY
The validityof UK unemployment statisticswas the focus of intensedebate throughthe
1980s and again in 1992. This mainlyconcernedwhethertheywere open to politically
motivatedmanipulation.To some extentthisargumenthas misseda widerpoint.Frequent
changes of coverage and consistencybetween sources of informationwhich provide
measuresof unemploymenthave led to a paucity of data that are suitable for social
scientificstudywithits impliedcoststo our understanding of unemployment overthelast
15 years.It is hoped thatan increasingrelianceon specificallydesignedsurveytechniques,
ratherthanmeasuresbased on benefitadministration data, will overcomemanyof these
problemsfor the 1990s. The design of the Labour Force Surveyshould take on board
the questions raised by social scientists.In particular,the shortageof available work
shouldbe measuredbythebroaderindicatorsof social distressas wellas theidentification
of excess labour supplyin the labour market.
Keywords: DISCOURAGED WORKERS; STATISTICS; UNEMPLOYMENT
1. INTRODUCTION
For the last decadethe officialmonthly countof the unemployed has beenthe
subjectof seriouscontroversy.Thiscontroversy on theimpactthatthe
has centred
frequent,mostlyminor,changesin coveragehavehad on theheadlinenumberof
unemployed. Thisdebateoverthe'true'measureofthecountorwhatitwouldhave
beenontheoldbasishastendedto obscurewhatforeconomists is perhapsthemore
important issue:thattheclaimantcountis notand neverhas beena measureof
excesslaboursupply.It is rather
a by-productofan administrative
systemofbenefit
assessment andhencevulnerable to changesinthatsystem.Thechangesincoverage
of theclaimantcountoverthelast decadehave onlymadethisdistinction more
obvious.Chargesmadeagainstthecountmethodofestimating unemployment have
concerned allegationsaboutpolitically
inspired
manipulationofthefigures through
numerous changesin coverageduringthe1980s.The supporting evidenceforthese
chargeshas been thatall but one of thesechangeshave been unidirectional-
downwards. Criticshave expresseda viewthatcalculating
unemployment on the
old (pre-1982)coveragewouldresultin a considerablyhighertotalfigure, suchas
that of the Unemployment Unit (UU) index.
Here we argue that such a debate is somewhat fruitlessand that it misses the
damage that has been done to our understanding of unemployment through the
4000T
3500t
3000-
2500-
2000 /
1000-A._
500- 4
71 73 75 77 79 81 83 85 87 89 91
4000
3500-.
3000- -/
2500- \ '.- /
2000- \
82 83 84 85 86 87 88 89 90 91
tSource,theUU.
tEmployment Department's fromthechangeincounting
ofthesizeof theeffect
estimate ofunemploy-
methods
mentfiguresgivenin Table 1.
of theunemployed.
characteristics Thesedata sourcesfallintotwotypes-cohorts
and surveys.Therehavebeenperiodiccohortswithin theclaimantcountin thelast
15 years.The thirdis currently beingconstructed. This monitorstheprogression
of all newclaimants in a certainweekforup to a year,eveniftheyhaveleftthe
count.Hence,it providesvaluableinsightsintopropensities to leave the count,
the employment statethatpeople exitinto and the likelihoodof re-entry into
unemployment (see Narendranathan et al. (1985)).
260 GREGG [Part 2,
The mostimportant surveyforinformation on unemployment is theLFS. The
LFS surveys theemployment stateof 60000households in thespringof eachyear.
As it is drawnfromthe wholeof the Britishpopulation,it can assess search
unemployment, searchintensity,previouseconomicstate,occupation as wellas age,
sex,raceand broadregion.The mainproblemabouttheLFS was thatuntil1992
it had been onlypublishedannuallyand forthe yearsbefore1984it was only
biennial.Thereare also variations in thequestionsasked,whichmakea historical
seriesproblematical. Furthermore, itis subjectto theerrorsassociatedwithsample
surveys and therewas a year'sdelaybeforetheresultsbecameavailable.The great
advantageof the surveyapproachis thatit can measureunemployment on an
internationally acceptedbasisand as itis nottheindirectresultofthemachinations
of thebenefitsystemit is freefromany concernsof manipulation.
Othersurveysare also availablewhichprovideusefulinformation suchas the
FamilyExpenditure Surveyand theGeneralHouseholdSurvey.Thesesourcescan
addressissuesofbenefit uptakeandinterrelations betweenindividuals inhouseholds
(see forinstanceDilnotand Kell (1987) and Kell and Wright(1990)). The most
comprehensive surveyis thecensusof populationwhichcoversthewholepopula-
tion.Suchcomprehensive data are usefulin generatingweights to enableaggrega-
tionof othersurveysup to nationallevel.
Most countrieshave both surveyand registrant measuresforunemployment
butin thosewithmonthly surveys (USA, Canada, Australiaand Japan)relatively
littleuseis madeoftheregistrant measure.Amongthosewithlessfrequent surveys
the registrant measuretakes on a greaterrole and is based on registration at
government employment agencies.The UK is theexception, as it usesentitlement
to unemployment-related benefits. This meansthatthegap betweentheadminis-
trativemeasureand thesurveymeasureis muchlargerforwomenin theUK than
in othercountries.So, whereasin mostcountriesthe registrant countis higher
thanthe survey-based searchmeasure,forthe UK thisis not normally so. The
Netherlands is also unusualin thisrespect(see Whiteand Leyland(1992)formore
details).
spring1992;source,calculations
t Britain, madefromEmployment Gazette,November1992,p. 537,andLabour
ForceSurveyQuarterly Bulletinno. 1 and unpublished
LFS data.
t ILO-OECD definitions.
?Thiscategoryexcludesthoseseeking workon ILO definitions
butincludes thoseseeking
workbutunableto start
in thenext2 weeks.
??Unpaidfamily members includedin notwanting work.
androseagainsubsequently.
Thisimpliesthatthesocialimplicationsoftheshortage
of workmayneeda broaderanalysisthanconventional unemployment measures.
Thispointis further
broughtoutbytheinformation on changesof economicstate
between1990and 1991containedin theLFS. Amongthepopulationof working
age, about24%Vo
of womenwhowereinactivein 1990wereactivein 1991(18%Vo in
employment) and20%Vooftheunemployed wereinactivea yearlater.Formen,26%Vo
of thosewhowereinactivehad becomeactive(19%Vo in employment) and 13%Vo
of
thoseunemployed a yearbeforewerenowinactive.Hence,thedistinction between
inactivity
and unemployment is oftenslight.
Category Participation
ratet Claimants? Claimants?? Claimantsnotwanting
searching as % of wantingas % of as % of all not
all searching all wanting wanting
Male 73.8 80 34 4
Female 52.8 44 9 2
All 62.9 66 17 2
tBritain,spring1992.
tTheparticipationrateis thoseworking
or searching
on ILO definitions
as a percentage
oftheadultpopulation.
? ILO-OECD definitions.
??Thiscategory excludesthoseseekingworkon ILO definitions
butincludesthoseseekingworkbutunableto
startin thenext2 weeks.
264 GREGG [Part2,
Althoughtotalsfortheclaimantcountand theILO-OECD measureof those
seekingworkare similar(2.61 millionand 2.65 millionrespectively), thisobscures
thefactthatthenumber ofpeoplebothclaiming and searching was onlytwo-thirds
of all thoseclaiming.Thismeansthatabout850000claimants werenotsearching
forwork(or werein employment) and 890000weresearching forworkbut not
claimingbenefits.
Furthermore, as Table 6 shows,formalesmostof thosesearching areclaimants
butforwomenonlyaround40% claim.Claimants also constitute
a sizableminority
of thosenotsearching forworkbutwantingit, buta relatively smallproportion
ofthosenotwanting workat all. Acrossall thesecategories malesarealwaysmore
likelyto be claimants thanfemales.Therefore, evenifthetotalfigures arebroadly
similar,disaggregations, by sexin particular, are highlyunreliable. Womenmake
up aroundtwo-thirds of thosesearching forworkbutnoteligibleforbenefits, and
menconstitute two-thirds of thoseclaimingbut not searching forwork.
Consequently thereis a largegroupof peoplewho are commonto boththe
unemployed claimant countanda measureofthosewhoaresearching. Thisensures
thatthetwomeasuresof theunemployed willnormally movein thesamedirection
but the differences are sufficientlyimportant to producesubstantial divergences
between thetwoseries;see Table 7. Hence,in 1984an estimate of theILO-OECD
measureof theunemployed searching forworkwas about340000higherthanthe
claimant count.By 1986thishad beenerodedso thattheclaimantcountwas 10000
higher andreversed againby1990and theILO-OECD measurewas370000higher.
The NationalInstitute (1983) reportedresultsfromthe 1971 and 1981censuses
whichindicated thatsuchinstabilities also existedinthe1970s.Theclaimant register
roseby 1.68 millionbetweenthesetwo dateswhereasthecensusesmeasuredan
increaseof 1.31millionon theold labourforcedefinition of searchunemployment
(broadlyspeakingthismeasuredactivesearchin thelast2 weeks).Theyattributed
mostof thedifference overthisperiodto an increasein thepropensity of women
to register.
The relationship betweentheclaimantcountand themeasureof searchunem-
ployment hasbeenhighly unstableinthelate1980s.No doubtthisis inpartbecause
therestart initiativeetc. encouragedsearchactivity amongclaimants,but it still
impliesthateventheconsistent seriesof thecounthas proveda poorguideto the
rateof changeof excesslaboursupplyoverthisperiod.However,it remainsa
reasonableindicatorof thedirection of change.For instance,therewere900000
TABLE 7
ILO-OECD measuresfor Britaint
and claimantunemployment
ILO-OECD 3.11 2.98 2.98 2.89 2.38 1.98 1.87 2.30 2.65
Claimant 2.77 2.91 2.99 2.81 2.28 1.75 1.50 2.05 2.59
ILO - claimant 0.34 0.07 -0.01 0.08 0.10 0.23 0.37 0.25 0.06
7. LOOKING TO FUTURE
The futureis not as barrenas the past, forwe now have an opportunity to
overhaultheunemployment statistics. manyofthe
Thiswillenableus to overcome
pastproblems.The principalsourceof newinformationforthisprocesswillbe the
new quarterlyLFS. Fromthe springof 1992the previously annualpublication
266 GREGG [Part2,
becamequarterly, witha muchshorter timelag beforepublication, and thisallows
for,amongotherthings,thefrequent measurement of searchunemployment. In
addition,it offersan opportunity to put the unemployment statisticsbeyond
suspicionof politicalintervention.
Changesin themonthly claimantcountcan be usedas firstapproximations of
changesin searchunemployment on internationally accepteddefinitions. Subse-
quently thesewouldbe revised eachquarter whentheLFS results comein.Therefore
any discontinuity betweenfigureswill onlyhave a shortrunimpact.This is a
commontechnique in deriving dataandis alreadyusedbytheOECD and
statistical
EEC in ascertaining unemployment estimates fortheUK. This techniqueshould
enablethe creationof a back seriesat leastuntil1984. (An estimateof search
unemployment was includedin theLFS before1984buton a different definition
of searchin thelast 2 weekscalledthe'GB labourforcemeasure'.)
Furthermore, overtimethequarterly LFS willprovidetimeseriesmeasuresof
mostoftheissuesin whichwe areinterested, concerning theincidence and concen-
trationof unemployment. It willcoverall aspectsof the sex, age and previous
occupationcharacteristics of theunemployed, as wellas unemployment duration
andmeasures ofintensityofjob search.(In additionto thenewquarterly LFS from
1994the new computerized claimantcountserieswill also providemeasuresof
occupational unemployment forthisseries.)The remaining issuesof flowsbetween
states,exitprobabilities
andre-entry ratescanbe addressed byoneotherkeyfeature
of thenewLFS: a movingpanel in whichindividuals willappearin up to four
consecutive LFS samples.In thiswaytheirprogresscan therefore be monitored.
Finally,as a surveyofall individuals and notjustclaimants, itcan provideinsights
intopeople'scharacteristics whennot in workand evenwhennot searching for
work.(Seasonalvariations intheILO-OECD measureofunemployment havebeen
computed byusingunpublished quarterly data availablesince1984.Although the
samplesizesinvolvedwereconsidered too smallto warrantpublication, theyare
deemedsufficient to produceseasonaladjustments forrelevantseries.)
Thereare,however,severalproblems(at leastin theshortrun)associatedwith
themoveto thequarterly LFS. First,sampling variationina survey measuremeans
thattherewillbe a marginof erroraroundthecentralestimateforany chosen
measure.Thiswilltakecarefulmonitoring in theshorttermuntilthelikelybounds
forerrorbecomeestablished. Second,theshorttimeseries,evenif data back to
1984areused,meansthatrelatively littletimeseriesanalysisis possible.Thismakes
thedistinction betweencyclicalstructural changesproblematical in an analysisof
thedata,thoughthiscan be partlyovercome byusingthecross-sectional variation.
Thisnewsourceof information raisesanotherpotentialinsight intounemploy-
ment.In 1976JuliusShiskin,a former US Commissioner forLabour Statistics,
raisedthe idea of alternative measuresof unemployment for different needs.
He advocatedtheestablishment of severalalternative measures,each designedto
be of use in different circumstances (Shiskin,1976). None of thesewouldbe a
measureof trueunemployment becauseof a lack of a uniquedefinition of 'true
unemployment'. This has beenadoptedin theUSA (see Sorrentino (1993)).
Sucha system was advocatedfortheUK byMiller(1988)and is currently used
in the UK formoneysupplyfigures(MO, Ml etc.). Indeednotationcould be
borrowed fromthesedata. A revivalof thisidea is opportune, giventhenewdata
sourceand the conflictbetweenunemployment statistics
as measuresof excess
1994] UNEMPLOYMENT STATISTICS IN UK 267
TABLE 8
Benefitrecipientsnot includedin the claimantcount, 1980-91t
iS
1991 616 1130 1746
1989 526 981 1507
1986t 368 732 1100
1980t 138 432 571
Sicknessand invalidity
1991 1020 472 1492
1989 891 374 1265
1986 807 267 1076
1980 800 255 1055
tMen 16-64 yearsold, women 16-59 yearsold; source,Annual Abstractof Statistics,1993, and
Social SecurityStatistics,various years.
t Supplementary benefitratherthan IS applied in theseyears. Claimantstatusis as definedat the
point in time.
268 GREGG [Part2,
counthaverisensteadily throughout the1980sbyaround1.2 million.Not all these
are availableor wanting to workbut,althoughsocialtrendstowardssingle-parent
familieswillincreasetheIS dependence, it is difficult
to believethatsicknessand
disabilityare risingso dramaticallyin theUK. The mainreasonforthisincrease
is thelongerdurationsof claimsas outflowsintoemployment havedeclined(see
Disneyand Webb (1991)).
A measureof widersocial distressdue to shortagesof availableworkshould
includeall familymembers wherean absenceof workor a shortfall in thedesired
amountof workby one or moremembers is causinglow householdincome.This
wouldthenexcludethosefamilieswhereone memberis out of workwhileothers
have earnedincomesand would includeall thoserequiring benefits(e.g. non-
claimantIS, family and sicknessbenefit)
credit,invalidity to raisetheincomelevel
fromthelowbasecausedthrough lackofavailableworkforoneormoremembers.
This measurewouldbe calculablefromavailablesources,althoughtheywould
requiredetailsof incomeand otherextraquestionsin theLFS. Wherever possible,
disaggregations in linewiththecriteriain Section2 shouldbe made.
8. CONCLUSIONS
In the 1980stherehas been a deterioration in the usefulnessin the monthly
claimant countas a sourceofinformation usedforstudying majorissuesconcerned
withunemployment. Although theGSS has notbeeninvolvedin anyunreasonable
manipulation thishas notmeantthatthedatahavebeenfreefrom
ofthestatistics,
potentialpoliticalinterference.This is becausethe raw data are not underthe
controloftheGSS butdependon theadministrative rulesforentitlement to benefit
in forceat any pointin time.This is at least partlydetermined by Ministerial
decisions.
Ratherthanlookingbackto a previousincarnation of thecount-whichwas in
itselfa poor measureof underlying unemployment-we shouldgo forwards to
makingthe new quarterlyLFS the primesource for answeringquestionson
unemployment. Thisdatasourceproducesunemployment whicharetotally
statistics
independent.
The claimantcountis usefulforits speed of production, forits local detail
and measurement of dependenceon benefits.However,it should not be the
principalsourceof data whenwe are trying to understand unemployment and its
effects.The new quarterly LFS providesan opportunity forthe GSS to regain
in relationto unemployment
credibility data. It also offersthepotential to answer
broaderquestionsof interestto social scientistsand others,withrespectto
unemployment.
Although theclaimantcountshouldcontinue,it shouldbe replacedas thekey
measureof excesslaboursupplyin theeconomyby a searchmeasureon interna-
tionaldefinitions.Thiswouldusefully be supplemented by a broadermeasureof
theunsatisfieddemandforworkand a measureof theextentof distress causedby
absence of work. None of thesemeasuresalone wouldascertainthelevelof true
unemployment, forin realitythereis no suchuniquemeasure.Insteaddifferent
indicatorswouldprovideinformation on differing aspectsof theeffectsof lackof
workon theUK's economyand society.
1994] UNEMPLOYMENT STATISTICS IN UK 269
ACKNOWLEDGEMENTS
I would like to thank Paul Allin, AndrewBritton,the Editor and the referees
forcommentson a previousdraftof thispaper. All the seriesin Figs 1 and 2 were
providingby Dr David Taylor of the UU.
APPENDIX A
fromtheEmployment
are someusefulreferences
The following the
Gazetteconcerning
monthlyunemployment withthe LFS measuresof unem-
count(includingcomparisons
ployment).
May 1980 'A reviewof unemployment and vacancystatistics'
September1982 'Compilationof theunemployment statistics'
December1982 'Changedbasis of theunemployment statistics'
June1983 'The unemployed: surveyestimates for 1981comparedwiththe
monthly count'
August1984 'The unemployed: surveyestimates for 1983comparedwiththe
monthly count'
July1985 'Unemployment adjustedfordiscontinuities and seasonality'
October1985 'Unemployment: estimatesfromthe LFS comparedwiththe
monthly claimantcount'
1986
March-April 'Change in the compilationof the monthlyunemployment
statistics'
October1986 'Unemployment figures:theclaimantcountandtheLabourForce
Survey'
January1988 'Measuresof unemployment of the un-
and the characteristics
employed'
October1988 'Measuresof unemployment of the un-
and the characteristics
employed'
December1988 'Unemployment statistics:revisionsto the seasonallyadjusted
series'
August1989 'Measuresof unemployment: claimantcountand Labour Force
Survey'
October1990 'Measuresof unemployment'
December1990 'Monthly unemployment statistics:
maintaining a consistent
series'
November1991 'Measuresof unemployment: claimantcountand Labour Force
Survey'
July1992 'Measuresof unemployment: claimantcountand Labour Force
Survey'
September1992 'How unemployment is measuredin othercountries'
June1993 of theILO unemployed'
'Characteristics
REFERENCES
Dilnot,A. and Kell, M. (1987) Male unemployment and women'swork.Fisc. Stud., 8, no. 3,
1-16.
Disney,R. and Webb,S. (1991)Whyare thereso manylongtermsickin Britain?Econ. J., 101,
252-263.
Employment (1992)Figuring
PolicyInstitute Econ. Rep., 7, no. 2, Sept.
out unemployment.
Hussmans, standards
R. (1990)International of economicactivity,
on themeasurement employment,
unemployment and underemployment.In Developments LabourStatistics
inInternational (ed. R.
Turvey).London:Pinter.
270 GREGG [Part 2,
InternationalLabour Organization(1990) Surveysof EconomicallyActivePopulation, Employment,
Unemployment an ILO Manualof ConceptsandMethods,p. 107.Geneva:
and Underemployment:
InternationalLabour Office.
Kell, M. and Wright,J. (1990) Benefitsand thelabour supplyof womenmarriedto unemployedmen.
Econ. J., 100, 119-126.
Keynes,J. M. (1937) The generaltheoryof employment.Q. J. Econ., Feb.
Knight,K. G. (1987) Unemployment:an Economic Analysis. London: Croom Helm.
Lawlor, J. (1990) Monthlyunemployment statistics:maintaininga consistentseries.EmplymntGaz.,
98, 601-608.
Metcalf,D. (1984) On the measurementof employmentand unemployment. Natn. Inst. Econ. Rev.,
no. 109, Aug., 59-67.
Miller,R. (1988) The End of Unemployment.London: Atlas Economic Research Foundation.
Narendranathan,W., Nickell,S. and Stern,J. (1985) Unemployment benefitsrevisited.Econ. J., 95,
307-329.
National Institute(1983) Unemployment.Natn. Inst. Econ. Rev., Nov., 39-47.
Shiskin,J. (1976) Shiskinon the unemploymentnumbers.New York Times, Jan. 18th.
Sorrentino,C. (1993) Internationalcomparisonsof unemployment indicators.MthlyLab. Rev., Mar.,
3-24.
Taylor, D. (1984) The unemploymentnumbers game. UnemploymentUnit Bulletin 11, Jan.
UnemploymentUnit, London.
(1986) Creativecounting.Unemployment Bulletin20, summer.Unemployment Unit, London.
The Times (1991) The Times, Apr. 19th.
Wadsworth,J. (1991) Unemployment benefitsand searcheffortin theUK labour market.Economica,
58, 17-34.
White,A. and-Leyland,J. (1992) How unemployment is measuredin differentcountries.Emplymnt
Gaz., 100, 421-433.
G. D. N. (1976)TheConceptand Measurement
Worswick, of Involuntary Boulder:
Unemployment.
West View.