The Development and Analysis of Economic Indicators · Moore: The Development and Analysis of Economic Indicators 229 asked the National Bureau of Economic Research a private, nonprofit

The Developmentand Analysisof Economic Indicators *

Geoffrey H. Moore Centerfor InternationalBusinessCycleResearch,GraduateSchoolofBusiness,Columbia University

INDICATORS AND BUSINESS CYCLES

Businesscycles, large and small, appearto be a continuing feature of theeconomiclandscape.A turn up or down in the economyis clearly an eventof major social significance.Considerableinterest therefore attachesto themeanswhereby an economicturn can be forecast and its extent can beestimated.That is the roleof economicindicators, which reston thenumerousmeasurementsof the pulse of the economymadeby governmentagencies,privateorganizations,andindividual economists.Theanalysisof economicindicatorsis a well-developedtechniquefor ascertainingwhat the many pulsereadingsare saying about the stateof the economy.

* This essayis adaptedandupdatedfrom theauthor’sarticle, "The Analysis of EconomicIndicators,"publishedin ScientificAmerican, 232:1 January1975 copyright 1974 by ScientificAmerican,Inc., all rights reserved.

227

228 PART THREE: OUR SOCIAL WORLD

Economicindicators havecome to embracevirtually all the quantitativemeasuresof economicchangethat are continuouslyavailable.One can finddaily, weekly, monthly, and quarterly indicators; they measureproduction,prices, incomes,employment,investment,inventories,sales,andso on; andthey recordplans,commitments,and anticipationsas well as recenttransactions. Someof the indicators,suchas theunemploymentrateor theConsumerPriceIndex, are calculatedby theFederalGovernmenton thebasisof elaboratesamplingsurveysconductedeachmonth.Others,suchas the indexesof stockmarketpricesand the surveysof purchasingagents’ views of prices,orders,andinventories,are constructedby privateorganizationson thebasisof information they collect or obtain as a by-product.

As a result economistsor businesspeopleinterestedin forecastingchangeare faced,like weatherforecasters,with a massof factual informationthatpoursin constantly.They mustassessin somesystematicway what the informationsaysaboutthepresentandthefuture. Thetechniqueof indicator analysisembracesvarious systematicways of looking at this information with a view toassessingthe presentsituationand discerningsignificant futuredevelopmentsin the businesscycle.

Oneof the earliestsystemsof the kind, devisedshortly beforeWorld War I,cameto be known as theHarvardABC curves.The A curve wasan indexrepresentingspeculation,more specifically stock prices. The B curve representedbusinessactivity, measuredby the dollar volume of checks drawn on bankdeposits.The C curve representedthe moneymarket, measuredby the rateofintereston short-termcommercialloans.Historical studies,particularly thosecarriedout by WarrenPersonsat HarvardUniversity, showedthat thesethreecurves typically movedin sequence:stock pricesfirst, bank debitsnext, andinterestrateslast, with thelagging turns in interestratesprecedingtheoppositeturns in stock prices.The economiclogic of the sequencewasthat tight moneyandhigh interestratesled to adeclinein businessprospectsanda drop in stockprices,which led to cutbacksin investmentand a recessionin business.Therecessionin turn led to easiermoneyandlower interest rates, which eventually improvedbusinessprospects,lifted stock prices,andgenerateda newexpansionof economicactivity.

The systemcameto grief in theGreatDepressionof 1929 becausethe interpretersof the curvestooktoo optimistic a view andfailedto foreseethedebacle.Economistsgenerallyregardthe episodeas oneof thegreatforecastingfailuresof all time.

Curiously,however,thesequenceof eventson which thesystemwasoriginally basedhasin largemeasurepersisted.This is not to saythat theABC curveswould still suffice if they wererevived. Far more comprehensivesystemsofindicatorshavebeendevelopedsince 1929, andthe empiricalandtheoreticalbaseon which theystandhasbeenmorethoroughlystudied,documented,andtested.

Thesharprecessionof 1937-1938,which occurredbeforethe economyhadfully recoveredfrom the GreatDepression,helpedto spur that development.In the autumnof 1937,HenryMorgenthau,Jr., the Secretaryof the Treasury,

Moore: The DevelopmentandAnalysis of EconomicIndicators 229

askedthe NationalBureauof EconomicResearcha private,nonprofit researchagencyto devisea systemof indicatorsthat would signalwhenthe recessionwasnearingan end.At that time the quantitativeanalysisof economicperformancein the United Statesdid not approachtoday’s standards.The Government’snational incomeandproductaccounts,which form the foundationofmuch of moderneconomicanalysis,werejust being established.Othervitaleconomicstatistics,including unemploymentrates,were being developedorrefinedby public agenciestrying to provide information that would be usefulin fighting the Depression.Few statisticalserieswere issuedin seasonallyadjustedform, asthey are now. Comprehensiveeconometricmodelssystemsofequationsexpressingquantitativerelationsamongeconomicvariables,whichare widely employednow to forecastthe economyand to evaluateeconomicpolicies, were virtually unknown then.

Underthe leadershipof WesleyC. Mitchell andArthur F. Burns,theNationalBureauof EconomicResearchhad since the 1920s assembledand analyzedmonthly, quarterly,andannualdataon prices,employment,production,andother factors as part of a major researcheffort aimed at gaining a betterunderstandingof businesscycles.This projectenabledMitchell and Burns toselectanumberof seriesthat, on thebasisof pastperformanceandof relevancein thebusinesscycle,promisedto be fairly reliableindicatorsof businessrevival.The list wasgiven to the TreasuryDepartmentlate in 1937 in responseto Morgenthau’srequestandwaspublishedin May 1938.Thus originatedthe systemof leading,coincident,andlaggingindicatorswidely employedtoday in analyzing the economic situation, determining what factors are favorable or unfavorable, and forecastingshort-termdevelopments.

Since 1938 the availability and the use of economicindicatorshavebeengreatly expandedunder the leadershipof the National Bureauof EconomicResearch,the U.S. Departmentof Commerce,the Organizationfor EconomicCooperationandDevelopmentin Paris,andotherpublic andprivateagencies.The list of indicatorsassembledin 1937 wasrevisedin 1950, 1960, 1966, and1975 to take into accountnew economicseries,new researchfindings, andchangesin the structureof the economy.A new evaluationhasrecentlybeenconductedby the Centerfor InternationalBusinessCycle Researchat Columbia University.With eachrevisiontheperformanceof theindicatorsbothbeforeandafter the dateof their selectionhasbeencarefully examinedand exposedto public scrutiny.

In 1957RaymondJ.Saulnier,who wasthenchairmanof the President’sCouncil of EconomicAdvisers, askedthe Bureauof the Censusto developmethodswherebythe appraisalof current businessfluctuationscould take advantageof the large-scaleelectronicdataprocessingthat wasbecomingavailable,withthe resultsto beissuedin amonthly report. Experimentalwork doneoverthenext few years under the leadershipof Julius Shiskin, who was the chiefeconomicstatisticianof the Bureauof the Census,resultedin 1961 in themonthly publication by the Department of Commerceof BusinessCycleDevelopments. It is now called Business Conditions Digest; under bothnameseconomistsrefer to it asBCD. Thispublicationhasgreatlyincreasedthe

230 PART THREE: OUR SOCIAL WORLD

accessibilityof currentindicatordataandof variousstatisticaldevicesthat aidin their interpretation.As a result the indicatorshavebecomeamajor economicforecastingtool.

USEFUL QUALITIES OF INDICATORS

As notedabove,the analysisof economicindicatorsrestson bothanempiricalanda theoreticalfoundation.The selectionof particularindicatorsandtheemphasisgiven to them havebeenguidedby what is understoodof the causesof businesscycles.Obviously onewould wish to examinerecentchangesinany economicprocessthat is believedto play a significant role in anywidelyacceptedexplanationof cyclical fluctuations.

Many differentexplanationshavebeenadvancedfor thesefluctuations.Someof them placeprimarystresson the swingsin investmentin inventory andnewplant and equipmentthat both determineandare determinedby movementsin final demand.Othersassigna centralrole to the supplyof moneyandcredit,or to Governmentspendingandtax policies,or to relationsamongprices,costs,and profits.

All thesefactorsundoubtedlyinfluencethe courseof businessactivity. Someof them may bemoreimportantat a given time thanothers.No consensusexists,however,on which is the mostimportantor evenon how they all interact.Henceit is prudentto work with a variety of indicatorsrepresentinga broadrangeof influences.Readyaccessto awiderangeof indicatordataenablesoneto test competing or complementaryhypothesesabout current economicfluctuations.

With thisprinciple in mind, economicactivitiescanbe classifiedinto a fewbroad categoriesof closely related processesthat are significant from thebusinesscycle point of view. Indicatorshavebeenselectedfrom eachgroup.The principalcategoriesnow includedin BusinessConditionsDigestareshownin the lefthandcolumn of Table 1. Note that thesecategoriesdo not includeall aspectsof the economy.For example,statisticson agriculture; Federal,state,andlocal government;foreign trade;andpopulationandwealthare omitted.Nevertheless,the categoriesdo providea frameworkof factorsthat enterintotheoriesof the businesscycle andare important in assessingthe performanceof the economy.The omittedcategoriesare important, too, butsupply few indicatorsthat are systematicallyrelated to the businesscycle.

Within eachcategory,researchon businesscycleshasuncoveredindicatorsthat behavein a systematicway. Thesefindingshaveprovideda basisfor selecting particularindicatorsand classifyingthem accordingto their characteristiccyclical behavior,as in Table 1. Two of the chief characteristicsone looks forare the regularity with which the indicator conformsto businesscyclesandthe consistencywith which it leads,coincides,or lags at turningpoints in thecycles. Other relevantconsiderationsare the statisticaladequacyof the datasince the statisticalunderpinningof an indicator has a bearingon how wellthe indicator representsthe processit is supposedto reflect, the smoothness

Jlible 1 Cross-classificationof economicindicators

Relation to BusinessCycleEconomicProcess Leading RoughlyCoincident Lagging

Employmentand Averageworkweek Total employment Long-durationunemployment and overtime unemployment

Hiring and layoff ratesNew unemployment

insuranceclaimsProductionand Real GNPincome Industrial production

Personalincome

Consumption,trade, New orders,con- Retail salesorders, and deliveries surnergoods Manufacturing and

Vendor performance tradesalesFixed capital New investmentcorn- Backlog of investmentinvestment mitments commitments

Formationof business Investmentexpendienterprises tures

ResidentialconstructionInventoriesand Inventory investment Inventorylevelsinventory investment andpurchasing

Prices,costs,and Changein industrial Changein producer Changein consumerprofits materialsprices prices prices

Stockprices Changein unit laborProfits andprofit costs

marginsMoney andcredit Money andcredit Outstandingdebt

flows InterestratesCredit delinquencies

andbusinessfailuresBond prices

232 PART THREE: OUR SOCIAL WORLD

of the datasincehighly erraticseriesare difficult to interpretcorrectly, andthepromptnesswith which the figures arepublishedsinceout-of-datefigureshavea limited bearingon the current situation.

Empirical measuresof thesecharacteristicshavebeenconstructedfor largenumbersof indicators.Such measureshavebeenemployedin the attempt toobtain data capableof conveyingan adequatepicture of the changesin theeconomyasit movesthrough stagesof prosperityand recession.In addition,thebehaviorof the indicatorsafter theyhavebeenselectedhasbeenmonitoredclosely.Many of theindicatorshavesurvivedseveralsuccessiveevaluations.Forexample,measuresof theaverageworkweek, constructioncontracts,andstockpriceshavebeenon every one of the successivelists of indicatorsthat havebeendrawn up by the NationalBureauof Economic Researchsince 1937.

The samelists of indicatorshavealso beentestedby their performanceinothercountries.Every new recessionor slowdown, whether in this countryor abroad,providesadditional evidenceagainstwhich the indicatorscan beassessed,as doesevery upturn. This examinationand reexaminationhas accumulateda largeamount of empirical evidencethat demonstratesboth thevalue of the indicatorsand their limitations.

SEASONAL ADJUSTMENT AND SMOOTHING

A samplingof this evidenceis containedin the accompanyingillustrations.Letusconsideran indicatorsuchas thenumberof buildingpermits issuedfor newhousesseeFigure 1. The raw dataare statistically decomposedin order tomeasureandeliminateregularseasonalvariationsthatare repeatedevery year.Whenthis factor is removed,the indicator revealsmuchmore clearlythe tendencyfor permits to diminish during the recessionsof 1980 and 1981-1982and to increasewhen more prosperousconditions returnedin 1983.

Most economicindicators today are available in seasonallyadjustedform.Someof them are seldomreportedin any otherway. Examplesof indicatorsthat are invariablyadjustedfor seasonalfactorsinclude the grossnationalproduct, the unemploymentrate, andthe index of industrialproduction. Oneofthe computerprogramsthat is widely used to make seasonaladjustments,developedby the CensusBureau, is calledthe X-11 program.Seealso thearticle by Leon and Rones for a discussionof seasonaladjustment.

The smoothingof irregularmovementsis less commonlypracticedbecausethe techniquesare somewhatless routine. Certain statisticalseries,however,are subject to much wider irregular movementsthan othersbecauseof differencesin samplingerror or in the effectsof suchfactors as unusualweatheror labor disputes.It is thereforeuseful in interpretingcurrentchangesto havea standardmeasureof the size of theseirregular fluctuationscomparedwiththe size of the movementsthat reflect long-termtrendsandthe eventsof thebusinesscycle, which are often called trend-cycle movements.

Two measuresof this kind are provided for all the indicatorscarried inBusinessConditions Digest. Oneshowshow largetheaveragemonthlychange

Moore: The Developmentand Analysisof EconomicIndicators 233

‘I’0x

80

Figure 1 Building permitsfor new houses,before andafter seasonaladjustment.Source:Departmentof Commerce.

in the irregularcomponentis with respectto the averagemonthly changeinthe trend-cyclecomponent.The othershowshow many monthsmustelapseon the averagebeforethe changein the trend-cyclecomponent,which buildsup overa periodof time, exceedsthe irregular component,which doesnot.For example,the measuresshow that monthly changesin housingstartsarelikely to be dominatedby "noise" such as randomvariation, but that whenthesechangesaremeasuredoverspansof four months,the trend-cycle"signal"becomesdominant.On the other hand,the index of industrialproduction ismuch lessaffectedby noise, sothat monthly movementsare moresignificant.

LEADS AND LAGS

The most importantcharacteristicof an indicator from the point of view offorecastingis of coursethe evidenceit providesconcerningfuturechangesineconomicactivity. Indicatorsdiffer in this respectfor numerousreasons.Certain types,suchas housingstarts,contractsfor construction,andnew ordersfor machineryandequipment,representan earlystagein the processof making decisionson investment.Sinceit takestime to build a houseor a factoryor a turbine, the actualproductionor completion or shipping usually lagsbehind the ordersor contracts.The lag dependson, amongother things, thevolume of unfilled ordersor contractsstill to be completed.Wheregoodsare

120100

so

120

100

SA5OMAL FfrCro.5

197 1980 1981 198a 1983 198q 1985 1986 1987

234 PARTTHREE: OUR SOCIAL WORLD

madefor stock ratherthan to order theremay be no lag becauseordersarefilled as they are received.

Anotherkind of lead-lagrelationexistsbetweenchangesin the workweekonthe onehandandemploymenton theother. In manyenterprisesemployerscanincreaseor decreasehoursof work morequickly, morecheaply,andwith less ofa commitmentthan they can hire or fire workers.Hencein mostmanufacturing industriesthe averagelength of theworkweekusuallybeginsto increaseordecreasebeforeacorrespondingchangein the level of employment.The workweek is thereforea leadingindicator with respectto the unemploymentrate.

Many bilateralrelationsof thiskind havebeentracedseeFigure2. Thematterobviouslybecomesmore complex,however,whenthe relationsare multilateral. Indexesof stock marketprices, for example,have exhibited a longstandingtendencyto leadchangesin businessactivities the HarvardABC curvesrelied in part on this tendency,but the explanationseemsto requirethe interactionof movementsin profits andin interestrates,andotherfactorsaswell.A cyclical declinein profits often startsbefore abusinessexpansioncomestoan end; the proximatecauseis usually a rapid risein the costsof production.Interestratesalso are likely to rise sharply.Both factorsoperateto reducetheattractivenessof common stocks and depresstheir prices, even though thevolume of businessactivity is still rising. Near the endof a recessionthe oppositetendenciescomeinto play and lift stockpricesbefore businessbeginsto improve. Since interestrates are generallycountedamongthe lagging indicators,this caseillustrates that even lagging indicatorscan play a role inforecasting,by affecting the movementsof the leading indicators.

Forthepurposeof measuringleadsandlagsa chronologyof businesscycleshas proved useful. The National Bureauof EconomicResearchhas definedbusinesscyclesin sucha way thatpeaksandtroughscanbedatedwith reasonableobjectivity. Indeed,somepartsof the dating procedurecan now be carriedout by computer.Sincethe vast majority of indicatorsthat are of interestshow cyclical movementsconformingto businesscycles,the peaksandtroughsin eachindicatorcanbematchedwith thoseof thebusinesscycle to determinecharacteristicleadsand lags.

Following this plan,groupsof indicator seriesthat typically lead,coincidewith, or lag behindturns in thebusinesscycle havebeenidentified,as illustratedin Table 1. Compositeindexesconstructedfrom thesegroupsseeFigure 3 canbe employedas individual indicatorscanto measurethe relativeseverityofan economicdownturnas it progressesfrom month to month. With such amonitoringschemeonecanobservetherelativeseverityof thecurrentdeclineanddrawcertaininferencesbasedon the factthat the severityrankingsamongdifferent recessionshaveusuallynot changeda greatdealafter the first fewmonths.

It is of courseessentialin anyappraisalof theeconomicoutlookto takeintoaccountactualandprospectivepolicy actionsby theGovernment.Suchactionsincludetax reductionsor increases,changesin requiredbankreserves,changesin military expenditures,and the establishmentof programsof public employment. Such actionsoften do not fit readily into the frameworkof indicators,

Moore: The Developmentand Analysis of EconomicIndicators 235

RESIDEnTIAL COW5rIJCTIO,EXPENDITURES1 IN 1982 $

196’I 1963 1986 1187

I&o

1.50

P-iS P -r

p979 1980 Il 1983 1984 I98 lqS’ 1’87

CONTRACTS AND OCERS R*RANT EQPMEwr, IN 1962 $

30

* 25NCD0’

.4000

36ô

CD

300

250

EXPETtRS NEW PLAI’4T 4EQUIP,’11NT1 IN 38Z $

1919 1480 I’1 1992 1983I I_

41

40

39

z

-1

21

20

9

-v-vA’ERA6E WORKWEEK,MANUFCTUR1PlG

EMPLOYMENT MA1JFAC!.lRING

1479 980 1481 982. 983 1984 j995 I9$ 1987

Figure 2 Threeleading indicators andtheactivitiestheylead. Note: Vertical linesare businesscyclepeaksP andtroughsT. Sources: Departmentof CommerceandBureauof Labor Statistics

236 PARTTHREE: OUR SOCIAL WORLD

1979 1980 1981 1982 I93 1984 198& 1986 1987

FIgure 3 Leading,coincident,andlaggingindexesfor the UnitedStates.Note: Vertical lines representbusinesscyclepeaksP andtroughsT. Source: Departmentof Commerce.

although their effects, together with other influences, may be registeredpromptlyin orders,contracts,housingstarts,stockprices,andsoon. Still, certain indicatorsdo provideanearlycontinuousreadingon Governmentactivities,although they havegenerallynot beenfound sufficiently systematicin theirrelationto businesscyclesto warrant selectionas leading,coincident,or lagging indicators.

INTERNATIONAL ASPECTS

Economicindicatorsof all typesare followed morewidely in the UnitedStatesthan in most othercountries. The growth of trade, travel, and internationalfinance,however,has increasedthe needfor promptly availablestatisticsoninternationaltransactionsandeconomicdevelopmentin other countries. In1973 the National Bureauof Economic Researchbegana programof assembling andanalyzingindicatorsfor adozenindustrial countries,andsince 1979thisprogramhasbeencarriedforwardby theCenterfor InternationalBusinessCycle Research.

Fortunatelytheindicatorapproachis sufficiently flexible to be adaptedreadilyto situationswhere,as in manycountriessinceWorld War II, economicrecessionshavetaken theform of retardedgrowth of aggregateactivity ratherthanabsolute declines and where such retardationmay have been deliberatelyinducedby governmentpolicies in order to cool off inflation or restore a

Moore: The DevelopmentandAnalysis of EconomicIndicators 237

deterioratingtradebalance.Moreover,the approachis flexible enoughto accommodatedifferencesamongcountriesin thetypesof indicatordatathatareavailableor are mostrevealing.Forexample,in Europe,statisticson job vacanciesare relied on morethan in theUnited States,anddataon the internationalmigrationof workersare moresignificantbecausemigrant workersareasignificantpart of the work force.

The pursuitof economicindicatoranalysison an internationalscale,by internationalagenciesaswell as by domesticinstitutions,hasdemonstratedthefeasibility of the approachand its potential valuein observingandappraisinginternationalfluctuations in economicgrowth rates and the accompanyingtrendsin price levels, foreign trade,capital investment,andemployment.Onecan envision the evolutionof a worldwide systemof indicators,built on theplan originally developedfor the United States, to support the analysis ofeconomicindicatorson a global scale.

PROBLEMS

1. Why are seasonallyadjusteddataused in analyzingbusinesscycles?

2. What do the leading indicatorslead?

3. What qualifications should a good economicindicator possess?

4. Political candidateA says a recessionhasbegunbecausestock priceshavedropped. B says no, becauseemployment is still rising. How would youdecidewho is right?

5. The following figures are from theJuly 1975 issueof BusinessConditionsDigest.

CoincidentLeading Index LaggingIndex 1967 = 100 Index

July 1974 145.3 138.8 210.5Aug. 1974 140.4 138.2 214.5Sept. 1974 135.0 137.4 216.3Oct. 1974 130.1 136.2 219.0Nov. 1974 126.0 132.3 220.4Dec. 1974 123.6 128.2 220.0Jan. 1975 118.7 125.2 217.8Feb. 1975 118.6 124.1 212.9Mar. 1975 120.3 122.0 210.1Apr. 1975 124.8 122.2 205.5May 1975 127.9 122.4 201.5June 1975 130.7 124.6 200.2

Consideringthe factthat a recessionbeganin November1973,do you thinkit hasendedandif so, when?Whatconfirmingevidencewould you like to have?

238 PARTTHREE: OUR SOCIAL WORLD

6. How canleadingindicators for anothercountry, say,Japan,be helpful tobusinessfirms in the United States?

7. In deciding whether to invest in stocks or in bonds, what economicindicatorswould you examine?

RELATED REFERENCES AND DATA SOURCES

Bureauof EconomicAnalysis, U.S. Departmentof Commerce, 1984. Handbookof Cyclical Indicators. Washington,D.C.

Bureauof EconomicAnalysis, U. S. Departmentof Commerce.BusinessConditionsDigest,monthly.Washington,D.C.

Philip A. Klein and GeoffreyH. Moore. 1985. MonitoringGrowth Cyclesin Market-OrientedCountries. Cambridge,MA: Ballinger Publishing.

Geoffrey H. Moore. 1983.BusinessCycles,Inflation, and Forecasting,2nd edition. Cambridge,MA: Ballinger Publishing.

Organizationfor EconomicCooperationandDevelopment.Main EconomicIndicators,monthly.Paris.