Speech

Perspectives on Productivity “TheNewEconomyinaNewCentury:ImpactsofTechnologyonWhatandHowWeTeach” AconferencesponsoredbytheOfficeofEconomicEducationandBusinessResearch,SIUE South-WesternCollegePublishing SouthernIllinoisUniversity–Edwardsville Edwardsville,Illinois April7,2000 What in the world has happened in wemustdigintoitifwearetomakeprogressin theUnitedStatestoraiseeconomic understandingthekeydisputesoverwhetherwe growthsosubstantiallyduringthe reallyare,orarenot,livingina“new”economy. last few years? Real output has Beforeproceeding,Iwanttoemphasizethat risen beyond virtually all forecasts; unemploy- theviewsIexpressherearemineanddonot ment has fallen to levels that only a few years necessarilyreflectofficialpositionsoftheFederal ago were thought to be unattainable; labor pro- ReserveSystem.Ithankmycolleaguesatthe ductivitygrowthhasincreasedtorateslastexpe- FederalReserveBankofSt.Louisfortheircomriencedmanyyearsago,duringthe“GoldenAge” ments—especiallyDickAnderson,whoisacoof 1950-73; and inflation has remained low and authorofthisspeech.Iretainfullresponsibility evenfallenslightly.Thisunlikelycoincidenceof forerrors. economic outcomes has led some observers to conclude that the economy has entered a “New Age.” In this age, technical progress in infor- MEASURING PRODUCTIVITY: mation and communication technology—often AVERAGE LABOR PRODUCTIVITY referred as “ICT” for short— has, some believe, removedallpreviousspeedlimitsontheeconomy. Tobeginourdiscussion,letmeposethis Fromthistimeforward,potentialrealoutputwill question:Whatdowemeanby“productivity?” grow more rapidly than in the past, and the Themostcommonlydiscussedmeasureislabor unemployment rate will be forever lower. And, productivity.Laborproductivityissimplythe even more remarkably, these events will occur outputofeitheranindustryortheaggregateeconwithout any expansionary monetary policy that omydividedbylaborinput.Achangeinlabor might raise fears of higher inflation. productivityreflectsanychangeinoutputthat PerhapsI’veofferedanoverblownpictureof cannotbeaccountedforbyachangeinlabor thestorysometell,butmycharacterizationisnot input;suchchangesmaybedue,forexample,to overblownbymuch.Mygoaltonightistobring changesintheamountofcapitalusedperperson someperspectivetotheseeventsbyaskinghow employed. unusualtheyarewithinbothU.S.historyand Istartwiththisdefinitionalpointbecause relativetotheexperienceofothercountriesduring mostpopulardiscussionsofproductivitylaunch the1990s.I’llconcentrateontheproductivitypart rightintotheanalysisofthegloriousneweconomy ofthestory,forthatparthascertainlycaptured withoutnotinghowimportantthedefinitionisto everyone’sattention.Iwilldistinguishcarefully theanalysis.Ittakesonlyhalfasecondofthought betweenlaborproductivityandtotalfactorpro- tounderstandthatitreallymatterswhethera ductivity.Thisissuemaysoundtechnical,but ditchdigger’shigherproductivitycomesfrom 1

ECONOMICGROWTH substitutingabackhoeforashovelorforsome therateofinvestmentinnewphysicalcapital,and otherreason. thegrowthrateoflaborproductivity.Hestresses Let’sstartwithafewnumbers.Laborproduc- thattechnologicalinnovationstypicallyareput tivitygrowthintheUnitedStateshasrisenduring intoservice“embodied”innewcapitalequipthelastseveralyears.Usingannualaveragedata, ment.Incontrast,newwaysofdoingthingsthat in1999laborproductivity—outputperhourof enhancetheproductivityoftheexistingcapital laborinput—increased3.0percentinthenonfarm stockrarelyhavelargeeffects.Stronginvestment businesssector.During1998,theincreasewas spendingbefore1973,bothintheUnitedStates 2.8percent;during1997,2.0percent.Thegrowth andworldwide,contributedtorapidproductivity oflaborproductivityisnotonlyhigherthanit growthbydisseminatingnewtechnology.Worldusedtobe,butforeachofthethreemostrecent wideinvestmentspendingslowedsharplyafter yearsithasbeenhigherthanthepreviousyear. 1973,andwithitworldwidelaborproductivity Howunusualisthishigherrateoflaborprogrowth.Thereboundofbusinessinvestment ductivitygrowth?TheavailableU.S.historical duringthe1990s,bothintheUnitedStatesand data,beginningin1929,showthatlaborproduc- Europe,cameatatimeofmajortechnicalprogress tivitygrowthishighlyvolatile.Increasesofthis (andpricedecreases)intheproductionofsemisizeintheannualgrowthrateforayearortwo conductorsandcomputers.Thisinvestment arenotunusualinthisrecord,especiallyduring reboundhasundoubtedlycontributedtothe theearlyyearsofbusinesscyclerecoveries.But, recentincreaseoflaborproductivitygrowth. suchincreasesarehighlyunusualduringthelater Changesinlaborproductivityinforeigncounyearsofaneconomicexpansion.Themostrecent triesprovideasecondyardstickagainstwhichto EconomicReportofthePresident,forexample, judgewhethertherecentgrowthofU.S.labor notesthatlaborproductivityinthenonfarmbusiproductivityisunusualornot.Letmecautionyou, nesssectorincreasedbyonly1.3percentand1.1 however,thatnotallmeasuresoflaborproducpercentperyear,respectively,duringthefinaltwo tivityarethesame.IntheUnitedStates,themost yearsoftheexpansionsthatendedinlate1969 commonlydiscussedmeasureistheoneproduced and1990. bytheProductivitySectionoftheBureauofLabor Longer-runcomparisonsalsosuggestthat Statistics.Inthismeasure,thenumeratoristhe somethingunusualhasoccurred.During1950-73, outputofthenonfarmbusinesssectorandthe forexample—theso-called“GoldenAge”ofU.S. denominatoristotalhoursofworkbyemployees productivitygrowth—laborproductivityinthe inthatsector.InEuropeanpublications,however, nonfarmbusinesssectorincreasedatacompound themostcommonlycitedmeasureistheratioof averageannualrateofapproximately3percent. aggregaterealGDPtothenumberofemployed From1973through1995,productivitygrowth slowedtoabouta1percentpace.Duringthelast persons.Evenmoreconfusing,insomepublicatwoyears,laborproductivitygrowthhasreturned tions,analystsfocusonrealGDPpercapita, toitsGoldenAgerate.Thequestionsoftheday ratherthanperemployedperson,asameasure are:Whydidlaborproductivitygrowthrise? ofproductivity. Willthehigherratecontinue? APh.D.ineconomicsisnotrequiredto Onehypothesisisthattheresurgenceoflabor appreciatethatthesevariousmeasuresoflabor productivitygrowthiscloselyconnectedtothe productivitywilldifferwhenvariablessuchas strongpaceofbusinessequipmentinvestment thenumberofannualhoursperemployeeor duringthe1990s,bothintheUnitedStatesand thepercentageoftheworking-agepopulation Europe.AngusMaddison,thewell-knowneco- employeddifferacrosscountries.TheU.S.Bureau nomicgrowthhistorian,hasdemonstratedstrong ofLaborStatisticsisoneofthefewagenciesthat longer-runcorrelationsamongthreevariablesin seekstoadjustforthesedifferencesandprovide aneconomy:thepaceoftechnologicalinnovation, consistentcross-countrycomparisons. 2

PerspectivesonProductivity Unfortunately,duetodatalimitations,the mentintheservicesindustries.Suchstudiesare BLSpublishesthecomparativestatisticsonly hamperedbythefactthatoutputofthesesectors onceeachyearandincludesonlymanufacturing. islesswell-measuredthanthatofmanufacturing. FortheUnitedStates,from1990to1998,labor Thestudiesgenerallyhavereachedsimilarconproductivityinmanufacturingshowsa3.3per- clusions,however:Servicesectorshaveextensively centannualgrowthrateofoutputperhour.For purchasednewICTequipmentanddeployedit certaincountries,therateismorerapid:4.6per- toincreaselaborproductivity,bothintheUnited centforSweden,3.9percentforFrance,and3.8 StatesandEurope. percentfortheNetherlands.Forsomeothersitis slower:3.2percentforGermany,3.0percentfor JapanandBelgium,2.2percentfortheUnited MEASURING PRODUCTIVITY: Kingdom,and2.0percentforCanada.Onan TOTAL FACTOR PRODUCTIVITY annualbasis,thedataareerratic:manufacturing Tothispoint,Ihaveconcentratedonlabor productivitygrowthintheUnitedStates,forexamproductivity.Thecontributionstooutputofother ple,risesto6.1percentin1995,slowsto2.1perinputs,includingphysicalcapital,havebeen centin1996,andrisesagainto4.1percentinboth ignored.Forsomepurposesandduringsome 1997and1998.Insomeindividualyearssince timeperiods,theseomissionsmightbeunimpor- 1995,growthratesforothercountriesagainexceed tant.Unfortunately,asIhintedintheintroduction theU.S.rate:7.2percentforFranceand5.9perofthisspeech,thisisnotoneofthoseperiods. centforGermanyin1997and4.3percentfor Theoutstandingfactofthepastdecadeisthe Germanyin1998. sharpfallinthepriceofICTrelativetotheprices Inpart,cross-countrydifferencesreflectdifferofotherproductiveinputs.Totheextentthat entstagesoftherespectivebusinesscycles.In thesepricedecreaseshaveinducedfirmstosubaddition,fluctuationsininternationaleconomic stituteICT-typecapitalforlabor,thestoryofthe conditionsmakecomparisonsduringthe1990s “neweconomy”isoneofinputsubstitution,not tenuous.Ingeneral,Europeannationshavebeen productivitygrowth.Althoughditchdiggersmay morestronglyaffectedbyinternationaleconomic nothavebiggerandbetterbackhoes,mostworkers crisessuchasoccurredinMexicoin1995,East havebeenequippedwithbettercommunications Asiain1997,andRussiain1998.Onbalance, equipment,bettercomputers,andbettersoftware. though,comparisonsofU.S.andEuropeanpro- Recallthatchangesinlaborproductivity ductivitygrowthsuggestsimilarpatterns. reflectallchangesinoutputthatcannotbeattrib- Additionalconfirmationofthecommon utedtochangesinlaborinput.Changesininput impactofICTtechnologyisevidentinacomparipricesthatinducesubstitutionbyfirmsofcapital sonoftheanalysisinthemostrecentEconomic forlaborwillincreasemeasuredlaborproductiv- ReportofthePresidentwithsimilaranalysisin ity.(Ofcourse,ifoutputandsalesdonotincrease theEuropeanUnion’sDecember1999Joint sufficiently,higherproductivitywillalsoresult EmploymentReport.Bothreportsareconcerned infewerworkersinthatfirmorindustry,ashas withrestructuring,layoffsofworkers,andthe beenthecontinuingcaseinagriculturefor150 needfor“lifelong”educationtoadapttoachang- years.)Thesesubstitutions,however,arenotwhat ingworld.TheEU’sEmploymentReportalso economistscommonlyrefertoas“technical reviewsthelargenumberofinitiativesbymem- change,”noraretheylikelytoincreasethelongbernationstopromoteentrepreneurshipinthe termtrendgrowthoftheeconomy’spotential new“informationsociety,”includingeasingthe output.Instead,weneedtomeasuretheeffectof administrativeburdensthathamperjobcreation newtechnologyonoutputafteraccountingforall bysmallbusiness. thegrowthofinputduetoincreasesinbothcapital Finally,Imustnotethatanincreasingnumber andlabor.Theconceptoftotalfactorproductivity ofstudieshaveaddressedtheroleofICTinvest- seekstodoso. 3

ECONOMICGROWTH Totalfactorproductivitymeasureschanges Akeyfindingfromthisresearchisthatthe inoutputrelativetoanindexnumberthatmeas- outstandingfactintheUnitedStatessince1995 uresthecombinedinputsoflaborandcapital. hasbeena28percentperyeardecreaseinthe Changesintotalfactorproductivityreflectchanges costsofICTequipment,qualityadjusted.These intheorganizationofproductionorintechnology studiesconcludethattheoverwhelmingfactor thatarenotduetochangesineitherlaboror behindtheincreaseinU.S.laborproductivity capitalinputs. duringthe1990shasbeentheincreaseinthe Theconceptoftotalfactorproductivityis capital/laborratio.Althoughdifferinginthe notnew.EconomistsoftenrefertoTFPasthe details,thesestudiesonbalanceconcludethat “Solowresidual,”followingfromRobertSolow’s totalfactorproductivitygrowthhasincreased seminal1956growththeoryarticle.But,tounder- moderatelysince1995,andinthepastthreeyears hascontributedperhaps0.75to1percentagepoint standTFP,wemustbackuponestep:Whatdo ofthegrowthofaggregaterealGDP. wemeanby“capital”?Inearlygrowththeory Ifconfirmedbyfurtherresearch,theseresults research,suchasSolow’s1956and1957articles haveimportantimplicationsforthefuture.In andEdwardDenison’slaterworkongrowth particular,theresultsresurrectacommontheme accounting,physicalcapitalwastakentobea inthegrowththeoryliteratureofthelast45years: homogenousinput.Thatis,intheanalysis,differthecriticalimportanceofnotconfusingchanges entvintagesofcapitalwerenotseparated.Many inlevelswithchangesinratesofgrowth.Sharp, analystshavearguedthatthiswas(andis)aseriunexpecteddecreasesinthepriceofanimportant ousmistake.Newtechnologyisprimarilyputinto input—ICTequipment—mayinducealong-lasting, useintheformofnewphysicalcapital,anddifferperhapspermanent,increaseintheleveloflabor ingvintagesofcapitalmustbeadjustedcarefully productivityandpotentialoutput.Butthesame fortheirproductivecapacity.Ironically,Robert decreasesmaysuggestlittleornoincrease,except Solowhimselfarguedforthispointofviewin fornear-termadjustments,inthegrowthrateof threelaterarticles,publishedbetween1960and eitherlaborproductivityorpotentialoutput.For 1963. thegrowthrateitselftobepermanentlyhigher,the Inrecenteconometricstudiesoftotalfactor rapiddeclineinthepriceofICTequipmentand productivity,laborandcapitalinputshavebeen consequentcontinuingincreaseinthecapital/ measuredwithindex-numbermethodssimilar laborratiowouldhavetobepermanent. tothoseintroducedbytheBureauofEconomic Inrelated,widelypublicizedresearch,Robert Analysisin1995tomeasurerealGDP.Suchmeas- GordonhasarguedthatalloftheU.S.economy’s uresareparticularlyappropriatewhenseveral productivitygainduringthe1990sisdueto typesofeachinputmustbeincluded—suchas improvementsintheproductionofcomputers, workerswithdifferingamountsofeducationor ormorespecifically,intheproductionofthesemiseveraldifferentvintagesofphysicalcapital— conductorcomponentsofcomputers.Heargues andtherelativepricesofthevariousinputsare thatmostothersegmentsoftheeconomyhave changing.Theeconometricmethodsattemptto experiencedlittle,ifany,gainfromtheuseof separatechangesinlaborproductivityintotwo computers.Althoughthesemorerecenteconometparts:thepartduetotheincreaseintheamount ricresultssuggestsomewhathighertotalfactor ofcapitalperunitoflaborandthepartdueto productivitygrowththanGordon’sestimates, organizationalchangespermittedbynewtech- theyareconsistentwithGordon’sfindings.In nology.Thefirstpartmeasurestheincreaseinout- sectorsoutsidesemiconductormanufacturing, putasaconsequenceoftheapplicationofmore theimprovementincomputershasinducedthe capital.Thesecondpartisthe“Solowresidual,” substitutionoflessexpensiveformoreexpensive thatpartofadditionaloutputwhichisnot inputs;net(econometrically)oftheinputsubstiaccountedforbyinputsoflaborandcapital. tutioneffect,mostofthesesectorshavegained 4

PerspectivesonProductivity littlefromthenewtechnology.Thatis,newtech- Thesedevelopmentsareobviouslyextremely nologyhasnotshiftedtheproductionfunctionin important.Theywillpermanentlyincreasethe thesesectorsbuthassimplyledtotheapplication leveloflaborproductivity.Butgivenourcurrent ofmorecapitalperunitoflabortotheproduction understandingoftheseissues,theyareunlikely process.Semiconductormanufacturing,onthe toyieldapermanentlyhighergrowthrateofprootherhand,hasbenefitedsignificantlybecause ductivity.Itseemsprobabletomethateventually— theoutputofthatsector—bettercomputers—isa althoughIhavenoideawhen—thetechnical veryimportantinputtothesamesector:new, advancesthathavebeencreatingthesharp morepowerfulcomputersareessentialtothe declinesinthepriceofcomputingpowerwill designandmanufactureofevennewer,more slow,perhapsasaconsequenceoffundamental powerfulcomputers.Inturn,thesecomputers lawsofphysicsthatcontrolhowmanytransistors willeasetheproductionofevennewermachines. canbepackedontomicrochips.Whentheprice Asnewercomputerscontinuetoenhanceour declineofcomputingpowerslows,theapplicaabilitytodesignandbuildevermorepowerful tionofadditionalcomputingpowerperunitof machines,the“perunit”costofcomputinglikely laborwillslow,andsoalsowilltherateofgrowth willcontinuetodecrease.In1990,theper-unit oflaborproductivity.Butletmeemphasizeagain decreaseinthepriceofcomputingpowerwas thatIhavenoideawhenthegrowthwillslowor runningatapproximatelya15percentannual evenifourcurrentunderstandingoftheseissues rate;bythemiddleofthedecade,thepriceofcom- isontarget. putingpowerwasfallingatnearlya30percent annualrate.Asthepricecontinuestodecrease, itseemslikelythatmuchofthisnewpowerwill HISTORICAL ANALYSIS OF beusedtomanageandcatalogknowledgeand FACTORS IN ECONOMIC informationonincreasinglysophisticatedcom- GROWTH AND PRODUCTIVITY puternetworks.(Theseadvancesmighteven reducetheflowofpaperacrossmydesk!). I’llconcludemyremarksbyattemptingto Inexpensiveandpowerfulcomputersalready placetheeventsofthepastdecadeinthecontext arepermittingthedesignandimplementationof ofinternationaleconomichistory.Althoughdata knowledge-anddocument-managementsystems areincomplete,itseemsunlikelythatinthepast atindividual-departmentlevels,aluxurythat anytechnologyhasexperiencedpricedecreases wouldhavebeendifficulttoaffordonlyafew ofthescalewehaverecentlyexperiencedforICT yearsago.Morepowerfulcomputersalsowill equipmentandsoftware.Electricityspreadfairly continuetoimprovecommunications,permitting rapidlyintobothhomesandfactoriesduringthe fasterandmoreprecisetransmissionoflarge twodecadesafterEdisonandWestinghouse amountsofinformationoverlongdistances.The agreedonindustrystandardsduringthelate lowercostofhigh-speedinternationaldatalinks 1890s.But,followingtheintroductionofthese alreadyispermittingmajormulti-nationalcor- standards,thecostofelectricalequipmentand porationstobettershareinformationaroundthe appliancesdidn’tdecreasesharplyrelativeto worldinrealtime,andperhapshascontributed othergoodsorwagerates,duringthefirstquarter tosomeofthemega-mergersofrecentyears.At ofthetwentiethcentury. thesametime,thelowerpriceofICTequipment Theworldinthepasthasexperiencedsharp likelyhasloweredthecostsofstartingandoper- fluctuationsinproductivitygrowthandlikelywill atingasmallbusiness.Accountingsoftwarecan again.Thehistoricalrecordsuggests,however, replaceanaccountant;awordprocessingprogram thatsustainingmorethana2percentannual canreplaceatypist;webpagescanreplacea growthrateoflaborproductivityforanextended printedcatalog;ande-mailcanreplaceatelephone periodisunlikely.Economichistorianstellus receptionist. thattheworldenteredanew“capitalist”erain 5

ECONOMICGROWTH about1820.Beforethen,forcenturies,realoutput about2percentannualgrowth,astheUnited perworkerhadincreasedverylittle.Sincethen, Stateswaspullingahead,technologically,of therehasbeenafairlysteadyincreaseinthe GreatBritain.Perhapssurprisingly,theslow1 amountofphysicalcapitalperlaborhour,and percentproductivitygrowthduring1973to1992 worldwideoutput(GDP)perpersonemployed stillranksthaterainthirdplace. hasincreasedmorethaneightfold.Historians Inthebestofallworlds,wecanonlyhope agreethatGreatBritainwasthetechnological thattherecent3percentannualgrowthinlabor leaderuptoabout1890;theyalsoagreethatlabor productivitycontinues.FortheUnitedStates, productivitygrewatonlyabouta1.4percent thelong-runtrendoflaborproductivitygrowth annualrateduringthatperiod. seemstobeapproximately2percent.Beforethe About1890,theUnitedStatestookthemantle GoldenAgeof1950to1973,productivitygrew of“technologicalleader”awayfromBritain.Ata atabouta2¼percentannualratebetween1890 conferenceofeducators,itseemsappropriateto and1929andaboutaof1percentratebetween notethathistoriansgivemuchofthecredittothe 1929and1938.Evena2percenttrendwouldbe U.S.highereducationsystem.TheMorrillActof superiortotheeconomy’sperformanceduring 1862,whichcreatedlandgrantuniversities,stim- the1970sand1980s. ulatedteachingandresearchinbothagriculture Animportantbottomlineformeasapolicyandengineering.Inengineering,withinadecade makeristhatthestateoftheeconomicscienceof afterpassageoftheMorrillAct,thenumberof productivitygrowthcontainsmanyhugegaps. engineeringschoolswentfrom6to70,andthen Thestateofknowledgedoesnotjustifyfirmconto126by1917.U.S.engineeringschoolsgraduated victionsaboutanyproductivitygrowthforecast, 100studentsin1870;by1917,thenumberwas althougharangeof3to4percentoverthenext 4,300.Asearlyas1890,theratioofuniversity fewyearsdoesseemsensibletome. studentsper1,000primaryschoolstudentsin Givenourincompleteknowledge,itisimportheUnitedStateswastwotothreetimesthatof tantthatwenotlockourselvesintoamonetary anyothercountry.Inagriculture,aslateas1914, policythatdependsonanyparticularrateof theUnitedStateswaswellbehindEuropein productivitygrowth.Itiscertainlyquitepossible scientificagriculture—agenerationlater,the thattoday’sproductivitygrowthwillbemain- UnitedStateswastheworldleader. tainedforthenextdecade,orrisefurther.Bythe Since1890,U.S.productivitygrowthhasnot sametoken,itisobviouslydangeroustosimply beensmooth.Productivitygrewmostrapidly assumethatourgoodtimeswillrollonforthe during1950to1973,whenmuchoftheworldwas foreseeablefuture.Itisnotverysatisfyingfora catchingup,technologically,totheUnitedStates. formeracademictosaythatwe’regoingtohave SomeoftheU.S.growthduringthisperiodmay towatchcarefully,andthatsuchwatchingis havereflectedarecoveryprocessafterthedisrup- aboutthebestadvicewecantake.Butthatdoes tionsoftheGreatDepressionandWorldWarII. seemtometobeoursituation,likeitornot. Thesecondbestperiodwas1870to1913with 6