IV
　　Ihavementionedinpassingthatthenewsystemmightbemorefavourabletopeacethantheoldhasbeen。Itisworthwhiletorepeatandemphasisethataspect。Warhasseveralcauses。Dictatorsandotherssuch，towhomwaroffers，inexpectationatleast，apleasurableexcitement，finditeasytoworkonthenaturalbellicosityoftheirpeoples。But，overandabovethis，facilitatingtheirtaskoffanningthepopularflame，aretheeconomiccausesofwar，namely，thepressureofpopulationandthecompetitivestruggleformarkets。Itisthesecondfactor，whichprobablyplayedapre-
　　[Page382]THE
　　GENERALTHEORYOFEMPLOYMENT
　　dominantpartinthenineteenthcentury，andmightagain，thatisgermanetothisdiscussion。
　　Ihavepointedoutintheprecedingchapterthat，underthesystemofdomesticlaissez-faireandaninternationalgoldstandardsuchaswasorthodoxinthelatterhalfofthenineteenthcentury，therewasnomeansopentoagovernmentwherebytomitigateeconomicdistressathomeexceptthroughthecompetitivestruggleformarkets。Forallmeasureshelpfultoastateofchronicorintermittentunder-employmentwereruledout，exceptmeasurestoimprovethebalanceoftradeonincomeaccount。
　　Thus，whilsteconomistswereaccustomedtoapplaudtheprevailinginternationalsystemasfurnishingthefruitsoftheinternationaldivisionoflabourandharmonisingatthesametimetheinterestsofdifferentnations，therelayconcealedalessbenigninfluence；andthosestatesmenweremovedbycommonsenseandacorrectapprehensionofthetruecourseofevents，whobelievedthatifarich，oldcountryweretoneglectthestruggleformarketsitsprosperitywoulddroopandfail。Butifnationscanlearntoprovidethemselveswithfullemploymentbytheirdomesticpolicyand，wemustadd，iftheycanalsoattainequilibriuminthetrendoftheirpopulation，thereneedbenoimportanteconomicforcescalculatedtosettheinterestofonecountryagainstthatofitsneighbours。Therewouldstillberoomfortheinternationaldivisionoflabourandforinternationallendinginappropriateconditions。Buttherewouldnolongerbeapressingmotivewhyonecountryneedforceitswaresonanotherorrepulsetheofferingsofitsneighbour，notbecausethiswasnecessarytoenableittopayforwhatitwishedtopurchase，butwiththeexpressobjectofupsettingtheequilibriumofpaymentssoastodevelopabalanceoftradeinitsownfavour。Internationaltradewouldceasetobewhatitis，namely，adesperateexpedienttomaintainemploymentathomebyforcingsalesonforeign[Page383]CONCLUDING
　　NOTES
　　marketsandrestrictingpurchases，which，ifsuccessful，willmerelyshifttheproblemofunemploymenttotheneighbourwhichisworstedinthestruggle，butawillingandunimpededexchangeofgoodsandservicesinconditionsofmutualadvantage。
　　V
　　Isthefulfilmentoftheseideasavisionaryhope？Havetheyinsufficientrootsinthemotiveswhichgoverntheevolutionofpoliticalsociety？Aretheinterestswhichtheywillthwartstrongerandmoreobviousthanthosewhichtheywillserve？
　　Idonotattemptananswerinthisplace。Itwouldneedavolumeofadifferentcharacterfromthisonetoindicateeveninoutlinethepracticalmeasuresinwhichtheymightbegraduallyclothed。Butiftheideasarecorrect¾；anhypothesisonwhichtheauthorhimselfmustnecessarilybasewhathewrites¾；itwouldbeamistake，Ipredict，todisputetheirpotencyoveraperiodoftime。Atthepresentmomentpeopleareunusuallyexpectantofamorefundamentaldiagnosis；moreparticularlyreadytoreceiveit；eagertotryitout，ifitshouldbeevenplausible。Butapartfromthiscontemporarymood，theideasofeconomistsandpoliticalphilosophers，bothwhentheyarerightandwhentheyarewrong，aremorepowerfulthaniscommonlyunderstood。
　　Indeedtheworldisruledbylittleelse。Practicalmen，whobelievethemselvestobequiteexemptfromanyintellectualinfluences，areusuallytheslavesofsomedefuncteconomist。Madmeninauthority，whohearvoicesintheair，aredistillingtheirfrenzyfromsomeacademicscribblerofafewyearsback。Iamsurethatthepowerofvestedinterestsisvastlyexaggeratedcomparedwiththegradualencroachmentofideas。Not，indeed，immediately，butafteracertaininterval；forinthefieldofeconomicandpoliticalphilosophytherearenotmanywhoare[Page384]THE
　　GENERALTHEORYOFEMPLOYMENT
　　influencedbynewtheoriesaftertheyaretwenty-fiveorthirtyyearsofage，sothattheideaswhichcivilservantsandpoliticiansandevenagitatorsapplytocurrenteventsarenotlikelytobethenewest。But，soonorlate，itisideas，notvestedinterests，whicharedangerousforgoodorevil。
　　PrintingErrorsinFirstEditionAppendixI
　　PRINTINGERRORSINTHEFIRSTEDITIONCORRECTEDINTHEPRESENTEDITION
　　PageLineCorrection[60]6For'possession'read'possessions'
　　[83]12For'has'read'had'
　　[126]13For'23'read'19'
　　[126]footnoteI，line2For'th'read'the'
　　[172]21For'security'read'precautionary'
　　[212]9For'than'read'that'
　　[229]32For'output'read'thestockofassetsingeneral'
　　[233]25For'their'read'its'
　　[237]31For'or'read'of'
　　[267]28For'three'read'four'
　　[271]4For'technique'read'techniques'
　　[319]23For'income'read'incomes'
　　[341]7For'Mercantilist'read'Mercantilists'ThesecorrectionscometolightinpreparingvariousforeigneditionsofTheGeneralTheory，inpreparingthevariorumversionofearlierdraftswhichappearsinvolumeXIV，orinsettingthisbookforpress。
　　Thecorrectionsdonotcovermoresubstantialerrorssuchastheunsatisfactorypresentationofaggregatesupplyanddemandonpage29ortheinadequatederivationoftheequationsonpage305。
　　FluctuationsinU。S。NetInvestment[Page386]
　　Appendix2
　　FromTheEconomicJournal，September1936
　　FLUCTUATIONSINNETINVESTMENTINTHEUNITEDSTATES
　　InmyGeneralTheoryofEmployment，InterestandMoney，chap。
　　8，pp。98-104，Imadeabriefattempttoillustratethewiderangeoffluctuationsinnetinvestment，basingmyselfoncertaincalculationsbyMrColinClarkforGreatBritainandbyMrKuznetsfortheUnitedStates。[208]
　　InthecaseofMrKuznets'figuresIpointedoutp。103thathisallowancesfordepreciation，etc。，included'nodeductionatallinrespectofhousesandotherdurablecommoditiesinthehandsofindividuals'。Butthetablewhichimmediatelyfollowedthisdidnotmakeitsufficientlycleartothereaderthatthefirstlinerelatingto'grosscapitalformation'comprisedmuchwidercategoriesofcapitalgoodsthanthesecondlinerelatingtoentrepreneurs'depreciation，etc。'；andIwasmyselfmisledonthenextpage，whereIexpresseddoubtsastothesufficiencyofthelatteriteminrelationtotheformerforgettingthatthelatterrelatedonlytoapartoftheformer。Theresultwasthatthetableasprintedconsiderablyunder-statedtheforceofthephenomenonwhichIwasconcernedtodescribe，sinceacompletecalculationinrespectofdepreciation，etc。，coveringalltheitemsinthefirstlineofthetable，wouldleadtomuchlargerfiguresthanthosegiveninthesecondline。SomecorrespondencewithMrKuznetsnowenablesmetoexplaintheseimportantfiguresmorefullyandclearly，andinthelightoflaterinformation。
　　MrKuznetsdivideshisaggregateofgrosscapitalformationashecallsitfortheUnitedStatesintoanumberofcategoriesasfollows:
　　1Consumers'DurableCommoditiesThesecomprisemotor-cars，furnitureandhouseequipmentandothermoreorlessdurablearticles，apartfromhouses，pur-
　　[Page387]APPENDIX
　　2
　　chasedandownedbythosewhoconsumethem。Whetherornottheseitemsshouldbeincludedininvestmentdependssofarasthedefinitionisconcerned
　　onwhethertheexpenditureonthemwhenitisinitiallymadeisincludedincurrentsavingorincurrentexpenditure；anditdependssofarasthepracticalapplicationisconcernedonwhetherinsubsequentyearstheownersfeelunderamotivetomakeprovisionforcurrentdepreciationoutoftheirincomesevenwhentheyarenotreplacingorrenewingthem。
　　Doubtlessitisnotpossibletodrawahard-and-fastline。Butitisprobablethatfewindividualsfeelitnecessaryinsuchcasestomakeafinancialprovisionfordepreciationapartfromactualrepairsandrenewals。This，incombinationwiththedifficultyofobtainingproperstatisticsandofdrawingaclearline，makesitpreferable，Ithink，toexcludesuchequipmentfrominvestmentandtoincludeitinconsumption-expenditureintheyearinwhichitisincurred。ThisisinaccordancewiththedefinitionofconsumptiongiveninmyGeneralTheory，p。54。[209]
　　Ishall，therefore，excludethiscategoryfromthefinalcalculation；
　　[210]thoughIhopetodealwiththeproblemmorethoroughlyatalatertime。NeverthelessitmaybeinterestingtoquoteMrKuznets'estimates，whichareofsubstantialmagnitude:
　　Millionsofdollars。
　　192519261927192819291930193119321933
　　Consumer'sdurablecommodities8，6649，3168，8879，17510，0587，8925，8854，0223，737Theabovefigurefor1929includes3，400milliondollarsformotor-cars，whilstthedepreciationinrespectofthesameitemforthatyearisestimatedat2，500milliondollars。
　　2ResidentialConstructionThisisanimportantandhighlyfluctuatingitemwhichshouldundoubtedlybeincludedininvestment，andnotinconsumptionexpenditure，sincehousesareusuallyregardedas[Page388]THE
　　GENERALTHEORYOFEMPLOYMENT
　　purchasedoutofsavingsandnotoutofincome，andareoftenownedbyothersthantheoccupiers。IntheBulletinfromwhichthesefiguresaretakenMrKuznetsgivesnoestimatefortheannualrateofdepreciation，etc。Morerecently，however，hiscolleague，MrSolomonFabricant，haspublishedsuchestimates，[211]whichIhaveusedinthefollowingtable:
　　Millionsofdollars。
　　192519261927192819291930193119321933
　　Residentialconstruction3，0502，9652，8563，0952，1271，222900311276
　　Depreciation*1，5541，6761，7541，8421，9111，9011，6981，4601，567
　　Netinvestment1，4961，2891，1021，253216-679-798-1，149-1，291*Thesefiguresarecalculatedintermsofcurrentreproductioncosts。
　　MrFabricanthasalsoprovidedestimatesintermsoforiginalcost，whichfortheyearspriorto1932areconsiderablylower。
　　3BusinessFixedCapitalMrKuznetsheredistinguishesexpenditureonnewproducers'durablegoodsandbusinessconstructionfromthenetchangein'businessinventories，'
　　i。e。inworkingandliquidcapital；andweshall，therefore，dealwiththelatterunderaseparateheading。
　　Theamountofthedeductiontoobtainnetinvestmentinrespectofparts，repairsandservicing，andrepairsandmaintenanceofbusinessconstructionasdistinctfromdepreciationanddepletion，whichisnotmadegood，depends，ofcourse，onwhethertheformerhavebeenincludedingrossinvestment。
　　MrKuznetsgivesapartialestimatefortheformerbutthefiguresgivenbelowexcludetheseitemsbothfromgrossandfromnetinvestment。Butwhilsttheresultofdeductingboththerepairsitemandthedepreciationitemprobablycorrespondsfairlycloselytomynetinvestment，thetwodeductionstakenseparatelydonotcloselycorrespondtomydeductionsforusercostandsupplementarycost；sothatitisnotpossibletocalculatefromMrKuznets'dataafigurecorrespondingtomygrossinvestment。
　　Thefollowingtablegivesinthefirstline'theformationofgrosscapitaldestinedforbusinessuse，exclusiveofparts，repairsandservicing，andrepairsandmaintenanceofbusinessconstruction，andexcludingchangesinbusinessinventories'；
　　[Page389]APPENDIX
　　2
　　andinthesecondlinetheestimated'depreciationanddepletion'onthesameitems:
　　Millionsofdollars。
　　192519261927192819291930193119321933
　　Grossbusinesscapitalformationasabove9，0709，8159，55510，01911，3969，3365，9333，2052，894
　　Depreciationanddepletion*5，6856，2696，3126，4477，0396，7126，1545，0924，971
　　Netinvestment3，3853，5463，2433，5724，3572，624-221-1，887-2，077*ThesefiguresarenottakenfromMrKuznets'memoranda，hutfromMrFabricant'slaterandrevisedestimates。Asbeforetheyareintermsofcurrentreplacement
　　cost。Intermsoforiginalcosttheyareappreciablylowerpriorto1931
　　andhighersubsequently。
　　4BusinessInventoriesForthefinancialgainsorlossesarisingoutofthisitemthereappeartobefairlyadequatestatisticsintheUnitedStates，thoughnotinthiscountry。MrKuznets'figuresareasfollows:
　　Millionsofdollars。
　　192519261927192819291930193119321933
　　Netgainorlossinbusinessinventories9162，664-1765111，800100-500-2，250-2，250Thistablecoversnotonlymanufacturers'stocksbutalsostocksoffarmers，mines，traders，governmentagencies，etc。From1929onwardsthefiguresgiveninMrKuznets'memorandumof1934provedtorequirecorrection。Thosegivenaboveareprovisionalandapproximateestimates，pendingthepublicationofrevisedfiguresbytheNationalBureau。
　　5PublicConstructionandBorrowingTherelevantfigureinthiscontextisnotsomuchthegrossornet
　　expenditureonconstruction，astheamountofexpendituremetoutofanetincreaseinborrowing。Thatistosayinthecaseofpublicauthoritiesandthelike，theirnetinvestmentmaybebestregardedasbeingmeasuredbythenetincreaseintheirborrowing。Insofarastheirexpendituresaremetbycompulsorytransferfromthecurrentincomeofthepublic，theyhavenocorrelativeinprivatesaving；whilstpublicsaving，ifweweretofindasatisfactorydefinitionforthisconcept，wouldbesubject[Page390]THE
　　GENERALTHEORYOFEMPLOYMENT
　　toquitedifferentpsychologicalinfluencesfromprivatesaving。IhavetouchedontheprobleminmyGeneralTheory，p。128，footnote。I
　　propose，therefore，toinsertinplaceofthefiguresofpublicconstructionthe'loanexpenditure'ofpublicbodies。
　　MrKuznetshasverykindlysuppliedmewithfiguresforthenetchangesintheamountofpublicdebtFederal，StateandlocaloutstandingintheUnitedStates，which，exceptforminorchangesintheGovernment'scashbalances，representtheamountofpublicexpenditurenotcoveredbytaxesandotherrevenues。[212]Thisisgivenbelowinparallelwithhisestimatesoftheamountofconstructionbypublicauthorities。Theinterestingresultemergesthatupto1928therewasanetreductioninthepublicdebtinspiteofalargeexpenditureonpublicconstruction，andthatevenupto1931somepartofpublicconstructionwasmetoutofrevenue。Theexcessofborrowingoverconstructionin1932and1933represents，ofcourse，variousmeasuresofpublicrelief。
　　Millionsofdollars。
　　192519261927192819291930193119321933
　　Publicconstruction2，7172，6123，0453，0232，7763，3002，9062，0971，659
　　Netchangeinoutstandingpublicdebt**-43-280-244-504411，7122，8222，5652，796*SeeMrKuznets'Bulletin，TableII，line22，broughtuptodateonthebasisofmorerecentdata。
　　**Seecol。9ofthetablegivenintheappendixbelow。
　　6ForeignInvestmentFinally，wehavethenetchangeinclaimscountries，estimatedbyMrKuznetsasfollows:
　　Millionsofdollars。
　　192519261927192819291930193119321933
　　42844606957312371326402937AggregateNetInvestmentWearenowinapositiontocombinetheaboveitemsintoasingleaggregate。
　　Thistotalisnotquitecomprehensive，sinceitexcludesconstructionbysemi-publicagencies，andasmall[Page391]APPENDIX
　　2
　　amountunallocableconstruction。ButMrKuznetsisoftheopinionthatbothomissionsarequiteminorincharacterandcouldnotmuchaffectthemovementsofnetinvestmentinthetablewhichnowfollows。
　　Millionsofdollars。
　　192519261927192819291930193119321933
　　Residentialconstruction1，4961，2891，1021，253216-679-798-1，149-1，291
　　Businessfixedcapital3，3853，5463，2433，5724，3572，624-221-1，887-2，077
　　Businessinventories9162，664-1765111，800100-500-2，250-2，250
　　Netloanexpendituresbypublicauthorities-43-280-244-104411，7122，8222，5652，796
　　Foreigninvestment4284460695731237132640293
　　Aggregatenetinvestment6，1827，2634，5316，2837，1264，1281，629-2，681-2，529Itisevidentthatthistableisoffirst-classimportancefortheinterpretationofbusinessfluctuationsintheUnitedStates。Inmattersofdetailthefollowingpointsstandout:
　　aThearrearsofresidentialconstructionattheendof1933musthavebeenenormous。Fortherehadbeennonetinvestmentinthisfieldsince1925。Thisdoesnotmean，ofcourse，thattheactualstateofhousingwassobadasthis。Somegrossinvestmentinhousingcontinuedthroughout，andthegradualdeteriorationinthestateofaccommodation，throughobsolescenceanddecaynotmadegood，doesnotimpairforthwithtoanequalextenttheactualaccommodationavailableforthetimebeing。
　　bThepartplayedbyfluctuationsinbusinessinventoriesisverymarked，especiallyinaccentuatingthedepressionatthebottomoftheslump。Theincreaseininventoriesin1929wasprobablyforthemostpartdesignedtomeetdemandwhichdidnotfullymaterialise；whilstthesmallfurtherincreasein1930representedaccumulationsofunsoldstocks。In1932and1933，manufacturersmetcurrentdemandtoanextraordinaryextentoutofstocks，sothateffectivedemandfelllargelybehindactualconsumption。Butthis，fortunately，isastateofaffairswhichcouldnotcontinueindefinitely。Afurtherdepletionofstocksonthisscalecouldnotpossiblytakeplace，sincethestockswerenolongerthere。AlevelofbusinessinventoriessolowasthatwhichexistedintheUnitedStatesattheendof1933wasanalmostcertainheraldofsomemeasureofrecovery。
　　Ingeneralanaggregateofnetinvestmentwhichisbasedonanincreaseinbusinessinventoriesbeyondnormalisclearlyprecarious；andit[Page392]THE
　　GENERALTHEORYOFEMPLOYMENT
　　iseasytoseeinretrospectthatalargegrowthofinventoriesin1929，coupledwithadeclineinresidentialconstruction，wasominous。Thefiguresfor1934，1935，and1936willbemostinterestingwhenwehavethem。Onewouldexpectthattherecoveryofthetwoformeryearshasbeenbasedonareturnofinventoriestonormalandonpublicloanexpenditure，butthatby1936durableinvestmentwasbeginningtosupplantinventoriesinmakingupthetotal。Itisonthecontinuedsteadinessofthefirsttwoitemsoftheabovetableatfiguresnotlessthanthoseof1925to1928thatthemaintenanceofprosperitymustdepend；anditisforthisreasonthatalowlong-termrateofinterestissovitallyimportant。
　　cThemannerinwhichthechangesinpublicloanexpenditurecameintomoderatethefluctuations，whichwouldhaveoccurredotherwise，isveryapparent。Themannerinwhichfrom1931FederalborrowingtooktheplaceofStateandlocalborrowing，asshownintheAppendixbelow，isstriking。From30June1924，to30June1930，Federalloansoutstandingfellfrom21to15billions，whilstinthesameperiodStateandlocalloansrosefrom10to16billions，thetotalremainingunchanged；whereasfrom30June1930to30June1935，Federalloansrosefrom15to26billionsandtheothersfrom16onlyto17billions。Theappendix，whichgivesthefiguresofpublicborrowingupto30June1935，shows¾；contrary，perhaps，tothegeneralimpression¾；thatpublicborrowingwasatitsheightin1931，andthatin1934-35
　　itwasbutlittlemorethanin1929-30。
　　dWhencomparablefiguresofincomeareavailable，weshallbeabletomakesomecomputationsastothevalueoftheMultiplierintheconditionsoftheUnitedStates，thoughtherearemanystatisticaldifficultiesstilltoovercome。If，however，asaverycrude，preliminarytestwetaketheDept。ofCommerceestimatesofincomeuncorrectedforpricechanges，wefindthatduringthelargemovementsoftheyearsfrom1929to1932thechangesinmoney-incomeswerefromthreetofivetimesthechangesinnetinvestmentshownabove。In1933incomesandinvestmentbothincreasedslightly，butthemovementsweretoonarrowtoallowtheratiooftheonetotheothertobecalculatedwithinareasonablemarginoferror。
　　J。M。KEYNES
　　[Page393]APPENDIX
　　2
　　ANotetoAppendix2
　　TOTALANDNETOUTSTANDINGISSUESOFPUBLICDEBT
　　Millionsofdollars。
　　TotaloutstandingissuesNetoutstandingissuesDate30June1
　　Federal2State，county，city，etc。3
　　Combined4
　　Federal5State，county，city，etc。6
　　Combined7Netchange8Averageforcalendaryear9
　　192420，98211，63332，61520，6279，92130，548——
　　192520，21112，83033，04119，73710，97530，712164-43
　　192619，38413，66433，04818，79011，67230，462-250-280
　　192718，25114，73532，98617，54212，61030，152-310-244
　　192817，31815，69933，01716，52213，45229，974-178-10
　　192916，63916，76033，39915，77314，35830，131157441
　　193015，92217，98533，90714，96915，88730，8567251，712
　　193116，52019，18835，70816，09817，45733，5552，6992，822
　　193219，16119，63535，79618，67317，82836，5012，9462，565
　　193322，15819，10741，26521，61317，07236，6852，1842，796
　　193426，48018，94245，42225，32316，77142，0943，4092，173
　　193527，64519，27746，92226，13716，89543，032938-Source:ReportoftheSecretaryoftheTreasuryforyearended30June1935，p。424。
　　Totaloutstandingissuesexcludeasmallvolumeofmaturedandnon-interestbearingobligationsseeibid。，p。379。
　　NetoutstandingissuesareequaltototaloutstandingissueslessthoseheldinU。S。Governmenttrustfunds，orownedbyU。S。Governmentorbygovernmentalagenciesandheldinsinkingfunds。
　　ThetableabovedoesnotincludethecontingentdebtoftheFederalGovernment，i。e。obligationsguaranteedbytheUnitedStates。These，comprisinglargelydebtissuesoftheFederalFarmMortgageCorporation，HomeOwnersLoanCorporationandtheReconstructionFinanceCorporation，wereasfollows:
　　DateMillionsofdollars30June1934691
　　31December19343，079
　　30June19354，151
　　31December19354，525SeeCostofGovernmentintheUnitedStates，bytheNationalIndustrialConferenceBoard，pub。no。223，NewYork，1936，Table26，p。68。
　　MovementsofRealWagesandOutput[Page394]
　　Appendix3
　　FromTheEconomicJournal，March1939
　　RELATIVEMOVEMENTSOFREALWAGESANDOUTPUT
　　AnarticlebyMrJ。G。DunlopinthisJournalSeptember1938，Vol。XLVIII，p。413onTheMovementofRealandMoneyWageRates，andthenotebyMrL。Tarshisprintedbelow[intheEconomicJournal，March1939]p。150，[213]clearlyindicatethatacommonbelieftowhichIaccededinmyGeneralTheoryofEmploymentneedstobereconsidered。
　　Itheresaid:Itwouldbeinterestingtoseetheresultsofastatisticalenquiryintotheactualrelationshipbetweenchangesinmoneywagesandchangesinrealwages。Inthecaseofachangepeculiartoaparticularindustryonewouldexpectthechangeinrealwagestobeinthesamedirectionasthechangeinmoneywages。Butinthecaseofchangesinthegenerallevelofwages，itwillbefound，Ithink，thatthechangeinrealwagesassociatedwithachangeinmoneywages，sofarfrombeingusuallyinthesamedirection，isalmostalwaysintheoppositedirection……Thisisbecause，intheshortperiod，fallingmoneywagesandrisingrealwagesareeach，forindependentreasons，likelytoaccompanydecreasingemployment；
　　labourbeingreadiertoacceptwage-cutswhenemploymentisfallingoff，yetrealwagesinevitablyrisinginthesamecircumstancesonaccountoftheincreasingmarginalreturntoagivencapitalequipmentwhenoutputisdiminished。ButMrDunlop'sinvestigationsintotheBritishstatisticsappeartoshowthat，whenmoneywagesarerising，realwageshaveusuallyrisenalso；
　　whilst，whenmoneywagesarefalling，realwagesarenomorelikelytorisethantofall。AndMrTarshishasreachedbroadlysimilarresultsinrespectofrecentyearsintheUnitedStates。
　　InthepassagequotedabovefrommyGeneralTheoryIwasaccepting，withouttakingcaretocheckthefactsformyself，abeliefwhichhasbeenwidelyheldbyBritisheconomistsuptothelastyearortwo。SincethematerialonwhichMrDunlopmainlydepends¾；namely，theindicesofrealandmoneywages[Page395]APPENDIX
　　3
　　preparedbyMrG。H。WoodandProf。Bowley-¾；havebeenavailabletoallofusformanyyears，itisstrangethatthecorrectionhasnotbeenmadebefore。[214]Buttheunderlyingproblemisnotsimple，andisnotcompletelydisposedofbythestatisticalstudiesinquestion。
　　Firstofallitisnecessarytodistinguishbetweentwodifferentproblems。
　　InthepassagequotedaboveIwasdealingwiththereactionofrealwagestochangesinoutput，andhadinmindsituationswherechangesinrealandmoneywageswereareflectionofchangesinthelevelofemploymentcausedbychangesineffectivedemand。Thisis，infact，thecasewhich，ifIunderstandthemrightly，MrDunlopandMrTarshishaveprimarilyinview。[215]Butthereisalsothecasewherechangesinwagesreflectchangesinpricesorintheconditionsgoverningthewagebargainwhichdonotcorrespondto，orarenotprimarilytheresultof，changesinthelevelofoutputandemploymentandarenotcausedbythoughtheymaycause
　　changesineffectivedemand。ThisquestionIdiscussedinadifferentpartofmyGeneralTheorynamelychapter19，'ChangesinMoneyWages'，whereIreachedtheconclusionthatwagechanges，whicharenotinthefirstinstanceduetochangesinoutput，havecomplexreactionsonoutputwhichmaybeineitherdirectionaccordingtocircumstancesandaboutwhichitisdifficulttogeneralise。ItiswiththefirstproblemonlythatI
　　amconcernedinwhatfollows。[216]
　　Thequestionoftheinfluenceonrealwagesofperiodsofboomanddepressionhasalonghistory。Butweneednotgofartherbackthantheperiodofthe'eightiesand'ninetiesofthelastcentury，whenitwasthesubjectofinvestigationbyvariousofficialbodiesbeforewhichMarshallgaveevidenceorintheworkofwhichhetookpart。IwasmyselfbroughtupupontheevidencehegavebeforetheGoldandSilverCommissionin1887
　　andtheIndianCurrencyCommitteein1899。
　　[217]Itisnotalwaysclear[Page396]THE
　　GENERALTHEORYOFEMPLOYMENT
　　whetherMarshallhasinmindariseinmoneywagesassociatedwithariseinoutput，oronewhichmerelyreflectsachangeinpricesdue，forexample，toachangeinthestandardwhichwastheparticularsubjectonwhichhewasgivingevidence；butinsomepassagesitisevidentthatheisdealingwithchangesinrealwagesattimeswhenoutputisexpanding。
　　Itisclear，however，thathisconclusionisbased，notlikesomelaterargumentsonà；priorigroundsarisingoutofincreasingmarginalcostintheshortperiod，butonstatisticalgroundswhichshowed¾；sohethought¾；thatintheshortperiodwageswerestickierthanprices。InhispreliminarymemorandumfortheGoldandSilverCommissionOfficialPapers，p。19hewrote:'[Duringaslowandgradualfallofprices]apowerfulfrictiontendstopreventmoneywagesinmosttradesfromfallingasfastasprices；andthistendsalmostimperceptiblytoestablishahigherstandardoflivingamongtheworkingclasses，andtodiminishtheinequalitiesofwealth。Thesebenefitsareoftenignored；butinmyopiniontheyareoftennearlyasimportantastheevilswhichresultfromthatgradualfallofpriceswhichissometimescalledadepressionoftrade。'AndwhenMrChaplinaskedhimop。cit。，p。99，'Youthinkthatduringaperiodofdepressiontheemployedworkingclasseshavebeengettingmorethantheydidbefore？'hereplied，'Morethantheydidbefore，ontheaverage。'
　　Subsequently，asappearsfromanimportantletterofApril1897hithertounpublishedtoFoxwell，[218]whoheldsomewhatstronglytheoppositeopinion，Marshall'sopinionbecamerathermoretentative；thoughthefollowingextractrefersmoretohisgeneralattitudetowardsrisingpricesthantotheirparticulareffectonrealwages:Youknow，myviewsonthismatterareanotveryconfident，bnotverywarmlyadvocatedbyme，cnotveryold，d
　　basedentirelyonnon-academicarguments&；observation。
　　Intheyears68to77Iwasstronglyonthesideyounowadvocate。TheobservationofeventsinBristolmademedoubt。In85，or86IwroteaMemnfortheComnon[Page397]APPENDIX
　　3Depressionshowingaslightpreferenceforrisingprices。ButinthefollowingtwoyearsIstudiedthematterclosely，IreadandanalysedtheevidenceofbusinessmenbeforethatCommission；&；bythetimetheGold&；SilverCommissioncame，Ihadjustturnedthecorner。
　　SincethenIhavereadagreatdeal，butalmostexclusivelyofanon-academicorderonthesubject:&；wasthinkingaboutitduhngagreatpartoftheevidencegivenbybusinessmen&；workingmenbeforetheLabourCommission。Ihavefoundagooddealthatisnewtostrengthenmynewconviction，nothingtoshakeit。IamfarfromcertainIamright。IamabsolutelycertainthattheevidencebroughtforwardinprinttothecontraryinEnglandandAmericaIhavenotreadlargelyforothercountriesdoesnotprovewhatitclaimsto，&；doesnotmeetoranticipatemyarguments，inthesimplewayyouseemtoimagine。ShortlyafterwardshebegantoworkathisevidencefortheIndianCurrencyCommitteewhichseemstohavehadtheeffectofconfirminghiminhispreviousopinion。HisfinalconsideredopinionisgiveninQuestion11，781:[219
　　]Iwillconfessthat，fortenorfifteenyearsafterIbegantostudypoliticaleconomy，Iheldthecommondoctrine，thatariseofpriceswasgenerallybeneficialtobusinessmendirectly，andindirectlytotheworkingclasses。But，afterthattime，Ichangedmyviews，andI
　　havebeenconfirmedinmynewopinionsbyfindingthattheyarelargelyheldinAmerica，whichhasrecentlypassedthroughexperiencessomewhatsimilartothoseofEnglandearlyinthecentury。Thereasonsforthechangeinmyopinionareratherlong，andIgavethematsomelengthbeforetheGoldandSilverCommission。Ithink，perhaps，Ihadbettercontentmyselfnowwithcallingyourattentiontothefactthatthestatisticalaspectofthematterisinadifferentpositionnow。Theassertionsthatariseinpricesincreasedtherealwagesoftheworkerweresoconsonantwiththecommonopinionofpeoplewhohadnotspeciallystudiedthematter，thatitwasacceptedalmostasanaxiom；but，withinthelasttenyears，thestatisticsofwageshavebeencarriedsofarincertaincountries，andespeciallyinEnglandandAmerica，thatweareabletobringittothetest。Ihaveaccumulatedagreatnumberoffacts，butnearlyeverythingIhaveaccumulatedisimpliedinthistable。ItiscopiedfromthearticlebyMrBowleyintheEconomicJournalforlastDecember。Itistheresultofworkthathasbeengoingonforanumberofyears，andseemstometobepracticallydecisive。It[Page398]THE
　　GENERALTHEORYOFEMPLOYMENTcollectstheaveragewagesinEnglandfromtheyear1844totheyear1891，andthencalculateswhatpurchasingpowerthewageswouldgiveatthedifferenttimes，anditshowsthattheriseofrealwagesafter1873whenpriceswerefallingwasgreaterthanbefore1873whenpriceswererising。HerefollowsatablefromProf。Bowley'sarticleinthisJournalforDecember1898。Marshall'sfinalconclusionwascrystallisedinapassageinthePrinciplesBookVI，ch。VIII，§；6:'[Whenpricesrisetheemployer]willthereforebemoreableandmorewillingtopaythehighwages；andwageswilltendupwards。Butexperienceshowsthatwhethertheyaregovernedbyslidingscalesornot
　　theyseldomriseasmuchinproportionasprices；andthereforetheydonotrisenearlyasmuchinproportionasprofits。'AlthoughMarshall'sevidencebeforetheIndianCurrencyCommitteewasgivenin1899，Prof。Bowley'sstatisticsonwhichhewasrelyingdonotrelateeffectivelytoadatelaterthan1891or1893atlatest。Itisclear，Ithink，thatMarshall'sgeneralisationwasbasedonexperiencefrom1880
　　to1886whichdidinfactbearitout。Ifwedividetheyearsfrom1880
　　to1914intosuccessiveperiodsofrecoveryanddepression，thebroadresult，allowingfortrend，appearstobeasfollows:
　　Realwages1880-1884RecoveryFalling1884-1886DepressionRising1886-1890RecoveryRising1890-1896DepressionFalling1896-1899RecoveryRising1899-1905DepressionFalling1905-1907RecoveryRising1907-1910DepressionFalling1910-1914RecoveryRisingAccordingtothis，Marshall'sgeneralisationholdsfortheperiodsfrom1880to1884andfrom1884to1886，butfornosubsequentperiods。[220
　　]Itseemsthatwehavebeenlivingalltheseyearsonageneralisationwhichheldgood，byexception，intheyears1880-86，whichwastheformativeperiodinMarshall'sthoughtinthismatter，buthasneveronceheldgoodinthefiftyyearssincehecrystallisedit！ForMarshall'sviewmainlyprevailed，andFoxwell'scontraryopinionwasdiscardedastheheresyofan[Page399]APPENDIX
　　3
　　inflationist。ItistobeobservedthatMarshallofferedhisgeneralisationmerelyasanobservedstatisticalfact，and，beyondexplainingitasprobablyduetowagesbeingstickierthanprices，hedidnotattempttosupportitbyà；priorireasoning。Thefactthatithassurvivedasadogmaconfidentlyacceptedbymygenerationmustbeexplained，Ithink，bythemoretheoreticalsupportwhichithassubsequentlyreceived。
　　TomystatementthatMarshall'sgeneralisationhasremaineduncorrecteduntilrecentlythereis，however，animportantexception。InhisIndustrialFluctuations，publishedin1927，ProfessorPigoupointedoutp。217
　　that'theupperhalvesoftradecycleshave，onthewhole，beenassociatedwithhigherratesofrealwagesthanthelowerhalves，'andheprintedinsupportofthisalargescalechartfortheperiodfrom1850to1910。
　　Subsequently，however，heseemstohaverevertedtotheMarshalliantradition，andinhisTheoryofUnemployment，publishedin1933，hewritesp。296:Ingeneral，thetranslationofinertiafromrealwage-ratestomoneywage-ratescausesrealratestomoveinamannernotcompensatory，butcomplementary，tomovementsintherealdemandfunction。Realwage-ratesnotmerelyfailtofallwhentherealdemandforlabourisfalling，butactuallyrise；and，inlikemanner，whentherealdemandforlabourisexpanding，realwage-ratesfall。AboutthattimeM。Rueffhadattractedmuchattentionbythepublicationofstatisticswhichpurportedtoshowthatariseinrealwagestendedtogowithanincreaseinunemployment，Prof。PigoupointsoutthatthesestatisticsarevitiatedbythefactthatM。Rueffdividedmoneywagesbythewholesaleindexinsteadofbythecost-of-livingindex，andhedoesnotagreewithM。Rueffthattheobservedriseinrealwageswasthemaincauseoftheincreasedunemploymentwithwhichitwasassociated。Butheconcludes，neverthelessp。300，onabalanceofconsiderations，that'therecanbelittledoubtthatinmodernindustrialcommunitiesthislattertendencyi。e。forshiftsinrealdemandtobeassociatedwithshiftsintheoppositesenseintherateofrealwagesforwhichworkpeoplestipulateispredominant'。
　　LikeMarshall，Prof。Pigoubasedhisconclusionprimarilyonthestickinessofmoneywagesrelativelytoprices。Butmyownreadinesstoaccepttheprevailinggeneralisation，atthetimewhenIwaswritingmyGeneralTheory，wasmuchin-
　　[Page400]THE
　　GENERALTHEORYOFEMPLOYMENT
　　fluencedbyanà；prioriargument，whichhadrecentlywonwideacceptance，tobefoundinMrR。F。Kahn'sarticleon'TheRelationofHomeInvestmenttoEmployment，'publishedintheEconomicJournalforJune1931。[221]Thesupposedempiricalfact，thatintheshortperiodrealwagestendtomoveintheoppositedirectiontothelevelofoutput，appeared，thatistosay，tobeinconformitywiththemorefundamentalgeneralisationsthatindustryissubjecttoincreasingmarginalcostintheshortperiod，thatforaclosed[222]systemasawholemarginalcostintheshortperiodissubstantiallythesamethingasmarginalwagecost，andthatincompetitiveconditionspricesaregovernedbymarginalcost；allthisbeingsubject，ofcourse，tovariousqualificationsinparticularcases，butremainingareliablegeneralisationbyandlarge。
　　Inowrecognisethattheconclusionistoosimple，anddoesnotallowsufficientlyforthecomplexityofthefacts。ButIstillholdtothemainstructureoftheargument，andbelievethatitneedstobeamendedratherthandiscarded。ThatIwasaneasyvictimofthetraditionalconclusionbecauseitfittedmytheoryistheoppositeofthetruth。Formyowntheorythisconclusionwasinconvenient，sinceithadatendencytooffsettheinfluenceofthemainforceswhichIwasdiscussingandmadeitnecessaryformetointroducequalifications，whichIneednothavetroubledwithifIcouldhaveadoptedthecontrarygeneralisationfavouredbyFoxwell，MrDunlopandMrTarshis。Inparticular，thetraditionalconclusionplayedanimportantpart，itwillberemembered，inthediscussions，sometenyearsago，astotheeffectofexpansionistpoliciesonemployment，atatimewhenIhadnotdevelopedmyownargumentinascompleteaformasIdidsubsequently。Iwasalreadyarguingatthattimethatthegoodeffectofanexpansionistinvestmentpolicyonemployment，thefactofwhichnoonedenied，wasduetothestimulantwhichitgavetoeffectivedemand。
　　Prof。Pigou，ontheotherhand，andmanyothereconomistsexplainedtheobservedresultbythereductioninrealwagescovertlyeffectedbytheriseinpriceswhichensuedontheincreaseineffectivedemand。Itwasheldthat[Page401]APPENDIX
　　3
　　publicinvestmentpoliciesandalsoanimprovementinthetradebalancethroughtariffsproducedtheireffectbydeceiving，sotospeak，theworkingclassesintoacceptingalowerrealwage，effectingbythismeansthesamefavourableinfluenceonemploymentwhich，accordingtotheseeconomists，wouldhaveresultedfromamoredirectattackonrealwagese。g。byreducingmoneywageswhilstenforcingacreditpolicycalculatedtoleavepricesunchanged。Ifthefallingtendencyofrealwagesinperiodsofrisingdemandisdenied，thisalternativeexplanationmust，ofcourse，falltotheground。SinceIsharedatthetimetheprevailingbeliefastothefacts，Iwasnotinapositiontomakethisdenial。If，however，itprovesrighttoadoptthecontrarygeneralisation，itwouldbepossibletosimplifyconsiderablythemorecomplicatedversionofmyfundamentalexplanationwhichIhaveexpoundedinmyGeneralTheory。[223]Mypracticalconclusionswouldhave，inthatcase，à；fortioriforce。IfwecanadvancefartherontheroadtowardsfullemploymentthanIhadpreviouslysupposedwithoutseriouslyaffectingrealhourlywagesortherateofprofitsperunitofoutput，thewarningsoftheanti-expansionistsneedcauseuslessanxiety。
　　Nevertheless，weshould，Isubmit，hesitatesomewhatandcarryourinquiriesfurtherbeforewediscardtoomuchofourformerconclusionswhich，subjecttotherightqualifications，haveà；priorisupportandhavesurvivedformanyyearsthescrutinyofexperienceandcommonsense。I
　　offer，[224]therefore，forfurtherstatisticalinvestigationananalysisoftheelementsoftheproblemwithaviewtodiscoveringatwhatpointstheweaknessesoftheformerargumentemerge。Therearefiveheadswhichdeserveseparateconsideration。
　　I
　　Firstofall，arethestatisticsonwhichMrDunlopandMrTarshisarerelyingsufficientlyaccurateandsufficientlyuniformintheirindicationstoformthebasisofareliableinduction？
　　Forexample，insorecentacompilationastheLeagueofNationsWorldEconomicSurvey1937-38，preparedbyMrJ。E。Meade，thetraditionalconclusionreceivessupport，notonà；priori[Page402]THE
　　GENERALTHEORYOFEMPLOYMENT
　　grounds，butonthebasisofthemostrecentlyavailablestatistics。
　　Iquotethefollowingfrompp。54-55:Duringthegreatdepressionafter1929，thedemandforgoodsandservicesdiminished，andinconsequencethepriceofcommoditiesfellrapidly。Inmostcountries，ascanbeseenfromthegraphonp。52，hourlymoneywageswerereducedasthedemandforlabourfell；butineverycasetherewasagreaterfallinprices，sothathourlyrealwagesrose……[Itisthenexplainedthatthesamewasnottrueofweeklywages。]……Sincetherecovery，theoppositemovementsmaybeobserved。Inmostcountries，increaseddemandforgoodsandserviceshascausedcommoditypricestorisemorerapidlythanhourlymoneywages，andthehourlyrealwagehasfallen……IntheUnitedStates[225]andFrance，[226]however，theriseinmoneywageswassorapidbetween1936and1937thatthehourlyrealwagecontinuedtorise……Whenrealhourlywagesareraised¾；i。e。
　　whenthemarginbetweencommoditypricesandthemoney-wagecostbecomeslessfavourable¾；employersarelikelytodiminishtheamountofemploymentwhichtheyoffertolabour。Whiletherewere，nodoubt，otherinfluencesaffectingthedemandforlabour，theimportanceofthisfactoriswellillustratedbythegraphonp。53。
　　Inthecaseofallthecountriesrepresentedforwhichinformationisavailable，thefallincommoditypricesbetween1929and1932causedariseinthehourlyrealwage，andthiswasaccompaniedbyadiminutioninemployment……itisshownthatontherecoverytherehasbeenagreatervarietyofexperience……Thisauthoritativestudyhavinginternationalscopeindicatesthatthenewgeneralisationsmustbeacceptedwithreserve。InanycaseMrTarshis'sscatterdiagramprintedbelow[intheEconomicJournal，March1939]
　　p。150，whilstitshowsadefinitepreponderanceinthesouth-westandnorth-eastcompartmentsandahighcoefficientofassociation，includesaconsiderablenumberofdivergentcases，andtheabsoluterangeofmostofthescatterisextremelysmall，withamarkedclusteringintheneighbourhoodofthezerolineforchangesinrealwages；andmuchthesameistrueofMrDunlop'sresults。ThegreatmajorityofMrTarshis'sobservationsrelatetochangesoflessthan1。5percent。IntheintroductiontohisWagesandIncomeintheUnitedKingdomsince1860，Prof。Bowleyindicatesthatthisisprobablylessthanthemarginoferrorforstatisticsofthiskind。Thisgeneralconclusion[Page403]APPENDIX
　　3
　　isreinforcedbythefactthatitishourlywageswhicharerelevantinthepresentcontext，forwhichaccuratestatisticsarenotavailable。
　　[227]Moreover，inthepost-scriptumtohisnote，MrTarshisexplainsthatwhilstrealwagestendtomoveinthesamedirectionasmoneywages，theymoveintheoppositedirection，thoughonlyslightly，tothelevelofoutputasmeasuredbyman-hoursofemployment；fromwhichitappearsthatMrTarshis'sfinalresultisinconformitywithmyoriginalassumption，whichis，ofcourse，concernedwithhourlywages。Itseemspossible，therefore，takingaccountofMrMeade'sresults，thatImaynot，afterall，havebeenseriouslywrong。
　　Furthermore，forreasonsgivenbelow，itisimportanttoseparatetheobservationsaccordingastheabsolutelevelofemploymentisdistinctlygoodoronlymediocre。Itmaybethatwecananalyseourresultssoastogivetwodistinctgeneralisationsaccordingtotheabsolutelevelreachedbyemployment。If，atthepresentstageoftheinquiry，wearetomakeanysinglestatisticalgeneralisation，Ishouldpreferonetotheeffectthat，forfluctuationswithintherangewhichhasbeenusualintheperiodsinvestigatedwhichseldomapproachconditionsoffullemployment，short-periodchangesinrealwagesareusuallysosmallcomparedwiththechangesinotherfactorsthatweshallnotoftengofarwrongifwetreatrealwagesassubstantiallyconstantintheshortperiodaveryhelpfulsimplificationifitisjustified。Theconclusion，thatchangesinrealwagesarenotusuallyanimportantfactorinshort-periodfluctuationsuntilthepointoffullemploymentisapproaching，isonewhichhasbeenalreadyreachedbyDrKaleckionthebasisofhisowninvestigations。[228]
　　II
　　Itmaybethatwehaveunder-estimatedthequantitativeeffectofafactorofwhichwehavealwaysbeenaware。Ourargumentassumedthat，broadlyspeaking，labourisremuneratedintermsofitsowncompositeproduct，oratleastthatthepriceofwage-goodsmovesinthesamewayasthepriceofoutputasawhole。Butnoonehassupposedthatthiswasstrictlythecaseorwasbetterthananapproximation；anditmaybethatthe[Page404]THE
　　GENERALTHEORYOFEMPLOYMENT
　　proportionofwage-goods，whicharenotthecurrentproductofthelabourinquestionandthepricesofwhicharenotgovernedbythemarginalcostofsuchproduct，issogreatastointerferewiththereliabilityofourapproximation。House-rentandgoodsimportedonchangingtermsoftradeareleadingexamplesofthisfactor。Ifintheshortperiodrentsareconstantandthetermsoftradetendtoimprovewhenmoneywagesriseandtodeterioratewhenmoneywagesfall，ourconclusionwillbeupsetinpracticeinspiteoftherestofourpremissesholdinggood。
　　Inthecaseofthiscountryonehasbeeninthehabitofsupposingthatthesetwofactorshaveinfacttendedtooffsetoneanother，thoughtheoppositemightbethecaseintheraw-materialcountries。Forwhereasrents，beinglargelyfixed，riseandfalllessthanmoneywages，thepriceofimportedfood-stuffstendstorisemorethanmoneywagesinperiodsofactivityandtofallmoreinperiodsofdepression。AtanyratebothMrDunlopandMrTarshisclaimtoshowthatfluctuationsinthetermsoftradetermsofforeigntradeinMrDunlop'sBritishinquiryandtermsoftradebetweenindustryandagricultureinMrTarshis'sAmericaninquiryarenotsufficienttoaffectthegeneraltendencyoftheirresults，thoughtheyclearlymodifythemquantitativelytoaconsiderableextent。[229
　　]Nevertheless，theeffectofexpenditureonitemssuchasrent，gas，electricity，water，transport，etc。，ofwhichthepricesdonotchangemateriallyintheshortperiod，needstobeseparatelycalculatedbeforewecanbeclear。
　　Ifitshouldemergethatitisthisfactorwhichexplainstheresults，therestofourfundamentalgeneralisationswouldremainundisturbed。Itisimportant，therefore，ifwearetounderstandthesituation，thatthestatisticiansshouldendeavourtocalculatewagesintermsoftheactualproductofthelabourinquestion。
　　III
　　Hastheidentificationofmarginalcostwithmarginalwagecostintroducedarelevanterror？InmyGeneralTheoryofEmployment，chapter6
　　Appendix，IhavearguedthatthisidentificationisdangerousinthatitignoresafactorwhichIhavecalled'marginalusercost'。Itisunlikely，however，thatthiscanhelpusinthepresentcontext。Formarginalusercostislikelytoincreasewhenoutputisincreasing，sothatthisfactor[Page405]APPENDIX
　　3
　　wouldworkintheoppositedirectionfromthatrequiredtoexplainourpresentproblem，andwouldbeanadditionalreasonforexpectingpricestorisemorethanwages。Indeed，onewould，ongeneralgrounds，expectmarginaltotalcosttoincreasemore，andnotless，thanmarginalwagecost。
　　IV
　　Isittheassumptionofincreasingmarginalrealcostintheshortperiodwhichweoughttosuspect？MrTarshisfindspartoftheexplanationhere；
　　andDrKaleckiisinclinedtoinferapproximatelyconstantmarginalrealcost。[230]Butthereisanimportantdistinctionwhichwehavetomake。
　　Weshouldallagreethatifwestartfromalevelofoutputverygreatlybelowcapacity，sothateventhemostefficientplantandlabourareonlypartiallyemployed，marginalrealcostmaybeexpectedtodeclinewithincreasingoutput，or，attheworst，remainconstant。Butapointmustsurelycome，longbeforeplantandlabourarefullyemployed，whenlessefficientplantandlabourhavetobebroughtintocommissionortheefficientorganisationemployedbeyondtheoptimumdegreeofintensiveness。Evenifoneconcedesthatthecourseoftheshort-periodmarginalcostcurveisdownwardsinitsearlyreaches，MrKahn'sassumptionthatiteventuallyturnsupwardsis，ongeneralcommon-sensegrounds，surelybeyondreasonablequestion；andthatthishappens，moreover，onapartofthecurvewhichishighlyrelevantforpracticalpurposes。Certainlyitwouldrequiremoreconvincingevidencethanyetexiststopersuademetogiveupthispresumption。
　　Nevertheless，itisofgreatpracticalimportancethatthestatisticiansshouldendeavourtodetermineatwhatlevelofemploymentandoutputtheshort-periodmarginal-costcurveforthecompositeproductasawholebeginstoturnupwardandhowsharplyitrisesaftertheturning-pointhasbeenreached。Thisknowledgeisessentialfortheinterpretationofthetradecycle。ItisforthisreasonthatIsuggestedabovethattheobservationsoftherelativemovementofrealandmoneywagesshouldbeseparatelyclassifiedaccordingtotheaveragelevelofemploymentwhichhadbeenreached。
　　Itmayprove，indeed，atanyrateinthecaseofstatisticsrelatingtorecentyearsthatthelevelofemploymenthasbeen[Page406]THE
　　GENERALTHEORYOFEMPLOYMENT
　　preponderantlysolowthatwehavebeenlivingmoreoftenthannotonthereachesofthecurvebeforethecriticalpointofupturnhasbeenattained。
　　ItshouldbenoticedthatMrTarshis'sAmericanfiguresrelateonlytotheperiodfrom1932to1938，duringthewholeofwhichperiodtherehasbeensuchintenseunemploymentintheUnitedStates，bothoflabourandofplant，thatitwouldbequiteplausibletosupposethatthecriticalpointofthemarginalcostcurvehadneverbeenreached。Ifthishasbeenthecase，itisimportantthatweshouldknowit。Butsuchanexperiencemustnotmisleadusintosupposingthatthismustnecessarilybethecase，orintoforgettingthesharplydifferenttheorywhichbecomesapplicableaftertheturning-pointhasbeenreached。
　　If，indeed，theshapeofthemarginal-costcurveprovestobesuchthatwetendtobeliving，withconditionsastheyareatpresent，moreoftentotheleftthantotherightofitscriticalpoint，thepracticalcaseforaplannedexpansionistpolicyisconsiderablyreinforced；formanycaveatstowhichwemustattendafterthispointhasbeenreachedcanbe，inthatcase，frequentlyneglected。Intakingitasmygeneralassumptionthatweareoftenontherightofthecriticalpoint，IhavebeentakingthecaseinwhichthepracticalpolicywhichIhaveadvocatedneedsthemostcarefulhandling。Inparticularthewarningsgiven，quiterightly，byMrD。H。Robertsonofthedangerswhichmayarisewhenweencourageorallowtheactivityofthesystemtoadvancetoorapidlyalongtheupwardslopesofthemarginal-costcurvetowardsthegoaloffullemployment，canbemoreoftenneglected，forthetimebeingatleast，whentheassumptionwhichIhavepreviouslyadmittedasnormalandreasonableisabandoned。
　　V
　　Thereremainsthequestionwhetherthemistakeliesintheapproximateidentificationofmarginalcostwithprice，orratherintheassumptionthatforoutputasawholetheybearamoreorlessproportionaterelationshiptooneanotherirrespectiveoftheintensityofoutput。Foritmaybethecasethatthepracticalworkingsofthelawsofimperfectcompetitioninthemodernquasi-competitivesystemaresuchthat，whenoutputincreasesandmoneywagesrise，pricesriselessthaninproportiontotheincreaseinmarginalmoneycost。Itisscarcelylikely，perhaps，thatthe[Page407]APPENDIX
　　3
　　narrowinggapcouldbesufficienttopreventadeclineinrealwagesinaphaseinwhichmarginalrealcostwasincreasingrapidly。Butitmightbesufficienttooffsettheeffectonrealwagesofamodestriseinmarginalrealcost，andeventodominatethesituationintheeventofthemarginalrealcostcurveprovingtobealmosthorizontaloverasubstantialportionofitsrelevantlength。
　　Itisevidentlypossiblethatsomesuchfactorshouldexist。Itmightbe，inasense，merelyanextensionofthestickinessofpricesofwhichwehavealreadytakenaccountinIIabove。Apartfromthosepriceswhicharevirtuallyconstantintheshortperiod，thereareobviouslymanyotherswhichare，forvariousreasons，moreorlesssticky。Butthisfactorwouldbeparticularlylikelytoemergewhenoutputincreases，insofarasproducersareinfluencedintheirpracticalpricepoliciesandintheirexploitationoftheopportunitiesgiventhembytheimperfectionsofcompetition，bytheirlong-periodaveragecost，andarelessattentivethaneconomiststotheirshort-periodmarginalcost。Indeed，itisrareforanyonebutaneconomisttosupposethatpriceispredominantlygovernedbymarginalcost。Mostbusinessmenaresurprisedbythesuggestionthatitisaclosecalculationofshort-periodmarginalcostorofmarginalrevenuewhichshoulddominatetheirpricepolicies。Theymaintainthatsuchapolicywouldrapidlylandinbankruptcyanyonewhopractisedit。Andifitistruethattheyareproducingmoreoftenthannotonascaleatwhichmarginalcostisfallingwithanincreaseinoutput，theywouldclearlyberight；
　　foritwouldbeonlyonrareoccasionsthattheywouldbecollectinganythingwhatevertowardstheiroverhead。Itis，beyonddoubt，thepracticalassumptionoftheproducerthathispricepolicyoughttobeinfluencedbythefactthatheisnormallyoperatingsubjecttodecreasingaveragecost，evenifintheshort-periodhismarginalcostisrising。Hiseffortistomaintainpriceswhenoutputfallsand，whenoutputincreases，hemayraisethembylessthanthefullamountrequiredtooffsethighercostsincludinghigherwages。Hewouldadmitthatthis，regardedbyhimasthereasonable，prudentandfar-sightedpolicy，goesbytheboardwhen，attheheightoftheboom，heisoverwhelmedbymoreordersthanhecansupply；butevensoheisfilledwithforebodingastotheultimateconsequencesofhisbeingforcedsofarfromtherightandreasonablepolicyoffixinghispricesbyreferencetohislong-periodoverheadaswellashiscurrentcosts。Rightlyorderedcompetitionconsists，inhis[Page408]THE
　　GENERALTHEORYOFEMPLOYMENT
　　opinion，inaproperpressuretosecureanadjustmentofpricestochangesinlong-periodaveragecost；andthesuggestionthatheisbecomingadangerousandanti-socialmonopolistwhenever，byopenortacitagreementwithhiscompetitors，heendeavourstopreventpricesfromhollowingshort-periodmarginalcost，howevermuchthismayfallawayfromlong-periodaveragecost，strikeshimasdisastrous。ItisthefailureofthelatestphaseoftheNewDealintheUnitedStates，incontrasttotheearliestphase，ofwhichtheoppositeistrue，todistinguishbetweenpriceagreementsformaintainingpricesinrightrelationtoaveragelong-periodcostandthosewhichaimatobtainingamonopolisticprofitinexcessofaveragelong-periodcostwhichstrikeshimasparticularlyunfair。
　　Thus，sinceitistheavowedpolicyofindustrialiststobecontentwithasmallergrossprofitperunitofoutputwhenoutputincreasesthanwhenitdeclines，itisnotunlikelythatthispolicymaybe，atleastpartially，operative。Itwouldbeofgreatinterestifthestatisticianscouldshowindetailinwhatwaygrossprofitperunitofoutputchangesindifferentindustrieswithachangingratiobetweenactualandcapacityoutput。Suchaninvestigationshoulddistinguish，ifpossible，betweentheeffectofincreasingoutputonunit-profitandthatofhighercostsintheshapeofhighermoneywagesandotherexpenses。IfitshouldappearthatIncreasingoutputassuchhasatendencytodecreaseunit-profit，itwouldfollowthatthepolicysuggestedaboveisactualaswellasprofessed。
　　If，however，thedeclineinunit-profitappearstobemainlytheresultofatendencyofpricestooffsethighercostsincompletely，irrespectiveofchangesinthelevelofoutput，thenwehavemerelyanexampleofthestickinessofpricesarisingoutoftheimperfectionofcompetitionintrinsictothemarketconditions。Unfortunatelyitisoftendifficultorimpossibletodistinguishclearlybetweentheeffectsofthetwoinfluences，sincehighermoneycostsandincreasingoutputwillgenerallygotogether。
　　Awell-knownstatisticalphenomenonwhichoughttohaveputmeonmyguardconfirmstheprobabilityofconstantordiminishing，ratherthanincreasing，profitperunitofoutputwhenoutputincreases。Imeanthestabilityoftheproportionofthenationaldividendaccruingtolabour，irrespectiveapparentlyofthelevelofoutputasawholeandofthephaseofthetradecycle。Thisisoneofthemostsurprising，yetbest-established，[Page409]APPENDIX
　　3
　　factsinthewholerangeofeconomicstatistics，bothforGreatBritainandfortheUnitedStates。Thefollowingfiguressummarisebrieflywhatare，Ibelieve，theundisputedfacts:[231]
　　RELATIVESHAREOFMANUALLABOURINTHENATIONALINCOMEOFGREATBRITAIN
　　[232]
　　191140。7192443。0192843。0193243。0
　　192540。8192942。4193342。7
　　192642。0193041。1193442。0
　　192743。0193143。7193541。8RELATIVESHAREOFMANUALLABOURINTHENATIONALINCOMEOFU。S。A。[233
　　]
　　191934。9192339。3192737。0193134。9
　　192037。4192437。6192835。8193236。0
　　192135。0192537。1192936。1193337。2
　　192237。0192636。7193035。0193435。8Thefluctuationsinthesefiguresfromyeartoyearappeartobeofarandomcharacter，andcertainlygivenosignificantindicationsofanytendencytomoveagainstlabourinyearsofincreasingoutput。Itisthestabilityoftheratioforeachcountrywhichischieflyremarkable，andthisappearstobealong-run，andnotmerelyashort-period，phenomenon。[234]Moreover，itwouldbeinterestingtodiscoverwhetherthedifferencebetweentheBritishandtheAmericanratioisduetoadiscrepancyinthebasisofreckoningadoptedinthetwosetsofstatisticsortoasignificantdifferenceinthedegreesofmonopolyprevalentinthetwocountriesortotechnicalconditions。
　　Inanycase，thesefactsdonotsupporttherecentlyprevailingassumptionsastotherelativemovementsofrealwagesandoutput，andareinconsistentwiththeideaoftherebeinganymarkedtendencytoincreasingunit-profitwithincreasingoutput。Indeed，eveninthelightoftheaboveconsiderations，theresultremainsabitofamiracle。Forevenifpricepoliciesaresuchas[Page410]THE
　　GENERALTHEORYOFEMPLOYMENT
　　tocauseunit-profittodecreaseinthesamecircumstancesasthoseinwhichmarginalrealcostisincreasing，whyshouldthetwoquantitiesbesorelatedthat，regardlessofotherconditions，themovementoftheonealmostexactlyoffsetsthemovementoftheother？Irecentlyofferedtheproblemofexplainingthis，asEdgeworthwouldhavecalledit，totheresearchstudentsatCambridge。TheonlysolutionwasofferedbyDrKaleckiinthebrilliantarticlewhichhasbeenpublishedinEconometrica。
　　[235]DrKaleckihereemploysahighlyoriginaltechniqueofanalysisintothedistributionalproblembetweenthefactorsofproductioninconditionsofimperfectcompetition，whichmayprovetobeanimportantpieceofpioneerwork。ButthemainupshotiswhatIhaveindicatedabove，andDrKaleckimakes，tothebestofmyunderstanding，nodefiniteprogresstowardsexplainingwhy，whenthereisachangeintheratioofactualtocapacityoutput，thecorrespondingchangesinthedegreeoftheimperfectionofcompetitionshouldsoexactlyoffsetotherchanges。Nordoesheexplainwhythedistributionoftheproductbetweencapitalandlabourshouldbestableinthelongrun，beyondsuggestionthatchangesofonekindalwaysjustservetooffsetchangesofanother；yetitisverysurprisingthatonbalancethereshouldhavebeenaconstantdegreeofmonopolyoverthelasttwentyyearsorlonger。Hisownexplanationisbasedontheassumptionsthatmarginalrealcostsareconstant，thatthedegreeoftheimperfectionofthemarketchangesintheoppositedirectiontooutput，butthatthischangeispreciselyoffsetbythefactthatthepricesofbasicrawmaterialspurchasedbythesystemfromoutsiderelativelytomoneywagesincreaseanddecreasewithoutput。Yetthereisnoobviousreasonwhythesechangesshouldsonearlyoffsetoneanother；anditwouldseemsafernottoassumethatmarginalrealcostsareconstant，buttoconcludethatinactualfact，whenoutputchanges，thechangeinthedegreeoftheimperfectionofthemarketissuchastooffsetthecombinedeffectofchangesinmarginalcostsandofchangesinthepricesofmaterialsboughtfromoutsidethesystemrelativelytomoneywages。ItmaybenoticedthatDrKalecki'sargumentassumestheexistenceofanoppositechangeinthedegreeoftheimperfectionofcompetitionorinthedegreeinwhichproducerstakeadvantageofitwhenoutputincreasesfromthatexpectedbyMrR。F。
　　[Page411]APPENDIX