Chapter10
THEMARGINALPROPENSITYTOCONSUME
ANDTHEMULTIPLIER
Weestablishedinchapter8thatemploymentcanonlyincreaseparipassuwithinvestmentunlessthereisachangeinthepropensitytoconsume。Wecannowcarrythislineofthoughtastagefurther。Foringivencircumstancesadefiniteratio，tobecalledthemultiplier，canbeestablishedbetweenincomeandinvestmentand，subjecttocertainsimplifications，betweenthetotalemploymentandtheemploymentdirectlyemployedoninvestmentwhichweshallcalltheprimaryemployment。
Thisfurtherstepisanintegralpartofourtheoryofemployment，sinceitestablishesapreciserelationship，giventhepropensitytoconsume，betweenaggregateemploymentandincomeandtherateofinvestment。TheconceptionofthemultiplierwasfirstintroducedintoeconomictheorybyMrR。F。Kahninhisarticleon'TheRelationofHomeInvestmenttoUnemployment'EconomicJournal，June1931。Hisargumentinthisarticledependedonthefundamentalnotionthat，ifthepropensitytoconsumeinvarioushypotheticalcircumstancesistogetherwithcertainotherconditions
takenasgivenandweconceivethemonetaryorotherpublicauthoritytotakestepstostimulateortoretardinvestment，thechangeintheamountofemploymentwillbeafunctionofthenetchangeintheamountofinvestment；
anditaimedatlayingdowngeneralprinciplesbywhichtoestimatetheactualquantitativerelationshipbetweenanincre-
[Page114]THE
GENERALTHEORYOFEMPLOYMENT
mentofnetinvestmentandtheincrementofaggregateemploymentwhichwillbeassociatedwithit。Beforecomingtothemultiplier，however，itwillbeconvenienttointroducetheconceptionofthemarginalpropensitytoconsume。
I
Thefluctuationsinrealincomeunderconsiderationinthisbookarethosewhichresultfromapplyingdifferentquantitiesofemploymenti。e。
oflabour-unitstoagivencapitalequipment，sothatrealincomeincreasesanddecreaseswiththenumberoflabour-unitsemployed。If，asweassumeingeneral，thereisadecreasingreturnatthemarginasthenumberoflabour-unitsemployedonthegivencapitalequipmentisincreased，incomemeasuredintermsofwage-unitswillincreasemorethaninproportiontotheamountofemployment，which，inturn，willincreasemorethaninproportiontotheamountofrealincomemeasuredifthatispossibleintermsofproduct。Realincomemeasuredintermsofproductandincomemeasuredintermsofwage-unitswill，however，increaseanddecreasetogetherintheshortperiodwhencapitalequipmentisvirtuallyunchanged。Since，therefore，realincome，intermsofproduct，maybeincapableofprecisenumericalmeasurement，itisoftenconvenienttoregardincomeintermsofwage-unitsYwasanadequateworkingindexofchangesinrealincome。Incertaincontextswemustnotoverlookthefactthat，ingeneral，Ywincreasesanddecreasesinagreaterproportionthanrealincome；butinothercontextsthefactthattheyalwaysincreaseanddecreasetogetherrendersthemvirtuallyinterchangeable。
Ournormalpsychologicallawthat，whentherealincomeofthecommunityincreasesordecreases，itsconsumptionwillincreaseordecreasebutnotsofast，can，therefore，betranslated¾；not，indeed，withabsoluteaccuracybutsubjecttoqualificationswhichareobvious[Page115]THE
MARGINALPROPENSITYTOCONSUME
andcaneasilybestatedinaformallycompletefashionintothepropositionsthatDCwandDYwhavethesamesign，butDYw>DCw，whereCwistheconsumptionintermsofwage-units。Thisismerelyarepetitionofthepropositionalreadyestablishedonp。29above。Letusdefine，then，dCw/dYwasthemarginalpropensitytoconsume。
Thisquantityisofconsiderableimportance，becauseittellsushowthenextincrementofoutputwillhavetobedividedbetweenconsumptionandinvestment。ForDYw=DCwDIw，whereCwandIwaretheincrementsofconsumptionandinvestment；sothatwecanwriteDYw=kDIw，where1-1/kisequaltothemarginalpropensitytoconsume。
Letuscallktheinvestmentmultiplier。Ittellsusthat，whenthereisanincrementofaggregateinvestment，incomewillincreasebyanamountwhichisktimestheincrementofinvestment。
II
MrKahn'smultiplierisalittledifferentfromthis，beingwhatwemaycalltheemploymentmultiplierdesignatedbyk'，sinceitmeasurestheratiooftheincrementoftotalemploymentwhichisassociatedwithagivenincrementofprimaryemploymentintheinvestmentindustries。
Thatistosay，iftheincrementofinvestmentDIwleadstoanincrementofprimaryemploymentDN2intheinvestmentindustries，theincrementoftotalemploymentDN
=k'DN2。
Thereisnoreasoningeneraltosupposethatk=k'。
Forthereisnonecessarypresumptionthattheshapesoftherelevantportionsoftheaggregatesupplyfunctionsfordifferenttypesofindustryaresuchthattheratiooftheincrementofemploymentintheonesetofindustriestotheincrementofdemandwhichhas[Page116]THE
GENERALTHEORYOFEMPLOYMENT
stimulateditwillbethesameasintheothersetofindustries。[
72]Itiseasy，indeed，toconceiveofcases，as，forexample，wherethemarginalpropensitytoconsumeiswidelydifferentfromtheaveragepropensity，inwhichtherewouldbeapresumptioninfavourofsomeinequalitybetweenDYw/DNandDIw/DN2，sincetherewouldbeverydivergentproportionatechangesinthedemandsforconsumption-goodsandinvestment-goodsrespectively。Ifwewishtotakeaccountofsuchpossibledifferencesintheshapesoftherelevantportionsoftheaggregatesupplyfunctionsforthetwogroupsofindustriesrespectively，thereisnodifficultyinrewritingthefollowingargumentinthemoregeneralisedform。Buttoelucidatetheideasinvolved，itwillbeconvenienttodealwiththesimplifiedcasewherek=k'。
Itfollows，therefore，that，iftheconsumptionpsychologyofthecommunityissuchthattheywillchoosetoconsume，e。g。nine-tenthsofanincrementofincome，[73]thenthemultiplierkis10；andthetotalemploymentcausedbye。g。increasedpublicworkswillbetentimestheprimaryemploymentprovidedby[Page117]THE
MARGINALPROPENSITYTOCONSUME
thepublicworksthemselves，assumingnoreductionofinvestmentinotherdirections。Onlyintheeventofthecommunitymaintainingtheirconsumptionunchangedinspiteoftheincreaseinemploymentandhenceinrealincome，willtheincreaseofemploymentberestrictedtotheprimaryemploymentprovidedbythepublicworks。If，ontheotherhand，theyseektoconsumethewholeofanyincrementofincome，therewillbenopointofstabilityandpriceswillrisewithoutlimit。Withnormalpsychologicalsuppositions，anincreaseinemploymentwillonlybeassociatedwithadeclineinconsumptionifthereisatthesametimeachangeinthepropensitytoconsume¾；astheresult，forinstance，ofpropagandaintimeofwarinfavourofrestrictingindividualconsumption；
anditisonlyinthiseventthattheincreasedemploymentininvestmentwillbeassociatedwithanunfavourablerepercussiononemploymentintheindustriesproducingforconsumption。
Thisonlysumsupinaformulawhatshouldbynowbeobvioustothereaderongeneralgrounds。Anincrementofinvestmentintermsofwage-unitscannotoccurunlessthepublicarepreparedtoincreasetheirsavingsintermsofwage-units。Ordinarilyspeaking，thepublicwillnotdothisunlesstheiraggregateincomeintermsofwage-unitsisincreasing。Thustheirefforttoconsumeapartoftheirincreasedincomeswillstimulateoutputuntilthenewlevelanddistributionofincomesprovidesamarginofsavingsufficienttocorrespondtotheincreasedinvestment。Themultipliertellsusbyhowmuchtheiremploymenthastobeincreasedtoyieldanincreaseinrealincomesufficienttoinducethemtodothenecessaryextrasaving，andisafunctionoftheirpsychologicalpropensities。[74]Ifsavingisthepillandconsumptionisthejam，theextrajamhastobeproportionedtothesizeofthe[Page118]THE
GENERALTHEORYOFEMPLOYMENT
additionalpill。Unlessthepsychologicalpropensitiesofthepublicaredifferentfromwhatwearesupposing，wehavehereestablishedthelawthatincreasedemploymentforinvestmentmustnecessarilystimulatetheindustriesproducingforconsumptionandthusleadtoatotalincreaseofemploymentwhichisamultipleoftheprimaryemploymentrequiredbytheinvestmentitself。
Itfollowsfromtheabovethat，ifthemarginalpropensitytoconsumeisnotfarshortofunity，smallfluctuationsininvestmentwillleadtowidefluctuationsinemployment；but，atthesametime，acomparativelysmallincrementofinvestmentwillleadtofullemployment。If，ontheotherhand，themarginalpropensitytoconsumeisnotmuchabovezero，smallfluctuationsininvestmentwillleadtocorrespondinglysmallfluctuationsinemployment；but，atthesametime，itmayrequirealargeincrementofinvestmenttoproducefullemployment。Intheformercaseinvoluntaryunemploymentwouldbeaneasilyremediedmalady，thoughliabletobetroublesomeifitisallowedtodevelop。Inthelattercase，employmentmaybelessvariablebutliabletosettledownatalowlevelandtoproverecalcitranttoanybutthemostdrasticremedies。Inactualfactthemarginalpropensitytoconsumeseemstoliesomewherebetweenthesetwoextremes，thoughmuchnearertounitythantozero；withtheresultthatwehave，inasense，theworstofbothworlds，fluctuationsinemploymentbeingconsiderableand，atthesametime，theincrementininvestmentrequiredtoproducefullemploymentbeingtoogreattobeeasilyhandled。Unfortunatelythefluctuationshavebeensufficienttopreventthenatureofthemaladyfrombeingobvious，whilstitsseverityissuchthatitcannotberemediedunlessitsnatureisunderstood。
Whenfullemploymentisreached，anyattempttoincreaseinvestmentstillfurtherwillsetupatendencyinmoney-pricestorisewithoutlimit，irrespectiveofthemarginalpropensitytoconsume；i。e。weshall[Page119]THE
MARGINALPROPENSITYTOCONSUME
havereachedastateoftrueinflation。[75]Uptothispoint，however，risingpriceswillbeassociatedwithanincreasingaggregaterealincome。
III
Wehavebeendealingsofarwithanetincrementofinvestment。
If，therefore，wewishtoapplytheabovewithoutqualificationtotheeffectofe。g。increasedpublicworks，wehavetoassumethatthereisnooffsetthroughdecreasedinvestmentinotherdirections，¾；andalso，ofcourse，noassociatedchangeinthepropensityofthecommunitytoconsume。MrKahnwasmainlyconcernedinthearticlereferredtoaboveinconsideringwhatoffsetsweoughttotakeintoaccountaslikelytobeimportant，andinsuggestingquantitativeestimates。Forinanactualcasethereareseveralfactorsbesidessomespecificincreaseofinvestmentofagivenkindwhichenterintothefinalresult。If，forexample，agovernmentemploys100，000additionalmenonpublicworks，andifthemultiplierasdefinedaboveis4，itisnotsafetoassumethataggregateemploymentwillincreaseby400，000。Forthenewpolicymayhaveadversereactionsoninvestmentinotherdirections。
ItwouldseemfollowingMrKahnthatthefollowingarelikelyinamoderncommunitytobethefactorswhichitismostimportantnottooverlookthoughthefirsttwowillnotbefullyintelligibleuntilafterBookIV
hasbeenreached:
iThemethodoffinancingthepolicyandtheincreasedworkingcash，requiredbytheincreasedemploymentandtheassociatedriseofprices，mayhavetheeffectofincreasingtherateofinterestandsoretardinginvestmentinotherdirections，unlessthemonetaryauthoritytakesstepstothecontrary；whilst，atthesametime，theincreasedcostofcapitalgoodswillreducetheirmarginalefficiencytotheprivatein-
[Page120]THE
GENERALTHEORYOFEMPLOYMENT
vestor，andthiswillrequireanactualfallintherateofinteresttooffsetit。
iiWiththeconfusedpsychologywhichoftenprevails，thegovernmentprogrammemay，throughitseffecton'confidence'，increaseliquidity-preferenceordiminishthemarginalefficiencyofcapital，which，again，mayretardotherinvestmentunlessmeasuresaretakentooffsetit。
iiiInanopensystemwithforeign-traderelations，somepartofthemultiplieroftheincreasedinvestmentwillaccruetothebenefitofemploymentinforeigncountries，sinceaproportionoftheincreasedconsumptionwilldiminishourowncountry'sfavourableforeignbalance；
sothat，ifweconsideronlytheeffectondomesticemploymentasdistinctfromworldemployment，wemustdiminishthefullfigureofthemultiplier。
Ontheotherhandourowncountrymayrecoveraportionofthisleakagethroughfavourablerepercussionsduetotheactionofthemultiplierintheforeigncountryinincreasingitseconomicactivity。
Furthermore，ifweareconsideringchangesofasubstantialamount，wehavetoallowforaprogressivechangeinthemarginalpropensitytoconsume，asthepositionofthemarginisgraduallyshifted；andhenceinthemultiplier。Themarginalpropensitytoconsumeisnotconstantforalllevelsofemployment，anditisprobablethattherewillbe，asarule，atendencyforittodiminishasemploymentincreases；whenrealincomeincreases，thatistosay，thecommunitywillwishtoconsumeagraduallydiminishingproportionofit。
Therearealsootherfactors，overandabovetheoperationofthegeneralruleJustmentioned，whichmayoperatetomodifythemarginalpropensitytoconsume，andhencethemultiplier；andtheseotherfactorsseemlikely，asarule，toaccentuatethetendencyofthegeneralruleratherthantooffsetjt。For，inthefirstplace，theincreaseofemploymentwilltend，owing[Page121]THE
MARGINALPROPENSITYTOCONSUME
totheeffectofdiminishingreturnsintheshortperiod，toincreasetheproportionofaggregateincomewhichaccruestotheentrepreneurs，whoseindividualmarginalpropensitytoconsumeisprobablylessthantheaverageforthecommunityasawhole。Inthesecondplace，unemploymentislikelytobeassociatedwithnegativesavingincertainquarters，privateorpublic，becausetheunemployedmaybelivingeitheronthesavingsofthemselvesandtheirfriendsoronpublicreliefwhichispartlyfinancedoutofloans；withtheresultthatre-employmentwillgraduallydiminishtheseparticularactsofnegativesavingandreduce，therefore，themarginalpropensitytoconsumemorerapidlythanwouldhaveoccurredfromanequalincreaseinthecommunity'srealincomeaccruingindifferentcircumstances。
Inanycase，themultiplierislikelytobegreaterforasmallnetincrementofinvestmentthanforalargeincrement；sothat，wheresubstantialchangesareinview，wemustbeguidedbytheaveragevalueofthemultiplierbasedontheaveragemarginalpropensitytoconsumeovertherangeinquestion。
MrKahnhasexaminedtheprobablequantitativeresultofsuchfactorsastheseincertainhypotheticalspecialcases。But，clearly，itisnotpossibletocarryanygeneralisationveryfar。Onecanonlysay，forexample，thatatypicalmoderncommunitywouldprobablytendtoconsumenotmuchlessthan80percentofanyincrementofrealincome，ifitwereaclosedsystemwiththeconsumptionoftheunemployedpaidforbytransfersfromtheconsumptionofotherconsumers，sothatthemultiplierafterallowingforoffsetswouldnotbemuchlessthan5。Inacountry，however，whereforeigntradeaccountsfor，say，20percentofconsumptionandwheretheunemployedreceiveoutofloansortheirequivalentupto，say，50percentoftheirnormalconsumptionwheninwork，themultipliermayfallaslowas2or3timestheemploymentpro-
[Page122]THE
GENERALTHEORYOFEMPLOYMENT
videdbyaspecificnewinvestment。Thusagivenfluctuationofinvestmentwillbeassociatedwithamuchlessviolentfluctuationofemploymentinacountryinwhichforeigntradeplaysalargepartandunemploymentreliefisfinancedonalargerscaleoutofborrowingaswasthecase，e。g。inGreatBritainin1931，thaninacountryinwhichthesefactorsarelessimportantasintheUnitedStatesin1932。[76]
Itis，however，tothegeneralprincipleofthemultipliertowhichwehavetolookforanexplanationofhowfluctuationsintheamountofinvestment，whichareacomparativelysmallproportionofthenationalincome，arecapableofgeneratingfluctuationsinaggregateemploymentandincomesomuchgreaterinamplitudethanthemselves。
IV
Thediscussionhasbeencarriedon，sofar，onthebasisofachangeinaggregateinvestmentwhichhasbeenforeseensufficientlyinadvancefortheconsumptionindustriestoadvanceparipassuwiththecapital-goodsindustrieswithoutmoredisturbancetothepriceofconsumption-goodsthanisconsequential，inconditionsofdecreasingreturns，onanincreaseinthequantitywhichisproduced。
Ingeneral，however，wehavetotakeaccountofthecasewheretheinitiativecomesfromanincreaseintheoutputofthecapital-goodsindustrieswhichwasnotfullyforeseen。Itisobviousthataninitiativeofthisdescriptiononlyproducesitsfulleffectonemploymentoveraperiodoftime。Ihavefound，however，indiscussionthatthisobviousfactoftengivesrisetosomeconfusionbetweenthelogicaltheoryofthemultiplier，whichholdsgoodcontinuously，withouttime-lag，atallmomentsoftime，andtheconsequencesofanexpansioninthecapital-goodsindustrieswhichtakegradualeffect，subjecttotime-lagandonlyafteraninterval。
[Page123]THE
MARGINALPROPENSITYTOCONSUME
Therelationshipbetweenthesetwothingscanbeclearedupbypointingout，firstlythatanunforeseen，orimperfectlyforeseen，expansioninthecapital-goodsindustriesdoesnothaveaninstantaneouseffectofequalamountontheaggregateofinvestmentbutcausesagradualincreaseofthelatter；and，secondly，thatitmaycauseatemporarydepartureofthemarginalpropensitytoconsumeawayfromitsnormalvalue，followed，however，byagradualreturntoit。
Thusanexpansioninthecapital-goodsindustriescausesaseriesofincrementsinaggregateinvestmentoccurringinsuccessiveperiodsoveranintervaloftime，andaseriesofvaluesofthemarginalpropensitytoconsumeinthesesuccessiveperiodswhichdifferbothfromwhatthevalueswouldhavebeeniftheexpansionhadbeenforeseenandfromwhattheywillbewhenthecommunityhassettleddowntoanewsteadylevelofaggregateinvestment。Butineveryintervaloftimethetheoryofthemultiplierholdsgoodinthesensethattheincrementofaggregatedemandisequaltotheproductoftheincrementofaggregateinvestmentandthemultiplierasdeterminedbythemarginalpropensitytoconsume。
Theexplanationofthesetwosetsoffactscanbeseenmostclearlybytakingtheextremecasewheretheexpansionofemploymentinthecapital-goodsindustriesissoentirelyunforeseenthatinthefirstinstancethereisnoincreasewhateverintheoutputofconsumption-goods。Inthiseventtheeffortsofthosenewlyemployedinthecapital-goodsindustriestoconsumeaproportionoftheirincreasedincomeswillraisethepricesofconsumption-goodsuntilatemporaryequilibriumbetweendemandandsupplyhasbeenbroughtaboutpartlybythehighpricescausingapostponementofconsumption，partlybyaredistributionofincomeinfavourofthesavingclassesasaneffectoftheincreasedprofitsresultingfromthehigherprices，andpartlybythehigherpricescausingadepletion[Page124]THE
GENERALTHEORYOFEMPLOYMENT
ofstocks。Sofarasthebalanceisrestoredbyapostponementofconsumptionthereisatemporaryreductionofthemarginalpropensitytoconsume，i。e。
ofthemultiplieritself，andinsofarasthereisadepletionofstocks，aggregateinvestmentincreasesforthetimebeingbylessthantheincrementofinvestmentinthecapital-goodsindustries，¾；i。e。
thethingtobemultiplieddoesnotincreasebythefullincrementofinvestmentinthecapital-goodsindustries。Astimegoeson，however，theconsumption-goodsindustriesadjustthemselvestothenewdemand，sothatwhenthedeferredconsumptionisenjoyed，themarginalpropensitytoconsumerisestemporarilyaboveitsnormallevel，tocompensatefortheextenttowhichitpreviouslyfellbelowit，andeventuallyreturnstoitsnormallevel；whilsttherestorationofstockstotheirpreviousfigurecausestheincrementofaggregateinvestmenttobetemporarilygreaterthantheincrementofinvestmentinthecapital-goodsindustriestheincrementofworkingcapitalcorrespondingtothegreateroutputalsohavingtemporarilythesameeffect。
Thefactthatanunforeseenchangeonlyexercisesitsfulleffectonemploymentoveraperiodoftimeisimportantincertaincontexts；¾；inparticularitplaysapartintheanalysisofthetradecycleonlinessuchasIfollowedinmyTreatiseonMoney。Butitdoesnotinanywayaffectthesignificanceofthetheoryofthemultiplierassetforthinthischapter；norrenderitinapplicableasanindicatorofthetotalbenefittoemploymenttobeexpectedfromanexpansioninthecapital。
goodsindustries。Moreover，exceptinconditionswheretheconsumptionindustriesarealreadyworkingalmostatcapacitysothatanexpansionofoutputrequiresanexpansionofplantandnotmerelythemoreintensiveemploymentoftheexistingplant，thereisnoreasontosupposethatmorethanabriefintervaloftimencedelapsebeforeemploymentintheconsumptionindustriesisadvancingparipassuwith[Page125]THE
MARGINALPROPENSITYTOCONSUME
employmentinthecapital-goodsindustrieswiththemultiplieroperatingnearitsnormalfigure。
V
Wehaveseenabovethatthegreaterthemarginalpropensitytoconsume，thegreaterthemultiplier，andhencethegreaterthedisturbancetoemploymentcorrespondingtoagivenchangeininvestment。Thismightseemtoleadtotheparadoxicalconclusionthatapoorcommunityinwhichsavingisaverysmallproportionofincomewillbemoresubjecttoviolentfluctuationsthanawealthycommunitywheresavingisalargerproportionofincomeandthemultiplierconsequentlysmaller。
Thisconclusion，however，wouldoverlookthedistinctionbetweentheeffectsofthemarginalpropensitytoconsumeandthoseoftheaveragepropensitytoconsume。Forwhilstahighmarginalpropensitytoconsumeinvolvesalargerproportionateeffectfromagivenpercentagechangeininvestment，theabsoluteeffectwill，nevertheless，besmalliftheaveragepropensitytoconsumeisalsohigh。Thismaybeillustratedasfollowsbyanumericalexample。
Letussupposethatacommunity'spropensitytoconsumeissuchthat，solongasitsrealincomedoesnotexceedtheoutputfromemploying5，000，000
menonitsexistingcapitalequipment，itconsumesthewholeofitsincome；
thatoftheoutputofthenext100，000additionalmenemployeditconsumes99percent，ofthenext100，000afterthat98percent，ofthethird100，000
97percentandsoon；andthat10，000，000menemployedrepresentsfullemployment。Itfollowsfromthisthat，when5，000，000n100，000
menareemployed，themultiplieratthemarginis100/n，and[nni]/[250n]percentofthenationalincomeisinvested。
[Page126]THE
GENERALTHEORYOFEMPLOYMENT
Thuswhen5，200，000menareemployedthemultiplierisverylarge，namely50，butinvestmentisonlyatriflingproportionofcurrentincome，namely，0。06percent；withtheresultthatifinvestmentfallsoffbyalargeproportion，sayabouttwo-thirds，employmentwillonlydeclineto5，100，000，i。e。byabout2percent。Ontheotherhand，when9，000，000menareemployed，themarginalmultiplieriscomparativelysmall，namely2½；，butinvestmentisnowasubstantialproportionofcurrentincome，namely，9percent；
withtheresultthatifinvestmentfallsbytwo-thirds，employmentwilldeclineto6，900，000，namely，by19percent。Inthelimitwhereinvestmentfallsofftozero，employmentwilldeclinebyabout4percentintheformercase，whereasinthelattercaseitwilldeclineby44percent。[77]
Intheaboveexample，thepoorerofthetwocommunitiesundercomparisonispoorerbyreasonofunder-employment。Butthesamereasoningappliesbyeasyadaptationifthepovertyisduetoinferiorskill，techniqueorequipment。Thuswhilstthemultiplierislargerinapoorcommunity，theeffectonemploymentoffluctuationsininvestmentwillbemuchgreaterinawealthycommunity，assumingthatinthelattercurrentinvestmentrepresentsamuchlargerproportionofcurrentoutput。[78]
[Page127]THE
MARGINALPROPENSITYTOCONSUME
Itisalsoobviousfromtheabovethattheemploymentofagivennumberofmenonpublicworkswillontheassumptionsmadehaveamuchlargereffectonaggregateemploymentatatimewhenthereissevereunemployment，thanitwillhavelateronwhenfullemploymentisapproached。Intheaboveexample，if，atatimewhenemploymenthasfallento5，200，000，anadditional100，000menareemployedonpublicworks，totalemploymentwillriseto6，400，000。Butifemploymentisalready9，000，000whentheadditional100，000
menaretakenonforpublicworks，totalemploymentwillonlyriseto9，200，000。
Thuspublicworksevenofdoubtfulutilitymaypayforthemselvesoverandoveragainatatimeofsevereunemployment，ifonlyfromthediminishedcostofreliefexpenditure，providedthatwecanassumethatasmallerproportionofincomeissavedwhenunemploymentisgreater；buttheymaybecomeamoredoubtfulpropositionasastateoffullemploymentisapproached。
Furthermore，ifourassumptioniscorrectthatthemarginalpropensitytoconsumefallsoffsteadilyasweapproachfullemployment，itfollowsthatitwillbecomemoreandmoretroublesometosecureafurthergivenincreaseofemploymentbyfurtherincreasinginvestment。Itshouldnotbedifficulttocompileachartofthemarginalpropensitytoconsumeateachstageofatradecyclefromthestatisticsiftheywereavailable
ofaggregateincorneandaggregateinvestmentatsuccessivedates。Atpresent，however，ourstatisticsarenotaccurateenoughorcompiledsufficientlywiththisspecificobjectinviewtoallowustoinfermorethanhighlyapproximateestimates。Thebestforthepurpose，ofwhichIamaware，areMrKuznets'figuresfortheUnitedStatesalreadyreferredto，p。103above，thoughtheyare，nevertheless，veryprecarious。Takeninconjunctionwithestimatesofnationalincomethesesuggest，forwhattheyareworth，bothalowerfigureandamorestablefigurefortheinvestmentmultiplier[Page128]THE
GENERALTHEORYOFEMPLOYMENT
thanIshouldhaveexpected。Ifsingleyearsaretakeninisolation，theresultslookratherwild。Butiftheyaregroupedinpairs，themultiplierseemstohavebeenlessthan3andprobablyfairlystableintheneighbourhoodof2。5。Thissuggestsamarginalpropensitytoconsumenotexceeding6oto70percent¾；afigurequiteplausiblefortheboom，butsurprisingly，and，inmyjudgment，improbablylowfortheslump。Itispossible，however，thattheextremefinancialconservatismofcorporatefinanceintheUnitedStates，evenduringtheslump，mayaccountforit。Inotherwords，if，wheninvestmentisfallingheavilythroughafailuretoundertakerepairsandreplacements，financialprovisionismade，nevertheless，inrespectofsuchwastage，theeffectistopreventtheriseinthemarginalpropensitytoconsumewhichwouldhaveoccurredotherwise。IsuspectthatthisfactormayhaveplayedasignificantpartinaggravatingthedegreeoftherecentslumpintheUnitedStates。Ontheotherhand，itispossiblethatthestatisticssomewhatoverstatethedeclineininvestment，whichisallegedtohavefallenoffbymorethan75percentin1932comparedwith1929，whilstnet'capitalformation'
declinedbymorethan95percent；¾；amoderatechangeintheseestimatesbeingcapableofmakingasubstantialdifferencetothemultiplier。
VI
Wheninvoluntaryunemploymentexists，themarginaldisutilityoflabourisnecessarilylessthantheutilityofthemarginalproduct。Indeeditmaybemuchless。Foramanwhohasbeenlongunemployedsomemeasureoflabour，insteadofinvolvingdisutility，mayhaveapositiveutility。Ifthisisaccepted，theabovereasoningshowshow'wasteful'loanexpenditure[79]mayneverthelessenrichthecom-
[Page129]THE
MARGINALPROPENSITYTOCONSUME
munityonbalance。Pyramid-building，earthquakes，evenwarsmayservetoincreasewealth，iftheeducationofourstatesmenontheprinciplesoftheclassicaleconomicsstandsinthewayofanythingbetter。
Itiscurioushowcommonsense，wrigglingforanescapefromabsurdconclusions，hasbeenapttoreachapreferenceforwholly'wasteful'
formsofloanexpenditureratherthanforpartlywastefulforms，which，becausetheyarenotwhollywasteful，tendtobejudgedonstrict'business'principles。Forexample，unemploymentrelieffinancedbyloansismorereadilyacceptedthanthefinancingofimprovementsatachargebelowthecurrentrateofinterest；whilsttheformofdiggingholesinthegroundknownasgold-mining，whichnotonlyaddsnothingwhatevertotherealwealthoftheworldbutinvolvesthedisutilityoflabour，isthemostacceptableofallsolutions。
IftheTreasuryweretofilloldbottleswithbanknotes，burythematsuitabledepthsindisusedcoalmineswhicharethenfilleduptothesurfacewithtownrubbish，andleaveittoprivateenterpriseonwell-triedprinciplesoflaissez-fairetodigthenotesupagaintherighttodosobeingobtained，ofcourse，bytenderingforleasesofthenote-bearingterritory，thereneedbenomoreunemploymentand，withthehelpoftherepercussions，therealincomeofthecommunity，anditscapitalwealthalso，wouldprobablybecomeagooddealgreaterthanitactuallyis。Itwould，indeed，bemoresensibletobuildhousesandthelike；butiftherearepoliticalandpracticaldifficultiesinthewayofthis，theabovewouldbebetterthannothing。
Theanalogybetweenthisexpedientandthegold-
[Page130]THE
GENERALTHEORYOFEMPLOYMENT
minesoftherealworldiscomplete。Atperiodswhengoldisavailableatsuitabledepthsexperienceshowsthattherealwealthoftheworldincreasesrapidly；andwhenbutlittleofitissoavailableourwealthsuffersstagnationordecline。Thusgold-minesareofthegreatestvalueandimportancetocivilisation。Justaswarshavebeentheonlyformoflarge-scaleloanexpenditurewhichstatesmenhavethoughtjustifiable，sogold-miningistheonlypretextfordiggingholesinthegroundwhichhasrecommendeditselftobankersassoundfinance；andeachoftheseactivitieshasplayeditspartinprogress¾；failingsomethingbetter。Tomentionadetail，thetendencyinslumpsforthepriceofgoldtoriseintermsoflabourandmaterialsaidseventualrecovery，becauseitincreasesthedepthatwhichgold-diggingpaysandlowerstheminimumgradeoforewhichispayable。
Inadditiontotheprobableeffectofincreasedsuppliesofgoldontherateofinterest，gold-miningisfortworeasonsahighlypracticalformofinvestment，ifweareprecludedfromincreasingemploymentbymeanswhichatthesametimeincreaseourstockofusefulwealth。Inthefirstplace，owingtothegamblingattractionswhichitoffersitiscarriedonwithouttooclosearegardtotherulingrateofinterest。Inthesecondplacetheresult，namely，theincreasedstockofgold，doesnot，asinothercases，havetheeffectofdiminishingitsmarginalutility。Sincethevalueofahousedependsonitsutility，everyhousewhichisbuiltservestodiminishtheprospectiverentsobtainablefromfurtherhouse-buildingandthereforelessenstheattractionoffurthersimilarinvestmentunlesstherateofinterestisfallingparipassu。Butthefruitsofgold-miningdonotsufferfromthisdisadvantage，andacheckcanonlycomethroughariseofthewage-unitintermsofgold，whichisnotlikelytooccurunlessanduntilemploymentissubstantiallybetter。Moreover，thereisnosubsequentreverseeffectonaccountofprovisionfor[Page131]THE
MARGINALPROPENSITYTOCONSUME
userandsupplementarycosts，asinthecaseoflessdurableformsofwealth。
AncientEgyptwasdoublyfortunate，anddoubtlessowedtothisitsfabledwealth，inthatitpossessedtwoactivities，namely，pyramid-buildingaswellasthesearchforthepreciousmetals，thefruitsofwhich，sincetheycouldnotservetheneedsofmanbybeingconsumed，didnotstalewithabundance。TheMiddleAgesbuiltcathedralsandsangdirges。Twopyramids，twomassesforthedead，aretwiceasgoodasone；butnotsotworailwaysfromLondontoYork。Thuswearesosensible，haveschooledourselvestosocloseasemblanceofprudentfinanciers，takingcarefulthoughtbeforeweaddtothe'financial'burdensofposteritybybuildingthemhousestolivein，thatwehavenosucheasyescapefromthesufferingsofunemployment。
WehavetoacceptthemasaninevitableresultofapplyingtotheconductoftheStatethemaximswhicharebestcalculatedto'enrich'anindividualbyenablinghimtopileupclaimstoenjoymentwhichhedoesnotintendtoexerciseatanydefinitetime。
TheMarginalEfficiencyofCapital[Page135]
Chapter11
THEMARGINALEFFICIENCYOFCAPITAL
I
Whenamanbuysaninvestmentorcapital-asset，hepurchasestherighttotheseriesofprospectivereturns，whichheexpectstoobtainfromsellingitsoutput，afterdeductingtherunningexpensesofobtainingthatoutput，duringthelifeoftheasset。ThisseriesofannuitiesQ1，Q2，……Qnitisconvenienttocalltheprospectiveyieldoftheinvestment。
Overagainsttheprospectiveyieldoftheinvestmentwehavethesupplypriceofthecapital-asset，meaningbythis，notthemarket-priceatwhichanassetofthetypeinquestioncanactuallybepurchasedinthemarket，butthepricewhichwouldjustinduceamanufacturernewlytoproduceanadditionalunitofsuchassets，i。e。whatissometimescalleditsreplacementcost。Therelationbetweentheprospectiveyieldofacapital-assetanditssupplypriceorreplacementcost，i。e。therelationbetweentheprospectiveyieldofonemoreunitofthattypeofcapitalandthecostofproducingthatunit，furnishesuswiththemarginalefficiencyofcapitalofthattype。Moreprecisely，Idefinethemarginalefficiencyofcapitalasbeingequaltothatrateofdiscountwhichwouldmakethepresentvalueoftheseriesofannuitiesgivenbythereturnsexpectedfromthecapital-assetduringitslifejustequaltoitssupplyprice。
Thisgivesusthemarginalefficienciesofparticulartypesofcapital-assets。
Thegreatestof[Page136]THE
GENERALTHEORYOFEMPLOYMENT
thesemarginalefficienciescanthenberegardedasthemarginalefficiencyofcapitalingeneral。
Thereadershouldnotethatthemarginalefficiencyofcapitalisheredefinedintermsoftheexpectationofyieldandofthecurrentsupplypriceofthecapital-asset。Itdependsontherateofreturnexpectedtobeobtainableonmoneyifitwereinvestedinanewlyproducedasset；notonthehistoricalresultofwhataninvestmenthasyieldedonitsoriginalcostifwelookbackonitsrecordafteritslifeisover。
Ifthereisanincreasedinvestmentinanygiventypeofcapitalduringanyperiodoftime，themarginalefficiencyofthattypeofcapitalwilldiminishastheinvestmentinitisincreased，partlybecausetheprospectiveyieldwillfallasthesupplyofthattypeofcapitalisincreased，andpartlybecause，asarule，pressureonthefacilitiesforproducingthattypeofcapitalwillcauseitssupplypricetoincrease；thesecondofthesefactorsbeingusuallythemoreimportantinproducingequilibriumintheshortrun，butthelongertheperiodinviewthemoredoesthefirstfactortakeitsplace。Thusforeachtypeofcapitalwecanbuildupaschedule，showingbyhowmuchinvestmentinitwillhavetoincreasewithintheperiod，inorderthatitsmarginalefficiencyshouldfalltoanygivenfigure。Wecanthenaggregatetheseschedulesforallthedifferenttypesofcapital，soastoprovideaschedulerelatingtherateofaggregateinvestmenttothecorrespondingmarginalefficiencyofcapitalingeneralwhichthatrateofinvestmentwillestablish。Weshallcallthistheinvestmentdemand-schedule；or，alternatively，thescheduleofthemarginalefficiencyofcapital。
Nowitisobviousthattheactualrateofcurrentinvestmentwillbepushedtothepointwherethereisnolongeranyclassofcapital-assetofwhichthemarginalefficiencyexceedsthecurrentrateofinterest。
Inotherwords，therateofinvestmentwillbepushedtothe[Page137]THE
MARGINALEFFICIENCYOFCAPITAL
pointontheinvestmentdemand-schedulewherethemarginalefficiencyofcapitalingeneralisequaltothemarketrateofinterest。[80]
Thesamethingcanalsobeexpressedasfollows。IfQristheprospectiveyieldfromanassetattimer，anddristhepresentvalueof£；1deferredryearsatthecurrentrateofinterest，SQrdristhedemandpriceoftheinvestment；andinvestmentwillbecarriedtothepointwhereSQrdrbecomesequaltothesupplypriceoftheinvestmentasdefinedabove。If，ontheotherhand，SQrdrfallsshortofthesupplyprice，therewillbenocurrentinvestmentintheassetinquestion。
Itfollowsthattheinducementtoinvestdependspartlyontheinvestmentdemand-scheduleandpartlyontherateofinterest。OnlyattheconclusionofBookIVwillitbepossibletotakeacomprehensiveviewofthefactorsdeterminingtherateofinvestmentintheiractualcomplexity。Iwould，however，askthereadertonoteatoncethatneithertheknowledgeofanasset'sprospectiveyieldnortheknowledgeofthemarginalefficiencyoftheassetenablesustodeduceeithertherateofinterestorthepresentvalueoftheasset。Wemustascertaintherateofinterestfromsomeothersource，andonlythencanwevaluetheassetby'capitalising'itsprospectiveyield。
II
Howistheabovedefinitionofthemarginalefficiencyofcapitalrelatedtocommonusage？TheMarginalProductivityorYieldorEfficiencyorUtilityofCapitalarefamiliartermswhichwehaveallfrequentlyused。Butitisnoteasybysearchingtheliteratureofeconomicsto[Page138]THE
GENERALTHEORYOFEMPLOYMENT
findaclearstatementofwhateconomistshaveusuallyintendedbytheseterms。
Thereareatleastthreeambiguitiestoclearup。Thereis，tobeginwith，theambiguitywhetherweareconcernedwiththeincrementofphysicalproductperunitoftimeduetotheemploymentofonemorephysicalunitofcapital，orwiththeincrementofvalueduetotheemploymentofonemorevalueunitofcapital。Theformerinvolvesdifficultiesastothedefinitionofthephysicalunitofcapital，whichIbelievetobebothinsolubleandunnecessary。Itis，ofcourse，possibletosaythattenlabourerswillraisemorewheatfromagivenareawhentheyareinapositiontomakeuseofcertainadditionalmachines；butIknownomeansofreducingthistoanintelligiblearithmeticalratiowhichdoesnotbringinvalues。
Neverthelessmanydiscussionsofthissubjectseemtobemainlyconcernedwiththephysicalproductivityofcapitalinsomesense，thoughthewritersfailtomakethemselvesclear。
Secondly，thereisthequestionwhetherthemarginalefficiencyofcapitalissomeabsolutequantityoraratio。Thecontextsinwhichitisusedandthepracticeoftreatingitasbeingofthesamedimensionastherateofinterestseemtorequirethatitshouldbearatio。Yetitisnotusuallymadeclearwhatthetwotermsoftheratioaresupposedtobe。
Finally，thereisthedistinction，theneglectofwhichhasbeenthemaincauseofconfusionandmisunderstanding，betweentheincrementofvalueobtainablebyusinganadditionalquantityofcapitalintheexistingsituation，andtheseriesofincrementswhichitisexpectedtoobtainoverthewholelifeoftheadditionalcapitalasset；¾；i。e。
thedistinctionbetweenQ1andthecompleteseriesQ1，Q2，……Qr，……Thisinvolvesthewholequestionoftheplaceofexpectationineconomictheory。MostdiscussionsofthemarginalefficiencyofcapitalseemtopaynoattentiontoanymemberoftheseriesexceptQ1。Yetthiscannotbe[Page139]THE
MARGINALEFFICIENCYOFCAPITAL
legitimateexceptinaStatictheory，forwhichalltheQ'sareequal。Theordinarytheoryofdistribution，whereitisassumedthatcapitalisgettingnowitsmarginalproductivityinsomesenseorother，isonlyvalidinastationarystate。Theaggregatecurrentreturntocapitalhasnodirectrelationshiptoitsmarginalefficiency；whilstitscurrentreturnatthemarginofproductioni。e。thereturntocapitalwhichentersintothesupplypriceofoutputisitsmarginalusercost，whichalsohasnocloseconnectionwithitsmarginalefficiency。
Thereis，asIhavesaidabove，aremarkablelackofanyclearaccountofthematter。AtthesametimeIbelievethatthedefinitionwhichIhavegivenaboveisfairlyclosetowhatMarshallintendedtomeanbytheterm。
ThephrasewhichMarshallhimselfusesis'marginalnetefficiency'ofafactorofproduction；or，alternatively，the'marginalutilityofcapital'。
ThefollowingisasummaryofthemostrelevantpassagewhichIcanfindinhisPrinciples6thed。pp。519-520。
Ihaveruntogethersomenon-consecutivesentencestoconveythegistofwhathesays:
Inacertainfactoryanextra£；100worthofmachinerycanbeappliedsoasnottoinvolveanyotherextraexpense，andsoastoaddannually£；3worthtothenetoutputofthefactoryafterallowingforitsownwearandtear。Iftheinvestorsofcapitalpushitintoeveryoccupationinwhichitseemslikelytogainahighreward；andif，afterthishasbeendoneandequilibriumhasbeenfound，itstillpaysandonlyjustpaystoemploythismachinery，wecaninferfromthisfactthattheyearlyrateofinterestis3percent。Butillustrationsofthiskindmerelyindicatepartoftheactionofthegreatcauseswhichgovernvalue。Theycannotbemadeintoatheoryofinterest，anymorethanintoatheoryofwages，withoutreasoninginacircle……Supposethattherateofinterestis3percent。perannumonperfectlygoodsecurity；andthatthehat-makingtradeabsorbsacapitalofonemillionpounds。Thisimpliesthatthehat-makingtradecanturnthewholemillionpounds'worthofcapitaltosogoodaccountthattheywouldpay3percent。perannumnetforthe[Page140]THE
GENERALTHEORYOFEMPLOYMENT
useofitratherthangowithoutanyofit。Theremaybemachinerywhichthetradewouldhaverefusedtodispensewithiftherateofinteresthadbeen20percent。perannum。Iftheratehadbeen10percent。，morewouldhavebeenused；ifithadbeen6percent。，stillmore；if4percent。
stillmore；andfinally，theratebeing3percent。，theyusemorestill。
Whentheyhavethisamount，themarginalutilityofthemachinery，i。e。
theutilityofthatmachinerywhichitisonlyjustworththeirwhiletoemploy，ismeasuredby3percent。
ItisevidentfromtheabovethatMarshallwaswellawarethatweareinvolvedinacircularargumentifwetrytodeterminealongtheselineswhattherateofinterestactuallyis。[81]Inthispassageheappearstoaccepttheviewsetforthabove，thattherateofinterestdeterminesthepointtowhichnewinvestmentwillbepushed，giventhescheduleofthemarginalefficiencyofcapital。Iftherateofinterestis3percent，thismeansthatnoonewillpay£；100foramachineunlesshehopestherebytoadd£；3tohisannualnetoutputafterallowingforcostsanddepreciation。Butweshallseeinchapter14thatinotherpassagesMarshallwaslesscautious¾；thoughstilldrawingbackwhenhisargumentwasleadinghimontodubiousground。
Althoughhedoesnotcallitthe'marginalefficiencyofcapital'，ProfessorIrvingFisherhasgiveninhisTheoryofInterest1930adefinitionofwhathecalls'therateofreturnovercost'whichisidenticalwithmydefinition。'Therateofreturnovercost'，hewrites，[82]'isthatratewhich，employedincomputingthepresentworthofallthecostsandthepresentworthofallthereturns，willmakethesetwoequal。'ProfessorFisherexplainsthattheextentofinvestmentinanydirectionwilldependonacomparisonbetweentherateofreturnovercostandtherateofinterest。
Toinducenewinvestment'therateofreturnovercost[Page141]THE
MARGINALEFFICIENCYOFCAPITAL
mustexceedtherateofinterest'。[83]'Thisnewmagnitudeorfactor
inourstudyplaysthecentralrô；leontheinvestmentopportunitysideofinteresttheory。'[84]ThusProfessorFisheruseshis'rateofreturnovercostinthesamesenseandforpreciselythesamepurposeasIemploy'themarginalefficiencyofcapital'。
III
Themostimportantconfusionconcerningthemeaningandsignificanceofthemarginalefficiencyofcapitalhasensuedonthefailuretoseethatitdependsontheprospectiveyieldofcapital，andnotmerelyonitscurrentyield。Thiscanbebestillustratedbypointingouttheeffectonthemarginalefficiencyofcapitalofanexpectationofchangesintheprospectivecostofproduction，whetherthesechangesareexpectedtocomefromchangesinlabourcost，i。e。inthewage-unit，orfrominventionsandnewtechnique。Theoutputfromequipmentproducedto-daywillhavetocompete，inthecourseofitslife，withtheoutputfromequipmentproducedsubsequently，perhapsatalowerlabourcost，perhapsbyanimprovedtechnique，whichiscontentwithalowerpriceforitsoutputandwillbeincreasedinquantityuntilthepriceofitsoutputhasfallentothelowerfigurewithwhichitiscontent。Moreover，theentrepreneur'sprofitintermsofmoneyfromequipment，oldornew，willbereduced，ifalloutputcomestobeproducedmorecheaply。Insofarassuchdevelopmentsareforeseenasprobable，orevenaspossible，themarginalefficiencyofcapitalproducedto-dayisappropriatelydiminished。
Thisisthefactorthroughwhichtheexpectationofchangesinthevalueofmoneyinfluencesthevolumeofcurrentoutput。Theexpectationofafallinthevalueofmoneystimulatesinvestment，andhenceemploymentgenerally，becauseitraisesthescheduleofthe[Page142]THE
GENERALTHEORYOFEMPLOYMENT
marginalefficiencyofcapital，i。e。theinvestmentdemand-schedule；
andtheexpectationofariseinthevalueofmoneyisdepressing，becauseitlowersthescheduleofthemarginalefficiencyofcapital。
ThisisthetruthwhichliesbehindProfessorIrvingFisher'stheoryofwhatheoriginallycalled'AppreciationandInterest'¾；thedistinctionbetweenthemoneyrateofinterestandtherealrateofinterestwherethelatterisequaltotheformeraftercorrectionforchangesinthevalueofmoney。Itisdifficulttomakesenseofthistheoryasstated，becauseitisnotclearwhetherthechangeinthevalueofmoneyisorisnotassumedtobeforeseen。Thereisnoescapefromthedilemmathat，ifitisnotforeseen，therewillbenoeffectoncurrentaffairs；whilst，ifitisforeseen，thepricesofexistinggoodswillbeforthwithsoadjustedthattheadvantagesofholdingmoneyandofholdinggoodsareagainequalised，anditwillbetoolateforholdersofmoneytogainortosufferachangeintherateofinterestwhichwilloffsettheprospectivechangeduringtheperiodoftheloaninthevalueofthemoneylent。ForthedilemmaisnotsuccessfullyescapedbyProfessorPigou'sexpedientofsupposingthattheprospectivechangeinthevalueofmoneyisforeseenbyonesetofpeoplebutnotforeseenbyanother。
Themistakeliesinsupposingthatitistherateofinterestonwhichprospectivechangesinthevalueofmoneywilldirectlyreact，insteadofthemarginalefficiencyofagivenstockofcapital。Thepricesofexistingassetswillalwaysadjustthemselvestochangesinexpectationconcerningtheprospectivevalueofmoney。Thesignificanceofsuchchangesinexpectationliesintheireffectonthereadinesstoproducenewassetsthroughtheirreactiononthemarginalefficiencyofcapital。Thestimulatingeffectoftheexpectationofhigherpricesisdue，nottoitsraisingtherateofinterestthatwouldbeaparadoxicalwayofstimulatingoutput¾；insofarastherateofinterestrises，the[Page143]THE
MARGINALEFFICIENCYOFCAPITAL
stimulatingeffectistothatextentoffset，buttoitsraisingthemarginalefficiencyofagivenstockofcapital。Iftherateofinterestweretoriseparipassuwiththemarginalefficiencyofcapital，therewouldbenostimulatingeffectfromtheexpectationofrisingprices。Forthestimulustooutputdependsonthemarginalefficiencyofagivenstockofcapitalrisingrelativelytotherateofinterest。
IndeedProfessorFisher'stheorycouldbebestre-writtenintermsofa'realrateofinterest'definedasbeingtherateofinterestwhichwouldhavetorule，consequentlyonachangeinthestateofexpectationastothefuturevalueofmoney，inorderthatthischangeshouldhavenoeffectoncurrentoutput。[85]
Itisworthnotingthatanexpectationofafuturefallintherateofinterestwillhavetheeffectofloweringthescheduleofthemarginalefficiencyofcapital；sinceitmeansthattheoutputfromequipmentproducedto-daywillhavetocompeteduringpartofitslifewiththeoutputfromequipmentwhichiscontentwithalowerreturn。Thisexpectationwillhavenogreatdepressingeffect，sincetheexpectations，whichareheldconcerningthecomplexofratesofinterestforvarioustermswhichwillruleinthefuture，willbepartiallyreflectedinthecomplexofratesofinterestwhichruleto-day。Neverthelesstheremaybesomedepressingeffect，sincetheoutputfromequipmentproducedto-day，whichwillemergetowardstheendofthelifeofthisequipment，mayhavetocompetewiththeoutputofmuchyoungerequipmentwhichiscontentwithalowerreturnbecauseofthelowerrateofinterestwhichrulesforperiodssubsequenttotheendofthelifeofequipmentproducedto-day。
Itisimportanttounderstandthedependenceofthemarginalefficiencyofagivenstockofcapitalonchangesinexpectation，becauseitischieflythisdepend-
[Page144]THE
GENERALTHEORYOFEMPLOYMENT
encewhichrendersthemarginalefficiencyofcapitalsubjecttothesomewhatviolentfluctuationswhicharetheexplanationofthetradecycle。
Inchapter22belowweshallshowthatthesuccessionofboomandslumpcanbedescribedandanalysedintermsofthefluctuationsofthemarginalefficiencyofcapitalrelativelytotherateofinterest。
IV
Twotypesofriskaffectthevolumeofinvestmentwhichhavenotcommonlybeendistinguished，butwhichitisimportanttodistinguish。Thefirstistheentrepreneur'sorborrower'sriskandarisesoutofdoubtsinhisownmindastotheprobabilityofhisactuallyearningtheprospectiveyieldforwhichhehopes。Ifamanisventuringhisownmoney，thisistheonlyriskwhichisrelevant。
Butwhereasystemofborrowingandlendingexists，bywhichImeantherantingofloanswithamarginofrealorpersonalsecurity，asecondtypeofriskisrelevantwhichwemaycallthelender'srisk。Thismaybedueeithertomoralhazard，i。e。voluntarydefaultorothermeansofescape，possiblylawful，fromthefulfilmentoftheobligation，ortothepossibleinsufficiencyofthemarginofsecurity，i。e。involuntarydefaultduetothedisappointmentofexpectation。Athirdsourceofriskmightbeadded，namely，apossibleadversechangeinthevalueofthemonetarystandardwhichrendersamoney-loantothisextentlesssecurethanarealasset；thoughallormostofthisshouldbealreadyreflected，andthereforeabsorbed，inthepriceofdurablerealassets。
Nowthefirsttypeofriskis，inasense，arealsocialcost，thoughsusceptibletodiminutionbyaveragingaswellasbyanincreasedaccuracyofforesight。Thesecond，however，isapureadditiontothecostofinvestmentwhichwouldnotexistiftheborrowerandlenderwerethesameperson。
Moreover，itinvolvesinpart[Page145]THE
MARGINALEFFICIENCYOFCAPITAL
aduplicationofaproportionoftheentrepreneur'srisk，whichisaddedtwicetothepurerateofinteresttogivetheminimumprospectiveyieldwhichwillinducetheinvestment。Forifaventureisariskyone，theborrowerwillrequireawidermarginbetweenhisexpectationofyieldandtherateofinterestatwhichhewillthinkitworthhiswhiletoborrow；
whilsttheverysamereasonwillleadthelendertorequireawidermarginbetweenwhathechargesandthepurerateofinterestinordertoinducehimtolendexceptwheretheborrowerissostrongandwealthythatheisinapositiontoofferanexceptionalmarginofsecurity。Thehopeofaveryfavourableoutcome，whichmaybalancetheriskinthemindoftheborrower，isnotavailabletosolacethelender。
Thisduplicationofallowanceforaportionoftheriskhasnothithertobeenemphasised，sofarasIamaware；butitmaybeimportantincertaincircumstances。Duringaboomthepopularestimationofthemagnitudeofboththeserisks，bothborrower'sriskandlender'srisk，isapttobecomeunusuallyandimprudentlylow。
V
Thescheduleofthemarginalefficiencyofcapitalisoffundamentalimportancebecauseitismainlythroughthisfactormuchmorethanthroughtherateofinterestthattheexpectationofthefutureinfluencesthepresent。Themistakeofregardingthemarginalefficiencyofcapitalprimarilyintermsofthecurrentyieldofcapitalequipment，whichwouldbecorrectonlyinthestaticstatewherethereisnochangingfuturetoinfluencethepresent，hashadtheresultofbreakingthetheoreticallinkbetweento-dayandto-morrow。Eventherateofinterestis，virtually，[86]a[Page146]THE
GENERALTHEORYOFEMPLOYMENT
currentphenomenon；andifwereducethemarginalefficiencyofcapitaltothesamestatus，wecutourselvesofffromtakinganydirectaccountoftheinfluenceofthefutureinouranalysisoftheexistingequilibrium。
Thefactthattheassumptionsofthestaticstateoftenunderliepresent-dayeconomictheory，importsintoitalargeelementofunreality。Buttheintroductionoftheconceptsofusercostandofthemarginalefficiencyofcapital，asdefinedabove，willhavetheeffect，Ithink，ofbringingitbacktoreality，whilstreducingtoaminimumthenecessarydegreeofadaptation。
Itisbyreasonoftheexistenceofdurableequipmentthattheeconomicfutureislinkedtothepresent。Itis，therefore，consonantwith，andagreeableto，ourbroadprinciplesofthought，thattheexpectationofthefutureshouldaffectthepresentthroughthedemandpricefordurableequipment。
Long-TermExpectation[Page147]
Chapter12
THESTATEOFLONG-TERMEXPECTATION
I
Wehaveseeninthepreviouschapterthatthescaleofinvestmentdependsontherelationbetweentherateofinterestandthescheduleofthemarginalefficiencyofcapitalcorrespondingtodifferentscalesofcurrentinvestment，whilstthemarginalefficiencyofcapitaldependsontherelationbetweenthesupplypriceofacapital-assetanditsprospectiveyield。Inthischapterweshallconsiderinmoredetailsomeofthefactorswhichdeterminetheprospectiveyieldofanasset。
Theconsiderationsuponwhichexpectationsofprospectiveyieldsarebasedarepartlyexistingfactswhichwecanassumetobeknownmoreorlessforcertain，andpartlyfutureeventswhichcanonlybeforecastedwithmoreorlessconfidence。Amongstthefirstmaybementionedtheexistingstockofvarioustypesofcapital-assetsandofcapital-assetsingeneralandthestrengthoftheexistingconsumers'demandforgoodswhichrequirefortheirefficientproductionarelativelylargerassistancefromcapital。
Amongstthelatterarefuturechangesinthetypeandquantityofthestockofcapital-assetsandinthetastesoftheconsumer，thestrengthofeffectivedemandfromtimetotimeduringthelifeoftheinvestmentunderconsideration，andthechangesinthewage-unitintermsofmoneywhichmayoccurduringitslife。Wemaysumupthestateofpsychologicalexpectationwhichcoversthe[Page148]THE
GENERALTHEORYOFEMPLOYMENT
latterasbeingthestateoflong-termexpectation；¾；asdistinguishedfromtheshort-termexpectationuponthebasisofwhichaproducerestimateswhathewillgetforaproductwhenitisfinishedifhedecidestobeginproducingitto-daywiththeexistingplant，whichweexaminedinchapter5。
II
Itwouldbefoolish，informingourexpectations，toattachgreatweighttomatterswhichareveryuncertain。[87]Itisreasonable，therefore，tobeguidedtoaconsiderabledegreebythefactsaboutwhichwefeelsomewhatconfident，eventhoughtheymaybelessdecisivelyrelevanttotheissuethanotherfactsaboutwhichourknowledgeisvagueandscanty。
Forthisreasonthefactsoftheexistingsituationenter，inasensedisproportionately，intotheformationofourlong-termexpectations；ourusualpracticebeingtotaketheexistingsituationandtoprojectitintothefuture，modifiedonlytotheextentthatwehavemoreorlessdefinitereasonsforexpectingachange。
Thestateoflong-termexpectation，uponwhichourdecisionsarebased，doesnotsolelydepend，therefore，onthemostprobableforecastwecanmake。Italsodependsontheconfidencewithwhichwemakethisforecast¾；onhowhighlyweratethelikelihoodofourbestforecastturningoutquitewrong。Ifweexpectlargechangesbutareveryuncertainastowhatpreciseformthesechangeswilltake，thenourconfidencewillbeweak。
Thestateofconfidence，astheytermit，isamattertowhichpracticalmenalwayspaytheclosestandmostanxiousattention。Buteconomistshavenotanalyseditcarefullyandhavebeencontent，asarule，todiscuss[Page149]LONG-TERM
EXPECTATION
itingeneralterms。Inparticularithasnotbeenmadeclearthatitsrelevancetoeconomicproblemscomesinthroughitsimportantinfluenceonthescheduleofthemarginalefficiencyofcapital。Therearenottwoseparatefactorsaffectingtherateofinvestment，namely，thescheduleofthemarginalefficiencyofcapitalandthestateofconfidence。Thestateofconfidenceisrelevantbecauseitisoneofthemajorfactorsdeterminingtheformer，whichisthesamethingastheinvestmentdemand-schedule。
Thereis，however，notmuchtobesaidaboutthestateofconfidenceapriori。Ourconclusionsmustmainlydependupontheactualobservationofmarketsandbusinesspsychology。Thisisthereasonwhytheensuingdigressionisonadifferentlevelofabstractionfrommostofthisbook。
Forconvenienceofexpositionweshallassumeinthefollowingdiscussionofthestateofconfidencethattherearenochangesintherateofinterest；
andweshallwrite，throughoutthefollowingsections，asifchangesinthevaluesofinvestmentsweresolelyduetochangesintheexpectationoftheirprospectiveyieldsandnotatalltochangesintherateofinterestatwhichtheseprospectiveyieldsarecapitalised。Theeffectofchangesintherateofinterestis，however，easilysuperimposedontheeffectofchangesinthestateofconfidence。
III
Theoutstandingfactistheextremeprecariousnessofthebasisofknowledgeonwhichourestimatesofprospectiveyieldhavetobemade。Ourknowledgeofthefactorswhichwillgoverntheyieldofaninvestmentsomeyearshenceisusuallyveryslightandoftennegligible。Ifwespeakfrankly，wehavetoadmitthatourbasisofknowledgeforestimatingtheyieldtenyearshenceofarailway，acoppermine，atextilefactory，thegoodwillofapatentmedicine，anAtlantic[Page150]THE
GENERALTHEORYOFEMPLOYMENT
liner，abuildingintheCityofLondonamountstolittleandsometimestonothing；orevenfiveyearshence。Infact，thosewhoseriouslyattempttomakeanysuchestimateareoftensomuchintheminoritythattheirbehaviourdoesnotgovernthemarket。
Informertimes，whenenterprisesweremainlyownedbythosewhoundertookthemorbytheirfriendsandassociates，investmentdependedonasufficientsupplyofindividualsofsanguinetemperamentandconstructiveimpulseswhoembarkedonbusinessasawayoflife，notreallyrelyingonaprecisecalculationofprospectiveprofit。Theaffairwaspartlyalottery，thoughwiththeultimateresultlargelygovernedbywhethertheabilitiesandcharacterofthemanagerswereaboveorbelowtheaverage。Somewouldfailandsomewouldsucceed。Butevenaftertheeventnoonewouldknowwhethertheaverageresultsintermsofthesumsinvestedhadexceeded，equalledorfallenshortoftheprevailingrateofinterest；though，ifweexcludetheexploitationofnaturalresourcesandmonopolies，itisprobablethattheactualaverageresultsofinvestments，evenduringperiodsofprogressandprosperity，havedisappointedthehopeswhichpromptedthem。Businessmenplayamixedgameofskillandchance，theaverageresultsofwhichtotheplayersarenotknownbythosewhotakeahand。Ifhumannaturefeltnotemptationtotakeachance，nosatisfactionprofitapartinconstructingafactory，arailway，amineorafarm，theremightnotbemuchinvestmentmerelyasaresultofcoldcalculation。
Decisionstoinvestinprivatebusinessoftheold-fashionedtypewere，however，decisionslargelyirrevocable，notonlyforthecommunityasawhole，butalsofortheindividual。Withtheseparationbetweenownershipandmanagementwhichprevailsto-dayandwiththedevelopmentoforganisedinvestmentmarkets，anewfactorofgreatimportancehasenteredin，whichsometimesfacilitatesinvestmentbutsometimesadds[Page151]LONG-TERM
EXPECTATION
greatlytotheinstabilityofthesystem。Intheabsenceofsecuritymarkets，thereisnoobjectinfrequentlyattemptingtorevalueaninvestmenttowhichwearecommitted。ButtheStockExchangerevaluesmanyinvestmentseverydayandtherevaluationsgiveafrequentopportunitytotheindividualthoughnottothecommunityasawholetorevisehiscommitments。Itisasthoughafarmer，havingtappedhisbarometerafterbreakfast，coulddecidetoremovehiscapitalfromthefarmingbusinessbetween10andII
inthemorningandreconsiderwhetherheshouldreturntoitlaterintheweek。ButthedailyrevaluationsoftheStockExchange，thoughtheyareprimarilymadetofacilitatetransfersofoldinvestmentsbetweenoneindividualandanother，inevitablyexertadecisiveinfluenceontherateofcurrentinvestment。Forthereisnosenseinbuildingupanewenterpriseatacostgreaterthanthatatwhichasimilarexistingenterprisecanbepurchased；
whilstthereisaninducementtospendonanewprojectwhatmayseemanextravagantsum，ifitcanbefloatedoffontheStockExchangeatanimmediateprofit。[88]ThuscertainclassesofinvestmentaregovernedbytheaverageexpectationofthosewhodealontheStockExchangeasrevealedinthepriceofshares，ratherthanbythegenuineexpectationsoftheprofessionalentrepreneur。[89]Howthenarethesehighlysignificantdaily，evenhourly，revaluationsofexistinginvestmentscarriedoutinpractice？
[Page152]THE
GENERALTHEORYOFEMPLOYMENT
IV
Inpracticewehavetacitlyagreed，asarule，tofallbackonwhatis，intruth，aconvention。Theessenceofthisconvention¾；thoughitdoesnot，ofcourse，workoutquitesosimply¾；liesinassumingthattheexistingstateofaffairswillcontinueindefinitely，exceptinsofaraswehavespecificreasonstoexpectachange。Thisdoesnotmeanthatwereallybelievethattheexistingstateofaffairswillcontinueindefinitely。Weknowfromextensiveexperiencethatthisismostunlikely。Theactualresultsofaninvestmentoveralongtermofyearsveryseldomagreewiththeinitialexpectation。Norcanwerationaliseourbehaviourbyarguingthattoamaninastateofignoranceerrorsineitherdirectionareequallyprobable，sothatthereremainsameanactuarialexpectationbasedonequi-probabilities。Foritcaneasilybeshownthattheassumptionofarithmeticallyequalprobabilitiesbasedonastateofignoranceleadstoabsurdities。Weareassuming，ineffect，thattheexistingmarketvaluation，howeverarrivedat，isuniquelycorrectinrelationtoourexistingknowledgeofthefactswhichwillinfluencetheyieldoftheinvestment，andthatitwillonlychangeinproportiontochangesinthisknowledge；though，philosophicallyspeaking，itcannotbeuniquelycorrect，sinceourexistingknowledgedoesnotprovideasufficientbasisforacalculatedmathematicalexpectation。Inpointoffact，allsortsofconsiderationsenterintothemarketvaluationwhichareinnowayrelevanttotheprospectiveyield。
Neverthelesstheaboveconventionalmethodofcalculationwillbecompatiblewithaconsiderablemeasureofcontinuityandstabilityinouraffairs，solongaswecanrelyonthemaintenanceoftheconvention。
Forifthereexistorganisedinvestmentmarketsandifwecanrelyonthemaintenanceoftheconvention，aninvestorcanlegitimatelyencouragehimselfwiththe[Page153]LONG-TERM