Mr.

Keynes and Traditional Theory

Author(s): R. F. Harrod
Source: Econometrica, Vol. 5, No. 1 (Jan., 1937), pp. 74-86
Published by: The Econometric Society
Stable URL: http://www.jstor.org/stable/1905402 .
Accessed: 01/08/2014 15:12

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp

.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

The Econometric Society is collaborating with JSTOR to digitize, preserve and extend access to Econometrica.

http://www.jstor.org

This content downloaded from 129.210.6.137 on Fri, 1 Aug 2014 15:12:57 PM

All use subject to JSTOR Terms and Conditions
 MR. KEYNES AND TRADITIONAL THEORY
By R. F. HARROD

IN THIS PAPER I do not proposeto ask or answerthe question,has Mr.

Keynes succeeded in establishingthe propositionswhichhe claims to
have established?nor again, what kind of evidence is requiredto es-
tablish or to refutethose propositions?I shall confinemyselfto a nar-
rowerquestion,namely,what are the propositionswhich Mr. Keynes
claims to have established?And in orderto restrictmy subject matter
still further,I propose to confinemyselfto those propositions,which
he claims to have established,that are in conflictwith the theoryof
value in the formin whichit has hithertobeen commonlyaccepted by
most economists.In other words, my question is what modifications
in the generallyrecognizedtheoryof value would acceptance of the
propositionsthat Mr. Keynes claims to have establishedentail?
In orderto clarifythe issues involvedit may be well to divide com-
monly accepted theory into the general theory and its specialised
branches.The generaltheoryconsistsprimarilyof a numberof func-
tional equations expressingindividual preferenceschedules and a
numberof identities,such as that supply must be equal to demand,
and the elucidation of such questions as whetherthere are as many
equations as thereare unknownsand whetherthe solutionsare single
or multiple.The resultof these enquiriesshould make it clear whether
the equilibriumof the systemas a whole is stable or unstable or un-
determined,whetherthereare alternativepositionsof equilibrium,etc.
There may be some clues as to the generalformof some of the func-
tional equations, providedby such principlesas the Law of Diminish-
ing Utility, to use old-fashionedterminology,which may make it
possibleto predictthe directionof changesin the values of the various
unknownsdue to a given change in one of them. More precise pre-
dictioncan onlybe achieved if or whenit becomes possible,as a result
of the labors of such investigatorsas Dr. Schultz, to writedown the
actual terms of the functionalequations. Within the corpus of this
general theorymay be included the formulationof the market con-
ditionsthat are requiredforthe realisationof some kind of maximum.
Thus ifone individualA is indifferent whetherhe producescommodity
X or commodityY fora certainconsideration,and anotherindividual
B prefersX to Y, the maximumis not realisedifthe marketso operates
that A normallyproduces Y and not X forB. On this conditionthe
maxim of Free Trade is fairly securely founded,the more general
maximoflaissez fairemuchless securelyso.
In contrastwith the theoryof value in this very generalform,may
be set the special theoriesformulatedto deal with specificproblems
74

This content downloaded from 129.210.6.137 on Fri, 1 Aug 2014 15:12:57 PM

All use subject to JSTOR Terms and Conditions
 R. F. HARROD 75

such as interest,profit,joint production,discriminatingmonopoly,

etc. The normal method used in dealing with these departmental
studies is to assume that certainterms,which appear as variables in
the general systemof equations, may be treated as constantsforthe
special purpose in hand. For instance in studyingthe behaviour of
duopolisticproducersof a given commodity,it may be assumed that
the duopolistscan obtain the servicesof factorsof productionat rates
the determinationof which in the market will not be appreciably
affectedby the duopolists' behaviour. Such methods constituteshort
cuts to the unravellingof particularproblemsand they are oftenper-
fectlylegitimate.In the minds of most economists,other than those
who stand,so to speak, at the philosophicalend of the economicarray,
the conclusions reached by these short-cutmethods constitutethe
main findingsof economictheory.
I may say at once that in my opinionMr. Keynes' conclusionsneed
not be deemed to make a vast differenceto the general theory,but
that they do make a vast difference to a numberof short-cutconclu-
sions of leading importance.Thus to those whom I may perhaps call
without offencethe ordinaryworkingeconomiststhey ought, if ac-
cepted,to appear to constitutequite a revolution.Whethertheyentail
a substantialmodificationof the moregeneraltheorydepends on how
that is stated. I need hardlyobservethat thereis no authorisedversion.
Those whose main interestis in the generaltheory,may, if they have
laid theirfoundationswell and carefully,be able to look down with a
smile of indifference on the fulminationsof Mr. Keynes. Pavillioned
upon theirOlympianfastness,theyare not likelyto show muchirrita-
tion.
It is convenientto take Mr. Keynes' theoryof interestas the start-
ing pointofthisexposition.In the commonlyaccepted short-cuttheory
thereare two unknownsand two equations. The two unknownsare the
volume of saving (= the volume of investment)and the rate of in-
terest. Of the new-fangledview, sponsored by some out-of-the-way
definitionsin Mr. Keynes' Treatiseon Money,that the volume of sav-
ing may be unequal to the volume of investment,it is not necessary
to say anything,since it has played no part in the standard short-cut
formulationsof interesttheory(althoughit has figuredin recentwrit-
ings concerned with practical monetary problems). The commonly
accepted interesttheoryfromthe time of the early classical writers
onward entails that saving is always and necessarilyequal to invest-
ment.
The two equations in the traditionaltheoryof interestcorrespond
to the demand and supply schedules relating to a particular com-
modity.First thereis the demand equation:

This content downloaded from 129.210.6.137 on Fri, 1 Aug 2014 15:12:57 PM

All use subject to JSTOR Terms and Conditions
76 ECONOMETRICA

y=(x),
y, the marginalproductivityof capital,dependingon x, the amount of
capital investedper unit of time. So much capital will be investedthat
its marginalproductivityis equal to the rate of interest;that is,
Y=y',
wherey' is the rate of interest.Since both the traditionaltheoryand
Mr. Keynes hold that investmentis undertakenup to the point at
which the marginalproductivityof capital is equal to the rate of in-
terest,y' may be suppressed,and y made to stand forthe rate ofinter-
est whichis equal to the marginalproductivityof capital.
Then thereis the supply equation:
x =-(Y);
x, the amount whichindividuals choose to save, whichis equal to the
amount of investment,depends on the rate of interest.Thus there
are two unknowns,the rate of interestand the volume of saving, and
sufficientequations to determinethem. It is not necessary for the
presentpurposeto considercontroversiesconcerningthe formsofthese
equations,such as whethera rise in the rate of interesttends to cause
people to save moreor less.
This treatmentof interestand saving is analogous to that of the
priceof a particulararticleand the amount of it produced.The treat-
ment depends on the short-cutassumptionof ceterisparibus. This is
oftenlegitimatein the case of particularcommodities,althoughit is
recognisedthat in certaincases it is idle not to bringin certainother
variables, for instance the prices of close substitutes. Among the
"otherthings"whichare supposed to be "equal" is the level of income
in the communityunderdiscussion.In many cases it may be truethat
whenwe are tryingto determinehow much of a particularcommodity
a produceris likelyto produce,his decisionto producea littlemoreor
a littleless will not have a sufficiently large effecton the total income
of the communityto react on the marketforhis goods in such a way
as to make an appreciable difference to him. This particularshort-cut
is in that case justified.I suggestthat the most importantsinglepoint
in Mr. Keynes' analysisis the view that it is illegitimateto assume that
the level of incomein the communityis independentof the amount of
investmentdecided upon. No resultsachieved by the short-cutof such
an assumptioncan be of any value.
How does Mr. Keynes' analysis proceed? His firstequation is sub-
stantiallythe same as that of the traditionalanalysis
y =(x).
The marginal productivityof capital is a functionof the amount of
investmentundertaken.The marginalproductivityof capital appears
in Mr. Keynes' book underthe title of marginalefficiency. It does not

This content downloaded from 129.210.6.137 on Fri, 1 Aug 2014 15:12:57 PM

All use subject to JSTOR Terms and Conditions
 R. F. HARROD 77

appear that thereis a difference of principlehere. It is true that Mr.

Keynes makes an exhaustiveand interestinganalysis of this marginal
efficiencyand demonstratesthat its value depends on entrepreneurial
expectations.The stress which he lays on expectationsis sound, and
constitutesa great improvementin the definitionof marginal pro-
ductivity. This improvement,however, might be incorporated in
traditional theory without entailing important modificationsin its
other parts.
When we come to the second equation the level of income must be
introducedas an unknownterm,giving
x =(y, i),
wherei is the level of income.The amount of saving depends not only
on the rate of interest,but on the level of income in the community.
It mightbe thoughtthat to introducethe level of income as an un-
knownat this point is tantamountto abandoning all attempt to have
a departmentaltheoryof the volume of saving,since the level of total
income appears in all the equations of the general theory and it is
impossibleto determineits value withouttaking all factorsinto ac-
count. This would mean that we should have to leave the ordinary
workingeconomist without any departmentaltheory of saving and
interestwhich he could grasp, and to let him flounderin the maze of
nXrXs, etc., equations governingthe whole system.Mr. Keynes has,
however,come to the rescueand carved out a new short-cutofhis own.
In his view thevalue oftheunknownlevel ofincomecan be determined
in a legitimateand satisfactorymannerby the departmentalequations
relatingto saving and interestonly. To the legitimacyof this assump-
tion it will be necessaryto returnpresently.
Meanwhile,since there are threeunknownsand but two equations
in the savings/interestcomplex, another equation is needed. Before
proceedingto that, it may be well to recurto the second equation,
x-=(y, i).
This may be transposedinto the form
i=(,y).
The level of income depends on the amount of investment(=that of
saving) and the rate ofinterest.In thisformthe secondequation shows
itselfas the doctrineof the multiplier.The multiplieris the reciprocal
of the fractionexpressingthe proportionof any given income,which,
at a given rate of interest,people consume. If the value of the multi-
plier is knownfor any given rate of interestand level of income,the
actual level of income can be deduced directlyfromthe volume of
investment.Those to whom the doctrineof multiplierseems an alien
morselin the corpus of economicdoctrineshould rememberthat it is

This content downloaded from 129.210.6.137 on Fri, 1 Aug 2014 15:12:57 PM

All use subject to JSTOR Terms and Conditions
78 ECONOMETRICA

merelya disguisedformof the ordinarysupplyscheduleoffreecapital,

but withthe level of incometreatedas a variable.
In discussingthis doctrine,for the sake of a still shortercut, Mr.
Keynes is inclinedto let the rate of interestdrop out of sight.Thus the
equation becomes
i=gx);
the level of incomedepends on the volume of investment.The justifi-
cation for this procedureis that whereas the relation of the level of
incometo the amount of investmentis in the broadest sense known-
it may be assumed that people save a largerabsolute amount froma
larger income-the relation of the amount which people choose to
save to the rate of interestis a matterof controversy.Moreover in
Mr. Keynes' view the level of income has a more importanteffecton
the amount which people choose to save than the rate of interest.
However, thereis no need to pick a quarrelhere. The rate of interest
may be broughtback into this part of the picture without affecting
the main argument.The propensityto consume may be regardedas
dependingon the rate of interest,althoughforthe sake of brevityand
claritymentionof this need not be insisted on at every point in an
expositionof the doctrineof the multiplier.
What of the thirdequation? We have
y=x(m),
wherem is the quantityof money,a knownterm,dependingon bank-
ing policy. This is the liquidity preferenceschedule. Probably i, the
level of income,ought to be insertedin this equation, thus:
y = x(m, i),
since the amount of money required for active circulationby con-
sumersand tradersdepends on the level ofincome.Oughtnot the price
level to come in also? That may be taken to be subsumedunder i, the
level of income,in a mannerthat I shall presentlyexplain. The residue
of money,not requiredforactive circulation,is available forordinary
people who are discouragedby theirbrokersfromimmediateinvest-
ment,and, more important,forfirms,who may want cash forcapital
extensionsor similarpurposeswithinsix monthsor a year or two, and
are unwillingto hold theirreservesin the formof securitiesto which
some risk of depreciationwithin the prescribedperiod is attached.
Since the amount of money available for liquid reserves is strictly
limitedand cannotbe increasedby the meredesireon the part offirms
to hold more money than that, the prospectiveyield of less liquid re-
servesmustbe sufficient to confinethose who insiston a moneyreserve
to the amount of money available for that purpose. The less the
amount of moneyavailable the higherthe rate of interestwill have to
be, both because the high rate is a quid pro quo against the risk of

This content downloaded from 129.210.6.137 on Fri, 1 Aug 2014 15:12:57 PM

All use subject to JSTOR Terms and Conditions
 R. F. HARROD 79

depreciationofthe capital and also because the higherthe presentrate

the less probability is there of depreciation within the prescribed
period.
It is not necessaryto give a finalpronouncementon the significance
ofthe liquiditypreferenceequation. It appears that even ifsome modi-
ficationis required in this third equation, which determinesthe rate
of interest,a type of analysis similar in its general structureto that
of Mr. Keynes may be maintained.
We now have three equations to determinethe value of the three
unknowns,level of income, volume of saving (=volume of invest-
ment), and rate of interest(= marginalproductivityof capital).
For the workingeconomist these results may be set out in still
briefershorthand as follows.The amount of investment(=amount
of saving) depends on the marginal productivityof capital and the
rate of interest;the level of income is connectedwith the amount of
investmentby the multiplier,i.e., by the propensityto consume; and
the rate of interestdepends on the desire forliquid reservpsand the
amountofspare cash in the communityavailable to satisfythat desire.
The amount of this spare cash depends on the policy of the banks in
determiningthe quantity of their I.O.U.'s that are outstandingand
on the level of income (the higherthis, the more moneywill be taken
away into active circulation).
Thus ifthe schedulesexpressingthe marginalproductivityof capital,
the propensityto consume, and the liquidity preferenceare known
and the total quantity of money in the system is known also, the
amountof investment,the level of incomeand the rate ofinterestmay
readily be determined.
The next topic forconsiderationis the legitimacyof the assumption
that the level of income may be regardedas determinedby the com-
plex of considerationsexpressed in the savings/interestequations,
ratherthan by the whole systemof equations. In generalthe level of
activity is traditionallyconceived as depending on the preference
schedulesof the variousfactorsexpressingtheirwillingnessto do vari-
ous amounts of work in returnfor income, and on the schedules ex-
pressingthe relationbetweenthe amountof workdone and the income
accruingfromit (Laws ofReturns).In consideringtheformerschedules
we have to take into account all the factorsof production.Now in Mr.
Keynes' systemthe supply of capital has already been dealt with by
the savings/interest equations. For the supplyof risk-bearing,we may
provisionallycontentourselveswith the elegant device whichhe pro-
vides in his footnoteto p. 24. He writes,"by his (the entrepreneur's)
expectation of proceeds I mean, therefore,that expectation of pro-
ceeds, which,if it were held with certainty,would lead to the same

This content downloaded from 129.210.6.137 on Fri, 1 Aug 2014 15:12:57 PM

All use subject to JSTOR Terms and Conditions
80 ECONOMETRICA

behaviour as does the bundle of vague and more various possibilities

which actually makes up his state of expectationwhen he reaches his
decision." Thus considerationsaffectingthe supply of risk-bearingare
subsumedin the equations whichdeterminethe volume of investment.
There remainthe factorsotherthan those covered by the category
of investment.Of these we are only concernedwith those the supply
of whichcan be varied. Thus we are leftwiththose whichmay roughly
be designated prime factors. What is the nature of their supply-
schedule? What is the formof theirpreferenceforincome in relation
to the work required to obtain it?
In this fieldMr. Keynes' argumentis vitally dependenton his ob-
servation of real conditions. The work/incomepreferenceschedule
exertsits powerupon the economicsystemthroughthe termson which
the prime factorsare willingto sell their services. The contractsor
bargainsof the entrepreneurs withprimefactorsare normallyfixedin
money,with no proviso regardingthe general level of prices. In the
exceptional cases in which there is such a proviso,it is none the less
usually the case that a rise in pricesinvolves some fall in real rewards
to primefactorsand conversely.It is truethat in a timeof risingprices
the factorsmay press for a rise in rewards,but, even if they achieve
this,thereis still no provisoto safeguardthemagainst a furtherrise of
prices,and prices may, forall the new bargainslay down, and indeed
are veryoftenin factobservedto, runon ahead of rewards.Conversely
in a timeof fallingprices.This gives the supply schedulesof the prime
factorsa very special kind of indeterminacywhich underminestheir
powerto determinethe generallevel of activity.Mr. Keynes discusses
this matterin Book I and its importancein his logical edificejusti-
fieshim in givingit pride of place.
Consider next the second set of schedules determiningthe general
level of activity,namely those expressingthe relations between the
amount of work done and the income accruingfromit (Laws of Re-
turns). Since the bargains with primefactorsare expressedin money,
the returnsdue to theiremploymentshouldbe expressedin moneyalso.
But the moneyvalue of these returnsdepends on the level of prices.
The generalpricelevel mightbe regardedas determinedby the Quan-
tity Theory of Money; Mr. Keynes does not so regard it for reasons
whichwill be explained below. On the contraryhe regardsthe general
price level as completelymalleable and determinedby the equations
in the generalfieldwithoutreferenceto the quantityof money.
The consequence of the conclusionsyielded by the interest/savings
equations,iftheseare accepted,is, that thelevel ofincomeand activity
is determined.Now suppose the entrepreneurs decide to producemore
than the amount so determined.Owing to a deficientpropensityto

This content downloaded from 129.210.6.137 on Fri, 1 Aug 2014 15:12:57 PM

All use subject to JSTOR Terms and Conditions
 R. F. HARROD 81

consume, they will find deficientpurchasingpower, and either ac-

cumulatestocksor sell at a loss. If theydo the formerthe accumulation
of stocks will constitutean additional (involuntary)investmenton the
part of the community,whichwhenadded to the intendedinvestment,
makes the total investmentof the communitysuch as to be consistent,
in accordance with the interest/savingsequations, with the higher
level of activitywhich entrepreneursare choosingto indulge in. But
such a positionis unstable. So long as stocks are accumulating,they
will reduce activity and continueto do so, until it reaches the point
indicatedby the interest/savings equations. If on the otherhand they
sell at a loss, they will be dis-saving;the propensityto consume will
be temporarilyraised, so that the higherlevel of activity which they
are choosingto indulge in becomes consistentwith that required by
the interest/savings equations. But again the positionis unstable. The
marginalpropensityto consumewill not be permanentlysustained at
an abnormallyhigh figureby these means. To avoid losses, entre-
preneurswill restrictand continueto do so, until activityand income
are reduced to a level which satisfiesthe interest/savingsequations,
with the marginal propensityto consume normal for that level of
income. Converse argumentswould apply in the case of entrepreneurs
decidingto producetoo little.
Now if the level of activityso determinedis indeed the equilibrium
level of activity,the pricelevel must be appropriateto it. Let us sup-
pose that the price of each commodityis determinedby the marginal
moneycost of production,in the crude way that a tiro mightdescribe
erroneouslysupposinghimselfto be explainingthe true classical theory
of cost of production.If the law of diminishingreturnsprevailed on
balance, as Mr. Keynes supposes that it does anyhow in the short
period,the generalpricelevel would be expectedto rise withincreases
of output and to fall with decreases. To make the matterstill more
crude and common,suppose pricesto vary not merelyin proportionto
changes in the numberof units of factorsrequiredper unit of output,
as output varies, but also in proportionto changes in rates of reward
to the factors.In this case we should find,as output rose and diminish-
ing returnscame into play,that the riseof priceswould just sufficiently
exceed the rise of wages, etc., if any, to cover the increasedreal mar-
ginal cost of productionper unit. Factors might press for a rise of
rewards,but thoughthey mightgain on balance in sometrades,they
would always be beaten by the pricelevel in the systemas a whole.
Now thisis preciselywhat Mr. Keynes supposes actually to happen.
It is, however,"subject to the qualificationthat the equality (between
marginalcost and price) may be disturbed,in accordance with certain
principles,if competitionand marketsare imperfect"(p. 5). The ob-

This content downloaded from 129.210.6.137 on Fri, 1 Aug 2014 15:12:57 PM

All use subject to JSTOR Terms and Conditions
82 ECONOMETRICA

jectionsto thisviewwhichupholders oftheQuantityTheoryofMoney

mightraisemustbe considered. But firstobserveits relationto the
determination ofthelevelofactivity.
Take a periodwithinwhichprimefactorbargainsdo not change.
The supplyofeach ofthesein moneytermsmaythenbe represented
by a horizontal straightline.But ifpricesvaryin proportion to costs
(cost variationsincludingallowanceforovertimerates,the employ-
mentoflessefficient labour,etc.),thenthemoneyvalueofthemarginal
net productofeach factormustbe represented by a co-incident hori-
zontal straightline. Thereforeon these conditionsthe two sets of
schedulesleavethelevelofoutputentirely indeterminate. Ifthematter
is expressedin realtermsbothsetsofschedulesare downward sloping
to theright;theyare stillco-incident. If moneyrewardsto factorsare
raisedor loweredin responseto changesin the level of employment
and pricesare adjustedaccordingly, thesameresultensues.Thus this
complexofequationsdoesnotdetermine thelevelofactivity;therefore
it leavesthatlevelfreeto be determined by thesavings/interest com-
plex.Q.E.D.
Thus thecruxofthematterseemsto have shiftedto the Quantity
TheoryofMoney.The essenceofthedifference betweenthetraditional
theoryand Mr. Keynes'theorycan be put thus: In the traditional
theorythesupplyand demandschedulesofall thefactorsstandon the
samefooting; thelevelofactivityis an unknown, butthepricelevelis
determined bythemonetary equation.Thisdetermination oftheprice
level enablesthe level of activityto be determined by the factors'
moneysupplyschedules, and bytheirmarginal productivity schedules.
In Mr. Keynes'theorythelevelofactivityis determined bytheequa-
tionsgoverning the savings/interestcomplex.In the generalfield,in
whichwe are now only concernedwiththe demandand supplyof
primefactors, thelevelofactivityis conceivedas determined ab extra.
It is a knownquantity.But thepricelevelis conceived tobe completely
malleable.If it werenotthesystemin thegeneralfieldwouldbe over-
determined. Thusthemonetary equationis shornofitsformer powers.
The levelofactivitybeinga knownquantitythe pricelevelis deter-
minedby the moneycost of production, withsuitablemodifications
forimperfect competition.
What righthas Mr. Keynesto gutthe monetaryequationin this
way?Has, then,the bankingpolicyno powerto influence the situa-
tion?Yes, certainly it has. The factis thatthe powerresidingin the
monetary equationhas alreadybeenused up in Mr. Keynes'systemi
in theliquiditypreference equationand it cannottherefore exertany
directinfluence in the generalfield.To make it do so would be to use
itsdetermining influence twiceover.In factin Mr. Keynes'systemall
the old piecesreappear,but theyappearin different places.

This content downloaded from 129.210.6.137 on Fri, 1 Aug 2014 15:12:57 PM

All use subject to JSTOR Terms and Conditions
 R. F. HARROD 83

Explanationis necessary.It willbe remembered thataccordingto

the liquiditypreference equation,the rate of interestis determined
by thedesireofpeopleforliquidreservesand the quantityofmoney
availableforthatpurpose.The quantityof moneyavailableforthat
purposeis equal to thetotalquantityofmoneyin existenceless that
requiredforactive trade.' Now if the quantityrequiredforactive
tradewereperfectly indeterminate, as it mustbe by the Quantity
Theory-foraccording to thatthepriceleveldependson thequantity
ofmoneyavailableforactivetrade,and therefore it is unknownwhat
quantityofmoneyanygivenamountofactivetradewillabsorb-the
residuewouldbe indeterminate also. But ifthem in
y= x(m)
is indeterminate, therearetoo manyunknowns in theinterest/savings
set of equations.Thus it is necessaryto the validityof Mr. Keynes'
solutionof the problemof investment and interestthat the amount
ofmoneyavailableforliquidreservesshouldbe determinate, and that
involvesthatthepricelevelshouldbe determined otherwise thanby
themonetary equation.Andso,in Mr.Keynes'systemitis.
The mattermaybe putthus:The savings/interest equationssuffice
to determinethe level of activity,subject to the provisothat the
quantityofmoneywhichappearsin theliquiditypreference equation
is a knownquantity;and thiswill be knownif the pricelevel and
therefore theamountabsorbedinactivetradeis known.The equations
in thegeneralfieldsuffice to determinethepricelevel,subjectto the
provisothat the level of activityis known.Thus thereis afterall
mutualdependency. The levelof activitywillbe suchthatso much
moneyis absorbedin activetradethatthe amountleftoverenables
interest to standat a rateconsistentwiththatlevelofactivity.
The mutualinterdependency of thewholesystemremains,but the
short-cuts indispensable to thinkingaboutparticularproblems, as Mr.
Keyneshas carvedthemout,remainalso.
The amountofinvestment dependson themarginalproductivity of
capitaland the rate of interest.The level of incomeand activityis
relatedto the amountofinvestment by themultiplier,thatis by the
marginalpropensity to consume,thepricelevelis relatedto thelevel
ofactivityby themarginal' moneycostofproduction (whichdepends
I In his liquiditypreferenceequation Mr. Keynes includes the demand for
moneyforwhateverpurpose,and the quantityof moneythat appears in it is
the total quantityof moneyin the community.It has appeared simplerin this
part of the expositionto divide this total into two parts, the amount required
foractive circulationand the residue,to definethe quantityof moneywhich
appears in the liquiditypreferenceequation as that residue,and the demand
whichthe equation expressesas the demand forpurposesotherthan those of
active circulation.This re-definition of termsis merelyan expositorydevice
and does not implyany departurefromMr. Keynes' essentialdoctrines.

This content downloaded from 129.210.6.137 on Fri, 1 Aug 2014 15:12:57 PM

All use subject to JSTOR Terms and Conditions
84 ECONOMETRICA

on theamountofactivityundertaken), theamountofmoneyabsorbed
in activetradedependson thevolumeoftradeand thepricelevel,the
amountof moneyavailableforliquid reservesis equal to the total
amountof moneyin the systemless that requiredforactivetrade,
and therateofinterest dependson theamountofmoneyavailablefor
liquidreserves and theliquiditypreference schedule.
It maybe'wellto do someexercises.Supposethe banksto increase
thetotalamountofmoneyavailableby openmarketoperations. The
increment may eventuallybe dividedbetweenactivecirculation and
liquidreserves.An increaseofmoneyavailableforliquidreserveswill
tendto reducetherateofinterest; and so to increaseinvestment. This
will increase the level of income through the multiplier in accordance
withthemarginalpropensity to consume.If thefallin therateofin-
terest increases the marginal propensity to consume,the increaseof
incomewillbe pro tantogreater,but it is not certainthatit does so.
The increaseofincomeinvolvesan increaseofturnover, and ofprices
in accordance with the law of diminishing returns. This involvesan
increased use of money in active circulation. Thus the fall in therate
ofinterest will not be so great as it would be if all the new money went
into liquid reserves. The money will be divided between the two uses,
no
butthereis reason whatever to suppose that the increments in each
use willbe in proportion to the amounts of money previously employed
there,as is assumedin a QuantityTheoryusinga compendious index
number. The comparative size of the increments will depend on the
current elasticity of the liquidity preference schedule and the current
elasticityof the marginalproductivity of capitalschedule(whichin-
volvesexpectations).
Supposea fallin rewardsto primefactors.The pricelevelwilldrop.
Money will be releasedfromactive circulationforliquid reserves.
Thiswilltendtomaketherateofinterest fallandto reactonthelevelof
investment and activity accordingly. Thus the stimulusto activity
is veryindirectand its effectiveness depends on the same factorsas
that'provided by an increase in the quantityof money.This is very
different fromthe view that a reduction of rewardswill stimulate
activitybecausecosts fall while prices are sustained by the quantity
ofmoneyremaining thesame.
It appearsto me thatthe achievement of Mr. Keyneshas beento
considercertainfeaturesof traditional theory whichwereunsatis-
factory, becausetheproblems involved tended to be slurredover,and
to reconstruct thattheoryin a way which resolves theproblems. The
principalfeaturesso considered are (1) the assumption that the level
ofincomecouldbe takenas fixedinthedepartmental theory ofinterest
andsaving,(2) thepeculiarnatureofthesupply schedules of theprime

This content downloaded from 129.210.6.137 on Fri, 1 Aug 2014 15:12:57 PM

All use subject to JSTOR Terms and Conditions
 RB. F. HARROD 85

factorswhicharisesout of theirbargainsbeingfixedin moneywithout

provisoas to the pricelevel, and (3) the failureof monetarytheoryto
explainhow the total stockof moneyis dividedbetweenliquid reserves
and active circulation,or, in otherwords,the unsatisfactorycharacter
of the theoryof velocityof circulation.
I stated above that the old pieces in the traditionaltheoryreappear,
but sometimesin new places. It might at firstbe thoughtthat the
liquiditypreferenceschedule is a new piece, and that thereforeeither
the new systemis over-determined or the traditionalwritersmusthave
been wrongin supposingthat theirsystemwas determined.But it is
not really a new piece. The old theory pre-supposed that income
velocity of circulationwas somehow determined.But preciselyhow
was somethingof a mystery.Thus the old theoryassumed that there
was a piece therebut did not state exactly what it was. Mr. Keynes'
innovationmay thusbe regardedas a precisedefinition ofthe old piece.
By placing it where he does, he overcomesa difficulty, which has
been assuming an alarmingprominencein recenteconomic work. In
monetaryliteraturethe rate of interesthas been treated,and increas-
inglyso, as an influenceof vital importancein the monetarysituation.
But in traditionaltheory,neitherin the general systemof equations
nor in the departmentaltheoryof interestdoes it appear that the rate
of interestis more intimatelyconnected with the numerairethan the
priceof any otherfactorof production.This is a strikingdiscrepancy.
Mr. Keynes introducesthe liquidity preferenceschedule at a point
which makes it a vital link between the general systemof equations
and monetarytheory.His treatmentis in harmonywithrecentlitera-
ture in that he justifiesthe special connexionof the price of this par-
ticularfactorwith monetaryproblems.It is an immenseadvance on
recent literaturebecause it removes the discrepancy between the
treatmentof interestin the two branchesof study.
In my judgementMr. Keynes has not affecteda revolutionin funda-
mental economictheorybut a re-adjustmentand a shiftof emphasis.
Yet to affecta re-adjustmentin a system,whichin its broad outlines,
despite differences of terminology, has receivedthe approval of many
powerful minds, Marshall, Edgeworth, and Pigou, the Austrian
School, the School of Lausanne, Wicksell, Pantaleoni, Taussig, and
Clark, to mention but a few, is itself a notable and distinguished
achievement.And in the sphereof departmentaleconomicsand short-
cuts,whichare ofgreatestconcernforthe ordinaryworkingeconomist,
Mr. Keynes' views constitutea genuinerevolutionin many fields.
The foregoingaccount has attemptedto expound,not to appraise.
The only criticismof Mr. Keynes which I ventureto offeris that his
systemis still static. Note has been taken of the fact that at certain

This content downloaded from 129.210.6.137 on Fri, 1 Aug 2014 15:12:57 PM

All use subject to JSTOR Terms and Conditions
86 ECONOMETRICA

important points,e.g., in his definitionof the marginalefficiency of

capital,Mr. Keyneslays greatstresson the importance of anticipa-
tionsin determining thepresentequilibrium.
But reference to anticipation is not enoughto makea theorydy-
namic.For it is stilla staticequilibrium whichtheanticipations along
withothercircumstances serveto determine; we are stillseekingto
ascertainwhat amountsof the variouscommodities and factorsof
production willbe exchangedor used and whatpriceswillobtain,so
longas the conditions, includinganticipations, remainthesame.But
in the dynamictheory,as I envisageit,one of the determinands will
be therateofgrowth oftheseamounts.Ourquestionwillthenbe,what
rateofgrowthcan continue to obtain,so longas thevarioussurround-
ingcircumstances, including thepropensity to save,remainthesame?
Savingessentially entailsgrowth, at leastin someofthemagnitudes
underconsideration. No theoryregarding the equilibrium amountof
savingcan be valid,whichassumesthat withinthe periodin which
equilibrium is established,otherthings, suchas thelevelofincome,do
notgrowbut remainconstant.
I envisagein the futuretwo departments of economicprinciples.
The first,thestatictheory, willbe elaboratedon theassumption that
thereis no growthand no saving.The assumption thatpeoplespend
thewholeoftheirincomewillbe rigidlymaintained. On thisbasisit
willbe possibleto evaluatetheequilibrium setofpricesand quantities
of the various commodities and factors,excludingsaving. In the
seconddepartment, dynamictheory,growthand savingwillbe taken
intoaccount.Equilibriumtheorywillbe concerned not merelywith
whatsize,but also withwhatrateofgrowthof certainmagnitudes is
consistent withthe surrounding circumstances. Thereappearsto be
no reasonwhythe dynamicprinciples shouldnot cometo be as pre-
ciselydefined and as rigidlydemonstrable as thestaticprinciples.The
distinguishing featureofthedynamictheorywillnotbe thatit takes
anticipations intoaccount,forthosemayaffect thestaticequilibrium
also,but thatit willembodynewtermsin itsfundamental equations,
rate of growth,acceleration, de-celeration,etc. If development pro-
ceedson theselinestherewillbe a closeparallelbetweenthe statics
and dynamicsof economicsand mechanics.
But to developthisthemefurther wouldtake me too farfrommy
subject.
R. F. HARROD
ChristChurch
Oxford, England

This content downloaded from 129.210.6.137 on Fri, 1 Aug 2014 15:12:57 PM

All use subject to JSTOR Terms and Conditions