Econometrica, Vol. 14, No.

2, 1946
CAPITAL EXPANSION, RATE OF GROWTH,
AND EMPLOYMENT'
By EVSEY D. DOMAR
I. INTRODUCTION
Thispaperdealswitha problemthatis bothold and new-the rela-
tionbetweencapitalaccumulation andemployment. In economiclitera-
tureithasbeendiscusseda numberoftimes,themostnotablecontribu-
tionbelonging to Marx.Morerecently, it wasbrought forthbyKeynes
and hisfollowers.
A thorough analysisofeconomicaspectsof capitalaccumulation is
a tremendous job. The onlywayin whichtheproblemcanbe examined
at all in a shortpaperlikethisis by isolatingit fromthegeneraleco-
nomicstructure and introducing a numberofsimplifying assumptions.
Some of themare not entirelynecessaryand, as the argumentpro-
gresses,thereaderwillsee howtheycan be modified or removed.
The following assumptions and definitions shouldbe notedat the
outset:(a) thereis a constantgeneralpricelevel; (b) no lags are pres-
ent; (c) savingsand investment referto theincomeofthesameperiod;
(d) bothare net,i.e., overand above depreciation; (e) depreciation
is
measurednotin respectto historical costs,but to the costofreplace-
mentof the depreciatedasset by anotherone of thesameproductive
capacity;2 (f) productive capacityofan assetor ofthewholeeconomy
is a measurableconcept.
The last assumption, on which(e) also depends,is notentirelysafe.
Whether a certainpieceofcapitalequipmentor thewholeeconomyis
considered, theirproductive capacitiesdependnotonlyonphysicaland
technicalfactors,but on thewholeinterplay ofeconomicand institu-
tionalforces,such as distribution of income,consumers' preferences,
1 This is a summaryof a paper presentedbeforea joint sessionof the Econo-
metricSocietyand the AmericanStatisticalAssociationin Clevelandon January
24, 1946. It containsthe logical essence of the argumentwith relativelylittle
economicdetail. I hope to developthe latterin a separatepaper to be published
in one of the othereconomicjournals.
Many thanksforhelp and criticismgo to my fellowmembersof the "Little
Seminar": Paul Baran, Svend Laursen,Lloyd A. Metzler,RichardA. Musgrave,
Mary S. Painter,Melvin W. Reder,Tibor de Scitovszky,AlfredSherrard,Mary
Wise Smelker,MerlinSmelker,and mostof all to JamesS. Duesenberry.
2 If the originalmachineworth$1,000 and producing100 unitsis replacedby

anotherone worthalso $1,000, but producing120 units, only $833.33 will be

regardedas replacement,and the remaining$166.67 as new investment.A simi-
lar correctionis made whenthe new machinecostsmoreor less than the original
one. The treatmentof depreciation,particularlywhen accompaniedby sharp
technologicaland price changes,presentsan extremelydifficult problem.It is
quite possible that our approach, while convenientforpresentpurposes,may
give rise to seriousdifficulties
in the future.

137
138 EVSEY D. DOMAR

wagerates,relativeprices,structure ofindustry,and so on, manyof

whichare in turnaffected by the behaviorof the variablesanalyzed
here.We shallnevertheless assumeall theseconditions as givenand
shallmeanby theproductive capacityofan economy(or an asset)its
totaloutputwhenall productive factors
arefullyemployed underthese
conditions.3
The economywillbe said to be in equilibrium whenits productive
capacityP equals its nationalincomeY. Our firsttaskis to discover
theconditions underwhichthisequilibrium canbe maintained, ormore
precisely,
therateofgrowth at whichtheeconomy mustexpandinorder
to remainin a continuousstateoffullemployment.
II. THE PROBLEM OF GROWTH

The idea that the preservation of fullemployment in a capitalist

economyrequiresa growing incomegoesback (in oneformoranother)
at least to Marx. It has been fullyrecognizedin numerousstudies
(recentlymade in Washingtonand elsewhere)of the magnitudeof
grossnational productneeded to maintainfull employment. But
thoughthevariousauthorscometo different numericalresults,they
all approachtheirproblemfromthe pointof view of the size of the
laborforce.The laborforce(man-hours worked)and its productivity
are supposedto increaseaccordingto one formulaor another,and if
fullemployment is to be maintained, nationalincomemustgrowat
thecombined rate.For practicalrelativelyshort-runpurposesthisis a
goodmethod, butitsanalyticalmeritsarenothigh,becauseit presents
a theoretically
incomplete system:sincean increasein laborforceor
in its productivityonlyraisesproductivecapacityand does not by
itselfgenerateincome(similarto that producedby investment), the
demandside oftheequationis missing.Nor is the difficulty disposed
ofby Mr. Kalecki'smethodaccording to whichcapitalshouldincrease
proportionallyto the increasein laborforceand its productivity.4 As
Mrs.Robinsonwellremarked, "The rateofincreasein productivity of
laboris notsomething givenby Nature."6Laborproductivity is nota
functionof technological progressin the abstract,but technological
progress embodiedin capitalgoods,and theamountofcapitalgoodsin
3 It should undoubtedlybe possibleto workout a more precisedefinition of
productivecapacity,but I preferto leave the matteropen,because a morepre-
cise definitionis not entirelynecessaryin this paper and can be workedout as
and whenneeded.
4 See his essay, "Three Ways to Full Employment"in The Economicsof Full
Employment, Oxford,1944,p. 47, and also his "Full Employmentby Stimulating
Private Investment?"in OxfordEconomicPapers, March, 1945,pp. 83-92.
5 See herreviewof The EconomicsofFull Employment, EconomicJournal,Vol.
55, April,1945,p. 79.
 CAPITAL EXPANSION, RATE OF GROWTH, AND EMPLOYMENT 139

general.Even withouttechnological progress,capitalaccumulation in-

creaseslabor productivity, at least to a certainpoint,bothbecause
morecapitalis usedperworkman in each industry and becausethere
is a shiftoflaborto industries thatuse morecapitaland can afford to
paya higher wage.So iflaborproductivity is affected
bycapitalaccum-
ulation,theformula thatthelattershouldproceedat thesamerateas
theformer (and as theincreaseinlaborforce)is notas helpfulas it ap-
pears.
The standardKeynesiansystemdoesnotprovideus withanytools
forderiving theequilibrium rateofgrowth.The problemofgrowthis
entirelyabsentfromit becauseoftheexplicitassumption thatemploy-
mentis a function ofnationalincome.Thisassumption can be justified
onlyovershortperiodsof time;it willresultin seriouserrorsovera
periodof a fewyears.Clearly,a full-employment level of incomeof
fiveyearsagowouldcreateconsiderable unemployment today.Weshall
assumeinsteadthatemployment is a function oftheratioofnationalin-
cometo productive capacity.Whilethisapproachseemsto me to be
superiorto thatof Keynes,it shouldbe lookeduponas a secondap-
proximation ratherthana finalsolution:it doesnotallowus toseparate
unusedcapacityinto idle machinesand idle men; dependingupon
variouscircumstances, thesameratioofincometo capacitymayyield
differentfractionsoflaborforceemployed.
Because investment in the Keynesiansystemis merelyan instru-
mentforgenerating income,the systemdoes not take into account
theextremely essential,elementary, and well-known factthatinvest-
mentalso increasesproductivecapacity.6This dual characterof the
investment processmakes the approachto the equilibriumrate of
growth fromtheinvestment (capital)pointofviewmorepromising: if
investment bothincreasesproductive capacityand generates income,
it providesus withbothsidesoftheequationthesolutionofwhichmay
yieldtherequiredrateofgrowth.
Let investment proceedat therateI peryear,and let theratioof
thepotentialnetvalue added (afterdepreciation), i.e., oftheproduc-
tivecapacityofthenewprojectsto capitalinvestedin them,i.e., to
I, be indicatedby s.? The netannualpotentialoutputoftheseprojects
willthenbe equal to Is. But theproductive capacityofthewholeecon-
6 Whethereverydollarinvestedincreasesproductivecapaoityis essentiallya

matterof definition. It can safelybe said that investmenttaken as a wholecer-

tainly does. To make this statementhold in regardto residentialhousing,im-
puted rentshould be includedin the nationalincome.See also note 19.
7 The use ofthe word"project" does not implythat investment is done by the
government, or thatit is alwaysmade in newundertakings. I am using"project"
(in the absenceof a betterterm)because investmentcan mean the act ofinvest-
ing and the resultof the act.
140 EVSEY D. DOMAR

omymayincreaseby a smalleramount,becausetheoperation ofthese

newprojectsmayinvolvea transferoflabor(and otherfactors)from
reduced.8We shall
capacityis therefore
otherplants,whoseproductive
definea-,the potentialsocial averageinvestment as
productivity
dP
dt
(1)

The following characteristics ofo-shouldbe noted:

1. Its use doesnotimplythatotherfactorsofproduction and tech-
nologyremainconstant.On thecontrary, its magnitude dependsto a
verygreatextenton technological progress. It wouldbe morecorrect
to saythato-refers to an increasein capacitywhichaccompanies rather
thanone whichis causedby investment.
2. o-refersto theincreasein potential capacity.Whetheror notthis
potentialincreaseresultsin a largerincomedependson the behavior
ofmoneyexpenditures.
3. o-is concernedwiththe increasein productivecapacityof the
wholesociety,and notwiththerateofreturnderivedorexpectedfrom
investment. Therefore o-is notaffected directlyby changesin distribu-
tionofincome.
4. s is themaximum thata-can attain.The difference betweenthem
willdependon the magnitudeof the rate of investment on the one
hand,and thegrowth ofotherfactors, suchas labor,naturalresources,
and technological progress on the other.A misdirection ofinvestment
willalso producea difference betweens and o-.
We shallmaketheheroicassumption thats and a are constant.
From(1) it followsthat
(2) dP I.
dt -
It is important to notethat,witha giveno-,dP/dtis a function of
I, and not of dI/dt.WhetherdI/dtis positiveor negative,dP/dtis
alwayspositiveso longas a and I are positive.
Expression(2) showing theincreasein productive capacityis essen-
tiallythesupplysideofoursystem.On thedemandsidewe have the
multiplier theory,too familiarto need any comment, exceptforan
emphasison theobviousbut oftenforgotten factthatwithanygiven
marginalpropensityto save, dY/dtis a functionnotof I, butof dI/dt.
themarginal
Indicating propensityto save bya, and assuming it to be
we havethesimplerelationship
constant,9 that
8 I am disregardingthe externaleconomies and diseconomiesof the older
plants due to the operationof the new projects.
' Overthe period1879-1941the averagepropensityto save (ratioofnetcapital
 CAPITAL EXPANSION, RATE OF GROWTH, AND EMPLOYMENT 141

dY dI 1
(3)=__
dt dt a
Let theeconomybe in an equilibrium
positionso that10
(4) Po = Yo.
To retaintheequilibrium
position,we musthave
dP dY
dt dt
Substituting
(2) and (3) into(5) we obtainourfundamental
equation
dl 1
(X6) I - -,
dt a
thesolutionofwhichgives
(7) I = Ioeaat.
ao-is theequilibrium
rateofgrowth.
So longas it remainsconstant,
themaintenanceoffull employment requiresinvestment
to growat a con-
stantcompound-interest
rate.
If, as a crudeestimate,a is takenat 12 per centand a-at some30
percent,theequilibrium rateofgrowthwillbe some3.6 percentper
year.lOa
The readerwillnowsee thatthe assumptionofconstanta and afis
notentirely
necessary, and thatthewholeproblemcan be workedout
withvariablea and a.

formationto nationalincome) was fairlyconstantand approximatelyequal to

some 12 per cent. See Simon Kuznets, National ProductSince 1869, National
Bureau of EconomicResearch (mimeographed,1945) p. II-89 and the Surveyof
CurrentBusiness,Vol. 22, May, 1942, and Vol. 24, April,1944. In a problemof
cyclicalcharacter,an assumptionofa constantpropensityto save wouldbe very
bad. Since we are interestedherein a secularproblemof continuousfullemploy-
ment,this assumptionis not too dangerous.
10 The problemcan be also workedout forthe case when PO> YO.
lOa Afterthis paper was sent to the printer, I founda veryinteresting article
by E. H. Stern,"Capital Requirementsin ProgressiveEconomies,"Economica,
Vol. 12, August,1945, pp. 163-171, in whichthe relationbetweencapital and
outputin the U. S. during1879-1929is expressed(in billionsofdollars)as capital
=3.274 income-3.55. My estimatesgave roughlysimilarresults.This would
place s around30 per cent,thoughthisfigureshouldbe raisedto accountforthe
underutilization of capital duringa part of that period.It is also not clear how
the junkingprocess(see p. 144) was reflectedin thesefigures.
The averagerate ofgrowthof real nationalincomeoverthe period1879-1941
was some 3.3 per cent. See Table V, p. 818, and AppendixB, pp. 826-827,in my
paper,"The 'Burden'ofthe Debt and the National Income," AmericanEconomic
Review,Vol. 34, December,1944.
142 EVSEY D. DOMAR

III. THE EFFECTS OF GROWTH

Ournextproblemis to explorewhathappenswheninvestment does

growat someconstantpercentage rater,which,however, is notneces-
sarilyequal to theequilibrium
rateac-. It willbe necessary to introduce
two additionalconcepts:average propensity to save IIY and the
averageratioof productivecapacityto capitalP/K. To simplify the
problem, we shallassumethat
1. IIY= a, so thataveragepropensity to save is equal to marginal.
2. P/K=s, i.e., the ratioof productive capacityto capitalforthe
wholeeconomyis equal to thatofthenewinvestment projects.
We shall considerfirstthe special simple case a-=s, and then the
more generalcase when -<8s.11
Case 1: a-= s. SinceI = oert,capital,beingthesumofall netinvest-
ments,equals
rt 10
(8) K =Ko + IoJ ertdt= Ko +- (ert-1).
O ~~~~r
As t becomeslarge,K willapproachtheexpression

(9) - ert,
r
so thatcapitalwillalso growat a rateapproaching r.
As Y= (l/a)Ioert,theratioofincometo capitalis
1
- Ioert
Y a
(10) .

Ko +-(ert-1)
r
and
y r C
lim- = -
t(11 goo K a

Thus so long as r and a remainconstant(or changein the same

proportion)no "deepening"of capital takes place. This, roughly
speaking,was thesituationin theUnitedStatesoverthelast seventy
yearsor so priorto thiswar.
11It is also possiblethat,owingto capital-savinginventionsin existingplants,
a>s. Formallythis case can be excluded by fallingback on the definitionof
depreciationgivenin note 2. This, however,is not a veryhappy solution,but
the approachused in thispaper willhardlyoffera betterone. I think,however,
that a in our societyis sufficiently
highto make of>s in a continuousstate offull
employmentmorean exceptionthan a rule.
 CAPITAL EXPANSION, RATE OF GROWTH, AND EMPLOYMENT 143

K = P/s into(11) we obtain

Substituting
Y r
(12) lim-=
t-*O P aS

Sincein thepresentcase o=s,

Y r
(13) lim-=-*
t- o P ao

The expression
r
(14) 0 =-

may be calledthe coefficient of utilization.Whenthe economygrows

at theequilibrium rate,so thatr= ao-, 0 = 100percentand productive
capacityis fullyutilized.But as r fallsbelowao-,a fractionofcapacity
(1-0) is gradually leftunused.12Thusthefailureoftheeconomy togrow
at therequiredratecreatesunusedcapacityand unemployment.
Case 2: o-<s. As investment proceedsat the rate I, new projects
witha productive capacityofIs arebuilt.Sincetheproductive capac-
ity ofthe wholeeconomyincreasesonlyby Io, it followsthatsome-
wherein the economy(not excludingthe new projects)productive
capacityis reducedby I(s - o). Therefore everyyear an amountof
capitalequal to I(s - o-)/sbecomesuseless.
The problemcan nowbe approachedfromtwopointsofview.The
amountsI(s-o)/s, can be lookedupon as capitallosses,whichare
nottakenintoaccountin calculating incomeand investment.13 In this
case,I stillindicatestherateofnetinvestment, and all othersymbols
retaintheirold meaning, exceptthatcapitalhas to be redefined as the
integralof investment minuscapital losses: everyyear chunksof
capital(overand above depreciation) are writtenoffand junked.The
annualadditionto capitalwillthenbe
dK I(s-) Of
(15) ~ ~ I- = - = I-,
dt s s
and

(16) K = Ko + Io-f ertdt= Ko + Io -(e't - 1).

s sr
1 It shouldbe notedthat ifr, a, and o are constant,0 is also a constant.Even
thoughthe economyfails to grow at the requiredrate, the relative disparity
betweenits capacity and income does not become wider,because its capital
also growsnot at the ao but at the r rate.
18 These losses are not necessarily
losses in the accountingsense. See note 14.
 144 EVSEY D. DOMAR

Also,
Y r s
(17) lim- = -.-,
)K a a
and
Y r
(18) lim-=-,
t c P ao

whichis exactlythesameresultwe had in (13).

The secondapproachconsistsin treating theamountsI(s - o)/s not
as capitallossesbutas a specialallowanceforobsolescence. Net invest-
mentwouldthenhave to be defined notas I, but as IoIs. Othersym-
bols wouldhave to be redefined accordingly,and the wholeproblem
couldthenbe reworked outin thesamewayas on pp. 142-143.
In a sensethe choicebetweenthesetwo methodsis a matterof
bookkeeping; depending upon the characterof the problemin hand,
oneortheothercan be used,thoughI suspectthatthesecondmethod
can easilybecomemisleading.The natureof the processwillbe the
samewhichever methodis used.The factis that,owingto a difference
betweens and o-,the construction ofnewinvestment projectsmakes
certainassets(notexcluding thenewprojectsthemselves) useless,be-
causeunderthenewconditions broughtaboutby changesin demand,
ora risein thewagerates,orboth,theproductsoftheseassetscannot
be sold.'4As statedon p. 140thedifference betweens and o-is created
eitherby misdirection of investment orbythelackofbalancebetween
thepropensity to save on theone hand,and thegrowthoflabor,dis-
coveryof naturalresources, and technologicalprogresson the other.
So longas mistakesaremadeorthislackofbalanceexists,thejunking
processis inevitable.
Froma socialpointof view,thejunkingprocessis not necessarily
undesirable. In thiscountry, wheresavinginvolveslittlehardship, it
may be perfectly justified.But it may presenta seriousobstacleto
the achievement of fullemployment, becausethe ownersof capital
assetsheadedforthejunk pilewilltryto avoid thelosses.So longas
theyconfinethemselves to changesin theiraccountingpractices,no
specialconsequences willfollow.But it is morelikelythattheywill
tryto accumulatelargerreserveseitherby reducingtheirown con-
14 To be strictlytrue, the statementin the text would requireconsiderable
divisibilityof capital assets. In the absence of such divisibility,the expression
"junking" shouldnot be taken too literally.
The fact that these assets may still be operatedto some extentor that their
productsare sold at lower pricesor that both these conditionsexist,does not
invalidateour argument,because o, beingexpressedin real terms,willbe higher
than it would be if the assets wereleftcompletelyunused.
 CAPITAL EXPANSION, RATE OF GROWTH, AND EMPLOYMENT 145

sumptionor by charging higherprices(or payinglowerwages).As a

result,thetotalpropensity to save mayrise.This willbe exactlythe
oppositemeasurefromwhatis neededto avoid the junkingprocess,
and willofcourselead to greatertrouble,thoughI am notpreparedto
say to what extentcapital ownerswill succeedin passingon these
losses.
In so faras theyare able to controlnewinvestment, theywilltry
to avoidlossesbypostponing therateofgrowth
it. Consequently, may
wellbe depressedbelowtherequiredao-,and unusedcapacitywillde-
velop.Ourpresent modeldoesnotallowus to separateunusedcapacity
intoidlecapitalandidlemen,thoughmostlikelybothwillbe present.'"
Becauseofhumanitarian we are moreconcerned
considerations, with
unemployedmen.But unemployed capitalis extremely because
important,
It presentsa gravedangerto a
its presenceinhibitsnew investment.'6
in a capitalistsociety.
equilibrium
full-employment

IV. GUARANTEED GROWTH OF INCOME

In thepreceding sectionsit was shownthata stateoffullemploy-

mentcan be maintained ifinvestment and incomegrowat an annual
rateao. The questionnowarisesas to whatextenttheargument can
be reversed:supposeincomeis guaranteed to growat theao rate;will
investment
thatcall forthsufficient to generatetheneededincome?
We are concerned herewitha situationwherespontaneous invest-
ment(i.e.,investment madein responseto changesin technique, shifts
in consumers'preferences,discoveryofnewresources, etc.) is notsuffi-
cient,and thereforea certainamountofinducedinvestment (madein
responseto a risein income)is also required."To simplify theargu-
ment,letus assumethatspontaneous investment is absentaltogether.
It shouldalso be madeclearthattheproblemis treatedfroma theo-
reticalpointofview,withoutconsidering thenumerous practicalques-
tionsthattheincomeguarantee wouldraise.
If an economystartsfroman equilibrium position,an expectedrise
in incomeofYao willrequirean investment equal to Yao/s.As before,
two caseshave to be considered.

15 The presenceof unemployed men may be obscuredby inefficientutilization

of labor, as in agriculture.
16 It is truethat a givencapital ownermay oftenhave a hardtimedistinguish-

ing betweencapital idle because of a<s, and capital idle because of r< aa. The
firstkindof idleness,however,is relativelypermanent,and cannotbe corrected
by greaterexpenditures, whilethe second is temporary(it is hoped) and is due
to poor fiscaland monetarypolicies.
17 Cf. Alvin H. Hansen, Fiscal Policy and Business Cycles,New York, 1944,
Part Three,and particularlyp. 297.
146 EVSEY D. DOMAR

1. If a is equal or reasonablyclose to s, the resulting

amountof
investment of Ya willequal thevolumeofsavingsthatwillbe made
at thatlevelofincome,and equilibrium willbe maintained.18
Thusa
mereguaranteeof a rise in income(if taken seriouslyby the investors)
will actuallygenerateenoughinvestment and incometomaketheguarantee
goodwithoutnecessarilyresorting to a government deficit.
2. If u-is appreciablybelows, investment willprobablyfallshortof
savingsand equilibrium willbe destroyed. The difficultyarisesbecause
a full-employment rateofinvestment in thefaceofa u <s makesthe
junkingprocess(discussedon pp. 143-145)inevitable, whilea mere
guaranteeof a risein income,as a generalrule,lackstheinstrument
to forcethecapitalownersto discardtheirequipment. Theywillsimply
investYao-Isinsteadof Ya. Onlyifin theeconomyas a wholethere
is a considerable numberofproductsthe demandforwhichis highly
elasticwithrespectto income,and therefore a goodnumberofothers
thedemandforwhichis negatively elasticwithrespectto income,will
a largeramountthan YaorIs be investedand a corresponding amount
of capitaljunked.Of course,if therisein incomeis accompaniedby
shiftsin consumers'preferences, the appearanceof new products,
aggressive competition, and otherchanges,thejunkingprocesswillbe
speededup, but ifthesechangesdo take place theymaygiveriseto
spontaneous investment oftheirownand theguaranteed riseinincome
willnotbe important. Still,theassuranceofa highand risingincome
is undoubtedly one of the best methodsforencouraging investment.
As explainedbefore, a substantial differencebetweens and ufsimply
indicatesthatwiththe availablelaborforceand the current progress
of technology, the maintenance of fullemployment undera givena
requirestheaccumulation ofcapitalat a fasterratethanit canbe used.
As a ge.neralrule,thisappliesequallywellto bothprivateand public
investment, thoughtheremay be special cases when,owingto the
development ofparticular consumers' preferences (e.g.,forvacations),
or to technological reasons(e.g.,need forpower),or to institutional
conditions(as in urbanredevelopment), considerable needforpublic
investment stillexists.'9
18 There is a slighterrorin the magnitudesin the text because of the use of

discontinuousfunctions.
19As soon as the government entersthe picturewe findourselvesin a maze of
definitionalproblems.From the point of view of this paper, saving and invest-
ment should be understoodin referenceto the whole economy,includingthe
government, and not to its privatesectoronly.But whichgovernmentexpendi-
turesshouldbe regardedas investment?The difficulty is presentin the private
sectoras well,exceptthat therewe can take refugein formaldefinitions, which
cannot be well applied to government.I leave the question open. Certainly,
investmentneed not be limitedto inventories, steel,and concrete.
 CAPITAL EXPANSION, RATE OF GROWTH, AND EMPLOYMENT 147

I am not preparedto say whetherwe alreadyare or shallsoonbe

facedwitha seriousdifference betweens and o-,thoughI doubtthat
it was an important problemin thepast,exceptperhapsfortheshort
boomyears.My ownguessis thatwe shallbe moreconcerned withthe
disparity betweenao-and r, thatis withthefailureofincometo grow
at therequiredrate.
If, however,thedifference betweena-and s becomesseriousandin-
hibitsinvestment, or if the junkingprocessproceedsat a fasterrate
thanis deemedsociallydesirable,thesocietywillhave at itsdisposal
two methodsnot mutuallyexclusive:(1) the reductionof the pro-
pensityto save,or (2) thespeedingup oftechnological I hope
progress.
thatthemainemphasiswillbe placedon thelatter.

This paperattemptedto analyzethe relationbetweeninvestment,

rate of growth,and employment. The analysiswas carriedout on a
veryabstractand simplified level-a procedurewhichmaybe justified
at thebeginning ofan investigation,
butwhichmustbe corrected later
on. In general,thereis no sucha thingas an absolutelygood or bad
assumption:whatmaybe safein one kindof a problemcan become
fatalin another.Oftheseveralassumptions madehere,thatregarding
depreciationis likelyto cause thegreatestdifficulties,
but it is by no
meanstheonlyone. I hopeto developthewholesubjectfurther at a
laterdate.
The centralthemeof the paper was therateof growth, a concept
whichhas been littleused in economictheory,and in whichI put
muchfaithas an extremely usefulinstrument of economicanalysis.
One does not have to be a Keynesianto believethatemployment is
somehowdependent on nationalincome,and thatnationalincomehas
something to do withinvestment. But as soonas investment comesin,
growthcannotbe leftout,becauseforan individualfirminvestment
maymeanmorecapitaland lesslabor,butfortheeconomyas a whole
(as a generalcase) investment meansmorecapitaland notlesslabor.If
bothareto be profitably employed, a growthofincomemusttakeplace.
D. C.
Washington,