Вы находитесь на странице: 1из 13

SAVING EQUALS INVESTMENT

SUMMARY
I. Why SI4, 297.-Difficulties of seeing this: the confusionbetween
stocks and flows, 299; the paradox of individual freedom and social
necessity, 300; the habit of labeling expenditureas "out of" particular
income receipts, 302; the failure to recognize the mathematical or
analytical nature of the proposition, 304; a misunderstandingof argu-
ments about equilibrium, 305. - II. Miss Curtis' condition that "all
income is spent," 305; her failure to see the place of "wishes" in eco-
nomic analysis, 307.- III. An attempt to salvage one of Miss Curtis'
results, 308.

In a recent articleI in this Journal, Miss Myra Curtis


attacks the formulationput forwardby Mr. Keynes in his
book,2and repeatedby me in an article,3accordingto which
saving and investment, for a whole economy, are always
equal. This equality has appeared paradoxicalto many
economists,and many difficultieshave hinderedits general
acceptance. In the first part of this article I shall endeavor
to clear up some of these difficulties. In the second part I
shall considertwo of Miss Curtis'criticismsat greaterlength;
and in the concludingsection I shall discussthe possibilityof
salvaging one point which does not rest entirely upon mis-
understandingof Mr. Keynes' argument.
Mr. Keynes and I and most people wouldsay that a man
saves somethingin a given period,if he spendson consump-
tion (consumes)in that period less than his income in the
period. The only unambiguousmeasureof the amountof his
saving is obtained by subtracting his (expenditureon) con-
sumption in. the period from his income in the period.
y (Income) - c (Consumption) = s (Saving) by definition. If
he consumesmore than his income, he is doing the opposite
1. "Is Money Saving Equal to Investment?" Quarterly Journal of
Economics, August, 1937.
2. The General Theory of Employment Interest and Money.
3. "Mr. Keynes"GeneralTheory of Employment,"' I. L. 0. Review,
October, 1936.
297
298 QUARTERLYJOURNALOF ECONOMICS

of saving, dissaving,and we similarlymeasurethe amountof


his dissavingby subtractinghis incomefromhis consumption.
c-y= -s is the same equation (with the signs changed)in
which - s can be called dissaving.
Investment is the expenditureof money on things other
than consumption. There is no reason why, for any indi-
vidual, there should be any particularrelationshipbetween
his investment (i) and the items y, c, and s mentionedabove.
But when we considera whole (closed)economy,we see that
there emergesa relationshipbetween these items that does
not appearto exist for the individual.
The equationy- c = s, since it is true for every individual
in the economy,is also true for any two

Y2-c2=82

(YI+Y2) - (CI+C2) = (81+82)


or any other numberof individualsin the economy. If we
take all the individualstogether and add up their incomes
and consumptionsand savings (using capital letters to rep-
resent these sums for the whole economy),we get Y- C= S.
In this respect, then, the whole economy is like any indi-
vidual.
But for the whole economy there is another relationship.
The sum of the incomesof all the individualsin the economy,
Y. is equal to the sum of the expendituresof all kinds by
the individualsof the economy,since these expendituresare
nothing but the payments, the receipt of which constitutes
all the incomes. The sum of all the paymentsmust be equal
to the sum of all the receiptsin the same period,since these
are the same thing, only lookedat fromdifferentangles. The
sum of expendituresof all kinds, which is equal to Y, must
consistof C, the sum of expenditureson consumption,plus I,
the sum of expenditureson things other than consumption,
sincethese two makeup all possibleexpenditures. This gives
us the equation Y=C+I or Y-C=I. We know that Y-C
is also equal to S, and since quantitiesthat are equal to the
same quantity are equal to one another, we get the result
SAVING EQUALS INVESTMENT 299

that S = 1. The sum of the savings of all individualsis equal


to the sum of their investmentsin the same period.
The resistance that this piece of very simple arithmetic
arousesin many people can usually be traced back to one
or more of the followingfive causes:
(1) A failure to recognize that all the items considered
are payments (or differences between payments) over a
period,and neveramountsexistingat somepointof time (such
as the beginning or the end or some intermediate point
within the period to which our propositionrelates). They
are all of the nature of flows which can be measuredeither
as so much in a given period (as in the simplest case examined
above) or as so much per unit of time (if we suppose the flows
to continueat an unchangedrate over severalunits of time).
They can never be measured as so much existing at any
moment of time. That can only be done of stocks,not of
flows, and our propositiondeals only with flows.
This failure to keep clear of irrelevant considerationsof
stocks (of money) may take the form of
(a) An insistenceon the discussionof the velocity of circu-
lation of money. The velocity of circulationis nothing but
the ratio betweensome total of money paymentsin a period
(which,beinga flow,may be relevantto our proposition)and
some stockof money existing at some point of time (which,
since it is a stock, is on a differentplane and can have no
relevanceto our proposition).
(b) An insistence on the discussion of "hoarding"(and
"dishoarding"). Sometimes "hoarding"means a reduction
in the velocity of circulation,the irrelevanceof which has
been shown. Sometimesit means simply holding stocks of
money. Sometimesit meansincreasingone'sstock of money.
And most frequentlyit mysteriouslymeansall three of these
simultaneously,as well as a lengtheningof the period an
individual holds particularcoins or notes. The concept of
stocksinherentin all of these usages rendersirrelevantany
validity that the particular meaning of "hoarding"may
retain.
Lack of clarityas to whetherstocks or flowsare being dis-
300 QUARTERLY JOURNAL OF ECONOMICS

cussed has played a great part in feeding useless discussions


in economics in the past. The Wages Fund is a conspicuous
example of an ambiguous word used to cover such a con-
fusion, and in modern theory of capital the same confusion
is a great stumbling block. The proposition that I=S is a
proposition about flows and has nothing to do with stocks.
(2) A failure to understand the paradox that, while each
individual separately is free to save either more or less than
he invests, all the individuals together are not so free, since
the sum of the investments I is always equal to the sum of
their savings S. How does this compulsion work? If it does
not affect any individual, how can it affect the whole econ-
omy, which is simply the sum of the individuals?
To understand paradoxes of this nature is the special
province of the economist, and many other similar paradoxes
have by their familiarity ceased to terrify and become part
of the stock in trade of all economists. Any country is free
to import more goods than it exports or vice versa, but world
imports always equal world exports (plus freight charges,
etc.). Any individual can take his money out of the bank
tomorrow morning, but all individuals together cannot. And
we have the converse paradox. One bank or one country
cannot expand its credit indefinitely; all the banks, or all
the countries, keeping in step, can do so. To insist that what
is true of each individual must be true of all individuals
together is the simple fallacy of composition.
But how does the compulsion work, if it does not affect
individuals? This question leaves many students uneasy.
The answer is that the individual is by no means as free to
decide how much he is going to save as has been suggested.
There are very few individuals who would not like to have
larger incomes than they actually have and to save more out
of these larger incomes. Each individual is constrained to
save the amount that he does by the size of his income; and
the size of his income is determined by other people's expendi-
tures on the goods that he produces. Each individual con-
siders his own income as given and independent of his own
expenditure (since in a large community the repercussions
SAVING EQUALS INVESTMENT 301

on his own income of any variations in an individual's expend-


iture are in general likely to be small enough to be legiti-
mately neglected); and, not being interested in the effect of
his expenditure in creating income for somebody else, he sees
no connection between income and expenditure. This does
not mean that the connection does not exist for the individual.
It merely means that he is not interested in it, insofar as his
own expenditure affects other people's incomes (tho he
may have a lively interest in the effect of other people's
expenditure on his own income). The economist has a wider
outlook, must concern himself equally with the incomes of
all the individuals, and so must recognize that for the whole
community the excess of total incomes over that part of
incomes that is created by expenditure or consumption must
have been created by investment (or expenditure on other
things), so that I =S.
The failure to face up to the paradox of social necessity
with apparent individual freedom sometimes takes the form
of trying to extract from the total of an individual's actual
saving (i.e., the excess of his income in a period over his con-
sumption in the period) some part of it that really is "free"
or "voluntary" or "ex ante" and declaring that the rest of
his saving is "forced" or "involuntary" or "really saved
by somebody else" (i.e., the investor who produced something
that cannot be consumed) so that it should not be counted.
All such attempts necessarily fall to the ground for the lack
of any situation, to serve as a basis for comparison, in which
the individual can with any plausibility be said to be uncon-
strained in his saving or even less constrained than in the
period discussed. It is much more satisfactory to recognize
that in a determinate universe all saving, like everything
else, is "forced" and that free will is nothing but a pleasant
illusion.
Related to this difficulty is the unconscious assumption-
taken from the point of view of the individual and illegiti-
mately transferred to society - that while expenditure is
varied, income remains the same. From this it would follow
that an increase in saving always means a reduction in con-
302 QUARTERLY JOURNAL OF ECONOMICS

sumption4(and never an increase in income, consumption


remainingconstant). The assumptionof constant income is
then dropped and the fall in consumptionis allowed to
diminish income, so that any increase in saving appears
necessarilyto involve a diminutionof income.5 From this
type of argumentany numberof surprisingresults naturally
follow, such as that altho there is an increaseof saving (to
which the fall in expenditureis due) there is no change in
saving (since incomehas fallen as much as consumption).6
(3) A tendencyto regardexpenditure,not as a flow during
a periodcoincidentwith the flow of income duringthe same
period,but as somethingcoming"outof" the incomereceived
in the period. "Saving"is on this view the incomereceivedin
a periodminus the expendituremade "out of" that income.
One possible meaning of this is simply that only that
expenditureis to be countedwhich takes place aftersome or
all of the income is received. If the ambiguitiesin this are
overcome - as they can be - by some arbitrary ruling as
to when we are-to begin countingthe expenditure,we will,
of course, find that "SS"so definedis greater than I by all
the expenditurethat took place too early in the periodto be
consideredto be "out of" the income receivedin the period,
so that we did not count it. If this procedurewere carried
to its logical conclusion,this expenditure,not beng "out of
income,"would have to be consideredas dissavingand sub-
stractedfrom "S" and so reduceit to exactly the same value
as I. It is, however,.not usually carriedto its logical conclu-
sion, and is consideredto be a demonstrationof the falsity
of our propositionthat I =S.
4. E.g., "An increase in S must be accompanied by a reduction in
expenditure on consumption goods."-Miss Curtis, Quarterly Journal
of Economics, August, 1937, p. 617.
5. E.g., "An increase in S depressesincome."- Ibid., p. 617.
6. Miss Curtis seems to hold that the converse of the proposition
quoted in footnote 4 above is also true, and to suppose that I hold the
same view. Thus she says that when I speak of a diminution of con-
sumption I must mean an increase of S ("or why the reduction in total
consumption and in income?"- Ibid., p. 617). This enables her, when
she realizes that the contradictionbetween the two propositionsquoted
SAVING EQUALS IN VESTMENT 303

Anothermeaningof the insistenceon countingC, I and S


only insofaras they come "out of" the income receiveddur-
ing the period is that we must count only the expenditure
(or laying up) of the particularnotes or coins received as
incomeduringthe period. Thus if somethingis bought with
money received before the period began, it is not expendi-
ture "out of" income. On this line of analysis, Peter, who
took his wagesbag to the grocer,has spent all his incomeand
saved nothing, while Paul, who put his wages bag into his
safe and took last week'sbag to the grocer,has saved all his
income. At this stage of the argumentit is again not neces-
sary for I to equal S. Of course,if this method of counting
were carriedto its logical conclusionand the expenditureof
coins other than those receivedas incomein the periodwere
regardedas dissaving, we would find ourselvesback at our
arithmeticalresult that I=S; but to do that would be to
destroy the whole purposeof this new method of counting.
Correlativewith the objection to countingspendingthat
is not "out of income" is an objectionto countingas saving
the unspent income with which a man is caught at the end
of a period,even tho he may not have the slightestintention
of saving it.7 This looks like a seriousdivergencefrom the
ordinaryman's idea of what is meant by saving, and has
inspired Mr. Robertsonto another of his delightful quota-
tions from "Alice."8 This would be justified if by saving
were meant particularcoins or notes receivedas income and
not spent. But we are not interestedin particularnotes or
coins, and what is included in the saving of an individual,
apart from the saving that he has used to buy assets other
than money, is the excessof the money he holds at the end
of a period over the money he had at the beginning. If a
man started a periodwith ?20 and finds himself at the end
of the periodwith ?25, it does not conflictwith commonsense
leads to the absurditiesnoted in the text of my article, to attribute the
same confusion to me.
7. I am grateful to Dr. H. W. Singer for drawing my attention to
this form of my third type of difficulty - an important form which I
had overlooked.
8. Economic Journal, September, 1937.
304 QUARTERLY JOURNAL OF ECONOMICS

to say that he has saved ?5 in that period, even if it is his


intention to spend the whole ?25 (or more) in succeeding
periods. Of course if we take highly artificial periods - say,
of ten minutes each - our definitions acquire an artificial
flavor too. We would then have to say that in the ten-
minute period in which a man receives his weekly wage he
saves (nearly) all of it, and that in all the other ten-minute
periods in which he makes any expenditures he dissaves. But
if we take reasonable periods, this artificiality disappears.
There is, of course, a sound idea underlying the notion of
considering only such expenditure as comes after or "out of"
income. It is that an individual's expenditure is determined
more by income in the past, which is known and has been
received, than by income in the present, which is uncertain.
This may be true to a certain extent, altho the effect of
expected income on a man's expenditure must not be left
out altogether. It is important for the real economic prob-
lems of forecasting expenditure and income, and has its
place in economic theory, much more important than our
piece of simple arithmetic; but it cannot be used to show
that two and two are five.
(4) The failure to realize that the proposition S=I is only
an analytical proposition, and not about the real world at all.
What is taken to be a statement about the real world excogi-
tated from an armchair is naturally looked upon with sus-
picion. Our proposition is not based upon observation of the
real world. It therefore cannot tell us anything we did not
know; neither can it turn out to be mistaken. It follows from
and is implicit in our definitions of income, consumption,
savings and investment, and the postulate that in any period
moneys paid out are equal to moneys received. It is a propo-
sition of the same order as the proposition that the area of
the square of the hypotenuse of a right-angled triangle is
equal to the sum of the areas of the squares on the other two
sides. It has been called a truism, often in tones of contempt,
telling us nothing but that something is equal to itself. In a
sense this accusation is justified. All the propositions of
mathematics are similarly truisms, since they tell us nothing
SAVING EQUALS INVESTMENT 305

that is not impliedin the basic definitionsand postulates. To


one who sees these implicationsin the postulatesthemselves
the enunciationof the propositionsof mathematicsare noth-
ing but an arrayof truismsand a waste of time, and I under-
stand that there are born mathematiciansfor whom proposi-
tions like Pythagoras' and the multiplication table are
unnecessaryencumbrances. The usefulnessof propositions
of this mathematicalnature is an inverse function of their
obviousness. The abundant discussionthat has grown up
aroundthe proposition,made famous by Mr. Keynes, that
S=I is abundantproofthat its truth is not instantaneously
obviousto all men; andif withoutadducingany new informa-
tion it leads them to see some implicationspreviouslyover-
looked,it carriesout the purposefor which it was designed.
(5) A belief that the short period equilibriumwhich is
discussedin the analysis that makes use of our proposition
is a necessary conditionfor the realizationof the equality.
This would indeed be suspicious, since the proof of the
equality-e.g., as given in the first pages of this article
does not mention equilibrium. This seems to go with a
belief that it is the ultimate goal of Mr. Keynes and his
followersto show that I = S and then to retirefrom the field
of economics. The equality of I to S has nothing whatever
to do with any kind of equilibrium. Equilibriumis discussed
as a conditionfor some kind of stability of Y and C (and con-
sequentlyalso of I and S). The equationof I to S is always
true and serves as a check, since any resultthat makesthem
unequalmust involve a mistakeeitherin logic or in counting.

II
A first readingof Miss Curtis'articlegives the impression
of a daring attack on the propositionthat I= S. A closer
examinationshows, however,that Miss Curtis puts forward
all of the objections and difficultiesdiscussed in the first
first section of this paper (as well as some minor confusions
more peculiarto herself) and in orderto be able to dismiss
the propositionas a truism she has ultimatelyto admit it as
true. This she does on page;616 wherefor the first time she
306 QUARTERLY JOURNAL OF ECONOMICS

definessaving in the way Keynes does (S" in her notation),


having spent more than half of the article in trying to show
that the equation does not hold for other definitionsof S
(saving "out of income" as discussedin I (3) above, not
carriedto its logical conclusion;and a hybrid between this
and a stock of money as discussedin I (1) above- "total
amount withheldfrom consumptionin the period,"p. 615).
There are, however,two points in her article that I should
like to discussfurther.
Miss Curtis'main slip is to be foundin her statementthat
"a hidden conditionis attached to the equations- namely
that all income (and nothing from other sources)is spent in
the period" (p. 607). "The conditionis, however, not one
that can be expectedto be uniformlyfulfilled. For if it were,
spending would be constant and incomes would never
change." (p. 610)
This looks at first like a denial that total income must be
equal to total expenditure(i.e., C+I). Miss Curtis shows,
however,that she does not make as simplea mistake as this.
She says, indeed, "Tho all expenditurebecomesincome, all
income need not become expenditure"(p. 608), and in her
arithmeticalexampleswhich are designedto show the falsity
of our proposition,she wisely avoids any internal contra-
dictionsby makingincome (Y) equalthroughoutto spending
(C+I) in the same period. She does not deny that incomes
and total spendings in any period must be equal to each
other.; What she denies is that incomes must be equal to
spending"out of" the incomesreceived. This condition,she
says, "cannotbe expectedto be uniformlyfulfilled"(p. 610);
"receipts of income . .. may be passed on during the period
either wholly, or in part, or not at all, to form the income of
others." (p. 608)
In this Miss Curtisdoes not go far enough. It is impossible
for all incomereceipts "to be passedon" during any period,
for the act of passingincome on by one personis the act of
receivingincomeby the other person,and wheneverthe gong
goes to mark the end of our period, there must always be
somebodyleft with unspent income. In a game of musical
SAVING EQUALS INVESTMENT 307

chairs,the players cannot beat the band, howeverfast they


run. The condition describedby Miss Curtis as one that
"cannot be expected to be uniformlyfulfilled"is one that
can neverbe fulfilled. Fortunatelythere is no need for the
fulfillmentof this impossiblecondition. The meaningbehind
Miss Curtis' conclusionthat if this (as we have seen impos-
sible) conditionwere fulfilled, "spendingwould be constant
and income would never change" will be examinedin the
next section.
The other point is concerned with a criticism both of
Mr. Keynes and of myselffor speakingof "attempts"to save
and "wishing"to save amounts differentfrom the amount
of investments. Miss Curtisis, of course,absolutelyright in
suggesting(p. 619)that "attempting"and "wishing"to spend
have in themselvesno effect on anything, except insofar as
they are translated into actual spending. She is not on
such firm ground when she applies the same argumentto
saving or, in her language,"withholdingof income." Saving,
or the "withholdingof income," is not an action that has
effect and that can be contrastedwith the mere"attempting"
or "wishing"to save in the same way as actually spending
money can be contrasted with merely wishing to spend.
Income receiversare free to spend on consumptionas much
or as little as they wish (within certainlimits, of course),but
they are not, taken altogether, able to save any desired
amount as simply as they are able to spend any desired
amount. This is becausetheir saving depends,not only upon
their spending,but also upon income, sinceit is the difference
between these. With the rate of investmentgiven (it being
undertakenby other people, or at any rate determinedby
other forces), income receivers are not able to save either
more or less than is being invested. They may wish to save
more and may try to implementthis wish by actually spend-
ing less. This, however,has the effectof diminishingincomes
by exactly the same amount, so that while there has been an
actualdiminutionin expenditure(anda correspondingactual
diminutionof Incomes),the increasein saving has neverthe-
less remaineda wish. In the same way a wish to save less
may result in increasedspendingand increasedincomes,but
308 QUARTERLY JOURNAL OF ECONOMICS

not in any diminution in saving - that decrease always


remains in the realm of wishes.
An equilibrium is reached, but not by translating the wish
for a larger (or smaller) amount of saving into an actuality
and so satisfying the wish and thereby dispelling the driving
force that is incompatible with equilibrium. Equilibrium
is reached by a fall (rise) in incomes and consumption as
far as is necessary to make the income receivers give up
their wish. If they are obstinate, they merely make a greater
fall (rise) in incomes necessary before they change their mind.
When they acquiesce, we have equilibrium. Reality has not
been adjusted to fit the wish; the wish has been adjusted to fit
reality.
Equilibrium here merely means that there is no longer
any tendency for Y and C to move down (or up) together.
To suppose, as Miss Curtis does (p. 620), that this equilibrium
is necessary to make I=S is to misunderstand the whole
point. That equality has nothing to do with equilibrium,
and is undisturbed however violent may have been the move-
ment towards equilibrium, if indeed it is ever reached. For
the parallel movement of Y and C does not in the least affect
the gap between them which is S and equal to I. The exam-
ination of all this movement towards equilibrium is not car-
ried out in the least for the purpose of demonstrating our
little bit of arithmetic (tho it may be observed there as any-
where else), but to consider the effect on the economy of
wishes on the part of individual income receivers to vary the
amount they are saving - not by telepathy but through the
changes in actual spending that result from those wishes.

III
There seems to be one point that Miss Curtis is perhaps
trying to make which does not rest entirely upon misunder-
standing. When she says that Mr. Keynes' definition of sav-
ing (the excess of income over expenditure on consumption
goods) "has nothing to do with saving in the ordinary sense
of withholding money income from consumption expendi-
ture" (p. 616), she may mean that people's expenditure is a
more stable function of income received at some time or in
SAVING EQUALS INVESTMENT 309

some period in the past than of income received in the same


period. Individuals consider, say, last week's income when
they make their decisions as to how much they are going to
spend, and consider themselves to have saved the difference
between last week's income and this week's spending.
Some such interpretation gives sense to Miss Curtis'
impossible condition "If ... all income of the period (and
no more than income) is spent and recreates itself as income
in the period" (p. 610), and gives validity to her otherwise
baseless conclusion "spending would be constant and incomes
would never change." For if spending (here total expendi-
ture, C2+I2) in period Two is equal to income (Y1) in period
One, income in period Two (Y2=C2+I2) is equal to income
in period One (Y1) and income is unchanged.
Such a salvaging of Miss Curtis' argument, however,
involves the adoption of a technique of analysis in terms of
successive periods, of the kind developed by Mr. Robertson
and the Swedish writers. Of this there is no trace in Miss
Curtis' article, which throughout considers simply "the
period." Again the difference between these writers and
Mr. Keynes and his followers is on quite another level. I
am rather skeptical concerning the usefulness of analysis in
terms of successive periods or "days," because it seems to me
to complicate and confuse matters rather than to clarify
them. But that is merely a hunch on my part as to which is
the more promising (or perhaps more pleasant!) road for
research. I may easily be mistaken in this and am ready to
welcome any results that those who like this technique may
produce. There is no question here of right or wrong or of
logical errors committed.
Miss Curtis claims, however, not a difference of technique
adopted but the correction of an error. It is therefore prob-
ably best to give up our attempt at salvage and to say that
Miss Curtis' denials that S=I are simply the result of con-
fusion of thought. Her various definitions of saving, if car-
ried through to their logical conclusions, all come to the same
as Mr. Keynes' definition, and the equation applies just as
much to these as to Mr. Keynes'.
LONDONSCHOOLOF ECONOMICS A. P. LERNER.

Вам также может понравиться