Kaldor - 1939 - Speculation and Economic Stability

8/10/2019 Kaldor - 1939 - Speculation and Economic Stability

1/28

The Review of Economic Studies Ltd.

Speculation and Economic StabilityAuthor(s): Nicholas KaldorSource: The Review of Economic Studies, Vol. 7, No. 1 (Oct., 1939), pp. 1-27Published by: Oxford University PressStable URL: http://www.jstor.org/stable/2967593 .Accessed: 29/11/2014 08:31

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 formsof scholarship. For more information about JSTOR, please contact [email protected].

.

Oxford University Press and The Review of Economic Studies, Ltd. are collaborating with JSTOR to digitize,preserve and extend access to The Review of Economic Studies.

http://www.jstor.org

This content downloaded from 142 .150.190.39 on Sat, 29 Nov 20 14 08:31:05 AMAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/action/showPublisher?publisherCode=ouphttp://www.jstor.org/stable/2967593?origin=JSTOR-pdfhttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/stable/2967593?origin=JSTOR-pdfhttp://www.jstor.org/action/showPublisher?publisherCode=oup


2/28


3/28

2 THE REVIEW OF ECONOMIC STUDIES

such speculative activity would be attended by a loss, and not a gain; andsuch speculators would be speedily eliminated. Only the speculator withbetter than average foresight can hope to remain permanently in the market.And this implies that the effect of speculative activity must be price-stabilising,and in the above sense, wholly beneficial.

This argument, however, implies a state of affairs where speculativedemand or supply amount only to a small proportion of total demand orsupply, so that speculative activity, while it can influence the magnitude of theprice-change, cannot at any time change the direction of the price-change. Ifthis condition is not satisfied, the argument breaks down. It still remains truethat the speculator, in order to be permanently successful, must possess betterthan average foresight. But it will be quite sufficient for him to forecastcorrectly (or more correctly) the degree of foresight of other speculators, ratherthan the future course of the underlying non-speculative factors in the market.,If the proportion of speculative transactions n the total is large, it may become,in fact, more profitable for the individual speculator to concentrate on fore-casting the psychology of other speculators, rather than the trend of thenon-speculative elements. In such circumstances, even if speculation as awhole is attended by a net loss, rather than a net gain, this will not prove,even in the long-run, self-corrective. For the losses of a floating population ofunsuccessful speculators will be sufficient to maintain permanently a smallbody of successful speculators; and the existence of this body of successfulspeculators will be a sufficient attraction to secure a permanent supply of thisfloating population. So long as the speculators differ in their own degree offoresight, and so long as they are numerous, they need not prove successful inforecasting events outside; they can live on each other.

But the traditional theory can also be criticised from another point ofview. It ignored the effect of speculation on the general level of activity-orrather, it concentrated its attention on price-stability and assumed (implicitlyperhaps, rather than explicitly) that if speculation can be shown to exert astabilising influence upon price, it will ipso facto have a stabilising influence onactivity. This, however, will only be true under certain special assumptionsregarding monetary management which are certainly not fulfilled in the realworld. In the absence of those assumptions, as will be shown below, speculation,in so far as it succeeds in eliminating price fluctuations will, in many cases,generate fluctuations in the level of incomes. Its stabilising influence on pricewill be accompanied by a de-stabilising influence on activity. Hence thequestion of the effect of speculation on price-stability and its effect on thestability of employment ought to be treated, not as part of the same problem,but as separate problems.

In the subsequent sections of this paper we shall deal first with the condi-tions under which speculation can take place, secondly, with the effect ofspeculation on price-stability, and finally, with the influence of speculation oneconomic stability in general.

1 Cf. Keynes. General Theory, ch. 12 on Long Term Expectations. We have reached thethird degree [in the share markets] where we devote our intelligences to anticipating what averageopinion expects average opinion to be. (p. I56.)

http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp


4/28

SPECULATION AND ECONOMIC STABILITY 3

I. THE PRE-REQUISITES OF SPECULATION3. Not all economic goods are the objects of speculative activity; in fact

the range of things in which speculation, on any significant scale, is possible israther limited. The two main conditions which must be present in normalcircumstances in order that a particular good or asset should be the objectof speculation, is the existence of a perfect, or semi-perfect market, and lowcarrying cost. For if carrying costs are large, and/or the market is imperfect,and thus the difference between buying price and selling price is large, specula-tion becomes far too expensive to be undertaken-except perhaps sporadically,or under the stress of violent changes.'

The presence of these two conditions presupposes, on the other hand, anumber of attributes which only a limited number of goods possess simul-taneously. These attributes are: (i) The good must be fully standardised, orcapable of full standardisation; (z) It must be an article of general demand;(3) It must be durable; (4) It must be valuable in proportion to bulk.

The first two conditions are indispensable for anything resembling aperfect market to develop in exchange.2 The last two ensure low carrying cost.For the greater the durability the less is the wastage due to mere passage oftime, the greater the value in proportion to bulk the less the cost of storage.

These last two factors make up what might be called carrying costs proper.But net carrying cost also depends on a third factor: the yield of goods. Innormal circumstances, stocks of all goods possess a yield, measured in terms ofthemselves, and this yield which is a compensation to the holder of stocks,must be deducted from carrying costs proper in calculating net carrying cost.4The latter can, therefore, be negative or positive.

From the point of view of yield, it is important to distinguish betweentwo categories of goods: those which are used in production and those whichare used up in production. (There is no convenient English equivalent for theGerman distinction between Gebrauchsgi2ter nd Verbrauchsgiiter, both ofwhich can refer to durable goods.) Stocks of goods of the latter category also

I In conditions of hyper-inflation-as in Germany in 1923-the range of goods in whichpeculation takes place is, of course, very much extended.

2 In particular he degree of standardisation equired s very high. For the difference betweenthe simultaneous buying price and selling price can only be small when there is a large andsteady volume of transactions, in the same article, per unit of time, so that it pays individuals toundertake purchases and sales through the agency of an organised market. It is necessary for thisthat the commodity in question should have but few attributes of quality (that it should be simple)so that a specification of standard qualities ( grades ) can be drawn up without difficulty. Inthe organised exchanges of the world this standardisation s carried so far that buyers rarely seethe article they are actually buying-contracts are made between buyers and sellers with referenceto standard grades and places of delivery. It is also necessary that the amount bought by therepresentative individual in a single transaction should represent considerable value. Nobodywould go to the trouble of buying, say, cigarettes, through the agency of a central market, evenif this would imply some saving.

3 By defining yield as that return which goods obtain when measured n terms of themselves,we exclude here any return due to appreciation of value (in terms of some standard) whetherexpected or unexpected.

' Our definition of net carrying cost is, therefore, the negative of Mr. Keynes' own-rate ofown-interest in ch. I7 of the General Theory-except that no allowance is made here for thefactor termed liquidity premium. Our reason for deducting the yield from the carrying cost,and not the other way round, is because (as will be clear below) from the point of view of specula-tion, net carrying cost is the significant concept rather than the own-rate of interest.



5/28


have a yield, qua stocks, by enabling the producer to lay hands on them themoment they are wanted and thus saving the cost and trouble of orderingfrequent deliveries, or of waiting for deliveries.' But the important differenceis that with this latter category, the amount of stock which can be thus

useful is, in given circumstances, strictly limited; their marginal yield fallssharply with an increase in stock above requirements and may rise verysharply with a reduction of stocks below requirements. 2 When redundantstocks exist, the marginal yield is zero.3 With the other category of goods,items of fixed capital, the yield declines much more slowly with an increase instock, and it is normally always positive. Hence as we defined speculativestocks as the excess of stocks over normal requirements (i.e. that part ofstocks which is only held in the expectation of a price-rise and would not beheld otherwise) we may say that with working-capital-goods (Verbrauchsgiiter)carrying costs are likely to be positive, when speculative stocks are positive,and negative when they are negative; with fixed-capital-goods (Gebraucchs-giiter), carrying costs are normally negative, irrespective of whether specula-tive stocks are positive or negative.

It would follow from this that fixed-capital-goods ike machinesorbuildings,whose carrying cost is negative and invariant with respect to the size ofspeculative stocks, ought to be much better objects of speculation than raw-materials, whose carrying costs are so variable. The reason why they are not,is because the condition of high standardisation, necessary for a perfect market,is not satisfied, and hence the lag between buying price and selling price islarge. It is not that machines, etc., by being used, become second-hand,and thereby lose value, since the depreciation due to use is already allowedfor in calculating their net yield. The reason is that all second-hand machinesare to some extent de-standardised; it is very difficult to conceive a perfectmarket in such objects.4,5 The same lack of standardisation accounts for thecomparative absence of speculation in land and buildings.

This explains, I think, why in the real world there are only two classesof assets which satisfy the conditions necessary for large-scale speculation. Thefirst consists of certain raw-materials, dealt in at organised produce exchanges.

1 There is, of course, in addition, the stock of goods in the course of production (goods inprocess) which depends on the length of the production process, but with this we are not hereconcerned (since they are not standardised).

2 This, as we shall see below, is equally true of stocks of money, as of other commodities.a Mr. Keynes, in the Treatise on Money, uses the term working capital for stocks which

have a positive yield, and liquid capital for those which have a zero yield. (Vol. II, pp.I30.)4 I.e. the seller of a second-hand machine must not only allow for a reduction of value due to

depreciation, but also an extra loss due to the fact that he is selling the machine and not buying it.5 Mr. Keynes, in certain parts of the General Theory, appears to use the term liquidity in

a sense which comes very close to our concept of perfect marketability ; i.e. goods which canbe sold at any time for the same price, or nearly the same price, at which they can be bought. Yetit is obvious that this attribute of goods is not the same thing as what Mr. Keynes really wants tomean by liquidity. Certain gilt-edged securities can be bought on the Stock Exchange at aprice which is only a small fraction higher than the price at which they can be sold; on thisdefinition, therefore, they would have to be regarded as highly liquid assets. In fact it is verydifficult to find satisfactory definition of what constitutes liquidity -a difficulty, I think,which is inherent in the concept itself. As will be argued below, what appears to be the result ofa preference for liquidity may be explained as the consequence of certain speculative activitiesin which liquidity preference in any positive sense, plays a very small part.



6/28


The second consists of standardised future claims or titles to property, i.e.bonds and shares. It is also obvious that the suitability of the second class forspeculative purposes is much greater than that of the first. Bonds and sharesare perfect objects for speculation; they possess all the necessary attributesto a maximum degree. They are perfectly standardised (one particular shareof a company is just as good as any other); perfectly durable (if the paperthey are written on goes bad it can be easily replaced) ; their value is veryhigh in proportion to bulk (storage cost is zero or a nominal amount); andin addition they (normally) have a yield, which is invariant (in the shortperiod at any rate) with respect to the size of speculative commitments. Hencetheir net carrying cost can never be positive, and in the majority of cases, isnegative.

4. If expectations were quite certain, speculative activity would so adjustthe current price that the difference between expected price and current pricewould be equal to the sum of interest cost and carrying cost. For if the differ-ence is greater than this, it would pay speculators to enlarge their commitments;in the converse case, to reduce it. The interest rate relevant in calculating theinterest cost is always the short term rate of interest, since speculation isessentially a short period commitment.1 The carrying cost, as mentionedbefore, is equal to the sum of storage cost and primary depreciation, 2 minusthe yield.

If expectations are uncertain, the difference between expected price andcurrent price must cover, in addition, a certain risk premium, which will be thegreater (i) the greater the dispersion of expectations around the mean (the lessthe standard probability) ; (ii) the greater the size of commitments. Hence,given the degree of uncertainty, marginal risk premium s an increasing functionof the size of speculative stocks.

If uncertainty is present, the market organisation will develop specialfacilities for divorcing the risk attached to holding stocks from the holders ofthese stocks. This is specially important for those who, though committed, arenot able to deliver before some future date. This facility is supplied by forwardtransactions, and the operation by which the risk premium is divorced fromtotal holding cost, is called hedging.

By selling forward, holders of stocks free themselves of any uncertainty(apart from the risk, which we may treat as negligible, of contracts not beingfulfilled), hence the difference between forward price and current price must beequal to the sum of interest cost and carrying cost, and the forward price mustalways fall short of the expected price (for the same date) by the amount ofthe marginal risk premium.

It may be useful, perhaps, to re-state these relationships in algebraicalform. If we denote interest cost by i, carrying cost by c, and the marginal risk

1. If there is a difference between the speculators' borrowing rate and their lending rate, oneor the other will be relevant, according as the marginal unit of commitment is made with borrowedfunds or not. Hence when speculators reduce their commitments interest costs may fall, and thishas (even if the market rate is unchanged) a similar effect as a fall in carrying cost (rise in yield).

2 The term primary depreciation is Mr. Hawtrey's (Capital and Employment, p. 272), andis used to denote that part of depreciation which inevitably arises through the mere passage oftime.



7/28


premium by r (all these are marginal terms) the current price, expected priceand forward price by CP, EP, and FP respectively, the following relationsmust always be satisfied:

EP-CP=i+c+rFP-CP=i+c, hence FP=EP-r

If speculative stocks are zero, i.e. EP=CP, then -c=i+r, i.e. the negativeof carrying cost must be equal to the sum of interest cost and risk premium,and since i and r are always positive, the carrying cost must be negative, i.e.the yield must exceed the sum of storage cost and primary depreciation by therequired amount. In that case FP=CP-r, the forward price must fall shortof the current price by an amount which Mr. Keynes calls normal back-wardation.1

The above theory of the forward market, which is taken from Mr. Keynes'Treatise on Money, is subject, however, to a certain qualification.2 The proposi-tion that the forward price must fall short of the expected price by the amountof the marginal risk premium, so that if the current price is expected to- remainunchanged, the forward price must be below the spot price, is only necessarilytrue if the hedgers are forward sellers and not forward buyers, and- the

speculators 3 being forward buyers, are not current holders of stock. Thisis probably true in the majority of markets; in the case of certain industrialraw materials, however, where the outside buyers are contractors with givenorders for some period ahead, the hedgers may be predominantly forwardbuyers, and the speculators spot buyers and forward sellers. Now the

carrying cost for these speculators may be higher than the carrying costsfor the market generally. This is because the yield of stocks of raw materials(which in our definition is included in net carrying cost) consists of con-venience, the possibility of making use of them the moment they are wanted,and this convenience is largely lost if the stock held is already sold forward.Hence in markets of this type, there are two carrying costs : (i) those ofordinary holders, which consist of the costs or storage and wastage, minus theyield; (ii) those of forward speculators which consist of costs of storage andwastage only.

Taking this into account, a generalised theory of the forward market mightbe set out as follows. If we write q for the marginal yield and c' for marginalcarrying costs proper, so that c = c'-q,

EP-CP i+c'-q+r

In markets where the hedgers are forward sellers, or where the yieldconsists of a money return which is not affected by forward selling,

FP-CP = i+c'-q, hence FP = EP-r1 Cf. Keynes, Treatise on Money, Vol. II, p. I42-I44. In the terminology of the market there

is backwardation when the forward price falls short of the spot price, and contango whenit is the opposite.

2 I am indebted in this connection to certain criticisms made by Mr. Hawtrey.3 Here I am using the term speculators in a narrower sense than in the rest of the article:it refers only to those who enter the forward market with speculative intent (the counterpart ofthe hedgers ).



8/28


In markets where the hedgers are forward buyers and where the yieldconsists of convenience,

FP-CP = i + c', hence FP = EP-r + q

Hence in such markets the forward price can exceed the expected price bythe amount by which the marginal yield to ordinary holders exceeds themarginal risk premium. When redundant stocks exist (q=O) the forwardprice must be below the expected price, in all cases.'

5. The elasticity of speculative stocks may be defined as the proportionatechange in the amount of speculative stocks held as a result of a given percentagechange in the difference between the current price and the expected price. Thiselasticity will obviously depend on the variations in the terms i, c, and r, whichare associated with a change in speculative stocks, in other words on theelasticity of marginal interest cost, marginal carrying cost, and the marginalrisk premium, with respect to a change in speculative commitments.

Of these three factors the marginal interest cost, as we have seen, may besubject to discontinuous variation if the marginal speculator, in a particularmarket turns from a lender of money into a borrower, or vice versa, but apartfrom this its elasticity is likely to be fairly high, if not infinite.2 The marginalrisk premium is normally rising, and its elasticity probably differs greatlybetween different markets. The more numerous are speculators in a particularmarket, and the more steady the price on the basis of past experience,3 thehigher this elasticity is likely to be. Finally, the marginal carrying cost, as wehave seen, can be assumed to be constant in the case of securities, while it willrise sharply (at any rate over a certain range) in the case of raw materials andprimary products. Hence, taking all factors together, the elasticity of specula-tive stocks is likely to be much higher in the case of long-term securities thanin the case of raw-materials.

The higher the elasticity of speculative stocks, the greater the dependenceof the current price on the expected price. In the limiting case when thiselasticity is infinite, the current price may be said to be entirely determined bythe expected price; changes in the conditions of non-speculative demand orsupply can then have no direct influence on the current price at all (since

1 In the case of bonds and shares, where the yield is a money return which is independentof forward sale, the forward price ought always to be below the expected price. In the forwardtransactions of the London Stock Exchange, however, the yield is credited to the forward buyer(and not to the forward seller, who actually holds the stock), hence the forward price is equal tothe current price plus interest cost and there is a contango, (Since in this case c'=O,FP=EP-r+q= CP?+i) Backwardation can only arise on the Stock Exchange if a fall inprice is expected, and this expected fall, on account of a shortage of stock for immediate deliverywhich is known to be purely temporary, cannot be adequately reflected in the current price (e.g.in the case of transatlantic stocks, when arbitrageurs run out of stock and have to wait for freshsupplies to be sent across the Atlantic). It is always a sign, therefore, of the current price notbeing in equilibrium in relation to the expected price.

2 In certain markets-such as the market for long term bonds-the lending rate is normallyalways relevant, and not the borrowing rate. In this case the elasticity of interest cost can betaken as infinite.

3 In other words, the elasticity of the marginal risk premium is likely to vary inversely withthe amount of the risk premium. When the risk premium is low, its elasticity is also likely to behigh. Cf. ?IO, p.I5 below.



9/28


speculative stocks will immediately be so adjusted as to leave the price un-changed); any change in the current price must be the result of a change inprice-expectations.'

In the opposite limiting case, when the elasticity of speculative stocks iszero, changes in the expected price have no influence upon the current price;the latter is entirely determined by the non-speculative factors.

II. SPECULATION AND PRICE STABILITY6. We can now attempt to analyse the question of the effect of speculation

on price-stability. We have seen in ?2 that the argument that speculation mustexert its influence in a price-stabilising direction in order to be successful pre-supposes that speculative purchases (or sales) make up only a small fractionof total transactions, and does not hold otherwise. But this still leaves thequestion of whether the effect of speculation is price stabilising or destabilisingopen.

We have seen that in all circumstances speculation must have the effectof narrowing the range of fluctuations of the current price relatively to theexpected rice. Hence, if the expected price is taken as given, speculation mustnecessarily exert a stabilising influence: a rise in the current price will befollowed by a fall in speculative stocks, and vice versa.

In order that speculation should be price destabilising, therefore, we mustassume one of two things: either (a) that changes in the current price lead toa change in the expected price by more (in proportion) than the current pricehas changed; or (b) that there are spontaneous changes in the expected pricewhich are speculative in origin and are not justified or not fully justified in themovement of non-speculative factors.

The second of these needs further elucidation. It may be than an impendingchange in data is foreseen by speculators, but exaggerated in importance and,therefore, creates price movements greater than those which would haveoccurred in the absence of speculation. E.g. a good harvest which, in theabsence of speculative activity would have caused a IO per cent reduction inprice, will involve a price movement which is both smoother and smaller inextent, if it is correctly foreseen by speculators. But if speculators foresee a50 per cent price reduction, n consequence of the expected harvest, the resultantprice oscillation will be much greater than if they had not foreseen it at all.It may be also that the speculators expect that a certain event will react,favourably or unfavourably, upon a particular price, though in the absence ofthe expectation no such reaction would have occur-red t all. The day-to-daymovements on the Stock Exchange, where considerable changes in prices occurin accordance with the day's political news, could hardly be accounted for onany other ground but on the attempt of speculators to forecast the psychologyof other speculators. If one is selling this is because he thinks the other woulddo likewise. If speculators combined and formed a monopoly, these price-movements could not take place at all.

1 This does not imply, of course, that the current price must be equal to the expected price;this would only be the case if in addition, the sum of i+c+r were zero. Nor does it imply thatchanges in the non-speculative factors can have no influence upon price at all, for changes inthese factors may influence price-expectations.



10/28


The first of these factors-i.e. the reaction of expectations to changes nthe current price-may be measured by Professor Hicks' concept of the

elasticity of expectations. ' This elasticity is defined as unity when a changein the current price causes an equi-proportionate change in the expected price.Hence, if the elasticity of expectations is positive, but less than unity, specula-tion will still have a stabilising influence, though, of course, a weaker one thanif the elasticity is zero. The case of unity elasticity of expectations is, asProfessor Hicks has said, on the borderline between stability and instability.

Thus the elasticity of expectations and the elasticity of speculative stockstogether determine what may be termed the degree of price-stabilisinginfluence of speculation : the extent to which price-variations, due to outsidecauses, are eliminated by speculation. This may be measured by the pro-portionate change in stocks in response to a given change in the currentprice, since the larger this change, the smaller the extent to which any givenchange in outside factors (a shift in demand or supply) can affect the price.If we denote the degree of price-stabilising influence by S, the elasticity ofspeculative stocks (as above defined) by e, and the elasticity of expectationsby -, their relation is as follows:

S = -e (X-I)Since e cannot be negative, the expression is negative or positive according

as q is greater or less than I.7. It is not possible, however, to express the behaviour of expectations at

any given moment in terms of a single elasticity. For what this elasticity willbe, on a particular day, will depend on the magnitude of the price-change onthat day, the price-history of previous days, and on whether the price expecta-tion refers to next day, next month, or next year. This elasticity is thus likelyto be both large and small, at the same time, according as the price-change hasbeen large or small, and according as the expectation refers to the near futureor the more distant future. It will vary, moreover, with the cause of the price-change. For it is permissible to assume that in most markets speculatorsregard price-changes merely as indicators of certain forces at work, and thatthey attempt to form some idea as to the nature of these forces before adjustingtheir expectations. A given change in price will react differently on specula-tors' expectations according as they regard t as the result of speculative forces,or of outside demand or outside supply and so on.

If any generalisation can be made it is that expectations are likely to beless elastic as regards the more distant future than as regards the near future,and as regards larger changes in price than as regards smaller changes. Thesetwo factors moreover are not independent of each other. For the expectationsas regards the more distant future are likely to be more and more influencedby the speculators' idea as to the normal price -this normal price isdetermined by different factors in different markets, but as we shall see below,it is likely to function in most-and the larger the deviation of the current prtcefrom the normal, the longer must the period be which speculators xpect to elapse

1 Value and Capital, p. 205.



11/28

Io THE REVIEW OF ECONOMIC STUDIES

before he price reverts o normality. Beyond a certain point, therefore (whetherthe word point is understood to be a distance of time in the future or amagnitude of price-change), expectations become insensitive; the elasticity ofexpectations becomes zero or may even become negative.

Speculation, therefore, is much more likely to operate in a destabilisingdirection when we consider price-fluctuations within smaller ranges, thanlarger ranges; and when we consider the movements over a shorter period,than over a longer period. This is so not only because it is the short-periodexpectations which are most elastic (show the strongest reaction to price-changes), but also because it is these short-period expectations which are mostflexible (are most liable to spontaneous changes). Those changes in expecta-tions which are caused by the speculators' own attempts to anticipate eachother's reactions, and thus set up purely spurious price-movements, are essen-tially short-period expectations; and they are responsible for short periodmovements.

Hence to our question: does speculation exert a price-stabilising nfluence,or the opposite ? the most likely answer is that it is neither, or rather that it isboth simultaneously. It is probable that in every market there is a certainrange ofprice-oscillation within which speculation works n a destabilising directionwhile outside that range it has a stabilising effect. Where markets differ, is inthe magnitude of this critical range of price-oscillation. In some markets(among which the market for long-term bonds is conspicuous) this range is innormal circumstances relatively small, the stabilising forces of speculationdominate over the others. In other cases (and here one may count perhaps themarkets for certain classes of ordinary shares) the range is large and thestabilising forces are relatively weak. The explanation of these differences iesin the varying degree of influence exerted by the idea of the normal pricein the different markets. It is to this factor that we must now turn.'

8. In the case of many commodities, and in particular, non-agriculturalraw materials, the elasticity of supply is rather high if the period of adjustmentis allowed for, though supply is inelastic for periods shorter than this period ofadjustment.2 The upper limit of the expected price, for periods longer thanthis period of adjustment, is given by this supply price. The stability, therefore,depends on the general belief that the normal supply price in the future willnot be very different from the normal supply price of the past; it is ultimatelya belief in the stability of money wages.3 It is in this way that the rigidity ofmoney wages contributes to the stability of the economic system, by inducingthe forces of speculation to operate in a much more stabilising fashion than theywould do if money wages were flexible.

If the current price is in excess of the normal supply price, the future date

1 The following owes much to Hicks, Value and Capital, pp. 270-2.2 This period of adjustment varies, I believe, for different commodities from six months to

anything up to two years, though it is likely to be between six and twelve months in the majorityof cases. The case of rubber and tin, where the period is several years, is exceptional. But inthese last two cases, the short-period elasticity (through more or less intensive tapping or

plucking ) is considerable.8 For it is the level of money wages which governs money supply prices if the elasticity of

supply is high.



12/28

SPECULATION AND ECONOMIC STABILITY II

at which price is expected to return to the normal is given by the period ofadjustment. If the current is below the normal supply price, the date at whichthe price is expected to return to the normal is determined by the speculators'expectations as to the period of absorption of excessive stocks. Hence in

markets where the belief in the existence of a normal supply price is strong,and where speculators are in a position to form a reasonable opinion as to thesize of redundant stocks, we should expect speculation to work mainly in astabilising direction.

Such is the case with the main industrial raw-materials; and yet we findthat here price-fluctuations are more violent than in most other markets.' Theexplanation for this, however, lies in the sudden changes to which the yield ofstocks of such raw materials is liable, combined with the low elasticity ofspeculative stocks. Traders' requirements of such stocks, as stated before, area fairly fixed proportion of the expected turnover; and a relatively slightreduction in the expected turnover is sufficient to bring their marginal yielddown to zero. But this necessitates a very sharp reduction in the current price,since net carrying costs, which were negative before, have now become positive.Assuming that previously the price was equal to the normal supply price, sothat the forward price fell short of the current price by the amount of the

normal backwardation; he price must fall far enough for the backwardationto turn into a contango, sufficient to cover interest plus carrying costs. If the

normal backwardation (i.e. the marginal risk premium) was equal to io percent per annum2 and the interest cost plus carrying cost (when the yield iszero) is also equal to I0 per cent per annum, the current price must fall by20 per cent if the excess stocks are expected to be absorbed in one year, and40 per cent if they are expected to be absorbed in two years.3 In the conversecase, where owing to a rise in turnover, stocks fall below the normal proportion,there might be an equally sharp rise in price due to the rapid rise in the yield ofstocks. According to Mr. Keynes4 it is easy to quote cases where the back-wardation amounted to 30 per cent per annum; and this implies that themarginal yield of stocks must have risen to 40 per cent. Hence the explanationfor the wide fluctuations in raw material prices need not be sought in anyinelasticity of supply (except, of course, in the short period), or in destabilisinginfluence of speculative activities, but simply in instability of demand and thelow elasticity of speculative stocks.

9. In the case of agricultural crops, the supply curve is much less elasticand is subject to frequent and unpredictable shifts, due to the weather. It isimpossible to foretell the size of the crop a year ahead, and it is not possible tosay when prices will revert to normal. Hence, in this case, the effect of specula-

I Mr. Keynes stated in the Economic Journal, September, I938, p. 45I, that the differencebetween the highest and lowest price, in the same year, in the case of rubber, cotton, wheat,and lead amounted in the average to 67 per cent over the last ten years.

2 In markets where the range of price-fluctuations is large, the marginal risk premium is alsolikely to be large, thus making the range of fluctuations still larger. io per cent is a normalfigure in the case of seasonal crops; I to 2 per cent may be regarded as the appropriate figure inthe case of long-term government bonds.

3 Keynes, Treatise on Money, Vol. II, p. I44.' Ibid, p. I43.



13/28

I2 THE REVIEW OF ECONOMIC STUDIES

tion on price-stability is much more doubtful. If, nevertheless, the recordedrange of price-fluctuations is no greater than in the case of industrial raw-materials, the explanation might be sought in the fact, as an article in TheEconomist recently suggested,' that these goods are mostly foodstuffs thedemand for which is much more stable than the demand for raw materials.2The fluctuations in the size of stocks are caused by changes in the supply side;and since the Southern Hemisphere became an important source of supply, theextent of the annual fluctuation in world output is much less considerable. Itwould probably be found, however, that in the case of agricultural crops thesame percentage changes in stocks are associated with greater variations ofprices than in the case of industrial raw materials (not because the elasticity ofspeculative stock is less, but the elasticity of expectations is probably greater).

IO. In the market for securities (bonds and shares) there s no such exter-nal determinant of normal price as the producers' supply price in the case ofcommodities. Yet in the market for long-term bonds the notion of a normalprice operates very strongly; only thus can the remarkable stability of thelong-term rate, relatively to the short-term rate, be explained. Since theelasticity of speculative stocks, in this case, is very large, the current price islargely determined by the expected price; if the current price is stable, thismust be because the expected price is insensitive to short-period variations ofthe current price. But how is this expected price determined ?

The simplest explanation which suggests itself is that the expected priceis determined by some average of past prices. The longer the period in thepast which enters into the calculation of this average, the less sensitive is theexpected price to movements in the current price.

This hypothesis, however, taken by itself, is not very satisfactory. In thefirst place, as Professor Robertson has pointed out, it appears to leave thelong-term rate of interest hanging by its own bootstraps. If the currentprice of consols is determined by the expected price, and the expected priceby an average of past prices, how were these past prices determined ? Howdid the rate of interest come to settle at one particular level rather than atsome other level ? In the second place, why should expectations be so inelasticin the long-term bond market in particular ? The fact that the price of manyindustrial shares moves fairly closely with fluctuations in current earnings,does not suggest that this is a characteristic attribute of security markets ingeneral.

These difficulties can be resolved, however, if account is taken of thedependence of the current long-term rate on the expected future short-termrates. A loan for a particular duration, say for two years, can be regarded, asProfessor Hicks has shown,3 as being made up of a spot short-term loan(say for three months) plus a series of forward short-term loans; a short-term loan renewed so many times, and contracted forward. Hence the existenceof a long-term loan market implies the existence of a series of forward markets

I The Economtist, August igth, I939, p. 350.2 Industrial crops, such as cotton, have shown inthe past more violent price-fluctuationthan

e.ither foodstuffs or industrial raw-materials.3 Value and Capital, pp. I45. Cf. also Pigou, Industrial Fluctuations, pp. 230-2.



14/28

SPECULATION AND ECONOMIC STABILITY I3

in short-term loans. And since the forward price of anything, if uncertainty ispresent, must be below the expected price, this implies that the current rateof interest on loans of any particular duration, must be above the average ofexpected future short-term rates, over the same period.

This explains why the yield of bonds with a currency of, say, twentyyears or over is so stable. For it depends on the expected average of short-term rates over the next twenty years and we need not suppose that theelasticity of expectations with respect to the short-term rate is small in orderthat this average should be stable. E.g. if the current short-term rate is halfthe normal rate, and is expected to remain in operation for as long as fiveyears, this will only reduce the average over the next twenty years by one-eighth. Thus the explanation is not that interest-expectations are particularlyinelastic; the expectations regarding the short-term rate would need to beextremely elastic in order that the expected long-term rate should be responsiveto changes in the current rate.'

This also answers the objection that the expectation-theory leaves thestructure of interest rates, current and expected, hanging in the air by its ownbootstraps. For while the current long rate depends on the expected shortrates, the current short rate is not dependent either on the expected shortrates or the expected long rate.2 The latter is not dependent on expectations

1 We can regard the current long-term rate as being determined either by the expected futurelong-term rate, or the average of expected future short-term rates. The two come to the samething since the expected long-term rate also depends on the average of short rates.

If R, is the current yield of a bond repayable in ten years' time and R2 is its expected yieldnext year, r, is the short term rate this year, r2 . . . etc., are the forward short-term rates insubsequent years (i.e. the expected short-term rates plus the risk premium), their relation isgiven by the following equations:

(I + RI)10 _ (I +rl) (i +r2) i* (I+,0)(i+R2)9 = (I+r2)(I+r3) * * * (i+rIO)

Hence,

(i+R2)9=

Thus the expected long-term rate will exceed the current long-term rate if the current short-term rate is below its expected average, and vice versa. (Cf. Hicks, op. cit., p. I52; also Hicks,

Mr. Hawtrey on Bank Rate and the Long-Term Rate of Interest, Manchester School, I939.)2 The current short rate is not dependent on the expected short rates, simply because the life-

time of short-term bills is much too short for expectations to have much influence. The expectedrate on bills next year can have no influence on determining the rate on a three-months' bill to-day;while the expected rates for the next three months are very largely determined by the current rate.It is only in exceptional circumstances that the market expects a definite change in the short-termdiscount rates within the next few months. It is just because the elasticity of expectations forvery short periods is generally so near to unity, that the short-term interest market is largelynon-speculative.

Similarly, a change in the long-term rate (either the current or the expected rate) cannotreact back on the short-term rate except perhaps indirectly by causing a change in the level ofincome and, hence, in the demand for cash. For, supposing the change in the long rate causesspeculators to sell long-term investments, this could only affect the short rate if they substitutedthe holding of cash for the holding of long-term bonds; it cannot affect the short rate if thesubstitution takes place in favour of short-term investments other than cash (savings deposits,etc.). But there is no reason to expect, in normal circumstances at any rate, that the substitutionwill be in favour of cash. Idle balances -i.e. that part of short-term holdings which the ownerdoes not require for transaction purposes-can be kept in forms such as savings deposits, whichoffer the same advantages as cash (as far as the preservation of capital value is concerned) andyield a return in addition. It is only when the short rate is so low that investment in savings



15/28


at all (or only to a very minor extent) but only on the current demand forcash balances (for transaction purposes) and the current supply. And sincethe elasticity of supply of cash with respect to the short-term rate, is normallymuch larger than the elasticity of demand, the current short-term rate can betreated simply as a datum, determined by the policy of the central bank.'

How is this related to Mr. Keynes' theory of the long-term rate of interestbeing determined by liquidity-preference ? It leads to much the same result(i.e. that the rate, in the short period, is not determined by savings and invest-ment) but the route by which it is reached is rather different. The insensitive-ness of the long-term rate to outside influences (i.e. the supply and demandfor savings ) is not due to any liquidity premium attached to money orshort-term bills; in fact, the notion of a liquidity premium appears entirelyabsent. It may be objected that it is merely replaced by the notion of a marginalrisk premium, and that Mr. Keynes' liquidity premium on the holding ofshort-term assets is merely the negative of our marginal risk premium on theholding of long-term bonds. But in this case, the peculiar behaviour of thelong-term interest rate is certainly not explained by the existence of this risk-premium.2 For let us suppose that subjective expectations are quite certain,so that this risk premium is completely absent. In this case the current

deposits is no longer considered worth while (see footnote below) that there can be a net substitu-tion in favour of cash; but precisely in those circumstances the elasticity of substitution betweencash and savings deposits is likely to be so high that this cannot have any appreciable effect onthe short-term rate.

Thus, while the current short rate does determine the relation between the current long

rate and the expected long rate, this is not true the other way round.1 The nature of the equilibrium in the short-term interestmarket is shown in the accompanying diagram, wherethequantityof cash is measured along Ox, the short-term interest rate along y DOy. DD and SS stand for the demand and supply of money,respectively. The demand curve is drawn on the assumptionthat the volume of money transactions (i.e. the level of income)is given. This demand curve is inelastic, since the marginal yieldof money declines fairly rapidly with an increase in the proportion \of the stock to turnover. Below a certain point (g) the demandcurve becomes elastic, however, since the holding of short-termassets tother than money is always connected with some risk q(and, perhaps, inconvenience), and individuals will not invest qshort term if the short-term rate is lower than the necessary \compensation for this. There is a certain minimum, therefore,below which the short-term rate cannot fall, though thisminimum might be very low. (The dotted line shows Xwhat the demand curve would be if this risk were entirelyabsent.) Hence, when the short-term rate is very low, it can be said to be determined by therisk premium attaching to the holding of the safest short-term asset; otherwise it is determinedby the supply-price of money (i.e. banking policy) and changes in the short-term rate are bestregarded as due to shifts in this supply price. (The elasticity of the supply of money in a modembankingsystemis ensured partlyby the open market operations of the central bank, partly by thecommercial banks not holding to a strict reserve ratio in the face of fluctuation in the demand forloans, and partly it is a consequence of the fact that under present banking practices a switch-overfrom current deposits to savings deposits automatically reduces the amount of deposit money inexistence, and vice versa.)

2 Moreover, the long-term rate is never equal to this risk premium (or liquidity premium);this only accounts for the difference between the long-term rate and the expected average ofshort-term rates. Professor Hicks has calculated this risk premium to have been about I per centin Great Britain before the last war, and 2 per cent after the war (Manchester School, Vol. X,No. I, p. 3I).



16/28


long-term rate, instead of being above the average of expected short-termrates, will be equal to that average. But the elasticity of demand for idlebalances with respect to the long-term interest rate, in the sense in whichMr. Keynes employs this concept, would certainly not be removed thereby;

on the contrary, it would be rendered infinite. For in the absence of un-certainty, when the marginal risk premium is zero, the elasticity of speculativestocks in the bond market would be infinite (since the other factors determiningthe elasticity of speculative stocks, i.e. interest costs and carrying costs, are,in this case, also infinitely elastic). The current price of bonds would be entirelygoverned by the expected price; if we assume the expectations concerningfuture interest rates as given, the current long-term rate would be uniquelydetermined.' Changes in the rate of savings or in the marginal efficiency ofcapital could not affect the long-term rate at all, except through the slow anddubious channel of changing the level of incomes, and, hence, the demand for

cash; this reacting on the short-term rate of interest, and this, in turn, graduallyaffecting the interest expectations for the future.

It is therefore not so much the uncertainty conceming future interestrates as the inelasticity of interest expectations which is responsible forMr. Keynes' liquidity preference function, i.e. which causes the demandfor short-term funds to be elastic with respect to the long-term rate.2 Theuncertainty of expectations works rather in the opposite direction. For whenuncertainty is high not only is the amount of the marginal risk premium high,but the elasticity of the marginal risk premium curve (and hence the elasticityof speculative stocks) is low; speculators are much less willing to extend theircommitments or to reduce them. It is in such periods that variations in theoutside demand or supply can be expected to have the most marked effecton bond prices.3

Nor is it quite accurate to represent the forces determining the long-termrate as the demand for liquid funds (money + bills) vis-a-vis the supply ofsuch funds. We have seen that the proposition that the rate of interest isthe agency which equates the supply and demand for money is true if under

rate of interest is meant the short rate of interest, and under moneythe sum total of things which mainly serve as a means of payment (i.e. banknotes and current deposits).4 But in the case of the long-term rate, we are notreally confronted with a demand curve for liquid funds but with a supply curve

1Toassumethatsubjectiveexpectations are certain is not the same thing, of course, as assumingperfect foresight or assuming that the expected prices are the same as current prices. Suppose

e.g. that, despite the current short-term rate being 2 per cent, speculators are absolutely certainthat a year hence, and forever afterwards, the rate will be 4 per cent. The existence of this expecta-tion will not cause any change in the current short-term rate; but it will invariably fix the levelof the current long-term rate at 3.92 per cent.

2 Mr. Keynes appears to be aware of this, when he speaks of the influence of the notion ofthe safe rate of interest in determining the current rate. (General Theory, p. 201.) But inthe general statement on the liquidity preference function he relies on the factor of uncertaintyof which money is free and, hence, commands a premium.

8 It is well known that in periods of great political uncertainty it becomes impossible eitherto buy or to sell gilt-edged in larger amounts without thereby directly affecting the price.

4 The proposition is true, partly because both the demand and the supply of money varieswith the rate of interest and also because any factor influencing the short-term rate can do so onlyby affecting the demand or the supply of cash. As mentioned before, emphasis should be placed



17/28

i6 THE REVIEW OF ECONOMIC STUDIES

of long-term assets by speculators; 1 and the two methods of representationdo not always yield the same conclusions. It is not clear, for example, whyan increase in the aggregate amount of liquid funds should have any directeffect on the speculators' willingness to hold long-term assets, other thanthrough the change in the short-term rate. An increase in the supply of moneywill directly reduce the short-term rate (unless t is already so low that it cannotfall further) ; this will have a minor effect on the long-term rate (for the currentprice of bonds will now be higher relatively to the expected price); a majoreffect will only come about slowly as expectations are affected.2

To sum up. In the market for long-term bonds, the elasticity of expecta-tions is normally small, and the elasticity of speculative stocks is large, bothabsolutely and relatively to the elasticity of the non-speculative demand andsupply (i.e. the elasticity of supply of savings, and the elasticity of the producers'demand for long-term funds, as a function of the rate interest). Hence theprice in the short period is largely determined by speculative influences. Itwould not be correct to say that these outside factors have no influence uponprice ;3 only that their influence is largely overshadowed by speculative forces.And for reasons given in the next section, the inability of these forces toinfluence the price in the short period will much weaken their price-determininginfluence in the long period as well.4

ii. There remains to examine the situation in the market for shares;and after what has been said above, it should not be difficult to account forthe apparent contrast between the stability in the gilt-edged market and theinstability of the share market. Shares are also subject to the same influencesas bond prices; changes in the expectations regarding future interest ratesshould affect the one just as much as the other. But in addition they are also

on the elasticity of the supply of money (with respect to the short-term rate) rather than on theelasticity of demand. The latter, except when the rate is very low, is relatively small: which isshown by the fact that the short-term rate can be varied within fairly wide limits without thesechanges being associated by equally significant variations in the quantity of money in existenceor in the level of incomes.

1 If the price of bonds is regarded as determined by the supply and demand for liquid funds,the price of any other thing which is subject to the price stabilising influence of speculation couldbe equally regarded as such. In all these cases it is the speculators' willingness to undertakecommitments in a particular direction, and not their preference for avoiding commitments ingeneral (i.e. their preference for liquidity ) which is relevant.

The liquidity premium on money and savings deposits, etc., in the sense in whichMr. Keynes employs this concept, may be regarded as z per cent per annum if the market forConsols is chosen as the standard with reference to which this liquidity premium is measured.But it will be IO per cent per annum if, say, the market for wheat is chosen as a standard, andsomething else again if the tin market is chosen. But is there any particular reason for selectingthe gilt-edged market as the standard ?

2 It might be argued that an increase in the proportion of short-term assets in the total ofassets will reduce the marginal risk premium on long-term assets and, hence, lower their yield.But it is doubtful whether this effect could be appreciable, unless the change in this proportion islarge; and since short-term assets amount in any case only to a small proportion of total wealth,a large change in their amount can only cause a small change in this proportion. In the case ofbanks and other financial institutions, however, which keep a large proportion of their assets onshort term, the effect may be considerable.

8 Here I am not thinking of the day-to-day fluctuations in gilt-edged prices which are, ofcourse, also due to speculative influences-they either reflect genuine variations in the degreeof certainty with which expectations are held, or the expectation of such variations on behalfof others-but changes which sometimes follow from the pressure of new issues.

4 Cf. S?I5-I6 below.



18/28

SPECULATION AND ECONOMIC STABILITY x7

subject to changes in the expectations concerning the expected level of profits.And here experience suggests that the elasticity of expectations concerningthe future level of profits is fairly high; share prices correlate fairly well withfluctuations in current earnings.

There are several explanations for this. In the first place the course ofprofits of any particular company is likely to exhibit a trend, in addition tofluctuations; and this trend may make it impossible for the expectation ofa long-run normal level to develop such as is formed in the case of the short-term interest rate where there is no such trend. In the second place, owing tothe absence of any means of information concerning the more distant future,the profits relating to the more distant periods are likely to be heavily dis-counted for uncertainty; thus the profits relating to the near future are moreheavily weighted than would follow from the mere fact of discounting atcompound interest. And, lastly, just because the short-period price fluctuationsare large, investors are likely to concentrate attention on the short-periodexpectations concerning capital value, rather than the long-period prospects ofenterprise; speculative motives get inseparably mixed up with other motivesand it is no longer possible to distinguish between speculative demand andnon-speculative demand. It is in such circumstances that the destabilisinginfluence of speculation will dominate over a wide range of price oscillation.

III. SPECULATION AND INCOME-STABILITY

i2. There remains, finally, to examine the effects of speculation on thestability of the general level of economic activity. Speculation affects the levelof activity through the variations in the size of speculative stocks held. Wehave seen above that if there is speculation, any change in the conditions of theoutside demand or supply will be followed by a change in the amount of stockheld; if speculation is price-stabilising (the elasticity of expectations less thanunity), a reduction in demand or an increase in supply will be followed by anincrease n stocks; if speculation s price-destabilising elasticity of expectationsgreater than unity) a reduction in demand or increase in supply will be followedby a decrease in stocks, and vice versa.

An increase in speculative stocks of any commodity implies an increaseof investment in that commodity, or, to use Mr. Hawtrey's expression, a

release of cash'> by the market. (Conversely, a reduction of stocks impliesan absorption of cash.) Unless this increase is simultaneously compensatedby a decrease in the amount of stocks held in other commodities, it must implya corresponding change in the level of investment for the system as a whole.Such compensation can take place in two ways. (i) If there are several goodswhose markets are speculative, an increase in speculative stocks in one marketmay, in certain cases, which we shall analyse below, involve a consequentialdecrease n speculative stocks in other markets. (ii) If the monetary authoritiesraise the short-term rate of interest they can induce a reduction in stocks inthe system as a whole. Without making some assumptions about monetarypolicy, it is impossible, therefore, to make any generalisations about theeffects of changes in the amount of speculative stocks. In the present section

2 Vol. 7



19/28

i8 THE REVIEW OF ECONOMIC STUDIES

we shall assume that the monetary authorities maintain the short-term rateof interest constant. In the last section we shall deal with the question of howfar the instabilities due to speculation can be counteracted if a different mone-tary policy is adopted.

I3. Before proceeding further we must introduce the distinction betweenincome goods and capital goods. There are certain categories of goods

in the case of which it can be assumed that spontaneous hanges in the amountof money spent on them (per unit period) will be associated with equivalentand opposite changes in the amount of money spent on other goods. There areother categories of goods for which this is not (or not necessarily) the case.The first category may be said to consist of goods which are ordinarily bought

out of income, the second category, of goods bought out of capital(i.e. whose purchase is charged to a capital account, and not to an incomeaccount). For under the assumption that all the income of individuals is spentone way or another 1and the level of income is given 2 change in the amountof money spent on anything must be associated by an opposite change in theamount of money spent on something else. In the case of goods purchased ona capital account this is not so, for variations in the amount of money spenton them may represent variations in the total volume of borrowing. Thus ourdistinction between income goods and capital goods comes close tothe usual distinction between consumption goods and capital goods,though we must remember that in our case, it is not the character or thedestination of the goods which is at the basis of the distinction but simplywhether spontaneous changes in the amount of money spent on them implysimultaneous changes in the total amount of expenditure or not.

I4. In the case of income-goods a change in speculative stocks, whichhas a stabilising effect on price must exert a destabilising influence on thelevel of economic activity, irrespective of whether the change was due to achange in the conditions of demand or supply. For an increase in the demand(a shift in the outside demand curve to the right) by causing a reduction inspeculative stocks, also entails a reduction in the level of incomes, since theincrease in the demand is associated with a reduction in the amount spent,and thus of incomes earned, in the production of other things, without anycompensating increase in incomes earned in the particular commodity inquestion. Conversely, a reduction in demand will involve an increase in thelevel of incomes, for the same reason. An increase in supply on the other hand(a shift in the outside supply curve to the right) will involve an increase inspeculative stocks and thus an increase in the level of incomes; a reductionin supply a reduction in incomes. None of these changes in incomes couldhave taken place in the absence of speculation; and the extent of the changein incomes, following upon a given shift in supply or demand, will be the

1 Income saved is also spent on something in so far as it is spent on the purchase of securitiesor other income-yielding assets. What we are assuming here, therefore, is that there is no attemptto hoard ; that changes in the disposition of incomes between the different lines of expendituredo not involve changes in the demand for money.

2 This is the meaning of the word spontaneous in the above definition. If there is a changein the amount spent as a result of a change in the level of incomes, there need not be, of course,any reduction in the amount spent on other things.



20/28


greater the larger the price-stabilising influence of speculation as abovedefined.'

We can conclude, therefore, that in the case of income-goods, an increasein demand or a reduction of supply will cause a reduction in incomes, andvice versa, the magnitude of which depends, inter alia, on the degree ofprice-stabilising influence. When this is negative (speculation is price-destabilising) a reduction in demand or an increase in supply will cause areduction in incomes, and an increase in demand or a reduction in supplyan increase in incomes. Only when the degree of price-stabilising influenceis zero will a shift in demand or supply leave the level of incomes unchanged.

I5. It might be objected that this effect of speculative activity on incomeswill be purely temporary; for while a discrepancy between outside demandand outside supply lasts, speculative stocks must increase (or decrease, asthe case may be) continuously and this process cannot go on for ever. Thereis no market which has either resources or the inclination to go on absorbingcash, or releasing cash indefinitely; sooner or later, the elasticity ofspeculative stocks, and, hence, the degree of price-stabilising influence, mustbecome zero. This is true, but it does not follow from this that either theprice-stabilising or the income-raising (or income-lowering) effect of speculationis purely temporary. For in certain cases the change in incomes itself providesa mechanism which brings the discrepancy between demand and supplygradually to an end. For commodities whose income-elasticity of demand isgreater than zero (or to use another terminology, where the marginal propensityto consume is positive) an increase in incomes will involve an expansion indemand, and vice versa. And since we have seen that if the price-stabilisinginfluence of speculation is positive, an excess of supply over demand must leadto an increase in incomes, and an excess of demand over supply to a reductionin incomes, in both cases the discrepancy between demand and supply will tendto get adjusted through the variations in demand caused by the changein incomes.2

This is nothing else but the well-known doctrine of the Multiplier putforward in a generalised form. The latter, strictly speaking, assumes that thedegree of price-stabilising nfluence is infinite, so that the price can be regardedas constant. In that case the increase in incomes created by a given increasein supply (or a given reduction in demand) will be (ultimately) as many timesthe value of the original increase in supply (or reduction in demand) as thereciprocal of the proportion of marginal income spent on the commodity inquestion. This is so because so long as the excess of supply over demandcontinues, stocks are accumulated and entrepreneur's receipts continue to rise.When income has risen in the ratio stated above, the discrepancy betweendemand and supply disappears (owing to the adjustment in demand) and theaccumulation of stocks ceases.

The doctrine of the Multiplier could therefore be restated as follows. The

ICf. p. 9 above.2These equalising forces operate if (a) S (the price-stabilising nfluence) is positive and (b) the

income-elasticity of demand is positive. If the income-elasticity of demand is negative, theequalising forces operate if S is negative.



21/28

2o THE REVIEW OF ECONOMIC STUDIES

demand and supply of anything always tends towards equality. In normalcases this equality is secured by adjustments in the price of that commodity.But if changes in the price of a particular commodity are attended by variationsin the amount of stocks held, the equality may be secured-to a greater or lesserextent-by adjustments in the level of incomes. Onlyif thestabilisinginfluenceof speculation is infinite and the income-elasticity of demand greater thanzero can the adjustment come about entirely through a change in incomes.In general, following upon a change in the demand or the supply schedule, themechanism of adjustment is all the more likely to operate through a change inincomes, rather than a change in price: (i) the greater the degree of price-stabilising influence of speculation; (ii) the greater the proportion of marginalincome spent on the good; (iii) the less is the elasticity of the (outside) demandand supply schedules.

i6. This doctrine is subject, however, to an important qualification. Ifthere is more than one good whose price is stabilised, or quasi-stabilised,through speculative activities, the mechanism of the demand-and-supplyequalisation through a change in incomes will not operate in the same way as inthe case of only a single good. Suppose that out of a large number of thingson which consumers spend their incomes, coffee is the sole good whose priceis stabilised by compensatory stock-variation. In that case, whenever there is,say, an increase in the supply of coffee, stocks of coffee will increase, incomeswill rise, and this process will go on until the demand for coffee, in consequence,has risen in the same proportion n which the supply has increased. But if theprice of sugar is also subject to the same influences, and consumers spend partof their increased incomes on that commodity, then whenever there is anincrease in the stocks of coffee and a consequent increase in incomes there willalso be a decrease in the stocks of sugar, which will act as a brake on the risein incomes. Irn his case the level of incomes can never rise sufficiently (as aconsequence of the increase in the supply of coffee) to adjust demand fullyto the increase in supply. Hence, the existence of more than one multiplierweakens the operation of any one of them.

In the real world the most important case of income goods whoseprices are stabilised through speculation are long-term bonds dealt in thecapital market. On our definition the things bought and sold in the capitalmarket are not capital goods, but income goods, for the outside ornon-speculative demand for securities consists of savings, savings are paidout of income, a change in the rate of savings out of a given income must implyan opposite change in the amount spent on other income goods. And sinceboth the proportion of marginal income going to savings and the elasticity ofspeculative stocks for bonds is large, it is in the case of the long-term investmentmarket that one would expect disequilibria due to changes in supply or demand(i.e. in the propensity to save or in the investment-demand schedule) to beadjusted through variations in the level of incomes, rather than in the long-term rate of interest.

This, however, cannot be entirely true, except for a closed system. Inthe case of an individual country which is on the Gold Standard, or where theprice of foreign exchange is stabilised through the operation of an Exchange



22/28

SPECULATION AND ECONOMIC STABILITY 2I

Equalisation Fund, the price of foreign exchange, and, hence, the price ofimported goods, behave in much the same manner as the prices of goodsstabilised by speculation. Hence, in addition to the savings-investmentmultiplier there exists a foreign trade multiplier, and the operation ofthe latter must weaken the price-stabilising forces operating in the former(and vice versa).

This can be best elucidated by an example. Let us suppose that for thecommunity as a whole, 25 per cent of additional income is saved, and that thereis an increase in the rate of long-term investment (financed through the saleof securities) by L' million per week. This will immediately increase incomesin the investment good industries by LI million and savings by ?250,000.Thus, while in the first week the investment market was required tofurnish the whole of the additional expenditure of L' million out of its specula-tive funds, in the second week it only has to furnish three-quarters of thatamount; one quarter will be furnished through the additional savings.' Ifwe suppose that all the income which is not saved is spent on home-producedgoods, there will be a further increase in incomes by 750,000 in the second week,and 562,500 in the third week, and so on.2 Similarly, the outside demand forsecurities will expand by I87,500 in the third week, I40,625 in the fourth week,and so on. As a result, after a certain number of weeks, total incomes willhave expanded by ?4 millions per week, and total savings by L' million; theoutside demand for securities will ultimately have increased in the same rateas the outside supply. The size of speculative commitments has increased, ofcourse, during this process, but this increase will come to a halt once the increasein outside demand has caught up with the increase in the rate of supply;the contribution which the speculative resources of the market have to makeis limited. Provided that the total required increase in the size of speculativestocks is not too large relatively to the resources of the market (i.e. providedit does not impair the degree of price-stabilising influence) there will be nopressure on the price of securities (i.e. no tendency for the rate of interest torise) either in the long run, or in the short run.

Let us suppose now, however, that not three-quarters, but only a half ofthe marginal income is spent on home-produced goods; and while 25 per centis saved, another 25 per cent is spent on additional imports. In that case the

multiplier will not be 4, but only 2 ; the ultimate increase in incomes,following upon a Li million increase in the rate of investment, will be ?2millions (per week) and of savings ?500,000. Hence, even after incomes havebeen fully adjusted to the change in the level of investment the rate of outsidedemand for securities will have only increased by one-half of the increase inthe rate of supply; if the rate of interest is to remain unchanged, speculatorswill have to furnish ?500,000 per week, indefinitely.

It is true that in this case the increase in speculative stocks in the securitiesmarket is attended by a decrease in stocks (of gold and foreign exchange) in

1 On the assumption, of course, that all increase in genuine savings is directed at thepurchase of long-term assets.

2 Abstracting from any involuntary reduction in stocks in the hands of retailers and whole-salers, which is purely temporary; this will merely delay adjustment, not prevent it.



23/28


the foreign exchange market; while in the first market speculators continuouslyborrow (on short term) in the second market they will continuously lend;and since the one must be numerically equal to the other, there can be notendency for the short-term interest rate to change.' In Keynesian terminologyone might also say that Investment and Savings will be e,qual; for if the exportsurplus is added to Investment, the net increase in the rate of Investment willnot be ?I,ooo,ooo, but only ?500,000. Yet the situation wil obviously not beone of equilibrium; it could not be indefinitely maintained. As the stocks ofgold and foreign exchange continuously decrease, and the speculative commit-ments in the long-term investment market increase, one of two things musthappen: either the price of gold (and foreign exchange) must rise, thus raisingthe prices of imported goods and thereby the home-investment multiplier, orthe price of securities must fall (the long-term rate of interest rise) and thusthe level of home investment will be reduced. Whether equilibrium is finallyreached through a rise in the price of foreign exchange, or of the rate of interest,will depend on the relative elasticity of speculative stocks in the two markets(or rather on the range over which in the two markets speculative stocks areelastic). Assuming that it takes place through the second and not the first,it would be wrong to attribute the rise in the interest rate to any shortage ofcash. In the situation contemplated there is no shortage of cash, and no pressureon the short-term rate of interest. The long-term rate rises relatively to theshort-term rate simply because, owing to a shortage of savings, speculatorsare required to expand continuously the size of their commitments: and thereare limits to the extent to which this is possible.

It is not quite accurate, therefore, when Mr. Keynes says that theinvestment market can be congested on account of a shortage of cash. It cannever be congested on account of a shortage of savings. 2 This propositionwould only hold if it were assumed, in addition to the degree of price-stabilisinginfluence being infinite, that the long-term investment market is the onlymarket which is subject to such speculative influences; in other words, thatdC (the marginal propensity to consume) which is relevant for calculating the

home-investment multiplier, is necessarily equal to i-dS (where S is the

mdY nbtdYmarginal propensity to save). This, as we have seen, need not be the case.3

1 There will be an initial increase in the demand for short-term funds with the increase inincomes; but provided this demand is satisfied by the banking system, no further contributionis required to keep the short-term rate stable. The continuous increase in the short-term indebted-ness of the speculators in securities will be exactly offset by the decrease in the short-termborrowings of the foreign exchange market.

2 The Ex-Ante Theory of the Rate of Interest, Economic Journal, December, I937,p. 669.

3 Mr. Keynes' own proof of this proposition is that there must always be exactly enoughex-post saving to take up the ex-post investment and so release the finance which the latter hadbeen previously employing (ibid., p. 669). Ex-post investment and ex-post saving will alwaysbe equal if ex-post investment is so defined as to include consequential changes in stocks (inour example, foreign exchange balances). But in order that the funds released through the reduc-tion in stocks should be available for the finance of long-term investment somebody must perform



24/28


Mr. Keynes' General Theory, therefore, in so far as it concerns his theoryof the rate of interest and the theory of the multiplier, is rather in the nature ofa special case ; strictly it only holds if S is infinite in the case of long-termbonds and zero in the case of everything else. But if we abstract from the caseof foreign exchange under a gold standard or quasi-gold-standard regime, it istrue that the degree of price-stabilising nfluence, though not perhaps infinite,is very much larger in the case of long-term bonds than for any other com-modity; and this means that the Keynesian theory, though a special case,gives, nevertheless, a fair approximation to reality.

I7. In the case of capital goods, a change in speculative stocks whichhas a stabilising effect on price will also have a stabilisinginfluence on activity,if the change in stocks was due to a shift in demand, and a destabilisinginfluence, if it was due to a change in supply. This difference s due to the factthat changes in the demand for capital goods are not (normally) associatedwith opposite changes in the demand for other goods; they represent variationsin the investment-demand schedule (in the marginal efficiency of capital) andsuch variations, in a regime where the long-term interest rate is itself subjectto the price-stabilising influence of speculation, and/or where the short-termrate of interest is held constant, must imply variations in the total scale ofexpenditures. Hence, an increase in the demand for capital goods involves anincrease in the level of incomes, when S is zero (in the absence of speculation);in so far as S is positive (the increase in demand is associated with a reductionin stocks), the increase in incomes will be less than it would be otherwise. Thus,the existence of speculation in capital goods, if price-stabilising, damps downthe multiplier and hence the range of income fluctuations, in much thesame way as the existence of a stabilised price for for

Date post:	02-Jun-2018
Category:	Documents
Upload:	jpkoning
View:	245 times
Download:	0 times

Kaldor - 1939 - Speculation and Economic Stability

Documents