7/27/2019 1816058.pdf

1/11

7/27/2019 1816058.pdf

2/11

PERSPECTIVES N TWENTIETHCENTURYECONOMICTHEORY*C a p i t a l Controversies: n c i e n t

a n d M o d e r nBy JOHNHICKS**

In what must by the rules be a fairlyshort paper, I cannot make a survey of allcontroversies about capital, from (say)Ricardo-Malthus to Joan Robinson-Solow.All I can attempt is something much moremodest. What I propose to do is to takeone particular point, which has figured (asI shall show) in many such controversies,and to use it as a means of pulling a partof the story together. It is a matter whichwould seem to be appropriate for discus-sion at this joint meeting, for it is an in-teresting illustration of the way in whichthe history of his subject can be of use tothe modern economist.Economists do indeed have a special usefor the history of economics, somethingmore than the general use that can bemade of their own history by students ofother subjects such as mathematics andthe natural sciences. The history of science,certainly, is no mere antiquarianism; one islearning science when one learns in whatways scientific discoveries have been made.The history of economics has that use, andit has other uses. It has, of course, a purehistorical use; the greatest economists,Smith or Marx or Keynes, have changedthe course of history; they are as worthythe attention of the pure historian asLouis Napoleon or Woodrow Wilson. But

this again is not the economist's use. Thatis something different.Economics is a social science, and a par-

ticular kind of social science, in that it isconcerned with the rational actions, thecalculated actions, of human beings, andwith their consequences. This has the re-sult that those whom we study can hearwhat we say. We may speak to each otherin our private languages, but private con-versations are no more than goods inprocess; while we speak only to each otherwe have not finished our job. The ideas ofeconomics, the powerful ideas of economics,come from the market-place, the "realworld," and to the "real world" they goback. So there is a dialogue between econo-mists and their subject-matter. It is adialogue in which there are important in-termediaries; statisticians are one kind ofintermediary, journalists another, account-ants (as we shall see) another; the econo-mist-statistician and the economist-journa-list do much of the intermediation them-selves. In the course of the dialogue ideasacquire associations; they cease to be freeideas, which can be defined at choice. It isnot in our power to say with Humpty-Dumpty, "When I use a word it meansjust what I choose it to mean"; we cannotescape the associations.I do not mean that there is not suchdialogue, and such associations, in the caseof other social sciences. Clearly there is; inpolitical science, say, as much, if not more,

* The paper given at this session by Joseph J. Speng-ler of Duke University will appear in the Journal ofEconomicIssutes.** All Souls College, Oxford.307

7/27/2019 1816058.pdf

3/11

308 AMERICAN ECONOMIC ASSOCIATION MAY 1974than in economics. Political ideas are in-deed so rich in associations that the studyof politics seems sometimes to consist oflittle else. Economics has (relatively) muchmore that is positive to offer, but we shouldnot allow our passion for quantification toblind us to the fact that economic ideasshare this characteristic of political ideasand do so for the same reason.

We cannot escape the associations, butwe can try to understand them, so as to bemasters of them. That is what, in my view,the history of economics is for. It is whatit is for, for the economist. We need toknow the history of our concepts in orderto know what it is that we are handling.The history of economics, so understood,cannot be discovered by poring over oldtextbooks, even old "classics." That is nomore than a part of what has to be done.The books must be read against theirbackground, the events which promptedthe analysis, and what happened to theanalysis when it went out into the world.All that is part of the tradition which wehave inherited, and from which, if we areto do our job, we cannot escape.I turn, I hope in that spirit, to my par-ticular topic.

One must begin with a distinction whichhas been fully understood only in quite re-cent years. Suppose we start by saying (asmany would say) that the capital of aneconomy is its stock of real goods, withpower of producing further goods (orutilities) in the future-the stock of suchgoods existing in the economy at a momentin time. That, in strictness, is no more thana list-what in English, but not in Ameri-can, we would call an inventory. How dowe aggregate it, as for macroeconomic pur-poses we have to do? We cannot aggregateexcept by adding money values; how do wedeflate that money aggregate, so as to geta measure of Real Capital? There are, inprinciple, two main ways.The first, which on the analogy of other

aggregates, such as consumption, mayseem the more natural, is to deflate by anindex of the prices of the capital goodsthemselves. (It may not be easy to find asuitable index, but that is not here myconcern.) It is, theoretically, a possiblemeasure; to distinguish it from the other,to which I shall be coming, I call it Volumeof Capital. It has the property, it will benoticed, that as between two economieswhich have capital stocks that are physi-cally identical, Volume of Capital must bethe same.

It may well be objected, and has oftenbeen objected, that Volume of Capitalmisses the essential fact about capital--that the utilities of capital goods are in-direct. The values of capital goods arederived values, capitalized values of futurenet products. If these future products arevalued at current prices of products, theresultant capital values should be betterindicators of the true values of the capitalgoods than their actual market values,which are influenced by expectations ofchanging values of products. These "cor-rected" capital values could be aggregated;but since they have been built up fromproduct prices, not capital good prices,they can only be deflated by an index ofproduct prices, if they are to be used as ameasure of Real Capital. Real Capital, inthis sense, does not have the invarianceproperty; it may be changed, without anychange in the physical goods, by the mereadmission of new information. I call thisother measure Value of Capital.

When the distinction is expressed in thisstatistical (or quasi-statistical) manner, itwould seem that we need have no difficultyin living with it. We can keep both in play,using one for some purposes, the other forothers. We may nevertheless have somedifficulty in explaining ourselves. Both, asdescribed, are measures of Real Capital;but what is Real Capital? We cannot saythat the two measures are two measures

7/27/2019 1816058.pdf

4/11

VOL. 64 NO. 2 TWENTIETH CENTURY ECONOMIC THEORY 309

of the same thing-as one may say thatone buys a pound of somethilng, or a dol-lar's worth of it, when the price of thething is a dollar a pound. The Real Capitalthat is being measured is different.If it is capital in the volume sense thatis being measured, capital is physical goods;but in the value sense capital is not physi-cal goods. It is a sum of values which mayconveniently be described as a Fund. AFund that may be embodied in physicalgoods in different ways. There are thesetwo senses of Real Capital which need tobe distinguished.I do of course borrow the term Fundfrom the history, and to the history I nowturn. I am going to maintain that the dis-tinction is quite ancient; it divides econo-mists, ancient and modern, into two camps.There are some for whom Real Capital is aFund I shall call them Fundists; andthere are some for whom it consists ofphysical goods. It is tempting to call thelatter Realists;1 but since one wants toemphasize that both concepts are real, thisis not satisfactory. I shall venture in thispaper to call them Materialists. (Material-ists, I mean in the sense of Dr. Johnson'srefutation of Berkeley's idealism-"strik-ing his foot with mighty force against alarge stone, till he rebounded from it, Irefute it thus." There will be some at leastof my Materialists who are worthy fol-lowers of Dr. Johnson.)One of these was Edwin Cannan, theteacher of Lionel Robbins and the founderof the economics school of the LondonSchool of Economics and Political Science.A beautiful illustration of the oppositionwith which I am here concerned is to befound in Cannan's comments on capital inAdam Smith.2 Cannan was convinced thatSmith was in a muddle. I do not think that

he was in a muddle; he was simply athoroughgoing Fundist, and to the Ma-terialist Cannan the Fundist position wasquite incomprehensible. Cannan: "Thecapital [in Smith] is often spoken of as ifit were something other than the goodsthemselves." That is just the point.Not only Adam Smith, but all (or nearlyall) of the British Classical Economistswere Fundists; so was Marx (how elseshould he have invented "Capitalism"?);so was Jevons. It was after 1870 that therewas a Materialist Revolution. It is notthe same as the Marginalist Revolution;for some of the Marginalists, such asJevons and Bohm-Bawerk, kept the Fund-ist flag flying. But most economists, inEngland and in America, went Material-ist.' Materialism, indeed, is characteristicof what is nowadays reckoned to be the"neoclassical" position. Not only Cannan,but Marshall and Pigou, and J. B. Clark,were clearly materialists. Anyone, indeed,who uses a Production Function, in whichProduct is shown as a function of labor,capital and technology, supposed separa-ble, confesses himself to be (at least whilehe is using it) a Materialist.

What about Keynes? Keynes, of course,was brought up as a Materialist, and thereare no more than slight signs, in the Gen-eral Theory, that he had departed from theMaterialist position. So it is perfectlypossible to be a Keynesian and yet to be aMaterialist. But the rethinking of capitaltheory and of growth theory, which fol-lowed from Keynes, and from Harrodon Keynes, led to a revival of Fundism. Ifthe Production Function is a hallmark ofMaterialism, the capital-output ratio is ahallmark of modern Fundism. That, in thebriefest of outline, is the story; how it hap-pened I shall now attempt to explain.

1 As I have done myself in a short passage in miCapital and Timiie, ). 13. I have since become convincedthat Materialist is better.2 Cannan, pp. 145-50.

I But there was at least one important AmericanFundist, F. W. Taussig. Irving Fisher is harder toplace, since he, at least sometimes, could see both sides.But it is interesting to find that Cannan thoughtlisher, like himself, to be a Materialist.

7/27/2019 1816058.pdf

5/11

310 AMERICANECONOMIC ASSOCIATION MAY 1974Let us go back to the Classics. Why werethe Classical Economists Fundists? It isnot easy to see, just from looking at their

works; they take their Fundism so muchfor granted that they do not need tojustify it. Surely the reason is that it cameto them from outside-from business prac-tice, from accounting practice.Even to this day, accountants areFundists. It is not true, accountants willinsist, that the plant and machinery of afirm are capital; they are not capital, theyare assets. Capital, to the accountant, ap-pears on the liabilities side of the balancesheet; plant and machinery appear on theassets side. Capital, accordingly, is a Fundthat is embodied in the assets.The origin of accounting is in the busi-ness of the merchant; accounting cate-gories, to this day, bear the mark of theirmercantile origin. It was the merchant whowas the original Fundist. It is the merchantwho thinks of his capital as a Fund that isinvested in a stock of salable goods. It is inthe Fund sense that capital "circulates";the physical goods do not circulate, butthe Fund does. It is the Fund that is"turned over." The stock of goods in themerchant's possession is one thing (themost he will admit is that it is the formthat is taken by his capital at the moment);his capital, he will surely say, is somethingmore permanent.These were the business terms whichcame naturally to the Classical Econo-mists. They had no reason to depart un-necessarily from the businessman's lan-guage. This was the system of thought thatthe businessmen of their time were using;they just followed it. It is true that theywere thinking of the whole (national)economy, not of the single business; butthis, they surely felt, made no difference.It did not need to make much difference.They could think (as Henry Thornton, inparticular, most surely did think) of the

whole economy as having a balance sheet,constructed by consolidation of the bal-ance sheets of the single businesses; in theconsolidation, the liabilities of one unitwould be cancelled against thd assets ofanother, but no item would be transferredfrom the liabilities to the assets side. So,even when all debts and paper claims hadbeen cancelled, there would remain on theassets side the real goods (and balance ofexternal claims), on the liabilities side theCapital-still a Fund. It need not bethought of as a debt owed by the nation toitself; it is the same kind of thing as theCapital of the single business.The way would thus be open for Classi-cal Fundism if the whole economy con-sisted of merchants; how far, however,could it cope with businesses of otherkinds? It was necessary, from the start, todeal with businesses of other kinds; but forthe first of the extensions that were needed,Fundism did not do at all badly. It is oftenthought that the notion of capital as "ad-vances to laborers" took its origin fromobservation of agriculture, so it is labelledphysiocratic; and it is true that if one looksonly at the economic literature it is with theFrench physiocrats that it seems to comein. But so far at least as the British Classi-cal Economists are concerned, it is moreconvincingly interpreted as a fitting ofagricultural experience into the mercantilepattern. The farmer, like the merchant,"turns over" his capital, buying the servicesof labor, as the merchant buys his stock intrade; selling the product of that laborwhen it is ready to be marketed. So theFundist concept of capital could be carriedover to agriculture, surviving the transi-tion. It seemed, on the whole, quite a goodfit.

The farmer, of course, used land as wellas capital, but that land was a separatefactor of production no one doubted. Therent of the land must be deducted from

7/27/2019 1816058.pdf

6/11

VOL. 64 NO. 2 TWENTIETH CENTURY ECONOMIC THEORY 311the "gross" profit on any agricultural op-eration before the "net" profit was arrivedat; it would be the rate of net profit on theCapital Fund that competition would tendto equalize.

Classical economics was three-factoreconomics, and we can now see that thetriad had deeper roots than is commonlysupposed. Labor is a flow, land is a stock(as stock and flow are used in moderneconomics); but capital is neither stock norflow-it is a Fund. Each of the three fac-tors has its own attribute, applicable to it-self but to neither of the others. Laborworks on land through capital, not on capi-tal nor with capital. The place of each ofthe factors in the productive process issharply distinguished.The Classical Economists, so interpreted,are rather consistent; among their succes-sors, in the latter part of the century, con-sistency disappears. Not all of those whowent Marginalist went Materialist, andthose who did go Materialist did so indifferent ways. The case of Walras, forinstance, is quite peculiar. I feel fairlysure that he is to be reckoned as a Ma-terialist; but the reason for his Materialismis his interest in particular capital goods,appearing (of course) in his work as a partof his general determination to work withan n-goods model. Yet his Materialismmay antedate that determination, andmay have been one of the things which im-pelled him towards it. He says that he tookhis view of capital from his father andAuguste Walras (writing in 1849) wouldcertainly seem to have been a Materialist,even an extreme Materialist. "Capitals"(capitaux) to him are capital goods; "in-comes" (revenus) are income goods; theyare distinguished by multiple (successive)uses against single uses and by that alone.Carriages, carts, steam-engines are capi-tals; a glass of wine, a round of beef, acandle are incomes. "Le revenu, ainsi que

son nom l'indique, c'est ce qui revient; or, cequi revient, c'est ce qui s'en va." (pp. 53-54).4One must yet beware, in father and sonalike, of mistaking for a theoretical ap-proach what is no more than a peculiarityof the French language, the restrictiveness,the deliberate restrictiveness, of its vocab-ulary. It may well be that much FrenchMaterialism is only apparent, a matter oflinguistics rather than economics.The case of Marshall is here more inter-esting. Marshall's Materialism is muchmore clearly to be explained by events-bythe now achieved Industrial Revolution-the rise in importance of plant and ma-chinery. The Classical schema, as we haveseen, began with trade and was extendedto agriculture; so long as stocks and workin progress were the main part of the man-ufacturer's physical assets it could be ex-tended, in much the same way, to manu-facture also. But when a large part of hiscapital became fixed in plant and ma-chinery, a candidate had appeared forfactor status, which did not fit into theclassical triad. What was to be done?Before considering what happened in theeconomics, it will be useful to turn againto the accounting aspect. The rise of theMachine had already presented the ac-countant with a parallel problem. Whatdid the accountant do about it?

So long as he had nothing to considerbut mercantile transactions, his task inprinciple was simple. For it is character-istic of the business of the merchant thatit is divisible into separate units. Everybale of cotton or pound of cheese whichever forms part of his stock is acquired at aparticular date and sold at a particulardate; purchase, retention and sale con-stitute a separable transaction. (CompleteI ("Revenue [the return], as its name indicates, isthat which comes back; now, that which comes back,

is that which went off."-Editors.)

7/27/2019 1816058.pdf

7/11

312 AMERICAN ECONOMIC ASSOCIATION MAY 1974separation is of course not attained inpractice, since there are overheads whichhave to be allocated; but it is so nearlyattained that it sets the pattern.) So theaccounts of the merchant may be regardedas a bundle of separate accounts. Pur-chases and sales are indeed going on con-tinually, so that if they are set out in atime sequence, the separate accounts willoverlap; it is only if an account were pre-pared for the whole history of the business,from first setting-up to final closing down,that a record of purchases and sales wouldtell the whole story. In any arbitrary, say,annual, period there must be transactionswhich have started before the beginning ofthe period, but are completed within theperiod; there must similarly be transac-tions which are begun within the periodbut not completed when it is ended. These,however, in the case of a mercantile busi-ness cause the minimum of trouble. Theycan be dealt with by the accepted rule ofnever taking a profit on any transactionbefore it is completed. The initial stock ofthe year will then be brought in at cost; andthe final stock will be valued at cost in thesame manner.5But what was to be done, on these prin-ciples, with plant and machinery? The useof land, being regarded as a permanency,could be brought in as a regular charge;but the plant and machinery is not ex-pected to last indefinitely, though its useis spread over a time which is longer thanthe accounting period. It is important toobserve that it is the extension of the useto a duration which is longer than the ac-counting period which creates the diffi-culty. There would be no difficulty, here asin the mercantile case, if the account weredrawn up for the whole life of the business,

from first setting-up to final closing-down.6It is for the annual account that there isthe problem. The cost of the machine hasto be set against a series of sales, the salesof the outputs to which it contributes, butsome of these sales are sales of the presentyear, some are later and some, maybe,earlier. There is thus a problem of imputa-tion; how much is to be reckoned into thecosts of this year, and how much into thecosts of other years? It is just the sameproblem as the allocation of overheads, andto that, as is now well known, there is nofirm economic solution.Neither has the accountant found a solu-tion only a name and a set of (essentiallyarbitrary) rules. The "depreciation quotas"must add to unity, but that is all that isknown, at all firmly, about them. The formof the account is preserved, but only bybringing in, as the capital which is sup-posed to be invested in the machine at thebeginning of the year, that part which hasnot been absorbed (by being allocated as acost to the output of preceding years) andby reckoning as the capital invested at theend of the year that part which has notbeen absorbed in those years nor in thisyear so that it is left to be carried for-ward to the future. This is in fact what ac-countants did (probably what they had todo) as soon as they were confronted withthe problem. It is what they still do, evento this day.We have had plenty of opportunity inthe present century to understand howarbitrary the accountant's depreciationquotas are. We have seen them batteredby inflation, and we have seen themmanipulated by tax authorities in the in-terests of fiscal policy. Late nineteenth cen-tury economists had much less of this ex-

I There is of course the qualification that an ex-psected oss mav be taken in advance: "cost or marketvalue whichever is the lower." But this does not affectthe principle.

6 It was the realization that the economist, unlike theaccountant, need not be bound down by annuality,unless he chooses, that prompted me to write Capitaland Time.

7/27/2019 1816058.pdf

8/11

VOL. 64 NO. 2 TWENTIETH CENTURY ECONOMIC THEORY 313perience, so it is not surprising to find thatthey began by taking the accountant's de-preciation quotas much too seriously. TIhatis true of both schools. It is true of the sur-viving Fundists (such as the Austrians)who conceived of investment in fixed capi-tal as equivalent to a bundle of invest-ments in circulating capital. If one-tenthof the cost of a machine of ten-year lifecould be imputed as cost to the product ofeach year, the machine was equivalent toten investments in circulating capital, onewith an investment period of one year, oneof two years . .. and one of ten years. Itwas the accountant's depreciation quotaswhich did the trick.It is also true of Marshall. He also reliedupon the accountant's "solution," but in adifferent way. It was an essential elementin his concept of long-period equilibrium.In the short period, Marshall tells us,when "the producers have to adjust theirsupply to the demand as best they can withthe appliances at their disposal," the "in-come" derived from those "appliances" isa quasi-rent.7 It is called a quasi-rent byanalogy with the rent of land; in all of hisshort-period theory Marshall has Ricar-do's rent theory very much in mind.Now the rent which is determined as asurplus, in Ricardo's manner, makes noallowance for depreciation; in Ricardo,land being "indestructible," no such allow-ance is of course required. Marshall, how-ever, does think of a deduction being made,evenin the short period, though Ricardo hasgiven him no help in determining whatthat deduction should be. It is true thatMarshall is so reticent on the matter thathe can easily be misunderstood; one has toread him quite carefully to discoverwhether quasi-rents are to be taken grossor net. He does however say in a defini-tional chapter of Principles:

We cannot properly speak of the

interest y-ieldedby( a machine. If we uisethe term interest at all, it mutst be inrelation not to the machine itself but toits monev valtue.For instance if the workdone by a machine which cost ?100 isworth ?4 a year net, that machine isyielding a quasi-rent of ?4 which isequivalent to interest at four per cent,on its original cost; but if the machineis worth only ?80 now, it is yieldingfive per cent on its present value.[pp. 74-75]

This seems conclusive.It is indeed quite remarkable how little

there is in Marshall's book about deprecia-tion. There is a footnote in which he recog-nizes that it is a problem, but the footnotejust ends with a reference to an accountingtextbook.8 He has evidently decided thatfor his purposes, the accountant's solutionwill do. Gross can in that way be reducedto net, and it is net returns that areequalized by competition in the longperiod.That is what Marshall says, but in theforty years which followed on the publica-tion of his Principles, the strangest thingsmust have happened to teaching on thispoint, even among his closest followers.One had heard rumors that there was agood deal of confusion between "gross"and "net" among Cambridge teachers, anda strong piece of evidence in support ofthem has now come to hand. In his chapteron the marginal efficiency of capital in theGeneral Theory, Keynes is careful not tocall his Q's quasi-rents, and rightly so,since they are gross of depreciation, so theyare not what Marshall meant by quasi-rent. But why are they symbolized by Q?It has naturally been assumed that Qstands for quasi-rent; so the Q's are calledquasi-rents in many post-Keynesian writ-ings. We now know that in an earlier draftthe Q's were called quasi-rents.9 Keynesmade a correction, or semicorrection, in

7 Marshall, 1).376.8 Marshall, pp. 354-55.9 Keynes, pp. 425-26.

7/27/2019 1816058.pdf

9/11

314 AMERICAN ECONOMIC ASSOCIATION MAY 1974the final version, but the confusion is be-trayed.The fact is that until the new wind be-gan to blow, in the mid-twenties, verylittle interest had been taken in Cam-bridge in capital theory. Take the case ofPigou. One of the remarkable changes in-troduced into the later editions of TheEconomics of Welfare is a chapter entitled"Maintaining Capital Intact." There wasno such chapter in the first or second edi-tion. It makes its appearance in the third(1928), by which time the problem of de-fining saving, in conditions of changingprices, had been brought to light in thework of Robertson (1926). In the fourthedition (1932) it is considerably altered;and there is a further version, in an im-portant article (1935) which looks as if itwas intended as the basis of another re-vision.10 Pigou, it is clear, was verybothered; and one can see why.It was only a part of the problem of capi-tal with which he thought himself to beconcerned; he was trying to look at thatone part in isolation. In spite of its "Wel-fare" coloring, the subject of his book wasthe Social Product (or "dividend" as hecalls it): how it is measured, what makesit large or small, how it is distributed. TheSocial Product of one particular period( "year") is considered, almost, in isola-tion. Having chosen that way of posinghis problem, he is led (almost inevitably)to what would nowadays be called a "Pro-duction Function" approach. Even so, hemight have found himself confronting thegeneral problem of measuring capital-forhow can we make a static comparison, be-tween two economies whose capital stocksare different, without having some meansof comparing their capital stocks? Pigoudid not, at first, raise this wide question,though Robertson, at much the same date,

could already see that it was involved.1'He confined himself to what arose in themeasurement of income, in the single year

the measurement of the investment com-ponent, the reduction of gross investmentto net. (This, of course, already implies aproblem of capital measurement thecomparison of the beginning-year and theend-year capital stocks.)Pigou's approach is strictly Materialist.He does indeed recognize that the businessconcept of capital is different, but

for economics the stock of capital exist-ing at any time is a collection of objects,the extent of which is a purely physicalfact. . . . the size of the stock is not . . .aflected by its value; it is exactly thesame . .. whether that value is large orsmall. [1935, p. 235]

He goes on to draw from this the importantconclusion, echoes of which are to be foundin many more recent writings:A distinction should be drawn betweenchanges which, while leaving the ele-

ment still as productive as ever, bringnearerthe day of sudden and final break-down, and physical changes whichreduce its current productivity and sorentable value. With the former sort ofchange, uintilthe breakdown occurs, thecapital stock is, I suggest, best regardedas intact, just as it is best regarded asintact despite the nearer approach of adav that will make a part of it obsolete.[1935, p. 238]I shall not discuss Pigou's treatment in

detail; enough has been said to indicate itsMaterialist character. For the purposewhich Pigou had in mind, it may well bedefensible (as I shall show), but it canhardly be regarded as general; so it is notsurprising that there were quarters whereit was not well received. I think not so

10 Pigou (1935).

1' See Robertson's paper "Wage Grumbles," reprintedin Robertson (1931). I have a particular affection forthat paper since it was through it that I myself firstcame into touch with its author.

7/27/2019 1816058.pdf

10/11

7/27/2019 1816058.pdf

11/11

316 AMERICANECONOMIC ASSOCIATION MAY 1974example of this in the book by J. R. S.Revell, The Wealth of the Nation (1967), aninquiry conducted in Cambridge, England,a place that must often appear on this sideof the Atlantic as a headquarters of theNew Fundism. Revell's calculation of theNational Capital of Britain would haveappealed to Pigou;15 he had to be Material-ist, because as a statistician he could benothing else.16 There is nothing else thatcan be used in macroeconometrics; so it ismacroeconometrics itself which is on trial-but that, perhaps, is as it should be.

REFERENCESE. Cannan, A Review of Economic Theory,London 1929.

F. Hayek, "The Maintenance of Capital,"Economica, Aug. 1935, 2, 241-74.-----, "Maintaining Capital Intact," Eco-

nomica, Aug. 1941a, 8, 276-80.Pure Theory of Capital, London1941b.J. R. Hicks, "Maintaining Capital Intact,"Economica, May 1942, 9, 174-79., Social Framework,4th ed., New York1971., Capital and Time, Oxford 1973.J. M. Keynes, Collected Writings, Londonand New York 1971-, 13.A. Marshall, Principles, 8th ed., London,1922.A. C. Pigou, Economics of Welfare, 3rd and4th ed., London 1928, 1932."Net Income and Capital Depletion,"Econ. J., June 1935, 178."Maintaining Capital Intact," Eco-nomica, Aug. 1941, 8, 271-75.D. H. Robertson, Banking Policy and thePrice Level, New York 1926., Economic Fragments, London 1931.F. W. Taussig, Wagesand Capital, New York1896.A. Walras, Theorie de la Richesse Sociale,Paris 1849.

15 I am aware that in his last note on the subject,Pigou (1941) dissociated himself from the replacementcost measure of capital, pointing to the case where thearticle to be replaced "had become impossible to pro-duce" as a reason for rejecting it. I do not think he waswell advised to make this concession. He would havedone better to insist that it was one of the "odd cases"that he was leaving out.16 For further discussion of this matter of measure-

ment, see Hicks (1973), Chapter 13, and (1971),Appendix D.