100 <strong>Piero</strong> <strong>Sraffa</strong>directly required for its production, plus its commodity inputs; theoperation is then repeated on the latter, until we have a series (aslong as we like) of dated labour inputs and a residuum (as small aswe like) of commodities. <strong>Sraffa</strong> (1960: 34) calls this procedure ‘reductionto dated quantities of labour’. We can then compute the averageperiod of production by taking a weighted average of the intervalsof time between the date of each direct labour input and the dateon which the output is obtained, where for each interval the correspondingamount of direct labour input is utilised as weight, once thetotal amount of labour directly or indirectly required to obtain thecommodity under consideration has been set equal to one. 5 Austriancapital theory then interprets the average period of production as ameasure of the quantity of capital employed in the production process,thus considering ‘time’, together with labour, as the factors ofproduction.The rate of interest is thus obtained by the balancing of two forces.On the one hand we have the supply of capital, namely waiting, correspondingto the readiness of economic agents to postpone consumption:the length of time agents are willing to wait is assumed to be apositive function of the rate of interest. On the other hand, we have thedemand for capital, namely the relationship between additional waiting(increased length of the average period of production) and additionalproduct; the postulate of decreasing marginal productivity implies adecreasing relation between the average period of production and therate of interest. Thus, the rate of interest can be considered as the priceof ‘capital’, determined by the usual mechanism of equilibrium betweensupply and demand.This construction is criticised by <strong>Sraffa</strong> (1960: 37–8). The point isthat the average period of production is computed without allowingfor compound interest; when it is considered, the results may changedramatically. Thus <strong>Sraffa</strong> shows that if the inputs of the various productiveprocesses are reduced to dated quantities of labour, when the rate ofprofits changes we can have ‘complicated patterns of price-movementswith several ups and downs’. This is shown with an example, where theprice of product a (‘old wine’) at first rises, then falls, then rises againrelatively to product b (‘oak chest’) as the rate of profits increases fromzero to its maximum value. The reversals in the direction of the movementof relative prices, in the face of unchanged methods of production,cannot be reconciled with any notion of capital as a measurable5For an algebraic treatment cf. Kurz and Salvadori (1995: 437).Critique of the Marginalist Approach 101quantity independent of distribution and prices’. 6 The difficulty hadalready been sensed by Wicksell (1901), but later exponents of theAustrian school went on utilising the notion of the average period ofproduction. In particular, Hayek (1931a) built his analysis of employmentand the trade cycle on it.The full implications of <strong>Sraffa</strong>’s criticism were not immediatelygrasped. In a review of <strong>Sraffa</strong>’s book, Harrod (1961) tried to defendthe average period of production by pointing out that it can alwaysbe calculated, given the rate of profits. Apparently, Harrod failed torealise that in such conditions the average period of production canno longer be used to explain the distribution of income, for its verydefinition depends on an exogenously given rate of profits, as <strong>Sraffa</strong>(1962) pointed out in a short reply to Harrod. This, of course, is preciselythe import of <strong>Sraffa</strong>’s original criticism of the Austrian methodof measuring ‘capital’.The difficulties illustrated above must be borne in mind also whenevaluating later attempts at utilising dated quantities of labour forthe analysis of dynamic issues. Reference here is to the so-called ‘neo-Austrian’ approach proposed by Hicks (1973) for the analysis of suchissues as the transition between different technologies. In fact, Hicks’smodel involves both the use of a static framework for the analysis ofdynamic issues and a serious underevaluation of the capital theorydifficulties mentioned above. Let us consider this issue in somewhatmore detail.<strong>Sraffa</strong>’s analysis makes it clear – and indeed the point was deniedneither by Böhm-Bawerk nor by Hayek – that the reduction to datedquantities of labour is a theoretical construct, simply presenting in adifferent way the technology which underlies the Sraffian system ofsimultaneous equations illustrated earlier (§ 3.2) and not a historicalreconstruction of the way in which the different means of productionhave actually been obtained. Marginalist capital theory aims atdetermining static equilibrium solutions, hence marginalist analysisof technical change refers to static substitution between capital andlabour; technological changes over historical time are not considered.This should be borne in mind for two reasons. First, it is clear6A critique similar to <strong>Sraffa</strong>’s was developed by Garegnani (1960), with a directanalysis of the theories of the various authors who made similar attempts toconstruct a theory of distribution based on this conception. The criticisms ofthe average period of production are now generally accepted. Cf. for exampleSamuelson (1966).
102 <strong>Piero</strong> <strong>Sraffa</strong>that the difficulties in capital theory stemming from the existence ofa multiplicity of commodities cannot be overcome by shifting to apresentation of the technology in terms of dated labour inputs: if nonew restrictive assumption is in one way or another introduced in theshift, such difficulties cannot but reappear in the latter presentationas well. 7 Second, what is analysed with the reduction to dated quantitiesof labour are the implications of a given technology, ruling at agiven moment in time: the presence of dated quantities is an analyticalconstruct, which does not correspond to periods of historical time.Comparison between two different technologies is simply an exercisein static comparative analysis. 8Analysis of what Hicks calls ‘traverse’, namely the transition betweentwo different technologies, involves either historical analysis, leavingaside any attempt at theoretical construction, or an exercise in comparativestatic analysis: the comparison, that is, between an initial anda final equilibrium. The latter case implies that there must be a uniqueequilibrium both in the initial and in the final position. Also, analysisof the ‘out of equilibrium’ transition between the two techniquesrequires specific assumptions concerning the ‘laws of movement’ of thevariables out of the equilibrium position, which in turn can give definiteresults only under restrictive assumptions. Typical in this respect isthe (usually tacit) assumption of no basic commodities in the model – anassumption even more restrictive than that of a one-(basic)-commodityworld.Even in the presence of just one basic commodity, the series oflabour inputs is potentially infinite: the residuum of commodities,though small as we like, can never be fully eliminated; however small,it becomes all-important in the determination of the price system whenthe rate of profits is at its maximum. This leads us to conclude that,7For illustration of how Hicks’s (1973) simplifying assumptions allow him to circumventreswitching and other problems in capital theory, and more generallyfor a detailed analysis of his book, cf. Burmeister (1974).8In modern terminology, it is common to place Böhm-Bawerk’s theory amongthe analyses of intertemporal equilibrium due to the fact that the rate of interestis interpreted, on the consumption side, as a rate of intertemporal preference.However, on the side of production the choice between alternative techniques(represented by different lengths of the production period and the correspondingdifferent levels of productivity: more ‘roundabout’ techniques are assumed to bemore productive) takes place on the basis of a given – hence static – technicalknowledge. Thus, intertemporal equilibrium is a static construct, being determinedwith reference to a given state of technology, and so to a given momentin time.Critique of the Marginalist Approach 103while the simultaneous equations method and the series of dated labourinputs can be considered as equivalent ways of representing technology,9 the former method is safer. In other words, no result derived underthe second method that cannot also be reached by the simultaneousequations method can be accepted as having full generality in a multicommodityworld with basic commodities.6.3 Critique of capital as a factor of productionThe traditional marginalist theories (those theories that Keynes misleadinglycalled ‘classical’, pointing as an example to Pigou’s analysis)have as their central tenet the thesis that an economic system whereperfect competition prevails, externalities are absent and which is notsubject to repeated exogenous disturbances, tends to an equilibriumposition endowed with characteristics of Pareto optimality, in thesense that it is not possible to improve the position of any economicagent without worsening the position of some other. In particular,traditional marginalist theories maintain that under perfect competitionreal wages move towards a level which ensures equality betweendemand and supply of labour, or in other words full employment.Among the automatic equilibrating mechanisms bringing the economytowards full employment, traditional marginalist theories stress theflexibility of the capital–labour ratio: if the real wage falls under thepressure of unemployment, firms will find it more profitable to adoptproductive techniques with a lower capital–labour ratio, so that a givenendowment of capital becomes compatible with the employment ofan increasing number of workers; increase in real wages and the consequentfall in the capital–labour ratio proceeds until full employmentis reached.This thesis takes different forms with authors belonging to differentstreams of the marginalist approach. Garegnani (1960) examines thetheories developed by a few representative writers within this tradition(Walras, Böhm-Bawerk, Wicksell), bringing out explicitly the criticismsformulated in their most essential terms in <strong>Sraffa</strong> (1960). Here it is alsoworth pointing out that <strong>Sraffa</strong>’s critique is more general than that developed(on at least partly parallel lines) by Joan Robinson (1953), whodirectly refers to the aggregate notion of capital used in the so-calledaggregate production function. <strong>Sraffa</strong>’s critique of marginalist theoriesrefers more generally to the very idea that the ‘prices’ of ‘factors of9For a clear illustration of this fact, cf. Kurz and Salvadori (1995, Chapter 6).
- Page 1 and 2:
Piero SraffaAlessandro Roncaglia
- Page 3 and 4:
ContentsList of FiguresIntroduction
- Page 5 and 6: Introduction ixWith this degree of
- Page 7 and 8: 2 Piero Sraffa(1874-1961), professo
- Page 9 and 10: 6 Piero Sraffarevaluation of the li
- Page 11 and 12: 10 Piero Sraffaadministration of th
- Page 13 and 14: 14 Piero Sraffa1.4 Imperfect compet
- Page 15: 18 Piero SraffaIn many fields of ec
- Page 18 and 19: 24 Piero SraffaAn Italian in Cambri
- Page 20 and 21: 28 Piero Sraffanot something fixed,
- Page 22 and 23: 32 Piero Sraffamonetary factors on
- Page 24 and 25: 36 Piero Sraffapartnered in his lab
- Page 26 and 27: 40 Piero SraffaActually, there was
- Page 28 and 29: 44 Piero Sraffadistribution of the
- Page 30 and 31: 48 Piero SraffaLet us recall at thi
- Page 32 and 33: 52 Piero Sraffathe other hand, the
- Page 34 and 35: 56 Piero Sraffaof production. 24 Bu
- Page 36 and 37: 4Basic and Non-Basic Products4.1 Ba
- Page 38 and 39: 64 Piero SraffaA line of argument s
- Page 40 and 41: 68 Piero Sraffathe system stemming
- Page 42 and 43: 72 Piero Sraffaplan that would yiel
- Page 44 and 45: 76 Piero Sraffaproduced less quanti
- Page 46 and 47: 80 Piero Sraffaterms of labour comm
- Page 48 and 49: 84 Piero Sraffaof value is, and mus
- Page 50 and 51: 88 Piero Sraffabeing invariant to c
- Page 52 and 53: 92 Piero Sraffa(variable plus const
- Page 54 and 55: 96 Piero Sraffaconsumption goods),
- Page 58 and 59: 104 Piero Sraffaproduction’ (iden
- Page 60 and 61: 108 Piero SraffaCritique of the Mar
- Page 62 and 63: 112 Piero SraffaThe growing remoten
- Page 64 and 65: 116 Piero Sraffareturns: Sraffa’s
- Page 66 and 67: 120 Piero SraffaFurthermore, the cl
- Page 68 and 69: 124 Piero SraffaIn this way the pro
- Page 70 and 71: 128 Piero SraffaSraffa raised again
- Page 72 and 73: 132 Piero Sraffaconnected, but can
- Page 74 and 75: 136 Piero SraffaThe bridge between
- Page 76 and 77: 140 Piero SraffaSraffa’s work for
- Page 78 and 79: 144 Piero SraffaThis debate is stil
- Page 80 and 81: 148 Piero SraffaObviously the ‘Ma
- Page 82 and 83: 152 Piero SraffaIn comparison to th
- Page 84 and 85: 156 Piero Sraffaof the path actuall
- Page 86 and 87: 160 Piero SraffaHowever, this const
- Page 88 and 89: 164 ReferencesReferences 165——
- Page 90 and 91: 168 ReferencesReferences 169Levhari
- Page 92 and 93: 172 ReferencesReferences 173——
- Page 94 and 95: 176 ReferencesReferences 177——
- Page 96: 180 IndexIndex 181Marx K., 10, 29,